diff --git a/include/sys/spa.h b/include/sys/spa.h index 1faf0420f21a..c989c0b35cd4 100644 --- a/include/sys/spa.h +++ b/include/sys/spa.h @@ -1,904 +1,905 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2014 by Delphix. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. */ #ifndef _SYS_SPA_H #define _SYS_SPA_H #include #include +#include #include #include #include #include #ifdef __cplusplus extern "C" { #endif /* * Forward references that lots of things need. */ typedef struct spa spa_t; typedef struct vdev vdev_t; typedef struct metaslab metaslab_t; typedef struct metaslab_group metaslab_group_t; typedef struct metaslab_class metaslab_class_t; typedef struct zio zio_t; typedef struct zilog zilog_t; typedef struct spa_aux_vdev spa_aux_vdev_t; typedef struct ddt ddt_t; typedef struct ddt_entry ddt_entry_t; typedef struct zbookmark_phys zbookmark_phys_t; struct dsl_pool; struct dsl_dataset; /* * General-purpose 32-bit and 64-bit bitfield encodings. */ #define BF32_DECODE(x, low, len) P2PHASE((x) >> (low), 1U << (len)) #define BF64_DECODE(x, low, len) P2PHASE((x) >> (low), 1ULL << (len)) #define BF32_ENCODE(x, low, len) (P2PHASE((x), 1U << (len)) << (low)) #define BF64_ENCODE(x, low, len) (P2PHASE((x), 1ULL << (len)) << (low)) #define BF32_GET(x, low, len) BF32_DECODE(x, low, len) #define BF64_GET(x, low, len) BF64_DECODE(x, low, len) #define BF32_SET(x, low, len, val) do { \ ASSERT3U(val, <, 1U << (len)); \ ASSERT3U(low + len, <=, 32); \ (x) ^= BF32_ENCODE((x >> low) ^ (val), low, len); \ _NOTE(CONSTCOND) } while (0) #define BF64_SET(x, low, len, val) do { \ ASSERT3U(val, <, 1ULL << (len)); \ ASSERT3U(low + len, <=, 64); \ ((x) ^= BF64_ENCODE((x >> low) ^ (val), low, len)); \ _NOTE(CONSTCOND) } while (0) #define BF32_GET_SB(x, low, len, shift, bias) \ ((BF32_GET(x, low, len) + (bias)) << (shift)) #define BF64_GET_SB(x, low, len, shift, bias) \ ((BF64_GET(x, low, len) + (bias)) << (shift)) #define BF32_SET_SB(x, low, len, shift, bias, val) do { \ ASSERT(IS_P2ALIGNED(val, 1U << shift)); \ ASSERT3S((val) >> (shift), >=, bias); \ BF32_SET(x, low, len, ((val) >> (shift)) - (bias)); \ _NOTE(CONSTCOND) } while (0) #define BF64_SET_SB(x, low, len, shift, bias, val) do { \ ASSERT(IS_P2ALIGNED(val, 1ULL << shift)); \ ASSERT3S((val) >> (shift), >=, bias); \ BF64_SET(x, low, len, ((val) >> (shift)) - (bias)); \ _NOTE(CONSTCOND) } while (0) /* * We currently support nine block sizes, from 512 bytes to 128K. * We could go higher, but the benefits are near-zero and the cost * of COWing a giant block to modify one byte would become excessive. */ #define SPA_MINBLOCKSHIFT 9 #define SPA_MAXBLOCKSHIFT 17 #define SPA_MINBLOCKSIZE (1ULL << SPA_MINBLOCKSHIFT) #define SPA_MAXBLOCKSIZE (1ULL << SPA_MAXBLOCKSHIFT) #define SPA_BLOCKSIZES (SPA_MAXBLOCKSHIFT - SPA_MINBLOCKSHIFT + 1) /* * Size of block to hold the configuration data (a packed nvlist) */ #define SPA_CONFIG_BLOCKSIZE (1ULL << 14) /* * The DVA size encodings for LSIZE and PSIZE support blocks up to 32MB. * The ASIZE encoding should be at least 64 times larger (6 more bits) * to support up to 4-way RAID-Z mirror mode with worst-case gang block * overhead, three DVAs per bp, plus one more bit in case we do anything * else that expands the ASIZE. */ #define SPA_LSIZEBITS 16 /* LSIZE up to 32M (2^16 * 512) */ #define SPA_PSIZEBITS 16 /* PSIZE up to 32M (2^16 * 512) */ #define SPA_ASIZEBITS 24 /* ASIZE up to 64 times larger */ /* * All SPA data is represented by 128-bit data virtual addresses (DVAs). * The members of the dva_t should be considered opaque outside the SPA. */ typedef struct dva { uint64_t dva_word[2]; } dva_t; /* * Each block has a 256-bit checksum -- strong enough for cryptographic hashes. */ typedef struct zio_cksum { uint64_t zc_word[4]; } zio_cksum_t; /* * Each block is described by its DVAs, time of birth, checksum, etc. * The word-by-word, bit-by-bit layout of the blkptr is as follows: * * 64 56 48 40 32 24 16 8 0 * +-------+-------+-------+-------+-------+-------+-------+-------+ * 0 | vdev1 | GRID | ASIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 1 |G| offset1 | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 2 | vdev2 | GRID | ASIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 3 |G| offset2 | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 4 | vdev3 | GRID | ASIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 5 |G| offset3 | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 6 |BDX|lvl| type | cksum |E| comp| PSIZE | LSIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 7 | padding | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 8 | padding | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 9 | physical birth txg | * +-------+-------+-------+-------+-------+-------+-------+-------+ * a | logical birth txg | * +-------+-------+-------+-------+-------+-------+-------+-------+ * b | fill count | * +-------+-------+-------+-------+-------+-------+-------+-------+ * c | checksum[0] | * +-------+-------+-------+-------+-------+-------+-------+-------+ * d | checksum[1] | * +-------+-------+-------+-------+-------+-------+-------+-------+ * e | checksum[2] | * +-------+-------+-------+-------+-------+-------+-------+-------+ * f | checksum[3] | * +-------+-------+-------+-------+-------+-------+-------+-------+ * * Legend: * * vdev virtual device ID * offset offset into virtual device * LSIZE logical size * PSIZE physical size (after compression) * ASIZE allocated size (including RAID-Z parity and gang block headers) * GRID RAID-Z layout information (reserved for future use) * cksum checksum function * comp compression function * G gang block indicator * B byteorder (endianness) * D dedup * X encryption (on version 30, which is not supported) * E blkptr_t contains embedded data (see below) * lvl level of indirection * type DMU object type * phys birth txg of block allocation; zero if same as logical birth txg * log. birth transaction group in which the block was logically born * fill count number of non-zero blocks under this bp * checksum[4] 256-bit checksum of the data this bp describes */ /* * "Embedded" blkptr_t's don't actually point to a block, instead they * have a data payload embedded in the blkptr_t itself. See the comment * in blkptr.c for more details. * * The blkptr_t is laid out as follows: * * 64 56 48 40 32 24 16 8 0 * +-------+-------+-------+-------+-------+-------+-------+-------+ * 0 | payload | * 1 | payload | * 2 | payload | * 3 | payload | * 4 | payload | * 5 | payload | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 6 |BDX|lvl| type | etype |E| comp| PSIZE| LSIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 7 | payload | * 8 | payload | * 9 | payload | * +-------+-------+-------+-------+-------+-------+-------+-------+ * a | logical birth txg | * +-------+-------+-------+-------+-------+-------+-------+-------+ * b | payload | * c | payload | * d | payload | * e | payload | * f | payload | * +-------+-------+-------+-------+-------+-------+-------+-------+ * * Legend: * * payload contains the embedded data * B (byteorder) byteorder (endianness) * D (dedup) padding (set to zero) * X encryption (set to zero; see above) * E (embedded) set to one * lvl indirection level * type DMU object type * etype how to interpret embedded data (BP_EMBEDDED_TYPE_*) * comp compression function of payload * PSIZE size of payload after compression, in bytes * LSIZE logical size of payload, in bytes * note that 25 bits is enough to store the largest * "normal" BP's LSIZE (2^16 * 2^9) in bytes * log. birth transaction group in which the block was logically born * * Note that LSIZE and PSIZE are stored in bytes, whereas for non-embedded * bp's they are stored in units of SPA_MINBLOCKSHIFT. * Generally, the generic BP_GET_*() macros can be used on embedded BP's. * The B, D, X, lvl, type, and comp fields are stored the same as with normal * BP's so the BP_SET_* macros can be used with them. etype, PSIZE, LSIZE must * be set with the BPE_SET_* macros. BP_SET_EMBEDDED() should be called before * other macros, as they assert that they are only used on BP's of the correct * "embedded-ness". */ #define BPE_GET_ETYPE(bp) \ (ASSERT(BP_IS_EMBEDDED(bp)), \ BF64_GET((bp)->blk_prop, 40, 8)) #define BPE_SET_ETYPE(bp, t) do { \ ASSERT(BP_IS_EMBEDDED(bp)); \ BF64_SET((bp)->blk_prop, 40, 8, t); \ _NOTE(CONSTCOND) } while (0) #define BPE_GET_LSIZE(bp) \ (ASSERT(BP_IS_EMBEDDED(bp)), \ BF64_GET_SB((bp)->blk_prop, 0, 25, 0, 1)) #define BPE_SET_LSIZE(bp, x) do { \ ASSERT(BP_IS_EMBEDDED(bp)); \ BF64_SET_SB((bp)->blk_prop, 0, 25, 0, 1, x); \ _NOTE(CONSTCOND) } while (0) #define BPE_GET_PSIZE(bp) \ (ASSERT(BP_IS_EMBEDDED(bp)), \ BF64_GET_SB((bp)->blk_prop, 25, 7, 0, 1)) #define BPE_SET_PSIZE(bp, x) do { \ ASSERT(BP_IS_EMBEDDED(bp)); \ BF64_SET_SB((bp)->blk_prop, 25, 7, 0, 1, x); \ _NOTE(CONSTCOND) } while (0) typedef enum bp_embedded_type { BP_EMBEDDED_TYPE_DATA, BP_EMBEDDED_TYPE_RESERVED, /* Reserved for an unintegrated feature. */ NUM_BP_EMBEDDED_TYPES = BP_EMBEDDED_TYPE_RESERVED } bp_embedded_type_t; #define BPE_NUM_WORDS 14 #define BPE_PAYLOAD_SIZE (BPE_NUM_WORDS * sizeof (uint64_t)) #define BPE_IS_PAYLOADWORD(bp, wp) \ ((wp) != &(bp)->blk_prop && (wp) != &(bp)->blk_birth) #define SPA_BLKPTRSHIFT 7 /* blkptr_t is 128 bytes */ #define SPA_DVAS_PER_BP 3 /* Number of DVAs in a bp */ /* * A block is a hole when it has either 1) never been written to, or * 2) is zero-filled. In both cases, ZFS can return all zeroes for all reads * without physically allocating disk space. Holes are represented in the * blkptr_t structure by zeroed blk_dva. Correct checking for holes is * done through the BP_IS_HOLE macro. For holes, the logical size, level, * DMU object type, and birth times are all also stored for holes that * were written to at some point (i.e. were punched after having been filled). */ typedef struct blkptr { dva_t blk_dva[SPA_DVAS_PER_BP]; /* Data Virtual Addresses */ uint64_t blk_prop; /* size, compression, type, etc */ uint64_t blk_pad[2]; /* Extra space for the future */ uint64_t blk_phys_birth; /* txg when block was allocated */ uint64_t blk_birth; /* transaction group at birth */ uint64_t blk_fill; /* fill count */ zio_cksum_t blk_cksum; /* 256-bit checksum */ } blkptr_t; /* * Macros to get and set fields in a bp or DVA. */ #define DVA_GET_ASIZE(dva) \ BF64_GET_SB((dva)->dva_word[0], 0, SPA_ASIZEBITS, SPA_MINBLOCKSHIFT, 0) #define DVA_SET_ASIZE(dva, x) \ BF64_SET_SB((dva)->dva_word[0], 0, SPA_ASIZEBITS, \ SPA_MINBLOCKSHIFT, 0, x) #define DVA_GET_GRID(dva) BF64_GET((dva)->dva_word[0], 24, 8) #define DVA_SET_GRID(dva, x) BF64_SET((dva)->dva_word[0], 24, 8, x) #define DVA_GET_VDEV(dva) BF64_GET((dva)->dva_word[0], 32, 32) #define DVA_SET_VDEV(dva, x) BF64_SET((dva)->dva_word[0], 32, 32, x) #define DVA_GET_OFFSET(dva) \ BF64_GET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0) #define DVA_SET_OFFSET(dva, x) \ BF64_SET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0, x) #define DVA_GET_GANG(dva) BF64_GET((dva)->dva_word[1], 63, 1) #define DVA_SET_GANG(dva, x) BF64_SET((dva)->dva_word[1], 63, 1, x) #define BP_GET_LSIZE(bp) \ (BP_IS_EMBEDDED(bp) ? \ (BPE_GET_ETYPE(bp) == BP_EMBEDDED_TYPE_DATA ? BPE_GET_LSIZE(bp) : 0): \ BF64_GET_SB((bp)->blk_prop, 0, SPA_LSIZEBITS, SPA_MINBLOCKSHIFT, 1)) #define BP_SET_LSIZE(bp, x) do { \ ASSERT(!BP_IS_EMBEDDED(bp)); \ BF64_SET_SB((bp)->blk_prop, \ 0, SPA_LSIZEBITS, SPA_MINBLOCKSHIFT, 1, x); \ _NOTE(CONSTCOND) } while (0) #define BP_GET_PSIZE(bp) \ (BP_IS_EMBEDDED(bp) ? 0 : \ BF64_GET_SB((bp)->blk_prop, 16, SPA_PSIZEBITS, SPA_MINBLOCKSHIFT, 1)) #define BP_SET_PSIZE(bp, x) do { \ ASSERT(!BP_IS_EMBEDDED(bp)); \ BF64_SET_SB((bp)->blk_prop, \ 16, SPA_PSIZEBITS, SPA_MINBLOCKSHIFT, 1, x); \ _NOTE(CONSTCOND) } while (0) #define BP_GET_COMPRESS(bp) BF64_GET((bp)->blk_prop, 32, 7) #define BP_SET_COMPRESS(bp, x) BF64_SET((bp)->blk_prop, 32, 7, x) #define BP_IS_EMBEDDED(bp) BF64_GET((bp)->blk_prop, 39, 1) #define BP_SET_EMBEDDED(bp, x) BF64_SET((bp)->blk_prop, 39, 1, x) #define BP_GET_CHECKSUM(bp) \ (BP_IS_EMBEDDED(bp) ? ZIO_CHECKSUM_OFF : \ BF64_GET((bp)->blk_prop, 40, 8)) #define BP_SET_CHECKSUM(bp, x) do { \ ASSERT(!BP_IS_EMBEDDED(bp)); \ BF64_SET((bp)->blk_prop, 40, 8, x); \ _NOTE(CONSTCOND) } while (0) #define BP_GET_TYPE(bp) BF64_GET((bp)->blk_prop, 48, 8) #define BP_SET_TYPE(bp, x) BF64_SET((bp)->blk_prop, 48, 8, x) #define BP_GET_LEVEL(bp) BF64_GET((bp)->blk_prop, 56, 5) #define BP_SET_LEVEL(bp, x) BF64_SET((bp)->blk_prop, 56, 5, x) #define BP_GET_DEDUP(bp) BF64_GET((bp)->blk_prop, 62, 1) #define BP_SET_DEDUP(bp, x) BF64_SET((bp)->blk_prop, 62, 1, x) #define BP_GET_BYTEORDER(bp) BF64_GET((bp)->blk_prop, 63, 1) #define BP_SET_BYTEORDER(bp, x) BF64_SET((bp)->blk_prop, 63, 1, x) #define BP_PHYSICAL_BIRTH(bp) \ (BP_IS_EMBEDDED(bp) ? 0 : \ (bp)->blk_phys_birth ? (bp)->blk_phys_birth : (bp)->blk_birth) #define BP_SET_BIRTH(bp, logical, physical) \ { \ ASSERT(!BP_IS_EMBEDDED(bp)); \ (bp)->blk_birth = (logical); \ (bp)->blk_phys_birth = ((logical) == (physical) ? 0 : (physical)); \ } #define BP_GET_FILL(bp) (BP_IS_EMBEDDED(bp) ? 1 : (bp)->blk_fill) #define BP_GET_ASIZE(bp) \ (BP_IS_EMBEDDED(bp) ? 0 : \ DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \ DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \ DVA_GET_ASIZE(&(bp)->blk_dva[2])) #define BP_GET_UCSIZE(bp) \ ((BP_GET_LEVEL(bp) > 0 || DMU_OT_IS_METADATA(BP_GET_TYPE(bp))) ? \ BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp)) #define BP_GET_NDVAS(bp) \ (BP_IS_EMBEDDED(bp) ? 0 : \ !!DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \ !!DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \ !!DVA_GET_ASIZE(&(bp)->blk_dva[2])) #define BP_COUNT_GANG(bp) \ (BP_IS_EMBEDDED(bp) ? 0 : \ (DVA_GET_GANG(&(bp)->blk_dva[0]) + \ DVA_GET_GANG(&(bp)->blk_dva[1]) + \ DVA_GET_GANG(&(bp)->blk_dva[2]))) #define DVA_EQUAL(dva1, dva2) \ ((dva1)->dva_word[1] == (dva2)->dva_word[1] && \ (dva1)->dva_word[0] == (dva2)->dva_word[0]) #define BP_EQUAL(bp1, bp2) \ (BP_PHYSICAL_BIRTH(bp1) == BP_PHYSICAL_BIRTH(bp2) && \ (bp1)->blk_birth == (bp2)->blk_birth && \ DVA_EQUAL(&(bp1)->blk_dva[0], &(bp2)->blk_dva[0]) && \ DVA_EQUAL(&(bp1)->blk_dva[1], &(bp2)->blk_dva[1]) && \ DVA_EQUAL(&(bp1)->blk_dva[2], &(bp2)->blk_dva[2])) #define ZIO_CHECKSUM_EQUAL(zc1, zc2) \ (0 == (((zc1).zc_word[0] - (zc2).zc_word[0]) | \ ((zc1).zc_word[1] - (zc2).zc_word[1]) | \ ((zc1).zc_word[2] - (zc2).zc_word[2]) | \ ((zc1).zc_word[3] - (zc2).zc_word[3]))) #define DVA_IS_VALID(dva) (DVA_GET_ASIZE(dva) != 0) #define ZIO_SET_CHECKSUM(zcp, w0, w1, w2, w3) \ { \ (zcp)->zc_word[0] = w0; \ (zcp)->zc_word[1] = w1; \ (zcp)->zc_word[2] = w2; \ (zcp)->zc_word[3] = w3; \ } #define BP_IDENTITY(bp) (ASSERT(!BP_IS_EMBEDDED(bp)), &(bp)->blk_dva[0]) #define BP_IS_GANG(bp) \ (BP_IS_EMBEDDED(bp) ? B_FALSE : DVA_GET_GANG(BP_IDENTITY(bp))) #define DVA_IS_EMPTY(dva) ((dva)->dva_word[0] == 0ULL && \ (dva)->dva_word[1] == 0ULL) #define BP_IS_HOLE(bp) \ (!BP_IS_EMBEDDED(bp) && DVA_IS_EMPTY(BP_IDENTITY(bp))) /* BP_IS_RAIDZ(bp) assumes no block compression */ #define BP_IS_RAIDZ(bp) (DVA_GET_ASIZE(&(bp)->blk_dva[0]) > \ BP_GET_PSIZE(bp)) #define BP_ZERO(bp) \ { \ (bp)->blk_dva[0].dva_word[0] = 0; \ (bp)->blk_dva[0].dva_word[1] = 0; \ (bp)->blk_dva[1].dva_word[0] = 0; \ (bp)->blk_dva[1].dva_word[1] = 0; \ (bp)->blk_dva[2].dva_word[0] = 0; \ (bp)->blk_dva[2].dva_word[1] = 0; \ (bp)->blk_prop = 0; \ (bp)->blk_pad[0] = 0; \ (bp)->blk_pad[1] = 0; \ (bp)->blk_phys_birth = 0; \ (bp)->blk_birth = 0; \ (bp)->blk_fill = 0; \ ZIO_SET_CHECKSUM(&(bp)->blk_cksum, 0, 0, 0, 0); \ } #ifdef _BIG_ENDIAN #define ZFS_HOST_BYTEORDER (0ULL) #else #define ZFS_HOST_BYTEORDER (1ULL) #endif #define BP_SHOULD_BYTESWAP(bp) (BP_GET_BYTEORDER(bp) != ZFS_HOST_BYTEORDER) #define BP_SPRINTF_LEN 320 /* * This macro allows code sharing between zfs, libzpool, and mdb. * 'func' is either snprintf() or mdb_snprintf(). * 'ws' (whitespace) can be ' ' for single-line format, '\n' for multi-line. */ #define SNPRINTF_BLKPTR(func, ws, buf, size, bp, type, checksum, compress) \ { \ static const char *copyname[] = \ { "zero", "single", "double", "triple" }; \ int len = 0; \ int copies = 0; \ int d; \ \ if (bp == NULL) { \ len += func(buf + len, size - len, ""); \ } else if (BP_IS_HOLE(bp)) { \ len += func(buf + len, size - len, ""); \ if (bp->blk_birth > 0) { \ len += func(buf + len, size - len, \ " birth=%lluL", \ (u_longlong_t)bp->blk_birth); \ } \ } else if (BP_IS_EMBEDDED(bp)) { \ len = func(buf + len, size - len, \ "EMBEDDED [L%llu %s] et=%u %s " \ "size=%llxL/%llxP birth=%lluL", \ (u_longlong_t)BP_GET_LEVEL(bp), \ type, \ (int)BPE_GET_ETYPE(bp), \ compress, \ (u_longlong_t)BPE_GET_LSIZE(bp), \ (u_longlong_t)BPE_GET_PSIZE(bp), \ (u_longlong_t)bp->blk_birth); \ } else { \ for (d = 0; d < BP_GET_NDVAS(bp); d++) { \ const dva_t *dva = &bp->blk_dva[d]; \ if (DVA_IS_VALID(dva)) \ copies++; \ len += func(buf + len, size - len, \ "DVA[%d]=<%llu:%llx:%llx>%c", d, \ (u_longlong_t)DVA_GET_VDEV(dva), \ (u_longlong_t)DVA_GET_OFFSET(dva), \ (u_longlong_t)DVA_GET_ASIZE(dva), \ ws); \ } \ if (BP_IS_GANG(bp) && \ DVA_GET_ASIZE(&bp->blk_dva[2]) <= \ DVA_GET_ASIZE(&bp->blk_dva[1]) / 2) \ copies--; \ len += func(buf + len, size - len, \ "[L%llu %s] %s %s %s %s %s %s%c" \ "size=%llxL/%llxP birth=%lluL/%lluP fill=%llu%c" \ "cksum=%llx:%llx:%llx:%llx", \ (u_longlong_t)BP_GET_LEVEL(bp), \ type, \ checksum, \ compress, \ BP_GET_BYTEORDER(bp) == 0 ? "BE" : "LE", \ BP_IS_GANG(bp) ? "gang" : "contiguous", \ BP_GET_DEDUP(bp) ? "dedup" : "unique", \ copyname[copies], \ ws, \ (u_longlong_t)BP_GET_LSIZE(bp), \ (u_longlong_t)BP_GET_PSIZE(bp), \ (u_longlong_t)bp->blk_birth, \ (u_longlong_t)BP_PHYSICAL_BIRTH(bp), \ (u_longlong_t)BP_GET_FILL(bp), \ ws, \ (u_longlong_t)bp->blk_cksum.zc_word[0], \ (u_longlong_t)bp->blk_cksum.zc_word[1], \ (u_longlong_t)bp->blk_cksum.zc_word[2], \ (u_longlong_t)bp->blk_cksum.zc_word[3]); \ } \ ASSERT(len < size); \ } #include #define BP_GET_BUFC_TYPE(bp) \ (((BP_GET_LEVEL(bp) > 0) || (DMU_OT_IS_METADATA(BP_GET_TYPE(bp)))) ? \ ARC_BUFC_METADATA : ARC_BUFC_DATA) typedef enum spa_import_type { SPA_IMPORT_EXISTING, SPA_IMPORT_ASSEMBLE } spa_import_type_t; /* state manipulation functions */ extern int spa_open(const char *pool, spa_t **, void *tag); extern int spa_open_rewind(const char *pool, spa_t **, void *tag, nvlist_t *policy, nvlist_t **config); extern int spa_get_stats(const char *pool, nvlist_t **config, char *altroot, size_t buflen); extern int spa_create(const char *pool, nvlist_t *config, nvlist_t *props, nvlist_t *zplprops); extern int spa_import_rootpool(char *devpath, char *devid); extern int spa_import(char *pool, nvlist_t *config, nvlist_t *props, uint64_t flags); extern nvlist_t *spa_tryimport(nvlist_t *tryconfig); extern int spa_destroy(char *pool); extern int spa_export(char *pool, nvlist_t **oldconfig, boolean_t force, boolean_t hardforce); extern int spa_reset(char *pool); extern void spa_async_request(spa_t *spa, int flag); extern void spa_async_unrequest(spa_t *spa, int flag); extern void spa_async_suspend(spa_t *spa); extern void spa_async_resume(spa_t *spa); extern spa_t *spa_inject_addref(char *pool); extern void spa_inject_delref(spa_t *spa); extern void spa_scan_stat_init(spa_t *spa); extern int spa_scan_get_stats(spa_t *spa, pool_scan_stat_t *ps); #define SPA_ASYNC_CONFIG_UPDATE 0x01 #define SPA_ASYNC_REMOVE 0x02 #define SPA_ASYNC_PROBE 0x04 #define SPA_ASYNC_RESILVER_DONE 0x08 #define SPA_ASYNC_RESILVER 0x10 #define SPA_ASYNC_AUTOEXPAND 0x20 #define SPA_ASYNC_REMOVE_DONE 0x40 #define SPA_ASYNC_REMOVE_STOP 0x80 /* * Controls the behavior of spa_vdev_remove(). */ #define SPA_REMOVE_UNSPARE 0x01 #define SPA_REMOVE_DONE 0x02 /* device manipulation */ extern int spa_vdev_add(spa_t *spa, nvlist_t *nvroot); extern int spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, int replacing); extern int spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done); extern int spa_vdev_remove(spa_t *spa, uint64_t guid, boolean_t unspare); extern boolean_t spa_vdev_remove_active(spa_t *spa); extern int spa_vdev_setpath(spa_t *spa, uint64_t guid, const char *newpath); extern int spa_vdev_setfru(spa_t *spa, uint64_t guid, const char *newfru); extern int spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config, nvlist_t *props, boolean_t exp); /* spare state (which is global across all pools) */ extern void spa_spare_add(vdev_t *vd); extern void spa_spare_remove(vdev_t *vd); extern boolean_t spa_spare_exists(uint64_t guid, uint64_t *pool, int *refcnt); extern void spa_spare_activate(vdev_t *vd); /* L2ARC state (which is global across all pools) */ extern void spa_l2cache_add(vdev_t *vd); extern void spa_l2cache_remove(vdev_t *vd); extern boolean_t spa_l2cache_exists(uint64_t guid, uint64_t *pool); extern void spa_l2cache_activate(vdev_t *vd); extern void spa_l2cache_drop(spa_t *spa); /* scanning */ extern int spa_scan(spa_t *spa, pool_scan_func_t func); extern int spa_scan_stop(spa_t *spa); /* spa syncing */ extern void spa_sync(spa_t *spa, uint64_t txg); /* only for DMU use */ extern void spa_sync_allpools(void); extern int zfs_sync_pass_deferred_free; /* spa namespace global mutex */ extern kmutex_t spa_namespace_lock; /* * SPA configuration functions in spa_config.c */ #define SPA_CONFIG_UPDATE_POOL 0 #define SPA_CONFIG_UPDATE_VDEVS 1 extern void spa_config_sync(spa_t *, boolean_t, boolean_t); extern void spa_config_load(void); extern nvlist_t *spa_all_configs(uint64_t *); extern void spa_config_set(spa_t *spa, nvlist_t *config); extern nvlist_t *spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg, int getstats); extern void spa_config_update(spa_t *spa, int what); /* * Miscellaneous SPA routines in spa_misc.c */ /* Namespace manipulation */ extern spa_t *spa_lookup(const char *name); extern spa_t *spa_add(const char *name, nvlist_t *config, const char *altroot); extern void spa_remove(spa_t *spa); extern spa_t *spa_next(spa_t *prev); /* Refcount functions */ extern void spa_open_ref(spa_t *spa, void *tag); extern void spa_close(spa_t *spa, void *tag); extern boolean_t spa_refcount_zero(spa_t *spa); #define SCL_NONE 0x00 #define SCL_CONFIG 0x01 #define SCL_STATE 0x02 #define SCL_L2ARC 0x04 /* hack until L2ARC 2.0 */ #define SCL_ALLOC 0x08 #define SCL_ZIO 0x10 #define SCL_FREE 0x20 #define SCL_VDEV 0x40 #define SCL_LOCKS 7 #define SCL_ALL ((1 << SCL_LOCKS) - 1) #define SCL_STATE_ALL (SCL_STATE | SCL_L2ARC | SCL_ZIO) /* Historical pool statistics */ typedef struct spa_stats_history { kmutex_t lock; uint64_t count; uint64_t size; kstat_t *kstat; void *private; list_t list; } spa_stats_history_t; typedef struct spa_stats { spa_stats_history_t read_history; spa_stats_history_t txg_history; spa_stats_history_t tx_assign_histogram; spa_stats_history_t io_history; } spa_stats_t; typedef enum txg_state { TXG_STATE_BIRTH = 0, TXG_STATE_OPEN = 1, TXG_STATE_QUIESCED = 2, TXG_STATE_WAIT_FOR_SYNC = 3, TXG_STATE_SYNCED = 4, TXG_STATE_COMMITTED = 5, } txg_state_t; extern void spa_stats_init(spa_t *spa); extern void spa_stats_destroy(spa_t *spa); extern void spa_read_history_add(spa_t *spa, const zbookmark_phys_t *zb, uint32_t aflags); extern void spa_txg_history_add(spa_t *spa, uint64_t txg, hrtime_t birth_time); extern int spa_txg_history_set(spa_t *spa, uint64_t txg, txg_state_t completed_state, hrtime_t completed_time); extern int spa_txg_history_set_io(spa_t *spa, uint64_t txg, uint64_t nread, uint64_t nwritten, uint64_t reads, uint64_t writes, uint64_t ndirty); extern void spa_tx_assign_add_nsecs(spa_t *spa, uint64_t nsecs); /* Pool configuration locks */ extern int spa_config_tryenter(spa_t *spa, int locks, void *tag, krw_t rw); extern void spa_config_enter(spa_t *spa, int locks, void *tag, krw_t rw); extern void spa_config_exit(spa_t *spa, int locks, void *tag); extern int spa_config_held(spa_t *spa, int locks, krw_t rw); /* Pool vdev add/remove lock */ extern uint64_t spa_vdev_enter(spa_t *spa); extern uint64_t spa_vdev_config_enter(spa_t *spa); extern void spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error, char *tag); extern int spa_vdev_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error); /* Pool vdev state change lock */ extern void spa_vdev_state_enter(spa_t *spa, int oplock); extern int spa_vdev_state_exit(spa_t *spa, vdev_t *vd, int error); /* Log state */ typedef enum spa_log_state { SPA_LOG_UNKNOWN = 0, /* unknown log state */ SPA_LOG_MISSING, /* missing log(s) */ SPA_LOG_CLEAR, /* clear the log(s) */ SPA_LOG_GOOD, /* log(s) are good */ } spa_log_state_t; extern spa_log_state_t spa_get_log_state(spa_t *spa); extern void spa_set_log_state(spa_t *spa, spa_log_state_t state); extern int spa_offline_log(spa_t *spa); /* Log claim callback */ extern void spa_claim_notify(zio_t *zio); extern void spa_deadman(void *); /* Accessor functions */ extern boolean_t spa_shutting_down(spa_t *spa); extern struct dsl_pool *spa_get_dsl(spa_t *spa); extern boolean_t spa_is_initializing(spa_t *spa); extern blkptr_t *spa_get_rootblkptr(spa_t *spa); extern void spa_set_rootblkptr(spa_t *spa, const blkptr_t *bp); extern void spa_altroot(spa_t *, char *, size_t); extern int spa_sync_pass(spa_t *spa); extern char *spa_name(spa_t *spa); extern uint64_t spa_guid(spa_t *spa); extern uint64_t spa_load_guid(spa_t *spa); extern uint64_t spa_last_synced_txg(spa_t *spa); extern uint64_t spa_first_txg(spa_t *spa); extern uint64_t spa_syncing_txg(spa_t *spa); extern uint64_t spa_version(spa_t *spa); extern pool_state_t spa_state(spa_t *spa); extern spa_load_state_t spa_load_state(spa_t *spa); extern uint64_t spa_freeze_txg(spa_t *spa); extern uint64_t spa_get_asize(spa_t *spa, uint64_t lsize); extern uint64_t spa_get_dspace(spa_t *spa); extern void spa_update_dspace(spa_t *spa); extern uint64_t spa_version(spa_t *spa); extern boolean_t spa_deflate(spa_t *spa); extern metaslab_class_t *spa_normal_class(spa_t *spa); extern metaslab_class_t *spa_log_class(spa_t *spa); extern int spa_max_replication(spa_t *spa); extern int spa_prev_software_version(spa_t *spa); extern int spa_busy(void); extern uint8_t spa_get_failmode(spa_t *spa); extern boolean_t spa_suspended(spa_t *spa); extern uint64_t spa_bootfs(spa_t *spa); extern uint64_t spa_delegation(spa_t *spa); extern objset_t *spa_meta_objset(spa_t *spa); extern uint64_t spa_deadman_synctime(spa_t *spa); /* Miscellaneous support routines */ extern void spa_activate_mos_feature(spa_t *spa, const char *feature, dmu_tx_t *tx); extern void spa_deactivate_mos_feature(spa_t *spa, const char *feature); extern int spa_rename(const char *oldname, const char *newname); extern spa_t *spa_by_guid(uint64_t pool_guid, uint64_t device_guid); extern boolean_t spa_guid_exists(uint64_t pool_guid, uint64_t device_guid); extern char *spa_strdup(const char *); extern void spa_strfree(char *); extern uint64_t spa_get_random(uint64_t range); extern uint64_t spa_generate_guid(spa_t *spa); extern void snprintf_blkptr(char *buf, size_t buflen, const blkptr_t *bp); extern void spa_freeze(spa_t *spa); extern int spa_change_guid(spa_t *spa); extern void spa_upgrade(spa_t *spa, uint64_t version); extern void spa_evict_all(void); extern vdev_t *spa_lookup_by_guid(spa_t *spa, uint64_t guid, boolean_t l2cache); extern boolean_t spa_has_spare(spa_t *, uint64_t guid); extern uint64_t dva_get_dsize_sync(spa_t *spa, const dva_t *dva); extern uint64_t bp_get_dsize_sync(spa_t *spa, const blkptr_t *bp); extern uint64_t bp_get_dsize(spa_t *spa, const blkptr_t *bp); extern boolean_t spa_has_slogs(spa_t *spa); extern boolean_t spa_is_root(spa_t *spa); extern boolean_t spa_writeable(spa_t *spa); extern boolean_t spa_has_pending_synctask(spa_t *spa); extern int spa_mode(spa_t *spa); extern uint64_t strtonum(const char *str, char **nptr); extern char *spa_his_ievent_table[]; extern void spa_history_create_obj(spa_t *spa, dmu_tx_t *tx); extern int spa_history_get(spa_t *spa, uint64_t *offset, uint64_t *len_read, char *his_buf); extern int spa_history_log(spa_t *spa, const char *his_buf); extern int spa_history_log_nvl(spa_t *spa, nvlist_t *nvl); extern void spa_history_log_version(spa_t *spa, const char *operation); extern void spa_history_log_internal(spa_t *spa, const char *operation, dmu_tx_t *tx, const char *fmt, ...); extern void spa_history_log_internal_ds(struct dsl_dataset *ds, const char *op, dmu_tx_t *tx, const char *fmt, ...); extern void spa_history_log_internal_dd(dsl_dir_t *dd, const char *operation, dmu_tx_t *tx, const char *fmt, ...); /* error handling */ struct zbookmark_phys; extern void spa_log_error(spa_t *spa, zio_t *zio); extern void zfs_ereport_post(const char *class, spa_t *spa, vdev_t *vd, zio_t *zio, uint64_t stateoroffset, uint64_t length); extern void zfs_post_remove(spa_t *spa, vdev_t *vd); extern void zfs_post_state_change(spa_t *spa, vdev_t *vd); extern void zfs_post_autoreplace(spa_t *spa, vdev_t *vd); extern uint64_t spa_get_errlog_size(spa_t *spa); extern int spa_get_errlog(spa_t *spa, void *uaddr, size_t *count); extern void spa_errlog_rotate(spa_t *spa); extern void spa_errlog_drain(spa_t *spa); extern void spa_errlog_sync(spa_t *spa, uint64_t txg); extern void spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub); /* vdev cache */ extern void vdev_cache_stat_init(void); extern void vdev_cache_stat_fini(void); /* Initialization and termination */ extern void spa_init(int flags); extern void spa_fini(void); extern void spa_boot_init(void); /* properties */ extern int spa_prop_set(spa_t *spa, nvlist_t *nvp); extern int spa_prop_get(spa_t *spa, nvlist_t **nvp); extern void spa_prop_clear_bootfs(spa_t *spa, uint64_t obj, dmu_tx_t *tx); extern void spa_configfile_set(spa_t *, nvlist_t *, boolean_t); /* asynchronous event notification */ extern void spa_event_notify(spa_t *spa, vdev_t *vdev, const char *name); #ifdef ZFS_DEBUG #define dprintf_bp(bp, fmt, ...) do { \ if (zfs_flags & ZFS_DEBUG_DPRINTF) { \ char *__blkbuf = kmem_alloc(BP_SPRINTF_LEN, KM_PUSHPAGE); \ snprintf_blkptr(__blkbuf, BP_SPRINTF_LEN, (bp)); \ dprintf(fmt " %s\n", __VA_ARGS__, __blkbuf); \ kmem_free(__blkbuf, BP_SPRINTF_LEN); \ } \ _NOTE(CONSTCOND) } while (0) #else #define dprintf_bp(bp, fmt, ...) #endif extern boolean_t spa_debug_enabled(spa_t *spa); #define spa_dbgmsg(spa, ...) \ { \ if (spa_debug_enabled(spa)) \ zfs_dbgmsg(__VA_ARGS__); \ } extern int spa_mode_global; /* mode, e.g. FREAD | FWRITE */ #ifdef __cplusplus } #endif #endif /* _SYS_SPA_H */ diff --git a/include/sys/zfs_context.h b/include/sys/zfs_context.h index 1487a99f4a64..dd02e6f1bcc4 100644 --- a/include/sys/zfs_context.h +++ b/include/sys/zfs_context.h @@ -1,743 +1,743 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright 2009 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ /* * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2012, Joyent, Inc. All rights reserved. * Copyright (c) 2012, 2014 by Delphix. All rights reserved. */ #ifndef _SYS_ZFS_CONTEXT_H #define _SYS_ZFS_CONTEXT_H #ifdef __KERNEL__ #include #include #include #include #include #include #include #include +#include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #else /* _KERNEL */ #define _SYS_MUTEX_H #define _SYS_RWLOCK_H #define _SYS_CONDVAR_H #define _SYS_SYSTM_H #define _SYS_T_LOCK_H #define _SYS_VNODE_H #define _SYS_VFS_H #define _SYS_SUNDDI_H #define _SYS_CALLB_H #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * Stack */ #define noinline __attribute__((noinline)) /* * Debugging */ /* * Note that we are not using the debugging levels. */ #define CE_CONT 0 /* continuation */ #define CE_NOTE 1 /* notice */ #define CE_WARN 2 /* warning */ #define CE_PANIC 3 /* panic */ #define CE_IGNORE 4 /* print nothing */ /* * ZFS debugging */ extern void dprintf_setup(int *argc, char **argv); extern void cmn_err(int, const char *, ...); extern void vcmn_err(int, const char *, va_list); extern void panic(const char *, ...); extern void vpanic(const char *, va_list); #define fm_panic panic extern int aok; /* * DTrace SDT probes have different signatures in userland than they do in * kernel. If they're being used in kernel code, re-define them out of * existence for their counterparts in libzpool. */ #ifdef DTRACE_PROBE #undef DTRACE_PROBE #endif /* DTRACE_PROBE */ #define DTRACE_PROBE(a) \ ZFS_PROBE0(#a) #ifdef DTRACE_PROBE1 #undef DTRACE_PROBE1 #endif /* DTRACE_PROBE1 */ #define DTRACE_PROBE1(a, b, c) \ ZFS_PROBE1(#a, (unsigned long)c) #ifdef DTRACE_PROBE2 #undef DTRACE_PROBE2 #endif /* DTRACE_PROBE2 */ #define DTRACE_PROBE2(a, b, c, d, e) \ ZFS_PROBE2(#a, (unsigned long)c, (unsigned long)e) #ifdef DTRACE_PROBE3 #undef DTRACE_PROBE3 #endif /* DTRACE_PROBE3 */ #define DTRACE_PROBE3(a, b, c, d, e, f, g) \ ZFS_PROBE3(#a, (unsigned long)c, (unsigned long)e, (unsigned long)g) #ifdef DTRACE_PROBE4 #undef DTRACE_PROBE4 #endif /* DTRACE_PROBE4 */ #define DTRACE_PROBE4(a, b, c, d, e, f, g, h, i) \ ZFS_PROBE4(#a, (unsigned long)c, (unsigned long)e, (unsigned long)g, \ (unsigned long)i) /* * We use the comma operator so that this macro can be used without much * additional code. For example, "return (EINVAL);" becomes * "return (SET_ERROR(EINVAL));". Note that the argument will be evaluated * twice, so it should not have side effects (e.g. something like: * "return (SET_ERROR(log_error(EINVAL, info)));" would log the error twice). */ #define SET_ERROR(err) (ZFS_SET_ERROR(err), err) /* * Threads. TS_STACK_MIN is dictated by the minimum allowed pthread stack * size. While TS_STACK_MAX is somewhat arbitrary, it was selected to be * large enough for the expected stack depth while small enough to avoid * exhausting address space with high thread counts. */ #define TS_MAGIC 0x72f158ab4261e538ull #define TS_RUN 0x00000002 #define TS_STACK_MIN PTHREAD_STACK_MIN #define TS_STACK_MAX (256 * 1024) /* in libzpool, p0 exists only to have its address taken */ typedef struct proc { uintptr_t this_is_never_used_dont_dereference_it; } proc_t; extern struct proc p0; #define curproc (&p0) typedef void (*thread_func_t)(void *); typedef void (*thread_func_arg_t)(void *); typedef pthread_t kt_did_t; #define kpreempt(x) ((void)0) typedef struct kthread { kt_did_t t_tid; thread_func_t t_func; void * t_arg; } kthread_t; #define curthread zk_thread_current() #define getcomm() "unknown" #define thread_exit zk_thread_exit #define thread_create(stk, stksize, func, arg, len, pp, state, pri) \ zk_thread_create(stk, stksize, (thread_func_t)func, arg, \ len, NULL, state, pri, PTHREAD_CREATE_DETACHED) #define thread_join(t) zk_thread_join(t) #define newproc(f, a, cid, pri, ctp, pid) (ENOSYS) extern kthread_t *zk_thread_current(void); extern void zk_thread_exit(void); extern kthread_t *zk_thread_create(caddr_t stk, size_t stksize, thread_func_t func, void *arg, size_t len, proc_t *pp, int state, pri_t pri, int detachstate); extern void zk_thread_join(kt_did_t tid); #define kpreempt_disable() ((void)0) #define kpreempt_enable() ((void)0) #define PS_NONE -1 #define issig(why) (FALSE) #define ISSIG(thr, why) (FALSE) /* * Mutexes */ #define MTX_MAGIC 0x9522f51362a6e326ull #define MTX_INIT ((void *)NULL) #define MTX_DEST ((void *)-1UL) typedef struct kmutex { void *m_owner; uint64_t m_magic; pthread_mutex_t m_lock; } kmutex_t; #define MUTEX_DEFAULT 0 #define MUTEX_HELD(m) ((m)->m_owner == curthread) #define MUTEX_NOT_HELD(m) (!MUTEX_HELD(m)) extern void mutex_init(kmutex_t *mp, char *name, int type, void *cookie); extern void mutex_destroy(kmutex_t *mp); extern void mutex_enter(kmutex_t *mp); extern void mutex_exit(kmutex_t *mp); extern int mutex_tryenter(kmutex_t *mp); extern void *mutex_owner(kmutex_t *mp); extern int mutex_held(kmutex_t *mp); /* * RW locks */ #define RW_MAGIC 0x4d31fb123648e78aull #define RW_INIT ((void *)NULL) #define RW_DEST ((void *)-1UL) typedef struct krwlock { void *rw_owner; void *rw_wr_owner; uint64_t rw_magic; pthread_rwlock_t rw_lock; uint_t rw_readers; } krwlock_t; typedef int krw_t; #define RW_READER 0 #define RW_WRITER 1 #define RW_DEFAULT RW_READER #define RW_READ_HELD(x) ((x)->rw_readers > 0) #define RW_WRITE_HELD(x) ((x)->rw_wr_owner == curthread) #define RW_LOCK_HELD(x) (RW_READ_HELD(x) || RW_WRITE_HELD(x)) #undef RW_LOCK_HELD #define RW_LOCK_HELD(x) (RW_READ_HELD(x) || RW_WRITE_HELD(x)) #undef RW_LOCK_HELD #define RW_LOCK_HELD(x) (RW_READ_HELD(x) || RW_WRITE_HELD(x)) extern void rw_init(krwlock_t *rwlp, char *name, int type, void *arg); extern void rw_destroy(krwlock_t *rwlp); extern void rw_enter(krwlock_t *rwlp, krw_t rw); extern int rw_tryenter(krwlock_t *rwlp, krw_t rw); extern int rw_tryupgrade(krwlock_t *rwlp); extern void rw_exit(krwlock_t *rwlp); #define rw_downgrade(rwlp) do { } while (0) extern uid_t crgetuid(cred_t *cr); extern uid_t crgetruid(cred_t *cr); extern gid_t crgetgid(cred_t *cr); extern int crgetngroups(cred_t *cr); extern gid_t *crgetgroups(cred_t *cr); /* * Condition variables */ #define CV_MAGIC 0xd31ea9a83b1b30c4ull typedef struct kcondvar { uint64_t cv_magic; pthread_cond_t cv; } kcondvar_t; #define CV_DEFAULT 0 extern void cv_init(kcondvar_t *cv, char *name, int type, void *arg); extern void cv_destroy(kcondvar_t *cv); extern void cv_wait(kcondvar_t *cv, kmutex_t *mp); extern clock_t cv_timedwait(kcondvar_t *cv, kmutex_t *mp, clock_t abstime); extern clock_t cv_timedwait_hires(kcondvar_t *cvp, kmutex_t *mp, hrtime_t tim, hrtime_t res, int flag); extern void cv_signal(kcondvar_t *cv); extern void cv_broadcast(kcondvar_t *cv); #define cv_timedwait_interruptible(cv, mp, at) cv_timedwait(cv, mp, at) #define cv_wait_interruptible(cv, mp) cv_wait(cv, mp) #define cv_wait_io(cv, mp) cv_wait(cv, mp) /* * Thread-specific data */ #define tsd_get(k) pthread_getspecific(k) #define tsd_set(k, v) pthread_setspecific(k, v) #define tsd_create(kp, d) pthread_key_create(kp, d) #define tsd_destroy(kp) /* nothing */ /* * Thread-specific data */ #define tsd_get(k) pthread_getspecific(k) #define tsd_set(k, v) pthread_setspecific(k, v) #define tsd_create(kp, d) pthread_key_create(kp, d) #define tsd_destroy(kp) /* nothing */ /* * kstat creation, installation and deletion */ extern kstat_t *kstat_create(const char *, int, const char *, const char *, uchar_t, ulong_t, uchar_t); extern void kstat_install(kstat_t *); extern void kstat_delete(kstat_t *); extern void kstat_waitq_enter(kstat_io_t *); extern void kstat_waitq_exit(kstat_io_t *); extern void kstat_runq_enter(kstat_io_t *); extern void kstat_runq_exit(kstat_io_t *); extern void kstat_waitq_to_runq(kstat_io_t *); extern void kstat_runq_back_to_waitq(kstat_io_t *); extern void kstat_set_raw_ops(kstat_t *ksp, int (*headers)(char *buf, size_t size), int (*data)(char *buf, size_t size, void *data), void *(*addr)(kstat_t *ksp, loff_t index)); /* * Kernel memory */ #define KM_SLEEP UMEM_NOFAIL #define KM_PUSHPAGE KM_SLEEP #define KM_NOSLEEP UMEM_DEFAULT -#define KM_NODEBUG 0x0 #define KMC_NODEBUG UMC_NODEBUG #define KMC_KMEM 0x0 #define KMC_VMEM 0x0 #define kmem_alloc(_s, _f) umem_alloc(_s, _f) #define kmem_zalloc(_s, _f) umem_zalloc(_s, _f) #define kmem_free(_b, _s) umem_free(_b, _s) #define vmem_alloc(_s, _f) kmem_alloc(_s, _f) #define vmem_zalloc(_s, _f) kmem_zalloc(_s, _f) #define vmem_free(_b, _s) kmem_free(_b, _s) #define kmem_cache_create(_a, _b, _c, _d, _e, _f, _g, _h, _i) \ umem_cache_create(_a, _b, _c, _d, _e, _f, _g, _h, _i) #define kmem_cache_destroy(_c) umem_cache_destroy(_c) #define kmem_cache_alloc(_c, _f) umem_cache_alloc(_c, _f) #define kmem_cache_free(_c, _b) umem_cache_free(_c, _b) #define kmem_debugging() 0 #define kmem_cache_reap_now(_c) /* nothing */ #define kmem_cache_set_move(_c, _cb) /* nothing */ #define POINTER_INVALIDATE(_pp) /* nothing */ #define POINTER_IS_VALID(_p) 0 typedef umem_cache_t kmem_cache_t; typedef enum kmem_cbrc { KMEM_CBRC_YES, KMEM_CBRC_NO, KMEM_CBRC_LATER, KMEM_CBRC_DONT_NEED, KMEM_CBRC_DONT_KNOW } kmem_cbrc_t; /* * Task queues */ typedef struct taskq taskq_t; typedef uintptr_t taskqid_t; typedef void (task_func_t)(void *); typedef struct taskq_ent { struct taskq_ent *tqent_next; struct taskq_ent *tqent_prev; task_func_t *tqent_func; void *tqent_arg; uintptr_t tqent_flags; } taskq_ent_t; #define TQENT_FLAG_PREALLOC 0x1 /* taskq_dispatch_ent used */ #define TASKQ_PREPOPULATE 0x0001 #define TASKQ_CPR_SAFE 0x0002 /* Use CPR safe protocol */ #define TASKQ_DYNAMIC 0x0004 /* Use dynamic thread scheduling */ #define TASKQ_THREADS_CPU_PCT 0x0008 /* Scale # threads by # cpus */ #define TASKQ_DC_BATCH 0x0010 /* Mark threads as batch */ #define TQ_SLEEP KM_SLEEP /* Can block for memory */ #define TQ_NOSLEEP KM_NOSLEEP /* cannot block for memory; may fail */ #define TQ_PUSHPAGE KM_PUSHPAGE /* Cannot perform I/O */ #define TQ_NOQUEUE 0x02 /* Do not enqueue if can't dispatch */ #define TQ_FRONT 0x08 /* Queue in front */ extern taskq_t *system_taskq; extern taskq_t *taskq_create(const char *, int, pri_t, int, int, uint_t); #define taskq_create_proc(a, b, c, d, e, p, f) \ (taskq_create(a, b, c, d, e, f)) #define taskq_create_sysdc(a, b, d, e, p, dc, f) \ (taskq_create(a, b, maxclsyspri, d, e, f)) extern taskqid_t taskq_dispatch(taskq_t *, task_func_t, void *, uint_t); extern taskqid_t taskq_dispatch_delay(taskq_t *, task_func_t, void *, uint_t, clock_t); extern void taskq_dispatch_ent(taskq_t *, task_func_t, void *, uint_t, taskq_ent_t *); extern int taskq_empty_ent(taskq_ent_t *); extern void taskq_init_ent(taskq_ent_t *); extern void taskq_destroy(taskq_t *); extern void taskq_wait(taskq_t *); extern void taskq_wait_id(taskq_t *, taskqid_t); extern int taskq_member(taskq_t *, kthread_t *); extern int taskq_cancel_id(taskq_t *, taskqid_t); extern void system_taskq_init(void); extern void system_taskq_fini(void); #define XVA_MAPSIZE 3 #define XVA_MAGIC 0x78766174 /* * vnodes */ typedef struct vnode { uint64_t v_size; int v_fd; char *v_path; } vnode_t; #define AV_SCANSTAMP_SZ 32 /* length of anti-virus scanstamp */ typedef struct xoptattr { timestruc_t xoa_createtime; /* Create time of file */ uint8_t xoa_archive; uint8_t xoa_system; uint8_t xoa_readonly; uint8_t xoa_hidden; uint8_t xoa_nounlink; uint8_t xoa_immutable; uint8_t xoa_appendonly; uint8_t xoa_nodump; uint8_t xoa_settable; uint8_t xoa_opaque; uint8_t xoa_av_quarantined; uint8_t xoa_av_modified; uint8_t xoa_av_scanstamp[AV_SCANSTAMP_SZ]; uint8_t xoa_reparse; uint8_t xoa_offline; uint8_t xoa_sparse; } xoptattr_t; typedef struct vattr { uint_t va_mask; /* bit-mask of attributes */ u_offset_t va_size; /* file size in bytes */ } vattr_t; typedef struct xvattr { vattr_t xva_vattr; /* Embedded vattr structure */ uint32_t xva_magic; /* Magic Number */ uint32_t xva_mapsize; /* Size of attr bitmap (32-bit words) */ uint32_t *xva_rtnattrmapp; /* Ptr to xva_rtnattrmap[] */ uint32_t xva_reqattrmap[XVA_MAPSIZE]; /* Requested attrs */ uint32_t xva_rtnattrmap[XVA_MAPSIZE]; /* Returned attrs */ xoptattr_t xva_xoptattrs; /* Optional attributes */ } xvattr_t; typedef struct vsecattr { uint_t vsa_mask; /* See below */ int vsa_aclcnt; /* ACL entry count */ void *vsa_aclentp; /* pointer to ACL entries */ int vsa_dfaclcnt; /* default ACL entry count */ void *vsa_dfaclentp; /* pointer to default ACL entries */ size_t vsa_aclentsz; /* ACE size in bytes of vsa_aclentp */ } vsecattr_t; #define AT_TYPE 0x00001 #define AT_MODE 0x00002 #define AT_UID 0x00004 #define AT_GID 0x00008 #define AT_FSID 0x00010 #define AT_NODEID 0x00020 #define AT_NLINK 0x00040 #define AT_SIZE 0x00080 #define AT_ATIME 0x00100 #define AT_MTIME 0x00200 #define AT_CTIME 0x00400 #define AT_RDEV 0x00800 #define AT_BLKSIZE 0x01000 #define AT_NBLOCKS 0x02000 #define AT_SEQ 0x08000 #define AT_XVATTR 0x10000 #define CRCREAT 0 extern int fop_getattr(vnode_t *vp, vattr_t *vap); #define VOP_CLOSE(vp, f, c, o, cr, ct) vn_close(vp) #define VOP_PUTPAGE(vp, of, sz, fl, cr, ct) 0 #define VOP_GETATTR(vp, vap, fl, cr, ct) fop_getattr((vp), (vap)); #define VOP_FSYNC(vp, f, cr, ct) fsync((vp)->v_fd) #define VN_RELE(vp) vn_close(vp) extern int vn_open(char *path, int x1, int oflags, int mode, vnode_t **vpp, int x2, int x3); extern int vn_openat(char *path, int x1, int oflags, int mode, vnode_t **vpp, int x2, int x3, vnode_t *vp, int fd); extern int vn_rdwr(int uio, vnode_t *vp, void *addr, ssize_t len, offset_t offset, int x1, int x2, rlim64_t x3, void *x4, ssize_t *residp); extern void vn_close(vnode_t *vp); #define vn_remove(path, x1, x2) remove(path) #define vn_rename(from, to, seg) rename((from), (to)) #define vn_is_readonly(vp) B_FALSE extern vnode_t *rootdir; #include /* for FREAD, FWRITE, etc */ /* * Random stuff */ #define ddi_get_lbolt() (gethrtime() >> 23) #define ddi_get_lbolt64() (gethrtime() >> 23) #define hz 119 /* frequency when using gethrtime() >> 23 for lbolt */ #define ddi_time_before(a, b) (a < b) #define ddi_time_after(a, b) ddi_time_before(b, a) #define ddi_time_before_eq(a, b) (!ddi_time_after(a, b)) #define ddi_time_after_eq(a, b) ddi_time_before_eq(b, a) #define ddi_time_before64(a, b) (a < b) #define ddi_time_after64(a, b) ddi_time_before64(b, a) #define ddi_time_before_eq64(a, b) (!ddi_time_after64(a, b)) #define ddi_time_after_eq64(a, b) ddi_time_before_eq64(b, a) extern void delay(clock_t ticks); #define SEC_TO_TICK(sec) ((sec) * hz) #define MSEC_TO_TICK(msec) ((msec) / (MILLISEC / hz)) #define USEC_TO_TICK(usec) ((usec) / (MICROSEC / hz)) #define NSEC_TO_TICK(usec) ((usec) / (NANOSEC / hz)) #define gethrestime_sec() time(NULL) #define gethrestime(t) \ do {\ (t)->tv_sec = gethrestime_sec();\ (t)->tv_nsec = 0;\ } while (0); #define max_ncpus 64 #define minclsyspri 60 #define maxclsyspri 99 #define CPU_SEQID (pthread_self() & (max_ncpus - 1)) #define kcred NULL #define CRED() NULL #define ptob(x) ((x) * PAGESIZE) extern uint64_t physmem; extern int highbit64(uint64_t i); extern int random_get_bytes(uint8_t *ptr, size_t len); extern int random_get_pseudo_bytes(uint8_t *ptr, size_t len); extern void kernel_init(int); extern void kernel_fini(void); struct spa; extern void nicenum(uint64_t num, char *buf); extern void show_pool_stats(struct spa *); typedef struct callb_cpr { kmutex_t *cc_lockp; } callb_cpr_t; #define CALLB_CPR_INIT(cp, lockp, func, name) { \ (cp)->cc_lockp = lockp; \ } #define CALLB_CPR_SAFE_BEGIN(cp) { \ ASSERT(MUTEX_HELD((cp)->cc_lockp)); \ } #define CALLB_CPR_SAFE_END(cp, lockp) { \ ASSERT(MUTEX_HELD((cp)->cc_lockp)); \ } #define CALLB_CPR_EXIT(cp) { \ ASSERT(MUTEX_HELD((cp)->cc_lockp)); \ mutex_exit((cp)->cc_lockp); \ } #define zone_dataset_visible(x, y) (1) #define INGLOBALZONE(z) (1) extern char *kmem_vasprintf(const char *fmt, va_list adx); extern char *kmem_asprintf(const char *fmt, ...); #define strfree(str) kmem_free((str), strlen(str) + 1) /* * Hostname information */ extern char hw_serial[]; /* for userland-emulated hostid access */ extern int ddi_strtoul(const char *str, char **nptr, int base, unsigned long *result); extern int ddi_strtoull(const char *str, char **nptr, int base, u_longlong_t *result); typedef struct utsname utsname_t; extern utsname_t *utsname(void); /* ZFS Boot Related stuff. */ struct _buf { intptr_t _fd; }; struct bootstat { uint64_t st_size; }; typedef struct ace_object { uid_t a_who; uint32_t a_access_mask; uint16_t a_flags; uint16_t a_type; uint8_t a_obj_type[16]; uint8_t a_inherit_obj_type[16]; } ace_object_t; #define ACE_ACCESS_ALLOWED_OBJECT_ACE_TYPE 0x05 #define ACE_ACCESS_DENIED_OBJECT_ACE_TYPE 0x06 #define ACE_SYSTEM_AUDIT_OBJECT_ACE_TYPE 0x07 #define ACE_SYSTEM_ALARM_OBJECT_ACE_TYPE 0x08 extern struct _buf *kobj_open_file(char *name); extern int kobj_read_file(struct _buf *file, char *buf, unsigned size, unsigned off); extern void kobj_close_file(struct _buf *file); extern int kobj_get_filesize(struct _buf *file, uint64_t *size); extern int zfs_secpolicy_snapshot_perms(const char *name, cred_t *cr); extern int zfs_secpolicy_rename_perms(const char *from, const char *to, cred_t *cr); extern int zfs_secpolicy_destroy_perms(const char *name, cred_t *cr); extern zoneid_t getzoneid(void); /* SID stuff */ typedef struct ksiddomain { uint_t kd_ref; uint_t kd_len; char *kd_name; } ksiddomain_t; ksiddomain_t *ksid_lookupdomain(const char *); void ksiddomain_rele(ksiddomain_t *); #define DDI_SLEEP KM_SLEEP #define ddi_log_sysevent(_a, _b, _c, _d, _e, _f, _g) \ sysevent_post_event(_c, _d, _b, "libzpool", _e, _f) #define zfs_sleep_until(wakeup) \ do { \ hrtime_t delta = wakeup - gethrtime(); \ struct timespec ts; \ ts.tv_sec = delta / NANOSEC; \ ts.tv_nsec = delta % NANOSEC; \ (void) nanosleep(&ts, NULL); \ } while (0) typedef int fstrans_cookie_t; extern fstrans_cookie_t spl_fstrans_mark(void); extern void spl_fstrans_unmark(fstrans_cookie_t); extern int spl_fstrans_check(void); #endif /* _KERNEL */ #endif /* _SYS_ZFS_CONTEXT_H */ diff --git a/module/nvpair/nvpair_alloc_spl.c b/module/nvpair/nvpair_alloc_spl.c index a75b4a6c7fd7..f9055b94f100 100644 --- a/module/nvpair/nvpair_alloc_spl.c +++ b/module/nvpair/nvpair_alloc_spl.c @@ -1,95 +1,95 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License, Version 1.0 only * (the "License"). You may not use this file except in compliance * with the License. * * You can obtain a copy of the license at * usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright 2004 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ #include #include static void * nv_alloc_sleep_spl(nv_alloc_t *nva, size_t size) { - return (kmem_alloc(size, KM_SLEEP | KM_NODEBUG)); + return (kmem_alloc(size, KM_SLEEP)); } static void * nv_alloc_pushpage_spl(nv_alloc_t *nva, size_t size) { - return (kmem_alloc(size, KM_PUSHPAGE | KM_NODEBUG)); + return (kmem_alloc(size, KM_PUSHPAGE)); } static void * nv_alloc_nosleep_spl(nv_alloc_t *nva, size_t size) { return (kmem_alloc(size, KM_NOSLEEP)); } static void nv_free_spl(nv_alloc_t *nva, void *buf, size_t size) { kmem_free(buf, size); } const nv_alloc_ops_t spl_sleep_ops_def = { NULL, /* nv_ao_init() */ NULL, /* nv_ao_fini() */ nv_alloc_sleep_spl, /* nv_ao_alloc() */ nv_free_spl, /* nv_ao_free() */ NULL /* nv_ao_reset() */ }; const nv_alloc_ops_t spl_pushpage_ops_def = { NULL, /* nv_ao_init() */ NULL, /* nv_ao_fini() */ nv_alloc_pushpage_spl, /* nv_ao_alloc() */ nv_free_spl, /* nv_ao_free() */ NULL /* nv_ao_reset() */ }; const nv_alloc_ops_t spl_nosleep_ops_def = { NULL, /* nv_ao_init() */ NULL, /* nv_ao_fini() */ nv_alloc_nosleep_spl, /* nv_ao_alloc() */ nv_free_spl, /* nv_ao_free() */ NULL /* nv_ao_reset() */ }; nv_alloc_t nv_alloc_sleep_def = { &spl_sleep_ops_def, NULL }; nv_alloc_t nv_alloc_pushpage_def = { &spl_pushpage_ops_def, NULL }; nv_alloc_t nv_alloc_nosleep_def = { &spl_nosleep_ops_def, NULL }; nv_alloc_t *nv_alloc_sleep = &nv_alloc_sleep_def; nv_alloc_t *nv_alloc_pushpage = &nv_alloc_pushpage_def; nv_alloc_t *nv_alloc_nosleep = &nv_alloc_nosleep_def; diff --git a/module/zfs/ddt.c b/module/zfs/ddt.c index 0ce8ca1a98ef..3ecb5390931a 100644 --- a/module/zfs/ddt.c +++ b/module/zfs/ddt.c @@ -1,1234 +1,1234 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2012, 2014 by Delphix. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include static kmem_cache_t *ddt_cache; static kmem_cache_t *ddt_entry_cache; /* * Enable/disable prefetching of dedup-ed blocks which are going to be freed. */ int zfs_dedup_prefetch = 0; static const ddt_ops_t *ddt_ops[DDT_TYPES] = { &ddt_zap_ops, }; static const char *ddt_class_name[DDT_CLASSES] = { "ditto", "duplicate", "unique", }; static void ddt_object_create(ddt_t *ddt, enum ddt_type type, enum ddt_class class, dmu_tx_t *tx) { spa_t *spa = ddt->ddt_spa; objset_t *os = ddt->ddt_os; uint64_t *objectp = &ddt->ddt_object[type][class]; boolean_t prehash = zio_checksum_table[ddt->ddt_checksum].ci_dedup; char name[DDT_NAMELEN]; ddt_object_name(ddt, type, class, name); ASSERT(*objectp == 0); VERIFY(ddt_ops[type]->ddt_op_create(os, objectp, tx, prehash) == 0); ASSERT(*objectp != 0); VERIFY(zap_add(os, DMU_POOL_DIRECTORY_OBJECT, name, sizeof (uint64_t), 1, objectp, tx) == 0); VERIFY(zap_add(os, spa->spa_ddt_stat_object, name, sizeof (uint64_t), sizeof (ddt_histogram_t) / sizeof (uint64_t), &ddt->ddt_histogram[type][class], tx) == 0); } static void ddt_object_destroy(ddt_t *ddt, enum ddt_type type, enum ddt_class class, dmu_tx_t *tx) { spa_t *spa = ddt->ddt_spa; objset_t *os = ddt->ddt_os; uint64_t *objectp = &ddt->ddt_object[type][class]; uint64_t count; char name[DDT_NAMELEN]; ddt_object_name(ddt, type, class, name); ASSERT(*objectp != 0); ASSERT(ddt_histogram_empty(&ddt->ddt_histogram[type][class])); VERIFY(ddt_object_count(ddt, type, class, &count) == 0 && count == 0); VERIFY(zap_remove(os, DMU_POOL_DIRECTORY_OBJECT, name, tx) == 0); VERIFY(zap_remove(os, spa->spa_ddt_stat_object, name, tx) == 0); VERIFY(ddt_ops[type]->ddt_op_destroy(os, *objectp, tx) == 0); bzero(&ddt->ddt_object_stats[type][class], sizeof (ddt_object_t)); *objectp = 0; } static int ddt_object_load(ddt_t *ddt, enum ddt_type type, enum ddt_class class) { ddt_object_t *ddo = &ddt->ddt_object_stats[type][class]; dmu_object_info_t doi; uint64_t count; char name[DDT_NAMELEN]; int error; ddt_object_name(ddt, type, class, name); error = zap_lookup(ddt->ddt_os, DMU_POOL_DIRECTORY_OBJECT, name, sizeof (uint64_t), 1, &ddt->ddt_object[type][class]); if (error != 0) return (error); VERIFY0(zap_lookup(ddt->ddt_os, ddt->ddt_spa->spa_ddt_stat_object, name, sizeof (uint64_t), sizeof (ddt_histogram_t) / sizeof (uint64_t), &ddt->ddt_histogram[type][class])); /* * Seed the cached statistics. */ error = ddt_object_info(ddt, type, class, &doi); if (error) return (error); error = ddt_object_count(ddt, type, class, &count); if (error) return (error); ddo->ddo_count = count; ddo->ddo_dspace = doi.doi_physical_blocks_512 << 9; ddo->ddo_mspace = doi.doi_fill_count * doi.doi_data_block_size; return (0); } static void ddt_object_sync(ddt_t *ddt, enum ddt_type type, enum ddt_class class, dmu_tx_t *tx) { ddt_object_t *ddo = &ddt->ddt_object_stats[type][class]; dmu_object_info_t doi; uint64_t count; char name[DDT_NAMELEN]; ddt_object_name(ddt, type, class, name); VERIFY(zap_update(ddt->ddt_os, ddt->ddt_spa->spa_ddt_stat_object, name, sizeof (uint64_t), sizeof (ddt_histogram_t) / sizeof (uint64_t), &ddt->ddt_histogram[type][class], tx) == 0); /* * Cache DDT statistics; this is the only time they'll change. */ VERIFY(ddt_object_info(ddt, type, class, &doi) == 0); VERIFY(ddt_object_count(ddt, type, class, &count) == 0); ddo->ddo_count = count; ddo->ddo_dspace = doi.doi_physical_blocks_512 << 9; ddo->ddo_mspace = doi.doi_fill_count * doi.doi_data_block_size; } static int ddt_object_lookup(ddt_t *ddt, enum ddt_type type, enum ddt_class class, ddt_entry_t *dde) { if (!ddt_object_exists(ddt, type, class)) return (SET_ERROR(ENOENT)); return (ddt_ops[type]->ddt_op_lookup(ddt->ddt_os, ddt->ddt_object[type][class], dde)); } static void ddt_object_prefetch(ddt_t *ddt, enum ddt_type type, enum ddt_class class, ddt_entry_t *dde) { if (!ddt_object_exists(ddt, type, class)) return; ddt_ops[type]->ddt_op_prefetch(ddt->ddt_os, ddt->ddt_object[type][class], dde); } int ddt_object_update(ddt_t *ddt, enum ddt_type type, enum ddt_class class, ddt_entry_t *dde, dmu_tx_t *tx) { ASSERT(ddt_object_exists(ddt, type, class)); return (ddt_ops[type]->ddt_op_update(ddt->ddt_os, ddt->ddt_object[type][class], dde, tx)); } static int ddt_object_remove(ddt_t *ddt, enum ddt_type type, enum ddt_class class, ddt_entry_t *dde, dmu_tx_t *tx) { ASSERT(ddt_object_exists(ddt, type, class)); return (ddt_ops[type]->ddt_op_remove(ddt->ddt_os, ddt->ddt_object[type][class], dde, tx)); } int ddt_object_walk(ddt_t *ddt, enum ddt_type type, enum ddt_class class, uint64_t *walk, ddt_entry_t *dde) { ASSERT(ddt_object_exists(ddt, type, class)); return (ddt_ops[type]->ddt_op_walk(ddt->ddt_os, ddt->ddt_object[type][class], dde, walk)); } int ddt_object_count(ddt_t *ddt, enum ddt_type type, enum ddt_class class, uint64_t *count) { ASSERT(ddt_object_exists(ddt, type, class)); return (ddt_ops[type]->ddt_op_count(ddt->ddt_os, ddt->ddt_object[type][class], count)); } int ddt_object_info(ddt_t *ddt, enum ddt_type type, enum ddt_class class, dmu_object_info_t *doi) { if (!ddt_object_exists(ddt, type, class)) return (SET_ERROR(ENOENT)); return (dmu_object_info(ddt->ddt_os, ddt->ddt_object[type][class], doi)); } boolean_t ddt_object_exists(ddt_t *ddt, enum ddt_type type, enum ddt_class class) { return (!!ddt->ddt_object[type][class]); } void ddt_object_name(ddt_t *ddt, enum ddt_type type, enum ddt_class class, char *name) { (void) sprintf(name, DMU_POOL_DDT, zio_checksum_table[ddt->ddt_checksum].ci_name, ddt_ops[type]->ddt_op_name, ddt_class_name[class]); } void ddt_bp_fill(const ddt_phys_t *ddp, blkptr_t *bp, uint64_t txg) { int d; ASSERT(txg != 0); for (d = 0; d < SPA_DVAS_PER_BP; d++) bp->blk_dva[d] = ddp->ddp_dva[d]; BP_SET_BIRTH(bp, txg, ddp->ddp_phys_birth); } void ddt_bp_create(enum zio_checksum checksum, const ddt_key_t *ddk, const ddt_phys_t *ddp, blkptr_t *bp) { BP_ZERO(bp); if (ddp != NULL) ddt_bp_fill(ddp, bp, ddp->ddp_phys_birth); bp->blk_cksum = ddk->ddk_cksum; bp->blk_fill = 1; BP_SET_LSIZE(bp, DDK_GET_LSIZE(ddk)); BP_SET_PSIZE(bp, DDK_GET_PSIZE(ddk)); BP_SET_COMPRESS(bp, DDK_GET_COMPRESS(ddk)); BP_SET_CHECKSUM(bp, checksum); BP_SET_TYPE(bp, DMU_OT_DEDUP); BP_SET_LEVEL(bp, 0); BP_SET_DEDUP(bp, 0); BP_SET_BYTEORDER(bp, ZFS_HOST_BYTEORDER); } void ddt_key_fill(ddt_key_t *ddk, const blkptr_t *bp) { ddk->ddk_cksum = bp->blk_cksum; ddk->ddk_prop = 0; DDK_SET_LSIZE(ddk, BP_GET_LSIZE(bp)); DDK_SET_PSIZE(ddk, BP_GET_PSIZE(bp)); DDK_SET_COMPRESS(ddk, BP_GET_COMPRESS(bp)); } void ddt_phys_fill(ddt_phys_t *ddp, const blkptr_t *bp) { int d; ASSERT(ddp->ddp_phys_birth == 0); for (d = 0; d < SPA_DVAS_PER_BP; d++) ddp->ddp_dva[d] = bp->blk_dva[d]; ddp->ddp_phys_birth = BP_PHYSICAL_BIRTH(bp); } void ddt_phys_clear(ddt_phys_t *ddp) { bzero(ddp, sizeof (*ddp)); } void ddt_phys_addref(ddt_phys_t *ddp) { ddp->ddp_refcnt++; } void ddt_phys_decref(ddt_phys_t *ddp) { if (ddp) { ASSERT(ddp->ddp_refcnt > 0); ddp->ddp_refcnt--; } } void ddt_phys_free(ddt_t *ddt, ddt_key_t *ddk, ddt_phys_t *ddp, uint64_t txg) { blkptr_t blk; ddt_bp_create(ddt->ddt_checksum, ddk, ddp, &blk); ddt_phys_clear(ddp); zio_free(ddt->ddt_spa, txg, &blk); } ddt_phys_t * ddt_phys_select(const ddt_entry_t *dde, const blkptr_t *bp) { ddt_phys_t *ddp = (ddt_phys_t *)dde->dde_phys; int p; for (p = 0; p < DDT_PHYS_TYPES; p++, ddp++) { if (DVA_EQUAL(BP_IDENTITY(bp), &ddp->ddp_dva[0]) && BP_PHYSICAL_BIRTH(bp) == ddp->ddp_phys_birth) return (ddp); } return (NULL); } uint64_t ddt_phys_total_refcnt(const ddt_entry_t *dde) { uint64_t refcnt = 0; int p; for (p = DDT_PHYS_SINGLE; p <= DDT_PHYS_TRIPLE; p++) refcnt += dde->dde_phys[p].ddp_refcnt; return (refcnt); } static void ddt_stat_generate(ddt_t *ddt, ddt_entry_t *dde, ddt_stat_t *dds) { spa_t *spa = ddt->ddt_spa; ddt_phys_t *ddp = dde->dde_phys; ddt_key_t *ddk = &dde->dde_key; uint64_t lsize = DDK_GET_LSIZE(ddk); uint64_t psize = DDK_GET_PSIZE(ddk); int p, d; bzero(dds, sizeof (*dds)); for (p = 0; p < DDT_PHYS_TYPES; p++, ddp++) { uint64_t dsize = 0; uint64_t refcnt = ddp->ddp_refcnt; if (ddp->ddp_phys_birth == 0) continue; for (d = 0; d < SPA_DVAS_PER_BP; d++) dsize += dva_get_dsize_sync(spa, &ddp->ddp_dva[d]); dds->dds_blocks += 1; dds->dds_lsize += lsize; dds->dds_psize += psize; dds->dds_dsize += dsize; dds->dds_ref_blocks += refcnt; dds->dds_ref_lsize += lsize * refcnt; dds->dds_ref_psize += psize * refcnt; dds->dds_ref_dsize += dsize * refcnt; } } void ddt_stat_add(ddt_stat_t *dst, const ddt_stat_t *src, uint64_t neg) { const uint64_t *s = (const uint64_t *)src; uint64_t *d = (uint64_t *)dst; uint64_t *d_end = (uint64_t *)(dst + 1); ASSERT(neg == 0 || neg == -1ULL); /* add or subtract */ while (d < d_end) *d++ += (*s++ ^ neg) - neg; } static void ddt_stat_update(ddt_t *ddt, ddt_entry_t *dde, uint64_t neg) { ddt_stat_t dds; ddt_histogram_t *ddh; int bucket; ddt_stat_generate(ddt, dde, &dds); bucket = highbit64(dds.dds_ref_blocks) - 1; ASSERT(bucket >= 0); ddh = &ddt->ddt_histogram[dde->dde_type][dde->dde_class]; ddt_stat_add(&ddh->ddh_stat[bucket], &dds, neg); } void ddt_histogram_add(ddt_histogram_t *dst, const ddt_histogram_t *src) { int h; for (h = 0; h < 64; h++) ddt_stat_add(&dst->ddh_stat[h], &src->ddh_stat[h], 0); } void ddt_histogram_stat(ddt_stat_t *dds, const ddt_histogram_t *ddh) { int h; bzero(dds, sizeof (*dds)); for (h = 0; h < 64; h++) ddt_stat_add(dds, &ddh->ddh_stat[h], 0); } boolean_t ddt_histogram_empty(const ddt_histogram_t *ddh) { const uint64_t *s = (const uint64_t *)ddh; const uint64_t *s_end = (const uint64_t *)(ddh + 1); while (s < s_end) if (*s++ != 0) return (B_FALSE); return (B_TRUE); } void ddt_get_dedup_object_stats(spa_t *spa, ddt_object_t *ddo_total) { enum zio_checksum c; enum ddt_type type; enum ddt_class class; /* Sum the statistics we cached in ddt_object_sync(). */ for (c = 0; c < ZIO_CHECKSUM_FUNCTIONS; c++) { ddt_t *ddt = spa->spa_ddt[c]; for (type = 0; type < DDT_TYPES; type++) { for (class = 0; class < DDT_CLASSES; class++) { ddt_object_t *ddo = &ddt->ddt_object_stats[type][class]; ddo_total->ddo_count += ddo->ddo_count; ddo_total->ddo_dspace += ddo->ddo_dspace; ddo_total->ddo_mspace += ddo->ddo_mspace; } } } /* ... and compute the averages. */ if (ddo_total->ddo_count != 0) { ddo_total->ddo_dspace /= ddo_total->ddo_count; ddo_total->ddo_mspace /= ddo_total->ddo_count; } } void ddt_get_dedup_histogram(spa_t *spa, ddt_histogram_t *ddh) { enum zio_checksum c; enum ddt_type type; enum ddt_class class; for (c = 0; c < ZIO_CHECKSUM_FUNCTIONS; c++) { ddt_t *ddt = spa->spa_ddt[c]; for (type = 0; type < DDT_TYPES; type++) { for (class = 0; class < DDT_CLASSES; class++) { ddt_histogram_add(ddh, &ddt->ddt_histogram_cache[type][class]); } } } } void ddt_get_dedup_stats(spa_t *spa, ddt_stat_t *dds_total) { ddt_histogram_t *ddh_total; ddh_total = kmem_zalloc(sizeof (ddt_histogram_t), KM_PUSHPAGE); ddt_get_dedup_histogram(spa, ddh_total); ddt_histogram_stat(dds_total, ddh_total); kmem_free(ddh_total, sizeof (ddt_histogram_t)); } uint64_t ddt_get_dedup_dspace(spa_t *spa) { ddt_stat_t dds_total = { 0 }; ddt_get_dedup_stats(spa, &dds_total); return (dds_total.dds_ref_dsize - dds_total.dds_dsize); } uint64_t ddt_get_pool_dedup_ratio(spa_t *spa) { ddt_stat_t dds_total = { 0 }; ddt_get_dedup_stats(spa, &dds_total); if (dds_total.dds_dsize == 0) return (100); return (dds_total.dds_ref_dsize * 100 / dds_total.dds_dsize); } int ddt_ditto_copies_needed(ddt_t *ddt, ddt_entry_t *dde, ddt_phys_t *ddp_willref) { spa_t *spa = ddt->ddt_spa; uint64_t total_refcnt = 0; uint64_t ditto = spa->spa_dedup_ditto; int total_copies = 0; int desired_copies = 0; int p; for (p = DDT_PHYS_SINGLE; p <= DDT_PHYS_TRIPLE; p++) { ddt_phys_t *ddp = &dde->dde_phys[p]; zio_t *zio = dde->dde_lead_zio[p]; uint64_t refcnt = ddp->ddp_refcnt; /* committed refs */ if (zio != NULL) refcnt += zio->io_parent_count; /* pending refs */ if (ddp == ddp_willref) refcnt++; /* caller's ref */ if (refcnt != 0) { total_refcnt += refcnt; total_copies += p; } } if (ditto == 0 || ditto > UINT32_MAX) ditto = UINT32_MAX; if (total_refcnt >= 1) desired_copies++; if (total_refcnt >= ditto) desired_copies++; if (total_refcnt >= ditto * ditto) desired_copies++; return (MAX(desired_copies, total_copies) - total_copies); } int ddt_ditto_copies_present(ddt_entry_t *dde) { ddt_phys_t *ddp = &dde->dde_phys[DDT_PHYS_DITTO]; dva_t *dva = ddp->ddp_dva; int copies = 0 - DVA_GET_GANG(dva); int d; for (d = 0; d < SPA_DVAS_PER_BP; d++, dva++) if (DVA_IS_VALID(dva)) copies++; ASSERT(copies >= 0 && copies < SPA_DVAS_PER_BP); return (copies); } size_t ddt_compress(void *src, uchar_t *dst, size_t s_len, size_t d_len) { uchar_t *version = dst++; int cpfunc = ZIO_COMPRESS_ZLE; zio_compress_info_t *ci = &zio_compress_table[cpfunc]; size_t c_len; ASSERT(d_len >= s_len + 1); /* no compression plus version byte */ c_len = ci->ci_compress(src, dst, s_len, d_len - 1, ci->ci_level); if (c_len == s_len) { cpfunc = ZIO_COMPRESS_OFF; bcopy(src, dst, s_len); } *version = cpfunc; /* CONSTCOND */ if (ZFS_HOST_BYTEORDER) *version |= DDT_COMPRESS_BYTEORDER_MASK; return (c_len + 1); } void ddt_decompress(uchar_t *src, void *dst, size_t s_len, size_t d_len) { uchar_t version = *src++; int cpfunc = version & DDT_COMPRESS_FUNCTION_MASK; zio_compress_info_t *ci = &zio_compress_table[cpfunc]; if (ci->ci_decompress != NULL) (void) ci->ci_decompress(src, dst, s_len, d_len, ci->ci_level); else bcopy(src, dst, d_len); if (((version & DDT_COMPRESS_BYTEORDER_MASK) != 0) != (ZFS_HOST_BYTEORDER != 0)) byteswap_uint64_array(dst, d_len); } ddt_t * ddt_select_by_checksum(spa_t *spa, enum zio_checksum c) { return (spa->spa_ddt[c]); } ddt_t * ddt_select(spa_t *spa, const blkptr_t *bp) { return (spa->spa_ddt[BP_GET_CHECKSUM(bp)]); } void ddt_enter(ddt_t *ddt) { mutex_enter(&ddt->ddt_lock); } void ddt_exit(ddt_t *ddt) { mutex_exit(&ddt->ddt_lock); } void ddt_init(void) { ddt_cache = kmem_cache_create("ddt_cache", sizeof (ddt_t), 0, NULL, NULL, NULL, NULL, NULL, 0); ddt_entry_cache = kmem_cache_create("ddt_entry_cache", sizeof (ddt_entry_t), 0, NULL, NULL, NULL, NULL, NULL, 0); } void ddt_fini(void) { kmem_cache_destroy(ddt_entry_cache); kmem_cache_destroy(ddt_cache); } static ddt_entry_t * ddt_alloc(const ddt_key_t *ddk) { ddt_entry_t *dde; dde = kmem_cache_alloc(ddt_entry_cache, KM_PUSHPAGE); bzero(dde, sizeof (ddt_entry_t)); cv_init(&dde->dde_cv, NULL, CV_DEFAULT, NULL); dde->dde_key = *ddk; return (dde); } static void ddt_free(ddt_entry_t *dde) { int p; ASSERT(!dde->dde_loading); for (p = 0; p < DDT_PHYS_TYPES; p++) ASSERT(dde->dde_lead_zio[p] == NULL); if (dde->dde_repair_data != NULL) zio_buf_free(dde->dde_repair_data, DDK_GET_PSIZE(&dde->dde_key)); cv_destroy(&dde->dde_cv); kmem_cache_free(ddt_entry_cache, dde); } void ddt_remove(ddt_t *ddt, ddt_entry_t *dde) { ASSERT(MUTEX_HELD(&ddt->ddt_lock)); avl_remove(&ddt->ddt_tree, dde); ddt_free(dde); } ddt_entry_t * ddt_lookup(ddt_t *ddt, const blkptr_t *bp, boolean_t add) { ddt_entry_t *dde, dde_search; enum ddt_type type; enum ddt_class class; avl_index_t where; int error; ASSERT(MUTEX_HELD(&ddt->ddt_lock)); ddt_key_fill(&dde_search.dde_key, bp); dde = avl_find(&ddt->ddt_tree, &dde_search, &where); if (dde == NULL) { if (!add) return (NULL); dde = ddt_alloc(&dde_search.dde_key); avl_insert(&ddt->ddt_tree, dde, where); } while (dde->dde_loading) cv_wait(&dde->dde_cv, &ddt->ddt_lock); if (dde->dde_loaded) return (dde); dde->dde_loading = B_TRUE; ddt_exit(ddt); error = ENOENT; for (type = 0; type < DDT_TYPES; type++) { for (class = 0; class < DDT_CLASSES; class++) { error = ddt_object_lookup(ddt, type, class, dde); if (error != ENOENT) break; } if (error != ENOENT) break; } ASSERT(error == 0 || error == ENOENT); ddt_enter(ddt); ASSERT(dde->dde_loaded == B_FALSE); ASSERT(dde->dde_loading == B_TRUE); dde->dde_type = type; /* will be DDT_TYPES if no entry found */ dde->dde_class = class; /* will be DDT_CLASSES if no entry found */ dde->dde_loaded = B_TRUE; dde->dde_loading = B_FALSE; if (error == 0) ddt_stat_update(ddt, dde, -1ULL); cv_broadcast(&dde->dde_cv); return (dde); } void ddt_prefetch(spa_t *spa, const blkptr_t *bp) { ddt_t *ddt; ddt_entry_t dde; enum ddt_type type; enum ddt_class class; if (!zfs_dedup_prefetch || bp == NULL || !BP_GET_DEDUP(bp)) return; /* * We only remove the DDT once all tables are empty and only * prefetch dedup blocks when there are entries in the DDT. * Thus no locking is required as the DDT can't disappear on us. */ ddt = ddt_select(spa, bp); ddt_key_fill(&dde.dde_key, bp); for (type = 0; type < DDT_TYPES; type++) { for (class = 0; class < DDT_CLASSES; class++) { ddt_object_prefetch(ddt, type, class, &dde); } } } int ddt_entry_compare(const void *x1, const void *x2) { const ddt_entry_t *dde1 = x1; const ddt_entry_t *dde2 = x2; const uint64_t *u1 = (const uint64_t *)&dde1->dde_key; const uint64_t *u2 = (const uint64_t *)&dde2->dde_key; int i; for (i = 0; i < DDT_KEY_WORDS; i++) { if (u1[i] < u2[i]) return (-1); if (u1[i] > u2[i]) return (1); } return (0); } static ddt_t * ddt_table_alloc(spa_t *spa, enum zio_checksum c) { ddt_t *ddt; - ddt = kmem_cache_alloc(ddt_cache, KM_PUSHPAGE | KM_NODEBUG); + ddt = kmem_cache_alloc(ddt_cache, KM_PUSHPAGE); bzero(ddt, sizeof (ddt_t)); mutex_init(&ddt->ddt_lock, NULL, MUTEX_DEFAULT, NULL); avl_create(&ddt->ddt_tree, ddt_entry_compare, sizeof (ddt_entry_t), offsetof(ddt_entry_t, dde_node)); avl_create(&ddt->ddt_repair_tree, ddt_entry_compare, sizeof (ddt_entry_t), offsetof(ddt_entry_t, dde_node)); ddt->ddt_checksum = c; ddt->ddt_spa = spa; ddt->ddt_os = spa->spa_meta_objset; return (ddt); } static void ddt_table_free(ddt_t *ddt) { ASSERT(avl_numnodes(&ddt->ddt_tree) == 0); ASSERT(avl_numnodes(&ddt->ddt_repair_tree) == 0); avl_destroy(&ddt->ddt_tree); avl_destroy(&ddt->ddt_repair_tree); mutex_destroy(&ddt->ddt_lock); kmem_cache_free(ddt_cache, ddt); } void ddt_create(spa_t *spa) { enum zio_checksum c; spa->spa_dedup_checksum = ZIO_DEDUPCHECKSUM; for (c = 0; c < ZIO_CHECKSUM_FUNCTIONS; c++) spa->spa_ddt[c] = ddt_table_alloc(spa, c); } int ddt_load(spa_t *spa) { enum zio_checksum c; enum ddt_type type; enum ddt_class class; int error; ddt_create(spa); error = zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_DDT_STATS, sizeof (uint64_t), 1, &spa->spa_ddt_stat_object); if (error) return (error == ENOENT ? 0 : error); for (c = 0; c < ZIO_CHECKSUM_FUNCTIONS; c++) { ddt_t *ddt = spa->spa_ddt[c]; for (type = 0; type < DDT_TYPES; type++) { for (class = 0; class < DDT_CLASSES; class++) { error = ddt_object_load(ddt, type, class); if (error != 0 && error != ENOENT) return (error); } } /* * Seed the cached histograms. */ bcopy(ddt->ddt_histogram, &ddt->ddt_histogram_cache, sizeof (ddt->ddt_histogram)); } return (0); } void ddt_unload(spa_t *spa) { enum zio_checksum c; for (c = 0; c < ZIO_CHECKSUM_FUNCTIONS; c++) { if (spa->spa_ddt[c]) { ddt_table_free(spa->spa_ddt[c]); spa->spa_ddt[c] = NULL; } } } boolean_t ddt_class_contains(spa_t *spa, enum ddt_class max_class, const blkptr_t *bp) { ddt_t *ddt; ddt_entry_t *dde; enum ddt_type type; enum ddt_class class; if (!BP_GET_DEDUP(bp)) return (B_FALSE); if (max_class == DDT_CLASS_UNIQUE) return (B_TRUE); ddt = spa->spa_ddt[BP_GET_CHECKSUM(bp)]; dde = kmem_cache_alloc(ddt_entry_cache, KM_PUSHPAGE); ddt_key_fill(&(dde->dde_key), bp); for (type = 0; type < DDT_TYPES; type++) { for (class = 0; class <= max_class; class++) { if (ddt_object_lookup(ddt, type, class, dde) == 0) { kmem_cache_free(ddt_entry_cache, dde); return (B_TRUE); } } } kmem_cache_free(ddt_entry_cache, dde); return (B_FALSE); } ddt_entry_t * ddt_repair_start(ddt_t *ddt, const blkptr_t *bp) { ddt_key_t ddk; ddt_entry_t *dde; enum ddt_type type; enum ddt_class class; ddt_key_fill(&ddk, bp); dde = ddt_alloc(&ddk); for (type = 0; type < DDT_TYPES; type++) { for (class = 0; class < DDT_CLASSES; class++) { /* * We can only do repair if there are multiple copies * of the block. For anything in the UNIQUE class, * there's definitely only one copy, so don't even try. */ if (class != DDT_CLASS_UNIQUE && ddt_object_lookup(ddt, type, class, dde) == 0) return (dde); } } bzero(dde->dde_phys, sizeof (dde->dde_phys)); return (dde); } void ddt_repair_done(ddt_t *ddt, ddt_entry_t *dde) { avl_index_t where; ddt_enter(ddt); if (dde->dde_repair_data != NULL && spa_writeable(ddt->ddt_spa) && avl_find(&ddt->ddt_repair_tree, dde, &where) == NULL) avl_insert(&ddt->ddt_repair_tree, dde, where); else ddt_free(dde); ddt_exit(ddt); } static void ddt_repair_entry_done(zio_t *zio) { ddt_entry_t *rdde = zio->io_private; ddt_free(rdde); } static void ddt_repair_entry(ddt_t *ddt, ddt_entry_t *dde, ddt_entry_t *rdde, zio_t *rio) { ddt_phys_t *ddp = dde->dde_phys; ddt_phys_t *rddp = rdde->dde_phys; ddt_key_t *ddk = &dde->dde_key; ddt_key_t *rddk = &rdde->dde_key; zio_t *zio; blkptr_t blk; int p; zio = zio_null(rio, rio->io_spa, NULL, ddt_repair_entry_done, rdde, rio->io_flags); for (p = 0; p < DDT_PHYS_TYPES; p++, ddp++, rddp++) { if (ddp->ddp_phys_birth == 0 || ddp->ddp_phys_birth != rddp->ddp_phys_birth || bcmp(ddp->ddp_dva, rddp->ddp_dva, sizeof (ddp->ddp_dva))) continue; ddt_bp_create(ddt->ddt_checksum, ddk, ddp, &blk); zio_nowait(zio_rewrite(zio, zio->io_spa, 0, &blk, rdde->dde_repair_data, DDK_GET_PSIZE(rddk), NULL, NULL, ZIO_PRIORITY_SYNC_WRITE, ZIO_DDT_CHILD_FLAGS(zio), NULL)); } zio_nowait(zio); } static void ddt_repair_table(ddt_t *ddt, zio_t *rio) { spa_t *spa = ddt->ddt_spa; ddt_entry_t *dde, *rdde_next, *rdde; avl_tree_t *t = &ddt->ddt_repair_tree; blkptr_t blk; if (spa_sync_pass(spa) > 1) return; ddt_enter(ddt); for (rdde = avl_first(t); rdde != NULL; rdde = rdde_next) { rdde_next = AVL_NEXT(t, rdde); avl_remove(&ddt->ddt_repair_tree, rdde); ddt_exit(ddt); ddt_bp_create(ddt->ddt_checksum, &rdde->dde_key, NULL, &blk); dde = ddt_repair_start(ddt, &blk); ddt_repair_entry(ddt, dde, rdde, rio); ddt_repair_done(ddt, dde); ddt_enter(ddt); } ddt_exit(ddt); } static void ddt_sync_entry(ddt_t *ddt, ddt_entry_t *dde, dmu_tx_t *tx, uint64_t txg) { dsl_pool_t *dp = ddt->ddt_spa->spa_dsl_pool; ddt_phys_t *ddp = dde->dde_phys; ddt_key_t *ddk = &dde->dde_key; enum ddt_type otype = dde->dde_type; enum ddt_type ntype = DDT_TYPE_CURRENT; enum ddt_class oclass = dde->dde_class; enum ddt_class nclass; uint64_t total_refcnt = 0; int p; ASSERT(dde->dde_loaded); ASSERT(!dde->dde_loading); for (p = 0; p < DDT_PHYS_TYPES; p++, ddp++) { ASSERT(dde->dde_lead_zio[p] == NULL); if (ddp->ddp_phys_birth == 0) { ASSERT(ddp->ddp_refcnt == 0); continue; } if (p == DDT_PHYS_DITTO) { if (ddt_ditto_copies_needed(ddt, dde, NULL) == 0) ddt_phys_free(ddt, ddk, ddp, txg); continue; } if (ddp->ddp_refcnt == 0) ddt_phys_free(ddt, ddk, ddp, txg); total_refcnt += ddp->ddp_refcnt; } if (dde->dde_phys[DDT_PHYS_DITTO].ddp_phys_birth != 0) nclass = DDT_CLASS_DITTO; else if (total_refcnt > 1) nclass = DDT_CLASS_DUPLICATE; else nclass = DDT_CLASS_UNIQUE; if (otype != DDT_TYPES && (otype != ntype || oclass != nclass || total_refcnt == 0)) { VERIFY(ddt_object_remove(ddt, otype, oclass, dde, tx) == 0); ASSERT(ddt_object_lookup(ddt, otype, oclass, dde) == ENOENT); } if (total_refcnt != 0) { dde->dde_type = ntype; dde->dde_class = nclass; ddt_stat_update(ddt, dde, 0); if (!ddt_object_exists(ddt, ntype, nclass)) ddt_object_create(ddt, ntype, nclass, tx); VERIFY(ddt_object_update(ddt, ntype, nclass, dde, tx) == 0); /* * If the class changes, the order that we scan this bp * changes. If it decreases, we could miss it, so * scan it right now. (This covers both class changing * while we are doing ddt_walk(), and when we are * traversing.) */ if (nclass < oclass) { dsl_scan_ddt_entry(dp->dp_scan, ddt->ddt_checksum, dde, tx); } } } static void ddt_sync_table(ddt_t *ddt, dmu_tx_t *tx, uint64_t txg) { spa_t *spa = ddt->ddt_spa; ddt_entry_t *dde; void *cookie = NULL; enum ddt_type type; enum ddt_class class; if (avl_numnodes(&ddt->ddt_tree) == 0) return; ASSERT(spa->spa_uberblock.ub_version >= SPA_VERSION_DEDUP); if (spa->spa_ddt_stat_object == 0) { spa->spa_ddt_stat_object = zap_create_link(ddt->ddt_os, DMU_OT_DDT_STATS, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_DDT_STATS, tx); } while ((dde = avl_destroy_nodes(&ddt->ddt_tree, &cookie)) != NULL) { ddt_sync_entry(ddt, dde, tx, txg); ddt_free(dde); } for (type = 0; type < DDT_TYPES; type++) { uint64_t add, count = 0; for (class = 0; class < DDT_CLASSES; class++) { if (ddt_object_exists(ddt, type, class)) { ddt_object_sync(ddt, type, class, tx); VERIFY(ddt_object_count(ddt, type, class, &add) == 0); count += add; } } for (class = 0; class < DDT_CLASSES; class++) { if (count == 0 && ddt_object_exists(ddt, type, class)) ddt_object_destroy(ddt, type, class, tx); } } bcopy(ddt->ddt_histogram, &ddt->ddt_histogram_cache, sizeof (ddt->ddt_histogram)); } void ddt_sync(spa_t *spa, uint64_t txg) { dmu_tx_t *tx; zio_t *rio = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE); enum zio_checksum c; ASSERT(spa_syncing_txg(spa) == txg); tx = dmu_tx_create_assigned(spa->spa_dsl_pool, txg); for (c = 0; c < ZIO_CHECKSUM_FUNCTIONS; c++) { ddt_t *ddt = spa->spa_ddt[c]; if (ddt == NULL) continue; ddt_sync_table(ddt, tx, txg); ddt_repair_table(ddt, rio); } (void) zio_wait(rio); dmu_tx_commit(tx); } int ddt_walk(spa_t *spa, ddt_bookmark_t *ddb, ddt_entry_t *dde) { do { do { do { ddt_t *ddt = spa->spa_ddt[ddb->ddb_checksum]; int error = ENOENT; if (ddt_object_exists(ddt, ddb->ddb_type, ddb->ddb_class)) { error = ddt_object_walk(ddt, ddb->ddb_type, ddb->ddb_class, &ddb->ddb_cursor, dde); } dde->dde_type = ddb->ddb_type; dde->dde_class = ddb->ddb_class; if (error == 0) return (0); if (error != ENOENT) return (error); ddb->ddb_cursor = 0; } while (++ddb->ddb_checksum < ZIO_CHECKSUM_FUNCTIONS); ddb->ddb_checksum = 0; } while (++ddb->ddb_type < DDT_TYPES); ddb->ddb_type = 0; } while (++ddb->ddb_class < DDT_CLASSES); return (SET_ERROR(ENOENT)); } #if defined(_KERNEL) && defined(HAVE_SPL) module_param(zfs_dedup_prefetch, int, 0644); MODULE_PARM_DESC(zfs_dedup_prefetch, "Enable prefetching dedup-ed blks"); #endif diff --git a/module/zfs/dmu.c b/module/zfs/dmu.c index 98defa3bbfd6..607da887b04b 100644 --- a/module/zfs/dmu.c +++ b/module/zfs/dmu.c @@ -1,2085 +1,2084 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2014 by Delphix. All rights reserved. * Copyright (c) 2013 by Saso Kiselkov. All rights reserved. * Copyright (c) 2014, Nexenta Systems, Inc. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef _KERNEL #include #include #endif /* * Enable/disable nopwrite feature. */ int zfs_nopwrite_enabled = 1; const dmu_object_type_info_t dmu_ot[DMU_OT_NUMTYPES] = { { DMU_BSWAP_UINT8, TRUE, "unallocated" }, { DMU_BSWAP_ZAP, TRUE, "object directory" }, { DMU_BSWAP_UINT64, TRUE, "object array" }, { DMU_BSWAP_UINT8, TRUE, "packed nvlist" }, { DMU_BSWAP_UINT64, TRUE, "packed nvlist size" }, { DMU_BSWAP_UINT64, TRUE, "bpobj" }, { DMU_BSWAP_UINT64, TRUE, "bpobj header" }, { DMU_BSWAP_UINT64, TRUE, "SPA space map header" }, { DMU_BSWAP_UINT64, TRUE, "SPA space map" }, { DMU_BSWAP_UINT64, TRUE, "ZIL intent log" }, { DMU_BSWAP_DNODE, TRUE, "DMU dnode" }, { DMU_BSWAP_OBJSET, TRUE, "DMU objset" }, { DMU_BSWAP_UINT64, TRUE, "DSL directory" }, { DMU_BSWAP_ZAP, TRUE, "DSL directory child map"}, { DMU_BSWAP_ZAP, TRUE, "DSL dataset snap map" }, { DMU_BSWAP_ZAP, TRUE, "DSL props" }, { DMU_BSWAP_UINT64, TRUE, "DSL dataset" }, { DMU_BSWAP_ZNODE, TRUE, "ZFS znode" }, { DMU_BSWAP_OLDACL, TRUE, "ZFS V0 ACL" }, { DMU_BSWAP_UINT8, FALSE, "ZFS plain file" }, { DMU_BSWAP_ZAP, TRUE, "ZFS directory" }, { DMU_BSWAP_ZAP, TRUE, "ZFS master node" }, { DMU_BSWAP_ZAP, TRUE, "ZFS delete queue" }, { DMU_BSWAP_UINT8, FALSE, "zvol object" }, { DMU_BSWAP_ZAP, TRUE, "zvol prop" }, { DMU_BSWAP_UINT8, FALSE, "other uint8[]" }, { DMU_BSWAP_UINT64, FALSE, "other uint64[]" }, { DMU_BSWAP_ZAP, TRUE, "other ZAP" }, { DMU_BSWAP_ZAP, TRUE, "persistent error log" }, { DMU_BSWAP_UINT8, TRUE, "SPA history" }, { DMU_BSWAP_UINT64, TRUE, "SPA history offsets" }, { DMU_BSWAP_ZAP, TRUE, "Pool properties" }, { DMU_BSWAP_ZAP, TRUE, "DSL permissions" }, { DMU_BSWAP_ACL, TRUE, "ZFS ACL" }, { DMU_BSWAP_UINT8, TRUE, "ZFS SYSACL" }, { DMU_BSWAP_UINT8, TRUE, "FUID table" }, { DMU_BSWAP_UINT64, TRUE, "FUID table size" }, { DMU_BSWAP_ZAP, TRUE, "DSL dataset next clones"}, { DMU_BSWAP_ZAP, TRUE, "scan work queue" }, { DMU_BSWAP_ZAP, TRUE, "ZFS user/group used" }, { DMU_BSWAP_ZAP, TRUE, "ZFS user/group quota" }, { DMU_BSWAP_ZAP, TRUE, "snapshot refcount tags"}, { DMU_BSWAP_ZAP, TRUE, "DDT ZAP algorithm" }, { DMU_BSWAP_ZAP, TRUE, "DDT statistics" }, { DMU_BSWAP_UINT8, TRUE, "System attributes" }, { DMU_BSWAP_ZAP, TRUE, "SA master node" }, { DMU_BSWAP_ZAP, TRUE, "SA attr registration" }, { DMU_BSWAP_ZAP, TRUE, "SA attr layouts" }, { DMU_BSWAP_ZAP, TRUE, "scan translations" }, { DMU_BSWAP_UINT8, FALSE, "deduplicated block" }, { DMU_BSWAP_ZAP, TRUE, "DSL deadlist map" }, { DMU_BSWAP_UINT64, TRUE, "DSL deadlist map hdr" }, { DMU_BSWAP_ZAP, TRUE, "DSL dir clones" }, { DMU_BSWAP_UINT64, TRUE, "bpobj subobj" } }; const dmu_object_byteswap_info_t dmu_ot_byteswap[DMU_BSWAP_NUMFUNCS] = { { byteswap_uint8_array, "uint8" }, { byteswap_uint16_array, "uint16" }, { byteswap_uint32_array, "uint32" }, { byteswap_uint64_array, "uint64" }, { zap_byteswap, "zap" }, { dnode_buf_byteswap, "dnode" }, { dmu_objset_byteswap, "objset" }, { zfs_znode_byteswap, "znode" }, { zfs_oldacl_byteswap, "oldacl" }, { zfs_acl_byteswap, "acl" } }; int dmu_buf_hold_noread(objset_t *os, uint64_t object, uint64_t offset, void *tag, dmu_buf_t **dbp) { dnode_t *dn; uint64_t blkid; dmu_buf_impl_t *db; int err; err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); blkid = dbuf_whichblock(dn, offset); rw_enter(&dn->dn_struct_rwlock, RW_READER); db = dbuf_hold(dn, blkid, tag); rw_exit(&dn->dn_struct_rwlock); dnode_rele(dn, FTAG); if (db == NULL) { *dbp = NULL; return (SET_ERROR(EIO)); } *dbp = &db->db; return (err); } int dmu_buf_hold(objset_t *os, uint64_t object, uint64_t offset, void *tag, dmu_buf_t **dbp, int flags) { int err; int db_flags = DB_RF_CANFAIL; if (flags & DMU_READ_NO_PREFETCH) db_flags |= DB_RF_NOPREFETCH; err = dmu_buf_hold_noread(os, object, offset, tag, dbp); if (err == 0) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)(*dbp); err = dbuf_read(db, NULL, db_flags); if (err != 0) { dbuf_rele(db, tag); *dbp = NULL; } } return (err); } int dmu_bonus_max(void) { return (DN_MAX_BONUSLEN); } int dmu_set_bonus(dmu_buf_t *db_fake, int newsize, dmu_tx_t *tx) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; dnode_t *dn; int error; DB_DNODE_ENTER(db); dn = DB_DNODE(db); if (dn->dn_bonus != db) { error = SET_ERROR(EINVAL); } else if (newsize < 0 || newsize > db_fake->db_size) { error = SET_ERROR(EINVAL); } else { dnode_setbonuslen(dn, newsize, tx); error = 0; } DB_DNODE_EXIT(db); return (error); } int dmu_set_bonustype(dmu_buf_t *db_fake, dmu_object_type_t type, dmu_tx_t *tx) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; dnode_t *dn; int error; DB_DNODE_ENTER(db); dn = DB_DNODE(db); if (!DMU_OT_IS_VALID(type)) { error = SET_ERROR(EINVAL); } else if (dn->dn_bonus != db) { error = SET_ERROR(EINVAL); } else { dnode_setbonus_type(dn, type, tx); error = 0; } DB_DNODE_EXIT(db); return (error); } dmu_object_type_t dmu_get_bonustype(dmu_buf_t *db_fake) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; dnode_t *dn; dmu_object_type_t type; DB_DNODE_ENTER(db); dn = DB_DNODE(db); type = dn->dn_bonustype; DB_DNODE_EXIT(db); return (type); } int dmu_rm_spill(objset_t *os, uint64_t object, dmu_tx_t *tx) { dnode_t *dn; int error; error = dnode_hold(os, object, FTAG, &dn); dbuf_rm_spill(dn, tx); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); dnode_rm_spill(dn, tx); rw_exit(&dn->dn_struct_rwlock); dnode_rele(dn, FTAG); return (error); } /* * returns ENOENT, EIO, or 0. */ int dmu_bonus_hold(objset_t *os, uint64_t object, void *tag, dmu_buf_t **dbp) { dnode_t *dn; dmu_buf_impl_t *db; int error; error = dnode_hold(os, object, FTAG, &dn); if (error) return (error); rw_enter(&dn->dn_struct_rwlock, RW_READER); if (dn->dn_bonus == NULL) { rw_exit(&dn->dn_struct_rwlock); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); if (dn->dn_bonus == NULL) dbuf_create_bonus(dn); } db = dn->dn_bonus; /* as long as the bonus buf is held, the dnode will be held */ if (refcount_add(&db->db_holds, tag) == 1) { VERIFY(dnode_add_ref(dn, db)); (void) atomic_inc_32_nv(&dn->dn_dbufs_count); } /* * Wait to drop dn_struct_rwlock until after adding the bonus dbuf's * hold and incrementing the dbuf count to ensure that dnode_move() sees * a dnode hold for every dbuf. */ rw_exit(&dn->dn_struct_rwlock); dnode_rele(dn, FTAG); VERIFY(0 == dbuf_read(db, NULL, DB_RF_MUST_SUCCEED | DB_RF_NOPREFETCH)); *dbp = &db->db; return (0); } /* * returns ENOENT, EIO, or 0. * * This interface will allocate a blank spill dbuf when a spill blk * doesn't already exist on the dnode. * * if you only want to find an already existing spill db, then * dmu_spill_hold_existing() should be used. */ int dmu_spill_hold_by_dnode(dnode_t *dn, uint32_t flags, void *tag, dmu_buf_t **dbp) { dmu_buf_impl_t *db = NULL; int err; if ((flags & DB_RF_HAVESTRUCT) == 0) rw_enter(&dn->dn_struct_rwlock, RW_READER); db = dbuf_hold(dn, DMU_SPILL_BLKID, tag); if ((flags & DB_RF_HAVESTRUCT) == 0) rw_exit(&dn->dn_struct_rwlock); ASSERT(db != NULL); err = dbuf_read(db, NULL, flags); if (err == 0) *dbp = &db->db; else dbuf_rele(db, tag); return (err); } int dmu_spill_hold_existing(dmu_buf_t *bonus, void *tag, dmu_buf_t **dbp) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)bonus; dnode_t *dn; int err; DB_DNODE_ENTER(db); dn = DB_DNODE(db); if (spa_version(dn->dn_objset->os_spa) < SPA_VERSION_SA) { err = SET_ERROR(EINVAL); } else { rw_enter(&dn->dn_struct_rwlock, RW_READER); if (!dn->dn_have_spill) { err = SET_ERROR(ENOENT); } else { err = dmu_spill_hold_by_dnode(dn, DB_RF_HAVESTRUCT | DB_RF_CANFAIL, tag, dbp); } rw_exit(&dn->dn_struct_rwlock); } DB_DNODE_EXIT(db); return (err); } int dmu_spill_hold_by_bonus(dmu_buf_t *bonus, void *tag, dmu_buf_t **dbp) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)bonus; dnode_t *dn; int err; DB_DNODE_ENTER(db); dn = DB_DNODE(db); err = dmu_spill_hold_by_dnode(dn, DB_RF_CANFAIL, tag, dbp); DB_DNODE_EXIT(db); return (err); } /* * Note: longer-term, we should modify all of the dmu_buf_*() interfaces * to take a held dnode rather than -- the lookup is wasteful, * and can induce severe lock contention when writing to several files * whose dnodes are in the same block. */ static int dmu_buf_hold_array_by_dnode(dnode_t *dn, uint64_t offset, uint64_t length, int read, void *tag, int *numbufsp, dmu_buf_t ***dbpp, uint32_t flags) { dmu_buf_t **dbp; uint64_t blkid, nblks, i; uint32_t dbuf_flags; int err; zio_t *zio; ASSERT(length <= DMU_MAX_ACCESS); dbuf_flags = DB_RF_CANFAIL | DB_RF_NEVERWAIT | DB_RF_HAVESTRUCT; if (flags & DMU_READ_NO_PREFETCH || length > zfetch_array_rd_sz) dbuf_flags |= DB_RF_NOPREFETCH; rw_enter(&dn->dn_struct_rwlock, RW_READER); if (dn->dn_datablkshift) { int blkshift = dn->dn_datablkshift; nblks = (P2ROUNDUP(offset+length, 1ULL<> blkshift; } else { if (offset + length > dn->dn_datablksz) { zfs_panic_recover("zfs: accessing past end of object " "%llx/%llx (size=%u access=%llu+%llu)", (longlong_t)dn->dn_objset-> os_dsl_dataset->ds_object, (longlong_t)dn->dn_object, dn->dn_datablksz, (longlong_t)offset, (longlong_t)length); rw_exit(&dn->dn_struct_rwlock); return (SET_ERROR(EIO)); } nblks = 1; } - dbp = kmem_zalloc(sizeof (dmu_buf_t *) * nblks, - KM_PUSHPAGE | KM_NODEBUG); + dbp = kmem_zalloc(sizeof (dmu_buf_t *) * nblks, KM_PUSHPAGE); zio = zio_root(dn->dn_objset->os_spa, NULL, NULL, ZIO_FLAG_CANFAIL); blkid = dbuf_whichblock(dn, offset); for (i = 0; i < nblks; i++) { dmu_buf_impl_t *db = dbuf_hold(dn, blkid+i, tag); if (db == NULL) { rw_exit(&dn->dn_struct_rwlock); dmu_buf_rele_array(dbp, nblks, tag); zio_nowait(zio); return (SET_ERROR(EIO)); } /* initiate async i/o */ if (read) { (void) dbuf_read(db, zio, dbuf_flags); } dbp[i] = &db->db; } rw_exit(&dn->dn_struct_rwlock); /* wait for async i/o */ err = zio_wait(zio); if (err) { dmu_buf_rele_array(dbp, nblks, tag); return (err); } /* wait for other io to complete */ if (read) { for (i = 0; i < nblks; i++) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)dbp[i]; mutex_enter(&db->db_mtx); while (db->db_state == DB_READ || db->db_state == DB_FILL) cv_wait(&db->db_changed, &db->db_mtx); if (db->db_state == DB_UNCACHED) err = SET_ERROR(EIO); mutex_exit(&db->db_mtx); if (err) { dmu_buf_rele_array(dbp, nblks, tag); return (err); } } } *numbufsp = nblks; *dbpp = dbp; return (0); } static int dmu_buf_hold_array(objset_t *os, uint64_t object, uint64_t offset, uint64_t length, int read, void *tag, int *numbufsp, dmu_buf_t ***dbpp) { dnode_t *dn; int err; err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); err = dmu_buf_hold_array_by_dnode(dn, offset, length, read, tag, numbufsp, dbpp, DMU_READ_PREFETCH); dnode_rele(dn, FTAG); return (err); } int dmu_buf_hold_array_by_bonus(dmu_buf_t *db_fake, uint64_t offset, uint64_t length, int read, void *tag, int *numbufsp, dmu_buf_t ***dbpp) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; dnode_t *dn; int err; DB_DNODE_ENTER(db); dn = DB_DNODE(db); err = dmu_buf_hold_array_by_dnode(dn, offset, length, read, tag, numbufsp, dbpp, DMU_READ_PREFETCH); DB_DNODE_EXIT(db); return (err); } void dmu_buf_rele_array(dmu_buf_t **dbp_fake, int numbufs, void *tag) { int i; dmu_buf_impl_t **dbp = (dmu_buf_impl_t **)dbp_fake; if (numbufs == 0) return; for (i = 0; i < numbufs; i++) { if (dbp[i]) dbuf_rele(dbp[i], tag); } kmem_free(dbp, sizeof (dmu_buf_t *) * numbufs); } /* * Issue prefetch i/os for the given blocks. * * Note: The assumption is that we *know* these blocks will be needed * almost immediately. Therefore, the prefetch i/os will be issued at * ZIO_PRIORITY_SYNC_READ * * Note: indirect blocks and other metadata will be read synchronously, * causing this function to block if they are not already cached. */ void dmu_prefetch(objset_t *os, uint64_t object, uint64_t offset, uint64_t len) { dnode_t *dn; uint64_t blkid; int nblks, err; if (zfs_prefetch_disable) return; if (len == 0) { /* they're interested in the bonus buffer */ dn = DMU_META_DNODE(os); if (object == 0 || object >= DN_MAX_OBJECT) return; rw_enter(&dn->dn_struct_rwlock, RW_READER); blkid = dbuf_whichblock(dn, object * sizeof (dnode_phys_t)); dbuf_prefetch(dn, blkid, ZIO_PRIORITY_SYNC_READ); rw_exit(&dn->dn_struct_rwlock); return; } /* * XXX - Note, if the dnode for the requested object is not * already cached, we will do a *synchronous* read in the * dnode_hold() call. The same is true for any indirects. */ err = dnode_hold(os, object, FTAG, &dn); if (err != 0) return; rw_enter(&dn->dn_struct_rwlock, RW_READER); if (dn->dn_datablkshift) { int blkshift = dn->dn_datablkshift; nblks = (P2ROUNDUP(offset + len, 1 << blkshift) - P2ALIGN(offset, 1 << blkshift)) >> blkshift; } else { nblks = (offset < dn->dn_datablksz); } if (nblks != 0) { int i; blkid = dbuf_whichblock(dn, offset); for (i = 0; i < nblks; i++) dbuf_prefetch(dn, blkid + i, ZIO_PRIORITY_SYNC_READ); } rw_exit(&dn->dn_struct_rwlock); dnode_rele(dn, FTAG); } /* * Get the next "chunk" of file data to free. We traverse the file from * the end so that the file gets shorter over time (if we crashes in the * middle, this will leave us in a better state). We find allocated file * data by simply searching the allocated level 1 indirects. * * On input, *start should be the first offset that does not need to be * freed (e.g. "offset + length"). On return, *start will be the first * offset that should be freed. */ static int get_next_chunk(dnode_t *dn, uint64_t *start, uint64_t minimum) { uint64_t maxblks = DMU_MAX_ACCESS >> (dn->dn_indblkshift + 1); /* bytes of data covered by a level-1 indirect block */ uint64_t iblkrange = dn->dn_datablksz * EPB(dn->dn_indblkshift, SPA_BLKPTRSHIFT); uint64_t blks; ASSERT3U(minimum, <=, *start); if (*start - minimum <= iblkrange * maxblks) { *start = minimum; return (0); } ASSERT(ISP2(iblkrange)); for (blks = 0; *start > minimum && blks < maxblks; blks++) { int err; /* * dnode_next_offset(BACKWARDS) will find an allocated L1 * indirect block at or before the input offset. We must * decrement *start so that it is at the end of the region * to search. */ (*start)--; err = dnode_next_offset(dn, DNODE_FIND_BACKWARDS, start, 2, 1, 0); /* if there are no indirect blocks before start, we are done */ if (err == ESRCH) { *start = minimum; break; } else if (err != 0) { return (err); } /* set start to the beginning of this L1 indirect */ *start = P2ALIGN(*start, iblkrange); } if (*start < minimum) *start = minimum; return (0); } static int dmu_free_long_range_impl(objset_t *os, dnode_t *dn, uint64_t offset, uint64_t length) { uint64_t object_size = (dn->dn_maxblkid + 1) * dn->dn_datablksz; int err; if (offset >= object_size) return (0); if (length == DMU_OBJECT_END || offset + length > object_size) length = object_size - offset; while (length != 0) { uint64_t chunk_end, chunk_begin; dmu_tx_t *tx; chunk_end = chunk_begin = offset + length; /* move chunk_begin backwards to the beginning of this chunk */ err = get_next_chunk(dn, &chunk_begin, offset); if (err) return (err); ASSERT3U(chunk_begin, >=, offset); ASSERT3U(chunk_begin, <=, chunk_end); tx = dmu_tx_create(os); dmu_tx_hold_free(tx, dn->dn_object, chunk_begin, chunk_end - chunk_begin); err = dmu_tx_assign(tx, TXG_WAIT); if (err) { dmu_tx_abort(tx); return (err); } dnode_free_range(dn, chunk_begin, chunk_end - chunk_begin, tx); dmu_tx_commit(tx); length -= chunk_end - chunk_begin; } return (0); } int dmu_free_long_range(objset_t *os, uint64_t object, uint64_t offset, uint64_t length) { dnode_t *dn; int err; err = dnode_hold(os, object, FTAG, &dn); if (err != 0) return (err); err = dmu_free_long_range_impl(os, dn, offset, length); /* * It is important to zero out the maxblkid when freeing the entire * file, so that (a) subsequent calls to dmu_free_long_range_impl() * will take the fast path, and (b) dnode_reallocate() can verify * that the entire file has been freed. */ if (err == 0 && offset == 0 && length == DMU_OBJECT_END) dn->dn_maxblkid = 0; dnode_rele(dn, FTAG); return (err); } int dmu_free_long_object(objset_t *os, uint64_t object) { dmu_tx_t *tx; int err; err = dmu_free_long_range(os, object, 0, DMU_OBJECT_END); if (err != 0) return (err); tx = dmu_tx_create(os); dmu_tx_hold_bonus(tx, object); dmu_tx_hold_free(tx, object, 0, DMU_OBJECT_END); err = dmu_tx_assign(tx, TXG_WAIT); if (err == 0) { err = dmu_object_free(os, object, tx); dmu_tx_commit(tx); } else { dmu_tx_abort(tx); } return (err); } int dmu_free_range(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, dmu_tx_t *tx) { dnode_t *dn; int err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); ASSERT(offset < UINT64_MAX); ASSERT(size == -1ULL || size <= UINT64_MAX - offset); dnode_free_range(dn, offset, size, tx); dnode_rele(dn, FTAG); return (0); } int dmu_read(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, void *buf, uint32_t flags) { dnode_t *dn; dmu_buf_t **dbp; int numbufs, err; err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); /* * Deal with odd block sizes, where there can't be data past the first * block. If we ever do the tail block optimization, we will need to * handle that here as well. */ if (dn->dn_maxblkid == 0) { int newsz = offset > dn->dn_datablksz ? 0 : MIN(size, dn->dn_datablksz - offset); bzero((char *)buf + newsz, size - newsz); size = newsz; } while (size > 0) { uint64_t mylen = MIN(size, DMU_MAX_ACCESS / 2); int i; /* * NB: we could do this block-at-a-time, but it's nice * to be reading in parallel. */ err = dmu_buf_hold_array_by_dnode(dn, offset, mylen, TRUE, FTAG, &numbufs, &dbp, flags); if (err) break; for (i = 0; i < numbufs; i++) { int tocpy; int bufoff; dmu_buf_t *db = dbp[i]; ASSERT(size > 0); bufoff = offset - db->db_offset; tocpy = (int)MIN(db->db_size - bufoff, size); bcopy((char *)db->db_data + bufoff, buf, tocpy); offset += tocpy; size -= tocpy; buf = (char *)buf + tocpy; } dmu_buf_rele_array(dbp, numbufs, FTAG); } dnode_rele(dn, FTAG); return (err); } void dmu_write(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, const void *buf, dmu_tx_t *tx) { dmu_buf_t **dbp; int numbufs, i; if (size == 0) return; VERIFY0(dmu_buf_hold_array(os, object, offset, size, FALSE, FTAG, &numbufs, &dbp)); for (i = 0; i < numbufs; i++) { int tocpy; int bufoff; dmu_buf_t *db = dbp[i]; ASSERT(size > 0); bufoff = offset - db->db_offset; tocpy = (int)MIN(db->db_size - bufoff, size); ASSERT(i == 0 || i == numbufs-1 || tocpy == db->db_size); if (tocpy == db->db_size) dmu_buf_will_fill(db, tx); else dmu_buf_will_dirty(db, tx); (void) memcpy((char *)db->db_data + bufoff, buf, tocpy); if (tocpy == db->db_size) dmu_buf_fill_done(db, tx); offset += tocpy; size -= tocpy; buf = (char *)buf + tocpy; } dmu_buf_rele_array(dbp, numbufs, FTAG); } void dmu_prealloc(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, dmu_tx_t *tx) { dmu_buf_t **dbp; int numbufs, i; if (size == 0) return; VERIFY(0 == dmu_buf_hold_array(os, object, offset, size, FALSE, FTAG, &numbufs, &dbp)); for (i = 0; i < numbufs; i++) { dmu_buf_t *db = dbp[i]; dmu_buf_will_not_fill(db, tx); } dmu_buf_rele_array(dbp, numbufs, FTAG); } void dmu_write_embedded(objset_t *os, uint64_t object, uint64_t offset, void *data, uint8_t etype, uint8_t comp, int uncompressed_size, int compressed_size, int byteorder, dmu_tx_t *tx) { dmu_buf_t *db; ASSERT3U(etype, <, NUM_BP_EMBEDDED_TYPES); ASSERT3U(comp, <, ZIO_COMPRESS_FUNCTIONS); VERIFY0(dmu_buf_hold_noread(os, object, offset, FTAG, &db)); dmu_buf_write_embedded(db, data, (bp_embedded_type_t)etype, (enum zio_compress)comp, uncompressed_size, compressed_size, byteorder, tx); dmu_buf_rele(db, FTAG); } /* * DMU support for xuio */ kstat_t *xuio_ksp = NULL; typedef struct xuio_stats { /* loaned yet not returned arc_buf */ kstat_named_t xuiostat_onloan_rbuf; kstat_named_t xuiostat_onloan_wbuf; /* whether a copy is made when loaning out a read buffer */ kstat_named_t xuiostat_rbuf_copied; kstat_named_t xuiostat_rbuf_nocopy; /* whether a copy is made when assigning a write buffer */ kstat_named_t xuiostat_wbuf_copied; kstat_named_t xuiostat_wbuf_nocopy; } xuio_stats_t; static xuio_stats_t xuio_stats = { { "onloan_read_buf", KSTAT_DATA_UINT64 }, { "onloan_write_buf", KSTAT_DATA_UINT64 }, { "read_buf_copied", KSTAT_DATA_UINT64 }, { "read_buf_nocopy", KSTAT_DATA_UINT64 }, { "write_buf_copied", KSTAT_DATA_UINT64 }, { "write_buf_nocopy", KSTAT_DATA_UINT64 } }; #define XUIOSTAT_INCR(stat, val) \ atomic_add_64(&xuio_stats.stat.value.ui64, (val)) #define XUIOSTAT_BUMP(stat) XUIOSTAT_INCR(stat, 1) int dmu_xuio_init(xuio_t *xuio, int nblk) { dmu_xuio_t *priv; uio_t *uio = &xuio->xu_uio; uio->uio_iovcnt = nblk; uio->uio_iov = kmem_zalloc(nblk * sizeof (iovec_t), KM_PUSHPAGE); priv = kmem_zalloc(sizeof (dmu_xuio_t), KM_PUSHPAGE); priv->cnt = nblk; priv->bufs = kmem_zalloc(nblk * sizeof (arc_buf_t *), KM_PUSHPAGE); priv->iovp = uio->uio_iov; XUIO_XUZC_PRIV(xuio) = priv; if (XUIO_XUZC_RW(xuio) == UIO_READ) XUIOSTAT_INCR(xuiostat_onloan_rbuf, nblk); else XUIOSTAT_INCR(xuiostat_onloan_wbuf, nblk); return (0); } void dmu_xuio_fini(xuio_t *xuio) { dmu_xuio_t *priv = XUIO_XUZC_PRIV(xuio); int nblk = priv->cnt; kmem_free(priv->iovp, nblk * sizeof (iovec_t)); kmem_free(priv->bufs, nblk * sizeof (arc_buf_t *)); kmem_free(priv, sizeof (dmu_xuio_t)); if (XUIO_XUZC_RW(xuio) == UIO_READ) XUIOSTAT_INCR(xuiostat_onloan_rbuf, -nblk); else XUIOSTAT_INCR(xuiostat_onloan_wbuf, -nblk); } /* * Initialize iov[priv->next] and priv->bufs[priv->next] with { off, n, abuf } * and increase priv->next by 1. */ int dmu_xuio_add(xuio_t *xuio, arc_buf_t *abuf, offset_t off, size_t n) { struct iovec *iov; uio_t *uio = &xuio->xu_uio; dmu_xuio_t *priv = XUIO_XUZC_PRIV(xuio); int i = priv->next++; ASSERT(i < priv->cnt); ASSERT(off + n <= arc_buf_size(abuf)); iov = uio->uio_iov + i; iov->iov_base = (char *)abuf->b_data + off; iov->iov_len = n; priv->bufs[i] = abuf; return (0); } int dmu_xuio_cnt(xuio_t *xuio) { dmu_xuio_t *priv = XUIO_XUZC_PRIV(xuio); return (priv->cnt); } arc_buf_t * dmu_xuio_arcbuf(xuio_t *xuio, int i) { dmu_xuio_t *priv = XUIO_XUZC_PRIV(xuio); ASSERT(i < priv->cnt); return (priv->bufs[i]); } void dmu_xuio_clear(xuio_t *xuio, int i) { dmu_xuio_t *priv = XUIO_XUZC_PRIV(xuio); ASSERT(i < priv->cnt); priv->bufs[i] = NULL; } static void xuio_stat_init(void) { xuio_ksp = kstat_create("zfs", 0, "xuio_stats", "misc", KSTAT_TYPE_NAMED, sizeof (xuio_stats) / sizeof (kstat_named_t), KSTAT_FLAG_VIRTUAL); if (xuio_ksp != NULL) { xuio_ksp->ks_data = &xuio_stats; kstat_install(xuio_ksp); } } static void xuio_stat_fini(void) { if (xuio_ksp != NULL) { kstat_delete(xuio_ksp); xuio_ksp = NULL; } } void xuio_stat_wbuf_copied() { XUIOSTAT_BUMP(xuiostat_wbuf_copied); } void xuio_stat_wbuf_nocopy() { XUIOSTAT_BUMP(xuiostat_wbuf_nocopy); } #ifdef _KERNEL /* * Copy up to size bytes between arg_buf and req based on the data direction * described by the req. If an entire req's data cannot be transfered in one * pass, you should pass in @req_offset to indicate where to continue. The * return value is the number of bytes successfully copied to arg_buf. */ static int dmu_req_copy(void *arg_buf, int size, struct request *req, size_t req_offset) { struct bio_vec bv, *bvp; struct req_iterator iter; char *bv_buf; int tocpy, bv_len, bv_offset; int offset = 0; rq_for_each_segment4(bv, bvp, req, iter) { /* * Fully consumed the passed arg_buf. We use goto here because * rq_for_each_segment is a double loop */ ASSERT3S(offset, <=, size); if (size == offset) goto out; /* Skip already copied bv */ if (req_offset >= bv.bv_len) { req_offset -= bv.bv_len; continue; } bv_len = bv.bv_len - req_offset; bv_offset = bv.bv_offset + req_offset; req_offset = 0; tocpy = MIN(bv_len, size - offset); ASSERT3S(tocpy, >=, 0); bv_buf = page_address(bv.bv_page) + bv_offset; ASSERT3P(bv_buf, !=, NULL); if (rq_data_dir(req) == WRITE) memcpy(arg_buf + offset, bv_buf, tocpy); else memcpy(bv_buf, arg_buf + offset, tocpy); offset += tocpy; } out: return (offset); } int dmu_read_req(objset_t *os, uint64_t object, struct request *req) { uint64_t size = blk_rq_bytes(req); uint64_t offset = blk_rq_pos(req) << 9; dmu_buf_t **dbp; int numbufs, i, err; size_t req_offset; /* * NB: we could do this block-at-a-time, but it's nice * to be reading in parallel. */ err = dmu_buf_hold_array(os, object, offset, size, TRUE, FTAG, &numbufs, &dbp); if (err) return (err); req_offset = 0; for (i = 0; i < numbufs; i++) { int tocpy, didcpy, bufoff; dmu_buf_t *db = dbp[i]; bufoff = offset - db->db_offset; ASSERT3S(bufoff, >=, 0); tocpy = (int)MIN(db->db_size - bufoff, size); if (tocpy == 0) break; didcpy = dmu_req_copy(db->db_data + bufoff, tocpy, req, req_offset); if (didcpy < tocpy) err = EIO; if (err) break; size -= tocpy; offset += didcpy; req_offset += didcpy; err = 0; } dmu_buf_rele_array(dbp, numbufs, FTAG); return (err); } int dmu_write_req(objset_t *os, uint64_t object, struct request *req, dmu_tx_t *tx) { uint64_t size = blk_rq_bytes(req); uint64_t offset = blk_rq_pos(req) << 9; dmu_buf_t **dbp; int numbufs, i, err; size_t req_offset; if (size == 0) return (0); err = dmu_buf_hold_array(os, object, offset, size, FALSE, FTAG, &numbufs, &dbp); if (err) return (err); req_offset = 0; for (i = 0; i < numbufs; i++) { int tocpy, didcpy, bufoff; dmu_buf_t *db = dbp[i]; bufoff = offset - db->db_offset; ASSERT3S(bufoff, >=, 0); tocpy = (int)MIN(db->db_size - bufoff, size); if (tocpy == 0) break; ASSERT(i == 0 || i == numbufs-1 || tocpy == db->db_size); if (tocpy == db->db_size) dmu_buf_will_fill(db, tx); else dmu_buf_will_dirty(db, tx); didcpy = dmu_req_copy(db->db_data + bufoff, tocpy, req, req_offset); if (tocpy == db->db_size) dmu_buf_fill_done(db, tx); if (didcpy < tocpy) err = EIO; if (err) break; size -= tocpy; offset += didcpy; req_offset += didcpy; err = 0; } dmu_buf_rele_array(dbp, numbufs, FTAG); return (err); } int dmu_read_uio(objset_t *os, uint64_t object, uio_t *uio, uint64_t size) { dmu_buf_t **dbp; int numbufs, i, err; xuio_t *xuio = NULL; /* * NB: we could do this block-at-a-time, but it's nice * to be reading in parallel. */ err = dmu_buf_hold_array(os, object, uio->uio_loffset, size, TRUE, FTAG, &numbufs, &dbp); if (err) return (err); for (i = 0; i < numbufs; i++) { int tocpy; int bufoff; dmu_buf_t *db = dbp[i]; ASSERT(size > 0); bufoff = uio->uio_loffset - db->db_offset; tocpy = (int)MIN(db->db_size - bufoff, size); if (xuio) { dmu_buf_impl_t *dbi = (dmu_buf_impl_t *)db; arc_buf_t *dbuf_abuf = dbi->db_buf; arc_buf_t *abuf = dbuf_loan_arcbuf(dbi); err = dmu_xuio_add(xuio, abuf, bufoff, tocpy); if (!err) { uio->uio_resid -= tocpy; uio->uio_loffset += tocpy; } if (abuf == dbuf_abuf) XUIOSTAT_BUMP(xuiostat_rbuf_nocopy); else XUIOSTAT_BUMP(xuiostat_rbuf_copied); } else { err = uiomove((char *)db->db_data + bufoff, tocpy, UIO_READ, uio); } if (err) break; size -= tocpy; } dmu_buf_rele_array(dbp, numbufs, FTAG); return (err); } static int dmu_write_uio_dnode(dnode_t *dn, uio_t *uio, uint64_t size, dmu_tx_t *tx) { dmu_buf_t **dbp; int numbufs; int err = 0; int i; err = dmu_buf_hold_array_by_dnode(dn, uio->uio_loffset, size, FALSE, FTAG, &numbufs, &dbp, DMU_READ_PREFETCH); if (err) return (err); for (i = 0; i < numbufs; i++) { int tocpy; int bufoff; dmu_buf_t *db = dbp[i]; ASSERT(size > 0); bufoff = uio->uio_loffset - db->db_offset; tocpy = (int)MIN(db->db_size - bufoff, size); ASSERT(i == 0 || i == numbufs-1 || tocpy == db->db_size); if (tocpy == db->db_size) dmu_buf_will_fill(db, tx); else dmu_buf_will_dirty(db, tx); /* * XXX uiomove could block forever (eg.nfs-backed * pages). There needs to be a uiolockdown() function * to lock the pages in memory, so that uiomove won't * block. */ err = uiomove((char *)db->db_data + bufoff, tocpy, UIO_WRITE, uio); if (tocpy == db->db_size) dmu_buf_fill_done(db, tx); if (err) break; size -= tocpy; } dmu_buf_rele_array(dbp, numbufs, FTAG); return (err); } int dmu_write_uio_dbuf(dmu_buf_t *zdb, uio_t *uio, uint64_t size, dmu_tx_t *tx) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)zdb; dnode_t *dn; int err; if (size == 0) return (0); DB_DNODE_ENTER(db); dn = DB_DNODE(db); err = dmu_write_uio_dnode(dn, uio, size, tx); DB_DNODE_EXIT(db); return (err); } int dmu_write_uio(objset_t *os, uint64_t object, uio_t *uio, uint64_t size, dmu_tx_t *tx) { dnode_t *dn; int err; if (size == 0) return (0); err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); err = dmu_write_uio_dnode(dn, uio, size, tx); dnode_rele(dn, FTAG); return (err); } #endif /* _KERNEL */ /* * Allocate a loaned anonymous arc buffer. */ arc_buf_t * dmu_request_arcbuf(dmu_buf_t *handle, int size) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)handle; return (arc_loan_buf(db->db_objset->os_spa, size)); } /* * Free a loaned arc buffer. */ void dmu_return_arcbuf(arc_buf_t *buf) { arc_return_buf(buf, FTAG); VERIFY(arc_buf_remove_ref(buf, FTAG)); } /* * When possible directly assign passed loaned arc buffer to a dbuf. * If this is not possible copy the contents of passed arc buf via * dmu_write(). */ void dmu_assign_arcbuf(dmu_buf_t *handle, uint64_t offset, arc_buf_t *buf, dmu_tx_t *tx) { dmu_buf_impl_t *dbuf = (dmu_buf_impl_t *)handle; dnode_t *dn; dmu_buf_impl_t *db; uint32_t blksz = (uint32_t)arc_buf_size(buf); uint64_t blkid; DB_DNODE_ENTER(dbuf); dn = DB_DNODE(dbuf); rw_enter(&dn->dn_struct_rwlock, RW_READER); blkid = dbuf_whichblock(dn, offset); VERIFY((db = dbuf_hold(dn, blkid, FTAG)) != NULL); rw_exit(&dn->dn_struct_rwlock); DB_DNODE_EXIT(dbuf); /* * We can only assign if the offset is aligned, the arc buf is the * same size as the dbuf, and the dbuf is not metadata. It * can't be metadata because the loaned arc buf comes from the * user-data kmem area. */ if (offset == db->db.db_offset && blksz == db->db.db_size && DBUF_GET_BUFC_TYPE(db) == ARC_BUFC_DATA) { dbuf_assign_arcbuf(db, buf, tx); dbuf_rele(db, FTAG); } else { objset_t *os; uint64_t object; DB_DNODE_ENTER(dbuf); dn = DB_DNODE(dbuf); os = dn->dn_objset; object = dn->dn_object; DB_DNODE_EXIT(dbuf); dbuf_rele(db, FTAG); dmu_write(os, object, offset, blksz, buf->b_data, tx); dmu_return_arcbuf(buf); XUIOSTAT_BUMP(xuiostat_wbuf_copied); } } typedef struct { dbuf_dirty_record_t *dsa_dr; dmu_sync_cb_t *dsa_done; zgd_t *dsa_zgd; dmu_tx_t *dsa_tx; } dmu_sync_arg_t; /* ARGSUSED */ static void dmu_sync_ready(zio_t *zio, arc_buf_t *buf, void *varg) { dmu_sync_arg_t *dsa = varg; dmu_buf_t *db = dsa->dsa_zgd->zgd_db; blkptr_t *bp = zio->io_bp; if (zio->io_error == 0) { if (BP_IS_HOLE(bp)) { /* * A block of zeros may compress to a hole, but the * block size still needs to be known for replay. */ BP_SET_LSIZE(bp, db->db_size); } else if (!BP_IS_EMBEDDED(bp)) { ASSERT(BP_GET_LEVEL(bp) == 0); bp->blk_fill = 1; } } } static void dmu_sync_late_arrival_ready(zio_t *zio) { dmu_sync_ready(zio, NULL, zio->io_private); } /* ARGSUSED */ static void dmu_sync_done(zio_t *zio, arc_buf_t *buf, void *varg) { dmu_sync_arg_t *dsa = varg; dbuf_dirty_record_t *dr = dsa->dsa_dr; dmu_buf_impl_t *db = dr->dr_dbuf; mutex_enter(&db->db_mtx); ASSERT(dr->dt.dl.dr_override_state == DR_IN_DMU_SYNC); if (zio->io_error == 0) { dr->dt.dl.dr_nopwrite = !!(zio->io_flags & ZIO_FLAG_NOPWRITE); if (dr->dt.dl.dr_nopwrite) { ASSERTV(blkptr_t *bp = zio->io_bp); ASSERTV(blkptr_t *bp_orig = &zio->io_bp_orig); ASSERTV(uint8_t chksum = BP_GET_CHECKSUM(bp_orig)); ASSERT(BP_EQUAL(bp, bp_orig)); ASSERT(zio->io_prop.zp_compress != ZIO_COMPRESS_OFF); ASSERT(zio_checksum_table[chksum].ci_dedup); } dr->dt.dl.dr_overridden_by = *zio->io_bp; dr->dt.dl.dr_override_state = DR_OVERRIDDEN; dr->dt.dl.dr_copies = zio->io_prop.zp_copies; if (BP_IS_HOLE(&dr->dt.dl.dr_overridden_by)) BP_ZERO(&dr->dt.dl.dr_overridden_by); } else { dr->dt.dl.dr_override_state = DR_NOT_OVERRIDDEN; } cv_broadcast(&db->db_changed); mutex_exit(&db->db_mtx); dsa->dsa_done(dsa->dsa_zgd, zio->io_error); kmem_free(dsa, sizeof (*dsa)); } static void dmu_sync_late_arrival_done(zio_t *zio) { blkptr_t *bp = zio->io_bp; dmu_sync_arg_t *dsa = zio->io_private; ASSERTV(blkptr_t *bp_orig = &zio->io_bp_orig); if (zio->io_error == 0 && !BP_IS_HOLE(bp)) { /* * If we didn't allocate a new block (i.e. ZIO_FLAG_NOPWRITE) * then there is nothing to do here. Otherwise, free the * newly allocated block in this txg. */ if (zio->io_flags & ZIO_FLAG_NOPWRITE) { ASSERT(BP_EQUAL(bp, bp_orig)); } else { ASSERT(BP_IS_HOLE(bp_orig) || !BP_EQUAL(bp, bp_orig)); ASSERT(zio->io_bp->blk_birth == zio->io_txg); ASSERT(zio->io_txg > spa_syncing_txg(zio->io_spa)); zio_free(zio->io_spa, zio->io_txg, zio->io_bp); } } dmu_tx_commit(dsa->dsa_tx); dsa->dsa_done(dsa->dsa_zgd, zio->io_error); kmem_free(dsa, sizeof (*dsa)); } static int dmu_sync_late_arrival(zio_t *pio, objset_t *os, dmu_sync_cb_t *done, zgd_t *zgd, zio_prop_t *zp, zbookmark_phys_t *zb) { dmu_sync_arg_t *dsa; dmu_tx_t *tx; tx = dmu_tx_create(os); dmu_tx_hold_space(tx, zgd->zgd_db->db_size); if (dmu_tx_assign(tx, TXG_WAIT) != 0) { dmu_tx_abort(tx); /* Make zl_get_data do txg_waited_synced() */ return (SET_ERROR(EIO)); } dsa = kmem_alloc(sizeof (dmu_sync_arg_t), KM_PUSHPAGE); dsa->dsa_dr = NULL; dsa->dsa_done = done; dsa->dsa_zgd = zgd; dsa->dsa_tx = tx; zio_nowait(zio_write(pio, os->os_spa, dmu_tx_get_txg(tx), zgd->zgd_bp, zgd->zgd_db->db_data, zgd->zgd_db->db_size, zp, dmu_sync_late_arrival_ready, NULL, dmu_sync_late_arrival_done, dsa, ZIO_PRIORITY_SYNC_WRITE, ZIO_FLAG_CANFAIL|ZIO_FLAG_FASTWRITE, zb)); return (0); } /* * Intent log support: sync the block associated with db to disk. * N.B. and XXX: the caller is responsible for making sure that the * data isn't changing while dmu_sync() is writing it. * * Return values: * * EEXIST: this txg has already been synced, so there's nothing to do. * The caller should not log the write. * * ENOENT: the block was dbuf_free_range()'d, so there's nothing to do. * The caller should not log the write. * * EALREADY: this block is already in the process of being synced. * The caller should track its progress (somehow). * * EIO: could not do the I/O. * The caller should do a txg_wait_synced(). * * 0: the I/O has been initiated. * The caller should log this blkptr in the done callback. * It is possible that the I/O will fail, in which case * the error will be reported to the done callback and * propagated to pio from zio_done(). */ int dmu_sync(zio_t *pio, uint64_t txg, dmu_sync_cb_t *done, zgd_t *zgd) { blkptr_t *bp = zgd->zgd_bp; dmu_buf_impl_t *db = (dmu_buf_impl_t *)zgd->zgd_db; objset_t *os = db->db_objset; dsl_dataset_t *ds = os->os_dsl_dataset; dbuf_dirty_record_t *dr; dmu_sync_arg_t *dsa; zbookmark_phys_t zb; zio_prop_t zp; dnode_t *dn; ASSERT(pio != NULL); ASSERT(txg != 0); SET_BOOKMARK(&zb, ds->ds_object, db->db.db_object, db->db_level, db->db_blkid); DB_DNODE_ENTER(db); dn = DB_DNODE(db); dmu_write_policy(os, dn, db->db_level, WP_DMU_SYNC, &zp); DB_DNODE_EXIT(db); /* * If we're frozen (running ziltest), we always need to generate a bp. */ if (txg > spa_freeze_txg(os->os_spa)) return (dmu_sync_late_arrival(pio, os, done, zgd, &zp, &zb)); /* * Grabbing db_mtx now provides a barrier between dbuf_sync_leaf() * and us. If we determine that this txg is not yet syncing, * but it begins to sync a moment later, that's OK because the * sync thread will block in dbuf_sync_leaf() until we drop db_mtx. */ mutex_enter(&db->db_mtx); if (txg <= spa_last_synced_txg(os->os_spa)) { /* * This txg has already synced. There's nothing to do. */ mutex_exit(&db->db_mtx); return (SET_ERROR(EEXIST)); } if (txg <= spa_syncing_txg(os->os_spa)) { /* * This txg is currently syncing, so we can't mess with * the dirty record anymore; just write a new log block. */ mutex_exit(&db->db_mtx); return (dmu_sync_late_arrival(pio, os, done, zgd, &zp, &zb)); } dr = db->db_last_dirty; while (dr && dr->dr_txg != txg) dr = dr->dr_next; if (dr == NULL) { /* * There's no dr for this dbuf, so it must have been freed. * There's no need to log writes to freed blocks, so we're done. */ mutex_exit(&db->db_mtx); return (SET_ERROR(ENOENT)); } ASSERT(dr->dr_next == NULL || dr->dr_next->dr_txg < txg); /* * Assume the on-disk data is X, the current syncing data is Y, * and the current in-memory data is Z (currently in dmu_sync). * X and Z are identical but Y is has been modified. Normally, * when X and Z are the same we will perform a nopwrite but if Y * is different we must disable nopwrite since the resulting write * of Y to disk can free the block containing X. If we allowed a * nopwrite to occur the block pointing to Z would reference a freed * block. Since this is a rare case we simplify this by disabling * nopwrite if the current dmu_sync-ing dbuf has been modified in * a previous transaction. */ if (dr->dr_next) zp.zp_nopwrite = B_FALSE; ASSERT(dr->dr_txg == txg); if (dr->dt.dl.dr_override_state == DR_IN_DMU_SYNC || dr->dt.dl.dr_override_state == DR_OVERRIDDEN) { /* * We have already issued a sync write for this buffer, * or this buffer has already been synced. It could not * have been dirtied since, or we would have cleared the state. */ mutex_exit(&db->db_mtx); return (SET_ERROR(EALREADY)); } ASSERT(dr->dt.dl.dr_override_state == DR_NOT_OVERRIDDEN); dr->dt.dl.dr_override_state = DR_IN_DMU_SYNC; mutex_exit(&db->db_mtx); dsa = kmem_alloc(sizeof (dmu_sync_arg_t), KM_PUSHPAGE); dsa->dsa_dr = dr; dsa->dsa_done = done; dsa->dsa_zgd = zgd; dsa->dsa_tx = NULL; zio_nowait(arc_write(pio, os->os_spa, txg, bp, dr->dt.dl.dr_data, DBUF_IS_L2CACHEABLE(db), DBUF_IS_L2COMPRESSIBLE(db), &zp, dmu_sync_ready, NULL, dmu_sync_done, dsa, ZIO_PRIORITY_SYNC_WRITE, ZIO_FLAG_CANFAIL, &zb)); return (0); } int dmu_object_set_blocksize(objset_t *os, uint64_t object, uint64_t size, int ibs, dmu_tx_t *tx) { dnode_t *dn; int err; err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); err = dnode_set_blksz(dn, size, ibs, tx); dnode_rele(dn, FTAG); return (err); } void dmu_object_set_checksum(objset_t *os, uint64_t object, uint8_t checksum, dmu_tx_t *tx) { dnode_t *dn; /* * Send streams include each object's checksum function. This * check ensures that the receiving system can understand the * checksum function transmitted. */ ASSERT3U(checksum, <, ZIO_CHECKSUM_LEGACY_FUNCTIONS); VERIFY0(dnode_hold(os, object, FTAG, &dn)); ASSERT3U(checksum, <, ZIO_CHECKSUM_FUNCTIONS); dn->dn_checksum = checksum; dnode_setdirty(dn, tx); dnode_rele(dn, FTAG); } void dmu_object_set_compress(objset_t *os, uint64_t object, uint8_t compress, dmu_tx_t *tx) { dnode_t *dn; /* * Send streams include each object's compression function. This * check ensures that the receiving system can understand the * compression function transmitted. */ ASSERT3U(compress, <, ZIO_COMPRESS_LEGACY_FUNCTIONS); VERIFY0(dnode_hold(os, object, FTAG, &dn)); dn->dn_compress = compress; dnode_setdirty(dn, tx); dnode_rele(dn, FTAG); } int zfs_mdcomp_disable = 0; /* * When the "redundant_metadata" property is set to "most", only indirect * blocks of this level and higher will have an additional ditto block. */ int zfs_redundant_metadata_most_ditto_level = 2; void dmu_write_policy(objset_t *os, dnode_t *dn, int level, int wp, zio_prop_t *zp) { dmu_object_type_t type = dn ? dn->dn_type : DMU_OT_OBJSET; boolean_t ismd = (level > 0 || DMU_OT_IS_METADATA(type) || (wp & WP_SPILL)); enum zio_checksum checksum = os->os_checksum; enum zio_compress compress = os->os_compress; enum zio_checksum dedup_checksum = os->os_dedup_checksum; boolean_t dedup = B_FALSE; boolean_t nopwrite = B_FALSE; boolean_t dedup_verify = os->os_dedup_verify; int copies = os->os_copies; /* * We maintain different write policies for each of the following * types of data: * 1. metadata * 2. preallocated blocks (i.e. level-0 blocks of a dump device) * 3. all other level 0 blocks */ if (ismd) { /* * XXX -- we should design a compression algorithm * that specializes in arrays of bps. */ boolean_t lz4_ac = spa_feature_is_active(os->os_spa, SPA_FEATURE_LZ4_COMPRESS); if (zfs_mdcomp_disable) { compress = ZIO_COMPRESS_EMPTY; } else if (lz4_ac) { compress = ZIO_COMPRESS_LZ4; } else { compress = ZIO_COMPRESS_LZJB; } /* * Metadata always gets checksummed. If the data * checksum is multi-bit correctable, and it's not a * ZBT-style checksum, then it's suitable for metadata * as well. Otherwise, the metadata checksum defaults * to fletcher4. */ if (zio_checksum_table[checksum].ci_correctable < 1 || zio_checksum_table[checksum].ci_eck) checksum = ZIO_CHECKSUM_FLETCHER_4; if (os->os_redundant_metadata == ZFS_REDUNDANT_METADATA_ALL || (os->os_redundant_metadata == ZFS_REDUNDANT_METADATA_MOST && (level >= zfs_redundant_metadata_most_ditto_level || DMU_OT_IS_METADATA(type) || (wp & WP_SPILL)))) copies++; } else if (wp & WP_NOFILL) { ASSERT(level == 0); /* * If we're writing preallocated blocks, we aren't actually * writing them so don't set any policy properties. These * blocks are currently only used by an external subsystem * outside of zfs (i.e. dump) and not written by the zio * pipeline. */ compress = ZIO_COMPRESS_OFF; checksum = ZIO_CHECKSUM_OFF; } else { compress = zio_compress_select(dn->dn_compress, compress); checksum = (dedup_checksum == ZIO_CHECKSUM_OFF) ? zio_checksum_select(dn->dn_checksum, checksum) : dedup_checksum; /* * Determine dedup setting. If we are in dmu_sync(), * we won't actually dedup now because that's all * done in syncing context; but we do want to use the * dedup checkum. If the checksum is not strong * enough to ensure unique signatures, force * dedup_verify. */ if (dedup_checksum != ZIO_CHECKSUM_OFF) { dedup = (wp & WP_DMU_SYNC) ? B_FALSE : B_TRUE; if (!zio_checksum_table[checksum].ci_dedup) dedup_verify = B_TRUE; } /* * Enable nopwrite if we have a cryptographically secure * checksum that has no known collisions (i.e. SHA-256) * and compression is enabled. We don't enable nopwrite if * dedup is enabled as the two features are mutually exclusive. */ nopwrite = (!dedup && zio_checksum_table[checksum].ci_dedup && compress != ZIO_COMPRESS_OFF && zfs_nopwrite_enabled); } zp->zp_checksum = checksum; zp->zp_compress = compress; zp->zp_type = (wp & WP_SPILL) ? dn->dn_bonustype : type; zp->zp_level = level; zp->zp_copies = MIN(copies, spa_max_replication(os->os_spa)); zp->zp_dedup = dedup; zp->zp_dedup_verify = dedup && dedup_verify; zp->zp_nopwrite = nopwrite; } int dmu_offset_next(objset_t *os, uint64_t object, boolean_t hole, uint64_t *off) { dnode_t *dn; int i, err; err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); /* * Sync any current changes before * we go trundling through the block pointers. */ for (i = 0; i < TXG_SIZE; i++) { if (list_link_active(&dn->dn_dirty_link[i])) break; } if (i != TXG_SIZE) { dnode_rele(dn, FTAG); txg_wait_synced(dmu_objset_pool(os), 0); err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); } err = dnode_next_offset(dn, (hole ? DNODE_FIND_HOLE : 0), off, 1, 1, 0); dnode_rele(dn, FTAG); return (err); } void __dmu_object_info_from_dnode(dnode_t *dn, dmu_object_info_t *doi) { dnode_phys_t *dnp = dn->dn_phys; int i; doi->doi_data_block_size = dn->dn_datablksz; doi->doi_metadata_block_size = dn->dn_indblkshift ? 1ULL << dn->dn_indblkshift : 0; doi->doi_type = dn->dn_type; doi->doi_bonus_type = dn->dn_bonustype; doi->doi_bonus_size = dn->dn_bonuslen; doi->doi_indirection = dn->dn_nlevels; doi->doi_checksum = dn->dn_checksum; doi->doi_compress = dn->dn_compress; doi->doi_nblkptr = dn->dn_nblkptr; doi->doi_physical_blocks_512 = (DN_USED_BYTES(dnp) + 256) >> 9; doi->doi_max_offset = (dn->dn_maxblkid + 1) * dn->dn_datablksz; doi->doi_fill_count = 0; for (i = 0; i < dnp->dn_nblkptr; i++) doi->doi_fill_count += BP_GET_FILL(&dnp->dn_blkptr[i]); } void dmu_object_info_from_dnode(dnode_t *dn, dmu_object_info_t *doi) { rw_enter(&dn->dn_struct_rwlock, RW_READER); mutex_enter(&dn->dn_mtx); __dmu_object_info_from_dnode(dn, doi); mutex_exit(&dn->dn_mtx); rw_exit(&dn->dn_struct_rwlock); } /* * Get information on a DMU object. * If doi is NULL, just indicates whether the object exists. */ int dmu_object_info(objset_t *os, uint64_t object, dmu_object_info_t *doi) { dnode_t *dn; int err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); if (doi != NULL) dmu_object_info_from_dnode(dn, doi); dnode_rele(dn, FTAG); return (0); } /* * As above, but faster; can be used when you have a held dbuf in hand. */ void dmu_object_info_from_db(dmu_buf_t *db_fake, dmu_object_info_t *doi) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; DB_DNODE_ENTER(db); dmu_object_info_from_dnode(DB_DNODE(db), doi); DB_DNODE_EXIT(db); } /* * Faster still when you only care about the size. * This is specifically optimized for zfs_getattr(). */ void dmu_object_size_from_db(dmu_buf_t *db_fake, uint32_t *blksize, u_longlong_t *nblk512) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; dnode_t *dn; DB_DNODE_ENTER(db); dn = DB_DNODE(db); *blksize = dn->dn_datablksz; /* add 1 for dnode space */ *nblk512 = ((DN_USED_BYTES(dn->dn_phys) + SPA_MINBLOCKSIZE/2) >> SPA_MINBLOCKSHIFT) + 1; DB_DNODE_EXIT(db); } void byteswap_uint64_array(void *vbuf, size_t size) { uint64_t *buf = vbuf; size_t count = size >> 3; int i; ASSERT((size & 7) == 0); for (i = 0; i < count; i++) buf[i] = BSWAP_64(buf[i]); } void byteswap_uint32_array(void *vbuf, size_t size) { uint32_t *buf = vbuf; size_t count = size >> 2; int i; ASSERT((size & 3) == 0); for (i = 0; i < count; i++) buf[i] = BSWAP_32(buf[i]); } void byteswap_uint16_array(void *vbuf, size_t size) { uint16_t *buf = vbuf; size_t count = size >> 1; int i; ASSERT((size & 1) == 0); for (i = 0; i < count; i++) buf[i] = BSWAP_16(buf[i]); } /* ARGSUSED */ void byteswap_uint8_array(void *vbuf, size_t size) { } void dmu_init(void) { zfs_dbgmsg_init(); sa_cache_init(); xuio_stat_init(); dmu_objset_init(); dnode_init(); dbuf_init(); zfetch_init(); dmu_tx_init(); l2arc_init(); arc_init(); } void dmu_fini(void) { arc_fini(); /* arc depends on l2arc, so arc must go first */ l2arc_fini(); dmu_tx_fini(); zfetch_fini(); dbuf_fini(); dnode_fini(); dmu_objset_fini(); xuio_stat_fini(); sa_cache_fini(); zfs_dbgmsg_fini(); } #if defined(_KERNEL) && defined(HAVE_SPL) EXPORT_SYMBOL(dmu_bonus_hold); EXPORT_SYMBOL(dmu_buf_hold_array_by_bonus); EXPORT_SYMBOL(dmu_buf_rele_array); EXPORT_SYMBOL(dmu_prefetch); EXPORT_SYMBOL(dmu_free_range); EXPORT_SYMBOL(dmu_free_long_range); EXPORT_SYMBOL(dmu_free_long_object); EXPORT_SYMBOL(dmu_read); EXPORT_SYMBOL(dmu_write); EXPORT_SYMBOL(dmu_prealloc); EXPORT_SYMBOL(dmu_object_info); EXPORT_SYMBOL(dmu_object_info_from_dnode); EXPORT_SYMBOL(dmu_object_info_from_db); EXPORT_SYMBOL(dmu_object_size_from_db); EXPORT_SYMBOL(dmu_object_set_blocksize); EXPORT_SYMBOL(dmu_object_set_checksum); EXPORT_SYMBOL(dmu_object_set_compress); EXPORT_SYMBOL(dmu_write_policy); EXPORT_SYMBOL(dmu_sync); EXPORT_SYMBOL(dmu_request_arcbuf); EXPORT_SYMBOL(dmu_return_arcbuf); EXPORT_SYMBOL(dmu_assign_arcbuf); EXPORT_SYMBOL(dmu_buf_hold); EXPORT_SYMBOL(dmu_ot); module_param(zfs_mdcomp_disable, int, 0644); MODULE_PARM_DESC(zfs_mdcomp_disable, "Disable meta data compression"); module_param(zfs_nopwrite_enabled, int, 0644); MODULE_PARM_DESC(zfs_nopwrite_enabled, "Enable NOP writes"); #endif diff --git a/module/zfs/dnode.c b/module/zfs/dnode.c index 0a87aa9b15d8..e5c212cc0bda 100644 --- a/module/zfs/dnode.c +++ b/module/zfs/dnode.c @@ -1,1923 +1,1922 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2012, 2014 by Delphix. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include static kmem_cache_t *dnode_cache; /* * Define DNODE_STATS to turn on statistic gathering. By default, it is only * turned on when DEBUG is also defined. */ #ifdef DEBUG #define DNODE_STATS #endif /* DEBUG */ #ifdef DNODE_STATS #define DNODE_STAT_ADD(stat) ((stat)++) #else #define DNODE_STAT_ADD(stat) /* nothing */ #endif /* DNODE_STATS */ ASSERTV(static dnode_phys_t dnode_phys_zero); int zfs_default_bs = SPA_MINBLOCKSHIFT; int zfs_default_ibs = DN_MAX_INDBLKSHIFT; #ifdef _KERNEL static kmem_cbrc_t dnode_move(void *, void *, size_t, void *); #endif /* _KERNEL */ /* ARGSUSED */ static int dnode_cons(void *arg, void *unused, int kmflag) { dnode_t *dn = arg; int i; rw_init(&dn->dn_struct_rwlock, NULL, RW_DEFAULT, NULL); mutex_init(&dn->dn_mtx, NULL, MUTEX_DEFAULT, NULL); mutex_init(&dn->dn_dbufs_mtx, NULL, MUTEX_DEFAULT, NULL); cv_init(&dn->dn_notxholds, NULL, CV_DEFAULT, NULL); /* * Every dbuf has a reference, and dropping a tracked reference is * O(number of references), so don't track dn_holds. */ refcount_create_untracked(&dn->dn_holds); refcount_create(&dn->dn_tx_holds); list_link_init(&dn->dn_link); bzero(&dn->dn_next_nblkptr[0], sizeof (dn->dn_next_nblkptr)); bzero(&dn->dn_next_nlevels[0], sizeof (dn->dn_next_nlevels)); bzero(&dn->dn_next_indblkshift[0], sizeof (dn->dn_next_indblkshift)); bzero(&dn->dn_next_bonustype[0], sizeof (dn->dn_next_bonustype)); bzero(&dn->dn_rm_spillblk[0], sizeof (dn->dn_rm_spillblk)); bzero(&dn->dn_next_bonuslen[0], sizeof (dn->dn_next_bonuslen)); bzero(&dn->dn_next_blksz[0], sizeof (dn->dn_next_blksz)); for (i = 0; i < TXG_SIZE; i++) { list_link_init(&dn->dn_dirty_link[i]); dn->dn_free_ranges[i] = NULL; list_create(&dn->dn_dirty_records[i], sizeof (dbuf_dirty_record_t), offsetof(dbuf_dirty_record_t, dr_dirty_node)); } dn->dn_allocated_txg = 0; dn->dn_free_txg = 0; dn->dn_assigned_txg = 0; dn->dn_dirtyctx = 0; dn->dn_dirtyctx_firstset = NULL; dn->dn_bonus = NULL; dn->dn_have_spill = B_FALSE; dn->dn_zio = NULL; dn->dn_oldused = 0; dn->dn_oldflags = 0; dn->dn_olduid = 0; dn->dn_oldgid = 0; dn->dn_newuid = 0; dn->dn_newgid = 0; dn->dn_id_flags = 0; dn->dn_dbufs_count = 0; dn->dn_unlisted_l0_blkid = 0; list_create(&dn->dn_dbufs, sizeof (dmu_buf_impl_t), offsetof(dmu_buf_impl_t, db_link)); dn->dn_moved = 0; return (0); } /* ARGSUSED */ static void dnode_dest(void *arg, void *unused) { int i; dnode_t *dn = arg; rw_destroy(&dn->dn_struct_rwlock); mutex_destroy(&dn->dn_mtx); mutex_destroy(&dn->dn_dbufs_mtx); cv_destroy(&dn->dn_notxholds); refcount_destroy(&dn->dn_holds); refcount_destroy(&dn->dn_tx_holds); ASSERT(!list_link_active(&dn->dn_link)); for (i = 0; i < TXG_SIZE; i++) { ASSERT(!list_link_active(&dn->dn_dirty_link[i])); ASSERT3P(dn->dn_free_ranges[i], ==, NULL); list_destroy(&dn->dn_dirty_records[i]); ASSERT0(dn->dn_next_nblkptr[i]); ASSERT0(dn->dn_next_nlevels[i]); ASSERT0(dn->dn_next_indblkshift[i]); ASSERT0(dn->dn_next_bonustype[i]); ASSERT0(dn->dn_rm_spillblk[i]); ASSERT0(dn->dn_next_bonuslen[i]); ASSERT0(dn->dn_next_blksz[i]); } ASSERT0(dn->dn_allocated_txg); ASSERT0(dn->dn_free_txg); ASSERT0(dn->dn_assigned_txg); ASSERT0(dn->dn_dirtyctx); ASSERT3P(dn->dn_dirtyctx_firstset, ==, NULL); ASSERT3P(dn->dn_bonus, ==, NULL); ASSERT(!dn->dn_have_spill); ASSERT3P(dn->dn_zio, ==, NULL); ASSERT0(dn->dn_oldused); ASSERT0(dn->dn_oldflags); ASSERT0(dn->dn_olduid); ASSERT0(dn->dn_oldgid); ASSERT0(dn->dn_newuid); ASSERT0(dn->dn_newgid); ASSERT0(dn->dn_id_flags); ASSERT0(dn->dn_dbufs_count); ASSERT0(dn->dn_unlisted_l0_blkid); list_destroy(&dn->dn_dbufs); } void dnode_init(void) { ASSERT(dnode_cache == NULL); dnode_cache = kmem_cache_create("dnode_t", sizeof (dnode_t), 0, dnode_cons, dnode_dest, NULL, NULL, NULL, 0); kmem_cache_set_move(dnode_cache, dnode_move); } void dnode_fini(void) { kmem_cache_destroy(dnode_cache); dnode_cache = NULL; } #ifdef ZFS_DEBUG void dnode_verify(dnode_t *dn) { int drop_struct_lock = FALSE; ASSERT(dn->dn_phys); ASSERT(dn->dn_objset); ASSERT(dn->dn_handle->dnh_dnode == dn); ASSERT(DMU_OT_IS_VALID(dn->dn_phys->dn_type)); if (!(zfs_flags & ZFS_DEBUG_DNODE_VERIFY)) return; if (!RW_WRITE_HELD(&dn->dn_struct_rwlock)) { rw_enter(&dn->dn_struct_rwlock, RW_READER); drop_struct_lock = TRUE; } if (dn->dn_phys->dn_type != DMU_OT_NONE || dn->dn_allocated_txg != 0) { int i; ASSERT3U(dn->dn_indblkshift, <=, SPA_MAXBLOCKSHIFT); if (dn->dn_datablkshift) { ASSERT3U(dn->dn_datablkshift, >=, SPA_MINBLOCKSHIFT); ASSERT3U(dn->dn_datablkshift, <=, SPA_MAXBLOCKSHIFT); ASSERT3U(1<dn_datablkshift, ==, dn->dn_datablksz); } ASSERT3U(dn->dn_nlevels, <=, 30); ASSERT(DMU_OT_IS_VALID(dn->dn_type)); ASSERT3U(dn->dn_nblkptr, >=, 1); ASSERT3U(dn->dn_nblkptr, <=, DN_MAX_NBLKPTR); ASSERT3U(dn->dn_bonuslen, <=, DN_MAX_BONUSLEN); ASSERT3U(dn->dn_datablksz, ==, dn->dn_datablkszsec << SPA_MINBLOCKSHIFT); ASSERT3U(ISP2(dn->dn_datablksz), ==, dn->dn_datablkshift != 0); ASSERT3U((dn->dn_nblkptr - 1) * sizeof (blkptr_t) + dn->dn_bonuslen, <=, DN_MAX_BONUSLEN); for (i = 0; i < TXG_SIZE; i++) { ASSERT3U(dn->dn_next_nlevels[i], <=, dn->dn_nlevels); } } if (dn->dn_phys->dn_type != DMU_OT_NONE) ASSERT3U(dn->dn_phys->dn_nlevels, <=, dn->dn_nlevels); ASSERT(DMU_OBJECT_IS_SPECIAL(dn->dn_object) || dn->dn_dbuf != NULL); if (dn->dn_dbuf != NULL) { ASSERT3P(dn->dn_phys, ==, (dnode_phys_t *)dn->dn_dbuf->db.db_data + (dn->dn_object % (dn->dn_dbuf->db.db_size >> DNODE_SHIFT))); } if (drop_struct_lock) rw_exit(&dn->dn_struct_rwlock); } #endif void dnode_byteswap(dnode_phys_t *dnp) { uint64_t *buf64 = (void*)&dnp->dn_blkptr; int i; if (dnp->dn_type == DMU_OT_NONE) { bzero(dnp, sizeof (dnode_phys_t)); return; } dnp->dn_datablkszsec = BSWAP_16(dnp->dn_datablkszsec); dnp->dn_bonuslen = BSWAP_16(dnp->dn_bonuslen); dnp->dn_maxblkid = BSWAP_64(dnp->dn_maxblkid); dnp->dn_used = BSWAP_64(dnp->dn_used); /* * dn_nblkptr is only one byte, so it's OK to read it in either * byte order. We can't read dn_bouslen. */ ASSERT(dnp->dn_indblkshift <= SPA_MAXBLOCKSHIFT); ASSERT(dnp->dn_nblkptr <= DN_MAX_NBLKPTR); for (i = 0; i < dnp->dn_nblkptr * sizeof (blkptr_t)/8; i++) buf64[i] = BSWAP_64(buf64[i]); /* * OK to check dn_bonuslen for zero, because it won't matter if * we have the wrong byte order. This is necessary because the * dnode dnode is smaller than a regular dnode. */ if (dnp->dn_bonuslen != 0) { /* * Note that the bonus length calculated here may be * longer than the actual bonus buffer. This is because * we always put the bonus buffer after the last block * pointer (instead of packing it against the end of the * dnode buffer). */ int off = (dnp->dn_nblkptr-1) * sizeof (blkptr_t); size_t len = DN_MAX_BONUSLEN - off; dmu_object_byteswap_t byteswap; ASSERT(DMU_OT_IS_VALID(dnp->dn_bonustype)); byteswap = DMU_OT_BYTESWAP(dnp->dn_bonustype); dmu_ot_byteswap[byteswap].ob_func(dnp->dn_bonus + off, len); } /* Swap SPILL block if we have one */ if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) byteswap_uint64_array(&dnp->dn_spill, sizeof (blkptr_t)); } void dnode_buf_byteswap(void *vbuf, size_t size) { dnode_phys_t *buf = vbuf; int i; ASSERT3U(sizeof (dnode_phys_t), ==, (1<>= DNODE_SHIFT; for (i = 0; i < size; i++) { dnode_byteswap(buf); buf++; } } void dnode_setbonuslen(dnode_t *dn, int newsize, dmu_tx_t *tx) { ASSERT3U(refcount_count(&dn->dn_holds), >=, 1); dnode_setdirty(dn, tx); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); ASSERT3U(newsize, <=, DN_MAX_BONUSLEN - (dn->dn_nblkptr-1) * sizeof (blkptr_t)); dn->dn_bonuslen = newsize; if (newsize == 0) dn->dn_next_bonuslen[tx->tx_txg & TXG_MASK] = DN_ZERO_BONUSLEN; else dn->dn_next_bonuslen[tx->tx_txg & TXG_MASK] = dn->dn_bonuslen; rw_exit(&dn->dn_struct_rwlock); } void dnode_setbonus_type(dnode_t *dn, dmu_object_type_t newtype, dmu_tx_t *tx) { ASSERT3U(refcount_count(&dn->dn_holds), >=, 1); dnode_setdirty(dn, tx); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); dn->dn_bonustype = newtype; dn->dn_next_bonustype[tx->tx_txg & TXG_MASK] = dn->dn_bonustype; rw_exit(&dn->dn_struct_rwlock); } void dnode_rm_spill(dnode_t *dn, dmu_tx_t *tx) { ASSERT3U(refcount_count(&dn->dn_holds), >=, 1); ASSERT(RW_WRITE_HELD(&dn->dn_struct_rwlock)); dnode_setdirty(dn, tx); dn->dn_rm_spillblk[tx->tx_txg&TXG_MASK] = DN_KILL_SPILLBLK; dn->dn_have_spill = B_FALSE; } static void dnode_setdblksz(dnode_t *dn, int size) { ASSERT0(P2PHASE(size, SPA_MINBLOCKSIZE)); ASSERT3U(size, <=, SPA_MAXBLOCKSIZE); ASSERT3U(size, >=, SPA_MINBLOCKSIZE); ASSERT3U(size >> SPA_MINBLOCKSHIFT, <, 1<<(sizeof (dn->dn_phys->dn_datablkszsec) * 8)); dn->dn_datablksz = size; dn->dn_datablkszsec = size >> SPA_MINBLOCKSHIFT; dn->dn_datablkshift = ISP2(size) ? highbit64(size - 1) : 0; } static dnode_t * dnode_create(objset_t *os, dnode_phys_t *dnp, dmu_buf_impl_t *db, uint64_t object, dnode_handle_t *dnh) { dnode_t *dn = kmem_cache_alloc(dnode_cache, KM_PUSHPAGE); ASSERT(!POINTER_IS_VALID(dn->dn_objset)); dn->dn_moved = 0; /* * Defer setting dn_objset until the dnode is ready to be a candidate * for the dnode_move() callback. */ dn->dn_object = object; dn->dn_dbuf = db; dn->dn_handle = dnh; dn->dn_phys = dnp; if (dnp->dn_datablkszsec) { dnode_setdblksz(dn, dnp->dn_datablkszsec << SPA_MINBLOCKSHIFT); } else { dn->dn_datablksz = 0; dn->dn_datablkszsec = 0; dn->dn_datablkshift = 0; } dn->dn_indblkshift = dnp->dn_indblkshift; dn->dn_nlevels = dnp->dn_nlevels; dn->dn_type = dnp->dn_type; dn->dn_nblkptr = dnp->dn_nblkptr; dn->dn_checksum = dnp->dn_checksum; dn->dn_compress = dnp->dn_compress; dn->dn_bonustype = dnp->dn_bonustype; dn->dn_bonuslen = dnp->dn_bonuslen; dn->dn_maxblkid = dnp->dn_maxblkid; dn->dn_have_spill = ((dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) != 0); dn->dn_id_flags = 0; dmu_zfetch_init(&dn->dn_zfetch, dn); ASSERT(DMU_OT_IS_VALID(dn->dn_phys->dn_type)); mutex_enter(&os->os_lock); list_insert_head(&os->os_dnodes, dn); membar_producer(); /* * Everything else must be valid before assigning dn_objset makes the * dnode eligible for dnode_move(). */ dn->dn_objset = os; mutex_exit(&os->os_lock); arc_space_consume(sizeof (dnode_t), ARC_SPACE_OTHER); return (dn); } /* * Caller must be holding the dnode handle, which is released upon return. */ static void dnode_destroy(dnode_t *dn) { objset_t *os = dn->dn_objset; ASSERT((dn->dn_id_flags & DN_ID_NEW_EXIST) == 0); mutex_enter(&os->os_lock); POINTER_INVALIDATE(&dn->dn_objset); list_remove(&os->os_dnodes, dn); mutex_exit(&os->os_lock); /* the dnode can no longer move, so we can release the handle */ zrl_remove(&dn->dn_handle->dnh_zrlock); dn->dn_allocated_txg = 0; dn->dn_free_txg = 0; dn->dn_assigned_txg = 0; dn->dn_dirtyctx = 0; if (dn->dn_dirtyctx_firstset != NULL) { kmem_free(dn->dn_dirtyctx_firstset, 1); dn->dn_dirtyctx_firstset = NULL; } if (dn->dn_bonus != NULL) { mutex_enter(&dn->dn_bonus->db_mtx); dbuf_evict(dn->dn_bonus); dn->dn_bonus = NULL; } dn->dn_zio = NULL; dn->dn_have_spill = B_FALSE; dn->dn_oldused = 0; dn->dn_oldflags = 0; dn->dn_olduid = 0; dn->dn_oldgid = 0; dn->dn_newuid = 0; dn->dn_newgid = 0; dn->dn_id_flags = 0; dn->dn_unlisted_l0_blkid = 0; dmu_zfetch_rele(&dn->dn_zfetch); kmem_cache_free(dnode_cache, dn); arc_space_return(sizeof (dnode_t), ARC_SPACE_OTHER); } void dnode_allocate(dnode_t *dn, dmu_object_type_t ot, int blocksize, int ibs, dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *tx) { int i; if (blocksize == 0) blocksize = 1 << zfs_default_bs; else if (blocksize > SPA_MAXBLOCKSIZE) blocksize = SPA_MAXBLOCKSIZE; else blocksize = P2ROUNDUP(blocksize, SPA_MINBLOCKSIZE); if (ibs == 0) ibs = zfs_default_ibs; ibs = MIN(MAX(ibs, DN_MIN_INDBLKSHIFT), DN_MAX_INDBLKSHIFT); dprintf("os=%p obj=%llu txg=%llu blocksize=%d ibs=%d\n", dn->dn_objset, dn->dn_object, tx->tx_txg, blocksize, ibs); ASSERT(dn->dn_type == DMU_OT_NONE); ASSERT(bcmp(dn->dn_phys, &dnode_phys_zero, sizeof (dnode_phys_t)) == 0); ASSERT(dn->dn_phys->dn_type == DMU_OT_NONE); ASSERT(ot != DMU_OT_NONE); ASSERT(DMU_OT_IS_VALID(ot)); ASSERT((bonustype == DMU_OT_NONE && bonuslen == 0) || (bonustype == DMU_OT_SA && bonuslen == 0) || (bonustype != DMU_OT_NONE && bonuslen != 0)); ASSERT(DMU_OT_IS_VALID(bonustype)); ASSERT3U(bonuslen, <=, DN_MAX_BONUSLEN); ASSERT(dn->dn_type == DMU_OT_NONE); ASSERT0(dn->dn_maxblkid); ASSERT0(dn->dn_allocated_txg); ASSERT0(dn->dn_assigned_txg); ASSERT(refcount_is_zero(&dn->dn_tx_holds)); ASSERT3U(refcount_count(&dn->dn_holds), <=, 1); ASSERT3P(list_head(&dn->dn_dbufs), ==, NULL); for (i = 0; i < TXG_SIZE; i++) { ASSERT0(dn->dn_next_nblkptr[i]); ASSERT0(dn->dn_next_nlevels[i]); ASSERT0(dn->dn_next_indblkshift[i]); ASSERT0(dn->dn_next_bonuslen[i]); ASSERT0(dn->dn_next_bonustype[i]); ASSERT0(dn->dn_rm_spillblk[i]); ASSERT0(dn->dn_next_blksz[i]); ASSERT(!list_link_active(&dn->dn_dirty_link[i])); ASSERT3P(list_head(&dn->dn_dirty_records[i]), ==, NULL); ASSERT3P(dn->dn_free_ranges[i], ==, NULL); } dn->dn_type = ot; dnode_setdblksz(dn, blocksize); dn->dn_indblkshift = ibs; dn->dn_nlevels = 1; if (bonustype == DMU_OT_SA) /* Maximize bonus space for SA */ dn->dn_nblkptr = 1; else dn->dn_nblkptr = 1 + ((DN_MAX_BONUSLEN - bonuslen) >> SPA_BLKPTRSHIFT); dn->dn_bonustype = bonustype; dn->dn_bonuslen = bonuslen; dn->dn_checksum = ZIO_CHECKSUM_INHERIT; dn->dn_compress = ZIO_COMPRESS_INHERIT; dn->dn_dirtyctx = 0; dn->dn_free_txg = 0; if (dn->dn_dirtyctx_firstset) { kmem_free(dn->dn_dirtyctx_firstset, 1); dn->dn_dirtyctx_firstset = NULL; } dn->dn_allocated_txg = tx->tx_txg; dn->dn_id_flags = 0; dnode_setdirty(dn, tx); dn->dn_next_indblkshift[tx->tx_txg & TXG_MASK] = ibs; dn->dn_next_bonuslen[tx->tx_txg & TXG_MASK] = dn->dn_bonuslen; dn->dn_next_bonustype[tx->tx_txg & TXG_MASK] = dn->dn_bonustype; dn->dn_next_blksz[tx->tx_txg & TXG_MASK] = dn->dn_datablksz; } void dnode_reallocate(dnode_t *dn, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *tx) { int nblkptr; ASSERT3U(blocksize, >=, SPA_MINBLOCKSIZE); ASSERT3U(blocksize, <=, SPA_MAXBLOCKSIZE); ASSERT0(blocksize % SPA_MINBLOCKSIZE); ASSERT(dn->dn_object != DMU_META_DNODE_OBJECT || dmu_tx_private_ok(tx)); ASSERT(tx->tx_txg != 0); ASSERT((bonustype == DMU_OT_NONE && bonuslen == 0) || (bonustype != DMU_OT_NONE && bonuslen != 0) || (bonustype == DMU_OT_SA && bonuslen == 0)); ASSERT(DMU_OT_IS_VALID(bonustype)); ASSERT3U(bonuslen, <=, DN_MAX_BONUSLEN); /* clean up any unreferenced dbufs */ dnode_evict_dbufs(dn); dn->dn_id_flags = 0; rw_enter(&dn->dn_struct_rwlock, RW_WRITER); dnode_setdirty(dn, tx); if (dn->dn_datablksz != blocksize) { /* change blocksize */ ASSERT(dn->dn_maxblkid == 0 && (BP_IS_HOLE(&dn->dn_phys->dn_blkptr[0]) || dnode_block_freed(dn, 0))); dnode_setdblksz(dn, blocksize); dn->dn_next_blksz[tx->tx_txg&TXG_MASK] = blocksize; } if (dn->dn_bonuslen != bonuslen) dn->dn_next_bonuslen[tx->tx_txg&TXG_MASK] = bonuslen; if (bonustype == DMU_OT_SA) /* Maximize bonus space for SA */ nblkptr = 1; else nblkptr = 1 + ((DN_MAX_BONUSLEN - bonuslen) >> SPA_BLKPTRSHIFT); if (dn->dn_bonustype != bonustype) dn->dn_next_bonustype[tx->tx_txg&TXG_MASK] = bonustype; if (dn->dn_nblkptr != nblkptr) dn->dn_next_nblkptr[tx->tx_txg&TXG_MASK] = nblkptr; if (dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR) { dbuf_rm_spill(dn, tx); dnode_rm_spill(dn, tx); } rw_exit(&dn->dn_struct_rwlock); /* change type */ dn->dn_type = ot; /* change bonus size and type */ mutex_enter(&dn->dn_mtx); dn->dn_bonustype = bonustype; dn->dn_bonuslen = bonuslen; dn->dn_nblkptr = nblkptr; dn->dn_checksum = ZIO_CHECKSUM_INHERIT; dn->dn_compress = ZIO_COMPRESS_INHERIT; ASSERT3U(dn->dn_nblkptr, <=, DN_MAX_NBLKPTR); /* fix up the bonus db_size */ if (dn->dn_bonus) { dn->dn_bonus->db.db_size = DN_MAX_BONUSLEN - (dn->dn_nblkptr-1) * sizeof (blkptr_t); ASSERT(dn->dn_bonuslen <= dn->dn_bonus->db.db_size); } dn->dn_allocated_txg = tx->tx_txg; mutex_exit(&dn->dn_mtx); } #ifdef _KERNEL #ifdef DNODE_STATS static struct { uint64_t dms_dnode_invalid; uint64_t dms_dnode_recheck1; uint64_t dms_dnode_recheck2; uint64_t dms_dnode_special; uint64_t dms_dnode_handle; uint64_t dms_dnode_rwlock; uint64_t dms_dnode_active; } dnode_move_stats; #endif /* DNODE_STATS */ static void dnode_move_impl(dnode_t *odn, dnode_t *ndn) { int i; ASSERT(!RW_LOCK_HELD(&odn->dn_struct_rwlock)); ASSERT(MUTEX_NOT_HELD(&odn->dn_mtx)); ASSERT(MUTEX_NOT_HELD(&odn->dn_dbufs_mtx)); ASSERT(!RW_LOCK_HELD(&odn->dn_zfetch.zf_rwlock)); /* Copy fields. */ ndn->dn_objset = odn->dn_objset; ndn->dn_object = odn->dn_object; ndn->dn_dbuf = odn->dn_dbuf; ndn->dn_handle = odn->dn_handle; ndn->dn_phys = odn->dn_phys; ndn->dn_type = odn->dn_type; ndn->dn_bonuslen = odn->dn_bonuslen; ndn->dn_bonustype = odn->dn_bonustype; ndn->dn_nblkptr = odn->dn_nblkptr; ndn->dn_checksum = odn->dn_checksum; ndn->dn_compress = odn->dn_compress; ndn->dn_nlevels = odn->dn_nlevels; ndn->dn_indblkshift = odn->dn_indblkshift; ndn->dn_datablkshift = odn->dn_datablkshift; ndn->dn_datablkszsec = odn->dn_datablkszsec; ndn->dn_datablksz = odn->dn_datablksz; ndn->dn_maxblkid = odn->dn_maxblkid; bcopy(&odn->dn_next_nblkptr[0], &ndn->dn_next_nblkptr[0], sizeof (odn->dn_next_nblkptr)); bcopy(&odn->dn_next_nlevels[0], &ndn->dn_next_nlevels[0], sizeof (odn->dn_next_nlevels)); bcopy(&odn->dn_next_indblkshift[0], &ndn->dn_next_indblkshift[0], sizeof (odn->dn_next_indblkshift)); bcopy(&odn->dn_next_bonustype[0], &ndn->dn_next_bonustype[0], sizeof (odn->dn_next_bonustype)); bcopy(&odn->dn_rm_spillblk[0], &ndn->dn_rm_spillblk[0], sizeof (odn->dn_rm_spillblk)); bcopy(&odn->dn_next_bonuslen[0], &ndn->dn_next_bonuslen[0], sizeof (odn->dn_next_bonuslen)); bcopy(&odn->dn_next_blksz[0], &ndn->dn_next_blksz[0], sizeof (odn->dn_next_blksz)); for (i = 0; i < TXG_SIZE; i++) { list_move_tail(&ndn->dn_dirty_records[i], &odn->dn_dirty_records[i]); } bcopy(&odn->dn_free_ranges[0], &ndn->dn_free_ranges[0], sizeof (odn->dn_free_ranges)); ndn->dn_allocated_txg = odn->dn_allocated_txg; ndn->dn_free_txg = odn->dn_free_txg; ndn->dn_assigned_txg = odn->dn_assigned_txg; ndn->dn_dirtyctx = odn->dn_dirtyctx; ndn->dn_dirtyctx_firstset = odn->dn_dirtyctx_firstset; ASSERT(refcount_count(&odn->dn_tx_holds) == 0); refcount_transfer(&ndn->dn_holds, &odn->dn_holds); ASSERT(list_is_empty(&ndn->dn_dbufs)); list_move_tail(&ndn->dn_dbufs, &odn->dn_dbufs); ndn->dn_dbufs_count = odn->dn_dbufs_count; ndn->dn_unlisted_l0_blkid = odn->dn_unlisted_l0_blkid; ndn->dn_bonus = odn->dn_bonus; ndn->dn_have_spill = odn->dn_have_spill; ndn->dn_zio = odn->dn_zio; ndn->dn_oldused = odn->dn_oldused; ndn->dn_oldflags = odn->dn_oldflags; ndn->dn_olduid = odn->dn_olduid; ndn->dn_oldgid = odn->dn_oldgid; ndn->dn_newuid = odn->dn_newuid; ndn->dn_newgid = odn->dn_newgid; ndn->dn_id_flags = odn->dn_id_flags; dmu_zfetch_init(&ndn->dn_zfetch, NULL); list_move_tail(&ndn->dn_zfetch.zf_stream, &odn->dn_zfetch.zf_stream); ndn->dn_zfetch.zf_dnode = odn->dn_zfetch.zf_dnode; ndn->dn_zfetch.zf_stream_cnt = odn->dn_zfetch.zf_stream_cnt; ndn->dn_zfetch.zf_alloc_fail = odn->dn_zfetch.zf_alloc_fail; /* * Update back pointers. Updating the handle fixes the back pointer of * every descendant dbuf as well as the bonus dbuf. */ ASSERT(ndn->dn_handle->dnh_dnode == odn); ndn->dn_handle->dnh_dnode = ndn; if (ndn->dn_zfetch.zf_dnode == odn) { ndn->dn_zfetch.zf_dnode = ndn; } /* * Invalidate the original dnode by clearing all of its back pointers. */ odn->dn_dbuf = NULL; odn->dn_handle = NULL; list_create(&odn->dn_dbufs, sizeof (dmu_buf_impl_t), offsetof(dmu_buf_impl_t, db_link)); odn->dn_dbufs_count = 0; odn->dn_unlisted_l0_blkid = 0; odn->dn_bonus = NULL; odn->dn_zfetch.zf_dnode = NULL; /* * Set the low bit of the objset pointer to ensure that dnode_move() * recognizes the dnode as invalid in any subsequent callback. */ POINTER_INVALIDATE(&odn->dn_objset); /* * Satisfy the destructor. */ for (i = 0; i < TXG_SIZE; i++) { list_create(&odn->dn_dirty_records[i], sizeof (dbuf_dirty_record_t), offsetof(dbuf_dirty_record_t, dr_dirty_node)); odn->dn_free_ranges[i] = NULL; odn->dn_next_nlevels[i] = 0; odn->dn_next_indblkshift[i] = 0; odn->dn_next_bonustype[i] = 0; odn->dn_rm_spillblk[i] = 0; odn->dn_next_bonuslen[i] = 0; odn->dn_next_blksz[i] = 0; } odn->dn_allocated_txg = 0; odn->dn_free_txg = 0; odn->dn_assigned_txg = 0; odn->dn_dirtyctx = 0; odn->dn_dirtyctx_firstset = NULL; odn->dn_have_spill = B_FALSE; odn->dn_zio = NULL; odn->dn_oldused = 0; odn->dn_oldflags = 0; odn->dn_olduid = 0; odn->dn_oldgid = 0; odn->dn_newuid = 0; odn->dn_newgid = 0; odn->dn_id_flags = 0; /* * Mark the dnode. */ ndn->dn_moved = 1; odn->dn_moved = (uint8_t)-1; } /*ARGSUSED*/ static kmem_cbrc_t dnode_move(void *buf, void *newbuf, size_t size, void *arg) { dnode_t *odn = buf, *ndn = newbuf; objset_t *os; int64_t refcount; uint32_t dbufs; /* * The dnode is on the objset's list of known dnodes if the objset * pointer is valid. We set the low bit of the objset pointer when * freeing the dnode to invalidate it, and the memory patterns written * by kmem (baddcafe and deadbeef) set at least one of the two low bits. * A newly created dnode sets the objset pointer last of all to indicate * that the dnode is known and in a valid state to be moved by this * function. */ os = odn->dn_objset; if (!POINTER_IS_VALID(os)) { DNODE_STAT_ADD(dnode_move_stats.dms_dnode_invalid); return (KMEM_CBRC_DONT_KNOW); } /* * Ensure that the objset does not go away during the move. */ rw_enter(&os_lock, RW_WRITER); if (os != odn->dn_objset) { rw_exit(&os_lock); DNODE_STAT_ADD(dnode_move_stats.dms_dnode_recheck1); return (KMEM_CBRC_DONT_KNOW); } /* * If the dnode is still valid, then so is the objset. We know that no * valid objset can be freed while we hold os_lock, so we can safely * ensure that the objset remains in use. */ mutex_enter(&os->os_lock); /* * Recheck the objset pointer in case the dnode was removed just before * acquiring the lock. */ if (os != odn->dn_objset) { mutex_exit(&os->os_lock); rw_exit(&os_lock); DNODE_STAT_ADD(dnode_move_stats.dms_dnode_recheck2); return (KMEM_CBRC_DONT_KNOW); } /* * At this point we know that as long as we hold os->os_lock, the dnode * cannot be freed and fields within the dnode can be safely accessed. * The objset listing this dnode cannot go away as long as this dnode is * on its list. */ rw_exit(&os_lock); if (DMU_OBJECT_IS_SPECIAL(odn->dn_object)) { mutex_exit(&os->os_lock); DNODE_STAT_ADD(dnode_move_stats.dms_dnode_special); return (KMEM_CBRC_NO); } ASSERT(odn->dn_dbuf != NULL); /* only "special" dnodes have no parent */ /* * Lock the dnode handle to prevent the dnode from obtaining any new * holds. This also prevents the descendant dbufs and the bonus dbuf * from accessing the dnode, so that we can discount their holds. The * handle is safe to access because we know that while the dnode cannot * go away, neither can its handle. Once we hold dnh_zrlock, we can * safely move any dnode referenced only by dbufs. */ if (!zrl_tryenter(&odn->dn_handle->dnh_zrlock)) { mutex_exit(&os->os_lock); DNODE_STAT_ADD(dnode_move_stats.dms_dnode_handle); return (KMEM_CBRC_LATER); } /* * Ensure a consistent view of the dnode's holds and the dnode's dbufs. * We need to guarantee that there is a hold for every dbuf in order to * determine whether the dnode is actively referenced. Falsely matching * a dbuf to an active hold would lead to an unsafe move. It's possible * that a thread already having an active dnode hold is about to add a * dbuf, and we can't compare hold and dbuf counts while the add is in * progress. */ if (!rw_tryenter(&odn->dn_struct_rwlock, RW_WRITER)) { zrl_exit(&odn->dn_handle->dnh_zrlock); mutex_exit(&os->os_lock); DNODE_STAT_ADD(dnode_move_stats.dms_dnode_rwlock); return (KMEM_CBRC_LATER); } /* * A dbuf may be removed (evicted) without an active dnode hold. In that * case, the dbuf count is decremented under the handle lock before the * dbuf's hold is released. This order ensures that if we count the hold * after the dbuf is removed but before its hold is released, we will * treat the unmatched hold as active and exit safely. If we count the * hold before the dbuf is removed, the hold is discounted, and the * removal is blocked until the move completes. */ refcount = refcount_count(&odn->dn_holds); ASSERT(refcount >= 0); dbufs = odn->dn_dbufs_count; /* We can't have more dbufs than dnode holds. */ ASSERT3U(dbufs, <=, refcount); DTRACE_PROBE3(dnode__move, dnode_t *, odn, int64_t, refcount, uint32_t, dbufs); if (refcount > dbufs) { rw_exit(&odn->dn_struct_rwlock); zrl_exit(&odn->dn_handle->dnh_zrlock); mutex_exit(&os->os_lock); DNODE_STAT_ADD(dnode_move_stats.dms_dnode_active); return (KMEM_CBRC_LATER); } rw_exit(&odn->dn_struct_rwlock); /* * At this point we know that anyone with a hold on the dnode is not * actively referencing it. The dnode is known and in a valid state to * move. We're holding the locks needed to execute the critical section. */ dnode_move_impl(odn, ndn); list_link_replace(&odn->dn_link, &ndn->dn_link); /* If the dnode was safe to move, the refcount cannot have changed. */ ASSERT(refcount == refcount_count(&ndn->dn_holds)); ASSERT(dbufs == ndn->dn_dbufs_count); zrl_exit(&ndn->dn_handle->dnh_zrlock); /* handle has moved */ mutex_exit(&os->os_lock); return (KMEM_CBRC_YES); } #endif /* _KERNEL */ void dnode_special_close(dnode_handle_t *dnh) { dnode_t *dn = dnh->dnh_dnode; /* * Wait for final references to the dnode to clear. This can * only happen if the arc is asyncronously evicting state that * has a hold on this dnode while we are trying to evict this * dnode. */ while (refcount_count(&dn->dn_holds) > 0) delay(1); zrl_add(&dnh->dnh_zrlock); dnode_destroy(dn); /* implicit zrl_remove() */ zrl_destroy(&dnh->dnh_zrlock); dnh->dnh_dnode = NULL; } dnode_t * dnode_special_open(objset_t *os, dnode_phys_t *dnp, uint64_t object, dnode_handle_t *dnh) { dnode_t *dn = dnode_create(os, dnp, NULL, object, dnh); dnh->dnh_dnode = dn; zrl_init(&dnh->dnh_zrlock); DNODE_VERIFY(dn); return (dn); } static void dnode_buf_pageout(dmu_buf_t *db, void *arg) { dnode_children_t *children_dnodes = arg; int i; int epb = db->db_size >> DNODE_SHIFT; ASSERT(epb == children_dnodes->dnc_count); for (i = 0; i < epb; i++) { dnode_handle_t *dnh = &children_dnodes->dnc_children[i]; dnode_t *dn; /* * The dnode handle lock guards against the dnode moving to * another valid address, so there is no need here to guard * against changes to or from NULL. */ if (dnh->dnh_dnode == NULL) { zrl_destroy(&dnh->dnh_zrlock); continue; } zrl_add(&dnh->dnh_zrlock); dn = dnh->dnh_dnode; /* * If there are holds on this dnode, then there should * be holds on the dnode's containing dbuf as well; thus * it wouldn't be eligible for eviction and this function * would not have been called. */ ASSERT(refcount_is_zero(&dn->dn_holds)); ASSERT(refcount_is_zero(&dn->dn_tx_holds)); dnode_destroy(dn); /* implicit zrl_remove() */ zrl_destroy(&dnh->dnh_zrlock); dnh->dnh_dnode = NULL; } kmem_free(children_dnodes, sizeof (dnode_children_t) + (epb - 1) * sizeof (dnode_handle_t)); } /* * errors: * EINVAL - invalid object number. * EIO - i/o error. * succeeds even for free dnodes. */ int dnode_hold_impl(objset_t *os, uint64_t object, int flag, void *tag, dnode_t **dnp) { int epb, idx, err; int drop_struct_lock = FALSE; int type; uint64_t blk; dnode_t *mdn, *dn; dmu_buf_impl_t *db; dnode_children_t *children_dnodes; dnode_handle_t *dnh; /* * If you are holding the spa config lock as writer, you shouldn't * be asking the DMU to do *anything* unless it's the root pool * which may require us to read from the root filesystem while * holding some (not all) of the locks as writer. */ ASSERT(spa_config_held(os->os_spa, SCL_ALL, RW_WRITER) == 0 || (spa_is_root(os->os_spa) && spa_config_held(os->os_spa, SCL_STATE, RW_WRITER))); if (object == DMU_USERUSED_OBJECT || object == DMU_GROUPUSED_OBJECT) { dn = (object == DMU_USERUSED_OBJECT) ? DMU_USERUSED_DNODE(os) : DMU_GROUPUSED_DNODE(os); if (dn == NULL) return (SET_ERROR(ENOENT)); type = dn->dn_type; if ((flag & DNODE_MUST_BE_ALLOCATED) && type == DMU_OT_NONE) return (SET_ERROR(ENOENT)); if ((flag & DNODE_MUST_BE_FREE) && type != DMU_OT_NONE) return (SET_ERROR(EEXIST)); DNODE_VERIFY(dn); (void) refcount_add(&dn->dn_holds, tag); *dnp = dn; return (0); } if (object == 0 || object >= DN_MAX_OBJECT) return (SET_ERROR(EINVAL)); mdn = DMU_META_DNODE(os); ASSERT(mdn->dn_object == DMU_META_DNODE_OBJECT); DNODE_VERIFY(mdn); if (!RW_WRITE_HELD(&mdn->dn_struct_rwlock)) { rw_enter(&mdn->dn_struct_rwlock, RW_READER); drop_struct_lock = TRUE; } blk = dbuf_whichblock(mdn, object * sizeof (dnode_phys_t)); db = dbuf_hold(mdn, blk, FTAG); if (drop_struct_lock) rw_exit(&mdn->dn_struct_rwlock); if (db == NULL) return (SET_ERROR(EIO)); err = dbuf_read(db, NULL, DB_RF_CANFAIL); if (err) { dbuf_rele(db, FTAG); return (err); } ASSERT3U(db->db.db_size, >=, 1<db.db_size >> DNODE_SHIFT; idx = object & (epb-1); ASSERT(DB_DNODE(db)->dn_type == DMU_OT_DNODE); children_dnodes = dmu_buf_get_user(&db->db); if (children_dnodes == NULL) { int i; dnode_children_t *winner; children_dnodes = kmem_alloc(sizeof (dnode_children_t) + - (epb - 1) * sizeof (dnode_handle_t), - KM_PUSHPAGE | KM_NODEBUG); + (epb - 1) * sizeof (dnode_handle_t), KM_PUSHPAGE); children_dnodes->dnc_count = epb; dnh = &children_dnodes->dnc_children[0]; for (i = 0; i < epb; i++) { zrl_init(&dnh[i].dnh_zrlock); dnh[i].dnh_dnode = NULL; } if ((winner = dmu_buf_set_user(&db->db, children_dnodes, NULL, dnode_buf_pageout))) { kmem_free(children_dnodes, sizeof (dnode_children_t) + (epb - 1) * sizeof (dnode_handle_t)); children_dnodes = winner; } } ASSERT(children_dnodes->dnc_count == epb); dnh = &children_dnodes->dnc_children[idx]; zrl_add(&dnh->dnh_zrlock); if ((dn = dnh->dnh_dnode) == NULL) { dnode_phys_t *phys = (dnode_phys_t *)db->db.db_data+idx; dnode_t *winner; dn = dnode_create(os, phys, db, object, dnh); winner = atomic_cas_ptr(&dnh->dnh_dnode, NULL, dn); if (winner != NULL) { zrl_add(&dnh->dnh_zrlock); dnode_destroy(dn); /* implicit zrl_remove() */ dn = winner; } } mutex_enter(&dn->dn_mtx); type = dn->dn_type; if (dn->dn_free_txg || ((flag & DNODE_MUST_BE_ALLOCATED) && type == DMU_OT_NONE) || ((flag & DNODE_MUST_BE_FREE) && (type != DMU_OT_NONE || !refcount_is_zero(&dn->dn_holds)))) { mutex_exit(&dn->dn_mtx); zrl_remove(&dnh->dnh_zrlock); dbuf_rele(db, FTAG); return (type == DMU_OT_NONE ? ENOENT : EEXIST); } mutex_exit(&dn->dn_mtx); if (refcount_add(&dn->dn_holds, tag) == 1) dbuf_add_ref(db, dnh); /* Now we can rely on the hold to prevent the dnode from moving. */ zrl_remove(&dnh->dnh_zrlock); DNODE_VERIFY(dn); ASSERT3P(dn->dn_dbuf, ==, db); ASSERT3U(dn->dn_object, ==, object); dbuf_rele(db, FTAG); *dnp = dn; return (0); } /* * Return held dnode if the object is allocated, NULL if not. */ int dnode_hold(objset_t *os, uint64_t object, void *tag, dnode_t **dnp) { return (dnode_hold_impl(os, object, DNODE_MUST_BE_ALLOCATED, tag, dnp)); } /* * Can only add a reference if there is already at least one * reference on the dnode. Returns FALSE if unable to add a * new reference. */ boolean_t dnode_add_ref(dnode_t *dn, void *tag) { mutex_enter(&dn->dn_mtx); if (refcount_is_zero(&dn->dn_holds)) { mutex_exit(&dn->dn_mtx); return (FALSE); } VERIFY(1 < refcount_add(&dn->dn_holds, tag)); mutex_exit(&dn->dn_mtx); return (TRUE); } void dnode_rele(dnode_t *dn, void *tag) { uint64_t refs; /* Get while the hold prevents the dnode from moving. */ dmu_buf_impl_t *db = dn->dn_dbuf; dnode_handle_t *dnh = dn->dn_handle; mutex_enter(&dn->dn_mtx); refs = refcount_remove(&dn->dn_holds, tag); mutex_exit(&dn->dn_mtx); /* * It's unsafe to release the last hold on a dnode by dnode_rele() or * indirectly by dbuf_rele() while relying on the dnode handle to * prevent the dnode from moving, since releasing the last hold could * result in the dnode's parent dbuf evicting its dnode handles. For * that reason anyone calling dnode_rele() or dbuf_rele() without some * other direct or indirect hold on the dnode must first drop the dnode * handle. */ ASSERT(refs > 0 || dnh->dnh_zrlock.zr_owner != curthread); /* NOTE: the DNODE_DNODE does not have a dn_dbuf */ if (refs == 0 && db != NULL) { /* * Another thread could add a hold to the dnode handle in * dnode_hold_impl() while holding the parent dbuf. Since the * hold on the parent dbuf prevents the handle from being * destroyed, the hold on the handle is OK. We can't yet assert * that the handle has zero references, but that will be * asserted anyway when the handle gets destroyed. */ dbuf_rele(db, dnh); } } void dnode_setdirty(dnode_t *dn, dmu_tx_t *tx) { objset_t *os = dn->dn_objset; uint64_t txg = tx->tx_txg; if (DMU_OBJECT_IS_SPECIAL(dn->dn_object)) { dsl_dataset_dirty(os->os_dsl_dataset, tx); return; } DNODE_VERIFY(dn); #ifdef ZFS_DEBUG mutex_enter(&dn->dn_mtx); ASSERT(dn->dn_phys->dn_type || dn->dn_allocated_txg); ASSERT(dn->dn_free_txg == 0 || dn->dn_free_txg >= txg); mutex_exit(&dn->dn_mtx); #endif /* * Determine old uid/gid when necessary */ dmu_objset_userquota_get_ids(dn, B_TRUE, tx); mutex_enter(&os->os_lock); /* * If we are already marked dirty, we're done. */ if (list_link_active(&dn->dn_dirty_link[txg & TXG_MASK])) { mutex_exit(&os->os_lock); return; } ASSERT(!refcount_is_zero(&dn->dn_holds) || list_head(&dn->dn_dbufs)); ASSERT(dn->dn_datablksz != 0); ASSERT0(dn->dn_next_bonuslen[txg&TXG_MASK]); ASSERT0(dn->dn_next_blksz[txg&TXG_MASK]); ASSERT0(dn->dn_next_bonustype[txg&TXG_MASK]); dprintf_ds(os->os_dsl_dataset, "obj=%llu txg=%llu\n", dn->dn_object, txg); if (dn->dn_free_txg > 0 && dn->dn_free_txg <= txg) { list_insert_tail(&os->os_free_dnodes[txg&TXG_MASK], dn); } else { list_insert_tail(&os->os_dirty_dnodes[txg&TXG_MASK], dn); } mutex_exit(&os->os_lock); /* * The dnode maintains a hold on its containing dbuf as * long as there are holds on it. Each instantiated child * dbuf maintains a hold on the dnode. When the last child * drops its hold, the dnode will drop its hold on the * containing dbuf. We add a "dirty hold" here so that the * dnode will hang around after we finish processing its * children. */ VERIFY(dnode_add_ref(dn, (void *)(uintptr_t)tx->tx_txg)); (void) dbuf_dirty(dn->dn_dbuf, tx); dsl_dataset_dirty(os->os_dsl_dataset, tx); } void dnode_free(dnode_t *dn, dmu_tx_t *tx) { int txgoff = tx->tx_txg & TXG_MASK; dprintf("dn=%p txg=%llu\n", dn, tx->tx_txg); /* we should be the only holder... hopefully */ /* ASSERT3U(refcount_count(&dn->dn_holds), ==, 1); */ mutex_enter(&dn->dn_mtx); if (dn->dn_type == DMU_OT_NONE || dn->dn_free_txg) { mutex_exit(&dn->dn_mtx); return; } dn->dn_free_txg = tx->tx_txg; mutex_exit(&dn->dn_mtx); /* * If the dnode is already dirty, it needs to be moved from * the dirty list to the free list. */ mutex_enter(&dn->dn_objset->os_lock); if (list_link_active(&dn->dn_dirty_link[txgoff])) { list_remove(&dn->dn_objset->os_dirty_dnodes[txgoff], dn); list_insert_tail(&dn->dn_objset->os_free_dnodes[txgoff], dn); mutex_exit(&dn->dn_objset->os_lock); } else { mutex_exit(&dn->dn_objset->os_lock); dnode_setdirty(dn, tx); } } /* * Try to change the block size for the indicated dnode. This can only * succeed if there are no blocks allocated or dirty beyond first block */ int dnode_set_blksz(dnode_t *dn, uint64_t size, int ibs, dmu_tx_t *tx) { dmu_buf_impl_t *db, *db_next; int err; if (size == 0) size = SPA_MINBLOCKSIZE; if (size > SPA_MAXBLOCKSIZE) size = SPA_MAXBLOCKSIZE; else size = P2ROUNDUP(size, SPA_MINBLOCKSIZE); if (ibs == dn->dn_indblkshift) ibs = 0; if (size >> SPA_MINBLOCKSHIFT == dn->dn_datablkszsec && ibs == 0) return (0); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); /* Check for any allocated blocks beyond the first */ if (dn->dn_maxblkid != 0) goto fail; mutex_enter(&dn->dn_dbufs_mtx); for (db = list_head(&dn->dn_dbufs); db; db = db_next) { db_next = list_next(&dn->dn_dbufs, db); if (db->db_blkid != 0 && db->db_blkid != DMU_BONUS_BLKID && db->db_blkid != DMU_SPILL_BLKID) { mutex_exit(&dn->dn_dbufs_mtx); goto fail; } } mutex_exit(&dn->dn_dbufs_mtx); if (ibs && dn->dn_nlevels != 1) goto fail; /* resize the old block */ err = dbuf_hold_impl(dn, 0, 0, TRUE, FTAG, &db); if (err == 0) dbuf_new_size(db, size, tx); else if (err != ENOENT) goto fail; dnode_setdblksz(dn, size); dnode_setdirty(dn, tx); dn->dn_next_blksz[tx->tx_txg&TXG_MASK] = size; if (ibs) { dn->dn_indblkshift = ibs; dn->dn_next_indblkshift[tx->tx_txg&TXG_MASK] = ibs; } /* rele after we have fixed the blocksize in the dnode */ if (db) dbuf_rele(db, FTAG); rw_exit(&dn->dn_struct_rwlock); return (0); fail: rw_exit(&dn->dn_struct_rwlock); return (SET_ERROR(ENOTSUP)); } /* read-holding callers must not rely on the lock being continuously held */ void dnode_new_blkid(dnode_t *dn, uint64_t blkid, dmu_tx_t *tx, boolean_t have_read) { uint64_t txgoff = tx->tx_txg & TXG_MASK; int epbs, new_nlevels; uint64_t sz; ASSERT(blkid != DMU_BONUS_BLKID); ASSERT(have_read ? RW_READ_HELD(&dn->dn_struct_rwlock) : RW_WRITE_HELD(&dn->dn_struct_rwlock)); /* * if we have a read-lock, check to see if we need to do any work * before upgrading to a write-lock. */ if (have_read) { if (blkid <= dn->dn_maxblkid) return; if (!rw_tryupgrade(&dn->dn_struct_rwlock)) { rw_exit(&dn->dn_struct_rwlock); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); } } if (blkid <= dn->dn_maxblkid) goto out; dn->dn_maxblkid = blkid; /* * Compute the number of levels necessary to support the new maxblkid. */ new_nlevels = 1; epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT; for (sz = dn->dn_nblkptr; sz <= blkid && sz >= dn->dn_nblkptr; sz <<= epbs) new_nlevels++; if (new_nlevels > dn->dn_nlevels) { int old_nlevels = dn->dn_nlevels; dmu_buf_impl_t *db; list_t *list; dbuf_dirty_record_t *new, *dr, *dr_next; dn->dn_nlevels = new_nlevels; ASSERT3U(new_nlevels, >, dn->dn_next_nlevels[txgoff]); dn->dn_next_nlevels[txgoff] = new_nlevels; /* dirty the left indirects */ db = dbuf_hold_level(dn, old_nlevels, 0, FTAG); ASSERT(db != NULL); new = dbuf_dirty(db, tx); dbuf_rele(db, FTAG); /* transfer the dirty records to the new indirect */ mutex_enter(&dn->dn_mtx); mutex_enter(&new->dt.di.dr_mtx); list = &dn->dn_dirty_records[txgoff]; for (dr = list_head(list); dr; dr = dr_next) { dr_next = list_next(&dn->dn_dirty_records[txgoff], dr); if (dr->dr_dbuf->db_level != new_nlevels-1 && dr->dr_dbuf->db_blkid != DMU_BONUS_BLKID && dr->dr_dbuf->db_blkid != DMU_SPILL_BLKID) { ASSERT(dr->dr_dbuf->db_level == old_nlevels-1); list_remove(&dn->dn_dirty_records[txgoff], dr); list_insert_tail(&new->dt.di.dr_children, dr); dr->dr_parent = new; } } mutex_exit(&new->dt.di.dr_mtx); mutex_exit(&dn->dn_mtx); } out: if (have_read) rw_downgrade(&dn->dn_struct_rwlock); } void dnode_free_range(dnode_t *dn, uint64_t off, uint64_t len, dmu_tx_t *tx) { dmu_buf_impl_t *db; uint64_t blkoff, blkid, nblks; int blksz, blkshift, head, tail; int trunc = FALSE; int epbs; rw_enter(&dn->dn_struct_rwlock, RW_WRITER); blksz = dn->dn_datablksz; blkshift = dn->dn_datablkshift; epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT; if (len == DMU_OBJECT_END) { len = UINT64_MAX - off; trunc = TRUE; } /* * First, block align the region to free: */ if (ISP2(blksz)) { head = P2NPHASE(off, blksz); blkoff = P2PHASE(off, blksz); if ((off >> blkshift) > dn->dn_maxblkid) goto out; } else { ASSERT(dn->dn_maxblkid == 0); if (off == 0 && len >= blksz) { /* * Freeing the whole block; fast-track this request. * Note that we won't dirty any indirect blocks, * which is fine because we will be freeing the entire * file and thus all indirect blocks will be freed * by free_children(). */ blkid = 0; nblks = 1; goto done; } else if (off >= blksz) { /* Freeing past end-of-data */ goto out; } else { /* Freeing part of the block. */ head = blksz - off; ASSERT3U(head, >, 0); } blkoff = off; } /* zero out any partial block data at the start of the range */ if (head) { ASSERT3U(blkoff + head, ==, blksz); if (len < head) head = len; if (dbuf_hold_impl(dn, 0, dbuf_whichblock(dn, off), TRUE, FTAG, &db) == 0) { caddr_t data; /* don't dirty if it isn't on disk and isn't dirty */ if (db->db_last_dirty || (db->db_blkptr && !BP_IS_HOLE(db->db_blkptr))) { rw_exit(&dn->dn_struct_rwlock); dmu_buf_will_dirty(&db->db, tx); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); data = db->db.db_data; bzero(data + blkoff, head); } dbuf_rele(db, FTAG); } off += head; len -= head; } /* If the range was less than one block, we're done */ if (len == 0) goto out; /* If the remaining range is past end of file, we're done */ if ((off >> blkshift) > dn->dn_maxblkid) goto out; ASSERT(ISP2(blksz)); if (trunc) tail = 0; else tail = P2PHASE(len, blksz); ASSERT0(P2PHASE(off, blksz)); /* zero out any partial block data at the end of the range */ if (tail) { if (len < tail) tail = len; if (dbuf_hold_impl(dn, 0, dbuf_whichblock(dn, off+len), TRUE, FTAG, &db) == 0) { /* don't dirty if not on disk and not dirty */ if (db->db_last_dirty || (db->db_blkptr && !BP_IS_HOLE(db->db_blkptr))) { rw_exit(&dn->dn_struct_rwlock); dmu_buf_will_dirty(&db->db, tx); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); bzero(db->db.db_data, tail); } dbuf_rele(db, FTAG); } len -= tail; } /* If the range did not include a full block, we are done */ if (len == 0) goto out; ASSERT(IS_P2ALIGNED(off, blksz)); ASSERT(trunc || IS_P2ALIGNED(len, blksz)); blkid = off >> blkshift; nblks = len >> blkshift; if (trunc) nblks += 1; /* * Dirty the first and last indirect blocks, as they (and/or their * parents) will need to be written out if they were only * partially freed. Interior indirect blocks will be themselves freed, * by free_children(), so they need not be dirtied. Note that these * interior blocks have already been prefetched by dmu_tx_hold_free(). */ if (dn->dn_nlevels > 1) { uint64_t first, last; first = blkid >> epbs; if ((db = dbuf_hold_level(dn, 1, first, FTAG))) { dmu_buf_will_dirty(&db->db, tx); dbuf_rele(db, FTAG); } if (trunc) last = dn->dn_maxblkid >> epbs; else last = (blkid + nblks - 1) >> epbs; if (last > first && (db = dbuf_hold_level(dn, 1, last, FTAG))) { dmu_buf_will_dirty(&db->db, tx); dbuf_rele(db, FTAG); } } done: /* * Add this range to the dnode range list. * We will finish up this free operation in the syncing phase. */ mutex_enter(&dn->dn_mtx); { int txgoff = tx->tx_txg & TXG_MASK; if (dn->dn_free_ranges[txgoff] == NULL) { dn->dn_free_ranges[txgoff] = range_tree_create(NULL, NULL, &dn->dn_mtx); } range_tree_clear(dn->dn_free_ranges[txgoff], blkid, nblks); range_tree_add(dn->dn_free_ranges[txgoff], blkid, nblks); } dprintf_dnode(dn, "blkid=%llu nblks=%llu txg=%llu\n", blkid, nblks, tx->tx_txg); mutex_exit(&dn->dn_mtx); dbuf_free_range(dn, blkid, blkid + nblks - 1, tx); dnode_setdirty(dn, tx); out: rw_exit(&dn->dn_struct_rwlock); } static boolean_t dnode_spill_freed(dnode_t *dn) { int i; mutex_enter(&dn->dn_mtx); for (i = 0; i < TXG_SIZE; i++) { if (dn->dn_rm_spillblk[i] == DN_KILL_SPILLBLK) break; } mutex_exit(&dn->dn_mtx); return (i < TXG_SIZE); } /* return TRUE if this blkid was freed in a recent txg, or FALSE if it wasn't */ uint64_t dnode_block_freed(dnode_t *dn, uint64_t blkid) { void *dp = spa_get_dsl(dn->dn_objset->os_spa); int i; if (blkid == DMU_BONUS_BLKID) return (FALSE); /* * If we're in the process of opening the pool, dp will not be * set yet, but there shouldn't be anything dirty. */ if (dp == NULL) return (FALSE); if (dn->dn_free_txg) return (TRUE); if (blkid == DMU_SPILL_BLKID) return (dnode_spill_freed(dn)); mutex_enter(&dn->dn_mtx); for (i = 0; i < TXG_SIZE; i++) { if (dn->dn_free_ranges[i] != NULL && range_tree_contains(dn->dn_free_ranges[i], blkid, 1)) break; } mutex_exit(&dn->dn_mtx); return (i < TXG_SIZE); } /* call from syncing context when we actually write/free space for this dnode */ void dnode_diduse_space(dnode_t *dn, int64_t delta) { uint64_t space; dprintf_dnode(dn, "dn=%p dnp=%p used=%llu delta=%lld\n", dn, dn->dn_phys, (u_longlong_t)dn->dn_phys->dn_used, (longlong_t)delta); mutex_enter(&dn->dn_mtx); space = DN_USED_BYTES(dn->dn_phys); if (delta > 0) { ASSERT3U(space + delta, >=, space); /* no overflow */ } else { ASSERT3U(space, >=, -delta); /* no underflow */ } space += delta; if (spa_version(dn->dn_objset->os_spa) < SPA_VERSION_DNODE_BYTES) { ASSERT((dn->dn_phys->dn_flags & DNODE_FLAG_USED_BYTES) == 0); ASSERT0(P2PHASE(space, 1<dn_phys->dn_used = space >> DEV_BSHIFT; } else { dn->dn_phys->dn_used = space; dn->dn_phys->dn_flags |= DNODE_FLAG_USED_BYTES; } mutex_exit(&dn->dn_mtx); } /* * Call when we think we're going to write/free space in open context to track * the amount of memory in use by the currently open txg. */ void dnode_willuse_space(dnode_t *dn, int64_t space, dmu_tx_t *tx) { objset_t *os = dn->dn_objset; dsl_dataset_t *ds = os->os_dsl_dataset; int64_t aspace = spa_get_asize(os->os_spa, space); if (ds != NULL) { dsl_dir_willuse_space(ds->ds_dir, aspace, tx); dsl_pool_dirty_space(dmu_tx_pool(tx), space, tx); } dmu_tx_willuse_space(tx, aspace); } /* * Scans a block at the indicated "level" looking for a hole or data, * depending on 'flags'. * * If level > 0, then we are scanning an indirect block looking at its * pointers. If level == 0, then we are looking at a block of dnodes. * * If we don't find what we are looking for in the block, we return ESRCH. * Otherwise, return with *offset pointing to the beginning (if searching * forwards) or end (if searching backwards) of the range covered by the * block pointer we matched on (or dnode). * * The basic search algorithm used below by dnode_next_offset() is to * use this function to search up the block tree (widen the search) until * we find something (i.e., we don't return ESRCH) and then search back * down the tree (narrow the search) until we reach our original search * level. */ static int dnode_next_offset_level(dnode_t *dn, int flags, uint64_t *offset, int lvl, uint64_t blkfill, uint64_t txg) { dmu_buf_impl_t *db = NULL; void *data = NULL; uint64_t epbs = dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT; uint64_t epb = 1ULL << epbs; uint64_t minfill, maxfill; boolean_t hole; int i, inc, error, span; dprintf("probing object %llu offset %llx level %d of %u\n", dn->dn_object, *offset, lvl, dn->dn_phys->dn_nlevels); hole = ((flags & DNODE_FIND_HOLE) != 0); inc = (flags & DNODE_FIND_BACKWARDS) ? -1 : 1; ASSERT(txg == 0 || !hole); if (lvl == dn->dn_phys->dn_nlevels) { error = 0; epb = dn->dn_phys->dn_nblkptr; data = dn->dn_phys->dn_blkptr; } else { uint64_t blkid = dbuf_whichblock(dn, *offset) >> (epbs * lvl); error = dbuf_hold_impl(dn, lvl, blkid, TRUE, FTAG, &db); if (error) { if (error != ENOENT) return (error); if (hole) return (0); /* * This can only happen when we are searching up * the block tree for data. We don't really need to * adjust the offset, as we will just end up looking * at the pointer to this block in its parent, and its * going to be unallocated, so we will skip over it. */ return (SET_ERROR(ESRCH)); } error = dbuf_read(db, NULL, DB_RF_CANFAIL | DB_RF_HAVESTRUCT); if (error) { dbuf_rele(db, FTAG); return (error); } data = db->db.db_data; } if (db != NULL && txg != 0 && (db->db_blkptr == NULL || db->db_blkptr->blk_birth <= txg || BP_IS_HOLE(db->db_blkptr))) { /* * This can only happen when we are searching up the tree * and these conditions mean that we need to keep climbing. */ error = SET_ERROR(ESRCH); } else if (lvl == 0) { dnode_phys_t *dnp = data; span = DNODE_SHIFT; ASSERT(dn->dn_type == DMU_OT_DNODE); for (i = (*offset >> span) & (blkfill - 1); i >= 0 && i < blkfill; i += inc) { if ((dnp[i].dn_type == DMU_OT_NONE) == hole) break; *offset += (1ULL << span) * inc; } if (i < 0 || i == blkfill) error = SET_ERROR(ESRCH); } else { blkptr_t *bp = data; uint64_t start = *offset; span = (lvl - 1) * epbs + dn->dn_datablkshift; minfill = 0; maxfill = blkfill << ((lvl - 1) * epbs); if (hole) maxfill--; else minfill++; *offset = *offset >> span; for (i = BF64_GET(*offset, 0, epbs); i >= 0 && i < epb; i += inc) { if (BP_GET_FILL(&bp[i]) >= minfill && BP_GET_FILL(&bp[i]) <= maxfill && (hole || bp[i].blk_birth > txg)) break; if (inc > 0 || *offset > 0) *offset += inc; } *offset = *offset << span; if (inc < 0) { /* traversing backwards; position offset at the end */ ASSERT3U(*offset, <=, start); *offset = MIN(*offset + (1ULL << span) - 1, start); } else if (*offset < start) { *offset = start; } if (i < 0 || i >= epb) error = SET_ERROR(ESRCH); } if (db) dbuf_rele(db, FTAG); return (error); } /* * Find the next hole, data, or sparse region at or after *offset. * The value 'blkfill' tells us how many items we expect to find * in an L0 data block; this value is 1 for normal objects, * DNODES_PER_BLOCK for the meta dnode, and some fraction of * DNODES_PER_BLOCK when searching for sparse regions thereof. * * Examples: * * dnode_next_offset(dn, flags, offset, 1, 1, 0); * Finds the next/previous hole/data in a file. * Used in dmu_offset_next(). * * dnode_next_offset(mdn, flags, offset, 0, DNODES_PER_BLOCK, txg); * Finds the next free/allocated dnode an objset's meta-dnode. * Only finds objects that have new contents since txg (ie. * bonus buffer changes and content removal are ignored). * Used in dmu_object_next(). * * dnode_next_offset(mdn, DNODE_FIND_HOLE, offset, 2, DNODES_PER_BLOCK >> 2, 0); * Finds the next L2 meta-dnode bp that's at most 1/4 full. * Used in dmu_object_alloc(). */ int dnode_next_offset(dnode_t *dn, int flags, uint64_t *offset, int minlvl, uint64_t blkfill, uint64_t txg) { uint64_t initial_offset = *offset; int lvl, maxlvl; int error = 0; if (!(flags & DNODE_FIND_HAVELOCK)) rw_enter(&dn->dn_struct_rwlock, RW_READER); if (dn->dn_phys->dn_nlevels == 0) { error = SET_ERROR(ESRCH); goto out; } if (dn->dn_datablkshift == 0) { if (*offset < dn->dn_datablksz) { if (flags & DNODE_FIND_HOLE) *offset = dn->dn_datablksz; } else { error = SET_ERROR(ESRCH); } goto out; } maxlvl = dn->dn_phys->dn_nlevels; for (lvl = minlvl; lvl <= maxlvl; lvl++) { error = dnode_next_offset_level(dn, flags, offset, lvl, blkfill, txg); if (error != ESRCH) break; } while (error == 0 && --lvl >= minlvl) { error = dnode_next_offset_level(dn, flags, offset, lvl, blkfill, txg); } /* * There's always a "virtual hole" at the end of the object, even * if all BP's which physically exist are non-holes. */ if ((flags & DNODE_FIND_HOLE) && error == ESRCH && txg == 0 && minlvl == 1 && blkfill == 1 && !(flags & DNODE_FIND_BACKWARDS)) { error = 0; } if (error == 0 && (flags & DNODE_FIND_BACKWARDS ? initial_offset < *offset : initial_offset > *offset)) error = SET_ERROR(ESRCH); out: if (!(flags & DNODE_FIND_HAVELOCK)) rw_exit(&dn->dn_struct_rwlock); return (error); } diff --git a/module/zfs/dsl_scan.c b/module/zfs/dsl_scan.c index 0e16002b3755..ebb6305eb16e 100644 --- a/module/zfs/dsl_scan.c +++ b/module/zfs/dsl_scan.c @@ -1,1880 +1,1880 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2008, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2014 by Delphix. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef _KERNEL #include #endif typedef int (scan_cb_t)(dsl_pool_t *, const blkptr_t *, const zbookmark_phys_t *); static scan_cb_t dsl_scan_scrub_cb; static void dsl_scan_cancel_sync(void *, dmu_tx_t *); static void dsl_scan_sync_state(dsl_scan_t *, dmu_tx_t *tx); int zfs_top_maxinflight = 32; /* maximum I/Os per top-level */ int zfs_resilver_delay = 2; /* number of ticks to delay resilver */ int zfs_scrub_delay = 4; /* number of ticks to delay scrub */ int zfs_scan_idle = 50; /* idle window in clock ticks */ int zfs_scan_min_time_ms = 1000; /* min millisecs to scrub per txg */ int zfs_free_min_time_ms = 1000; /* min millisecs to free per txg */ int zfs_resilver_min_time_ms = 3000; /* min millisecs to resilver per txg */ int zfs_no_scrub_io = B_FALSE; /* set to disable scrub i/o */ int zfs_no_scrub_prefetch = B_FALSE; /* set to disable scrub prefetch */ enum ddt_class zfs_scrub_ddt_class_max = DDT_CLASS_DUPLICATE; int dsl_scan_delay_completion = B_FALSE; /* set to delay scan completion */ /* max number of blocks to free in a single TXG */ ulong zfs_free_max_blocks = 100000; #define DSL_SCAN_IS_SCRUB_RESILVER(scn) \ ((scn)->scn_phys.scn_func == POOL_SCAN_SCRUB || \ (scn)->scn_phys.scn_func == POOL_SCAN_RESILVER) /* the order has to match pool_scan_type */ static scan_cb_t *scan_funcs[POOL_SCAN_FUNCS] = { NULL, dsl_scan_scrub_cb, /* POOL_SCAN_SCRUB */ dsl_scan_scrub_cb, /* POOL_SCAN_RESILVER */ }; int dsl_scan_init(dsl_pool_t *dp, uint64_t txg) { int err; dsl_scan_t *scn; spa_t *spa = dp->dp_spa; uint64_t f; scn = dp->dp_scan = kmem_zalloc(sizeof (dsl_scan_t), KM_SLEEP); scn->scn_dp = dp; /* * It's possible that we're resuming a scan after a reboot so * make sure that the scan_async_destroying flag is initialized * appropriately. */ ASSERT(!scn->scn_async_destroying); scn->scn_async_destroying = spa_feature_is_active(dp->dp_spa, SPA_FEATURE_ASYNC_DESTROY); err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, "scrub_func", sizeof (uint64_t), 1, &f); if (err == 0) { /* * There was an old-style scrub in progress. Restart a * new-style scrub from the beginning. */ scn->scn_restart_txg = txg; zfs_dbgmsg("old-style scrub was in progress; " "restarting new-style scrub in txg %llu", scn->scn_restart_txg); /* * Load the queue obj from the old location so that it * can be freed by dsl_scan_done(). */ (void) zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, "scrub_queue", sizeof (uint64_t), 1, &scn->scn_phys.scn_queue_obj); } else { err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_SCAN, sizeof (uint64_t), SCAN_PHYS_NUMINTS, &scn->scn_phys); /* * Detect if the pool contains the signature of #2094. If it * does properly update the scn->scn_phys structure and notify * the administrator by setting an errata for the pool. */ if (err == EOVERFLOW) { uint64_t zaptmp[SCAN_PHYS_NUMINTS + 1]; VERIFY3S(SCAN_PHYS_NUMINTS, ==, 24); VERIFY3S(offsetof(dsl_scan_phys_t, scn_flags), ==, (23 * sizeof (uint64_t))); err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_SCAN, sizeof (uint64_t), SCAN_PHYS_NUMINTS + 1, &zaptmp); if (err == 0) { uint64_t overflow = zaptmp[SCAN_PHYS_NUMINTS]; if (overflow & ~DSL_SCAN_FLAGS_MASK || scn->scn_async_destroying) { spa->spa_errata = ZPOOL_ERRATA_ZOL_2094_ASYNC_DESTROY; return (EOVERFLOW); } bcopy(zaptmp, &scn->scn_phys, SCAN_PHYS_NUMINTS * sizeof (uint64_t)); scn->scn_phys.scn_flags = overflow; /* Required scrub already in progress. */ if (scn->scn_phys.scn_state == DSS_FINISHED || scn->scn_phys.scn_state == DSS_CANCELED) spa->spa_errata = ZPOOL_ERRATA_ZOL_2094_SCRUB; } } if (err == ENOENT) return (0); else if (err) return (err); if (scn->scn_phys.scn_state == DSS_SCANNING && spa_prev_software_version(dp->dp_spa) < SPA_VERSION_SCAN) { /* * A new-type scrub was in progress on an old * pool, and the pool was accessed by old * software. Restart from the beginning, since * the old software may have changed the pool in * the meantime. */ scn->scn_restart_txg = txg; zfs_dbgmsg("new-style scrub was modified " "by old software; restarting in txg %llu", scn->scn_restart_txg); } } spa_scan_stat_init(spa); return (0); } void dsl_scan_fini(dsl_pool_t *dp) { if (dp->dp_scan) { kmem_free(dp->dp_scan, sizeof (dsl_scan_t)); dp->dp_scan = NULL; } } /* ARGSUSED */ static int dsl_scan_setup_check(void *arg, dmu_tx_t *tx) { dsl_scan_t *scn = dmu_tx_pool(tx)->dp_scan; if (scn->scn_phys.scn_state == DSS_SCANNING) return (SET_ERROR(EBUSY)); return (0); } static void dsl_scan_setup_sync(void *arg, dmu_tx_t *tx) { dsl_scan_t *scn = dmu_tx_pool(tx)->dp_scan; pool_scan_func_t *funcp = arg; dmu_object_type_t ot = 0; dsl_pool_t *dp = scn->scn_dp; spa_t *spa = dp->dp_spa; ASSERT(scn->scn_phys.scn_state != DSS_SCANNING); ASSERT(*funcp > POOL_SCAN_NONE && *funcp < POOL_SCAN_FUNCS); bzero(&scn->scn_phys, sizeof (scn->scn_phys)); scn->scn_phys.scn_func = *funcp; scn->scn_phys.scn_state = DSS_SCANNING; scn->scn_phys.scn_min_txg = 0; scn->scn_phys.scn_max_txg = tx->tx_txg; scn->scn_phys.scn_ddt_class_max = DDT_CLASSES - 1; /* the entire DDT */ scn->scn_phys.scn_start_time = gethrestime_sec(); scn->scn_phys.scn_errors = 0; scn->scn_phys.scn_to_examine = spa->spa_root_vdev->vdev_stat.vs_alloc; scn->scn_restart_txg = 0; scn->scn_done_txg = 0; spa_scan_stat_init(spa); if (DSL_SCAN_IS_SCRUB_RESILVER(scn)) { scn->scn_phys.scn_ddt_class_max = zfs_scrub_ddt_class_max; /* rewrite all disk labels */ vdev_config_dirty(spa->spa_root_vdev); if (vdev_resilver_needed(spa->spa_root_vdev, &scn->scn_phys.scn_min_txg, &scn->scn_phys.scn_max_txg)) { spa_event_notify(spa, NULL, FM_EREPORT_ZFS_RESILVER_START); } else { spa_event_notify(spa, NULL, FM_EREPORT_ZFS_SCRUB_START); } spa->spa_scrub_started = B_TRUE; /* * If this is an incremental scrub, limit the DDT scrub phase * to just the auto-ditto class (for correctness); the rest * of the scrub should go faster using top-down pruning. */ if (scn->scn_phys.scn_min_txg > TXG_INITIAL) scn->scn_phys.scn_ddt_class_max = DDT_CLASS_DITTO; } /* back to the generic stuff */ if (dp->dp_blkstats == NULL) { dp->dp_blkstats = kmem_alloc(sizeof (zfs_all_blkstats_t), - KM_PUSHPAGE | KM_NODEBUG); + KM_PUSHPAGE); } bzero(dp->dp_blkstats, sizeof (zfs_all_blkstats_t)); if (spa_version(spa) < SPA_VERSION_DSL_SCRUB) ot = DMU_OT_ZAP_OTHER; scn->scn_phys.scn_queue_obj = zap_create(dp->dp_meta_objset, ot ? ot : DMU_OT_SCAN_QUEUE, DMU_OT_NONE, 0, tx); dsl_scan_sync_state(scn, tx); spa_history_log_internal(spa, "scan setup", tx, "func=%u mintxg=%llu maxtxg=%llu", *funcp, scn->scn_phys.scn_min_txg, scn->scn_phys.scn_max_txg); } /* ARGSUSED */ static void dsl_scan_done(dsl_scan_t *scn, boolean_t complete, dmu_tx_t *tx) { static const char *old_names[] = { "scrub_bookmark", "scrub_ddt_bookmark", "scrub_ddt_class_max", "scrub_queue", "scrub_min_txg", "scrub_max_txg", "scrub_func", "scrub_errors", NULL }; dsl_pool_t *dp = scn->scn_dp; spa_t *spa = dp->dp_spa; int i; /* Remove any remnants of an old-style scrub. */ for (i = 0; old_names[i]; i++) { (void) zap_remove(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, old_names[i], tx); } if (scn->scn_phys.scn_queue_obj != 0) { VERIFY(0 == dmu_object_free(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, tx)); scn->scn_phys.scn_queue_obj = 0; } /* * If we were "restarted" from a stopped state, don't bother * with anything else. */ if (scn->scn_phys.scn_state != DSS_SCANNING) return; if (complete) scn->scn_phys.scn_state = DSS_FINISHED; else scn->scn_phys.scn_state = DSS_CANCELED; spa_history_log_internal(spa, "scan done", tx, "complete=%u", complete); if (DSL_SCAN_IS_SCRUB_RESILVER(scn)) { mutex_enter(&spa->spa_scrub_lock); while (spa->spa_scrub_inflight > 0) { cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock); } mutex_exit(&spa->spa_scrub_lock); spa->spa_scrub_started = B_FALSE; spa->spa_scrub_active = B_FALSE; /* * If the scrub/resilver completed, update all DTLs to * reflect this. Whether it succeeded or not, vacate * all temporary scrub DTLs. */ vdev_dtl_reassess(spa->spa_root_vdev, tx->tx_txg, complete ? scn->scn_phys.scn_max_txg : 0, B_TRUE); if (complete) { spa_event_notify(spa, NULL, scn->scn_phys.scn_min_txg ? FM_EREPORT_ZFS_RESILVER_FINISH : FM_EREPORT_ZFS_SCRUB_FINISH); } spa_errlog_rotate(spa); /* * We may have finished replacing a device. * Let the async thread assess this and handle the detach. */ spa_async_request(spa, SPA_ASYNC_RESILVER_DONE); } scn->scn_phys.scn_end_time = gethrestime_sec(); if (spa->spa_errata == ZPOOL_ERRATA_ZOL_2094_SCRUB) spa->spa_errata = 0; } /* ARGSUSED */ static int dsl_scan_cancel_check(void *arg, dmu_tx_t *tx) { dsl_scan_t *scn = dmu_tx_pool(tx)->dp_scan; if (scn->scn_phys.scn_state != DSS_SCANNING) return (SET_ERROR(ENOENT)); return (0); } /* ARGSUSED */ static void dsl_scan_cancel_sync(void *arg, dmu_tx_t *tx) { dsl_scan_t *scn = dmu_tx_pool(tx)->dp_scan; dsl_scan_done(scn, B_FALSE, tx); dsl_scan_sync_state(scn, tx); } int dsl_scan_cancel(dsl_pool_t *dp) { return (dsl_sync_task(spa_name(dp->dp_spa), dsl_scan_cancel_check, dsl_scan_cancel_sync, NULL, 3)); } static void dsl_scan_visitbp(blkptr_t *bp, const zbookmark_phys_t *zb, dnode_phys_t *dnp, dsl_dataset_t *ds, dsl_scan_t *scn, dmu_objset_type_t ostype, dmu_tx_t *tx); inline __attribute__((always_inline)) static void dsl_scan_visitdnode( dsl_scan_t *, dsl_dataset_t *ds, dmu_objset_type_t ostype, dnode_phys_t *dnp, uint64_t object, dmu_tx_t *tx); void dsl_free(dsl_pool_t *dp, uint64_t txg, const blkptr_t *bp) { zio_free(dp->dp_spa, txg, bp); } void dsl_free_sync(zio_t *pio, dsl_pool_t *dp, uint64_t txg, const blkptr_t *bpp) { ASSERT(dsl_pool_sync_context(dp)); zio_nowait(zio_free_sync(pio, dp->dp_spa, txg, bpp, pio->io_flags)); } static uint64_t dsl_scan_ds_maxtxg(dsl_dataset_t *ds) { uint64_t smt = ds->ds_dir->dd_pool->dp_scan->scn_phys.scn_max_txg; if (dsl_dataset_is_snapshot(ds)) return (MIN(smt, ds->ds_phys->ds_creation_txg)); return (smt); } static void dsl_scan_sync_state(dsl_scan_t *scn, dmu_tx_t *tx) { VERIFY0(zap_update(scn->scn_dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_SCAN, sizeof (uint64_t), SCAN_PHYS_NUMINTS, &scn->scn_phys, tx)); } static boolean_t dsl_scan_check_pause(dsl_scan_t *scn, const zbookmark_phys_t *zb) { uint64_t elapsed_nanosecs; int mintime; /* we never skip user/group accounting objects */ if (zb && (int64_t)zb->zb_object < 0) return (B_FALSE); if (scn->scn_pausing) return (B_TRUE); /* we're already pausing */ if (!ZB_IS_ZERO(&scn->scn_phys.scn_bookmark)) return (B_FALSE); /* we're resuming */ /* We only know how to resume from level-0 blocks. */ if (zb && zb->zb_level != 0) return (B_FALSE); mintime = (scn->scn_phys.scn_func == POOL_SCAN_RESILVER) ? zfs_resilver_min_time_ms : zfs_scan_min_time_ms; elapsed_nanosecs = gethrtime() - scn->scn_sync_start_time; if (elapsed_nanosecs / NANOSEC > zfs_txg_timeout || (NSEC2MSEC(elapsed_nanosecs) > mintime && txg_sync_waiting(scn->scn_dp)) || spa_shutting_down(scn->scn_dp->dp_spa)) { if (zb) { dprintf("pausing at bookmark %llx/%llx/%llx/%llx\n", (longlong_t)zb->zb_objset, (longlong_t)zb->zb_object, (longlong_t)zb->zb_level, (longlong_t)zb->zb_blkid); scn->scn_phys.scn_bookmark = *zb; } dprintf("pausing at DDT bookmark %llx/%llx/%llx/%llx\n", (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_class, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_type, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_checksum, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_cursor); scn->scn_pausing = B_TRUE; return (B_TRUE); } return (B_FALSE); } typedef struct zil_scan_arg { dsl_pool_t *zsa_dp; zil_header_t *zsa_zh; } zil_scan_arg_t; /* ARGSUSED */ static int dsl_scan_zil_block(zilog_t *zilog, blkptr_t *bp, void *arg, uint64_t claim_txg) { zil_scan_arg_t *zsa = arg; dsl_pool_t *dp = zsa->zsa_dp; dsl_scan_t *scn = dp->dp_scan; zil_header_t *zh = zsa->zsa_zh; zbookmark_phys_t zb; if (BP_IS_HOLE(bp) || bp->blk_birth <= scn->scn_phys.scn_cur_min_txg) return (0); /* * One block ("stubby") can be allocated a long time ago; we * want to visit that one because it has been allocated * (on-disk) even if it hasn't been claimed (even though for * scrub there's nothing to do to it). */ if (claim_txg == 0 && bp->blk_birth >= spa_first_txg(dp->dp_spa)) return (0); SET_BOOKMARK(&zb, zh->zh_log.blk_cksum.zc_word[ZIL_ZC_OBJSET], ZB_ZIL_OBJECT, ZB_ZIL_LEVEL, bp->blk_cksum.zc_word[ZIL_ZC_SEQ]); VERIFY(0 == scan_funcs[scn->scn_phys.scn_func](dp, bp, &zb)); return (0); } /* ARGSUSED */ static int dsl_scan_zil_record(zilog_t *zilog, lr_t *lrc, void *arg, uint64_t claim_txg) { if (lrc->lrc_txtype == TX_WRITE) { zil_scan_arg_t *zsa = arg; dsl_pool_t *dp = zsa->zsa_dp; dsl_scan_t *scn = dp->dp_scan; zil_header_t *zh = zsa->zsa_zh; lr_write_t *lr = (lr_write_t *)lrc; blkptr_t *bp = &lr->lr_blkptr; zbookmark_phys_t zb; if (BP_IS_HOLE(bp) || bp->blk_birth <= scn->scn_phys.scn_cur_min_txg) return (0); /* * birth can be < claim_txg if this record's txg is * already txg sync'ed (but this log block contains * other records that are not synced) */ if (claim_txg == 0 || bp->blk_birth < claim_txg) return (0); SET_BOOKMARK(&zb, zh->zh_log.blk_cksum.zc_word[ZIL_ZC_OBJSET], lr->lr_foid, ZB_ZIL_LEVEL, lr->lr_offset / BP_GET_LSIZE(bp)); VERIFY(0 == scan_funcs[scn->scn_phys.scn_func](dp, bp, &zb)); } return (0); } static void dsl_scan_zil(dsl_pool_t *dp, zil_header_t *zh) { uint64_t claim_txg = zh->zh_claim_txg; zil_scan_arg_t zsa = { dp, zh }; zilog_t *zilog; /* * We only want to visit blocks that have been claimed but not yet * replayed (or, in read-only mode, blocks that *would* be claimed). */ if (claim_txg == 0 && spa_writeable(dp->dp_spa)) return; zilog = zil_alloc(dp->dp_meta_objset, zh); (void) zil_parse(zilog, dsl_scan_zil_block, dsl_scan_zil_record, &zsa, claim_txg); zil_free(zilog); } /* ARGSUSED */ static void dsl_scan_prefetch(dsl_scan_t *scn, arc_buf_t *buf, blkptr_t *bp, uint64_t objset, uint64_t object, uint64_t blkid) { zbookmark_phys_t czb; uint32_t flags = ARC_NOWAIT | ARC_PREFETCH; if (zfs_no_scrub_prefetch) return; if (BP_IS_HOLE(bp) || bp->blk_birth <= scn->scn_phys.scn_min_txg || (BP_GET_LEVEL(bp) == 0 && BP_GET_TYPE(bp) != DMU_OT_DNODE)) return; SET_BOOKMARK(&czb, objset, object, BP_GET_LEVEL(bp), blkid); (void) arc_read(scn->scn_zio_root, scn->scn_dp->dp_spa, bp, NULL, NULL, ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL | ZIO_FLAG_SCAN_THREAD, &flags, &czb); } static boolean_t dsl_scan_check_resume(dsl_scan_t *scn, const dnode_phys_t *dnp, const zbookmark_phys_t *zb) { /* * We never skip over user/group accounting objects (obj<0) */ if (!ZB_IS_ZERO(&scn->scn_phys.scn_bookmark) && (int64_t)zb->zb_object >= 0) { /* * If we already visited this bp & everything below (in * a prior txg sync), don't bother doing it again. */ if (zbookmark_is_before(dnp, zb, &scn->scn_phys.scn_bookmark)) return (B_TRUE); /* * If we found the block we're trying to resume from, or * we went past it to a different object, zero it out to * indicate that it's OK to start checking for pausing * again. */ if (bcmp(zb, &scn->scn_phys.scn_bookmark, sizeof (*zb)) == 0 || zb->zb_object > scn->scn_phys.scn_bookmark.zb_object) { dprintf("resuming at %llx/%llx/%llx/%llx\n", (longlong_t)zb->zb_objset, (longlong_t)zb->zb_object, (longlong_t)zb->zb_level, (longlong_t)zb->zb_blkid); bzero(&scn->scn_phys.scn_bookmark, sizeof (*zb)); } } return (B_FALSE); } /* * Return nonzero on i/o error. * Return new buf to write out in *bufp. */ inline __attribute__((always_inline)) static int dsl_scan_recurse(dsl_scan_t *scn, dsl_dataset_t *ds, dmu_objset_type_t ostype, dnode_phys_t *dnp, const blkptr_t *bp, const zbookmark_phys_t *zb, dmu_tx_t *tx) { dsl_pool_t *dp = scn->scn_dp; int zio_flags = ZIO_FLAG_CANFAIL | ZIO_FLAG_SCAN_THREAD; int err; if (BP_GET_LEVEL(bp) > 0) { uint32_t flags = ARC_WAIT; int i; blkptr_t *cbp; int epb = BP_GET_LSIZE(bp) >> SPA_BLKPTRSHIFT; arc_buf_t *buf; err = arc_read(NULL, dp->dp_spa, bp, arc_getbuf_func, &buf, ZIO_PRIORITY_ASYNC_READ, zio_flags, &flags, zb); if (err) { scn->scn_phys.scn_errors++; return (err); } for (i = 0, cbp = buf->b_data; i < epb; i++, cbp++) { dsl_scan_prefetch(scn, buf, cbp, zb->zb_objset, zb->zb_object, zb->zb_blkid * epb + i); } for (i = 0, cbp = buf->b_data; i < epb; i++, cbp++) { zbookmark_phys_t czb; SET_BOOKMARK(&czb, zb->zb_objset, zb->zb_object, zb->zb_level - 1, zb->zb_blkid * epb + i); dsl_scan_visitbp(cbp, &czb, dnp, ds, scn, ostype, tx); } (void) arc_buf_remove_ref(buf, &buf); } else if (BP_GET_TYPE(bp) == DMU_OT_DNODE) { uint32_t flags = ARC_WAIT; dnode_phys_t *cdnp; int i, j; int epb = BP_GET_LSIZE(bp) >> DNODE_SHIFT; arc_buf_t *buf; err = arc_read(NULL, dp->dp_spa, bp, arc_getbuf_func, &buf, ZIO_PRIORITY_ASYNC_READ, zio_flags, &flags, zb); if (err) { scn->scn_phys.scn_errors++; return (err); } for (i = 0, cdnp = buf->b_data; i < epb; i++, cdnp++) { for (j = 0; j < cdnp->dn_nblkptr; j++) { blkptr_t *cbp = &cdnp->dn_blkptr[j]; dsl_scan_prefetch(scn, buf, cbp, zb->zb_objset, zb->zb_blkid * epb + i, j); } } for (i = 0, cdnp = buf->b_data; i < epb; i++, cdnp++) { dsl_scan_visitdnode(scn, ds, ostype, cdnp, zb->zb_blkid * epb + i, tx); } (void) arc_buf_remove_ref(buf, &buf); } else if (BP_GET_TYPE(bp) == DMU_OT_OBJSET) { uint32_t flags = ARC_WAIT; objset_phys_t *osp; arc_buf_t *buf; err = arc_read(NULL, dp->dp_spa, bp, arc_getbuf_func, &buf, ZIO_PRIORITY_ASYNC_READ, zio_flags, &flags, zb); if (err) { scn->scn_phys.scn_errors++; return (err); } osp = buf->b_data; dsl_scan_visitdnode(scn, ds, osp->os_type, &osp->os_meta_dnode, DMU_META_DNODE_OBJECT, tx); if (OBJSET_BUF_HAS_USERUSED(buf)) { /* * We also always visit user/group accounting * objects, and never skip them, even if we are * pausing. This is necessary so that the space * deltas from this txg get integrated. */ dsl_scan_visitdnode(scn, ds, osp->os_type, &osp->os_groupused_dnode, DMU_GROUPUSED_OBJECT, tx); dsl_scan_visitdnode(scn, ds, osp->os_type, &osp->os_userused_dnode, DMU_USERUSED_OBJECT, tx); } (void) arc_buf_remove_ref(buf, &buf); } return (0); } inline __attribute__((always_inline)) static void dsl_scan_visitdnode(dsl_scan_t *scn, dsl_dataset_t *ds, dmu_objset_type_t ostype, dnode_phys_t *dnp, uint64_t object, dmu_tx_t *tx) { int j; for (j = 0; j < dnp->dn_nblkptr; j++) { zbookmark_phys_t czb; SET_BOOKMARK(&czb, ds ? ds->ds_object : 0, object, dnp->dn_nlevels - 1, j); dsl_scan_visitbp(&dnp->dn_blkptr[j], &czb, dnp, ds, scn, ostype, tx); } if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) { zbookmark_phys_t czb; SET_BOOKMARK(&czb, ds ? ds->ds_object : 0, object, 0, DMU_SPILL_BLKID); dsl_scan_visitbp(&dnp->dn_spill, &czb, dnp, ds, scn, ostype, tx); } } /* * The arguments are in this order because mdb can only print the * first 5; we want them to be useful. */ static void dsl_scan_visitbp(blkptr_t *bp, const zbookmark_phys_t *zb, dnode_phys_t *dnp, dsl_dataset_t *ds, dsl_scan_t *scn, dmu_objset_type_t ostype, dmu_tx_t *tx) { dsl_pool_t *dp = scn->scn_dp; blkptr_t *bp_toread; bp_toread = kmem_alloc(sizeof (blkptr_t), KM_PUSHPAGE); *bp_toread = *bp; /* ASSERT(pbuf == NULL || arc_released(pbuf)); */ if (dsl_scan_check_pause(scn, zb)) goto out; if (dsl_scan_check_resume(scn, dnp, zb)) goto out; if (BP_IS_HOLE(bp)) goto out; scn->scn_visited_this_txg++; /* * This debugging is commented out to conserve stack space. This * function is called recursively and the debugging addes several * bytes to the stack for each call. It can be commented back in * if required to debug an issue in dsl_scan_visitbp(). * * dprintf_bp(bp, * "visiting ds=%p/%llu zb=%llx/%llx/%llx/%llx bp=%p", * ds, ds ? ds->ds_object : 0, * zb->zb_objset, zb->zb_object, zb->zb_level, zb->zb_blkid, * bp); */ if (bp->blk_birth <= scn->scn_phys.scn_cur_min_txg) goto out; if (dsl_scan_recurse(scn, ds, ostype, dnp, bp_toread, zb, tx) != 0) goto out; /* * If dsl_scan_ddt() has aready visited this block, it will have * already done any translations or scrubbing, so don't call the * callback again. */ if (ddt_class_contains(dp->dp_spa, scn->scn_phys.scn_ddt_class_max, bp)) { goto out; } /* * If this block is from the future (after cur_max_txg), then we * are doing this on behalf of a deleted snapshot, and we will * revisit the future block on the next pass of this dataset. * Don't scan it now unless we need to because something * under it was modified. */ if (BP_PHYSICAL_BIRTH(bp) <= scn->scn_phys.scn_cur_max_txg) { scan_funcs[scn->scn_phys.scn_func](dp, bp, zb); } out: kmem_free(bp_toread, sizeof (blkptr_t)); } static void dsl_scan_visit_rootbp(dsl_scan_t *scn, dsl_dataset_t *ds, blkptr_t *bp, dmu_tx_t *tx) { zbookmark_phys_t zb; SET_BOOKMARK(&zb, ds ? ds->ds_object : DMU_META_OBJSET, ZB_ROOT_OBJECT, ZB_ROOT_LEVEL, ZB_ROOT_BLKID); dsl_scan_visitbp(bp, &zb, NULL, ds, scn, DMU_OST_NONE, tx); dprintf_ds(ds, "finished scan%s", ""); } void dsl_scan_ds_destroyed(dsl_dataset_t *ds, dmu_tx_t *tx) { dsl_pool_t *dp = ds->ds_dir->dd_pool; dsl_scan_t *scn = dp->dp_scan; uint64_t mintxg; if (scn->scn_phys.scn_state != DSS_SCANNING) return; if (scn->scn_phys.scn_bookmark.zb_objset == ds->ds_object) { if (dsl_dataset_is_snapshot(ds)) { /* Note, scn_cur_{min,max}_txg stays the same. */ scn->scn_phys.scn_bookmark.zb_objset = ds->ds_phys->ds_next_snap_obj; zfs_dbgmsg("destroying ds %llu; currently traversing; " "reset zb_objset to %llu", (u_longlong_t)ds->ds_object, (u_longlong_t)ds->ds_phys->ds_next_snap_obj); scn->scn_phys.scn_flags |= DSF_VISIT_DS_AGAIN; } else { SET_BOOKMARK(&scn->scn_phys.scn_bookmark, ZB_DESTROYED_OBJSET, 0, 0, 0); zfs_dbgmsg("destroying ds %llu; currently traversing; " "reset bookmark to -1,0,0,0", (u_longlong_t)ds->ds_object); } } else if (zap_lookup_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, &mintxg) == 0) { ASSERT3U(ds->ds_phys->ds_num_children, <=, 1); VERIFY3U(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, tx)); if (dsl_dataset_is_snapshot(ds)) { /* * We keep the same mintxg; it could be > * ds_creation_txg if the previous snapshot was * deleted too. */ VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_phys->ds_next_snap_obj, mintxg, tx) == 0); zfs_dbgmsg("destroying ds %llu; in queue; " "replacing with %llu", (u_longlong_t)ds->ds_object, (u_longlong_t)ds->ds_phys->ds_next_snap_obj); } else { zfs_dbgmsg("destroying ds %llu; in queue; removing", (u_longlong_t)ds->ds_object); } } else { zfs_dbgmsg("destroying ds %llu; ignoring", (u_longlong_t)ds->ds_object); } /* * dsl_scan_sync() should be called after this, and should sync * out our changed state, but just to be safe, do it here. */ dsl_scan_sync_state(scn, tx); } void dsl_scan_ds_snapshotted(dsl_dataset_t *ds, dmu_tx_t *tx) { dsl_pool_t *dp = ds->ds_dir->dd_pool; dsl_scan_t *scn = dp->dp_scan; uint64_t mintxg; if (scn->scn_phys.scn_state != DSS_SCANNING) return; ASSERT(ds->ds_phys->ds_prev_snap_obj != 0); if (scn->scn_phys.scn_bookmark.zb_objset == ds->ds_object) { scn->scn_phys.scn_bookmark.zb_objset = ds->ds_phys->ds_prev_snap_obj; zfs_dbgmsg("snapshotting ds %llu; currently traversing; " "reset zb_objset to %llu", (u_longlong_t)ds->ds_object, (u_longlong_t)ds->ds_phys->ds_prev_snap_obj); } else if (zap_lookup_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, &mintxg) == 0) { VERIFY3U(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, tx)); VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_phys->ds_prev_snap_obj, mintxg, tx) == 0); zfs_dbgmsg("snapshotting ds %llu; in queue; " "replacing with %llu", (u_longlong_t)ds->ds_object, (u_longlong_t)ds->ds_phys->ds_prev_snap_obj); } dsl_scan_sync_state(scn, tx); } void dsl_scan_ds_clone_swapped(dsl_dataset_t *ds1, dsl_dataset_t *ds2, dmu_tx_t *tx) { dsl_pool_t *dp = ds1->ds_dir->dd_pool; dsl_scan_t *scn = dp->dp_scan; uint64_t mintxg; if (scn->scn_phys.scn_state != DSS_SCANNING) return; if (scn->scn_phys.scn_bookmark.zb_objset == ds1->ds_object) { scn->scn_phys.scn_bookmark.zb_objset = ds2->ds_object; zfs_dbgmsg("clone_swap ds %llu; currently traversing; " "reset zb_objset to %llu", (u_longlong_t)ds1->ds_object, (u_longlong_t)ds2->ds_object); } else if (scn->scn_phys.scn_bookmark.zb_objset == ds2->ds_object) { scn->scn_phys.scn_bookmark.zb_objset = ds1->ds_object; zfs_dbgmsg("clone_swap ds %llu; currently traversing; " "reset zb_objset to %llu", (u_longlong_t)ds2->ds_object, (u_longlong_t)ds1->ds_object); } if (zap_lookup_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds1->ds_object, &mintxg) == 0) { int err; ASSERT3U(mintxg, ==, ds1->ds_phys->ds_prev_snap_txg); ASSERT3U(mintxg, ==, ds2->ds_phys->ds_prev_snap_txg); VERIFY3U(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds1->ds_object, tx)); err = zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds2->ds_object, mintxg, tx); VERIFY(err == 0 || err == EEXIST); if (err == EEXIST) { /* Both were there to begin with */ VERIFY(0 == zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds1->ds_object, mintxg, tx)); } zfs_dbgmsg("clone_swap ds %llu; in queue; " "replacing with %llu", (u_longlong_t)ds1->ds_object, (u_longlong_t)ds2->ds_object); } else if (zap_lookup_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds2->ds_object, &mintxg) == 0) { ASSERT3U(mintxg, ==, ds1->ds_phys->ds_prev_snap_txg); ASSERT3U(mintxg, ==, ds2->ds_phys->ds_prev_snap_txg); VERIFY3U(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds2->ds_object, tx)); VERIFY(0 == zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds1->ds_object, mintxg, tx)); zfs_dbgmsg("clone_swap ds %llu; in queue; " "replacing with %llu", (u_longlong_t)ds2->ds_object, (u_longlong_t)ds1->ds_object); } dsl_scan_sync_state(scn, tx); } struct enqueue_clones_arg { dmu_tx_t *tx; uint64_t originobj; }; /* ARGSUSED */ static int enqueue_clones_cb(dsl_pool_t *dp, dsl_dataset_t *hds, void *arg) { struct enqueue_clones_arg *eca = arg; dsl_dataset_t *ds; int err; dsl_scan_t *scn = dp->dp_scan; if (hds->ds_dir->dd_phys->dd_origin_obj != eca->originobj) return (0); err = dsl_dataset_hold_obj(dp, hds->ds_object, FTAG, &ds); if (err) return (err); while (ds->ds_phys->ds_prev_snap_obj != eca->originobj) { dsl_dataset_t *prev; err = dsl_dataset_hold_obj(dp, ds->ds_phys->ds_prev_snap_obj, FTAG, &prev); dsl_dataset_rele(ds, FTAG); if (err) return (err); ds = prev; } VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, ds->ds_phys->ds_prev_snap_txg, eca->tx) == 0); dsl_dataset_rele(ds, FTAG); return (0); } static void dsl_scan_visitds(dsl_scan_t *scn, uint64_t dsobj, dmu_tx_t *tx) { dsl_pool_t *dp = scn->scn_dp; dsl_dataset_t *ds; objset_t *os; char *dsname; VERIFY3U(0, ==, dsl_dataset_hold_obj(dp, dsobj, FTAG, &ds)); if (dmu_objset_from_ds(ds, &os)) goto out; /* * Only the ZIL in the head (non-snapshot) is valid. Even though * snapshots can have ZIL block pointers (which may be the same * BP as in the head), they must be ignored. So we traverse the * ZIL here, rather than in scan_recurse(), because the regular * snapshot block-sharing rules don't apply to it. */ if (DSL_SCAN_IS_SCRUB_RESILVER(scn) && !dsl_dataset_is_snapshot(ds)) dsl_scan_zil(dp, &os->os_zil_header); /* * Iterate over the bps in this ds. */ dmu_buf_will_dirty(ds->ds_dbuf, tx); dsl_scan_visit_rootbp(scn, ds, &ds->ds_phys->ds_bp, tx); dsname = kmem_alloc(ZFS_MAXNAMELEN, KM_PUSHPAGE); dsl_dataset_name(ds, dsname); zfs_dbgmsg("scanned dataset %llu (%s) with min=%llu max=%llu; " "pausing=%u", (longlong_t)dsobj, dsname, (longlong_t)scn->scn_phys.scn_cur_min_txg, (longlong_t)scn->scn_phys.scn_cur_max_txg, (int)scn->scn_pausing); kmem_free(dsname, ZFS_MAXNAMELEN); if (scn->scn_pausing) goto out; /* * We've finished this pass over this dataset. */ /* * If we did not completely visit this dataset, do another pass. */ if (scn->scn_phys.scn_flags & DSF_VISIT_DS_AGAIN) { zfs_dbgmsg("incomplete pass; visiting again"); scn->scn_phys.scn_flags &= ~DSF_VISIT_DS_AGAIN; VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, scn->scn_phys.scn_cur_max_txg, tx) == 0); goto out; } /* * Add descendent datasets to work queue. */ if (ds->ds_phys->ds_next_snap_obj != 0) { VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_phys->ds_next_snap_obj, ds->ds_phys->ds_creation_txg, tx) == 0); } if (ds->ds_phys->ds_num_children > 1) { boolean_t usenext = B_FALSE; if (ds->ds_phys->ds_next_clones_obj != 0) { uint64_t count; /* * A bug in a previous version of the code could * cause upgrade_clones_cb() to not set * ds_next_snap_obj when it should, leading to a * missing entry. Therefore we can only use the * next_clones_obj when its count is correct. */ int err = zap_count(dp->dp_meta_objset, ds->ds_phys->ds_next_clones_obj, &count); if (err == 0 && count == ds->ds_phys->ds_num_children - 1) usenext = B_TRUE; } if (usenext) { VERIFY0(zap_join_key(dp->dp_meta_objset, ds->ds_phys->ds_next_clones_obj, scn->scn_phys.scn_queue_obj, ds->ds_phys->ds_creation_txg, tx)); } else { struct enqueue_clones_arg eca; eca.tx = tx; eca.originobj = ds->ds_object; VERIFY0(dmu_objset_find_dp(dp, dp->dp_root_dir_obj, enqueue_clones_cb, &eca, DS_FIND_CHILDREN)); } } out: dsl_dataset_rele(ds, FTAG); } /* ARGSUSED */ static int enqueue_cb(dsl_pool_t *dp, dsl_dataset_t *hds, void *arg) { dmu_tx_t *tx = arg; dsl_dataset_t *ds; int err; dsl_scan_t *scn = dp->dp_scan; err = dsl_dataset_hold_obj(dp, hds->ds_object, FTAG, &ds); if (err) return (err); while (ds->ds_phys->ds_prev_snap_obj != 0) { dsl_dataset_t *prev; err = dsl_dataset_hold_obj(dp, ds->ds_phys->ds_prev_snap_obj, FTAG, &prev); if (err) { dsl_dataset_rele(ds, FTAG); return (err); } /* * If this is a clone, we don't need to worry about it for now. */ if (prev->ds_phys->ds_next_snap_obj != ds->ds_object) { dsl_dataset_rele(ds, FTAG); dsl_dataset_rele(prev, FTAG); return (0); } dsl_dataset_rele(ds, FTAG); ds = prev; } VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, ds->ds_phys->ds_prev_snap_txg, tx) == 0); dsl_dataset_rele(ds, FTAG); return (0); } /* * Scrub/dedup interaction. * * If there are N references to a deduped block, we don't want to scrub it * N times -- ideally, we should scrub it exactly once. * * We leverage the fact that the dde's replication class (enum ddt_class) * is ordered from highest replication class (DDT_CLASS_DITTO) to lowest * (DDT_CLASS_UNIQUE) so that we may walk the DDT in that order. * * To prevent excess scrubbing, the scrub begins by walking the DDT * to find all blocks with refcnt > 1, and scrubs each of these once. * Since there are two replication classes which contain blocks with * refcnt > 1, we scrub the highest replication class (DDT_CLASS_DITTO) first. * Finally the top-down scrub begins, only visiting blocks with refcnt == 1. * * There would be nothing more to say if a block's refcnt couldn't change * during a scrub, but of course it can so we must account for changes * in a block's replication class. * * Here's an example of what can occur: * * If a block has refcnt > 1 during the DDT scrub phase, but has refcnt == 1 * when visited during the top-down scrub phase, it will be scrubbed twice. * This negates our scrub optimization, but is otherwise harmless. * * If a block has refcnt == 1 during the DDT scrub phase, but has refcnt > 1 * on each visit during the top-down scrub phase, it will never be scrubbed. * To catch this, ddt_sync_entry() notifies the scrub code whenever a block's * reference class transitions to a higher level (i.e DDT_CLASS_UNIQUE to * DDT_CLASS_DUPLICATE); if it transitions from refcnt == 1 to refcnt > 1 * while a scrub is in progress, it scrubs the block right then. */ static void dsl_scan_ddt(dsl_scan_t *scn, dmu_tx_t *tx) { ddt_bookmark_t *ddb = &scn->scn_phys.scn_ddt_bookmark; ddt_entry_t dde; int error; uint64_t n = 0; bzero(&dde, sizeof (ddt_entry_t)); while ((error = ddt_walk(scn->scn_dp->dp_spa, ddb, &dde)) == 0) { ddt_t *ddt; if (ddb->ddb_class > scn->scn_phys.scn_ddt_class_max) break; dprintf("visiting ddb=%llu/%llu/%llu/%llx\n", (longlong_t)ddb->ddb_class, (longlong_t)ddb->ddb_type, (longlong_t)ddb->ddb_checksum, (longlong_t)ddb->ddb_cursor); /* There should be no pending changes to the dedup table */ ddt = scn->scn_dp->dp_spa->spa_ddt[ddb->ddb_checksum]; ASSERT(avl_first(&ddt->ddt_tree) == NULL); dsl_scan_ddt_entry(scn, ddb->ddb_checksum, &dde, tx); n++; if (dsl_scan_check_pause(scn, NULL)) break; } zfs_dbgmsg("scanned %llu ddt entries with class_max = %u; pausing=%u", (longlong_t)n, (int)scn->scn_phys.scn_ddt_class_max, (int)scn->scn_pausing); ASSERT(error == 0 || error == ENOENT); ASSERT(error != ENOENT || ddb->ddb_class > scn->scn_phys.scn_ddt_class_max); } /* ARGSUSED */ void dsl_scan_ddt_entry(dsl_scan_t *scn, enum zio_checksum checksum, ddt_entry_t *dde, dmu_tx_t *tx) { const ddt_key_t *ddk = &dde->dde_key; ddt_phys_t *ddp = dde->dde_phys; blkptr_t bp; zbookmark_phys_t zb = { 0 }; int p; if (scn->scn_phys.scn_state != DSS_SCANNING) return; for (p = 0; p < DDT_PHYS_TYPES; p++, ddp++) { if (ddp->ddp_phys_birth == 0 || ddp->ddp_phys_birth > scn->scn_phys.scn_max_txg) continue; ddt_bp_create(checksum, ddk, ddp, &bp); scn->scn_visited_this_txg++; scan_funcs[scn->scn_phys.scn_func](scn->scn_dp, &bp, &zb); } } static void dsl_scan_visit(dsl_scan_t *scn, dmu_tx_t *tx) { dsl_pool_t *dp = scn->scn_dp; zap_cursor_t *zc; zap_attribute_t *za; if (scn->scn_phys.scn_ddt_bookmark.ddb_class <= scn->scn_phys.scn_ddt_class_max) { scn->scn_phys.scn_cur_min_txg = scn->scn_phys.scn_min_txg; scn->scn_phys.scn_cur_max_txg = scn->scn_phys.scn_max_txg; dsl_scan_ddt(scn, tx); if (scn->scn_pausing) return; } if (scn->scn_phys.scn_bookmark.zb_objset == DMU_META_OBJSET) { /* First do the MOS & ORIGIN */ scn->scn_phys.scn_cur_min_txg = scn->scn_phys.scn_min_txg; scn->scn_phys.scn_cur_max_txg = scn->scn_phys.scn_max_txg; dsl_scan_visit_rootbp(scn, NULL, &dp->dp_meta_rootbp, tx); spa_set_rootblkptr(dp->dp_spa, &dp->dp_meta_rootbp); if (scn->scn_pausing) return; if (spa_version(dp->dp_spa) < SPA_VERSION_DSL_SCRUB) { VERIFY0(dmu_objset_find_dp(dp, dp->dp_root_dir_obj, enqueue_cb, tx, DS_FIND_CHILDREN)); } else { dsl_scan_visitds(scn, dp->dp_origin_snap->ds_object, tx); } ASSERT(!scn->scn_pausing); } else if (scn->scn_phys.scn_bookmark.zb_objset != ZB_DESTROYED_OBJSET) { /* * If we were paused, continue from here. Note if the * ds we were paused on was deleted, the zb_objset may * be -1, so we will skip this and find a new objset * below. */ dsl_scan_visitds(scn, scn->scn_phys.scn_bookmark.zb_objset, tx); if (scn->scn_pausing) return; } /* * In case we were paused right at the end of the ds, zero the * bookmark so we don't think that we're still trying to resume. */ bzero(&scn->scn_phys.scn_bookmark, sizeof (zbookmark_phys_t)); zc = kmem_alloc(sizeof (zap_cursor_t), KM_PUSHPAGE); za = kmem_alloc(sizeof (zap_attribute_t), KM_PUSHPAGE); /* keep pulling things out of the zap-object-as-queue */ while (zap_cursor_init(zc, dp->dp_meta_objset, scn->scn_phys.scn_queue_obj), zap_cursor_retrieve(zc, za) == 0) { dsl_dataset_t *ds; uint64_t dsobj; dsobj = strtonum(za->za_name, NULL); VERIFY3U(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, dsobj, tx)); /* Set up min/max txg */ VERIFY3U(0, ==, dsl_dataset_hold_obj(dp, dsobj, FTAG, &ds)); if (za->za_first_integer != 0) { scn->scn_phys.scn_cur_min_txg = MAX(scn->scn_phys.scn_min_txg, za->za_first_integer); } else { scn->scn_phys.scn_cur_min_txg = MAX(scn->scn_phys.scn_min_txg, ds->ds_phys->ds_prev_snap_txg); } scn->scn_phys.scn_cur_max_txg = dsl_scan_ds_maxtxg(ds); dsl_dataset_rele(ds, FTAG); dsl_scan_visitds(scn, dsobj, tx); zap_cursor_fini(zc); if (scn->scn_pausing) goto out; } zap_cursor_fini(zc); out: kmem_free(za, sizeof (zap_attribute_t)); kmem_free(zc, sizeof (zap_cursor_t)); } static boolean_t dsl_scan_free_should_pause(dsl_scan_t *scn) { uint64_t elapsed_nanosecs; if (zfs_recover) return (B_FALSE); if (scn->scn_visited_this_txg >= zfs_free_max_blocks) return (B_TRUE); elapsed_nanosecs = gethrtime() - scn->scn_sync_start_time; return (elapsed_nanosecs / NANOSEC > zfs_txg_timeout || (NSEC2MSEC(elapsed_nanosecs) > zfs_free_min_time_ms && txg_sync_waiting(scn->scn_dp)) || spa_shutting_down(scn->scn_dp->dp_spa)); } static int dsl_scan_free_block_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx) { dsl_scan_t *scn = arg; if (!scn->scn_is_bptree || (BP_GET_LEVEL(bp) == 0 && BP_GET_TYPE(bp) != DMU_OT_OBJSET)) { if (dsl_scan_free_should_pause(scn)) return (SET_ERROR(ERESTART)); } zio_nowait(zio_free_sync(scn->scn_zio_root, scn->scn_dp->dp_spa, dmu_tx_get_txg(tx), bp, 0)); dsl_dir_diduse_space(tx->tx_pool->dp_free_dir, DD_USED_HEAD, -bp_get_dsize_sync(scn->scn_dp->dp_spa, bp), -BP_GET_PSIZE(bp), -BP_GET_UCSIZE(bp), tx); scn->scn_visited_this_txg++; return (0); } boolean_t dsl_scan_active(dsl_scan_t *scn) { spa_t *spa = scn->scn_dp->dp_spa; uint64_t used = 0, comp, uncomp; if (spa->spa_load_state != SPA_LOAD_NONE) return (B_FALSE); if (spa_shutting_down(spa)) return (B_FALSE); if (scn->scn_phys.scn_state == DSS_SCANNING || (scn->scn_async_destroying && !scn->scn_async_stalled)) return (B_TRUE); if (spa_version(scn->scn_dp->dp_spa) >= SPA_VERSION_DEADLISTS) { (void) bpobj_space(&scn->scn_dp->dp_free_bpobj, &used, &comp, &uncomp); } return (used != 0); } void dsl_scan_sync(dsl_pool_t *dp, dmu_tx_t *tx) { dsl_scan_t *scn = dp->dp_scan; spa_t *spa = dp->dp_spa; int err = 0; /* * Check for scn_restart_txg before checking spa_load_state, so * that we can restart an old-style scan while the pool is being * imported (see dsl_scan_init). */ if (scn->scn_restart_txg != 0 && scn->scn_restart_txg <= tx->tx_txg) { pool_scan_func_t func = POOL_SCAN_SCRUB; dsl_scan_done(scn, B_FALSE, tx); if (vdev_resilver_needed(spa->spa_root_vdev, NULL, NULL)) func = POOL_SCAN_RESILVER; zfs_dbgmsg("restarting scan func=%u txg=%llu", func, tx->tx_txg); dsl_scan_setup_sync(&func, tx); } /* * If the scan is inactive due to a stalled async destroy, try again. */ if ((!scn->scn_async_stalled && !dsl_scan_active(scn)) || spa_sync_pass(dp->dp_spa) > 1) return; scn->scn_visited_this_txg = 0; scn->scn_pausing = B_FALSE; scn->scn_sync_start_time = gethrtime(); spa->spa_scrub_active = B_TRUE; /* * First process the async destroys. If we pause, don't do * any scrubbing or resilvering. This ensures that there are no * async destroys while we are scanning, so the scan code doesn't * have to worry about traversing it. It is also faster to free the * blocks than to scrub them. */ if (spa_version(dp->dp_spa) >= SPA_VERSION_DEADLISTS) { scn->scn_is_bptree = B_FALSE; scn->scn_zio_root = zio_root(dp->dp_spa, NULL, NULL, ZIO_FLAG_MUSTSUCCEED); err = bpobj_iterate(&dp->dp_free_bpobj, dsl_scan_free_block_cb, scn, tx); VERIFY3U(0, ==, zio_wait(scn->scn_zio_root)); if (err != 0 && err != ERESTART) zfs_panic_recover("error %u from bpobj_iterate()", err); } if (err == 0 && spa_feature_is_active(spa, SPA_FEATURE_ASYNC_DESTROY)) { ASSERT(scn->scn_async_destroying); scn->scn_is_bptree = B_TRUE; scn->scn_zio_root = zio_root(dp->dp_spa, NULL, NULL, ZIO_FLAG_MUSTSUCCEED); err = bptree_iterate(dp->dp_meta_objset, dp->dp_bptree_obj, B_TRUE, dsl_scan_free_block_cb, scn, tx); VERIFY0(zio_wait(scn->scn_zio_root)); if (err == EIO || err == ECKSUM) { err = 0; } else if (err != 0 && err != ERESTART) { zfs_panic_recover("error %u from " "traverse_dataset_destroyed()", err); } if (bptree_is_empty(dp->dp_meta_objset, dp->dp_bptree_obj)) { /* finished; deactivate async destroy feature */ spa_feature_decr(spa, SPA_FEATURE_ASYNC_DESTROY, tx); ASSERT(!spa_feature_is_active(spa, SPA_FEATURE_ASYNC_DESTROY)); VERIFY0(zap_remove(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_BPTREE_OBJ, tx)); VERIFY0(bptree_free(dp->dp_meta_objset, dp->dp_bptree_obj, tx)); dp->dp_bptree_obj = 0; scn->scn_async_destroying = B_FALSE; } else { /* * If we didn't make progress, mark the async destroy as * stalled, so that we will not initiate a spa_sync() on * its behalf. */ scn->scn_async_stalled = (scn->scn_visited_this_txg == 0); } } if (scn->scn_visited_this_txg) { zfs_dbgmsg("freed %llu blocks in %llums from " "free_bpobj/bptree txg %llu; err=%u", (longlong_t)scn->scn_visited_this_txg, (longlong_t) NSEC2MSEC(gethrtime() - scn->scn_sync_start_time), (longlong_t)tx->tx_txg, err); scn->scn_visited_this_txg = 0; /* * Write out changes to the DDT that may be required as a * result of the blocks freed. This ensures that the DDT * is clean when a scrub/resilver runs. */ ddt_sync(spa, tx->tx_txg); } if (err != 0) return; if (!scn->scn_async_destroying && zfs_free_leak_on_eio && (dp->dp_free_dir->dd_phys->dd_used_bytes != 0 || dp->dp_free_dir->dd_phys->dd_compressed_bytes != 0 || dp->dp_free_dir->dd_phys->dd_uncompressed_bytes != 0)) { /* * We have finished background destroying, but there is still * some space left in the dp_free_dir. Transfer this leaked * space to the dp_leak_dir. */ if (dp->dp_leak_dir == NULL) { rrw_enter(&dp->dp_config_rwlock, RW_WRITER, FTAG); (void) dsl_dir_create_sync(dp, dp->dp_root_dir, LEAK_DIR_NAME, tx); VERIFY0(dsl_pool_open_special_dir(dp, LEAK_DIR_NAME, &dp->dp_leak_dir)); rrw_exit(&dp->dp_config_rwlock, FTAG); } dsl_dir_diduse_space(dp->dp_leak_dir, DD_USED_HEAD, dp->dp_free_dir->dd_phys->dd_used_bytes, dp->dp_free_dir->dd_phys->dd_compressed_bytes, dp->dp_free_dir->dd_phys->dd_uncompressed_bytes, tx); dsl_dir_diduse_space(dp->dp_free_dir, DD_USED_HEAD, -dp->dp_free_dir->dd_phys->dd_used_bytes, -dp->dp_free_dir->dd_phys->dd_compressed_bytes, -dp->dp_free_dir->dd_phys->dd_uncompressed_bytes, tx); } if (!scn->scn_async_destroying) { /* finished; verify that space accounting went to zero */ ASSERT0(dp->dp_free_dir->dd_phys->dd_used_bytes); ASSERT0(dp->dp_free_dir->dd_phys->dd_compressed_bytes); ASSERT0(dp->dp_free_dir->dd_phys->dd_uncompressed_bytes); } if (scn->scn_phys.scn_state != DSS_SCANNING) return; if (scn->scn_done_txg == tx->tx_txg) { ASSERT(!scn->scn_pausing); /* finished with scan. */ zfs_dbgmsg("txg %llu scan complete", tx->tx_txg); dsl_scan_done(scn, B_TRUE, tx); ASSERT3U(spa->spa_scrub_inflight, ==, 0); dsl_scan_sync_state(scn, tx); return; } if (scn->scn_phys.scn_ddt_bookmark.ddb_class <= scn->scn_phys.scn_ddt_class_max) { zfs_dbgmsg("doing scan sync txg %llu; " "ddt bm=%llu/%llu/%llu/%llx", (longlong_t)tx->tx_txg, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_class, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_type, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_checksum, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_cursor); ASSERT(scn->scn_phys.scn_bookmark.zb_objset == 0); ASSERT(scn->scn_phys.scn_bookmark.zb_object == 0); ASSERT(scn->scn_phys.scn_bookmark.zb_level == 0); ASSERT(scn->scn_phys.scn_bookmark.zb_blkid == 0); } else { zfs_dbgmsg("doing scan sync txg %llu; bm=%llu/%llu/%llu/%llu", (longlong_t)tx->tx_txg, (longlong_t)scn->scn_phys.scn_bookmark.zb_objset, (longlong_t)scn->scn_phys.scn_bookmark.zb_object, (longlong_t)scn->scn_phys.scn_bookmark.zb_level, (longlong_t)scn->scn_phys.scn_bookmark.zb_blkid); } scn->scn_zio_root = zio_root(dp->dp_spa, NULL, NULL, ZIO_FLAG_CANFAIL); dsl_pool_config_enter(dp, FTAG); dsl_scan_visit(scn, tx); dsl_pool_config_exit(dp, FTAG); (void) zio_wait(scn->scn_zio_root); scn->scn_zio_root = NULL; zfs_dbgmsg("visited %llu blocks in %llums", (longlong_t)scn->scn_visited_this_txg, (longlong_t)NSEC2MSEC(gethrtime() - scn->scn_sync_start_time)); if (!scn->scn_pausing) { scn->scn_done_txg = tx->tx_txg + 1; zfs_dbgmsg("txg %llu traversal complete, waiting till txg %llu", tx->tx_txg, scn->scn_done_txg); } if (DSL_SCAN_IS_SCRUB_RESILVER(scn)) { mutex_enter(&spa->spa_scrub_lock); while (spa->spa_scrub_inflight > 0) { cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock); } mutex_exit(&spa->spa_scrub_lock); } dsl_scan_sync_state(scn, tx); } /* * This will start a new scan, or restart an existing one. */ void dsl_resilver_restart(dsl_pool_t *dp, uint64_t txg) { if (txg == 0) { dmu_tx_t *tx; tx = dmu_tx_create_dd(dp->dp_mos_dir); VERIFY(0 == dmu_tx_assign(tx, TXG_WAIT)); txg = dmu_tx_get_txg(tx); dp->dp_scan->scn_restart_txg = txg; dmu_tx_commit(tx); } else { dp->dp_scan->scn_restart_txg = txg; } zfs_dbgmsg("restarting resilver txg=%llu", txg); } boolean_t dsl_scan_resilvering(dsl_pool_t *dp) { return (dp->dp_scan->scn_phys.scn_state == DSS_SCANNING && dp->dp_scan->scn_phys.scn_func == POOL_SCAN_RESILVER); } /* * scrub consumers */ static void count_block(zfs_all_blkstats_t *zab, const blkptr_t *bp) { int i; /* * If we resume after a reboot, zab will be NULL; don't record * incomplete stats in that case. */ if (zab == NULL) return; for (i = 0; i < 4; i++) { int l = (i < 2) ? BP_GET_LEVEL(bp) : DN_MAX_LEVELS; int t = (i & 1) ? BP_GET_TYPE(bp) : DMU_OT_TOTAL; int equal; zfs_blkstat_t *zb; if (t & DMU_OT_NEWTYPE) t = DMU_OT_OTHER; zb = &zab->zab_type[l][t]; zb->zb_count++; zb->zb_asize += BP_GET_ASIZE(bp); zb->zb_lsize += BP_GET_LSIZE(bp); zb->zb_psize += BP_GET_PSIZE(bp); zb->zb_gangs += BP_COUNT_GANG(bp); switch (BP_GET_NDVAS(bp)) { case 2: if (DVA_GET_VDEV(&bp->blk_dva[0]) == DVA_GET_VDEV(&bp->blk_dva[1])) zb->zb_ditto_2_of_2_samevdev++; break; case 3: equal = (DVA_GET_VDEV(&bp->blk_dva[0]) == DVA_GET_VDEV(&bp->blk_dva[1])) + (DVA_GET_VDEV(&bp->blk_dva[0]) == DVA_GET_VDEV(&bp->blk_dva[2])) + (DVA_GET_VDEV(&bp->blk_dva[1]) == DVA_GET_VDEV(&bp->blk_dva[2])); if (equal == 1) zb->zb_ditto_2_of_3_samevdev++; else if (equal == 3) zb->zb_ditto_3_of_3_samevdev++; break; } } } static void dsl_scan_scrub_done(zio_t *zio) { spa_t *spa = zio->io_spa; zio_data_buf_free(zio->io_data, zio->io_size); mutex_enter(&spa->spa_scrub_lock); spa->spa_scrub_inflight--; cv_broadcast(&spa->spa_scrub_io_cv); if (zio->io_error && (zio->io_error != ECKSUM || !(zio->io_flags & ZIO_FLAG_SPECULATIVE))) { spa->spa_dsl_pool->dp_scan->scn_phys.scn_errors++; } mutex_exit(&spa->spa_scrub_lock); } static int dsl_scan_scrub_cb(dsl_pool_t *dp, const blkptr_t *bp, const zbookmark_phys_t *zb) { dsl_scan_t *scn = dp->dp_scan; size_t size = BP_GET_PSIZE(bp); spa_t *spa = dp->dp_spa; uint64_t phys_birth = BP_PHYSICAL_BIRTH(bp); boolean_t needs_io = B_FALSE; int zio_flags = ZIO_FLAG_SCAN_THREAD | ZIO_FLAG_RAW | ZIO_FLAG_CANFAIL; int scan_delay = 0; int d; if (phys_birth <= scn->scn_phys.scn_min_txg || phys_birth >= scn->scn_phys.scn_max_txg) return (0); count_block(dp->dp_blkstats, bp); if (BP_IS_EMBEDDED(bp)) return (0); ASSERT(DSL_SCAN_IS_SCRUB_RESILVER(scn)); if (scn->scn_phys.scn_func == POOL_SCAN_SCRUB) { zio_flags |= ZIO_FLAG_SCRUB; needs_io = B_TRUE; scan_delay = zfs_scrub_delay; } else { ASSERT3U(scn->scn_phys.scn_func, ==, POOL_SCAN_RESILVER); zio_flags |= ZIO_FLAG_RESILVER; needs_io = B_FALSE; scan_delay = zfs_resilver_delay; } /* If it's an intent log block, failure is expected. */ if (zb->zb_level == ZB_ZIL_LEVEL) zio_flags |= ZIO_FLAG_SPECULATIVE; for (d = 0; d < BP_GET_NDVAS(bp); d++) { vdev_t *vd = vdev_lookup_top(spa, DVA_GET_VDEV(&bp->blk_dva[d])); /* * Keep track of how much data we've examined so that * zpool(1M) status can make useful progress reports. */ scn->scn_phys.scn_examined += DVA_GET_ASIZE(&bp->blk_dva[d]); spa->spa_scan_pass_exam += DVA_GET_ASIZE(&bp->blk_dva[d]); /* if it's a resilver, this may not be in the target range */ if (!needs_io) { if (DVA_GET_GANG(&bp->blk_dva[d])) { /* * Gang members may be spread across multiple * vdevs, so the best estimate we have is the * scrub range, which has already been checked. * XXX -- it would be better to change our * allocation policy to ensure that all * gang members reside on the same vdev. */ needs_io = B_TRUE; } else { needs_io = vdev_dtl_contains(vd, DTL_PARTIAL, phys_birth, 1); } } } if (needs_io && !zfs_no_scrub_io) { vdev_t *rvd = spa->spa_root_vdev; uint64_t maxinflight = rvd->vdev_children * zfs_top_maxinflight; void *data = zio_data_buf_alloc(size); mutex_enter(&spa->spa_scrub_lock); while (spa->spa_scrub_inflight >= maxinflight) cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock); spa->spa_scrub_inflight++; mutex_exit(&spa->spa_scrub_lock); /* * If we're seeing recent (zfs_scan_idle) "important" I/Os * then throttle our workload to limit the impact of a scan. */ if (ddi_get_lbolt64() - spa->spa_last_io <= zfs_scan_idle) delay(scan_delay); zio_nowait(zio_read(NULL, spa, bp, data, size, dsl_scan_scrub_done, NULL, ZIO_PRIORITY_SCRUB, zio_flags, zb)); } /* do not relocate this block */ return (0); } int dsl_scan(dsl_pool_t *dp, pool_scan_func_t func) { spa_t *spa = dp->dp_spa; /* * Purge all vdev caches and probe all devices. We do this here * rather than in sync context because this requires a writer lock * on the spa_config lock, which we can't do from sync context. The * spa_scrub_reopen flag indicates that vdev_open() should not * attempt to start another scrub. */ spa_vdev_state_enter(spa, SCL_NONE); spa->spa_scrub_reopen = B_TRUE; vdev_reopen(spa->spa_root_vdev); spa->spa_scrub_reopen = B_FALSE; (void) spa_vdev_state_exit(spa, NULL, 0); return (dsl_sync_task(spa_name(spa), dsl_scan_setup_check, dsl_scan_setup_sync, &func, 0)); } #if defined(_KERNEL) && defined(HAVE_SPL) module_param(zfs_top_maxinflight, int, 0644); MODULE_PARM_DESC(zfs_top_maxinflight, "Max I/Os per top-level"); module_param(zfs_resilver_delay, int, 0644); MODULE_PARM_DESC(zfs_resilver_delay, "Number of ticks to delay resilver"); module_param(zfs_scrub_delay, int, 0644); MODULE_PARM_DESC(zfs_scrub_delay, "Number of ticks to delay scrub"); module_param(zfs_scan_idle, int, 0644); MODULE_PARM_DESC(zfs_scan_idle, "Idle window in clock ticks"); module_param(zfs_scan_min_time_ms, int, 0644); MODULE_PARM_DESC(zfs_scan_min_time_ms, "Min millisecs to scrub per txg"); module_param(zfs_free_min_time_ms, int, 0644); MODULE_PARM_DESC(zfs_free_min_time_ms, "Min millisecs to free per txg"); module_param(zfs_resilver_min_time_ms, int, 0644); MODULE_PARM_DESC(zfs_resilver_min_time_ms, "Min millisecs to resilver per txg"); module_param(zfs_no_scrub_io, int, 0644); MODULE_PARM_DESC(zfs_no_scrub_io, "Set to disable scrub I/O"); module_param(zfs_no_scrub_prefetch, int, 0644); MODULE_PARM_DESC(zfs_no_scrub_prefetch, "Set to disable scrub prefetching"); module_param(zfs_free_max_blocks, ulong, 0644); MODULE_PARM_DESC(zfs_free_max_blocks, "Max number of blocks freed in one txg"); #endif diff --git a/module/zfs/spa.c b/module/zfs/spa.c index 776beda1179e..88641603abc9 100644 --- a/module/zfs/spa.c +++ b/module/zfs/spa.c @@ -1,6700 +1,6700 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2013 by Delphix. All rights reserved. * Copyright (c) 2013, 2014, Nexenta Systems, Inc. All rights reserved. */ /* * SPA: Storage Pool Allocator * * This file contains all the routines used when modifying on-disk SPA state. * This includes opening, importing, destroying, exporting a pool, and syncing a * pool. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef _KERNEL #include #include #include #include #include #include #endif /* _KERNEL */ #include "zfs_prop.h" #include "zfs_comutil.h" typedef enum zti_modes { ZTI_MODE_FIXED, /* value is # of threads (min 1) */ ZTI_MODE_BATCH, /* cpu-intensive; value is ignored */ ZTI_MODE_NULL, /* don't create a taskq */ ZTI_NMODES } zti_modes_t; #define ZTI_P(n, q) { ZTI_MODE_FIXED, (n), (q) } #define ZTI_PCT(n) { ZTI_MODE_ONLINE_PERCENT, (n), 1 } #define ZTI_BATCH { ZTI_MODE_BATCH, 0, 1 } #define ZTI_NULL { ZTI_MODE_NULL, 0, 0 } #define ZTI_N(n) ZTI_P(n, 1) #define ZTI_ONE ZTI_N(1) typedef struct zio_taskq_info { zti_modes_t zti_mode; uint_t zti_value; uint_t zti_count; } zio_taskq_info_t; static const char *const zio_taskq_types[ZIO_TASKQ_TYPES] = { "iss", "iss_h", "int", "int_h" }; /* * This table defines the taskq settings for each ZFS I/O type. When * initializing a pool, we use this table to create an appropriately sized * taskq. Some operations are low volume and therefore have a small, static * number of threads assigned to their taskqs using the ZTI_N(#) or ZTI_ONE * macros. Other operations process a large amount of data; the ZTI_BATCH * macro causes us to create a taskq oriented for throughput. Some operations * are so high frequency and short-lived that the taskq itself can become a a * point of lock contention. The ZTI_P(#, #) macro indicates that we need an * additional degree of parallelism specified by the number of threads per- * taskq and the number of taskqs; when dispatching an event in this case, the * particular taskq is chosen at random. * * The different taskq priorities are to handle the different contexts (issue * and interrupt) and then to reserve threads for ZIO_PRIORITY_NOW I/Os that * need to be handled with minimum delay. */ const zio_taskq_info_t zio_taskqs[ZIO_TYPES][ZIO_TASKQ_TYPES] = { /* ISSUE ISSUE_HIGH INTR INTR_HIGH */ { ZTI_ONE, ZTI_NULL, ZTI_ONE, ZTI_NULL }, /* NULL */ { ZTI_N(8), ZTI_NULL, ZTI_BATCH, ZTI_NULL }, /* READ */ { ZTI_BATCH, ZTI_N(5), ZTI_N(16), ZTI_N(5) }, /* WRITE */ { ZTI_P(4, 8), ZTI_NULL, ZTI_ONE, ZTI_NULL }, /* FREE */ { ZTI_ONE, ZTI_NULL, ZTI_ONE, ZTI_NULL }, /* CLAIM */ { ZTI_ONE, ZTI_NULL, ZTI_ONE, ZTI_NULL }, /* IOCTL */ }; static void spa_sync_version(void *arg, dmu_tx_t *tx); static void spa_sync_props(void *arg, dmu_tx_t *tx); static boolean_t spa_has_active_shared_spare(spa_t *spa); static inline int spa_load_impl(spa_t *spa, uint64_t, nvlist_t *config, spa_load_state_t state, spa_import_type_t type, boolean_t mosconfig, char **ereport); static void spa_vdev_resilver_done(spa_t *spa); uint_t zio_taskq_batch_pct = 75; /* 1 thread per cpu in pset */ id_t zio_taskq_psrset_bind = PS_NONE; boolean_t zio_taskq_sysdc = B_TRUE; /* use SDC scheduling class */ uint_t zio_taskq_basedc = 80; /* base duty cycle */ boolean_t spa_create_process = B_TRUE; /* no process ==> no sysdc */ /* * This (illegal) pool name is used when temporarily importing a spa_t in order * to get the vdev stats associated with the imported devices. */ #define TRYIMPORT_NAME "$import" /* * ========================================================================== * SPA properties routines * ========================================================================== */ /* * Add a (source=src, propname=propval) list to an nvlist. */ static void spa_prop_add_list(nvlist_t *nvl, zpool_prop_t prop, char *strval, uint64_t intval, zprop_source_t src) { const char *propname = zpool_prop_to_name(prop); nvlist_t *propval; VERIFY(nvlist_alloc(&propval, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_uint64(propval, ZPROP_SOURCE, src) == 0); if (strval != NULL) VERIFY(nvlist_add_string(propval, ZPROP_VALUE, strval) == 0); else VERIFY(nvlist_add_uint64(propval, ZPROP_VALUE, intval) == 0); VERIFY(nvlist_add_nvlist(nvl, propname, propval) == 0); nvlist_free(propval); } /* * Get property values from the spa configuration. */ static void spa_prop_get_config(spa_t *spa, nvlist_t **nvp) { vdev_t *rvd = spa->spa_root_vdev; dsl_pool_t *pool = spa->spa_dsl_pool; uint64_t size, alloc, cap, version; zprop_source_t src = ZPROP_SRC_NONE; spa_config_dirent_t *dp; metaslab_class_t *mc = spa_normal_class(spa); ASSERT(MUTEX_HELD(&spa->spa_props_lock)); if (rvd != NULL) { alloc = metaslab_class_get_alloc(spa_normal_class(spa)); size = metaslab_class_get_space(spa_normal_class(spa)); spa_prop_add_list(*nvp, ZPOOL_PROP_NAME, spa_name(spa), 0, src); spa_prop_add_list(*nvp, ZPOOL_PROP_SIZE, NULL, size, src); spa_prop_add_list(*nvp, ZPOOL_PROP_ALLOCATED, NULL, alloc, src); spa_prop_add_list(*nvp, ZPOOL_PROP_FREE, NULL, size - alloc, src); spa_prop_add_list(*nvp, ZPOOL_PROP_FRAGMENTATION, NULL, metaslab_class_fragmentation(mc), src); spa_prop_add_list(*nvp, ZPOOL_PROP_EXPANDSZ, NULL, metaslab_class_expandable_space(mc), src); spa_prop_add_list(*nvp, ZPOOL_PROP_READONLY, NULL, (spa_mode(spa) == FREAD), src); cap = (size == 0) ? 0 : (alloc * 100 / size); spa_prop_add_list(*nvp, ZPOOL_PROP_CAPACITY, NULL, cap, src); spa_prop_add_list(*nvp, ZPOOL_PROP_DEDUPRATIO, NULL, ddt_get_pool_dedup_ratio(spa), src); spa_prop_add_list(*nvp, ZPOOL_PROP_HEALTH, NULL, rvd->vdev_state, src); version = spa_version(spa); if (version == zpool_prop_default_numeric(ZPOOL_PROP_VERSION)) src = ZPROP_SRC_DEFAULT; else src = ZPROP_SRC_LOCAL; spa_prop_add_list(*nvp, ZPOOL_PROP_VERSION, NULL, version, src); } if (pool != NULL) { /* * The $FREE directory was introduced in SPA_VERSION_DEADLISTS, * when opening pools before this version freedir will be NULL. */ if (pool->dp_free_dir != NULL) { spa_prop_add_list(*nvp, ZPOOL_PROP_FREEING, NULL, pool->dp_free_dir->dd_phys->dd_used_bytes, src); } else { spa_prop_add_list(*nvp, ZPOOL_PROP_FREEING, NULL, 0, src); } if (pool->dp_leak_dir != NULL) { spa_prop_add_list(*nvp, ZPOOL_PROP_LEAKED, NULL, pool->dp_leak_dir->dd_phys->dd_used_bytes, src); } else { spa_prop_add_list(*nvp, ZPOOL_PROP_LEAKED, NULL, 0, src); } } spa_prop_add_list(*nvp, ZPOOL_PROP_GUID, NULL, spa_guid(spa), src); if (spa->spa_comment != NULL) { spa_prop_add_list(*nvp, ZPOOL_PROP_COMMENT, spa->spa_comment, 0, ZPROP_SRC_LOCAL); } if (spa->spa_root != NULL) spa_prop_add_list(*nvp, ZPOOL_PROP_ALTROOT, spa->spa_root, 0, ZPROP_SRC_LOCAL); if ((dp = list_head(&spa->spa_config_list)) != NULL) { if (dp->scd_path == NULL) { spa_prop_add_list(*nvp, ZPOOL_PROP_CACHEFILE, "none", 0, ZPROP_SRC_LOCAL); } else if (strcmp(dp->scd_path, spa_config_path) != 0) { spa_prop_add_list(*nvp, ZPOOL_PROP_CACHEFILE, dp->scd_path, 0, ZPROP_SRC_LOCAL); } } } /* * Get zpool property values. */ int spa_prop_get(spa_t *spa, nvlist_t **nvp) { objset_t *mos = spa->spa_meta_objset; zap_cursor_t zc; zap_attribute_t za; int err; err = nvlist_alloc(nvp, NV_UNIQUE_NAME, KM_PUSHPAGE); if (err) return (err); mutex_enter(&spa->spa_props_lock); /* * Get properties from the spa config. */ spa_prop_get_config(spa, nvp); /* If no pool property object, no more prop to get. */ if (mos == NULL || spa->spa_pool_props_object == 0) { mutex_exit(&spa->spa_props_lock); goto out; } /* * Get properties from the MOS pool property object. */ for (zap_cursor_init(&zc, mos, spa->spa_pool_props_object); (err = zap_cursor_retrieve(&zc, &za)) == 0; zap_cursor_advance(&zc)) { uint64_t intval = 0; char *strval = NULL; zprop_source_t src = ZPROP_SRC_DEFAULT; zpool_prop_t prop; if ((prop = zpool_name_to_prop(za.za_name)) == ZPROP_INVAL) continue; switch (za.za_integer_length) { case 8: /* integer property */ if (za.za_first_integer != zpool_prop_default_numeric(prop)) src = ZPROP_SRC_LOCAL; if (prop == ZPOOL_PROP_BOOTFS) { dsl_pool_t *dp; dsl_dataset_t *ds = NULL; dp = spa_get_dsl(spa); dsl_pool_config_enter(dp, FTAG); if ((err = dsl_dataset_hold_obj(dp, za.za_first_integer, FTAG, &ds))) { dsl_pool_config_exit(dp, FTAG); break; } strval = kmem_alloc( MAXNAMELEN + strlen(MOS_DIR_NAME) + 1, KM_PUSHPAGE); dsl_dataset_name(ds, strval); dsl_dataset_rele(ds, FTAG); dsl_pool_config_exit(dp, FTAG); } else { strval = NULL; intval = za.za_first_integer; } spa_prop_add_list(*nvp, prop, strval, intval, src); if (strval != NULL) kmem_free(strval, MAXNAMELEN + strlen(MOS_DIR_NAME) + 1); break; case 1: /* string property */ strval = kmem_alloc(za.za_num_integers, KM_PUSHPAGE); err = zap_lookup(mos, spa->spa_pool_props_object, za.za_name, 1, za.za_num_integers, strval); if (err) { kmem_free(strval, za.za_num_integers); break; } spa_prop_add_list(*nvp, prop, strval, 0, src); kmem_free(strval, za.za_num_integers); break; default: break; } } zap_cursor_fini(&zc); mutex_exit(&spa->spa_props_lock); out: if (err && err != ENOENT) { nvlist_free(*nvp); *nvp = NULL; return (err); } return (0); } /* * Validate the given pool properties nvlist and modify the list * for the property values to be set. */ static int spa_prop_validate(spa_t *spa, nvlist_t *props) { nvpair_t *elem; int error = 0, reset_bootfs = 0; uint64_t objnum = 0; boolean_t has_feature = B_FALSE; elem = NULL; while ((elem = nvlist_next_nvpair(props, elem)) != NULL) { uint64_t intval; char *strval, *slash, *check, *fname; const char *propname = nvpair_name(elem); zpool_prop_t prop = zpool_name_to_prop(propname); switch ((int)prop) { case ZPROP_INVAL: if (!zpool_prop_feature(propname)) { error = SET_ERROR(EINVAL); break; } /* * Sanitize the input. */ if (nvpair_type(elem) != DATA_TYPE_UINT64) { error = SET_ERROR(EINVAL); break; } if (nvpair_value_uint64(elem, &intval) != 0) { error = SET_ERROR(EINVAL); break; } if (intval != 0) { error = SET_ERROR(EINVAL); break; } fname = strchr(propname, '@') + 1; if (zfeature_lookup_name(fname, NULL) != 0) { error = SET_ERROR(EINVAL); break; } has_feature = B_TRUE; break; case ZPOOL_PROP_VERSION: error = nvpair_value_uint64(elem, &intval); if (!error && (intval < spa_version(spa) || intval > SPA_VERSION_BEFORE_FEATURES || has_feature)) error = SET_ERROR(EINVAL); break; case ZPOOL_PROP_DELEGATION: case ZPOOL_PROP_AUTOREPLACE: case ZPOOL_PROP_LISTSNAPS: case ZPOOL_PROP_AUTOEXPAND: error = nvpair_value_uint64(elem, &intval); if (!error && intval > 1) error = SET_ERROR(EINVAL); break; case ZPOOL_PROP_BOOTFS: /* * If the pool version is less than SPA_VERSION_BOOTFS, * or the pool is still being created (version == 0), * the bootfs property cannot be set. */ if (spa_version(spa) < SPA_VERSION_BOOTFS) { error = SET_ERROR(ENOTSUP); break; } /* * Make sure the vdev config is bootable */ if (!vdev_is_bootable(spa->spa_root_vdev)) { error = SET_ERROR(ENOTSUP); break; } reset_bootfs = 1; error = nvpair_value_string(elem, &strval); if (!error) { objset_t *os; uint64_t compress; if (strval == NULL || strval[0] == '\0') { objnum = zpool_prop_default_numeric( ZPOOL_PROP_BOOTFS); break; } error = dmu_objset_hold(strval, FTAG, &os); if (error) break; /* Must be ZPL and not gzip compressed. */ if (dmu_objset_type(os) != DMU_OST_ZFS) { error = SET_ERROR(ENOTSUP); } else if ((error = dsl_prop_get_int_ds(dmu_objset_ds(os), zfs_prop_to_name(ZFS_PROP_COMPRESSION), &compress)) == 0 && !BOOTFS_COMPRESS_VALID(compress)) { error = SET_ERROR(ENOTSUP); } else { objnum = dmu_objset_id(os); } dmu_objset_rele(os, FTAG); } break; case ZPOOL_PROP_FAILUREMODE: error = nvpair_value_uint64(elem, &intval); if (!error && (intval < ZIO_FAILURE_MODE_WAIT || intval > ZIO_FAILURE_MODE_PANIC)) error = SET_ERROR(EINVAL); /* * This is a special case which only occurs when * the pool has completely failed. This allows * the user to change the in-core failmode property * without syncing it out to disk (I/Os might * currently be blocked). We do this by returning * EIO to the caller (spa_prop_set) to trick it * into thinking we encountered a property validation * error. */ if (!error && spa_suspended(spa)) { spa->spa_failmode = intval; error = SET_ERROR(EIO); } break; case ZPOOL_PROP_CACHEFILE: if ((error = nvpair_value_string(elem, &strval)) != 0) break; if (strval[0] == '\0') break; if (strcmp(strval, "none") == 0) break; if (strval[0] != '/') { error = SET_ERROR(EINVAL); break; } slash = strrchr(strval, '/'); ASSERT(slash != NULL); if (slash[1] == '\0' || strcmp(slash, "/.") == 0 || strcmp(slash, "/..") == 0) error = SET_ERROR(EINVAL); break; case ZPOOL_PROP_COMMENT: if ((error = nvpair_value_string(elem, &strval)) != 0) break; for (check = strval; *check != '\0'; check++) { if (!isprint(*check)) { error = SET_ERROR(EINVAL); break; } check++; } if (strlen(strval) > ZPROP_MAX_COMMENT) error = SET_ERROR(E2BIG); break; case ZPOOL_PROP_DEDUPDITTO: if (spa_version(spa) < SPA_VERSION_DEDUP) error = SET_ERROR(ENOTSUP); else error = nvpair_value_uint64(elem, &intval); if (error == 0 && intval != 0 && intval < ZIO_DEDUPDITTO_MIN) error = SET_ERROR(EINVAL); break; default: break; } if (error) break; } if (!error && reset_bootfs) { error = nvlist_remove(props, zpool_prop_to_name(ZPOOL_PROP_BOOTFS), DATA_TYPE_STRING); if (!error) { error = nvlist_add_uint64(props, zpool_prop_to_name(ZPOOL_PROP_BOOTFS), objnum); } } return (error); } void spa_configfile_set(spa_t *spa, nvlist_t *nvp, boolean_t need_sync) { char *cachefile; spa_config_dirent_t *dp; if (nvlist_lookup_string(nvp, zpool_prop_to_name(ZPOOL_PROP_CACHEFILE), &cachefile) != 0) return; dp = kmem_alloc(sizeof (spa_config_dirent_t), KM_PUSHPAGE); if (cachefile[0] == '\0') dp->scd_path = spa_strdup(spa_config_path); else if (strcmp(cachefile, "none") == 0) dp->scd_path = NULL; else dp->scd_path = spa_strdup(cachefile); list_insert_head(&spa->spa_config_list, dp); if (need_sync) spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE); } int spa_prop_set(spa_t *spa, nvlist_t *nvp) { int error; nvpair_t *elem = NULL; boolean_t need_sync = B_FALSE; if ((error = spa_prop_validate(spa, nvp)) != 0) return (error); while ((elem = nvlist_next_nvpair(nvp, elem)) != NULL) { zpool_prop_t prop = zpool_name_to_prop(nvpair_name(elem)); if (prop == ZPOOL_PROP_CACHEFILE || prop == ZPOOL_PROP_ALTROOT || prop == ZPOOL_PROP_READONLY) continue; if (prop == ZPOOL_PROP_VERSION || prop == ZPROP_INVAL) { uint64_t ver; if (prop == ZPOOL_PROP_VERSION) { VERIFY(nvpair_value_uint64(elem, &ver) == 0); } else { ASSERT(zpool_prop_feature(nvpair_name(elem))); ver = SPA_VERSION_FEATURES; need_sync = B_TRUE; } /* Save time if the version is already set. */ if (ver == spa_version(spa)) continue; /* * In addition to the pool directory object, we might * create the pool properties object, the features for * read object, the features for write object, or the * feature descriptions object. */ error = dsl_sync_task(spa->spa_name, NULL, spa_sync_version, &ver, 6); if (error) return (error); continue; } need_sync = B_TRUE; break; } if (need_sync) { return (dsl_sync_task(spa->spa_name, NULL, spa_sync_props, nvp, 6)); } return (0); } /* * If the bootfs property value is dsobj, clear it. */ void spa_prop_clear_bootfs(spa_t *spa, uint64_t dsobj, dmu_tx_t *tx) { if (spa->spa_bootfs == dsobj && spa->spa_pool_props_object != 0) { VERIFY(zap_remove(spa->spa_meta_objset, spa->spa_pool_props_object, zpool_prop_to_name(ZPOOL_PROP_BOOTFS), tx) == 0); spa->spa_bootfs = 0; } } /*ARGSUSED*/ static int spa_change_guid_check(void *arg, dmu_tx_t *tx) { spa_t *spa = dmu_tx_pool(tx)->dp_spa; vdev_t *rvd = spa->spa_root_vdev; uint64_t vdev_state; ASSERTV(uint64_t *newguid = arg); spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); vdev_state = rvd->vdev_state; spa_config_exit(spa, SCL_STATE, FTAG); if (vdev_state != VDEV_STATE_HEALTHY) return (SET_ERROR(ENXIO)); ASSERT3U(spa_guid(spa), !=, *newguid); return (0); } static void spa_change_guid_sync(void *arg, dmu_tx_t *tx) { uint64_t *newguid = arg; spa_t *spa = dmu_tx_pool(tx)->dp_spa; uint64_t oldguid; vdev_t *rvd = spa->spa_root_vdev; oldguid = spa_guid(spa); spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); rvd->vdev_guid = *newguid; rvd->vdev_guid_sum += (*newguid - oldguid); vdev_config_dirty(rvd); spa_config_exit(spa, SCL_STATE, FTAG); spa_history_log_internal(spa, "guid change", tx, "old=%llu new=%llu", oldguid, *newguid); } /* * Change the GUID for the pool. This is done so that we can later * re-import a pool built from a clone of our own vdevs. We will modify * the root vdev's guid, our own pool guid, and then mark all of our * vdevs dirty. Note that we must make sure that all our vdevs are * online when we do this, or else any vdevs that weren't present * would be orphaned from our pool. We are also going to issue a * sysevent to update any watchers. */ int spa_change_guid(spa_t *spa) { int error; uint64_t guid; mutex_enter(&spa->spa_vdev_top_lock); mutex_enter(&spa_namespace_lock); guid = spa_generate_guid(NULL); error = dsl_sync_task(spa->spa_name, spa_change_guid_check, spa_change_guid_sync, &guid, 5); if (error == 0) { spa_config_sync(spa, B_FALSE, B_TRUE); spa_event_notify(spa, NULL, FM_EREPORT_ZFS_POOL_REGUID); } mutex_exit(&spa_namespace_lock); mutex_exit(&spa->spa_vdev_top_lock); return (error); } /* * ========================================================================== * SPA state manipulation (open/create/destroy/import/export) * ========================================================================== */ static int spa_error_entry_compare(const void *a, const void *b) { spa_error_entry_t *sa = (spa_error_entry_t *)a; spa_error_entry_t *sb = (spa_error_entry_t *)b; int ret; ret = bcmp(&sa->se_bookmark, &sb->se_bookmark, sizeof (zbookmark_phys_t)); if (ret < 0) return (-1); else if (ret > 0) return (1); else return (0); } /* * Utility function which retrieves copies of the current logs and * re-initializes them in the process. */ void spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub) { ASSERT(MUTEX_HELD(&spa->spa_errlist_lock)); bcopy(&spa->spa_errlist_last, last, sizeof (avl_tree_t)); bcopy(&spa->spa_errlist_scrub, scrub, sizeof (avl_tree_t)); avl_create(&spa->spa_errlist_scrub, spa_error_entry_compare, sizeof (spa_error_entry_t), offsetof(spa_error_entry_t, se_avl)); avl_create(&spa->spa_errlist_last, spa_error_entry_compare, sizeof (spa_error_entry_t), offsetof(spa_error_entry_t, se_avl)); } static void spa_taskqs_init(spa_t *spa, zio_type_t t, zio_taskq_type_t q) { const zio_taskq_info_t *ztip = &zio_taskqs[t][q]; enum zti_modes mode = ztip->zti_mode; uint_t value = ztip->zti_value; uint_t count = ztip->zti_count; spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q]; char name[32]; uint_t i, flags = 0; boolean_t batch = B_FALSE; if (mode == ZTI_MODE_NULL) { tqs->stqs_count = 0; tqs->stqs_taskq = NULL; return; } ASSERT3U(count, >, 0); tqs->stqs_count = count; tqs->stqs_taskq = kmem_alloc(count * sizeof (taskq_t *), KM_SLEEP); switch (mode) { case ZTI_MODE_FIXED: ASSERT3U(value, >=, 1); value = MAX(value, 1); break; case ZTI_MODE_BATCH: batch = B_TRUE; flags |= TASKQ_THREADS_CPU_PCT; value = zio_taskq_batch_pct; break; default: panic("unrecognized mode for %s_%s taskq (%u:%u) in " "spa_activate()", zio_type_name[t], zio_taskq_types[q], mode, value); break; } for (i = 0; i < count; i++) { taskq_t *tq; if (count > 1) { (void) snprintf(name, sizeof (name), "%s_%s_%u", zio_type_name[t], zio_taskq_types[q], i); } else { (void) snprintf(name, sizeof (name), "%s_%s", zio_type_name[t], zio_taskq_types[q]); } if (zio_taskq_sysdc && spa->spa_proc != &p0) { if (batch) flags |= TASKQ_DC_BATCH; tq = taskq_create_sysdc(name, value, 50, INT_MAX, spa->spa_proc, zio_taskq_basedc, flags); } else { pri_t pri = maxclsyspri; /* * The write issue taskq can be extremely CPU * intensive. Run it at slightly lower priority * than the other taskqs. */ if (t == ZIO_TYPE_WRITE && q == ZIO_TASKQ_ISSUE) pri--; tq = taskq_create_proc(name, value, pri, 50, INT_MAX, spa->spa_proc, flags); } tqs->stqs_taskq[i] = tq; } } static void spa_taskqs_fini(spa_t *spa, zio_type_t t, zio_taskq_type_t q) { spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q]; uint_t i; if (tqs->stqs_taskq == NULL) { ASSERT3U(tqs->stqs_count, ==, 0); return; } for (i = 0; i < tqs->stqs_count; i++) { ASSERT3P(tqs->stqs_taskq[i], !=, NULL); taskq_destroy(tqs->stqs_taskq[i]); } kmem_free(tqs->stqs_taskq, tqs->stqs_count * sizeof (taskq_t *)); tqs->stqs_taskq = NULL; } /* * Dispatch a task to the appropriate taskq for the ZFS I/O type and priority. * Note that a type may have multiple discrete taskqs to avoid lock contention * on the taskq itself. In that case we choose which taskq at random by using * the low bits of gethrtime(). */ void spa_taskq_dispatch_ent(spa_t *spa, zio_type_t t, zio_taskq_type_t q, task_func_t *func, void *arg, uint_t flags, taskq_ent_t *ent) { spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q]; taskq_t *tq; ASSERT3P(tqs->stqs_taskq, !=, NULL); ASSERT3U(tqs->stqs_count, !=, 0); if (tqs->stqs_count == 1) { tq = tqs->stqs_taskq[0]; } else { tq = tqs->stqs_taskq[((uint64_t)gethrtime()) % tqs->stqs_count]; } taskq_dispatch_ent(tq, func, arg, flags, ent); } /* * Same as spa_taskq_dispatch_ent() but block on the task until completion. */ void spa_taskq_dispatch_sync(spa_t *spa, zio_type_t t, zio_taskq_type_t q, task_func_t *func, void *arg, uint_t flags) { spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q]; taskq_t *tq; taskqid_t id; ASSERT3P(tqs->stqs_taskq, !=, NULL); ASSERT3U(tqs->stqs_count, !=, 0); if (tqs->stqs_count == 1) { tq = tqs->stqs_taskq[0]; } else { tq = tqs->stqs_taskq[((uint64_t)gethrtime()) % tqs->stqs_count]; } id = taskq_dispatch(tq, func, arg, flags); if (id) taskq_wait_id(tq, id); } static void spa_create_zio_taskqs(spa_t *spa) { int t, q; for (t = 0; t < ZIO_TYPES; t++) { for (q = 0; q < ZIO_TASKQ_TYPES; q++) { spa_taskqs_init(spa, t, q); } } } #if defined(_KERNEL) && defined(HAVE_SPA_THREAD) static void spa_thread(void *arg) { callb_cpr_t cprinfo; spa_t *spa = arg; user_t *pu = PTOU(curproc); CALLB_CPR_INIT(&cprinfo, &spa->spa_proc_lock, callb_generic_cpr, spa->spa_name); ASSERT(curproc != &p0); (void) snprintf(pu->u_psargs, sizeof (pu->u_psargs), "zpool-%s", spa->spa_name); (void) strlcpy(pu->u_comm, pu->u_psargs, sizeof (pu->u_comm)); /* bind this thread to the requested psrset */ if (zio_taskq_psrset_bind != PS_NONE) { pool_lock(); mutex_enter(&cpu_lock); mutex_enter(&pidlock); mutex_enter(&curproc->p_lock); if (cpupart_bind_thread(curthread, zio_taskq_psrset_bind, 0, NULL, NULL) == 0) { curthread->t_bind_pset = zio_taskq_psrset_bind; } else { cmn_err(CE_WARN, "Couldn't bind process for zfs pool \"%s\" to " "pset %d\n", spa->spa_name, zio_taskq_psrset_bind); } mutex_exit(&curproc->p_lock); mutex_exit(&pidlock); mutex_exit(&cpu_lock); pool_unlock(); } if (zio_taskq_sysdc) { sysdc_thread_enter(curthread, 100, 0); } spa->spa_proc = curproc; spa->spa_did = curthread->t_did; spa_create_zio_taskqs(spa); mutex_enter(&spa->spa_proc_lock); ASSERT(spa->spa_proc_state == SPA_PROC_CREATED); spa->spa_proc_state = SPA_PROC_ACTIVE; cv_broadcast(&spa->spa_proc_cv); CALLB_CPR_SAFE_BEGIN(&cprinfo); while (spa->spa_proc_state == SPA_PROC_ACTIVE) cv_wait(&spa->spa_proc_cv, &spa->spa_proc_lock); CALLB_CPR_SAFE_END(&cprinfo, &spa->spa_proc_lock); ASSERT(spa->spa_proc_state == SPA_PROC_DEACTIVATE); spa->spa_proc_state = SPA_PROC_GONE; spa->spa_proc = &p0; cv_broadcast(&spa->spa_proc_cv); CALLB_CPR_EXIT(&cprinfo); /* drops spa_proc_lock */ mutex_enter(&curproc->p_lock); lwp_exit(); } #endif /* * Activate an uninitialized pool. */ static void spa_activate(spa_t *spa, int mode) { ASSERT(spa->spa_state == POOL_STATE_UNINITIALIZED); spa->spa_state = POOL_STATE_ACTIVE; spa->spa_mode = mode; spa->spa_normal_class = metaslab_class_create(spa, zfs_metaslab_ops); spa->spa_log_class = metaslab_class_create(spa, zfs_metaslab_ops); /* Try to create a covering process */ mutex_enter(&spa->spa_proc_lock); ASSERT(spa->spa_proc_state == SPA_PROC_NONE); ASSERT(spa->spa_proc == &p0); spa->spa_did = 0; #ifdef HAVE_SPA_THREAD /* Only create a process if we're going to be around a while. */ if (spa_create_process && strcmp(spa->spa_name, TRYIMPORT_NAME) != 0) { if (newproc(spa_thread, (caddr_t)spa, syscid, maxclsyspri, NULL, 0) == 0) { spa->spa_proc_state = SPA_PROC_CREATED; while (spa->spa_proc_state == SPA_PROC_CREATED) { cv_wait(&spa->spa_proc_cv, &spa->spa_proc_lock); } ASSERT(spa->spa_proc_state == SPA_PROC_ACTIVE); ASSERT(spa->spa_proc != &p0); ASSERT(spa->spa_did != 0); } else { #ifdef _KERNEL cmn_err(CE_WARN, "Couldn't create process for zfs pool \"%s\"\n", spa->spa_name); #endif } } #endif /* HAVE_SPA_THREAD */ mutex_exit(&spa->spa_proc_lock); /* If we didn't create a process, we need to create our taskqs. */ if (spa->spa_proc == &p0) { spa_create_zio_taskqs(spa); } list_create(&spa->spa_config_dirty_list, sizeof (vdev_t), offsetof(vdev_t, vdev_config_dirty_node)); list_create(&spa->spa_state_dirty_list, sizeof (vdev_t), offsetof(vdev_t, vdev_state_dirty_node)); txg_list_create(&spa->spa_vdev_txg_list, offsetof(struct vdev, vdev_txg_node)); avl_create(&spa->spa_errlist_scrub, spa_error_entry_compare, sizeof (spa_error_entry_t), offsetof(spa_error_entry_t, se_avl)); avl_create(&spa->spa_errlist_last, spa_error_entry_compare, sizeof (spa_error_entry_t), offsetof(spa_error_entry_t, se_avl)); } /* * Opposite of spa_activate(). */ static void spa_deactivate(spa_t *spa) { int t, q; ASSERT(spa->spa_sync_on == B_FALSE); ASSERT(spa->spa_dsl_pool == NULL); ASSERT(spa->spa_root_vdev == NULL); ASSERT(spa->spa_async_zio_root == NULL); ASSERT(spa->spa_state != POOL_STATE_UNINITIALIZED); txg_list_destroy(&spa->spa_vdev_txg_list); list_destroy(&spa->spa_config_dirty_list); list_destroy(&spa->spa_state_dirty_list); taskq_cancel_id(system_taskq, spa->spa_deadman_tqid); for (t = 0; t < ZIO_TYPES; t++) { for (q = 0; q < ZIO_TASKQ_TYPES; q++) { spa_taskqs_fini(spa, t, q); } } metaslab_class_destroy(spa->spa_normal_class); spa->spa_normal_class = NULL; metaslab_class_destroy(spa->spa_log_class); spa->spa_log_class = NULL; /* * If this was part of an import or the open otherwise failed, we may * still have errors left in the queues. Empty them just in case. */ spa_errlog_drain(spa); avl_destroy(&spa->spa_errlist_scrub); avl_destroy(&spa->spa_errlist_last); spa->spa_state = POOL_STATE_UNINITIALIZED; mutex_enter(&spa->spa_proc_lock); if (spa->spa_proc_state != SPA_PROC_NONE) { ASSERT(spa->spa_proc_state == SPA_PROC_ACTIVE); spa->spa_proc_state = SPA_PROC_DEACTIVATE; cv_broadcast(&spa->spa_proc_cv); while (spa->spa_proc_state == SPA_PROC_DEACTIVATE) { ASSERT(spa->spa_proc != &p0); cv_wait(&spa->spa_proc_cv, &spa->spa_proc_lock); } ASSERT(spa->spa_proc_state == SPA_PROC_GONE); spa->spa_proc_state = SPA_PROC_NONE; } ASSERT(spa->spa_proc == &p0); mutex_exit(&spa->spa_proc_lock); /* * We want to make sure spa_thread() has actually exited the ZFS * module, so that the module can't be unloaded out from underneath * it. */ if (spa->spa_did != 0) { thread_join(spa->spa_did); spa->spa_did = 0; } } /* * Verify a pool configuration, and construct the vdev tree appropriately. This * will create all the necessary vdevs in the appropriate layout, with each vdev * in the CLOSED state. This will prep the pool before open/creation/import. * All vdev validation is done by the vdev_alloc() routine. */ static int spa_config_parse(spa_t *spa, vdev_t **vdp, nvlist_t *nv, vdev_t *parent, uint_t id, int atype) { nvlist_t **child; uint_t children; int error; int c; if ((error = vdev_alloc(spa, vdp, nv, parent, id, atype)) != 0) return (error); if ((*vdp)->vdev_ops->vdev_op_leaf) return (0); error = nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, &child, &children); if (error == ENOENT) return (0); if (error) { vdev_free(*vdp); *vdp = NULL; return (SET_ERROR(EINVAL)); } for (c = 0; c < children; c++) { vdev_t *vd; if ((error = spa_config_parse(spa, &vd, child[c], *vdp, c, atype)) != 0) { vdev_free(*vdp); *vdp = NULL; return (error); } } ASSERT(*vdp != NULL); return (0); } /* * Opposite of spa_load(). */ static void spa_unload(spa_t *spa) { int i; ASSERT(MUTEX_HELD(&spa_namespace_lock)); /* * Stop async tasks. */ spa_async_suspend(spa); /* * Stop syncing. */ if (spa->spa_sync_on) { txg_sync_stop(spa->spa_dsl_pool); spa->spa_sync_on = B_FALSE; } /* * Wait for any outstanding async I/O to complete. */ if (spa->spa_async_zio_root != NULL) { for (i = 0; i < max_ncpus; i++) (void) zio_wait(spa->spa_async_zio_root[i]); kmem_free(spa->spa_async_zio_root, max_ncpus * sizeof (void *)); spa->spa_async_zio_root = NULL; } bpobj_close(&spa->spa_deferred_bpobj); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); /* * Close all vdevs. */ if (spa->spa_root_vdev) vdev_free(spa->spa_root_vdev); ASSERT(spa->spa_root_vdev == NULL); /* * Close the dsl pool. */ if (spa->spa_dsl_pool) { dsl_pool_close(spa->spa_dsl_pool); spa->spa_dsl_pool = NULL; spa->spa_meta_objset = NULL; } ddt_unload(spa); /* * Drop and purge level 2 cache */ spa_l2cache_drop(spa); for (i = 0; i < spa->spa_spares.sav_count; i++) vdev_free(spa->spa_spares.sav_vdevs[i]); if (spa->spa_spares.sav_vdevs) { kmem_free(spa->spa_spares.sav_vdevs, spa->spa_spares.sav_count * sizeof (void *)); spa->spa_spares.sav_vdevs = NULL; } if (spa->spa_spares.sav_config) { nvlist_free(spa->spa_spares.sav_config); spa->spa_spares.sav_config = NULL; } spa->spa_spares.sav_count = 0; for (i = 0; i < spa->spa_l2cache.sav_count; i++) { vdev_clear_stats(spa->spa_l2cache.sav_vdevs[i]); vdev_free(spa->spa_l2cache.sav_vdevs[i]); } if (spa->spa_l2cache.sav_vdevs) { kmem_free(spa->spa_l2cache.sav_vdevs, spa->spa_l2cache.sav_count * sizeof (void *)); spa->spa_l2cache.sav_vdevs = NULL; } if (spa->spa_l2cache.sav_config) { nvlist_free(spa->spa_l2cache.sav_config); spa->spa_l2cache.sav_config = NULL; } spa->spa_l2cache.sav_count = 0; spa->spa_async_suspended = 0; if (spa->spa_comment != NULL) { spa_strfree(spa->spa_comment); spa->spa_comment = NULL; } spa_config_exit(spa, SCL_ALL, FTAG); } /* * Load (or re-load) the current list of vdevs describing the active spares for * this pool. When this is called, we have some form of basic information in * 'spa_spares.sav_config'. We parse this into vdevs, try to open them, and * then re-generate a more complete list including status information. */ static void spa_load_spares(spa_t *spa) { nvlist_t **spares; uint_t nspares; int i; vdev_t *vd, *tvd; ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); /* * First, close and free any existing spare vdevs. */ for (i = 0; i < spa->spa_spares.sav_count; i++) { vd = spa->spa_spares.sav_vdevs[i]; /* Undo the call to spa_activate() below */ if ((tvd = spa_lookup_by_guid(spa, vd->vdev_guid, B_FALSE)) != NULL && tvd->vdev_isspare) spa_spare_remove(tvd); vdev_close(vd); vdev_free(vd); } if (spa->spa_spares.sav_vdevs) kmem_free(spa->spa_spares.sav_vdevs, spa->spa_spares.sav_count * sizeof (void *)); if (spa->spa_spares.sav_config == NULL) nspares = 0; else VERIFY(nvlist_lookup_nvlist_array(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0); spa->spa_spares.sav_count = (int)nspares; spa->spa_spares.sav_vdevs = NULL; if (nspares == 0) return; /* * Construct the array of vdevs, opening them to get status in the * process. For each spare, there is potentially two different vdev_t * structures associated with it: one in the list of spares (used only * for basic validation purposes) and one in the active vdev * configuration (if it's spared in). During this phase we open and * validate each vdev on the spare list. If the vdev also exists in the * active configuration, then we also mark this vdev as an active spare. */ spa->spa_spares.sav_vdevs = kmem_zalloc(nspares * sizeof (void *), KM_PUSHPAGE); for (i = 0; i < spa->spa_spares.sav_count; i++) { VERIFY(spa_config_parse(spa, &vd, spares[i], NULL, 0, VDEV_ALLOC_SPARE) == 0); ASSERT(vd != NULL); spa->spa_spares.sav_vdevs[i] = vd; if ((tvd = spa_lookup_by_guid(spa, vd->vdev_guid, B_FALSE)) != NULL) { if (!tvd->vdev_isspare) spa_spare_add(tvd); /* * We only mark the spare active if we were successfully * able to load the vdev. Otherwise, importing a pool * with a bad active spare would result in strange * behavior, because multiple pool would think the spare * is actively in use. * * There is a vulnerability here to an equally bizarre * circumstance, where a dead active spare is later * brought back to life (onlined or otherwise). Given * the rarity of this scenario, and the extra complexity * it adds, we ignore the possibility. */ if (!vdev_is_dead(tvd)) spa_spare_activate(tvd); } vd->vdev_top = vd; vd->vdev_aux = &spa->spa_spares; if (vdev_open(vd) != 0) continue; if (vdev_validate_aux(vd) == 0) spa_spare_add(vd); } /* * Recompute the stashed list of spares, with status information * this time. */ VERIFY(nvlist_remove(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, DATA_TYPE_NVLIST_ARRAY) == 0); spares = kmem_alloc(spa->spa_spares.sav_count * sizeof (void *), KM_PUSHPAGE); for (i = 0; i < spa->spa_spares.sav_count; i++) spares[i] = vdev_config_generate(spa, spa->spa_spares.sav_vdevs[i], B_TRUE, VDEV_CONFIG_SPARE); VERIFY(nvlist_add_nvlist_array(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, spares, spa->spa_spares.sav_count) == 0); for (i = 0; i < spa->spa_spares.sav_count; i++) nvlist_free(spares[i]); kmem_free(spares, spa->spa_spares.sav_count * sizeof (void *)); } /* * Load (or re-load) the current list of vdevs describing the active l2cache for * this pool. When this is called, we have some form of basic information in * 'spa_l2cache.sav_config'. We parse this into vdevs, try to open them, and * then re-generate a more complete list including status information. * Devices which are already active have their details maintained, and are * not re-opened. */ static void spa_load_l2cache(spa_t *spa) { nvlist_t **l2cache; uint_t nl2cache; int i, j, oldnvdevs; uint64_t guid; vdev_t *vd, **oldvdevs, **newvdevs; spa_aux_vdev_t *sav = &spa->spa_l2cache; ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); if (sav->sav_config != NULL) { VERIFY(nvlist_lookup_nvlist_array(sav->sav_config, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0); newvdevs = kmem_alloc(nl2cache * sizeof (void *), KM_PUSHPAGE); } else { nl2cache = 0; newvdevs = NULL; } oldvdevs = sav->sav_vdevs; oldnvdevs = sav->sav_count; sav->sav_vdevs = NULL; sav->sav_count = 0; /* * Process new nvlist of vdevs. */ for (i = 0; i < nl2cache; i++) { VERIFY(nvlist_lookup_uint64(l2cache[i], ZPOOL_CONFIG_GUID, &guid) == 0); newvdevs[i] = NULL; for (j = 0; j < oldnvdevs; j++) { vd = oldvdevs[j]; if (vd != NULL && guid == vd->vdev_guid) { /* * Retain previous vdev for add/remove ops. */ newvdevs[i] = vd; oldvdevs[j] = NULL; break; } } if (newvdevs[i] == NULL) { /* * Create new vdev */ VERIFY(spa_config_parse(spa, &vd, l2cache[i], NULL, 0, VDEV_ALLOC_L2CACHE) == 0); ASSERT(vd != NULL); newvdevs[i] = vd; /* * Commit this vdev as an l2cache device, * even if it fails to open. */ spa_l2cache_add(vd); vd->vdev_top = vd; vd->vdev_aux = sav; spa_l2cache_activate(vd); if (vdev_open(vd) != 0) continue; (void) vdev_validate_aux(vd); if (!vdev_is_dead(vd)) l2arc_add_vdev(spa, vd); } } /* * Purge vdevs that were dropped */ for (i = 0; i < oldnvdevs; i++) { uint64_t pool; vd = oldvdevs[i]; if (vd != NULL) { ASSERT(vd->vdev_isl2cache); if (spa_l2cache_exists(vd->vdev_guid, &pool) && pool != 0ULL && l2arc_vdev_present(vd)) l2arc_remove_vdev(vd); vdev_clear_stats(vd); vdev_free(vd); } } if (oldvdevs) kmem_free(oldvdevs, oldnvdevs * sizeof (void *)); if (sav->sav_config == NULL) goto out; sav->sav_vdevs = newvdevs; sav->sav_count = (int)nl2cache; /* * Recompute the stashed list of l2cache devices, with status * information this time. */ VERIFY(nvlist_remove(sav->sav_config, ZPOOL_CONFIG_L2CACHE, DATA_TYPE_NVLIST_ARRAY) == 0); l2cache = kmem_alloc(sav->sav_count * sizeof (void *), KM_PUSHPAGE); for (i = 0; i < sav->sav_count; i++) l2cache[i] = vdev_config_generate(spa, sav->sav_vdevs[i], B_TRUE, VDEV_CONFIG_L2CACHE); VERIFY(nvlist_add_nvlist_array(sav->sav_config, ZPOOL_CONFIG_L2CACHE, l2cache, sav->sav_count) == 0); out: for (i = 0; i < sav->sav_count; i++) nvlist_free(l2cache[i]); if (sav->sav_count) kmem_free(l2cache, sav->sav_count * sizeof (void *)); } static int load_nvlist(spa_t *spa, uint64_t obj, nvlist_t **value) { dmu_buf_t *db; char *packed = NULL; size_t nvsize = 0; int error; *value = NULL; error = dmu_bonus_hold(spa->spa_meta_objset, obj, FTAG, &db); if (error) return (error); nvsize = *(uint64_t *)db->db_data; dmu_buf_rele(db, FTAG); - packed = kmem_alloc(nvsize, KM_PUSHPAGE | KM_NODEBUG); + packed = kmem_alloc(nvsize, KM_PUSHPAGE); error = dmu_read(spa->spa_meta_objset, obj, 0, nvsize, packed, DMU_READ_PREFETCH); if (error == 0) error = nvlist_unpack(packed, nvsize, value, 0); kmem_free(packed, nvsize); return (error); } /* * Checks to see if the given vdev could not be opened, in which case we post a * sysevent to notify the autoreplace code that the device has been removed. */ static void spa_check_removed(vdev_t *vd) { int c; for (c = 0; c < vd->vdev_children; c++) spa_check_removed(vd->vdev_child[c]); if (vd->vdev_ops->vdev_op_leaf && vdev_is_dead(vd) && !vd->vdev_ishole) { zfs_ereport_post(FM_EREPORT_RESOURCE_AUTOREPLACE, vd->vdev_spa, vd, NULL, 0, 0); spa_event_notify(vd->vdev_spa, vd, FM_EREPORT_ZFS_DEVICE_CHECK); } } /* * Validate the current config against the MOS config */ static boolean_t spa_config_valid(spa_t *spa, nvlist_t *config) { vdev_t *mrvd, *rvd = spa->spa_root_vdev; nvlist_t *nv; int c, i; VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nv) == 0); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); VERIFY(spa_config_parse(spa, &mrvd, nv, NULL, 0, VDEV_ALLOC_LOAD) == 0); ASSERT3U(rvd->vdev_children, ==, mrvd->vdev_children); /* * If we're doing a normal import, then build up any additional * diagnostic information about missing devices in this config. * We'll pass this up to the user for further processing. */ if (!(spa->spa_import_flags & ZFS_IMPORT_MISSING_LOG)) { nvlist_t **child, *nv; uint64_t idx = 0; child = kmem_alloc(rvd->vdev_children * sizeof (nvlist_t **), KM_PUSHPAGE); VERIFY(nvlist_alloc(&nv, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); for (c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; vdev_t *mtvd = mrvd->vdev_child[c]; if (tvd->vdev_ops == &vdev_missing_ops && mtvd->vdev_ops != &vdev_missing_ops && mtvd->vdev_islog) child[idx++] = vdev_config_generate(spa, mtvd, B_FALSE, 0); } if (idx) { VERIFY(nvlist_add_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, child, idx) == 0); VERIFY(nvlist_add_nvlist(spa->spa_load_info, ZPOOL_CONFIG_MISSING_DEVICES, nv) == 0); for (i = 0; i < idx; i++) nvlist_free(child[i]); } nvlist_free(nv); kmem_free(child, rvd->vdev_children * sizeof (char **)); } /* * Compare the root vdev tree with the information we have * from the MOS config (mrvd). Check each top-level vdev * with the corresponding MOS config top-level (mtvd). */ for (c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; vdev_t *mtvd = mrvd->vdev_child[c]; /* * Resolve any "missing" vdevs in the current configuration. * If we find that the MOS config has more accurate information * about the top-level vdev then use that vdev instead. */ if (tvd->vdev_ops == &vdev_missing_ops && mtvd->vdev_ops != &vdev_missing_ops) { if (!(spa->spa_import_flags & ZFS_IMPORT_MISSING_LOG)) continue; /* * Device specific actions. */ if (mtvd->vdev_islog) { spa_set_log_state(spa, SPA_LOG_CLEAR); } else { /* * XXX - once we have 'readonly' pool * support we should be able to handle * missing data devices by transitioning * the pool to readonly. */ continue; } /* * Swap the missing vdev with the data we were * able to obtain from the MOS config. */ vdev_remove_child(rvd, tvd); vdev_remove_child(mrvd, mtvd); vdev_add_child(rvd, mtvd); vdev_add_child(mrvd, tvd); spa_config_exit(spa, SCL_ALL, FTAG); vdev_load(mtvd); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); vdev_reopen(rvd); } else if (mtvd->vdev_islog) { /* * Load the slog device's state from the MOS config * since it's possible that the label does not * contain the most up-to-date information. */ vdev_load_log_state(tvd, mtvd); vdev_reopen(tvd); } } vdev_free(mrvd); spa_config_exit(spa, SCL_ALL, FTAG); /* * Ensure we were able to validate the config. */ return (rvd->vdev_guid_sum == spa->spa_uberblock.ub_guid_sum); } /* * Check for missing log devices */ static boolean_t spa_check_logs(spa_t *spa) { boolean_t rv = B_FALSE; switch (spa->spa_log_state) { default: break; case SPA_LOG_MISSING: /* need to recheck in case slog has been restored */ case SPA_LOG_UNKNOWN: rv = (dmu_objset_find(spa->spa_name, zil_check_log_chain, NULL, DS_FIND_CHILDREN) != 0); if (rv) spa_set_log_state(spa, SPA_LOG_MISSING); break; } return (rv); } static boolean_t spa_passivate_log(spa_t *spa) { vdev_t *rvd = spa->spa_root_vdev; boolean_t slog_found = B_FALSE; int c; ASSERT(spa_config_held(spa, SCL_ALLOC, RW_WRITER)); if (!spa_has_slogs(spa)) return (B_FALSE); for (c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; metaslab_group_t *mg = tvd->vdev_mg; if (tvd->vdev_islog) { metaslab_group_passivate(mg); slog_found = B_TRUE; } } return (slog_found); } static void spa_activate_log(spa_t *spa) { vdev_t *rvd = spa->spa_root_vdev; int c; ASSERT(spa_config_held(spa, SCL_ALLOC, RW_WRITER)); for (c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; metaslab_group_t *mg = tvd->vdev_mg; if (tvd->vdev_islog) metaslab_group_activate(mg); } } int spa_offline_log(spa_t *spa) { int error; error = dmu_objset_find(spa_name(spa), zil_vdev_offline, NULL, DS_FIND_CHILDREN); if (error == 0) { /* * We successfully offlined the log device, sync out the * current txg so that the "stubby" block can be removed * by zil_sync(). */ txg_wait_synced(spa->spa_dsl_pool, 0); } return (error); } static void spa_aux_check_removed(spa_aux_vdev_t *sav) { int i; for (i = 0; i < sav->sav_count; i++) spa_check_removed(sav->sav_vdevs[i]); } void spa_claim_notify(zio_t *zio) { spa_t *spa = zio->io_spa; if (zio->io_error) return; mutex_enter(&spa->spa_props_lock); /* any mutex will do */ if (spa->spa_claim_max_txg < zio->io_bp->blk_birth) spa->spa_claim_max_txg = zio->io_bp->blk_birth; mutex_exit(&spa->spa_props_lock); } typedef struct spa_load_error { uint64_t sle_meta_count; uint64_t sle_data_count; } spa_load_error_t; static void spa_load_verify_done(zio_t *zio) { blkptr_t *bp = zio->io_bp; spa_load_error_t *sle = zio->io_private; dmu_object_type_t type = BP_GET_TYPE(bp); int error = zio->io_error; spa_t *spa = zio->io_spa; if (error) { if ((BP_GET_LEVEL(bp) != 0 || DMU_OT_IS_METADATA(type)) && type != DMU_OT_INTENT_LOG) atomic_add_64(&sle->sle_meta_count, 1); else atomic_add_64(&sle->sle_data_count, 1); } zio_data_buf_free(zio->io_data, zio->io_size); mutex_enter(&spa->spa_scrub_lock); spa->spa_scrub_inflight--; cv_broadcast(&spa->spa_scrub_io_cv); mutex_exit(&spa->spa_scrub_lock); } /* * Maximum number of concurrent scrub i/os to create while verifying * a pool while importing it. */ int spa_load_verify_maxinflight = 10000; int spa_load_verify_metadata = B_TRUE; int spa_load_verify_data = B_TRUE; /*ARGSUSED*/ static int spa_load_verify_cb(spa_t *spa, zilog_t *zilog, const blkptr_t *bp, const zbookmark_phys_t *zb, const dnode_phys_t *dnp, void *arg) { zio_t *rio; size_t size; void *data; if (BP_IS_HOLE(bp) || BP_IS_EMBEDDED(bp)) return (0); /* * Note: normally this routine will not be called if * spa_load_verify_metadata is not set. However, it may be useful * to manually set the flag after the traversal has begun. */ if (!spa_load_verify_metadata) return (0); if (BP_GET_BUFC_TYPE(bp) == ARC_BUFC_DATA && !spa_load_verify_data) return (0); rio = arg; size = BP_GET_PSIZE(bp); data = zio_data_buf_alloc(size); mutex_enter(&spa->spa_scrub_lock); while (spa->spa_scrub_inflight >= spa_load_verify_maxinflight) cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock); spa->spa_scrub_inflight++; mutex_exit(&spa->spa_scrub_lock); zio_nowait(zio_read(rio, spa, bp, data, size, spa_load_verify_done, rio->io_private, ZIO_PRIORITY_SCRUB, ZIO_FLAG_SPECULATIVE | ZIO_FLAG_CANFAIL | ZIO_FLAG_SCRUB | ZIO_FLAG_RAW, zb)); return (0); } static int spa_load_verify(spa_t *spa) { zio_t *rio; spa_load_error_t sle = { 0 }; zpool_rewind_policy_t policy; boolean_t verify_ok = B_FALSE; int error = 0; zpool_get_rewind_policy(spa->spa_config, &policy); if (policy.zrp_request & ZPOOL_NEVER_REWIND) return (0); rio = zio_root(spa, NULL, &sle, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE); if (spa_load_verify_metadata) { error = traverse_pool(spa, spa->spa_verify_min_txg, TRAVERSE_PRE | TRAVERSE_PREFETCH_METADATA, spa_load_verify_cb, rio); } (void) zio_wait(rio); spa->spa_load_meta_errors = sle.sle_meta_count; spa->spa_load_data_errors = sle.sle_data_count; if (!error && sle.sle_meta_count <= policy.zrp_maxmeta && sle.sle_data_count <= policy.zrp_maxdata) { int64_t loss = 0; verify_ok = B_TRUE; spa->spa_load_txg = spa->spa_uberblock.ub_txg; spa->spa_load_txg_ts = spa->spa_uberblock.ub_timestamp; loss = spa->spa_last_ubsync_txg_ts - spa->spa_load_txg_ts; VERIFY(nvlist_add_uint64(spa->spa_load_info, ZPOOL_CONFIG_LOAD_TIME, spa->spa_load_txg_ts) == 0); VERIFY(nvlist_add_int64(spa->spa_load_info, ZPOOL_CONFIG_REWIND_TIME, loss) == 0); VERIFY(nvlist_add_uint64(spa->spa_load_info, ZPOOL_CONFIG_LOAD_DATA_ERRORS, sle.sle_data_count) == 0); } else { spa->spa_load_max_txg = spa->spa_uberblock.ub_txg; } if (error) { if (error != ENXIO && error != EIO) error = SET_ERROR(EIO); return (error); } return (verify_ok ? 0 : EIO); } /* * Find a value in the pool props object. */ static void spa_prop_find(spa_t *spa, zpool_prop_t prop, uint64_t *val) { (void) zap_lookup(spa->spa_meta_objset, spa->spa_pool_props_object, zpool_prop_to_name(prop), sizeof (uint64_t), 1, val); } /* * Find a value in the pool directory object. */ static int spa_dir_prop(spa_t *spa, const char *name, uint64_t *val) { return (zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, name, sizeof (uint64_t), 1, val)); } static int spa_vdev_err(vdev_t *vdev, vdev_aux_t aux, int err) { vdev_set_state(vdev, B_TRUE, VDEV_STATE_CANT_OPEN, aux); return (err); } /* * Fix up config after a partly-completed split. This is done with the * ZPOOL_CONFIG_SPLIT nvlist. Both the splitting pool and the split-off * pool have that entry in their config, but only the splitting one contains * a list of all the guids of the vdevs that are being split off. * * This function determines what to do with that list: either rejoin * all the disks to the pool, or complete the splitting process. To attempt * the rejoin, each disk that is offlined is marked online again, and * we do a reopen() call. If the vdev label for every disk that was * marked online indicates it was successfully split off (VDEV_AUX_SPLIT_POOL) * then we call vdev_split() on each disk, and complete the split. * * Otherwise we leave the config alone, with all the vdevs in place in * the original pool. */ static void spa_try_repair(spa_t *spa, nvlist_t *config) { uint_t extracted; uint64_t *glist; uint_t i, gcount; nvlist_t *nvl; vdev_t **vd; boolean_t attempt_reopen; if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_SPLIT, &nvl) != 0) return; /* check that the config is complete */ if (nvlist_lookup_uint64_array(nvl, ZPOOL_CONFIG_SPLIT_LIST, &glist, &gcount) != 0) return; vd = kmem_zalloc(gcount * sizeof (vdev_t *), KM_PUSHPAGE); /* attempt to online all the vdevs & validate */ attempt_reopen = B_TRUE; for (i = 0; i < gcount; i++) { if (glist[i] == 0) /* vdev is hole */ continue; vd[i] = spa_lookup_by_guid(spa, glist[i], B_FALSE); if (vd[i] == NULL) { /* * Don't bother attempting to reopen the disks; * just do the split. */ attempt_reopen = B_FALSE; } else { /* attempt to re-online it */ vd[i]->vdev_offline = B_FALSE; } } if (attempt_reopen) { vdev_reopen(spa->spa_root_vdev); /* check each device to see what state it's in */ for (extracted = 0, i = 0; i < gcount; i++) { if (vd[i] != NULL && vd[i]->vdev_stat.vs_aux != VDEV_AUX_SPLIT_POOL) break; ++extracted; } } /* * If every disk has been moved to the new pool, or if we never * even attempted to look at them, then we split them off for * good. */ if (!attempt_reopen || gcount == extracted) { for (i = 0; i < gcount; i++) if (vd[i] != NULL) vdev_split(vd[i]); vdev_reopen(spa->spa_root_vdev); } kmem_free(vd, gcount * sizeof (vdev_t *)); } static int spa_load(spa_t *spa, spa_load_state_t state, spa_import_type_t type, boolean_t mosconfig) { nvlist_t *config = spa->spa_config; char *ereport = FM_EREPORT_ZFS_POOL; char *comment; int error; uint64_t pool_guid; nvlist_t *nvl; if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &pool_guid)) return (SET_ERROR(EINVAL)); ASSERT(spa->spa_comment == NULL); if (nvlist_lookup_string(config, ZPOOL_CONFIG_COMMENT, &comment) == 0) spa->spa_comment = spa_strdup(comment); /* * Versioning wasn't explicitly added to the label until later, so if * it's not present treat it as the initial version. */ if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION, &spa->spa_ubsync.ub_version) != 0) spa->spa_ubsync.ub_version = SPA_VERSION_INITIAL; (void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_TXG, &spa->spa_config_txg); if ((state == SPA_LOAD_IMPORT || state == SPA_LOAD_TRYIMPORT) && spa_guid_exists(pool_guid, 0)) { error = SET_ERROR(EEXIST); } else { spa->spa_config_guid = pool_guid; if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_SPLIT, &nvl) == 0) { VERIFY(nvlist_dup(nvl, &spa->spa_config_splitting, KM_PUSHPAGE) == 0); } nvlist_free(spa->spa_load_info); spa->spa_load_info = fnvlist_alloc(); gethrestime(&spa->spa_loaded_ts); error = spa_load_impl(spa, pool_guid, config, state, type, mosconfig, &ereport); } spa->spa_minref = refcount_count(&spa->spa_refcount); if (error) { if (error != EEXIST) { spa->spa_loaded_ts.tv_sec = 0; spa->spa_loaded_ts.tv_nsec = 0; } if (error != EBADF) { zfs_ereport_post(ereport, spa, NULL, NULL, 0, 0); } } spa->spa_load_state = error ? SPA_LOAD_ERROR : SPA_LOAD_NONE; spa->spa_ena = 0; return (error); } /* * Load an existing storage pool, using the pool's builtin spa_config as a * source of configuration information. */ __attribute__((always_inline)) static inline int spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config, spa_load_state_t state, spa_import_type_t type, boolean_t mosconfig, char **ereport) { int error = 0; nvlist_t *nvroot = NULL; nvlist_t *label; vdev_t *rvd; uberblock_t *ub = &spa->spa_uberblock; uint64_t children, config_cache_txg = spa->spa_config_txg; int orig_mode = spa->spa_mode; int parse, i; uint64_t obj; boolean_t missing_feat_write = B_FALSE; /* * If this is an untrusted config, access the pool in read-only mode. * This prevents things like resilvering recently removed devices. */ if (!mosconfig) spa->spa_mode = FREAD; ASSERT(MUTEX_HELD(&spa_namespace_lock)); spa->spa_load_state = state; if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvroot)) return (SET_ERROR(EINVAL)); parse = (type == SPA_IMPORT_EXISTING ? VDEV_ALLOC_LOAD : VDEV_ALLOC_SPLIT); /* * Create "The Godfather" zio to hold all async IOs */ spa->spa_async_zio_root = kmem_alloc(max_ncpus * sizeof (void *), KM_SLEEP); for (i = 0; i < max_ncpus; i++) { spa->spa_async_zio_root[i] = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE | ZIO_FLAG_GODFATHER); } /* * Parse the configuration into a vdev tree. We explicitly set the * value that will be returned by spa_version() since parsing the * configuration requires knowing the version number. */ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); error = spa_config_parse(spa, &rvd, nvroot, NULL, 0, parse); spa_config_exit(spa, SCL_ALL, FTAG); if (error != 0) return (error); ASSERT(spa->spa_root_vdev == rvd); if (type != SPA_IMPORT_ASSEMBLE) { ASSERT(spa_guid(spa) == pool_guid); } /* * Try to open all vdevs, loading each label in the process. */ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); error = vdev_open(rvd); spa_config_exit(spa, SCL_ALL, FTAG); if (error != 0) return (error); /* * We need to validate the vdev labels against the configuration that * we have in hand, which is dependent on the setting of mosconfig. If * mosconfig is true then we're validating the vdev labels based on * that config. Otherwise, we're validating against the cached config * (zpool.cache) that was read when we loaded the zfs module, and then * later we will recursively call spa_load() and validate against * the vdev config. * * If we're assembling a new pool that's been split off from an * existing pool, the labels haven't yet been updated so we skip * validation for now. */ if (type != SPA_IMPORT_ASSEMBLE) { spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); error = vdev_validate(rvd, mosconfig); spa_config_exit(spa, SCL_ALL, FTAG); if (error != 0) return (error); if (rvd->vdev_state <= VDEV_STATE_CANT_OPEN) return (SET_ERROR(ENXIO)); } /* * Find the best uberblock. */ vdev_uberblock_load(rvd, ub, &label); /* * If we weren't able to find a single valid uberblock, return failure. */ if (ub->ub_txg == 0) { nvlist_free(label); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, ENXIO)); } /* * If the pool has an unsupported version we can't open it. */ if (!SPA_VERSION_IS_SUPPORTED(ub->ub_version)) { nvlist_free(label); return (spa_vdev_err(rvd, VDEV_AUX_VERSION_NEWER, ENOTSUP)); } if (ub->ub_version >= SPA_VERSION_FEATURES) { nvlist_t *features; /* * If we weren't able to find what's necessary for reading the * MOS in the label, return failure. */ if (label == NULL || nvlist_lookup_nvlist(label, ZPOOL_CONFIG_FEATURES_FOR_READ, &features) != 0) { nvlist_free(label); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, ENXIO)); } /* * Update our in-core representation with the definitive values * from the label. */ nvlist_free(spa->spa_label_features); VERIFY(nvlist_dup(features, &spa->spa_label_features, 0) == 0); } nvlist_free(label); /* * Look through entries in the label nvlist's features_for_read. If * there is a feature listed there which we don't understand then we * cannot open a pool. */ if (ub->ub_version >= SPA_VERSION_FEATURES) { nvlist_t *unsup_feat; nvpair_t *nvp; VERIFY(nvlist_alloc(&unsup_feat, NV_UNIQUE_NAME, KM_SLEEP) == 0); for (nvp = nvlist_next_nvpair(spa->spa_label_features, NULL); nvp != NULL; nvp = nvlist_next_nvpair(spa->spa_label_features, nvp)) { if (!zfeature_is_supported(nvpair_name(nvp))) { VERIFY(nvlist_add_string(unsup_feat, nvpair_name(nvp), "") == 0); } } if (!nvlist_empty(unsup_feat)) { VERIFY(nvlist_add_nvlist(spa->spa_load_info, ZPOOL_CONFIG_UNSUP_FEAT, unsup_feat) == 0); nvlist_free(unsup_feat); return (spa_vdev_err(rvd, VDEV_AUX_UNSUP_FEAT, ENOTSUP)); } nvlist_free(unsup_feat); } /* * If the vdev guid sum doesn't match the uberblock, we have an * incomplete configuration. We first check to see if the pool * is aware of the complete config (i.e ZPOOL_CONFIG_VDEV_CHILDREN). * If it is, defer the vdev_guid_sum check till later so we * can handle missing vdevs. */ if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_VDEV_CHILDREN, &children) != 0 && mosconfig && type != SPA_IMPORT_ASSEMBLE && rvd->vdev_guid_sum != ub->ub_guid_sum) return (spa_vdev_err(rvd, VDEV_AUX_BAD_GUID_SUM, ENXIO)); if (type != SPA_IMPORT_ASSEMBLE && spa->spa_config_splitting) { spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_try_repair(spa, config); spa_config_exit(spa, SCL_ALL, FTAG); nvlist_free(spa->spa_config_splitting); spa->spa_config_splitting = NULL; } /* * Initialize internal SPA structures. */ spa->spa_state = POOL_STATE_ACTIVE; spa->spa_ubsync = spa->spa_uberblock; spa->spa_verify_min_txg = spa->spa_extreme_rewind ? TXG_INITIAL - 1 : spa_last_synced_txg(spa) - TXG_DEFER_SIZE - 1; spa->spa_first_txg = spa->spa_last_ubsync_txg ? spa->spa_last_ubsync_txg : spa_last_synced_txg(spa) + 1; spa->spa_claim_max_txg = spa->spa_first_txg; spa->spa_prev_software_version = ub->ub_software_version; error = dsl_pool_init(spa, spa->spa_first_txg, &spa->spa_dsl_pool); if (error) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); spa->spa_meta_objset = spa->spa_dsl_pool->dp_meta_objset; if (spa_dir_prop(spa, DMU_POOL_CONFIG, &spa->spa_config_object) != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); if (spa_version(spa) >= SPA_VERSION_FEATURES) { boolean_t missing_feat_read = B_FALSE; nvlist_t *unsup_feat, *enabled_feat; spa_feature_t i; if (spa_dir_prop(spa, DMU_POOL_FEATURES_FOR_READ, &spa->spa_feat_for_read_obj) != 0) { return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } if (spa_dir_prop(spa, DMU_POOL_FEATURES_FOR_WRITE, &spa->spa_feat_for_write_obj) != 0) { return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } if (spa_dir_prop(spa, DMU_POOL_FEATURE_DESCRIPTIONS, &spa->spa_feat_desc_obj) != 0) { return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } enabled_feat = fnvlist_alloc(); unsup_feat = fnvlist_alloc(); if (!spa_features_check(spa, B_FALSE, unsup_feat, enabled_feat)) missing_feat_read = B_TRUE; if (spa_writeable(spa) || state == SPA_LOAD_TRYIMPORT) { if (!spa_features_check(spa, B_TRUE, unsup_feat, enabled_feat)) { missing_feat_write = B_TRUE; } } fnvlist_add_nvlist(spa->spa_load_info, ZPOOL_CONFIG_ENABLED_FEAT, enabled_feat); if (!nvlist_empty(unsup_feat)) { fnvlist_add_nvlist(spa->spa_load_info, ZPOOL_CONFIG_UNSUP_FEAT, unsup_feat); } fnvlist_free(enabled_feat); fnvlist_free(unsup_feat); if (!missing_feat_read) { fnvlist_add_boolean(spa->spa_load_info, ZPOOL_CONFIG_CAN_RDONLY); } /* * If the state is SPA_LOAD_TRYIMPORT, our objective is * twofold: to determine whether the pool is available for * import in read-write mode and (if it is not) whether the * pool is available for import in read-only mode. If the pool * is available for import in read-write mode, it is displayed * as available in userland; if it is not available for import * in read-only mode, it is displayed as unavailable in * userland. If the pool is available for import in read-only * mode but not read-write mode, it is displayed as unavailable * in userland with a special note that the pool is actually * available for open in read-only mode. * * As a result, if the state is SPA_LOAD_TRYIMPORT and we are * missing a feature for write, we must first determine whether * the pool can be opened read-only before returning to * userland in order to know whether to display the * abovementioned note. */ if (missing_feat_read || (missing_feat_write && spa_writeable(spa))) { return (spa_vdev_err(rvd, VDEV_AUX_UNSUP_FEAT, ENOTSUP)); } /* * Load refcounts for ZFS features from disk into an in-memory * cache during SPA initialization. */ for (i = 0; i < SPA_FEATURES; i++) { uint64_t refcount; error = feature_get_refcount_from_disk(spa, &spa_feature_table[i], &refcount); if (error == 0) { spa->spa_feat_refcount_cache[i] = refcount; } else if (error == ENOTSUP) { spa->spa_feat_refcount_cache[i] = SPA_FEATURE_DISABLED; } else { return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } } } if (spa_feature_is_active(spa, SPA_FEATURE_ENABLED_TXG)) { if (spa_dir_prop(spa, DMU_POOL_FEATURE_ENABLED_TXG, &spa->spa_feat_enabled_txg_obj) != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } spa->spa_is_initializing = B_TRUE; error = dsl_pool_open(spa->spa_dsl_pool); spa->spa_is_initializing = B_FALSE; if (error != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); if (!mosconfig) { uint64_t hostid; nvlist_t *policy = NULL, *nvconfig; if (load_nvlist(spa, spa->spa_config_object, &nvconfig) != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); if (!spa_is_root(spa) && nvlist_lookup_uint64(nvconfig, ZPOOL_CONFIG_HOSTID, &hostid) == 0) { char *hostname; unsigned long myhostid = 0; VERIFY(nvlist_lookup_string(nvconfig, ZPOOL_CONFIG_HOSTNAME, &hostname) == 0); #ifdef _KERNEL myhostid = zone_get_hostid(NULL); #else /* _KERNEL */ /* * We're emulating the system's hostid in userland, so * we can't use zone_get_hostid(). */ (void) ddi_strtoul(hw_serial, NULL, 10, &myhostid); #endif /* _KERNEL */ if (hostid != 0 && myhostid != 0 && hostid != myhostid) { nvlist_free(nvconfig); cmn_err(CE_WARN, "pool '%s' could not be " "loaded as it was last accessed by another " "system (host: %s hostid: 0x%lx). See: " "http://zfsonlinux.org/msg/ZFS-8000-EY", spa_name(spa), hostname, (unsigned long)hostid); return (SET_ERROR(EBADF)); } } if (nvlist_lookup_nvlist(spa->spa_config, ZPOOL_REWIND_POLICY, &policy) == 0) VERIFY(nvlist_add_nvlist(nvconfig, ZPOOL_REWIND_POLICY, policy) == 0); spa_config_set(spa, nvconfig); spa_unload(spa); spa_deactivate(spa); spa_activate(spa, orig_mode); return (spa_load(spa, state, SPA_IMPORT_EXISTING, B_TRUE)); } if (spa_dir_prop(spa, DMU_POOL_SYNC_BPOBJ, &obj) != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); error = bpobj_open(&spa->spa_deferred_bpobj, spa->spa_meta_objset, obj); if (error != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); /* * Load the bit that tells us to use the new accounting function * (raid-z deflation). If we have an older pool, this will not * be present. */ error = spa_dir_prop(spa, DMU_POOL_DEFLATE, &spa->spa_deflate); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); error = spa_dir_prop(spa, DMU_POOL_CREATION_VERSION, &spa->spa_creation_version); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); /* * Load the persistent error log. If we have an older pool, this will * not be present. */ error = spa_dir_prop(spa, DMU_POOL_ERRLOG_LAST, &spa->spa_errlog_last); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); error = spa_dir_prop(spa, DMU_POOL_ERRLOG_SCRUB, &spa->spa_errlog_scrub); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); /* * Load the history object. If we have an older pool, this * will not be present. */ error = spa_dir_prop(spa, DMU_POOL_HISTORY, &spa->spa_history); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); /* * If we're assembling the pool from the split-off vdevs of * an existing pool, we don't want to attach the spares & cache * devices. */ /* * Load any hot spares for this pool. */ error = spa_dir_prop(spa, DMU_POOL_SPARES, &spa->spa_spares.sav_object); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); if (error == 0 && type != SPA_IMPORT_ASSEMBLE) { ASSERT(spa_version(spa) >= SPA_VERSION_SPARES); if (load_nvlist(spa, spa->spa_spares.sav_object, &spa->spa_spares.sav_config) != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_spares(spa); spa_config_exit(spa, SCL_ALL, FTAG); } else if (error == 0) { spa->spa_spares.sav_sync = B_TRUE; } /* * Load any level 2 ARC devices for this pool. */ error = spa_dir_prop(spa, DMU_POOL_L2CACHE, &spa->spa_l2cache.sav_object); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); if (error == 0 && type != SPA_IMPORT_ASSEMBLE) { ASSERT(spa_version(spa) >= SPA_VERSION_L2CACHE); if (load_nvlist(spa, spa->spa_l2cache.sav_object, &spa->spa_l2cache.sav_config) != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_l2cache(spa); spa_config_exit(spa, SCL_ALL, FTAG); } else if (error == 0) { spa->spa_l2cache.sav_sync = B_TRUE; } spa->spa_delegation = zpool_prop_default_numeric(ZPOOL_PROP_DELEGATION); error = spa_dir_prop(spa, DMU_POOL_PROPS, &spa->spa_pool_props_object); if (error && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); if (error == 0) { uint64_t autoreplace = 0; spa_prop_find(spa, ZPOOL_PROP_BOOTFS, &spa->spa_bootfs); spa_prop_find(spa, ZPOOL_PROP_AUTOREPLACE, &autoreplace); spa_prop_find(spa, ZPOOL_PROP_DELEGATION, &spa->spa_delegation); spa_prop_find(spa, ZPOOL_PROP_FAILUREMODE, &spa->spa_failmode); spa_prop_find(spa, ZPOOL_PROP_AUTOEXPAND, &spa->spa_autoexpand); spa_prop_find(spa, ZPOOL_PROP_DEDUPDITTO, &spa->spa_dedup_ditto); spa->spa_autoreplace = (autoreplace != 0); } /* * If the 'autoreplace' property is set, then post a resource notifying * the ZFS DE that it should not issue any faults for unopenable * devices. We also iterate over the vdevs, and post a sysevent for any * unopenable vdevs so that the normal autoreplace handler can take * over. */ if (spa->spa_autoreplace && state != SPA_LOAD_TRYIMPORT) { spa_check_removed(spa->spa_root_vdev); /* * For the import case, this is done in spa_import(), because * at this point we're using the spare definitions from * the MOS config, not necessarily from the userland config. */ if (state != SPA_LOAD_IMPORT) { spa_aux_check_removed(&spa->spa_spares); spa_aux_check_removed(&spa->spa_l2cache); } } /* * Load the vdev state for all toplevel vdevs. */ vdev_load(rvd); /* * Propagate the leaf DTLs we just loaded all the way up the tree. */ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); vdev_dtl_reassess(rvd, 0, 0, B_FALSE); spa_config_exit(spa, SCL_ALL, FTAG); /* * Load the DDTs (dedup tables). */ error = ddt_load(spa); if (error != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); spa_update_dspace(spa); /* * Validate the config, using the MOS config to fill in any * information which might be missing. If we fail to validate * the config then declare the pool unfit for use. If we're * assembling a pool from a split, the log is not transferred * over. */ if (type != SPA_IMPORT_ASSEMBLE) { nvlist_t *nvconfig; if (load_nvlist(spa, spa->spa_config_object, &nvconfig) != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); if (!spa_config_valid(spa, nvconfig)) { nvlist_free(nvconfig); return (spa_vdev_err(rvd, VDEV_AUX_BAD_GUID_SUM, ENXIO)); } nvlist_free(nvconfig); /* * Now that we've validated the config, check the state of the * root vdev. If it can't be opened, it indicates one or * more toplevel vdevs are faulted. */ if (rvd->vdev_state <= VDEV_STATE_CANT_OPEN) return (SET_ERROR(ENXIO)); if (spa_check_logs(spa)) { *ereport = FM_EREPORT_ZFS_LOG_REPLAY; return (spa_vdev_err(rvd, VDEV_AUX_BAD_LOG, ENXIO)); } } if (missing_feat_write) { ASSERT(state == SPA_LOAD_TRYIMPORT); /* * At this point, we know that we can open the pool in * read-only mode but not read-write mode. We now have enough * information and can return to userland. */ return (spa_vdev_err(rvd, VDEV_AUX_UNSUP_FEAT, ENOTSUP)); } /* * We've successfully opened the pool, verify that we're ready * to start pushing transactions. */ if (state != SPA_LOAD_TRYIMPORT) { if ((error = spa_load_verify(spa))) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, error)); } if (spa_writeable(spa) && (state == SPA_LOAD_RECOVER || spa->spa_load_max_txg == UINT64_MAX)) { dmu_tx_t *tx; int need_update = B_FALSE; int c; ASSERT(state != SPA_LOAD_TRYIMPORT); /* * Claim log blocks that haven't been committed yet. * This must all happen in a single txg. * Note: spa_claim_max_txg is updated by spa_claim_notify(), * invoked from zil_claim_log_block()'s i/o done callback. * Price of rollback is that we abandon the log. */ spa->spa_claiming = B_TRUE; tx = dmu_tx_create_assigned(spa_get_dsl(spa), spa_first_txg(spa)); (void) dmu_objset_find(spa_name(spa), zil_claim, tx, DS_FIND_CHILDREN); dmu_tx_commit(tx); spa->spa_claiming = B_FALSE; spa_set_log_state(spa, SPA_LOG_GOOD); spa->spa_sync_on = B_TRUE; txg_sync_start(spa->spa_dsl_pool); /* * Wait for all claims to sync. We sync up to the highest * claimed log block birth time so that claimed log blocks * don't appear to be from the future. spa_claim_max_txg * will have been set for us by either zil_check_log_chain() * (invoked from spa_check_logs()) or zil_claim() above. */ txg_wait_synced(spa->spa_dsl_pool, spa->spa_claim_max_txg); /* * If the config cache is stale, or we have uninitialized * metaslabs (see spa_vdev_add()), then update the config. * * If this is a verbatim import, trust the current * in-core spa_config and update the disk labels. */ if (config_cache_txg != spa->spa_config_txg || state == SPA_LOAD_IMPORT || state == SPA_LOAD_RECOVER || (spa->spa_import_flags & ZFS_IMPORT_VERBATIM)) need_update = B_TRUE; for (c = 0; c < rvd->vdev_children; c++) if (rvd->vdev_child[c]->vdev_ms_array == 0) need_update = B_TRUE; /* * Update the config cache asychronously in case we're the * root pool, in which case the config cache isn't writable yet. */ if (need_update) spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE); /* * Check all DTLs to see if anything needs resilvering. */ if (!dsl_scan_resilvering(spa->spa_dsl_pool) && vdev_resilver_needed(rvd, NULL, NULL)) spa_async_request(spa, SPA_ASYNC_RESILVER); /* * Log the fact that we booted up (so that we can detect if * we rebooted in the middle of an operation). */ spa_history_log_version(spa, "open"); /* * Delete any inconsistent datasets. */ (void) dmu_objset_find(spa_name(spa), dsl_destroy_inconsistent, NULL, DS_FIND_CHILDREN); /* * Clean up any stale temporary dataset userrefs. */ dsl_pool_clean_tmp_userrefs(spa->spa_dsl_pool); } return (0); } static int spa_load_retry(spa_t *spa, spa_load_state_t state, int mosconfig) { int mode = spa->spa_mode; spa_unload(spa); spa_deactivate(spa); spa->spa_load_max_txg = spa->spa_uberblock.ub_txg - 1; spa_activate(spa, mode); spa_async_suspend(spa); return (spa_load(spa, state, SPA_IMPORT_EXISTING, mosconfig)); } /* * If spa_load() fails this function will try loading prior txg's. If * 'state' is SPA_LOAD_RECOVER and one of these loads succeeds the pool * will be rewound to that txg. If 'state' is not SPA_LOAD_RECOVER this * function will not rewind the pool and will return the same error as * spa_load(). */ static int spa_load_best(spa_t *spa, spa_load_state_t state, int mosconfig, uint64_t max_request, int rewind_flags) { nvlist_t *loadinfo = NULL; nvlist_t *config = NULL; int load_error, rewind_error; uint64_t safe_rewind_txg; uint64_t min_txg; if (spa->spa_load_txg && state == SPA_LOAD_RECOVER) { spa->spa_load_max_txg = spa->spa_load_txg; spa_set_log_state(spa, SPA_LOG_CLEAR); } else { spa->spa_load_max_txg = max_request; if (max_request != UINT64_MAX) spa->spa_extreme_rewind = B_TRUE; } load_error = rewind_error = spa_load(spa, state, SPA_IMPORT_EXISTING, mosconfig); if (load_error == 0) return (0); if (spa->spa_root_vdev != NULL) config = spa_config_generate(spa, NULL, -1ULL, B_TRUE); spa->spa_last_ubsync_txg = spa->spa_uberblock.ub_txg; spa->spa_last_ubsync_txg_ts = spa->spa_uberblock.ub_timestamp; if (rewind_flags & ZPOOL_NEVER_REWIND) { nvlist_free(config); return (load_error); } if (state == SPA_LOAD_RECOVER) { /* Price of rolling back is discarding txgs, including log */ spa_set_log_state(spa, SPA_LOG_CLEAR); } else { /* * If we aren't rolling back save the load info from our first * import attempt so that we can restore it after attempting * to rewind. */ loadinfo = spa->spa_load_info; spa->spa_load_info = fnvlist_alloc(); } spa->spa_load_max_txg = spa->spa_last_ubsync_txg; safe_rewind_txg = spa->spa_last_ubsync_txg - TXG_DEFER_SIZE; min_txg = (rewind_flags & ZPOOL_EXTREME_REWIND) ? TXG_INITIAL : safe_rewind_txg; /* * Continue as long as we're finding errors, we're still within * the acceptable rewind range, and we're still finding uberblocks */ while (rewind_error && spa->spa_uberblock.ub_txg >= min_txg && spa->spa_uberblock.ub_txg <= spa->spa_load_max_txg) { if (spa->spa_load_max_txg < safe_rewind_txg) spa->spa_extreme_rewind = B_TRUE; rewind_error = spa_load_retry(spa, state, mosconfig); } spa->spa_extreme_rewind = B_FALSE; spa->spa_load_max_txg = UINT64_MAX; if (config && (rewind_error || state != SPA_LOAD_RECOVER)) spa_config_set(spa, config); if (state == SPA_LOAD_RECOVER) { ASSERT3P(loadinfo, ==, NULL); return (rewind_error); } else { /* Store the rewind info as part of the initial load info */ fnvlist_add_nvlist(loadinfo, ZPOOL_CONFIG_REWIND_INFO, spa->spa_load_info); /* Restore the initial load info */ fnvlist_free(spa->spa_load_info); spa->spa_load_info = loadinfo; return (load_error); } } /* * Pool Open/Import * * The import case is identical to an open except that the configuration is sent * down from userland, instead of grabbed from the configuration cache. For the * case of an open, the pool configuration will exist in the * POOL_STATE_UNINITIALIZED state. * * The stats information (gen/count/ustats) is used to gather vdev statistics at * the same time open the pool, without having to keep around the spa_t in some * ambiguous state. */ static int spa_open_common(const char *pool, spa_t **spapp, void *tag, nvlist_t *nvpolicy, nvlist_t **config) { spa_t *spa; spa_load_state_t state = SPA_LOAD_OPEN; int error; int locked = B_FALSE; int firstopen = B_FALSE; *spapp = NULL; /* * As disgusting as this is, we need to support recursive calls to this * function because dsl_dir_open() is called during spa_load(), and ends * up calling spa_open() again. The real fix is to figure out how to * avoid dsl_dir_open() calling this in the first place. */ if (mutex_owner(&spa_namespace_lock) != curthread) { mutex_enter(&spa_namespace_lock); locked = B_TRUE; } if ((spa = spa_lookup(pool)) == NULL) { if (locked) mutex_exit(&spa_namespace_lock); return (SET_ERROR(ENOENT)); } if (spa->spa_state == POOL_STATE_UNINITIALIZED) { zpool_rewind_policy_t policy; firstopen = B_TRUE; zpool_get_rewind_policy(nvpolicy ? nvpolicy : spa->spa_config, &policy); if (policy.zrp_request & ZPOOL_DO_REWIND) state = SPA_LOAD_RECOVER; spa_activate(spa, spa_mode_global); if (state != SPA_LOAD_RECOVER) spa->spa_last_ubsync_txg = spa->spa_load_txg = 0; error = spa_load_best(spa, state, B_FALSE, policy.zrp_txg, policy.zrp_request); if (error == EBADF) { /* * If vdev_validate() returns failure (indicated by * EBADF), it indicates that one of the vdevs indicates * that the pool has been exported or destroyed. If * this is the case, the config cache is out of sync and * we should remove the pool from the namespace. */ spa_unload(spa); spa_deactivate(spa); spa_config_sync(spa, B_TRUE, B_TRUE); spa_remove(spa); if (locked) mutex_exit(&spa_namespace_lock); return (SET_ERROR(ENOENT)); } if (error) { /* * We can't open the pool, but we still have useful * information: the state of each vdev after the * attempted vdev_open(). Return this to the user. */ if (config != NULL && spa->spa_config) { VERIFY(nvlist_dup(spa->spa_config, config, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_nvlist(*config, ZPOOL_CONFIG_LOAD_INFO, spa->spa_load_info) == 0); } spa_unload(spa); spa_deactivate(spa); spa->spa_last_open_failed = error; if (locked) mutex_exit(&spa_namespace_lock); *spapp = NULL; return (error); } } spa_open_ref(spa, tag); if (config != NULL) *config = spa_config_generate(spa, NULL, -1ULL, B_TRUE); /* * If we've recovered the pool, pass back any information we * gathered while doing the load. */ if (state == SPA_LOAD_RECOVER) { VERIFY(nvlist_add_nvlist(*config, ZPOOL_CONFIG_LOAD_INFO, spa->spa_load_info) == 0); } if (locked) { spa->spa_last_open_failed = 0; spa->spa_last_ubsync_txg = 0; spa->spa_load_txg = 0; mutex_exit(&spa_namespace_lock); } #ifdef _KERNEL if (firstopen) zvol_create_minors(spa->spa_name); #endif *spapp = spa; return (0); } int spa_open_rewind(const char *name, spa_t **spapp, void *tag, nvlist_t *policy, nvlist_t **config) { return (spa_open_common(name, spapp, tag, policy, config)); } int spa_open(const char *name, spa_t **spapp, void *tag) { return (spa_open_common(name, spapp, tag, NULL, NULL)); } /* * Lookup the given spa_t, incrementing the inject count in the process, * preventing it from being exported or destroyed. */ spa_t * spa_inject_addref(char *name) { spa_t *spa; mutex_enter(&spa_namespace_lock); if ((spa = spa_lookup(name)) == NULL) { mutex_exit(&spa_namespace_lock); return (NULL); } spa->spa_inject_ref++; mutex_exit(&spa_namespace_lock); return (spa); } void spa_inject_delref(spa_t *spa) { mutex_enter(&spa_namespace_lock); spa->spa_inject_ref--; mutex_exit(&spa_namespace_lock); } /* * Add spares device information to the nvlist. */ static void spa_add_spares(spa_t *spa, nvlist_t *config) { nvlist_t **spares; uint_t i, nspares; nvlist_t *nvroot; uint64_t guid; vdev_stat_t *vs; uint_t vsc; uint64_t pool; ASSERT(spa_config_held(spa, SCL_CONFIG, RW_READER)); if (spa->spa_spares.sav_count == 0) return; VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0); VERIFY(nvlist_lookup_nvlist_array(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0); if (nspares != 0) { VERIFY(nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, spares, nspares) == 0); VERIFY(nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0); /* * Go through and find any spares which have since been * repurposed as an active spare. If this is the case, update * their status appropriately. */ for (i = 0; i < nspares; i++) { VERIFY(nvlist_lookup_uint64(spares[i], ZPOOL_CONFIG_GUID, &guid) == 0); if (spa_spare_exists(guid, &pool, NULL) && pool != 0ULL) { VERIFY(nvlist_lookup_uint64_array( spares[i], ZPOOL_CONFIG_VDEV_STATS, (uint64_t **)&vs, &vsc) == 0); vs->vs_state = VDEV_STATE_CANT_OPEN; vs->vs_aux = VDEV_AUX_SPARED; } } } } /* * Add l2cache device information to the nvlist, including vdev stats. */ static void spa_add_l2cache(spa_t *spa, nvlist_t *config) { nvlist_t **l2cache; uint_t i, j, nl2cache; nvlist_t *nvroot; uint64_t guid; vdev_t *vd; vdev_stat_t *vs; uint_t vsc; ASSERT(spa_config_held(spa, SCL_CONFIG, RW_READER)); if (spa->spa_l2cache.sav_count == 0) return; VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0); VERIFY(nvlist_lookup_nvlist_array(spa->spa_l2cache.sav_config, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0); if (nl2cache != 0) { VERIFY(nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, l2cache, nl2cache) == 0); VERIFY(nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0); /* * Update level 2 cache device stats. */ for (i = 0; i < nl2cache; i++) { VERIFY(nvlist_lookup_uint64(l2cache[i], ZPOOL_CONFIG_GUID, &guid) == 0); vd = NULL; for (j = 0; j < spa->spa_l2cache.sav_count; j++) { if (guid == spa->spa_l2cache.sav_vdevs[j]->vdev_guid) { vd = spa->spa_l2cache.sav_vdevs[j]; break; } } ASSERT(vd != NULL); VERIFY(nvlist_lookup_uint64_array(l2cache[i], ZPOOL_CONFIG_VDEV_STATS, (uint64_t **)&vs, &vsc) == 0); vdev_get_stats(vd, vs); } } } static void spa_add_feature_stats(spa_t *spa, nvlist_t *config) { nvlist_t *features; zap_cursor_t zc; zap_attribute_t za; ASSERT(spa_config_held(spa, SCL_CONFIG, RW_READER)); VERIFY(nvlist_alloc(&features, NV_UNIQUE_NAME, KM_SLEEP) == 0); if (spa->spa_feat_for_read_obj != 0) { for (zap_cursor_init(&zc, spa->spa_meta_objset, spa->spa_feat_for_read_obj); zap_cursor_retrieve(&zc, &za) == 0; zap_cursor_advance(&zc)) { ASSERT(za.za_integer_length == sizeof (uint64_t) && za.za_num_integers == 1); VERIFY3U(0, ==, nvlist_add_uint64(features, za.za_name, za.za_first_integer)); } zap_cursor_fini(&zc); } if (spa->spa_feat_for_write_obj != 0) { for (zap_cursor_init(&zc, spa->spa_meta_objset, spa->spa_feat_for_write_obj); zap_cursor_retrieve(&zc, &za) == 0; zap_cursor_advance(&zc)) { ASSERT(za.za_integer_length == sizeof (uint64_t) && za.za_num_integers == 1); VERIFY3U(0, ==, nvlist_add_uint64(features, za.za_name, za.za_first_integer)); } zap_cursor_fini(&zc); } VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_FEATURE_STATS, features) == 0); nvlist_free(features); } int spa_get_stats(const char *name, nvlist_t **config, char *altroot, size_t buflen) { int error; spa_t *spa; *config = NULL; error = spa_open_common(name, &spa, FTAG, NULL, config); if (spa != NULL) { /* * This still leaves a window of inconsistency where the spares * or l2cache devices could change and the config would be * self-inconsistent. */ spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); if (*config != NULL) { uint64_t loadtimes[2]; loadtimes[0] = spa->spa_loaded_ts.tv_sec; loadtimes[1] = spa->spa_loaded_ts.tv_nsec; VERIFY(nvlist_add_uint64_array(*config, ZPOOL_CONFIG_LOADED_TIME, loadtimes, 2) == 0); VERIFY(nvlist_add_uint64(*config, ZPOOL_CONFIG_ERRCOUNT, spa_get_errlog_size(spa)) == 0); if (spa_suspended(spa)) VERIFY(nvlist_add_uint64(*config, ZPOOL_CONFIG_SUSPENDED, spa->spa_failmode) == 0); spa_add_spares(spa, *config); spa_add_l2cache(spa, *config); spa_add_feature_stats(spa, *config); } } /* * We want to get the alternate root even for faulted pools, so we cheat * and call spa_lookup() directly. */ if (altroot) { if (spa == NULL) { mutex_enter(&spa_namespace_lock); spa = spa_lookup(name); if (spa) spa_altroot(spa, altroot, buflen); else altroot[0] = '\0'; spa = NULL; mutex_exit(&spa_namespace_lock); } else { spa_altroot(spa, altroot, buflen); } } if (spa != NULL) { spa_config_exit(spa, SCL_CONFIG, FTAG); spa_close(spa, FTAG); } return (error); } /* * Validate that the auxiliary device array is well formed. We must have an * array of nvlists, each which describes a valid leaf vdev. If this is an * import (mode is VDEV_ALLOC_SPARE), then we allow corrupted spares to be * specified, as long as they are well-formed. */ static int spa_validate_aux_devs(spa_t *spa, nvlist_t *nvroot, uint64_t crtxg, int mode, spa_aux_vdev_t *sav, const char *config, uint64_t version, vdev_labeltype_t label) { nvlist_t **dev; uint_t i, ndev; vdev_t *vd; int error; ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); /* * It's acceptable to have no devs specified. */ if (nvlist_lookup_nvlist_array(nvroot, config, &dev, &ndev) != 0) return (0); if (ndev == 0) return (SET_ERROR(EINVAL)); /* * Make sure the pool is formatted with a version that supports this * device type. */ if (spa_version(spa) < version) return (SET_ERROR(ENOTSUP)); /* * Set the pending device list so we correctly handle device in-use * checking. */ sav->sav_pending = dev; sav->sav_npending = ndev; for (i = 0; i < ndev; i++) { if ((error = spa_config_parse(spa, &vd, dev[i], NULL, 0, mode)) != 0) goto out; if (!vd->vdev_ops->vdev_op_leaf) { vdev_free(vd); error = SET_ERROR(EINVAL); goto out; } /* * The L2ARC currently only supports disk devices in * kernel context. For user-level testing, we allow it. */ #ifdef _KERNEL if ((strcmp(config, ZPOOL_CONFIG_L2CACHE) == 0) && strcmp(vd->vdev_ops->vdev_op_type, VDEV_TYPE_DISK) != 0) { error = SET_ERROR(ENOTBLK); vdev_free(vd); goto out; } #endif vd->vdev_top = vd; if ((error = vdev_open(vd)) == 0 && (error = vdev_label_init(vd, crtxg, label)) == 0) { VERIFY(nvlist_add_uint64(dev[i], ZPOOL_CONFIG_GUID, vd->vdev_guid) == 0); } vdev_free(vd); if (error && (mode != VDEV_ALLOC_SPARE && mode != VDEV_ALLOC_L2CACHE)) goto out; else error = 0; } out: sav->sav_pending = NULL; sav->sav_npending = 0; return (error); } static int spa_validate_aux(spa_t *spa, nvlist_t *nvroot, uint64_t crtxg, int mode) { int error; ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); if ((error = spa_validate_aux_devs(spa, nvroot, crtxg, mode, &spa->spa_spares, ZPOOL_CONFIG_SPARES, SPA_VERSION_SPARES, VDEV_LABEL_SPARE)) != 0) { return (error); } return (spa_validate_aux_devs(spa, nvroot, crtxg, mode, &spa->spa_l2cache, ZPOOL_CONFIG_L2CACHE, SPA_VERSION_L2CACHE, VDEV_LABEL_L2CACHE)); } static void spa_set_aux_vdevs(spa_aux_vdev_t *sav, nvlist_t **devs, int ndevs, const char *config) { int i; if (sav->sav_config != NULL) { nvlist_t **olddevs; uint_t oldndevs; nvlist_t **newdevs; /* * Generate new dev list by concatentating with the * current dev list. */ VERIFY(nvlist_lookup_nvlist_array(sav->sav_config, config, &olddevs, &oldndevs) == 0); newdevs = kmem_alloc(sizeof (void *) * (ndevs + oldndevs), KM_PUSHPAGE); for (i = 0; i < oldndevs; i++) VERIFY(nvlist_dup(olddevs[i], &newdevs[i], KM_PUSHPAGE) == 0); for (i = 0; i < ndevs; i++) VERIFY(nvlist_dup(devs[i], &newdevs[i + oldndevs], KM_PUSHPAGE) == 0); VERIFY(nvlist_remove(sav->sav_config, config, DATA_TYPE_NVLIST_ARRAY) == 0); VERIFY(nvlist_add_nvlist_array(sav->sav_config, config, newdevs, ndevs + oldndevs) == 0); for (i = 0; i < oldndevs + ndevs; i++) nvlist_free(newdevs[i]); kmem_free(newdevs, (oldndevs + ndevs) * sizeof (void *)); } else { /* * Generate a new dev list. */ VERIFY(nvlist_alloc(&sav->sav_config, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_nvlist_array(sav->sav_config, config, devs, ndevs) == 0); } } /* * Stop and drop level 2 ARC devices */ void spa_l2cache_drop(spa_t *spa) { vdev_t *vd; int i; spa_aux_vdev_t *sav = &spa->spa_l2cache; for (i = 0; i < sav->sav_count; i++) { uint64_t pool; vd = sav->sav_vdevs[i]; ASSERT(vd != NULL); if (spa_l2cache_exists(vd->vdev_guid, &pool) && pool != 0ULL && l2arc_vdev_present(vd)) l2arc_remove_vdev(vd); } } /* * Pool Creation */ int spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props, nvlist_t *zplprops) { spa_t *spa; char *altroot = NULL; vdev_t *rvd; dsl_pool_t *dp; dmu_tx_t *tx; int error = 0; uint64_t txg = TXG_INITIAL; nvlist_t **spares, **l2cache; uint_t nspares, nl2cache; uint64_t version, obj; boolean_t has_features; nvpair_t *elem; int c, i; char *poolname; nvlist_t *nvl; if (nvlist_lookup_string(props, "tname", &poolname) != 0) poolname = (char *)pool; /* * If this pool already exists, return failure. */ mutex_enter(&spa_namespace_lock); if (spa_lookup(poolname) != NULL) { mutex_exit(&spa_namespace_lock); return (SET_ERROR(EEXIST)); } /* * Allocate a new spa_t structure. */ nvl = fnvlist_alloc(); fnvlist_add_string(nvl, ZPOOL_CONFIG_POOL_NAME, pool); (void) nvlist_lookup_string(props, zpool_prop_to_name(ZPOOL_PROP_ALTROOT), &altroot); spa = spa_add(poolname, nvl, altroot); fnvlist_free(nvl); spa_activate(spa, spa_mode_global); if (props && (error = spa_prop_validate(spa, props))) { spa_deactivate(spa); spa_remove(spa); mutex_exit(&spa_namespace_lock); return (error); } /* * Temporary pool names should never be written to disk. */ if (poolname != pool) spa->spa_import_flags |= ZFS_IMPORT_TEMP_NAME; has_features = B_FALSE; for (elem = nvlist_next_nvpair(props, NULL); elem != NULL; elem = nvlist_next_nvpair(props, elem)) { if (zpool_prop_feature(nvpair_name(elem))) has_features = B_TRUE; } if (has_features || nvlist_lookup_uint64(props, zpool_prop_to_name(ZPOOL_PROP_VERSION), &version) != 0) { version = SPA_VERSION; } ASSERT(SPA_VERSION_IS_SUPPORTED(version)); spa->spa_first_txg = txg; spa->spa_uberblock.ub_txg = txg - 1; spa->spa_uberblock.ub_version = version; spa->spa_ubsync = spa->spa_uberblock; /* * Create "The Godfather" zio to hold all async IOs */ spa->spa_async_zio_root = kmem_alloc(max_ncpus * sizeof (void *), KM_SLEEP); for (i = 0; i < max_ncpus; i++) { spa->spa_async_zio_root[i] = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE | ZIO_FLAG_GODFATHER); } /* * Create the root vdev. */ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); error = spa_config_parse(spa, &rvd, nvroot, NULL, 0, VDEV_ALLOC_ADD); ASSERT(error != 0 || rvd != NULL); ASSERT(error != 0 || spa->spa_root_vdev == rvd); if (error == 0 && !zfs_allocatable_devs(nvroot)) error = SET_ERROR(EINVAL); if (error == 0 && (error = vdev_create(rvd, txg, B_FALSE)) == 0 && (error = spa_validate_aux(spa, nvroot, txg, VDEV_ALLOC_ADD)) == 0) { for (c = 0; c < rvd->vdev_children; c++) { vdev_metaslab_set_size(rvd->vdev_child[c]); vdev_expand(rvd->vdev_child[c], txg); } } spa_config_exit(spa, SCL_ALL, FTAG); if (error != 0) { spa_unload(spa); spa_deactivate(spa); spa_remove(spa); mutex_exit(&spa_namespace_lock); return (error); } /* * Get the list of spares, if specified. */ if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0) { VERIFY(nvlist_alloc(&spa->spa_spares.sav_config, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_nvlist_array(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, spares, nspares) == 0); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_spares(spa); spa_config_exit(spa, SCL_ALL, FTAG); spa->spa_spares.sav_sync = B_TRUE; } /* * Get the list of level 2 cache devices, if specified. */ if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0) { VERIFY(nvlist_alloc(&spa->spa_l2cache.sav_config, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_nvlist_array(spa->spa_l2cache.sav_config, ZPOOL_CONFIG_L2CACHE, l2cache, nl2cache) == 0); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_l2cache(spa); spa_config_exit(spa, SCL_ALL, FTAG); spa->spa_l2cache.sav_sync = B_TRUE; } spa->spa_is_initializing = B_TRUE; spa->spa_dsl_pool = dp = dsl_pool_create(spa, zplprops, txg); spa->spa_meta_objset = dp->dp_meta_objset; spa->spa_is_initializing = B_FALSE; /* * Create DDTs (dedup tables). */ ddt_create(spa); spa_update_dspace(spa); tx = dmu_tx_create_assigned(dp, txg); /* * Create the pool config object. */ spa->spa_config_object = dmu_object_alloc(spa->spa_meta_objset, DMU_OT_PACKED_NVLIST, SPA_CONFIG_BLOCKSIZE, DMU_OT_PACKED_NVLIST_SIZE, sizeof (uint64_t), tx); if (zap_add(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_CONFIG, sizeof (uint64_t), 1, &spa->spa_config_object, tx) != 0) { cmn_err(CE_PANIC, "failed to add pool config"); } if (spa_version(spa) >= SPA_VERSION_FEATURES) spa_feature_create_zap_objects(spa, tx); if (zap_add(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_CREATION_VERSION, sizeof (uint64_t), 1, &version, tx) != 0) { cmn_err(CE_PANIC, "failed to add pool version"); } /* Newly created pools with the right version are always deflated. */ if (version >= SPA_VERSION_RAIDZ_DEFLATE) { spa->spa_deflate = TRUE; if (zap_add(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_DEFLATE, sizeof (uint64_t), 1, &spa->spa_deflate, tx) != 0) { cmn_err(CE_PANIC, "failed to add deflate"); } } /* * Create the deferred-free bpobj. Turn off compression * because sync-to-convergence takes longer if the blocksize * keeps changing. */ obj = bpobj_alloc(spa->spa_meta_objset, 1 << 14, tx); dmu_object_set_compress(spa->spa_meta_objset, obj, ZIO_COMPRESS_OFF, tx); if (zap_add(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_SYNC_BPOBJ, sizeof (uint64_t), 1, &obj, tx) != 0) { cmn_err(CE_PANIC, "failed to add bpobj"); } VERIFY3U(0, ==, bpobj_open(&spa->spa_deferred_bpobj, spa->spa_meta_objset, obj)); /* * Create the pool's history object. */ if (version >= SPA_VERSION_ZPOOL_HISTORY) spa_history_create_obj(spa, tx); /* * Set pool properties. */ spa->spa_bootfs = zpool_prop_default_numeric(ZPOOL_PROP_BOOTFS); spa->spa_delegation = zpool_prop_default_numeric(ZPOOL_PROP_DELEGATION); spa->spa_failmode = zpool_prop_default_numeric(ZPOOL_PROP_FAILUREMODE); spa->spa_autoexpand = zpool_prop_default_numeric(ZPOOL_PROP_AUTOEXPAND); if (props != NULL) { spa_configfile_set(spa, props, B_FALSE); spa_sync_props(props, tx); } dmu_tx_commit(tx); spa->spa_sync_on = B_TRUE; txg_sync_start(spa->spa_dsl_pool); /* * We explicitly wait for the first transaction to complete so that our * bean counters are appropriately updated. */ txg_wait_synced(spa->spa_dsl_pool, txg); spa_config_sync(spa, B_FALSE, B_TRUE); spa_history_log_version(spa, "create"); spa->spa_minref = refcount_count(&spa->spa_refcount); mutex_exit(&spa_namespace_lock); return (0); } #ifdef _KERNEL /* * Get the root pool information from the root disk, then import the root pool * during the system boot up time. */ extern int vdev_disk_read_rootlabel(char *, char *, nvlist_t **); static nvlist_t * spa_generate_rootconf(char *devpath, char *devid, uint64_t *guid) { nvlist_t *config; nvlist_t *nvtop, *nvroot; uint64_t pgid; if (vdev_disk_read_rootlabel(devpath, devid, &config) != 0) return (NULL); /* * Add this top-level vdev to the child array. */ VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvtop) == 0); VERIFY(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &pgid) == 0); VERIFY(nvlist_lookup_uint64(config, ZPOOL_CONFIG_GUID, guid) == 0); /* * Put this pool's top-level vdevs into a root vdev. */ VERIFY(nvlist_alloc(&nvroot, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_string(nvroot, ZPOOL_CONFIG_TYPE, VDEV_TYPE_ROOT) == 0); VERIFY(nvlist_add_uint64(nvroot, ZPOOL_CONFIG_ID, 0ULL) == 0); VERIFY(nvlist_add_uint64(nvroot, ZPOOL_CONFIG_GUID, pgid) == 0); VERIFY(nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, &nvtop, 1) == 0); /* * Replace the existing vdev_tree with the new root vdev in * this pool's configuration (remove the old, add the new). */ VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, nvroot) == 0); nvlist_free(nvroot); return (config); } /* * Walk the vdev tree and see if we can find a device with "better" * configuration. A configuration is "better" if the label on that * device has a more recent txg. */ static void spa_alt_rootvdev(vdev_t *vd, vdev_t **avd, uint64_t *txg) { int c; for (c = 0; c < vd->vdev_children; c++) spa_alt_rootvdev(vd->vdev_child[c], avd, txg); if (vd->vdev_ops->vdev_op_leaf) { nvlist_t *label; uint64_t label_txg; if (vdev_disk_read_rootlabel(vd->vdev_physpath, vd->vdev_devid, &label) != 0) return; VERIFY(nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_TXG, &label_txg) == 0); /* * Do we have a better boot device? */ if (label_txg > *txg) { *txg = label_txg; *avd = vd; } nvlist_free(label); } } /* * Import a root pool. * * For x86. devpath_list will consist of devid and/or physpath name of * the vdev (e.g. "id1,sd@SSEAGATE..." or "/pci@1f,0/ide@d/disk@0,0:a"). * The GRUB "findroot" command will return the vdev we should boot. * * For Sparc, devpath_list consists the physpath name of the booting device * no matter the rootpool is a single device pool or a mirrored pool. * e.g. * "/pci@1f,0/ide@d/disk@0,0:a" */ int spa_import_rootpool(char *devpath, char *devid) { spa_t *spa; vdev_t *rvd, *bvd, *avd = NULL; nvlist_t *config, *nvtop; uint64_t guid, txg; char *pname; int error; /* * Read the label from the boot device and generate a configuration. */ config = spa_generate_rootconf(devpath, devid, &guid); #if defined(_OBP) && defined(_KERNEL) if (config == NULL) { if (strstr(devpath, "/iscsi/ssd") != NULL) { /* iscsi boot */ get_iscsi_bootpath_phy(devpath); config = spa_generate_rootconf(devpath, devid, &guid); } } #endif if (config == NULL) { cmn_err(CE_NOTE, "Cannot read the pool label from '%s'", devpath); return (SET_ERROR(EIO)); } VERIFY(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME, &pname) == 0); VERIFY(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_TXG, &txg) == 0); mutex_enter(&spa_namespace_lock); if ((spa = spa_lookup(pname)) != NULL) { /* * Remove the existing root pool from the namespace so that we * can replace it with the correct config we just read in. */ spa_remove(spa); } spa = spa_add(pname, config, NULL); spa->spa_is_root = B_TRUE; spa->spa_import_flags = ZFS_IMPORT_VERBATIM; /* * Build up a vdev tree based on the boot device's label config. */ VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvtop) == 0); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); error = spa_config_parse(spa, &rvd, nvtop, NULL, 0, VDEV_ALLOC_ROOTPOOL); spa_config_exit(spa, SCL_ALL, FTAG); if (error) { mutex_exit(&spa_namespace_lock); nvlist_free(config); cmn_err(CE_NOTE, "Can not parse the config for pool '%s'", pname); return (error); } /* * Get the boot vdev. */ if ((bvd = vdev_lookup_by_guid(rvd, guid)) == NULL) { cmn_err(CE_NOTE, "Can not find the boot vdev for guid %llu", (u_longlong_t)guid); error = SET_ERROR(ENOENT); goto out; } /* * Determine if there is a better boot device. */ avd = bvd; spa_alt_rootvdev(rvd, &avd, &txg); if (avd != bvd) { cmn_err(CE_NOTE, "The boot device is 'degraded'. Please " "try booting from '%s'", avd->vdev_path); error = SET_ERROR(EINVAL); goto out; } /* * If the boot device is part of a spare vdev then ensure that * we're booting off the active spare. */ if (bvd->vdev_parent->vdev_ops == &vdev_spare_ops && !bvd->vdev_isspare) { cmn_err(CE_NOTE, "The boot device is currently spared. Please " "try booting from '%s'", bvd->vdev_parent-> vdev_child[bvd->vdev_parent->vdev_children - 1]->vdev_path); error = SET_ERROR(EINVAL); goto out; } error = 0; out: spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); vdev_free(rvd); spa_config_exit(spa, SCL_ALL, FTAG); mutex_exit(&spa_namespace_lock); nvlist_free(config); return (error); } #endif /* * Import a non-root pool into the system. */ int spa_import(char *pool, nvlist_t *config, nvlist_t *props, uint64_t flags) { spa_t *spa; char *altroot = NULL; spa_load_state_t state = SPA_LOAD_IMPORT; zpool_rewind_policy_t policy; uint64_t mode = spa_mode_global; uint64_t readonly = B_FALSE; int error; nvlist_t *nvroot; nvlist_t **spares, **l2cache; uint_t nspares, nl2cache; /* * If a pool with this name exists, return failure. */ mutex_enter(&spa_namespace_lock); if (spa_lookup(pool) != NULL) { mutex_exit(&spa_namespace_lock); return (SET_ERROR(EEXIST)); } /* * Create and initialize the spa structure. */ (void) nvlist_lookup_string(props, zpool_prop_to_name(ZPOOL_PROP_ALTROOT), &altroot); (void) nvlist_lookup_uint64(props, zpool_prop_to_name(ZPOOL_PROP_READONLY), &readonly); if (readonly) mode = FREAD; spa = spa_add(pool, config, altroot); spa->spa_import_flags = flags; /* * Verbatim import - Take a pool and insert it into the namespace * as if it had been loaded at boot. */ if (spa->spa_import_flags & ZFS_IMPORT_VERBATIM) { if (props != NULL) spa_configfile_set(spa, props, B_FALSE); spa_config_sync(spa, B_FALSE, B_TRUE); mutex_exit(&spa_namespace_lock); return (0); } spa_activate(spa, mode); /* * Don't start async tasks until we know everything is healthy. */ spa_async_suspend(spa); zpool_get_rewind_policy(config, &policy); if (policy.zrp_request & ZPOOL_DO_REWIND) state = SPA_LOAD_RECOVER; /* * Pass off the heavy lifting to spa_load(). Pass TRUE for mosconfig * because the user-supplied config is actually the one to trust when * doing an import. */ if (state != SPA_LOAD_RECOVER) spa->spa_last_ubsync_txg = spa->spa_load_txg = 0; error = spa_load_best(spa, state, B_TRUE, policy.zrp_txg, policy.zrp_request); /* * Propagate anything learned while loading the pool and pass it * back to caller (i.e. rewind info, missing devices, etc). */ VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_LOAD_INFO, spa->spa_load_info) == 0); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); /* * Toss any existing sparelist, as it doesn't have any validity * anymore, and conflicts with spa_has_spare(). */ if (spa->spa_spares.sav_config) { nvlist_free(spa->spa_spares.sav_config); spa->spa_spares.sav_config = NULL; spa_load_spares(spa); } if (spa->spa_l2cache.sav_config) { nvlist_free(spa->spa_l2cache.sav_config); spa->spa_l2cache.sav_config = NULL; spa_load_l2cache(spa); } VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0); if (error == 0) error = spa_validate_aux(spa, nvroot, -1ULL, VDEV_ALLOC_SPARE); if (error == 0) error = spa_validate_aux(spa, nvroot, -1ULL, VDEV_ALLOC_L2CACHE); spa_config_exit(spa, SCL_ALL, FTAG); if (props != NULL) spa_configfile_set(spa, props, B_FALSE); if (error != 0 || (props && spa_writeable(spa) && (error = spa_prop_set(spa, props)))) { spa_unload(spa); spa_deactivate(spa); spa_remove(spa); mutex_exit(&spa_namespace_lock); return (error); } spa_async_resume(spa); /* * Override any spares and level 2 cache devices as specified by * the user, as these may have correct device names/devids, etc. */ if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0) { if (spa->spa_spares.sav_config) VERIFY(nvlist_remove(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, DATA_TYPE_NVLIST_ARRAY) == 0); else VERIFY(nvlist_alloc(&spa->spa_spares.sav_config, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_nvlist_array(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, spares, nspares) == 0); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_spares(spa); spa_config_exit(spa, SCL_ALL, FTAG); spa->spa_spares.sav_sync = B_TRUE; } if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0) { if (spa->spa_l2cache.sav_config) VERIFY(nvlist_remove(spa->spa_l2cache.sav_config, ZPOOL_CONFIG_L2CACHE, DATA_TYPE_NVLIST_ARRAY) == 0); else VERIFY(nvlist_alloc(&spa->spa_l2cache.sav_config, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_nvlist_array(spa->spa_l2cache.sav_config, ZPOOL_CONFIG_L2CACHE, l2cache, nl2cache) == 0); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_l2cache(spa); spa_config_exit(spa, SCL_ALL, FTAG); spa->spa_l2cache.sav_sync = B_TRUE; } /* * Check for any removed devices. */ if (spa->spa_autoreplace) { spa_aux_check_removed(&spa->spa_spares); spa_aux_check_removed(&spa->spa_l2cache); } if (spa_writeable(spa)) { /* * Update the config cache to include the newly-imported pool. */ spa_config_update(spa, SPA_CONFIG_UPDATE_POOL); } /* * It's possible that the pool was expanded while it was exported. * We kick off an async task to handle this for us. */ spa_async_request(spa, SPA_ASYNC_AUTOEXPAND); mutex_exit(&spa_namespace_lock); spa_history_log_version(spa, "import"); #ifdef _KERNEL zvol_create_minors(pool); #endif return (0); } nvlist_t * spa_tryimport(nvlist_t *tryconfig) { nvlist_t *config = NULL; char *poolname; spa_t *spa; uint64_t state; int error; if (nvlist_lookup_string(tryconfig, ZPOOL_CONFIG_POOL_NAME, &poolname)) return (NULL); if (nvlist_lookup_uint64(tryconfig, ZPOOL_CONFIG_POOL_STATE, &state)) return (NULL); /* * Create and initialize the spa structure. */ mutex_enter(&spa_namespace_lock); spa = spa_add(TRYIMPORT_NAME, tryconfig, NULL); spa_activate(spa, FREAD); /* * Pass off the heavy lifting to spa_load(). * Pass TRUE for mosconfig because the user-supplied config * is actually the one to trust when doing an import. */ error = spa_load(spa, SPA_LOAD_TRYIMPORT, SPA_IMPORT_EXISTING, B_TRUE); /* * If 'tryconfig' was at least parsable, return the current config. */ if (spa->spa_root_vdev != NULL) { config = spa_config_generate(spa, NULL, -1ULL, B_TRUE); VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME, poolname) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE, state) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_TIMESTAMP, spa->spa_uberblock.ub_timestamp) == 0); VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_LOAD_INFO, spa->spa_load_info) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_ERRATA, spa->spa_errata) == 0); /* * If the bootfs property exists on this pool then we * copy it out so that external consumers can tell which * pools are bootable. */ if ((!error || error == EEXIST) && spa->spa_bootfs) { char *tmpname = kmem_alloc(MAXPATHLEN, KM_PUSHPAGE); /* * We have to play games with the name since the * pool was opened as TRYIMPORT_NAME. */ if (dsl_dsobj_to_dsname(spa_name(spa), spa->spa_bootfs, tmpname) == 0) { char *cp; char *dsname; dsname = kmem_alloc(MAXPATHLEN, KM_PUSHPAGE); cp = strchr(tmpname, '/'); if (cp == NULL) { (void) strlcpy(dsname, tmpname, MAXPATHLEN); } else { (void) snprintf(dsname, MAXPATHLEN, "%s/%s", poolname, ++cp); } VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_BOOTFS, dsname) == 0); kmem_free(dsname, MAXPATHLEN); } kmem_free(tmpname, MAXPATHLEN); } /* * Add the list of hot spares and level 2 cache devices. */ spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); spa_add_spares(spa, config); spa_add_l2cache(spa, config); spa_config_exit(spa, SCL_CONFIG, FTAG); } spa_unload(spa); spa_deactivate(spa); spa_remove(spa); mutex_exit(&spa_namespace_lock); return (config); } /* * Pool export/destroy * * The act of destroying or exporting a pool is very simple. We make sure there * is no more pending I/O and any references to the pool are gone. Then, we * update the pool state and sync all the labels to disk, removing the * configuration from the cache afterwards. If the 'hardforce' flag is set, then * we don't sync the labels or remove the configuration cache. */ static int spa_export_common(char *pool, int new_state, nvlist_t **oldconfig, boolean_t force, boolean_t hardforce) { spa_t *spa; if (oldconfig) *oldconfig = NULL; if (!(spa_mode_global & FWRITE)) return (SET_ERROR(EROFS)); mutex_enter(&spa_namespace_lock); if ((spa = spa_lookup(pool)) == NULL) { mutex_exit(&spa_namespace_lock); return (SET_ERROR(ENOENT)); } /* * Put a hold on the pool, drop the namespace lock, stop async tasks, * reacquire the namespace lock, and see if we can export. */ spa_open_ref(spa, FTAG); mutex_exit(&spa_namespace_lock); spa_async_suspend(spa); mutex_enter(&spa_namespace_lock); spa_close(spa, FTAG); /* * The pool will be in core if it's openable, * in which case we can modify its state. */ if (spa->spa_state != POOL_STATE_UNINITIALIZED && spa->spa_sync_on) { /* * Objsets may be open only because they're dirty, so we * have to force it to sync before checking spa_refcnt. */ txg_wait_synced(spa->spa_dsl_pool, 0); /* * A pool cannot be exported or destroyed if there are active * references. If we are resetting a pool, allow references by * fault injection handlers. */ if (!spa_refcount_zero(spa) || (spa->spa_inject_ref != 0 && new_state != POOL_STATE_UNINITIALIZED)) { spa_async_resume(spa); mutex_exit(&spa_namespace_lock); return (SET_ERROR(EBUSY)); } /* * A pool cannot be exported if it has an active shared spare. * This is to prevent other pools stealing the active spare * from an exported pool. At user's own will, such pool can * be forcedly exported. */ if (!force && new_state == POOL_STATE_EXPORTED && spa_has_active_shared_spare(spa)) { spa_async_resume(spa); mutex_exit(&spa_namespace_lock); return (SET_ERROR(EXDEV)); } /* * We want this to be reflected on every label, * so mark them all dirty. spa_unload() will do the * final sync that pushes these changes out. */ if (new_state != POOL_STATE_UNINITIALIZED && !hardforce) { spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa->spa_state = new_state; spa->spa_final_txg = spa_last_synced_txg(spa) + TXG_DEFER_SIZE + 1; vdev_config_dirty(spa->spa_root_vdev); spa_config_exit(spa, SCL_ALL, FTAG); } } spa_event_notify(spa, NULL, FM_EREPORT_ZFS_POOL_DESTROY); if (spa->spa_state != POOL_STATE_UNINITIALIZED) { spa_unload(spa); spa_deactivate(spa); } if (oldconfig && spa->spa_config) VERIFY(nvlist_dup(spa->spa_config, oldconfig, 0) == 0); if (new_state != POOL_STATE_UNINITIALIZED) { if (!hardforce) spa_config_sync(spa, B_TRUE, B_TRUE); spa_remove(spa); } mutex_exit(&spa_namespace_lock); return (0); } /* * Destroy a storage pool. */ int spa_destroy(char *pool) { return (spa_export_common(pool, POOL_STATE_DESTROYED, NULL, B_FALSE, B_FALSE)); } /* * Export a storage pool. */ int spa_export(char *pool, nvlist_t **oldconfig, boolean_t force, boolean_t hardforce) { return (spa_export_common(pool, POOL_STATE_EXPORTED, oldconfig, force, hardforce)); } /* * Similar to spa_export(), this unloads the spa_t without actually removing it * from the namespace in any way. */ int spa_reset(char *pool) { return (spa_export_common(pool, POOL_STATE_UNINITIALIZED, NULL, B_FALSE, B_FALSE)); } /* * ========================================================================== * Device manipulation * ========================================================================== */ /* * Add a device to a storage pool. */ int spa_vdev_add(spa_t *spa, nvlist_t *nvroot) { uint64_t txg, id; int error; vdev_t *rvd = spa->spa_root_vdev; vdev_t *vd, *tvd; nvlist_t **spares, **l2cache; uint_t nspares, nl2cache; int c; ASSERT(spa_writeable(spa)); txg = spa_vdev_enter(spa); if ((error = spa_config_parse(spa, &vd, nvroot, NULL, 0, VDEV_ALLOC_ADD)) != 0) return (spa_vdev_exit(spa, NULL, txg, error)); spa->spa_pending_vdev = vd; /* spa_vdev_exit() will clear this */ if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares, &nspares) != 0) nspares = 0; if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) != 0) nl2cache = 0; if (vd->vdev_children == 0 && nspares == 0 && nl2cache == 0) return (spa_vdev_exit(spa, vd, txg, EINVAL)); if (vd->vdev_children != 0 && (error = vdev_create(vd, txg, B_FALSE)) != 0) return (spa_vdev_exit(spa, vd, txg, error)); /* * We must validate the spares and l2cache devices after checking the * children. Otherwise, vdev_inuse() will blindly overwrite the spare. */ if ((error = spa_validate_aux(spa, nvroot, txg, VDEV_ALLOC_ADD)) != 0) return (spa_vdev_exit(spa, vd, txg, error)); /* * Transfer each new top-level vdev from vd to rvd. */ for (c = 0; c < vd->vdev_children; c++) { /* * Set the vdev id to the first hole, if one exists. */ for (id = 0; id < rvd->vdev_children; id++) { if (rvd->vdev_child[id]->vdev_ishole) { vdev_free(rvd->vdev_child[id]); break; } } tvd = vd->vdev_child[c]; vdev_remove_child(vd, tvd); tvd->vdev_id = id; vdev_add_child(rvd, tvd); vdev_config_dirty(tvd); } if (nspares != 0) { spa_set_aux_vdevs(&spa->spa_spares, spares, nspares, ZPOOL_CONFIG_SPARES); spa_load_spares(spa); spa->spa_spares.sav_sync = B_TRUE; } if (nl2cache != 0) { spa_set_aux_vdevs(&spa->spa_l2cache, l2cache, nl2cache, ZPOOL_CONFIG_L2CACHE); spa_load_l2cache(spa); spa->spa_l2cache.sav_sync = B_TRUE; } /* * We have to be careful when adding new vdevs to an existing pool. * If other threads start allocating from these vdevs before we * sync the config cache, and we lose power, then upon reboot we may * fail to open the pool because there are DVAs that the config cache * can't translate. Therefore, we first add the vdevs without * initializing metaslabs; sync the config cache (via spa_vdev_exit()); * and then let spa_config_update() initialize the new metaslabs. * * spa_load() checks for added-but-not-initialized vdevs, so that * if we lose power at any point in this sequence, the remaining * steps will be completed the next time we load the pool. */ (void) spa_vdev_exit(spa, vd, txg, 0); mutex_enter(&spa_namespace_lock); spa_config_update(spa, SPA_CONFIG_UPDATE_POOL); mutex_exit(&spa_namespace_lock); return (0); } /* * Attach a device to a mirror. The arguments are the path to any device * in the mirror, and the nvroot for the new device. If the path specifies * a device that is not mirrored, we automatically insert the mirror vdev. * * If 'replacing' is specified, the new device is intended to replace the * existing device; in this case the two devices are made into their own * mirror using the 'replacing' vdev, which is functionally identical to * the mirror vdev (it actually reuses all the same ops) but has a few * extra rules: you can't attach to it after it's been created, and upon * completion of resilvering, the first disk (the one being replaced) * is automatically detached. */ int spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, int replacing) { uint64_t txg, dtl_max_txg; vdev_t *oldvd, *newvd, *newrootvd, *pvd, *tvd; vdev_ops_t *pvops; char *oldvdpath, *newvdpath; int newvd_isspare; int error; ASSERTV(vdev_t *rvd = spa->spa_root_vdev); ASSERT(spa_writeable(spa)); txg = spa_vdev_enter(spa); oldvd = spa_lookup_by_guid(spa, guid, B_FALSE); if (oldvd == NULL) return (spa_vdev_exit(spa, NULL, txg, ENODEV)); if (!oldvd->vdev_ops->vdev_op_leaf) return (spa_vdev_exit(spa, NULL, txg, ENOTSUP)); pvd = oldvd->vdev_parent; if ((error = spa_config_parse(spa, &newrootvd, nvroot, NULL, 0, VDEV_ALLOC_ATTACH)) != 0) return (spa_vdev_exit(spa, NULL, txg, EINVAL)); if (newrootvd->vdev_children != 1) return (spa_vdev_exit(spa, newrootvd, txg, EINVAL)); newvd = newrootvd->vdev_child[0]; if (!newvd->vdev_ops->vdev_op_leaf) return (spa_vdev_exit(spa, newrootvd, txg, EINVAL)); if ((error = vdev_create(newrootvd, txg, replacing)) != 0) return (spa_vdev_exit(spa, newrootvd, txg, error)); /* * Spares can't replace logs */ if (oldvd->vdev_top->vdev_islog && newvd->vdev_isspare) return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); if (!replacing) { /* * For attach, the only allowable parent is a mirror or the root * vdev. */ if (pvd->vdev_ops != &vdev_mirror_ops && pvd->vdev_ops != &vdev_root_ops) return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); pvops = &vdev_mirror_ops; } else { /* * Active hot spares can only be replaced by inactive hot * spares. */ if (pvd->vdev_ops == &vdev_spare_ops && oldvd->vdev_isspare && !spa_has_spare(spa, newvd->vdev_guid)) return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); /* * If the source is a hot spare, and the parent isn't already a * spare, then we want to create a new hot spare. Otherwise, we * want to create a replacing vdev. The user is not allowed to * attach to a spared vdev child unless the 'isspare' state is * the same (spare replaces spare, non-spare replaces * non-spare). */ if (pvd->vdev_ops == &vdev_replacing_ops && spa_version(spa) < SPA_VERSION_MULTI_REPLACE) { return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); } else if (pvd->vdev_ops == &vdev_spare_ops && newvd->vdev_isspare != oldvd->vdev_isspare) { return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); } if (newvd->vdev_isspare) pvops = &vdev_spare_ops; else pvops = &vdev_replacing_ops; } /* * Make sure the new device is big enough. */ if (newvd->vdev_asize < vdev_get_min_asize(oldvd)) return (spa_vdev_exit(spa, newrootvd, txg, EOVERFLOW)); /* * The new device cannot have a higher alignment requirement * than the top-level vdev. */ if (newvd->vdev_ashift > oldvd->vdev_top->vdev_ashift) return (spa_vdev_exit(spa, newrootvd, txg, EDOM)); /* * If this is an in-place replacement, update oldvd's path and devid * to make it distinguishable from newvd, and unopenable from now on. */ if (strcmp(oldvd->vdev_path, newvd->vdev_path) == 0) { spa_strfree(oldvd->vdev_path); oldvd->vdev_path = kmem_alloc(strlen(newvd->vdev_path) + 5, KM_PUSHPAGE); (void) sprintf(oldvd->vdev_path, "%s/%s", newvd->vdev_path, "old"); if (oldvd->vdev_devid != NULL) { spa_strfree(oldvd->vdev_devid); oldvd->vdev_devid = NULL; } } /* mark the device being resilvered */ newvd->vdev_resilver_txg = txg; /* * If the parent is not a mirror, or if we're replacing, insert the new * mirror/replacing/spare vdev above oldvd. */ if (pvd->vdev_ops != pvops) pvd = vdev_add_parent(oldvd, pvops); ASSERT(pvd->vdev_top->vdev_parent == rvd); ASSERT(pvd->vdev_ops == pvops); ASSERT(oldvd->vdev_parent == pvd); /* * Extract the new device from its root and add it to pvd. */ vdev_remove_child(newrootvd, newvd); newvd->vdev_id = pvd->vdev_children; newvd->vdev_crtxg = oldvd->vdev_crtxg; vdev_add_child(pvd, newvd); tvd = newvd->vdev_top; ASSERT(pvd->vdev_top == tvd); ASSERT(tvd->vdev_parent == rvd); vdev_config_dirty(tvd); /* * Set newvd's DTL to [TXG_INITIAL, dtl_max_txg) so that we account * for any dmu_sync-ed blocks. It will propagate upward when * spa_vdev_exit() calls vdev_dtl_reassess(). */ dtl_max_txg = txg + TXG_CONCURRENT_STATES; vdev_dtl_dirty(newvd, DTL_MISSING, TXG_INITIAL, dtl_max_txg - TXG_INITIAL); if (newvd->vdev_isspare) { spa_spare_activate(newvd); spa_event_notify(spa, newvd, FM_EREPORT_ZFS_DEVICE_SPARE); } oldvdpath = spa_strdup(oldvd->vdev_path); newvdpath = spa_strdup(newvd->vdev_path); newvd_isspare = newvd->vdev_isspare; /* * Mark newvd's DTL dirty in this txg. */ vdev_dirty(tvd, VDD_DTL, newvd, txg); /* * Schedule the resilver to restart in the future. We do this to * ensure that dmu_sync-ed blocks have been stitched into the * respective datasets. */ dsl_resilver_restart(spa->spa_dsl_pool, dtl_max_txg); /* * Commit the config */ (void) spa_vdev_exit(spa, newrootvd, dtl_max_txg, 0); spa_history_log_internal(spa, "vdev attach", NULL, "%s vdev=%s %s vdev=%s", replacing && newvd_isspare ? "spare in" : replacing ? "replace" : "attach", newvdpath, replacing ? "for" : "to", oldvdpath); spa_strfree(oldvdpath); spa_strfree(newvdpath); if (spa->spa_bootfs) spa_event_notify(spa, newvd, FM_EREPORT_ZFS_BOOTFS_VDEV_ATTACH); return (0); } /* * Detach a device from a mirror or replacing vdev. * * If 'replace_done' is specified, only detach if the parent * is a replacing vdev. */ int spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done) { uint64_t txg; int error; vdev_t *vd, *pvd, *cvd, *tvd; boolean_t unspare = B_FALSE; uint64_t unspare_guid = 0; char *vdpath; int c, t; ASSERTV(vdev_t *rvd = spa->spa_root_vdev); ASSERT(spa_writeable(spa)); txg = spa_vdev_enter(spa); vd = spa_lookup_by_guid(spa, guid, B_FALSE); if (vd == NULL) return (spa_vdev_exit(spa, NULL, txg, ENODEV)); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_exit(spa, NULL, txg, ENOTSUP)); pvd = vd->vdev_parent; /* * If the parent/child relationship is not as expected, don't do it. * Consider M(A,R(B,C)) -- that is, a mirror of A with a replacing * vdev that's replacing B with C. The user's intent in replacing * is to go from M(A,B) to M(A,C). If the user decides to cancel * the replace by detaching C, the expected behavior is to end up * M(A,B). But suppose that right after deciding to detach C, * the replacement of B completes. We would have M(A,C), and then * ask to detach C, which would leave us with just A -- not what * the user wanted. To prevent this, we make sure that the * parent/child relationship hasn't changed -- in this example, * that C's parent is still the replacing vdev R. */ if (pvd->vdev_guid != pguid && pguid != 0) return (spa_vdev_exit(spa, NULL, txg, EBUSY)); /* * Only 'replacing' or 'spare' vdevs can be replaced. */ if (replace_done && pvd->vdev_ops != &vdev_replacing_ops && pvd->vdev_ops != &vdev_spare_ops) return (spa_vdev_exit(spa, NULL, txg, ENOTSUP)); ASSERT(pvd->vdev_ops != &vdev_spare_ops || spa_version(spa) >= SPA_VERSION_SPARES); /* * Only mirror, replacing, and spare vdevs support detach. */ if (pvd->vdev_ops != &vdev_replacing_ops && pvd->vdev_ops != &vdev_mirror_ops && pvd->vdev_ops != &vdev_spare_ops) return (spa_vdev_exit(spa, NULL, txg, ENOTSUP)); /* * If this device has the only valid copy of some data, * we cannot safely detach it. */ if (vdev_dtl_required(vd)) return (spa_vdev_exit(spa, NULL, txg, EBUSY)); ASSERT(pvd->vdev_children >= 2); /* * If we are detaching the second disk from a replacing vdev, then * check to see if we changed the original vdev's path to have "/old" * at the end in spa_vdev_attach(). If so, undo that change now. */ if (pvd->vdev_ops == &vdev_replacing_ops && vd->vdev_id > 0 && vd->vdev_path != NULL) { size_t len = strlen(vd->vdev_path); for (c = 0; c < pvd->vdev_children; c++) { cvd = pvd->vdev_child[c]; if (cvd == vd || cvd->vdev_path == NULL) continue; if (strncmp(cvd->vdev_path, vd->vdev_path, len) == 0 && strcmp(cvd->vdev_path + len, "/old") == 0) { spa_strfree(cvd->vdev_path); cvd->vdev_path = spa_strdup(vd->vdev_path); break; } } } /* * If we are detaching the original disk from a spare, then it implies * that the spare should become a real disk, and be removed from the * active spare list for the pool. */ if (pvd->vdev_ops == &vdev_spare_ops && vd->vdev_id == 0 && pvd->vdev_child[pvd->vdev_children - 1]->vdev_isspare) unspare = B_TRUE; /* * Erase the disk labels so the disk can be used for other things. * This must be done after all other error cases are handled, * but before we disembowel vd (so we can still do I/O to it). * But if we can't do it, don't treat the error as fatal -- * it may be that the unwritability of the disk is the reason * it's being detached! */ error = vdev_label_init(vd, 0, VDEV_LABEL_REMOVE); /* * Remove vd from its parent and compact the parent's children. */ vdev_remove_child(pvd, vd); vdev_compact_children(pvd); /* * Remember one of the remaining children so we can get tvd below. */ cvd = pvd->vdev_child[pvd->vdev_children - 1]; /* * If we need to remove the remaining child from the list of hot spares, * do it now, marking the vdev as no longer a spare in the process. * We must do this before vdev_remove_parent(), because that can * change the GUID if it creates a new toplevel GUID. For a similar * reason, we must remove the spare now, in the same txg as the detach; * otherwise someone could attach a new sibling, change the GUID, and * the subsequent attempt to spa_vdev_remove(unspare_guid) would fail. */ if (unspare) { ASSERT(cvd->vdev_isspare); spa_spare_remove(cvd); unspare_guid = cvd->vdev_guid; (void) spa_vdev_remove(spa, unspare_guid, B_TRUE); cvd->vdev_unspare = B_TRUE; } /* * If the parent mirror/replacing vdev only has one child, * the parent is no longer needed. Remove it from the tree. */ if (pvd->vdev_children == 1) { if (pvd->vdev_ops == &vdev_spare_ops) cvd->vdev_unspare = B_FALSE; vdev_remove_parent(cvd); } /* * We don't set tvd until now because the parent we just removed * may have been the previous top-level vdev. */ tvd = cvd->vdev_top; ASSERT(tvd->vdev_parent == rvd); /* * Reevaluate the parent vdev state. */ vdev_propagate_state(cvd); /* * If the 'autoexpand' property is set on the pool then automatically * try to expand the size of the pool. For example if the device we * just detached was smaller than the others, it may be possible to * add metaslabs (i.e. grow the pool). We need to reopen the vdev * first so that we can obtain the updated sizes of the leaf vdevs. */ if (spa->spa_autoexpand) { vdev_reopen(tvd); vdev_expand(tvd, txg); } vdev_config_dirty(tvd); /* * Mark vd's DTL as dirty in this txg. vdev_dtl_sync() will see that * vd->vdev_detached is set and free vd's DTL object in syncing context. * But first make sure we're not on any *other* txg's DTL list, to * prevent vd from being accessed after it's freed. */ vdpath = spa_strdup(vd->vdev_path); for (t = 0; t < TXG_SIZE; t++) (void) txg_list_remove_this(&tvd->vdev_dtl_list, vd, t); vd->vdev_detached = B_TRUE; vdev_dirty(tvd, VDD_DTL, vd, txg); spa_event_notify(spa, vd, FM_EREPORT_ZFS_DEVICE_REMOVE); /* hang on to the spa before we release the lock */ spa_open_ref(spa, FTAG); error = spa_vdev_exit(spa, vd, txg, 0); spa_history_log_internal(spa, "detach", NULL, "vdev=%s", vdpath); spa_strfree(vdpath); /* * If this was the removal of the original device in a hot spare vdev, * then we want to go through and remove the device from the hot spare * list of every other pool. */ if (unspare) { spa_t *altspa = NULL; mutex_enter(&spa_namespace_lock); while ((altspa = spa_next(altspa)) != NULL) { if (altspa->spa_state != POOL_STATE_ACTIVE || altspa == spa) continue; spa_open_ref(altspa, FTAG); mutex_exit(&spa_namespace_lock); (void) spa_vdev_remove(altspa, unspare_guid, B_TRUE); mutex_enter(&spa_namespace_lock); spa_close(altspa, FTAG); } mutex_exit(&spa_namespace_lock); /* search the rest of the vdevs for spares to remove */ spa_vdev_resilver_done(spa); } /* all done with the spa; OK to release */ mutex_enter(&spa_namespace_lock); spa_close(spa, FTAG); mutex_exit(&spa_namespace_lock); return (error); } /* * Split a set of devices from their mirrors, and create a new pool from them. */ int spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config, nvlist_t *props, boolean_t exp) { int error = 0; uint64_t txg, *glist; spa_t *newspa; uint_t c, children, lastlog; nvlist_t **child, *nvl, *tmp; dmu_tx_t *tx; char *altroot = NULL; vdev_t *rvd, **vml = NULL; /* vdev modify list */ boolean_t activate_slog; ASSERT(spa_writeable(spa)); txg = spa_vdev_enter(spa); /* clear the log and flush everything up to now */ activate_slog = spa_passivate_log(spa); (void) spa_vdev_config_exit(spa, NULL, txg, 0, FTAG); error = spa_offline_log(spa); txg = spa_vdev_config_enter(spa); if (activate_slog) spa_activate_log(spa); if (error != 0) return (spa_vdev_exit(spa, NULL, txg, error)); /* check new spa name before going any further */ if (spa_lookup(newname) != NULL) return (spa_vdev_exit(spa, NULL, txg, EEXIST)); /* * scan through all the children to ensure they're all mirrors */ if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvl) != 0 || nvlist_lookup_nvlist_array(nvl, ZPOOL_CONFIG_CHILDREN, &child, &children) != 0) return (spa_vdev_exit(spa, NULL, txg, EINVAL)); /* first, check to ensure we've got the right child count */ rvd = spa->spa_root_vdev; lastlog = 0; for (c = 0; c < rvd->vdev_children; c++) { vdev_t *vd = rvd->vdev_child[c]; /* don't count the holes & logs as children */ if (vd->vdev_islog || vd->vdev_ishole) { if (lastlog == 0) lastlog = c; continue; } lastlog = 0; } if (children != (lastlog != 0 ? lastlog : rvd->vdev_children)) return (spa_vdev_exit(spa, NULL, txg, EINVAL)); /* next, ensure no spare or cache devices are part of the split */ if (nvlist_lookup_nvlist(nvl, ZPOOL_CONFIG_SPARES, &tmp) == 0 || nvlist_lookup_nvlist(nvl, ZPOOL_CONFIG_L2CACHE, &tmp) == 0) return (spa_vdev_exit(spa, NULL, txg, EINVAL)); vml = kmem_zalloc(children * sizeof (vdev_t *), KM_PUSHPAGE); glist = kmem_zalloc(children * sizeof (uint64_t), KM_PUSHPAGE); /* then, loop over each vdev and validate it */ for (c = 0; c < children; c++) { uint64_t is_hole = 0; (void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_HOLE, &is_hole); if (is_hole != 0) { if (spa->spa_root_vdev->vdev_child[c]->vdev_ishole || spa->spa_root_vdev->vdev_child[c]->vdev_islog) { continue; } else { error = SET_ERROR(EINVAL); break; } } /* which disk is going to be split? */ if (nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_GUID, &glist[c]) != 0) { error = SET_ERROR(EINVAL); break; } /* look it up in the spa */ vml[c] = spa_lookup_by_guid(spa, glist[c], B_FALSE); if (vml[c] == NULL) { error = SET_ERROR(ENODEV); break; } /* make sure there's nothing stopping the split */ if (vml[c]->vdev_parent->vdev_ops != &vdev_mirror_ops || vml[c]->vdev_islog || vml[c]->vdev_ishole || vml[c]->vdev_isspare || vml[c]->vdev_isl2cache || !vdev_writeable(vml[c]) || vml[c]->vdev_children != 0 || vml[c]->vdev_state != VDEV_STATE_HEALTHY || c != spa->spa_root_vdev->vdev_child[c]->vdev_id) { error = SET_ERROR(EINVAL); break; } if (vdev_dtl_required(vml[c])) { error = SET_ERROR(EBUSY); break; } /* we need certain info from the top level */ VERIFY(nvlist_add_uint64(child[c], ZPOOL_CONFIG_METASLAB_ARRAY, vml[c]->vdev_top->vdev_ms_array) == 0); VERIFY(nvlist_add_uint64(child[c], ZPOOL_CONFIG_METASLAB_SHIFT, vml[c]->vdev_top->vdev_ms_shift) == 0); VERIFY(nvlist_add_uint64(child[c], ZPOOL_CONFIG_ASIZE, vml[c]->vdev_top->vdev_asize) == 0); VERIFY(nvlist_add_uint64(child[c], ZPOOL_CONFIG_ASHIFT, vml[c]->vdev_top->vdev_ashift) == 0); } if (error != 0) { kmem_free(vml, children * sizeof (vdev_t *)); kmem_free(glist, children * sizeof (uint64_t)); return (spa_vdev_exit(spa, NULL, txg, error)); } /* stop writers from using the disks */ for (c = 0; c < children; c++) { if (vml[c] != NULL) vml[c]->vdev_offline = B_TRUE; } vdev_reopen(spa->spa_root_vdev); /* * Temporarily record the splitting vdevs in the spa config. This * will disappear once the config is regenerated. */ VERIFY(nvlist_alloc(&nvl, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_uint64_array(nvl, ZPOOL_CONFIG_SPLIT_LIST, glist, children) == 0); kmem_free(glist, children * sizeof (uint64_t)); mutex_enter(&spa->spa_props_lock); VERIFY(nvlist_add_nvlist(spa->spa_config, ZPOOL_CONFIG_SPLIT, nvl) == 0); mutex_exit(&spa->spa_props_lock); spa->spa_config_splitting = nvl; vdev_config_dirty(spa->spa_root_vdev); /* configure and create the new pool */ VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME, newname) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE, exp ? POOL_STATE_EXPORTED : POOL_STATE_ACTIVE) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_VERSION, spa_version(spa)) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_TXG, spa->spa_config_txg) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_GUID, spa_generate_guid(NULL)) == 0); (void) nvlist_lookup_string(props, zpool_prop_to_name(ZPOOL_PROP_ALTROOT), &altroot); /* add the new pool to the namespace */ newspa = spa_add(newname, config, altroot); newspa->spa_config_txg = spa->spa_config_txg; spa_set_log_state(newspa, SPA_LOG_CLEAR); /* release the spa config lock, retaining the namespace lock */ spa_vdev_config_exit(spa, NULL, txg, 0, FTAG); if (zio_injection_enabled) zio_handle_panic_injection(spa, FTAG, 1); spa_activate(newspa, spa_mode_global); spa_async_suspend(newspa); /* create the new pool from the disks of the original pool */ error = spa_load(newspa, SPA_LOAD_IMPORT, SPA_IMPORT_ASSEMBLE, B_TRUE); if (error) goto out; /* if that worked, generate a real config for the new pool */ if (newspa->spa_root_vdev != NULL) { VERIFY(nvlist_alloc(&newspa->spa_config_splitting, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_uint64(newspa->spa_config_splitting, ZPOOL_CONFIG_SPLIT_GUID, spa_guid(spa)) == 0); spa_config_set(newspa, spa_config_generate(newspa, NULL, -1ULL, B_TRUE)); } /* set the props */ if (props != NULL) { spa_configfile_set(newspa, props, B_FALSE); error = spa_prop_set(newspa, props); if (error) goto out; } /* flush everything */ txg = spa_vdev_config_enter(newspa); vdev_config_dirty(newspa->spa_root_vdev); (void) spa_vdev_config_exit(newspa, NULL, txg, 0, FTAG); if (zio_injection_enabled) zio_handle_panic_injection(spa, FTAG, 2); spa_async_resume(newspa); /* finally, update the original pool's config */ txg = spa_vdev_config_enter(spa); tx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir); error = dmu_tx_assign(tx, TXG_WAIT); if (error != 0) dmu_tx_abort(tx); for (c = 0; c < children; c++) { if (vml[c] != NULL) { vdev_split(vml[c]); if (error == 0) spa_history_log_internal(spa, "detach", tx, "vdev=%s", vml[c]->vdev_path); vdev_free(vml[c]); } } vdev_config_dirty(spa->spa_root_vdev); spa->spa_config_splitting = NULL; nvlist_free(nvl); if (error == 0) dmu_tx_commit(tx); (void) spa_vdev_exit(spa, NULL, txg, 0); if (zio_injection_enabled) zio_handle_panic_injection(spa, FTAG, 3); /* split is complete; log a history record */ spa_history_log_internal(newspa, "split", NULL, "from pool %s", spa_name(spa)); kmem_free(vml, children * sizeof (vdev_t *)); /* if we're not going to mount the filesystems in userland, export */ if (exp) error = spa_export_common(newname, POOL_STATE_EXPORTED, NULL, B_FALSE, B_FALSE); return (error); out: spa_unload(newspa); spa_deactivate(newspa); spa_remove(newspa); txg = spa_vdev_config_enter(spa); /* re-online all offlined disks */ for (c = 0; c < children; c++) { if (vml[c] != NULL) vml[c]->vdev_offline = B_FALSE; } vdev_reopen(spa->spa_root_vdev); nvlist_free(spa->spa_config_splitting); spa->spa_config_splitting = NULL; (void) spa_vdev_exit(spa, NULL, txg, error); kmem_free(vml, children * sizeof (vdev_t *)); return (error); } static nvlist_t * spa_nvlist_lookup_by_guid(nvlist_t **nvpp, int count, uint64_t target_guid) { int i; for (i = 0; i < count; i++) { uint64_t guid; VERIFY(nvlist_lookup_uint64(nvpp[i], ZPOOL_CONFIG_GUID, &guid) == 0); if (guid == target_guid) return (nvpp[i]); } return (NULL); } static void spa_vdev_remove_aux(nvlist_t *config, char *name, nvlist_t **dev, int count, nvlist_t *dev_to_remove) { nvlist_t **newdev = NULL; int i, j; if (count > 1) newdev = kmem_alloc((count - 1) * sizeof (void *), KM_PUSHPAGE); for (i = 0, j = 0; i < count; i++) { if (dev[i] == dev_to_remove) continue; VERIFY(nvlist_dup(dev[i], &newdev[j++], KM_PUSHPAGE) == 0); } VERIFY(nvlist_remove(config, name, DATA_TYPE_NVLIST_ARRAY) == 0); VERIFY(nvlist_add_nvlist_array(config, name, newdev, count - 1) == 0); for (i = 0; i < count - 1; i++) nvlist_free(newdev[i]); if (count > 1) kmem_free(newdev, (count - 1) * sizeof (void *)); } /* * Evacuate the device. */ static int spa_vdev_remove_evacuate(spa_t *spa, vdev_t *vd) { uint64_t txg; int error = 0; ASSERT(MUTEX_HELD(&spa_namespace_lock)); ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == 0); ASSERT(vd == vd->vdev_top); /* * Evacuate the device. We don't hold the config lock as writer * since we need to do I/O but we do keep the * spa_namespace_lock held. Once this completes the device * should no longer have any blocks allocated on it. */ if (vd->vdev_islog) { if (vd->vdev_stat.vs_alloc != 0) error = spa_offline_log(spa); } else { error = SET_ERROR(ENOTSUP); } if (error) return (error); /* * The evacuation succeeded. Remove any remaining MOS metadata * associated with this vdev, and wait for these changes to sync. */ ASSERT0(vd->vdev_stat.vs_alloc); txg = spa_vdev_config_enter(spa); vd->vdev_removing = B_TRUE; vdev_dirty_leaves(vd, VDD_DTL, txg); vdev_config_dirty(vd); spa_vdev_config_exit(spa, NULL, txg, 0, FTAG); return (0); } /* * Complete the removal by cleaning up the namespace. */ static void spa_vdev_remove_from_namespace(spa_t *spa, vdev_t *vd) { vdev_t *rvd = spa->spa_root_vdev; uint64_t id = vd->vdev_id; boolean_t last_vdev = (id == (rvd->vdev_children - 1)); ASSERT(MUTEX_HELD(&spa_namespace_lock)); ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); ASSERT(vd == vd->vdev_top); /* * Only remove any devices which are empty. */ if (vd->vdev_stat.vs_alloc != 0) return; (void) vdev_label_init(vd, 0, VDEV_LABEL_REMOVE); if (list_link_active(&vd->vdev_state_dirty_node)) vdev_state_clean(vd); if (list_link_active(&vd->vdev_config_dirty_node)) vdev_config_clean(vd); vdev_free(vd); if (last_vdev) { vdev_compact_children(rvd); } else { vd = vdev_alloc_common(spa, id, 0, &vdev_hole_ops); vdev_add_child(rvd, vd); } vdev_config_dirty(rvd); /* * Reassess the health of our root vdev. */ vdev_reopen(rvd); } /* * Remove a device from the pool - * * Removing a device from the vdev namespace requires several steps * and can take a significant amount of time. As a result we use * the spa_vdev_config_[enter/exit] functions which allow us to * grab and release the spa_config_lock while still holding the namespace * lock. During each step the configuration is synced out. * * Currently, this supports removing only hot spares, slogs, and level 2 ARC * devices. */ int spa_vdev_remove(spa_t *spa, uint64_t guid, boolean_t unspare) { vdev_t *vd; metaslab_group_t *mg; nvlist_t **spares, **l2cache, *nv; uint64_t txg = 0; uint_t nspares, nl2cache; int error = 0; boolean_t locked = MUTEX_HELD(&spa_namespace_lock); ASSERT(spa_writeable(spa)); if (!locked) txg = spa_vdev_enter(spa); vd = spa_lookup_by_guid(spa, guid, B_FALSE); if (spa->spa_spares.sav_vdevs != NULL && nvlist_lookup_nvlist_array(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0 && (nv = spa_nvlist_lookup_by_guid(spares, nspares, guid)) != NULL) { /* * Only remove the hot spare if it's not currently in use * in this pool. */ if (vd == NULL || unspare) { spa_vdev_remove_aux(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, spares, nspares, nv); spa_load_spares(spa); spa->spa_spares.sav_sync = B_TRUE; } else { error = SET_ERROR(EBUSY); } } else if (spa->spa_l2cache.sav_vdevs != NULL && nvlist_lookup_nvlist_array(spa->spa_l2cache.sav_config, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0 && (nv = spa_nvlist_lookup_by_guid(l2cache, nl2cache, guid)) != NULL) { /* * Cache devices can always be removed. */ spa_vdev_remove_aux(spa->spa_l2cache.sav_config, ZPOOL_CONFIG_L2CACHE, l2cache, nl2cache, nv); spa_load_l2cache(spa); spa->spa_l2cache.sav_sync = B_TRUE; } else if (vd != NULL && vd->vdev_islog) { ASSERT(!locked); ASSERT(vd == vd->vdev_top); mg = vd->vdev_mg; /* * Stop allocating from this vdev. */ metaslab_group_passivate(mg); /* * Wait for the youngest allocations and frees to sync, * and then wait for the deferral of those frees to finish. */ spa_vdev_config_exit(spa, NULL, txg + TXG_CONCURRENT_STATES + TXG_DEFER_SIZE, 0, FTAG); /* * Attempt to evacuate the vdev. */ error = spa_vdev_remove_evacuate(spa, vd); txg = spa_vdev_config_enter(spa); /* * If we couldn't evacuate the vdev, unwind. */ if (error) { metaslab_group_activate(mg); return (spa_vdev_exit(spa, NULL, txg, error)); } /* * Clean up the vdev namespace. */ spa_vdev_remove_from_namespace(spa, vd); } else if (vd != NULL) { /* * Normal vdevs cannot be removed (yet). */ error = SET_ERROR(ENOTSUP); } else { /* * There is no vdev of any kind with the specified guid. */ error = SET_ERROR(ENOENT); } if (!locked) return (spa_vdev_exit(spa, NULL, txg, error)); return (error); } /* * Find any device that's done replacing, or a vdev marked 'unspare' that's * currently spared, so we can detach it. */ static vdev_t * spa_vdev_resilver_done_hunt(vdev_t *vd) { vdev_t *newvd, *oldvd; int c; for (c = 0; c < vd->vdev_children; c++) { oldvd = spa_vdev_resilver_done_hunt(vd->vdev_child[c]); if (oldvd != NULL) return (oldvd); } /* * Check for a completed replacement. We always consider the first * vdev in the list to be the oldest vdev, and the last one to be * the newest (see spa_vdev_attach() for how that works). In * the case where the newest vdev is faulted, we will not automatically * remove it after a resilver completes. This is OK as it will require * user intervention to determine which disk the admin wishes to keep. */ if (vd->vdev_ops == &vdev_replacing_ops) { ASSERT(vd->vdev_children > 1); newvd = vd->vdev_child[vd->vdev_children - 1]; oldvd = vd->vdev_child[0]; if (vdev_dtl_empty(newvd, DTL_MISSING) && vdev_dtl_empty(newvd, DTL_OUTAGE) && !vdev_dtl_required(oldvd)) return (oldvd); } /* * Check for a completed resilver with the 'unspare' flag set. */ if (vd->vdev_ops == &vdev_spare_ops) { vdev_t *first = vd->vdev_child[0]; vdev_t *last = vd->vdev_child[vd->vdev_children - 1]; if (last->vdev_unspare) { oldvd = first; newvd = last; } else if (first->vdev_unspare) { oldvd = last; newvd = first; } else { oldvd = NULL; } if (oldvd != NULL && vdev_dtl_empty(newvd, DTL_MISSING) && vdev_dtl_empty(newvd, DTL_OUTAGE) && !vdev_dtl_required(oldvd)) return (oldvd); /* * If there are more than two spares attached to a disk, * and those spares are not required, then we want to * attempt to free them up now so that they can be used * by other pools. Once we're back down to a single * disk+spare, we stop removing them. */ if (vd->vdev_children > 2) { newvd = vd->vdev_child[1]; if (newvd->vdev_isspare && last->vdev_isspare && vdev_dtl_empty(last, DTL_MISSING) && vdev_dtl_empty(last, DTL_OUTAGE) && !vdev_dtl_required(newvd)) return (newvd); } } return (NULL); } static void spa_vdev_resilver_done(spa_t *spa) { vdev_t *vd, *pvd, *ppvd; uint64_t guid, sguid, pguid, ppguid; spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); while ((vd = spa_vdev_resilver_done_hunt(spa->spa_root_vdev)) != NULL) { pvd = vd->vdev_parent; ppvd = pvd->vdev_parent; guid = vd->vdev_guid; pguid = pvd->vdev_guid; ppguid = ppvd->vdev_guid; sguid = 0; /* * If we have just finished replacing a hot spared device, then * we need to detach the parent's first child (the original hot * spare) as well. */ if (ppvd->vdev_ops == &vdev_spare_ops && pvd->vdev_id == 0 && ppvd->vdev_children == 2) { ASSERT(pvd->vdev_ops == &vdev_replacing_ops); sguid = ppvd->vdev_child[1]->vdev_guid; } ASSERT(vd->vdev_resilver_txg == 0 || !vdev_dtl_required(vd)); spa_config_exit(spa, SCL_ALL, FTAG); if (spa_vdev_detach(spa, guid, pguid, B_TRUE) != 0) return; if (sguid && spa_vdev_detach(spa, sguid, ppguid, B_TRUE) != 0) return; spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); } spa_config_exit(spa, SCL_ALL, FTAG); } /* * Update the stored path or FRU for this vdev. */ int spa_vdev_set_common(spa_t *spa, uint64_t guid, const char *value, boolean_t ispath) { vdev_t *vd; boolean_t sync = B_FALSE; ASSERT(spa_writeable(spa)); spa_vdev_state_enter(spa, SCL_ALL); if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL) return (spa_vdev_state_exit(spa, NULL, ENOENT)); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_state_exit(spa, NULL, ENOTSUP)); if (ispath) { if (strcmp(value, vd->vdev_path) != 0) { spa_strfree(vd->vdev_path); vd->vdev_path = spa_strdup(value); sync = B_TRUE; } } else { if (vd->vdev_fru == NULL) { vd->vdev_fru = spa_strdup(value); sync = B_TRUE; } else if (strcmp(value, vd->vdev_fru) != 0) { spa_strfree(vd->vdev_fru); vd->vdev_fru = spa_strdup(value); sync = B_TRUE; } } return (spa_vdev_state_exit(spa, sync ? vd : NULL, 0)); } int spa_vdev_setpath(spa_t *spa, uint64_t guid, const char *newpath) { return (spa_vdev_set_common(spa, guid, newpath, B_TRUE)); } int spa_vdev_setfru(spa_t *spa, uint64_t guid, const char *newfru) { return (spa_vdev_set_common(spa, guid, newfru, B_FALSE)); } /* * ========================================================================== * SPA Scanning * ========================================================================== */ int spa_scan_stop(spa_t *spa) { ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == 0); if (dsl_scan_resilvering(spa->spa_dsl_pool)) return (SET_ERROR(EBUSY)); return (dsl_scan_cancel(spa->spa_dsl_pool)); } int spa_scan(spa_t *spa, pool_scan_func_t func) { ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == 0); if (func >= POOL_SCAN_FUNCS || func == POOL_SCAN_NONE) return (SET_ERROR(ENOTSUP)); /* * If a resilver was requested, but there is no DTL on a * writeable leaf device, we have nothing to do. */ if (func == POOL_SCAN_RESILVER && !vdev_resilver_needed(spa->spa_root_vdev, NULL, NULL)) { spa_async_request(spa, SPA_ASYNC_RESILVER_DONE); return (0); } return (dsl_scan(spa->spa_dsl_pool, func)); } /* * ========================================================================== * SPA async task processing * ========================================================================== */ static void spa_async_remove(spa_t *spa, vdev_t *vd) { int c; if (vd->vdev_remove_wanted) { vd->vdev_remove_wanted = B_FALSE; vd->vdev_delayed_close = B_FALSE; vdev_set_state(vd, B_FALSE, VDEV_STATE_REMOVED, VDEV_AUX_NONE); /* * We want to clear the stats, but we don't want to do a full * vdev_clear() as that will cause us to throw away * degraded/faulted state as well as attempt to reopen the * device, all of which is a waste. */ vd->vdev_stat.vs_read_errors = 0; vd->vdev_stat.vs_write_errors = 0; vd->vdev_stat.vs_checksum_errors = 0; vdev_state_dirty(vd->vdev_top); } for (c = 0; c < vd->vdev_children; c++) spa_async_remove(spa, vd->vdev_child[c]); } static void spa_async_probe(spa_t *spa, vdev_t *vd) { int c; if (vd->vdev_probe_wanted) { vd->vdev_probe_wanted = B_FALSE; vdev_reopen(vd); /* vdev_open() does the actual probe */ } for (c = 0; c < vd->vdev_children; c++) spa_async_probe(spa, vd->vdev_child[c]); } static void spa_async_autoexpand(spa_t *spa, vdev_t *vd) { int c; if (!spa->spa_autoexpand) return; for (c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; spa_async_autoexpand(spa, cvd); } if (!vd->vdev_ops->vdev_op_leaf || vd->vdev_physpath == NULL) return; spa_event_notify(vd->vdev_spa, vd, FM_EREPORT_ZFS_DEVICE_AUTOEXPAND); } static void spa_async_thread(spa_t *spa) { int tasks, i; ASSERT(spa->spa_sync_on); mutex_enter(&spa->spa_async_lock); tasks = spa->spa_async_tasks; spa->spa_async_tasks = 0; mutex_exit(&spa->spa_async_lock); /* * See if the config needs to be updated. */ if (tasks & SPA_ASYNC_CONFIG_UPDATE) { uint64_t old_space, new_space; mutex_enter(&spa_namespace_lock); old_space = metaslab_class_get_space(spa_normal_class(spa)); spa_config_update(spa, SPA_CONFIG_UPDATE_POOL); new_space = metaslab_class_get_space(spa_normal_class(spa)); mutex_exit(&spa_namespace_lock); /* * If the pool grew as a result of the config update, * then log an internal history event. */ if (new_space != old_space) { spa_history_log_internal(spa, "vdev online", NULL, "pool '%s' size: %llu(+%llu)", spa_name(spa), new_space, new_space - old_space); } } /* * See if any devices need to be marked REMOVED. */ if (tasks & SPA_ASYNC_REMOVE) { spa_vdev_state_enter(spa, SCL_NONE); spa_async_remove(spa, spa->spa_root_vdev); for (i = 0; i < spa->spa_l2cache.sav_count; i++) spa_async_remove(spa, spa->spa_l2cache.sav_vdevs[i]); for (i = 0; i < spa->spa_spares.sav_count; i++) spa_async_remove(spa, spa->spa_spares.sav_vdevs[i]); (void) spa_vdev_state_exit(spa, NULL, 0); } if ((tasks & SPA_ASYNC_AUTOEXPAND) && !spa_suspended(spa)) { spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); spa_async_autoexpand(spa, spa->spa_root_vdev); spa_config_exit(spa, SCL_CONFIG, FTAG); } /* * See if any devices need to be probed. */ if (tasks & SPA_ASYNC_PROBE) { spa_vdev_state_enter(spa, SCL_NONE); spa_async_probe(spa, spa->spa_root_vdev); (void) spa_vdev_state_exit(spa, NULL, 0); } /* * If any devices are done replacing, detach them. */ if (tasks & SPA_ASYNC_RESILVER_DONE) spa_vdev_resilver_done(spa); /* * Kick off a resilver. */ if (tasks & SPA_ASYNC_RESILVER) dsl_resilver_restart(spa->spa_dsl_pool, 0); /* * Let the world know that we're done. */ mutex_enter(&spa->spa_async_lock); spa->spa_async_thread = NULL; cv_broadcast(&spa->spa_async_cv); mutex_exit(&spa->spa_async_lock); thread_exit(); } void spa_async_suspend(spa_t *spa) { mutex_enter(&spa->spa_async_lock); spa->spa_async_suspended++; while (spa->spa_async_thread != NULL) cv_wait(&spa->spa_async_cv, &spa->spa_async_lock); mutex_exit(&spa->spa_async_lock); } void spa_async_resume(spa_t *spa) { mutex_enter(&spa->spa_async_lock); ASSERT(spa->spa_async_suspended != 0); spa->spa_async_suspended--; mutex_exit(&spa->spa_async_lock); } static void spa_async_dispatch(spa_t *spa) { mutex_enter(&spa->spa_async_lock); if (spa->spa_async_tasks && !spa->spa_async_suspended && spa->spa_async_thread == NULL && rootdir != NULL && !vn_is_readonly(rootdir)) spa->spa_async_thread = thread_create(NULL, 0, spa_async_thread, spa, 0, &p0, TS_RUN, maxclsyspri); mutex_exit(&spa->spa_async_lock); } void spa_async_request(spa_t *spa, int task) { zfs_dbgmsg("spa=%s async request task=%u", spa->spa_name, task); mutex_enter(&spa->spa_async_lock); spa->spa_async_tasks |= task; mutex_exit(&spa->spa_async_lock); } /* * ========================================================================== * SPA syncing routines * ========================================================================== */ static int bpobj_enqueue_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx) { bpobj_t *bpo = arg; bpobj_enqueue(bpo, bp, tx); return (0); } static int spa_free_sync_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx) { zio_t *zio = arg; zio_nowait(zio_free_sync(zio, zio->io_spa, dmu_tx_get_txg(tx), bp, zio->io_flags)); return (0); } /* * Note: this simple function is not inlined to make it easier to dtrace the * amount of time spent syncing frees. */ static void spa_sync_frees(spa_t *spa, bplist_t *bpl, dmu_tx_t *tx) { zio_t *zio = zio_root(spa, NULL, NULL, 0); bplist_iterate(bpl, spa_free_sync_cb, zio, tx); VERIFY(zio_wait(zio) == 0); } /* * Note: this simple function is not inlined to make it easier to dtrace the * amount of time spent syncing deferred frees. */ static void spa_sync_deferred_frees(spa_t *spa, dmu_tx_t *tx) { zio_t *zio = zio_root(spa, NULL, NULL, 0); VERIFY3U(bpobj_iterate(&spa->spa_deferred_bpobj, spa_free_sync_cb, zio, tx), ==, 0); VERIFY0(zio_wait(zio)); } static void spa_sync_nvlist(spa_t *spa, uint64_t obj, nvlist_t *nv, dmu_tx_t *tx) { char *packed = NULL; size_t bufsize; size_t nvsize = 0; dmu_buf_t *db; VERIFY(nvlist_size(nv, &nvsize, NV_ENCODE_XDR) == 0); /* * Write full (SPA_CONFIG_BLOCKSIZE) blocks of configuration * information. This avoids the dmu_buf_will_dirty() path and * saves us a pre-read to get data we don't actually care about. */ bufsize = P2ROUNDUP((uint64_t)nvsize, SPA_CONFIG_BLOCKSIZE); packed = vmem_alloc(bufsize, KM_PUSHPAGE); VERIFY(nvlist_pack(nv, &packed, &nvsize, NV_ENCODE_XDR, KM_PUSHPAGE) == 0); bzero(packed + nvsize, bufsize - nvsize); dmu_write(spa->spa_meta_objset, obj, 0, bufsize, packed, tx); vmem_free(packed, bufsize); VERIFY(0 == dmu_bonus_hold(spa->spa_meta_objset, obj, FTAG, &db)); dmu_buf_will_dirty(db, tx); *(uint64_t *)db->db_data = nvsize; dmu_buf_rele(db, FTAG); } static void spa_sync_aux_dev(spa_t *spa, spa_aux_vdev_t *sav, dmu_tx_t *tx, const char *config, const char *entry) { nvlist_t *nvroot; nvlist_t **list; int i; if (!sav->sav_sync) return; /* * Update the MOS nvlist describing the list of available devices. * spa_validate_aux() will have already made sure this nvlist is * valid and the vdevs are labeled appropriately. */ if (sav->sav_object == 0) { sav->sav_object = dmu_object_alloc(spa->spa_meta_objset, DMU_OT_PACKED_NVLIST, 1 << 14, DMU_OT_PACKED_NVLIST_SIZE, sizeof (uint64_t), tx); VERIFY(zap_update(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, entry, sizeof (uint64_t), 1, &sav->sav_object, tx) == 0); } VERIFY(nvlist_alloc(&nvroot, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); if (sav->sav_count == 0) { VERIFY(nvlist_add_nvlist_array(nvroot, config, NULL, 0) == 0); } else { list = kmem_alloc(sav->sav_count*sizeof (void *), KM_PUSHPAGE); for (i = 0; i < sav->sav_count; i++) list[i] = vdev_config_generate(spa, sav->sav_vdevs[i], B_FALSE, VDEV_CONFIG_L2CACHE); VERIFY(nvlist_add_nvlist_array(nvroot, config, list, sav->sav_count) == 0); for (i = 0; i < sav->sav_count; i++) nvlist_free(list[i]); kmem_free(list, sav->sav_count * sizeof (void *)); } spa_sync_nvlist(spa, sav->sav_object, nvroot, tx); nvlist_free(nvroot); sav->sav_sync = B_FALSE; } static void spa_sync_config_object(spa_t *spa, dmu_tx_t *tx) { nvlist_t *config; if (list_is_empty(&spa->spa_config_dirty_list)) return; spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); config = spa_config_generate(spa, spa->spa_root_vdev, dmu_tx_get_txg(tx), B_FALSE); /* * If we're upgrading the spa version then make sure that * the config object gets updated with the correct version. */ if (spa->spa_ubsync.ub_version < spa->spa_uberblock.ub_version) fnvlist_add_uint64(config, ZPOOL_CONFIG_VERSION, spa->spa_uberblock.ub_version); spa_config_exit(spa, SCL_STATE, FTAG); if (spa->spa_config_syncing) nvlist_free(spa->spa_config_syncing); spa->spa_config_syncing = config; spa_sync_nvlist(spa, spa->spa_config_object, config, tx); } static void spa_sync_version(void *arg, dmu_tx_t *tx) { uint64_t *versionp = arg; uint64_t version = *versionp; spa_t *spa = dmu_tx_pool(tx)->dp_spa; /* * Setting the version is special cased when first creating the pool. */ ASSERT(tx->tx_txg != TXG_INITIAL); ASSERT(SPA_VERSION_IS_SUPPORTED(version)); ASSERT(version >= spa_version(spa)); spa->spa_uberblock.ub_version = version; vdev_config_dirty(spa->spa_root_vdev); spa_history_log_internal(spa, "set", tx, "version=%lld", version); } /* * Set zpool properties. */ static void spa_sync_props(void *arg, dmu_tx_t *tx) { nvlist_t *nvp = arg; spa_t *spa = dmu_tx_pool(tx)->dp_spa; objset_t *mos = spa->spa_meta_objset; nvpair_t *elem = NULL; mutex_enter(&spa->spa_props_lock); while ((elem = nvlist_next_nvpair(nvp, elem))) { uint64_t intval; char *strval, *fname; zpool_prop_t prop; const char *propname; zprop_type_t proptype; spa_feature_t fid; prop = zpool_name_to_prop(nvpair_name(elem)); switch ((int)prop) { case ZPROP_INVAL: /* * We checked this earlier in spa_prop_validate(). */ ASSERT(zpool_prop_feature(nvpair_name(elem))); fname = strchr(nvpair_name(elem), '@') + 1; VERIFY0(zfeature_lookup_name(fname, &fid)); spa_feature_enable(spa, fid, tx); spa_history_log_internal(spa, "set", tx, "%s=enabled", nvpair_name(elem)); break; case ZPOOL_PROP_VERSION: intval = fnvpair_value_uint64(elem); /* * The version is synced seperatly before other * properties and should be correct by now. */ ASSERT3U(spa_version(spa), >=, intval); break; case ZPOOL_PROP_ALTROOT: /* * 'altroot' is a non-persistent property. It should * have been set temporarily at creation or import time. */ ASSERT(spa->spa_root != NULL); break; case ZPOOL_PROP_READONLY: case ZPOOL_PROP_CACHEFILE: /* * 'readonly' and 'cachefile' are also non-persisitent * properties. */ break; case ZPOOL_PROP_COMMENT: strval = fnvpair_value_string(elem); if (spa->spa_comment != NULL) spa_strfree(spa->spa_comment); spa->spa_comment = spa_strdup(strval); /* * We need to dirty the configuration on all the vdevs * so that their labels get updated. It's unnecessary * to do this for pool creation since the vdev's * configuratoin has already been dirtied. */ if (tx->tx_txg != TXG_INITIAL) vdev_config_dirty(spa->spa_root_vdev); spa_history_log_internal(spa, "set", tx, "%s=%s", nvpair_name(elem), strval); break; default: /* * Set pool property values in the poolprops mos object. */ if (spa->spa_pool_props_object == 0) { spa->spa_pool_props_object = zap_create_link(mos, DMU_OT_POOL_PROPS, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_PROPS, tx); } /* normalize the property name */ propname = zpool_prop_to_name(prop); proptype = zpool_prop_get_type(prop); if (nvpair_type(elem) == DATA_TYPE_STRING) { ASSERT(proptype == PROP_TYPE_STRING); strval = fnvpair_value_string(elem); VERIFY0(zap_update(mos, spa->spa_pool_props_object, propname, 1, strlen(strval) + 1, strval, tx)); spa_history_log_internal(spa, "set", tx, "%s=%s", nvpair_name(elem), strval); } else if (nvpair_type(elem) == DATA_TYPE_UINT64) { intval = fnvpair_value_uint64(elem); if (proptype == PROP_TYPE_INDEX) { const char *unused; VERIFY0(zpool_prop_index_to_string( prop, intval, &unused)); } VERIFY0(zap_update(mos, spa->spa_pool_props_object, propname, 8, 1, &intval, tx)); spa_history_log_internal(spa, "set", tx, "%s=%lld", nvpair_name(elem), intval); } else { ASSERT(0); /* not allowed */ } switch (prop) { case ZPOOL_PROP_DELEGATION: spa->spa_delegation = intval; break; case ZPOOL_PROP_BOOTFS: spa->spa_bootfs = intval; break; case ZPOOL_PROP_FAILUREMODE: spa->spa_failmode = intval; break; case ZPOOL_PROP_AUTOEXPAND: spa->spa_autoexpand = intval; if (tx->tx_txg != TXG_INITIAL) spa_async_request(spa, SPA_ASYNC_AUTOEXPAND); break; case ZPOOL_PROP_DEDUPDITTO: spa->spa_dedup_ditto = intval; break; default: break; } } } mutex_exit(&spa->spa_props_lock); } /* * Perform one-time upgrade on-disk changes. spa_version() does not * reflect the new version this txg, so there must be no changes this * txg to anything that the upgrade code depends on after it executes. * Therefore this must be called after dsl_pool_sync() does the sync * tasks. */ static void spa_sync_upgrades(spa_t *spa, dmu_tx_t *tx) { dsl_pool_t *dp = spa->spa_dsl_pool; ASSERT(spa->spa_sync_pass == 1); rrw_enter(&dp->dp_config_rwlock, RW_WRITER, FTAG); if (spa->spa_ubsync.ub_version < SPA_VERSION_ORIGIN && spa->spa_uberblock.ub_version >= SPA_VERSION_ORIGIN) { dsl_pool_create_origin(dp, tx); /* Keeping the origin open increases spa_minref */ spa->spa_minref += 3; } if (spa->spa_ubsync.ub_version < SPA_VERSION_NEXT_CLONES && spa->spa_uberblock.ub_version >= SPA_VERSION_NEXT_CLONES) { dsl_pool_upgrade_clones(dp, tx); } if (spa->spa_ubsync.ub_version < SPA_VERSION_DIR_CLONES && spa->spa_uberblock.ub_version >= SPA_VERSION_DIR_CLONES) { dsl_pool_upgrade_dir_clones(dp, tx); /* Keeping the freedir open increases spa_minref */ spa->spa_minref += 3; } if (spa->spa_ubsync.ub_version < SPA_VERSION_FEATURES && spa->spa_uberblock.ub_version >= SPA_VERSION_FEATURES) { spa_feature_create_zap_objects(spa, tx); } /* * LZ4_COMPRESS feature's behaviour was changed to activate_on_enable * when possibility to use lz4 compression for metadata was added * Old pools that have this feature enabled must be upgraded to have * this feature active */ if (spa->spa_uberblock.ub_version >= SPA_VERSION_FEATURES) { boolean_t lz4_en = spa_feature_is_enabled(spa, SPA_FEATURE_LZ4_COMPRESS); boolean_t lz4_ac = spa_feature_is_active(spa, SPA_FEATURE_LZ4_COMPRESS); if (lz4_en && !lz4_ac) spa_feature_incr(spa, SPA_FEATURE_LZ4_COMPRESS, tx); } rrw_exit(&dp->dp_config_rwlock, FTAG); } /* * Sync the specified transaction group. New blocks may be dirtied as * part of the process, so we iterate until it converges. */ void spa_sync(spa_t *spa, uint64_t txg) { dsl_pool_t *dp = spa->spa_dsl_pool; objset_t *mos = spa->spa_meta_objset; bplist_t *free_bpl = &spa->spa_free_bplist[txg & TXG_MASK]; vdev_t *rvd = spa->spa_root_vdev; vdev_t *vd; dmu_tx_t *tx; int error; int c; VERIFY(spa_writeable(spa)); /* * Lock out configuration changes. */ spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); spa->spa_syncing_txg = txg; spa->spa_sync_pass = 0; /* * If there are any pending vdev state changes, convert them * into config changes that go out with this transaction group. */ spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); while (list_head(&spa->spa_state_dirty_list) != NULL) { /* * We need the write lock here because, for aux vdevs, * calling vdev_config_dirty() modifies sav_config. * This is ugly and will become unnecessary when we * eliminate the aux vdev wart by integrating all vdevs * into the root vdev tree. */ spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_WRITER); while ((vd = list_head(&spa->spa_state_dirty_list)) != NULL) { vdev_state_clean(vd); vdev_config_dirty(vd); } spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_READER); } spa_config_exit(spa, SCL_STATE, FTAG); tx = dmu_tx_create_assigned(dp, txg); spa->spa_sync_starttime = gethrtime(); taskq_cancel_id(system_taskq, spa->spa_deadman_tqid); spa->spa_deadman_tqid = taskq_dispatch_delay(system_taskq, spa_deadman, spa, TQ_PUSHPAGE, ddi_get_lbolt() + NSEC_TO_TICK(spa->spa_deadman_synctime)); /* * If we are upgrading to SPA_VERSION_RAIDZ_DEFLATE this txg, * set spa_deflate if we have no raid-z vdevs. */ if (spa->spa_ubsync.ub_version < SPA_VERSION_RAIDZ_DEFLATE && spa->spa_uberblock.ub_version >= SPA_VERSION_RAIDZ_DEFLATE) { int i; for (i = 0; i < rvd->vdev_children; i++) { vd = rvd->vdev_child[i]; if (vd->vdev_deflate_ratio != SPA_MINBLOCKSIZE) break; } if (i == rvd->vdev_children) { spa->spa_deflate = TRUE; VERIFY(0 == zap_add(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_DEFLATE, sizeof (uint64_t), 1, &spa->spa_deflate, tx)); } } /* * If anything has changed in this txg, or if someone is waiting * for this txg to sync (eg, spa_vdev_remove()), push the * deferred frees from the previous txg. If not, leave them * alone so that we don't generate work on an otherwise idle * system. */ if (!txg_list_empty(&dp->dp_dirty_datasets, txg) || !txg_list_empty(&dp->dp_dirty_dirs, txg) || !txg_list_empty(&dp->dp_sync_tasks, txg) || ((dsl_scan_active(dp->dp_scan) || txg_sync_waiting(dp)) && !spa_shutting_down(spa))) { spa_sync_deferred_frees(spa, tx); } /* * Iterate to convergence. */ do { int pass = ++spa->spa_sync_pass; spa_sync_config_object(spa, tx); spa_sync_aux_dev(spa, &spa->spa_spares, tx, ZPOOL_CONFIG_SPARES, DMU_POOL_SPARES); spa_sync_aux_dev(spa, &spa->spa_l2cache, tx, ZPOOL_CONFIG_L2CACHE, DMU_POOL_L2CACHE); spa_errlog_sync(spa, txg); dsl_pool_sync(dp, txg); if (pass < zfs_sync_pass_deferred_free) { spa_sync_frees(spa, free_bpl, tx); } else { bplist_iterate(free_bpl, bpobj_enqueue_cb, &spa->spa_deferred_bpobj, tx); } ddt_sync(spa, txg); dsl_scan_sync(dp, tx); while ((vd = txg_list_remove(&spa->spa_vdev_txg_list, txg))) vdev_sync(vd, txg); if (pass == 1) spa_sync_upgrades(spa, tx); } while (dmu_objset_is_dirty(mos, txg)); /* * Rewrite the vdev configuration (which includes the uberblock) * to commit the transaction group. * * If there are no dirty vdevs, we sync the uberblock to a few * random top-level vdevs that are known to be visible in the * config cache (see spa_vdev_add() for a complete description). * If there *are* dirty vdevs, sync the uberblock to all vdevs. */ for (;;) { /* * We hold SCL_STATE to prevent vdev open/close/etc. * while we're attempting to write the vdev labels. */ spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); if (list_is_empty(&spa->spa_config_dirty_list)) { vdev_t *svd[SPA_DVAS_PER_BP]; int svdcount = 0; int children = rvd->vdev_children; int c0 = spa_get_random(children); for (c = 0; c < children; c++) { vd = rvd->vdev_child[(c0 + c) % children]; if (vd->vdev_ms_array == 0 || vd->vdev_islog) continue; svd[svdcount++] = vd; if (svdcount == SPA_DVAS_PER_BP) break; } error = vdev_config_sync(svd, svdcount, txg, B_FALSE); if (error != 0) error = vdev_config_sync(svd, svdcount, txg, B_TRUE); } else { error = vdev_config_sync(rvd->vdev_child, rvd->vdev_children, txg, B_FALSE); if (error != 0) error = vdev_config_sync(rvd->vdev_child, rvd->vdev_children, txg, B_TRUE); } if (error == 0) spa->spa_last_synced_guid = rvd->vdev_guid; spa_config_exit(spa, SCL_STATE, FTAG); if (error == 0) break; zio_suspend(spa, NULL); zio_resume_wait(spa); } dmu_tx_commit(tx); taskq_cancel_id(system_taskq, spa->spa_deadman_tqid); spa->spa_deadman_tqid = 0; /* * Clear the dirty config list. */ while ((vd = list_head(&spa->spa_config_dirty_list)) != NULL) vdev_config_clean(vd); /* * Now that the new config has synced transactionally, * let it become visible to the config cache. */ if (spa->spa_config_syncing != NULL) { spa_config_set(spa, spa->spa_config_syncing); spa->spa_config_txg = txg; spa->spa_config_syncing = NULL; } spa->spa_ubsync = spa->spa_uberblock; dsl_pool_sync_done(dp, txg); /* * Update usable space statistics. */ while ((vd = txg_list_remove(&spa->spa_vdev_txg_list, TXG_CLEAN(txg)))) vdev_sync_done(vd, txg); spa_update_dspace(spa); /* * It had better be the case that we didn't dirty anything * since vdev_config_sync(). */ ASSERT(txg_list_empty(&dp->dp_dirty_datasets, txg)); ASSERT(txg_list_empty(&dp->dp_dirty_dirs, txg)); ASSERT(txg_list_empty(&spa->spa_vdev_txg_list, txg)); spa->spa_sync_pass = 0; spa_config_exit(spa, SCL_CONFIG, FTAG); spa_handle_ignored_writes(spa); /* * If any async tasks have been requested, kick them off. */ spa_async_dispatch(spa); } /* * Sync all pools. We don't want to hold the namespace lock across these * operations, so we take a reference on the spa_t and drop the lock during the * sync. */ void spa_sync_allpools(void) { spa_t *spa = NULL; mutex_enter(&spa_namespace_lock); while ((spa = spa_next(spa)) != NULL) { if (spa_state(spa) != POOL_STATE_ACTIVE || !spa_writeable(spa) || spa_suspended(spa)) continue; spa_open_ref(spa, FTAG); mutex_exit(&spa_namespace_lock); txg_wait_synced(spa_get_dsl(spa), 0); mutex_enter(&spa_namespace_lock); spa_close(spa, FTAG); } mutex_exit(&spa_namespace_lock); } /* * ========================================================================== * Miscellaneous routines * ========================================================================== */ /* * Remove all pools in the system. */ void spa_evict_all(void) { spa_t *spa; /* * Remove all cached state. All pools should be closed now, * so every spa in the AVL tree should be unreferenced. */ mutex_enter(&spa_namespace_lock); while ((spa = spa_next(NULL)) != NULL) { /* * Stop async tasks. The async thread may need to detach * a device that's been replaced, which requires grabbing * spa_namespace_lock, so we must drop it here. */ spa_open_ref(spa, FTAG); mutex_exit(&spa_namespace_lock); spa_async_suspend(spa); mutex_enter(&spa_namespace_lock); spa_close(spa, FTAG); if (spa->spa_state != POOL_STATE_UNINITIALIZED) { spa_unload(spa); spa_deactivate(spa); } spa_remove(spa); } mutex_exit(&spa_namespace_lock); } vdev_t * spa_lookup_by_guid(spa_t *spa, uint64_t guid, boolean_t aux) { vdev_t *vd; int i; if ((vd = vdev_lookup_by_guid(spa->spa_root_vdev, guid)) != NULL) return (vd); if (aux) { for (i = 0; i < spa->spa_l2cache.sav_count; i++) { vd = spa->spa_l2cache.sav_vdevs[i]; if (vd->vdev_guid == guid) return (vd); } for (i = 0; i < spa->spa_spares.sav_count; i++) { vd = spa->spa_spares.sav_vdevs[i]; if (vd->vdev_guid == guid) return (vd); } } return (NULL); } void spa_upgrade(spa_t *spa, uint64_t version) { ASSERT(spa_writeable(spa)); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); /* * This should only be called for a non-faulted pool, and since a * future version would result in an unopenable pool, this shouldn't be * possible. */ ASSERT(SPA_VERSION_IS_SUPPORTED(spa->spa_uberblock.ub_version)); ASSERT3U(version, >=, spa->spa_uberblock.ub_version); spa->spa_uberblock.ub_version = version; vdev_config_dirty(spa->spa_root_vdev); spa_config_exit(spa, SCL_ALL, FTAG); txg_wait_synced(spa_get_dsl(spa), 0); } boolean_t spa_has_spare(spa_t *spa, uint64_t guid) { int i; uint64_t spareguid; spa_aux_vdev_t *sav = &spa->spa_spares; for (i = 0; i < sav->sav_count; i++) if (sav->sav_vdevs[i]->vdev_guid == guid) return (B_TRUE); for (i = 0; i < sav->sav_npending; i++) { if (nvlist_lookup_uint64(sav->sav_pending[i], ZPOOL_CONFIG_GUID, &spareguid) == 0 && spareguid == guid) return (B_TRUE); } return (B_FALSE); } /* * Check if a pool has an active shared spare device. * Note: reference count of an active spare is 2, as a spare and as a replace */ static boolean_t spa_has_active_shared_spare(spa_t *spa) { int i, refcnt; uint64_t pool; spa_aux_vdev_t *sav = &spa->spa_spares; for (i = 0; i < sav->sav_count; i++) { if (spa_spare_exists(sav->sav_vdevs[i]->vdev_guid, &pool, &refcnt) && pool != 0ULL && pool == spa_guid(spa) && refcnt > 2) return (B_TRUE); } return (B_FALSE); } /* * Post a FM_EREPORT_ZFS_* event from sys/fm/fs/zfs.h. The payload will be * filled in from the spa and (optionally) the vdev. This doesn't do anything * in the userland libzpool, as we don't want consumers to misinterpret ztest * or zdb as real changes. */ void spa_event_notify(spa_t *spa, vdev_t *vd, const char *name) { #ifdef _KERNEL zfs_ereport_post(name, spa, vd, NULL, 0, 0); #endif } #if defined(_KERNEL) && defined(HAVE_SPL) /* state manipulation functions */ EXPORT_SYMBOL(spa_open); EXPORT_SYMBOL(spa_open_rewind); EXPORT_SYMBOL(spa_get_stats); EXPORT_SYMBOL(spa_create); EXPORT_SYMBOL(spa_import_rootpool); EXPORT_SYMBOL(spa_import); EXPORT_SYMBOL(spa_tryimport); EXPORT_SYMBOL(spa_destroy); EXPORT_SYMBOL(spa_export); EXPORT_SYMBOL(spa_reset); EXPORT_SYMBOL(spa_async_request); EXPORT_SYMBOL(spa_async_suspend); EXPORT_SYMBOL(spa_async_resume); EXPORT_SYMBOL(spa_inject_addref); EXPORT_SYMBOL(spa_inject_delref); EXPORT_SYMBOL(spa_scan_stat_init); EXPORT_SYMBOL(spa_scan_get_stats); /* device maniion */ EXPORT_SYMBOL(spa_vdev_add); EXPORT_SYMBOL(spa_vdev_attach); EXPORT_SYMBOL(spa_vdev_detach); EXPORT_SYMBOL(spa_vdev_remove); EXPORT_SYMBOL(spa_vdev_setpath); EXPORT_SYMBOL(spa_vdev_setfru); EXPORT_SYMBOL(spa_vdev_split_mirror); /* spare statech is global across all pools) */ EXPORT_SYMBOL(spa_spare_add); EXPORT_SYMBOL(spa_spare_remove); EXPORT_SYMBOL(spa_spare_exists); EXPORT_SYMBOL(spa_spare_activate); /* L2ARC statech is global across all pools) */ EXPORT_SYMBOL(spa_l2cache_add); EXPORT_SYMBOL(spa_l2cache_remove); EXPORT_SYMBOL(spa_l2cache_exists); EXPORT_SYMBOL(spa_l2cache_activate); EXPORT_SYMBOL(spa_l2cache_drop); /* scanning */ EXPORT_SYMBOL(spa_scan); EXPORT_SYMBOL(spa_scan_stop); /* spa syncing */ EXPORT_SYMBOL(spa_sync); /* only for DMU use */ EXPORT_SYMBOL(spa_sync_allpools); /* properties */ EXPORT_SYMBOL(spa_prop_set); EXPORT_SYMBOL(spa_prop_get); EXPORT_SYMBOL(spa_prop_clear_bootfs); /* asynchronous event notification */ EXPORT_SYMBOL(spa_event_notify); #endif #if defined(_KERNEL) && defined(HAVE_SPL) module_param(spa_load_verify_maxinflight, int, 0644); MODULE_PARM_DESC(spa_load_verify_maxinflight, "Max concurrent traversal I/Os while verifying pool during import -X"); module_param(spa_load_verify_metadata, int, 0644); MODULE_PARM_DESC(spa_load_verify_metadata, "Set to traverse metadata on pool import"); module_param(spa_load_verify_data, int, 0644); MODULE_PARM_DESC(spa_load_verify_data, "Set to traverse data on pool import"); #endif diff --git a/module/zfs/spa_config.c b/module/zfs/spa_config.c index a08456d56f60..818a4308f116 100644 --- a/module/zfs/spa_config.c +++ b/module/zfs/spa_config.c @@ -1,558 +1,558 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2013 by Delphix. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #ifdef _KERNEL #include #include #endif /* * Pool configuration repository. * * Pool configuration is stored as a packed nvlist on the filesystem. By * default, all pools are stored in /etc/zfs/zpool.cache and loaded on boot * (when the ZFS module is loaded). Pools can also have the 'cachefile' * property set that allows them to be stored in an alternate location until * the control of external software. * * For each cache file, we have a single nvlist which holds all the * configuration information. When the module loads, we read this information * from /etc/zfs/zpool.cache and populate the SPA namespace. This namespace is * maintained independently in spa.c. Whenever the namespace is modified, or * the configuration of a pool is changed, we call spa_config_sync(), which * walks through all the active pools and writes the configuration to disk. */ static uint64_t spa_config_generation = 1; /* * This can be overridden in userland to preserve an alternate namespace for * userland pools when doing testing. */ char *spa_config_path = ZPOOL_CACHE; int zfs_autoimport_disable = 0; /* * Called when the module is first loaded, this routine loads the configuration * file into the SPA namespace. It does not actually open or load the pools; it * only populates the namespace. */ void spa_config_load(void) { void *buf = NULL; nvlist_t *nvlist, *child; nvpair_t *nvpair; char *pathname; struct _buf *file; uint64_t fsize; if (zfs_autoimport_disable) return; /* * Open the configuration file. */ pathname = kmem_alloc(MAXPATHLEN, KM_PUSHPAGE); (void) snprintf(pathname, MAXPATHLEN, "%s%s", (rootdir != NULL) ? "./" : "", spa_config_path); file = kobj_open_file(pathname); kmem_free(pathname, MAXPATHLEN); if (file == (struct _buf *)-1) return; if (kobj_get_filesize(file, &fsize) != 0) goto out; - buf = kmem_alloc(fsize, KM_PUSHPAGE | KM_NODEBUG); + buf = kmem_alloc(fsize, KM_PUSHPAGE); /* * Read the nvlist from the file. */ if (kobj_read_file(file, buf, fsize, 0) < 0) goto out; /* * Unpack the nvlist. */ if (nvlist_unpack(buf, fsize, &nvlist, KM_PUSHPAGE) != 0) goto out; /* * Iterate over all elements in the nvlist, creating a new spa_t for * each one with the specified configuration. */ mutex_enter(&spa_namespace_lock); nvpair = NULL; while ((nvpair = nvlist_next_nvpair(nvlist, nvpair)) != NULL) { if (nvpair_type(nvpair) != DATA_TYPE_NVLIST) continue; VERIFY(nvpair_value_nvlist(nvpair, &child) == 0); if (spa_lookup(nvpair_name(nvpair)) != NULL) continue; (void) spa_add(nvpair_name(nvpair), child, NULL); } mutex_exit(&spa_namespace_lock); nvlist_free(nvlist); out: if (buf != NULL) kmem_free(buf, fsize); kobj_close_file(file); } static void spa_config_write(spa_config_dirent_t *dp, nvlist_t *nvl) { size_t buflen; char *buf; vnode_t *vp; int oflags = FWRITE | FTRUNC | FCREAT | FOFFMAX; char *temp; /* * If the nvlist is empty (NULL), then remove the old cachefile. */ if (nvl == NULL) { (void) vn_remove(dp->scd_path, UIO_SYSSPACE, RMFILE); return; } /* * Pack the configuration into a buffer. */ VERIFY(nvlist_size(nvl, &buflen, NV_ENCODE_XDR) == 0); - buf = kmem_alloc(buflen, KM_PUSHPAGE | KM_NODEBUG); + buf = kmem_alloc(buflen, KM_PUSHPAGE); temp = kmem_zalloc(MAXPATHLEN, KM_PUSHPAGE); VERIFY(nvlist_pack(nvl, &buf, &buflen, NV_ENCODE_XDR, KM_PUSHPAGE) == 0); /* * Write the configuration to disk. We need to do the traditional * 'write to temporary file, sync, move over original' to make sure we * always have a consistent view of the data. */ (void) snprintf(temp, MAXPATHLEN, "%s.tmp", dp->scd_path); if (vn_open(temp, UIO_SYSSPACE, oflags, 0644, &vp, CRCREAT, 0) == 0) { if (vn_rdwr(UIO_WRITE, vp, buf, buflen, 0, UIO_SYSSPACE, 0, RLIM64_INFINITY, kcred, NULL) == 0 && VOP_FSYNC(vp, FSYNC, kcred, NULL) == 0) { (void) vn_rename(temp, dp->scd_path, UIO_SYSSPACE); } (void) VOP_CLOSE(vp, oflags, 1, 0, kcred, NULL); } (void) vn_remove(temp, UIO_SYSSPACE, RMFILE); kmem_free(buf, buflen); kmem_free(temp, MAXPATHLEN); } /* * Synchronize pool configuration to disk. This must be called with the * namespace lock held. Synchronizing the pool cache is typically done after * the configuration has been synced to the MOS. This exposes a window where * the MOS config will have been updated but the cache file has not. If * the system were to crash at that instant then the cached config may not * contain the correct information to open the pool and an explicity import * would be required. */ void spa_config_sync(spa_t *target, boolean_t removing, boolean_t postsysevent) { spa_config_dirent_t *dp, *tdp; nvlist_t *nvl; char *pool_name; ASSERT(MUTEX_HELD(&spa_namespace_lock)); if (rootdir == NULL || !(spa_mode_global & FWRITE)) return; /* * Iterate over all cachefiles for the pool, past or present. When the * cachefile is changed, the new one is pushed onto this list, allowing * us to update previous cachefiles that no longer contain this pool. */ for (dp = list_head(&target->spa_config_list); dp != NULL; dp = list_next(&target->spa_config_list, dp)) { spa_t *spa = NULL; if (dp->scd_path == NULL) continue; /* * Iterate over all pools, adding any matching pools to 'nvl'. */ nvl = NULL; while ((spa = spa_next(spa)) != NULL) { /* * Skip over our own pool if we're about to remove * ourselves from the spa namespace or any pool that * is readonly. Since we cannot guarantee that a * readonly pool would successfully import upon reboot, * we don't allow them to be written to the cache file. */ if ((spa == target && removing) || !spa_writeable(spa)) continue; mutex_enter(&spa->spa_props_lock); tdp = list_head(&spa->spa_config_list); if (spa->spa_config == NULL || tdp->scd_path == NULL || strcmp(tdp->scd_path, dp->scd_path) != 0) { mutex_exit(&spa->spa_props_lock); continue; } if (nvl == NULL) VERIFY(nvlist_alloc(&nvl, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); if (spa->spa_import_flags & ZFS_IMPORT_TEMP_NAME) { VERIFY0(nvlist_lookup_string(spa->spa_config, ZPOOL_CONFIG_POOL_NAME, &pool_name)); } else pool_name = spa_name(spa); VERIFY(nvlist_add_nvlist(nvl, pool_name, spa->spa_config) == 0); mutex_exit(&spa->spa_props_lock); } spa_config_write(dp, nvl); nvlist_free(nvl); } /* * Remove any config entries older than the current one. */ dp = list_head(&target->spa_config_list); while ((tdp = list_next(&target->spa_config_list, dp)) != NULL) { list_remove(&target->spa_config_list, tdp); if (tdp->scd_path != NULL) spa_strfree(tdp->scd_path); kmem_free(tdp, sizeof (spa_config_dirent_t)); } spa_config_generation++; if (postsysevent) spa_event_notify(target, NULL, FM_EREPORT_ZFS_CONFIG_SYNC); } /* * Sigh. Inside a local zone, we don't have access to /etc/zfs/zpool.cache, * and we don't want to allow the local zone to see all the pools anyway. * So we have to invent the ZFS_IOC_CONFIG ioctl to grab the configuration * information for all pool visible within the zone. */ nvlist_t * spa_all_configs(uint64_t *generation) { nvlist_t *pools; spa_t *spa = NULL; if (*generation == spa_config_generation) return (NULL); VERIFY(nvlist_alloc(&pools, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); mutex_enter(&spa_namespace_lock); while ((spa = spa_next(spa)) != NULL) { if (INGLOBALZONE(curproc) || zone_dataset_visible(spa_name(spa), NULL)) { mutex_enter(&spa->spa_props_lock); VERIFY(nvlist_add_nvlist(pools, spa_name(spa), spa->spa_config) == 0); mutex_exit(&spa->spa_props_lock); } } *generation = spa_config_generation; mutex_exit(&spa_namespace_lock); return (pools); } void spa_config_set(spa_t *spa, nvlist_t *config) { mutex_enter(&spa->spa_props_lock); if (spa->spa_config != NULL) nvlist_free(spa->spa_config); spa->spa_config = config; mutex_exit(&spa->spa_props_lock); } /* * Generate the pool's configuration based on the current in-core state. * * We infer whether to generate a complete config or just one top-level config * based on whether vd is the root vdev. */ nvlist_t * spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg, int getstats) { nvlist_t *config, *nvroot; vdev_t *rvd = spa->spa_root_vdev; unsigned long hostid = 0; boolean_t locked = B_FALSE; uint64_t split_guid; char *pool_name; if (vd == NULL) { vd = rvd; locked = B_TRUE; spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_READER); } ASSERT(spa_config_held(spa, SCL_CONFIG | SCL_STATE, RW_READER) == (SCL_CONFIG | SCL_STATE)); /* * If txg is -1, report the current value of spa->spa_config_txg. */ if (txg == -1ULL) txg = spa->spa_config_txg; /* * Originally, users had to handle spa namespace collisions by either * exporting the already imported pool or by specifying a new name for * the pool with a conflicting name. In the case of root pools from * virtual guests, neither approach to collision resolution is * reasonable. This is addressed by extending the new name syntax with * an option to specify that the new name is temporary. When specified, * ZFS_IMPORT_TEMP_NAME will be set in spa->spa_import_flags to tell us * to use the previous name, which we do below. */ if (spa->spa_import_flags & ZFS_IMPORT_TEMP_NAME) { VERIFY0(nvlist_lookup_string(spa->spa_config, ZPOOL_CONFIG_POOL_NAME, &pool_name)); } else pool_name = spa_name(spa); VERIFY(nvlist_alloc(&config, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_VERSION, spa_version(spa)) == 0); VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME, pool_name) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE, spa_state(spa)) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_TXG, txg) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_GUID, spa_guid(spa)) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_ERRATA, spa->spa_errata) == 0); VERIFY(spa->spa_comment == NULL || nvlist_add_string(config, ZPOOL_CONFIG_COMMENT, spa->spa_comment) == 0); #ifdef _KERNEL hostid = zone_get_hostid(NULL); #else /* _KERNEL */ /* * We're emulating the system's hostid in userland, so we can't use * zone_get_hostid(). */ (void) ddi_strtoul(hw_serial, NULL, 10, &hostid); #endif /* _KERNEL */ if (hostid != 0) { VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_HOSTID, hostid) == 0); } VERIFY0(nvlist_add_string(config, ZPOOL_CONFIG_HOSTNAME, utsname()->nodename)); if (vd != rvd) { VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_TOP_GUID, vd->vdev_top->vdev_guid) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_GUID, vd->vdev_guid) == 0); if (vd->vdev_isspare) VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_IS_SPARE, 1ULL) == 0); if (vd->vdev_islog) VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_IS_LOG, 1ULL) == 0); vd = vd->vdev_top; /* label contains top config */ } else { /* * Only add the (potentially large) split information * in the mos config, and not in the vdev labels */ if (spa->spa_config_splitting != NULL) VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_SPLIT, spa->spa_config_splitting) == 0); } /* * Add the top-level config. We even add this on pools which * don't support holes in the namespace. */ vdev_top_config_generate(spa, config); /* * If we're splitting, record the original pool's guid. */ if (spa->spa_config_splitting != NULL && nvlist_lookup_uint64(spa->spa_config_splitting, ZPOOL_CONFIG_SPLIT_GUID, &split_guid) == 0) { VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_SPLIT_GUID, split_guid) == 0); } nvroot = vdev_config_generate(spa, vd, getstats, 0); VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, nvroot) == 0); nvlist_free(nvroot); /* * Store what's necessary for reading the MOS in the label. */ VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_FEATURES_FOR_READ, spa->spa_label_features) == 0); if (getstats && spa_load_state(spa) == SPA_LOAD_NONE) { ddt_histogram_t *ddh; ddt_stat_t *dds; ddt_object_t *ddo; ddh = kmem_zalloc(sizeof (ddt_histogram_t), KM_PUSHPAGE); ddt_get_dedup_histogram(spa, ddh); VERIFY(nvlist_add_uint64_array(config, ZPOOL_CONFIG_DDT_HISTOGRAM, (uint64_t *)ddh, sizeof (*ddh) / sizeof (uint64_t)) == 0); kmem_free(ddh, sizeof (ddt_histogram_t)); ddo = kmem_zalloc(sizeof (ddt_object_t), KM_PUSHPAGE); ddt_get_dedup_object_stats(spa, ddo); VERIFY(nvlist_add_uint64_array(config, ZPOOL_CONFIG_DDT_OBJ_STATS, (uint64_t *)ddo, sizeof (*ddo) / sizeof (uint64_t)) == 0); kmem_free(ddo, sizeof (ddt_object_t)); dds = kmem_zalloc(sizeof (ddt_stat_t), KM_PUSHPAGE); ddt_get_dedup_stats(spa, dds); VERIFY(nvlist_add_uint64_array(config, ZPOOL_CONFIG_DDT_STATS, (uint64_t *)dds, sizeof (*dds) / sizeof (uint64_t)) == 0); kmem_free(dds, sizeof (ddt_stat_t)); } if (locked) spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); return (config); } /* * Update all disk labels, generate a fresh config based on the current * in-core state, and sync the global config cache (do not sync the config * cache if this is a booting rootpool). */ void spa_config_update(spa_t *spa, int what) { vdev_t *rvd = spa->spa_root_vdev; uint64_t txg; int c; ASSERT(MUTEX_HELD(&spa_namespace_lock)); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); txg = spa_last_synced_txg(spa) + 1; if (what == SPA_CONFIG_UPDATE_POOL) { vdev_config_dirty(rvd); } else { /* * If we have top-level vdevs that were added but have * not yet been prepared for allocation, do that now. * (It's safe now because the config cache is up to date, * so it will be able to translate the new DVAs.) * See comments in spa_vdev_add() for full details. */ for (c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; if (tvd->vdev_ms_array == 0) vdev_metaslab_set_size(tvd); vdev_expand(tvd, txg); } } spa_config_exit(spa, SCL_ALL, FTAG); /* * Wait for the mosconfig to be regenerated and synced. */ txg_wait_synced(spa->spa_dsl_pool, txg); /* * Update the global config cache to reflect the new mosconfig. */ if (!spa->spa_is_root) spa_config_sync(spa, B_FALSE, what != SPA_CONFIG_UPDATE_POOL); if (what == SPA_CONFIG_UPDATE_POOL) spa_config_update(spa, SPA_CONFIG_UPDATE_VDEVS); } #if defined(_KERNEL) && defined(HAVE_SPL) EXPORT_SYMBOL(spa_config_sync); EXPORT_SYMBOL(spa_config_load); EXPORT_SYMBOL(spa_all_configs); EXPORT_SYMBOL(spa_config_set); EXPORT_SYMBOL(spa_config_generate); EXPORT_SYMBOL(spa_config_update); module_param(spa_config_path, charp, 0444); MODULE_PARM_DESC(spa_config_path, "SPA config file (/etc/zfs/zpool.cache)"); module_param(zfs_autoimport_disable, int, 0644); MODULE_PARM_DESC(zfs_autoimport_disable, "Disable pool import at module load"); #endif diff --git a/module/zfs/spa_misc.c b/module/zfs/spa_misc.c index 2a2e72d5a359..363f982febfa 100644 --- a/module/zfs/spa_misc.c +++ b/module/zfs/spa_misc.c @@ -1,2002 +1,2002 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2014 by Delphix. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "zfs_prop.h" #include "zfeature_common.h" /* * SPA locking * * There are four basic locks for managing spa_t structures: * * spa_namespace_lock (global mutex) * * This lock must be acquired to do any of the following: * * - Lookup a spa_t by name * - Add or remove a spa_t from the namespace * - Increase spa_refcount from non-zero * - Check if spa_refcount is zero * - Rename a spa_t * - add/remove/attach/detach devices * - Held for the duration of create/destroy/import/export * * It does not need to handle recursion. A create or destroy may * reference objects (files or zvols) in other pools, but by * definition they must have an existing reference, and will never need * to lookup a spa_t by name. * * spa_refcount (per-spa refcount_t protected by mutex) * * This reference count keep track of any active users of the spa_t. The * spa_t cannot be destroyed or freed while this is non-zero. Internally, * the refcount is never really 'zero' - opening a pool implicitly keeps * some references in the DMU. Internally we check against spa_minref, but * present the image of a zero/non-zero value to consumers. * * spa_config_lock[] (per-spa array of rwlocks) * * This protects the spa_t from config changes, and must be held in * the following circumstances: * * - RW_READER to perform I/O to the spa * - RW_WRITER to change the vdev config * * The locking order is fairly straightforward: * * spa_namespace_lock -> spa_refcount * * The namespace lock must be acquired to increase the refcount from 0 * or to check if it is zero. * * spa_refcount -> spa_config_lock[] * * There must be at least one valid reference on the spa_t to acquire * the config lock. * * spa_namespace_lock -> spa_config_lock[] * * The namespace lock must always be taken before the config lock. * * * The spa_namespace_lock can be acquired directly and is globally visible. * * The namespace is manipulated using the following functions, all of which * require the spa_namespace_lock to be held. * * spa_lookup() Lookup a spa_t by name. * * spa_add() Create a new spa_t in the namespace. * * spa_remove() Remove a spa_t from the namespace. This also * frees up any memory associated with the spa_t. * * spa_next() Returns the next spa_t in the system, or the * first if NULL is passed. * * spa_evict_all() Shutdown and remove all spa_t structures in * the system. * * spa_guid_exists() Determine whether a pool/device guid exists. * * The spa_refcount is manipulated using the following functions: * * spa_open_ref() Adds a reference to the given spa_t. Must be * called with spa_namespace_lock held if the * refcount is currently zero. * * spa_close() Remove a reference from the spa_t. This will * not free the spa_t or remove it from the * namespace. No locking is required. * * spa_refcount_zero() Returns true if the refcount is currently * zero. Must be called with spa_namespace_lock * held. * * The spa_config_lock[] is an array of rwlocks, ordered as follows: * SCL_CONFIG > SCL_STATE > SCL_ALLOC > SCL_ZIO > SCL_FREE > SCL_VDEV. * spa_config_lock[] is manipulated with spa_config_{enter,exit,held}(). * * To read the configuration, it suffices to hold one of these locks as reader. * To modify the configuration, you must hold all locks as writer. To modify * vdev state without altering the vdev tree's topology (e.g. online/offline), * you must hold SCL_STATE and SCL_ZIO as writer. * * We use these distinct config locks to avoid recursive lock entry. * For example, spa_sync() (which holds SCL_CONFIG as reader) induces * block allocations (SCL_ALLOC), which may require reading space maps * from disk (dmu_read() -> zio_read() -> SCL_ZIO). * * The spa config locks cannot be normal rwlocks because we need the * ability to hand off ownership. For example, SCL_ZIO is acquired * by the issuing thread and later released by an interrupt thread. * They do, however, obey the usual write-wanted semantics to prevent * writer (i.e. system administrator) starvation. * * The lock acquisition rules are as follows: * * SCL_CONFIG * Protects changes to the vdev tree topology, such as vdev * add/remove/attach/detach. Protects the dirty config list * (spa_config_dirty_list) and the set of spares and l2arc devices. * * SCL_STATE * Protects changes to pool state and vdev state, such as vdev * online/offline/fault/degrade/clear. Protects the dirty state list * (spa_state_dirty_list) and global pool state (spa_state). * * SCL_ALLOC * Protects changes to metaslab groups and classes. * Held as reader by metaslab_alloc() and metaslab_claim(). * * SCL_ZIO * Held by bp-level zios (those which have no io_vd upon entry) * to prevent changes to the vdev tree. The bp-level zio implicitly * protects all of its vdev child zios, which do not hold SCL_ZIO. * * SCL_FREE * Protects changes to metaslab groups and classes. * Held as reader by metaslab_free(). SCL_FREE is distinct from * SCL_ALLOC, and lower than SCL_ZIO, so that we can safely free * blocks in zio_done() while another i/o that holds either * SCL_ALLOC or SCL_ZIO is waiting for this i/o to complete. * * SCL_VDEV * Held as reader to prevent changes to the vdev tree during trivial * inquiries such as bp_get_dsize(). SCL_VDEV is distinct from the * other locks, and lower than all of them, to ensure that it's safe * to acquire regardless of caller context. * * In addition, the following rules apply: * * (a) spa_props_lock protects pool properties, spa_config and spa_config_list. * The lock ordering is SCL_CONFIG > spa_props_lock. * * (b) I/O operations on leaf vdevs. For any zio operation that takes * an explicit vdev_t argument -- such as zio_ioctl(), zio_read_phys(), * or zio_write_phys() -- the caller must ensure that the config cannot * cannot change in the interim, and that the vdev cannot be reopened. * SCL_STATE as reader suffices for both. * * The vdev configuration is protected by spa_vdev_enter() / spa_vdev_exit(). * * spa_vdev_enter() Acquire the namespace lock and the config lock * for writing. * * spa_vdev_exit() Release the config lock, wait for all I/O * to complete, sync the updated configs to the * cache, and release the namespace lock. * * vdev state is protected by spa_vdev_state_enter() / spa_vdev_state_exit(). * Like spa_vdev_enter/exit, these are convenience wrappers -- the actual * locking is, always, based on spa_namespace_lock and spa_config_lock[]. * * spa_rename() is also implemented within this file since it requires * manipulation of the namespace. */ static avl_tree_t spa_namespace_avl; kmutex_t spa_namespace_lock; static kcondvar_t spa_namespace_cv; static int spa_active_count; int spa_max_replication_override = SPA_DVAS_PER_BP; static kmutex_t spa_spare_lock; static avl_tree_t spa_spare_avl; static kmutex_t spa_l2cache_lock; static avl_tree_t spa_l2cache_avl; kmem_cache_t *spa_buffer_pool; int spa_mode_global; #ifdef ZFS_DEBUG /* Everything except dprintf and spa is on by default in debug builds */ int zfs_flags = ~(ZFS_DEBUG_DPRINTF | ZFS_DEBUG_SPA); #else int zfs_flags = 0; #endif /* * zfs_recover can be set to nonzero to attempt to recover from * otherwise-fatal errors, typically caused by on-disk corruption. When * set, calls to zfs_panic_recover() will turn into warning messages. * This should only be used as a last resort, as it typically results * in leaked space, or worse. */ int zfs_recover = B_FALSE; /* * If destroy encounters an EIO while reading metadata (e.g. indirect * blocks), space referenced by the missing metadata can not be freed. * Normally this causes the background destroy to become "stalled", as * it is unable to make forward progress. While in this stalled state, * all remaining space to free from the error-encountering filesystem is * "temporarily leaked". Set this flag to cause it to ignore the EIO, * permanently leak the space from indirect blocks that can not be read, * and continue to free everything else that it can. * * The default, "stalling" behavior is useful if the storage partially * fails (i.e. some but not all i/os fail), and then later recovers. In * this case, we will be able to continue pool operations while it is * partially failed, and when it recovers, we can continue to free the * space, with no leaks. However, note that this case is actually * fairly rare. * * Typically pools either (a) fail completely (but perhaps temporarily, * e.g. a top-level vdev going offline), or (b) have localized, * permanent errors (e.g. disk returns the wrong data due to bit flip or * firmware bug). In case (a), this setting does not matter because the * pool will be suspended and the sync thread will not be able to make * forward progress regardless. In case (b), because the error is * permanent, the best we can do is leak the minimum amount of space, * which is what setting this flag will do. Therefore, it is reasonable * for this flag to normally be set, but we chose the more conservative * approach of not setting it, so that there is no possibility of * leaking space in the "partial temporary" failure case. */ int zfs_free_leak_on_eio = B_FALSE; /* * Expiration time in milliseconds. This value has two meanings. First it is * used to determine when the spa_deadman() logic should fire. By default the * spa_deadman() will fire if spa_sync() has not completed in 1000 seconds. * Secondly, the value determines if an I/O is considered "hung". Any I/O that * has not completed in zfs_deadman_synctime_ms is considered "hung" resulting * in a system panic. */ unsigned long zfs_deadman_synctime_ms = 1000000ULL; /* * By default the deadman is enabled. */ int zfs_deadman_enabled = 1; /* * The worst case is single-sector max-parity RAID-Z blocks, in which * case the space requirement is exactly (VDEV_RAIDZ_MAXPARITY + 1) * times the size; so just assume that. Add to this the fact that * we can have up to 3 DVAs per bp, and one more factor of 2 because * the block may be dittoed with up to 3 DVAs by ddt_sync(). All together, * the worst case is: * (VDEV_RAIDZ_MAXPARITY + 1) * SPA_DVAS_PER_BP * 2 == 24 */ int spa_asize_inflation = 24; /* * ========================================================================== * SPA config locking * ========================================================================== */ static void spa_config_lock_init(spa_t *spa) { int i; for (i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; mutex_init(&scl->scl_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&scl->scl_cv, NULL, CV_DEFAULT, NULL); refcount_create_untracked(&scl->scl_count); scl->scl_writer = NULL; scl->scl_write_wanted = 0; } } static void spa_config_lock_destroy(spa_t *spa) { int i; for (i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; mutex_destroy(&scl->scl_lock); cv_destroy(&scl->scl_cv); refcount_destroy(&scl->scl_count); ASSERT(scl->scl_writer == NULL); ASSERT(scl->scl_write_wanted == 0); } } int spa_config_tryenter(spa_t *spa, int locks, void *tag, krw_t rw) { int i; for (i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; if (!(locks & (1 << i))) continue; mutex_enter(&scl->scl_lock); if (rw == RW_READER) { if (scl->scl_writer || scl->scl_write_wanted) { mutex_exit(&scl->scl_lock); spa_config_exit(spa, locks ^ (1 << i), tag); return (0); } } else { ASSERT(scl->scl_writer != curthread); if (!refcount_is_zero(&scl->scl_count)) { mutex_exit(&scl->scl_lock); spa_config_exit(spa, locks ^ (1 << i), tag); return (0); } scl->scl_writer = curthread; } (void) refcount_add(&scl->scl_count, tag); mutex_exit(&scl->scl_lock); } return (1); } void spa_config_enter(spa_t *spa, int locks, void *tag, krw_t rw) { int wlocks_held = 0; int i; ASSERT3U(SCL_LOCKS, <, sizeof (wlocks_held) * NBBY); for (i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; if (scl->scl_writer == curthread) wlocks_held |= (1 << i); if (!(locks & (1 << i))) continue; mutex_enter(&scl->scl_lock); if (rw == RW_READER) { while (scl->scl_writer || scl->scl_write_wanted) { cv_wait(&scl->scl_cv, &scl->scl_lock); } } else { ASSERT(scl->scl_writer != curthread); while (!refcount_is_zero(&scl->scl_count)) { scl->scl_write_wanted++; cv_wait(&scl->scl_cv, &scl->scl_lock); scl->scl_write_wanted--; } scl->scl_writer = curthread; } (void) refcount_add(&scl->scl_count, tag); mutex_exit(&scl->scl_lock); } ASSERT(wlocks_held <= locks); } void spa_config_exit(spa_t *spa, int locks, void *tag) { int i; for (i = SCL_LOCKS - 1; i >= 0; i--) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; if (!(locks & (1 << i))) continue; mutex_enter(&scl->scl_lock); ASSERT(!refcount_is_zero(&scl->scl_count)); if (refcount_remove(&scl->scl_count, tag) == 0) { ASSERT(scl->scl_writer == NULL || scl->scl_writer == curthread); scl->scl_writer = NULL; /* OK in either case */ cv_broadcast(&scl->scl_cv); } mutex_exit(&scl->scl_lock); } } int spa_config_held(spa_t *spa, int locks, krw_t rw) { int i, locks_held = 0; for (i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; if (!(locks & (1 << i))) continue; if ((rw == RW_READER && !refcount_is_zero(&scl->scl_count)) || (rw == RW_WRITER && scl->scl_writer == curthread)) locks_held |= 1 << i; } return (locks_held); } /* * ========================================================================== * SPA namespace functions * ========================================================================== */ /* * Lookup the named spa_t in the AVL tree. The spa_namespace_lock must be held. * Returns NULL if no matching spa_t is found. */ spa_t * spa_lookup(const char *name) { static spa_t search; /* spa_t is large; don't allocate on stack */ spa_t *spa; avl_index_t where; char *cp; ASSERT(MUTEX_HELD(&spa_namespace_lock)); (void) strlcpy(search.spa_name, name, sizeof (search.spa_name)); /* * If it's a full dataset name, figure out the pool name and * just use that. */ cp = strpbrk(search.spa_name, "/@#"); if (cp != NULL) *cp = '\0'; spa = avl_find(&spa_namespace_avl, &search, &where); return (spa); } /* * Fires when spa_sync has not completed within zfs_deadman_synctime_ms. * If the zfs_deadman_enabled flag is set then it inspects all vdev queues * looking for potentially hung I/Os. */ void spa_deadman(void *arg) { spa_t *spa = arg; zfs_dbgmsg("slow spa_sync: started %llu seconds ago, calls %llu", (gethrtime() - spa->spa_sync_starttime) / NANOSEC, ++spa->spa_deadman_calls); if (zfs_deadman_enabled) vdev_deadman(spa->spa_root_vdev); spa->spa_deadman_tqid = taskq_dispatch_delay(system_taskq, spa_deadman, spa, TQ_PUSHPAGE, ddi_get_lbolt() + NSEC_TO_TICK(spa->spa_deadman_synctime)); } /* * Create an uninitialized spa_t with the given name. Requires * spa_namespace_lock. The caller must ensure that the spa_t doesn't already * exist by calling spa_lookup() first. */ spa_t * spa_add(const char *name, nvlist_t *config, const char *altroot) { spa_t *spa; spa_config_dirent_t *dp; int t; int i; ASSERT(MUTEX_HELD(&spa_namespace_lock)); - spa = kmem_zalloc(sizeof (spa_t), KM_PUSHPAGE | KM_NODEBUG); + spa = kmem_zalloc(sizeof (spa_t), KM_PUSHPAGE); mutex_init(&spa->spa_async_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_errlist_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_errlog_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_history_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_proc_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_props_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_scrub_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_suspend_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_vdev_top_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&spa->spa_async_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_proc_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_scrub_io_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_suspend_cv, NULL, CV_DEFAULT, NULL); for (t = 0; t < TXG_SIZE; t++) bplist_create(&spa->spa_free_bplist[t]); (void) strlcpy(spa->spa_name, name, sizeof (spa->spa_name)); spa->spa_state = POOL_STATE_UNINITIALIZED; spa->spa_freeze_txg = UINT64_MAX; spa->spa_final_txg = UINT64_MAX; spa->spa_load_max_txg = UINT64_MAX; spa->spa_proc = &p0; spa->spa_proc_state = SPA_PROC_NONE; spa->spa_deadman_synctime = MSEC2NSEC(zfs_deadman_synctime_ms); refcount_create(&spa->spa_refcount); spa_config_lock_init(spa); spa_stats_init(spa); avl_add(&spa_namespace_avl, spa); /* * Set the alternate root, if there is one. */ if (altroot) { spa->spa_root = spa_strdup(altroot); spa_active_count++; } /* * Every pool starts with the default cachefile */ list_create(&spa->spa_config_list, sizeof (spa_config_dirent_t), offsetof(spa_config_dirent_t, scd_link)); dp = kmem_zalloc(sizeof (spa_config_dirent_t), KM_PUSHPAGE); dp->scd_path = altroot ? NULL : spa_strdup(spa_config_path); list_insert_head(&spa->spa_config_list, dp); VERIFY(nvlist_alloc(&spa->spa_load_info, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); if (config != NULL) { nvlist_t *features; if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_FEATURES_FOR_READ, &features) == 0) { VERIFY(nvlist_dup(features, &spa->spa_label_features, 0) == 0); } VERIFY(nvlist_dup(config, &spa->spa_config, 0) == 0); } if (spa->spa_label_features == NULL) { VERIFY(nvlist_alloc(&spa->spa_label_features, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); } spa->spa_debug = ((zfs_flags & ZFS_DEBUG_SPA) != 0); /* * As a pool is being created, treat all features as disabled by * setting SPA_FEATURE_DISABLED for all entries in the feature * refcount cache. */ for (i = 0; i < SPA_FEATURES; i++) { spa->spa_feat_refcount_cache[i] = SPA_FEATURE_DISABLED; } return (spa); } /* * Removes a spa_t from the namespace, freeing up any memory used. Requires * spa_namespace_lock. This is called only after the spa_t has been closed and * deactivated. */ void spa_remove(spa_t *spa) { spa_config_dirent_t *dp; int t; ASSERT(MUTEX_HELD(&spa_namespace_lock)); ASSERT(spa->spa_state == POOL_STATE_UNINITIALIZED); nvlist_free(spa->spa_config_splitting); avl_remove(&spa_namespace_avl, spa); cv_broadcast(&spa_namespace_cv); if (spa->spa_root) { spa_strfree(spa->spa_root); spa_active_count--; } while ((dp = list_head(&spa->spa_config_list)) != NULL) { list_remove(&spa->spa_config_list, dp); if (dp->scd_path != NULL) spa_strfree(dp->scd_path); kmem_free(dp, sizeof (spa_config_dirent_t)); } list_destroy(&spa->spa_config_list); nvlist_free(spa->spa_label_features); nvlist_free(spa->spa_load_info); spa_config_set(spa, NULL); refcount_destroy(&spa->spa_refcount); spa_stats_destroy(spa); spa_config_lock_destroy(spa); for (t = 0; t < TXG_SIZE; t++) bplist_destroy(&spa->spa_free_bplist[t]); cv_destroy(&spa->spa_async_cv); cv_destroy(&spa->spa_proc_cv); cv_destroy(&spa->spa_scrub_io_cv); cv_destroy(&spa->spa_suspend_cv); mutex_destroy(&spa->spa_async_lock); mutex_destroy(&spa->spa_errlist_lock); mutex_destroy(&spa->spa_errlog_lock); mutex_destroy(&spa->spa_history_lock); mutex_destroy(&spa->spa_proc_lock); mutex_destroy(&spa->spa_props_lock); mutex_destroy(&spa->spa_scrub_lock); mutex_destroy(&spa->spa_suspend_lock); mutex_destroy(&spa->spa_vdev_top_lock); kmem_free(spa, sizeof (spa_t)); } /* * Given a pool, return the next pool in the namespace, or NULL if there is * none. If 'prev' is NULL, return the first pool. */ spa_t * spa_next(spa_t *prev) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); if (prev) return (AVL_NEXT(&spa_namespace_avl, prev)); else return (avl_first(&spa_namespace_avl)); } /* * ========================================================================== * SPA refcount functions * ========================================================================== */ /* * Add a reference to the given spa_t. Must have at least one reference, or * have the namespace lock held. */ void spa_open_ref(spa_t *spa, void *tag) { ASSERT(refcount_count(&spa->spa_refcount) >= spa->spa_minref || MUTEX_HELD(&spa_namespace_lock)); (void) refcount_add(&spa->spa_refcount, tag); } /* * Remove a reference to the given spa_t. Must have at least one reference, or * have the namespace lock held. */ void spa_close(spa_t *spa, void *tag) { ASSERT(refcount_count(&spa->spa_refcount) > spa->spa_minref || MUTEX_HELD(&spa_namespace_lock)); (void) refcount_remove(&spa->spa_refcount, tag); } /* * Check to see if the spa refcount is zero. Must be called with * spa_namespace_lock held. We really compare against spa_minref, which is the * number of references acquired when opening a pool */ boolean_t spa_refcount_zero(spa_t *spa) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); return (refcount_count(&spa->spa_refcount) == spa->spa_minref); } /* * ========================================================================== * SPA spare and l2cache tracking * ========================================================================== */ /* * Hot spares and cache devices are tracked using the same code below, * for 'auxiliary' devices. */ typedef struct spa_aux { uint64_t aux_guid; uint64_t aux_pool; avl_node_t aux_avl; int aux_count; } spa_aux_t; static int spa_aux_compare(const void *a, const void *b) { const spa_aux_t *sa = a; const spa_aux_t *sb = b; if (sa->aux_guid < sb->aux_guid) return (-1); else if (sa->aux_guid > sb->aux_guid) return (1); else return (0); } void spa_aux_add(vdev_t *vd, avl_tree_t *avl) { avl_index_t where; spa_aux_t search; spa_aux_t *aux; search.aux_guid = vd->vdev_guid; if ((aux = avl_find(avl, &search, &where)) != NULL) { aux->aux_count++; } else { aux = kmem_zalloc(sizeof (spa_aux_t), KM_PUSHPAGE); aux->aux_guid = vd->vdev_guid; aux->aux_count = 1; avl_insert(avl, aux, where); } } void spa_aux_remove(vdev_t *vd, avl_tree_t *avl) { spa_aux_t search; spa_aux_t *aux; avl_index_t where; search.aux_guid = vd->vdev_guid; aux = avl_find(avl, &search, &where); ASSERT(aux != NULL); if (--aux->aux_count == 0) { avl_remove(avl, aux); kmem_free(aux, sizeof (spa_aux_t)); } else if (aux->aux_pool == spa_guid(vd->vdev_spa)) { aux->aux_pool = 0ULL; } } boolean_t spa_aux_exists(uint64_t guid, uint64_t *pool, int *refcnt, avl_tree_t *avl) { spa_aux_t search, *found; search.aux_guid = guid; found = avl_find(avl, &search, NULL); if (pool) { if (found) *pool = found->aux_pool; else *pool = 0ULL; } if (refcnt) { if (found) *refcnt = found->aux_count; else *refcnt = 0; } return (found != NULL); } void spa_aux_activate(vdev_t *vd, avl_tree_t *avl) { spa_aux_t search, *found; avl_index_t where; search.aux_guid = vd->vdev_guid; found = avl_find(avl, &search, &where); ASSERT(found != NULL); ASSERT(found->aux_pool == 0ULL); found->aux_pool = spa_guid(vd->vdev_spa); } /* * Spares are tracked globally due to the following constraints: * * - A spare may be part of multiple pools. * - A spare may be added to a pool even if it's actively in use within * another pool. * - A spare in use in any pool can only be the source of a replacement if * the target is a spare in the same pool. * * We keep track of all spares on the system through the use of a reference * counted AVL tree. When a vdev is added as a spare, or used as a replacement * spare, then we bump the reference count in the AVL tree. In addition, we set * the 'vdev_isspare' member to indicate that the device is a spare (active or * inactive). When a spare is made active (used to replace a device in the * pool), we also keep track of which pool its been made a part of. * * The 'spa_spare_lock' protects the AVL tree. These functions are normally * called under the spa_namespace lock as part of vdev reconfiguration. The * separate spare lock exists for the status query path, which does not need to * be completely consistent with respect to other vdev configuration changes. */ static int spa_spare_compare(const void *a, const void *b) { return (spa_aux_compare(a, b)); } void spa_spare_add(vdev_t *vd) { mutex_enter(&spa_spare_lock); ASSERT(!vd->vdev_isspare); spa_aux_add(vd, &spa_spare_avl); vd->vdev_isspare = B_TRUE; mutex_exit(&spa_spare_lock); } void spa_spare_remove(vdev_t *vd) { mutex_enter(&spa_spare_lock); ASSERT(vd->vdev_isspare); spa_aux_remove(vd, &spa_spare_avl); vd->vdev_isspare = B_FALSE; mutex_exit(&spa_spare_lock); } boolean_t spa_spare_exists(uint64_t guid, uint64_t *pool, int *refcnt) { boolean_t found; mutex_enter(&spa_spare_lock); found = spa_aux_exists(guid, pool, refcnt, &spa_spare_avl); mutex_exit(&spa_spare_lock); return (found); } void spa_spare_activate(vdev_t *vd) { mutex_enter(&spa_spare_lock); ASSERT(vd->vdev_isspare); spa_aux_activate(vd, &spa_spare_avl); mutex_exit(&spa_spare_lock); } /* * Level 2 ARC devices are tracked globally for the same reasons as spares. * Cache devices currently only support one pool per cache device, and so * for these devices the aux reference count is currently unused beyond 1. */ static int spa_l2cache_compare(const void *a, const void *b) { return (spa_aux_compare(a, b)); } void spa_l2cache_add(vdev_t *vd) { mutex_enter(&spa_l2cache_lock); ASSERT(!vd->vdev_isl2cache); spa_aux_add(vd, &spa_l2cache_avl); vd->vdev_isl2cache = B_TRUE; mutex_exit(&spa_l2cache_lock); } void spa_l2cache_remove(vdev_t *vd) { mutex_enter(&spa_l2cache_lock); ASSERT(vd->vdev_isl2cache); spa_aux_remove(vd, &spa_l2cache_avl); vd->vdev_isl2cache = B_FALSE; mutex_exit(&spa_l2cache_lock); } boolean_t spa_l2cache_exists(uint64_t guid, uint64_t *pool) { boolean_t found; mutex_enter(&spa_l2cache_lock); found = spa_aux_exists(guid, pool, NULL, &spa_l2cache_avl); mutex_exit(&spa_l2cache_lock); return (found); } void spa_l2cache_activate(vdev_t *vd) { mutex_enter(&spa_l2cache_lock); ASSERT(vd->vdev_isl2cache); spa_aux_activate(vd, &spa_l2cache_avl); mutex_exit(&spa_l2cache_lock); } /* * ========================================================================== * SPA vdev locking * ========================================================================== */ /* * Lock the given spa_t for the purpose of adding or removing a vdev. * Grabs the global spa_namespace_lock plus the spa config lock for writing. * It returns the next transaction group for the spa_t. */ uint64_t spa_vdev_enter(spa_t *spa) { mutex_enter(&spa->spa_vdev_top_lock); mutex_enter(&spa_namespace_lock); return (spa_vdev_config_enter(spa)); } /* * Internal implementation for spa_vdev_enter(). Used when a vdev * operation requires multiple syncs (i.e. removing a device) while * keeping the spa_namespace_lock held. */ uint64_t spa_vdev_config_enter(spa_t *spa) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); spa_config_enter(spa, SCL_ALL, spa, RW_WRITER); return (spa_last_synced_txg(spa) + 1); } /* * Used in combination with spa_vdev_config_enter() to allow the syncing * of multiple transactions without releasing the spa_namespace_lock. */ void spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error, char *tag) { int config_changed = B_FALSE; ASSERT(MUTEX_HELD(&spa_namespace_lock)); ASSERT(txg > spa_last_synced_txg(spa)); spa->spa_pending_vdev = NULL; /* * Reassess the DTLs. */ vdev_dtl_reassess(spa->spa_root_vdev, 0, 0, B_FALSE); if (error == 0 && !list_is_empty(&spa->spa_config_dirty_list)) { config_changed = B_TRUE; spa->spa_config_generation++; } /* * Verify the metaslab classes. */ ASSERT(metaslab_class_validate(spa_normal_class(spa)) == 0); ASSERT(metaslab_class_validate(spa_log_class(spa)) == 0); spa_config_exit(spa, SCL_ALL, spa); /* * Panic the system if the specified tag requires it. This * is useful for ensuring that configurations are updated * transactionally. */ if (zio_injection_enabled) zio_handle_panic_injection(spa, tag, 0); /* * Note: this txg_wait_synced() is important because it ensures * that there won't be more than one config change per txg. * This allows us to use the txg as the generation number. */ if (error == 0) txg_wait_synced(spa->spa_dsl_pool, txg); if (vd != NULL) { ASSERT(!vd->vdev_detached || vd->vdev_dtl_sm == NULL); spa_config_enter(spa, SCL_ALL, spa, RW_WRITER); vdev_free(vd); spa_config_exit(spa, SCL_ALL, spa); } /* * If the config changed, update the config cache. */ if (config_changed) spa_config_sync(spa, B_FALSE, B_TRUE); } /* * Unlock the spa_t after adding or removing a vdev. Besides undoing the * locking of spa_vdev_enter(), we also want make sure the transactions have * synced to disk, and then update the global configuration cache with the new * information. */ int spa_vdev_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error) { spa_vdev_config_exit(spa, vd, txg, error, FTAG); mutex_exit(&spa_namespace_lock); mutex_exit(&spa->spa_vdev_top_lock); return (error); } /* * Lock the given spa_t for the purpose of changing vdev state. */ void spa_vdev_state_enter(spa_t *spa, int oplocks) { int locks = SCL_STATE_ALL | oplocks; /* * Root pools may need to read of the underlying devfs filesystem * when opening up a vdev. Unfortunately if we're holding the * SCL_ZIO lock it will result in a deadlock when we try to issue * the read from the root filesystem. Instead we "prefetch" * the associated vnodes that we need prior to opening the * underlying devices and cache them so that we can prevent * any I/O when we are doing the actual open. */ if (spa_is_root(spa)) { int low = locks & ~(SCL_ZIO - 1); int high = locks & ~low; spa_config_enter(spa, high, spa, RW_WRITER); vdev_hold(spa->spa_root_vdev); spa_config_enter(spa, low, spa, RW_WRITER); } else { spa_config_enter(spa, locks, spa, RW_WRITER); } spa->spa_vdev_locks = locks; } int spa_vdev_state_exit(spa_t *spa, vdev_t *vd, int error) { boolean_t config_changed = B_FALSE; if (vd != NULL || error == 0) vdev_dtl_reassess(vd ? vd->vdev_top : spa->spa_root_vdev, 0, 0, B_FALSE); if (vd != NULL) { vdev_state_dirty(vd->vdev_top); config_changed = B_TRUE; spa->spa_config_generation++; } if (spa_is_root(spa)) vdev_rele(spa->spa_root_vdev); ASSERT3U(spa->spa_vdev_locks, >=, SCL_STATE_ALL); spa_config_exit(spa, spa->spa_vdev_locks, spa); /* * If anything changed, wait for it to sync. This ensures that, * from the system administrator's perspective, zpool(1M) commands * are synchronous. This is important for things like zpool offline: * when the command completes, you expect no further I/O from ZFS. */ if (vd != NULL) txg_wait_synced(spa->spa_dsl_pool, 0); /* * If the config changed, update the config cache. */ if (config_changed) { mutex_enter(&spa_namespace_lock); spa_config_sync(spa, B_FALSE, B_TRUE); mutex_exit(&spa_namespace_lock); } return (error); } /* * ========================================================================== * Miscellaneous functions * ========================================================================== */ void spa_activate_mos_feature(spa_t *spa, const char *feature, dmu_tx_t *tx) { if (!nvlist_exists(spa->spa_label_features, feature)) { fnvlist_add_boolean(spa->spa_label_features, feature); /* * When we are creating the pool (tx_txg==TXG_INITIAL), we can't * dirty the vdev config because lock SCL_CONFIG is not held. * Thankfully, in this case we don't need to dirty the config * because it will be written out anyway when we finish * creating the pool. */ if (tx->tx_txg != TXG_INITIAL) vdev_config_dirty(spa->spa_root_vdev); } } void spa_deactivate_mos_feature(spa_t *spa, const char *feature) { if (nvlist_remove_all(spa->spa_label_features, feature) == 0) vdev_config_dirty(spa->spa_root_vdev); } /* * Rename a spa_t. */ int spa_rename(const char *name, const char *newname) { spa_t *spa; int err; /* * Lookup the spa_t and grab the config lock for writing. We need to * actually open the pool so that we can sync out the necessary labels. * It's OK to call spa_open() with the namespace lock held because we * allow recursive calls for other reasons. */ mutex_enter(&spa_namespace_lock); if ((err = spa_open(name, &spa, FTAG)) != 0) { mutex_exit(&spa_namespace_lock); return (err); } spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); avl_remove(&spa_namespace_avl, spa); (void) strlcpy(spa->spa_name, newname, sizeof (spa->spa_name)); avl_add(&spa_namespace_avl, spa); /* * Sync all labels to disk with the new names by marking the root vdev * dirty and waiting for it to sync. It will pick up the new pool name * during the sync. */ vdev_config_dirty(spa->spa_root_vdev); spa_config_exit(spa, SCL_ALL, FTAG); txg_wait_synced(spa->spa_dsl_pool, 0); /* * Sync the updated config cache. */ spa_config_sync(spa, B_FALSE, B_TRUE); spa_close(spa, FTAG); mutex_exit(&spa_namespace_lock); return (0); } /* * Return the spa_t associated with given pool_guid, if it exists. If * device_guid is non-zero, determine whether the pool exists *and* contains * a device with the specified device_guid. */ spa_t * spa_by_guid(uint64_t pool_guid, uint64_t device_guid) { spa_t *spa; avl_tree_t *t = &spa_namespace_avl; ASSERT(MUTEX_HELD(&spa_namespace_lock)); for (spa = avl_first(t); spa != NULL; spa = AVL_NEXT(t, spa)) { if (spa->spa_state == POOL_STATE_UNINITIALIZED) continue; if (spa->spa_root_vdev == NULL) continue; if (spa_guid(spa) == pool_guid) { if (device_guid == 0) break; if (vdev_lookup_by_guid(spa->spa_root_vdev, device_guid) != NULL) break; /* * Check any devices we may be in the process of adding. */ if (spa->spa_pending_vdev) { if (vdev_lookup_by_guid(spa->spa_pending_vdev, device_guid) != NULL) break; } } } return (spa); } /* * Determine whether a pool with the given pool_guid exists. */ boolean_t spa_guid_exists(uint64_t pool_guid, uint64_t device_guid) { return (spa_by_guid(pool_guid, device_guid) != NULL); } char * spa_strdup(const char *s) { size_t len; char *new; len = strlen(s); new = kmem_alloc(len + 1, KM_PUSHPAGE); bcopy(s, new, len); new[len] = '\0'; return (new); } void spa_strfree(char *s) { kmem_free(s, strlen(s) + 1); } uint64_t spa_get_random(uint64_t range) { uint64_t r; ASSERT(range != 0); (void) random_get_pseudo_bytes((void *)&r, sizeof (uint64_t)); return (r % range); } uint64_t spa_generate_guid(spa_t *spa) { uint64_t guid = spa_get_random(-1ULL); if (spa != NULL) { while (guid == 0 || spa_guid_exists(spa_guid(spa), guid)) guid = spa_get_random(-1ULL); } else { while (guid == 0 || spa_guid_exists(guid, 0)) guid = spa_get_random(-1ULL); } return (guid); } void snprintf_blkptr(char *buf, size_t buflen, const blkptr_t *bp) { char type[256]; char *checksum = NULL; char *compress = NULL; if (bp != NULL) { if (BP_GET_TYPE(bp) & DMU_OT_NEWTYPE) { dmu_object_byteswap_t bswap = DMU_OT_BYTESWAP(BP_GET_TYPE(bp)); (void) snprintf(type, sizeof (type), "bswap %s %s", DMU_OT_IS_METADATA(BP_GET_TYPE(bp)) ? "metadata" : "data", dmu_ot_byteswap[bswap].ob_name); } else { (void) strlcpy(type, dmu_ot[BP_GET_TYPE(bp)].ot_name, sizeof (type)); } if (!BP_IS_EMBEDDED(bp)) { checksum = zio_checksum_table[BP_GET_CHECKSUM(bp)].ci_name; } compress = zio_compress_table[BP_GET_COMPRESS(bp)].ci_name; } SNPRINTF_BLKPTR(snprintf, ' ', buf, buflen, bp, type, checksum, compress); } void spa_freeze(spa_t *spa) { uint64_t freeze_txg = 0; spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); if (spa->spa_freeze_txg == UINT64_MAX) { freeze_txg = spa_last_synced_txg(spa) + TXG_SIZE; spa->spa_freeze_txg = freeze_txg; } spa_config_exit(spa, SCL_ALL, FTAG); if (freeze_txg != 0) txg_wait_synced(spa_get_dsl(spa), freeze_txg); } void zfs_panic_recover(const char *fmt, ...) { va_list adx; va_start(adx, fmt); vcmn_err(zfs_recover ? CE_WARN : CE_PANIC, fmt, adx); va_end(adx); } /* * This is a stripped-down version of strtoull, suitable only for converting * lowercase hexadecimal numbers that don't overflow. */ uint64_t strtonum(const char *str, char **nptr) { uint64_t val = 0; char c; int digit; while ((c = *str) != '\0') { if (c >= '0' && c <= '9') digit = c - '0'; else if (c >= 'a' && c <= 'f') digit = 10 + c - 'a'; else break; val *= 16; val += digit; str++; } if (nptr) *nptr = (char *)str; return (val); } /* * ========================================================================== * Accessor functions * ========================================================================== */ boolean_t spa_shutting_down(spa_t *spa) { return (spa->spa_async_suspended); } dsl_pool_t * spa_get_dsl(spa_t *spa) { return (spa->spa_dsl_pool); } boolean_t spa_is_initializing(spa_t *spa) { return (spa->spa_is_initializing); } blkptr_t * spa_get_rootblkptr(spa_t *spa) { return (&spa->spa_ubsync.ub_rootbp); } void spa_set_rootblkptr(spa_t *spa, const blkptr_t *bp) { spa->spa_uberblock.ub_rootbp = *bp; } void spa_altroot(spa_t *spa, char *buf, size_t buflen) { if (spa->spa_root == NULL) buf[0] = '\0'; else (void) strncpy(buf, spa->spa_root, buflen); } int spa_sync_pass(spa_t *spa) { return (spa->spa_sync_pass); } char * spa_name(spa_t *spa) { return (spa->spa_name); } uint64_t spa_guid(spa_t *spa) { dsl_pool_t *dp = spa_get_dsl(spa); uint64_t guid; /* * If we fail to parse the config during spa_load(), we can go through * the error path (which posts an ereport) and end up here with no root * vdev. We stash the original pool guid in 'spa_config_guid' to handle * this case. */ if (spa->spa_root_vdev == NULL) return (spa->spa_config_guid); guid = spa->spa_last_synced_guid != 0 ? spa->spa_last_synced_guid : spa->spa_root_vdev->vdev_guid; /* * Return the most recently synced out guid unless we're * in syncing context. */ if (dp && dsl_pool_sync_context(dp)) return (spa->spa_root_vdev->vdev_guid); else return (guid); } uint64_t spa_load_guid(spa_t *spa) { /* * This is a GUID that exists solely as a reference for the * purposes of the arc. It is generated at load time, and * is never written to persistent storage. */ return (spa->spa_load_guid); } uint64_t spa_last_synced_txg(spa_t *spa) { return (spa->spa_ubsync.ub_txg); } uint64_t spa_first_txg(spa_t *spa) { return (spa->spa_first_txg); } uint64_t spa_syncing_txg(spa_t *spa) { return (spa->spa_syncing_txg); } pool_state_t spa_state(spa_t *spa) { return (spa->spa_state); } spa_load_state_t spa_load_state(spa_t *spa) { return (spa->spa_load_state); } uint64_t spa_freeze_txg(spa_t *spa) { return (spa->spa_freeze_txg); } /* ARGSUSED */ uint64_t spa_get_asize(spa_t *spa, uint64_t lsize) { return (lsize * spa_asize_inflation); } uint64_t spa_get_dspace(spa_t *spa) { return (spa->spa_dspace); } void spa_update_dspace(spa_t *spa) { spa->spa_dspace = metaslab_class_get_dspace(spa_normal_class(spa)) + ddt_get_dedup_dspace(spa); } /* * Return the failure mode that has been set to this pool. The default * behavior will be to block all I/Os when a complete failure occurs. */ uint8_t spa_get_failmode(spa_t *spa) { return (spa->spa_failmode); } boolean_t spa_suspended(spa_t *spa) { return (spa->spa_suspended); } uint64_t spa_version(spa_t *spa) { return (spa->spa_ubsync.ub_version); } boolean_t spa_deflate(spa_t *spa) { return (spa->spa_deflate); } metaslab_class_t * spa_normal_class(spa_t *spa) { return (spa->spa_normal_class); } metaslab_class_t * spa_log_class(spa_t *spa) { return (spa->spa_log_class); } int spa_max_replication(spa_t *spa) { /* * As of SPA_VERSION == SPA_VERSION_DITTO_BLOCKS, we are able to * handle BPs with more than one DVA allocated. Set our max * replication level accordingly. */ if (spa_version(spa) < SPA_VERSION_DITTO_BLOCKS) return (1); return (MIN(SPA_DVAS_PER_BP, spa_max_replication_override)); } int spa_prev_software_version(spa_t *spa) { return (spa->spa_prev_software_version); } uint64_t spa_deadman_synctime(spa_t *spa) { return (spa->spa_deadman_synctime); } uint64_t dva_get_dsize_sync(spa_t *spa, const dva_t *dva) { uint64_t asize = DVA_GET_ASIZE(dva); uint64_t dsize = asize; ASSERT(spa_config_held(spa, SCL_ALL, RW_READER) != 0); if (asize != 0 && spa->spa_deflate) { vdev_t *vd = vdev_lookup_top(spa, DVA_GET_VDEV(dva)); if (vd != NULL) dsize = (asize >> SPA_MINBLOCKSHIFT) * vd->vdev_deflate_ratio; } return (dsize); } uint64_t bp_get_dsize_sync(spa_t *spa, const blkptr_t *bp) { uint64_t dsize = 0; int d; for (d = 0; d < BP_GET_NDVAS(bp); d++) dsize += dva_get_dsize_sync(spa, &bp->blk_dva[d]); return (dsize); } uint64_t bp_get_dsize(spa_t *spa, const blkptr_t *bp) { uint64_t dsize = 0; int d; spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); for (d = 0; d < BP_GET_NDVAS(bp); d++) dsize += dva_get_dsize_sync(spa, &bp->blk_dva[d]); spa_config_exit(spa, SCL_VDEV, FTAG); return (dsize); } /* * ========================================================================== * Initialization and Termination * ========================================================================== */ static int spa_name_compare(const void *a1, const void *a2) { const spa_t *s1 = a1; const spa_t *s2 = a2; int s; s = strcmp(s1->spa_name, s2->spa_name); if (s > 0) return (1); if (s < 0) return (-1); return (0); } void spa_boot_init(void) { spa_config_load(); } void spa_init(int mode) { mutex_init(&spa_namespace_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa_spare_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa_l2cache_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&spa_namespace_cv, NULL, CV_DEFAULT, NULL); avl_create(&spa_namespace_avl, spa_name_compare, sizeof (spa_t), offsetof(spa_t, spa_avl)); avl_create(&spa_spare_avl, spa_spare_compare, sizeof (spa_aux_t), offsetof(spa_aux_t, aux_avl)); avl_create(&spa_l2cache_avl, spa_l2cache_compare, sizeof (spa_aux_t), offsetof(spa_aux_t, aux_avl)); spa_mode_global = mode; #ifndef _KERNEL if (spa_mode_global != FREAD && dprintf_find_string("watch")) { struct sigaction sa; sa.sa_flags = SA_SIGINFO; sigemptyset(&sa.sa_mask); sa.sa_sigaction = arc_buf_sigsegv; if (sigaction(SIGSEGV, &sa, NULL) == -1) { perror("could not enable watchpoints: " "sigaction(SIGSEGV, ...) = "); } else { arc_watch = B_TRUE; } } #endif fm_init(); refcount_init(); unique_init(); range_tree_init(); ddt_init(); zio_init(); dmu_init(); zil_init(); vdev_cache_stat_init(); vdev_file_init(); zfs_prop_init(); zpool_prop_init(); zpool_feature_init(); spa_config_load(); l2arc_start(); } void spa_fini(void) { l2arc_stop(); spa_evict_all(); vdev_file_fini(); vdev_cache_stat_fini(); zil_fini(); dmu_fini(); zio_fini(); ddt_fini(); range_tree_fini(); unique_fini(); refcount_fini(); fm_fini(); avl_destroy(&spa_namespace_avl); avl_destroy(&spa_spare_avl); avl_destroy(&spa_l2cache_avl); cv_destroy(&spa_namespace_cv); mutex_destroy(&spa_namespace_lock); mutex_destroy(&spa_spare_lock); mutex_destroy(&spa_l2cache_lock); } /* * Return whether this pool has slogs. No locking needed. * It's not a problem if the wrong answer is returned as it's only for * performance and not correctness */ boolean_t spa_has_slogs(spa_t *spa) { return (spa->spa_log_class->mc_rotor != NULL); } spa_log_state_t spa_get_log_state(spa_t *spa) { return (spa->spa_log_state); } void spa_set_log_state(spa_t *spa, spa_log_state_t state) { spa->spa_log_state = state; } boolean_t spa_is_root(spa_t *spa) { return (spa->spa_is_root); } boolean_t spa_writeable(spa_t *spa) { return (!!(spa->spa_mode & FWRITE)); } /* * Returns true if there is a pending sync task in any of the current * syncing txg, the current quiescing txg, or the current open txg. */ boolean_t spa_has_pending_synctask(spa_t *spa) { return (!txg_all_lists_empty(&spa->spa_dsl_pool->dp_sync_tasks)); } int spa_mode(spa_t *spa) { return (spa->spa_mode); } uint64_t spa_bootfs(spa_t *spa) { return (spa->spa_bootfs); } uint64_t spa_delegation(spa_t *spa) { return (spa->spa_delegation); } objset_t * spa_meta_objset(spa_t *spa) { return (spa->spa_meta_objset); } enum zio_checksum spa_dedup_checksum(spa_t *spa) { return (spa->spa_dedup_checksum); } /* * Reset pool scan stat per scan pass (or reboot). */ void spa_scan_stat_init(spa_t *spa) { /* data not stored on disk */ spa->spa_scan_pass_start = gethrestime_sec(); spa->spa_scan_pass_exam = 0; vdev_scan_stat_init(spa->spa_root_vdev); } /* * Get scan stats for zpool status reports */ int spa_scan_get_stats(spa_t *spa, pool_scan_stat_t *ps) { dsl_scan_t *scn = spa->spa_dsl_pool ? spa->spa_dsl_pool->dp_scan : NULL; if (scn == NULL || scn->scn_phys.scn_func == POOL_SCAN_NONE) return (SET_ERROR(ENOENT)); bzero(ps, sizeof (pool_scan_stat_t)); /* data stored on disk */ ps->pss_func = scn->scn_phys.scn_func; ps->pss_start_time = scn->scn_phys.scn_start_time; ps->pss_end_time = scn->scn_phys.scn_end_time; ps->pss_to_examine = scn->scn_phys.scn_to_examine; ps->pss_examined = scn->scn_phys.scn_examined; ps->pss_to_process = scn->scn_phys.scn_to_process; ps->pss_processed = scn->scn_phys.scn_processed; ps->pss_errors = scn->scn_phys.scn_errors; ps->pss_state = scn->scn_phys.scn_state; /* data not stored on disk */ ps->pss_pass_start = spa->spa_scan_pass_start; ps->pss_pass_exam = spa->spa_scan_pass_exam; return (0); } boolean_t spa_debug_enabled(spa_t *spa) { return (spa->spa_debug); } #if defined(_KERNEL) && defined(HAVE_SPL) /* Namespace manipulation */ EXPORT_SYMBOL(spa_lookup); EXPORT_SYMBOL(spa_add); EXPORT_SYMBOL(spa_remove); EXPORT_SYMBOL(spa_next); /* Refcount functions */ EXPORT_SYMBOL(spa_open_ref); EXPORT_SYMBOL(spa_close); EXPORT_SYMBOL(spa_refcount_zero); /* Pool configuration lock */ EXPORT_SYMBOL(spa_config_tryenter); EXPORT_SYMBOL(spa_config_enter); EXPORT_SYMBOL(spa_config_exit); EXPORT_SYMBOL(spa_config_held); /* Pool vdev add/remove lock */ EXPORT_SYMBOL(spa_vdev_enter); EXPORT_SYMBOL(spa_vdev_exit); /* Pool vdev state change lock */ EXPORT_SYMBOL(spa_vdev_state_enter); EXPORT_SYMBOL(spa_vdev_state_exit); /* Accessor functions */ EXPORT_SYMBOL(spa_shutting_down); EXPORT_SYMBOL(spa_get_dsl); EXPORT_SYMBOL(spa_get_rootblkptr); EXPORT_SYMBOL(spa_set_rootblkptr); EXPORT_SYMBOL(spa_altroot); EXPORT_SYMBOL(spa_sync_pass); EXPORT_SYMBOL(spa_name); EXPORT_SYMBOL(spa_guid); EXPORT_SYMBOL(spa_last_synced_txg); EXPORT_SYMBOL(spa_first_txg); EXPORT_SYMBOL(spa_syncing_txg); EXPORT_SYMBOL(spa_version); EXPORT_SYMBOL(spa_state); EXPORT_SYMBOL(spa_load_state); EXPORT_SYMBOL(spa_freeze_txg); EXPORT_SYMBOL(spa_get_asize); EXPORT_SYMBOL(spa_get_dspace); EXPORT_SYMBOL(spa_update_dspace); EXPORT_SYMBOL(spa_deflate); EXPORT_SYMBOL(spa_normal_class); EXPORT_SYMBOL(spa_log_class); EXPORT_SYMBOL(spa_max_replication); EXPORT_SYMBOL(spa_prev_software_version); EXPORT_SYMBOL(spa_get_failmode); EXPORT_SYMBOL(spa_suspended); EXPORT_SYMBOL(spa_bootfs); EXPORT_SYMBOL(spa_delegation); EXPORT_SYMBOL(spa_meta_objset); /* Miscellaneous support routines */ EXPORT_SYMBOL(spa_rename); EXPORT_SYMBOL(spa_guid_exists); EXPORT_SYMBOL(spa_strdup); EXPORT_SYMBOL(spa_strfree); EXPORT_SYMBOL(spa_get_random); EXPORT_SYMBOL(spa_generate_guid); EXPORT_SYMBOL(snprintf_blkptr); EXPORT_SYMBOL(spa_freeze); EXPORT_SYMBOL(spa_upgrade); EXPORT_SYMBOL(spa_evict_all); EXPORT_SYMBOL(spa_lookup_by_guid); EXPORT_SYMBOL(spa_has_spare); EXPORT_SYMBOL(dva_get_dsize_sync); EXPORT_SYMBOL(bp_get_dsize_sync); EXPORT_SYMBOL(bp_get_dsize); EXPORT_SYMBOL(spa_has_slogs); EXPORT_SYMBOL(spa_is_root); EXPORT_SYMBOL(spa_writeable); EXPORT_SYMBOL(spa_mode); EXPORT_SYMBOL(spa_namespace_lock); module_param(zfs_flags, uint, 0644); MODULE_PARM_DESC(zfs_flags, "Set additional debugging flags"); module_param(zfs_recover, int, 0644); MODULE_PARM_DESC(zfs_recover, "Set to attempt to recover from fatal errors"); module_param(zfs_free_leak_on_eio, int, 0644); MODULE_PARM_DESC(zfs_free_leak_on_eio, "Set to ignore IO errors during free and permanently leak the space"); module_param(zfs_deadman_synctime_ms, ulong, 0644); MODULE_PARM_DESC(zfs_deadman_synctime_ms, "Expiration time in milliseconds"); module_param(zfs_deadman_enabled, int, 0644); MODULE_PARM_DESC(zfs_deadman_enabled, "Enable deadman timer"); module_param(spa_asize_inflation, int, 0644); MODULE_PARM_DESC(spa_asize_inflation, "SPA size estimate multiplication factor"); #endif diff --git a/module/zfs/vdev.c b/module/zfs/vdev.c index 37ce49297390..18a246c333db 100644 --- a/module/zfs/vdev.c +++ b/module/zfs/vdev.c @@ -1,3405 +1,3405 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2011, 2014 by Delphix. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * When a vdev is added, it will be divided into approximately (but no * more than) this number of metaslabs. */ int metaslabs_per_vdev = 200; /* * Virtual device management. */ static vdev_ops_t *vdev_ops_table[] = { &vdev_root_ops, &vdev_raidz_ops, &vdev_mirror_ops, &vdev_replacing_ops, &vdev_spare_ops, &vdev_disk_ops, &vdev_file_ops, &vdev_missing_ops, &vdev_hole_ops, NULL }; /* * Given a vdev type, return the appropriate ops vector. */ static vdev_ops_t * vdev_getops(const char *type) { vdev_ops_t *ops, **opspp; for (opspp = vdev_ops_table; (ops = *opspp) != NULL; opspp++) if (strcmp(ops->vdev_op_type, type) == 0) break; return (ops); } /* * Default asize function: return the MAX of psize with the asize of * all children. This is what's used by anything other than RAID-Z. */ uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize) { uint64_t asize = P2ROUNDUP(psize, 1ULL << vd->vdev_top->vdev_ashift); uint64_t csize; int c; for (c = 0; c < vd->vdev_children; c++) { csize = vdev_psize_to_asize(vd->vdev_child[c], psize); asize = MAX(asize, csize); } return (asize); } /* * Get the minimum allocatable size. We define the allocatable size as * the vdev's asize rounded to the nearest metaslab. This allows us to * replace or attach devices which don't have the same physical size but * can still satisfy the same number of allocations. */ uint64_t vdev_get_min_asize(vdev_t *vd) { vdev_t *pvd = vd->vdev_parent; /* * If our parent is NULL (inactive spare or cache) or is the root, * just return our own asize. */ if (pvd == NULL) return (vd->vdev_asize); /* * The top-level vdev just returns the allocatable size rounded * to the nearest metaslab. */ if (vd == vd->vdev_top) return (P2ALIGN(vd->vdev_asize, 1ULL << vd->vdev_ms_shift)); /* * The allocatable space for a raidz vdev is N * sizeof(smallest child), * so each child must provide at least 1/Nth of its asize. */ if (pvd->vdev_ops == &vdev_raidz_ops) return (pvd->vdev_min_asize / pvd->vdev_children); return (pvd->vdev_min_asize); } void vdev_set_min_asize(vdev_t *vd) { int c; vd->vdev_min_asize = vdev_get_min_asize(vd); for (c = 0; c < vd->vdev_children; c++) vdev_set_min_asize(vd->vdev_child[c]); } vdev_t * vdev_lookup_top(spa_t *spa, uint64_t vdev) { vdev_t *rvd = spa->spa_root_vdev; ASSERT(spa_config_held(spa, SCL_ALL, RW_READER) != 0); if (vdev < rvd->vdev_children) { ASSERT(rvd->vdev_child[vdev] != NULL); return (rvd->vdev_child[vdev]); } return (NULL); } vdev_t * vdev_lookup_by_guid(vdev_t *vd, uint64_t guid) { vdev_t *mvd; int c; if (vd->vdev_guid == guid) return (vd); for (c = 0; c < vd->vdev_children; c++) if ((mvd = vdev_lookup_by_guid(vd->vdev_child[c], guid)) != NULL) return (mvd); return (NULL); } void vdev_add_child(vdev_t *pvd, vdev_t *cvd) { size_t oldsize, newsize; uint64_t id = cvd->vdev_id; vdev_t **newchild; ASSERT(spa_config_held(cvd->vdev_spa, SCL_ALL, RW_WRITER) == SCL_ALL); ASSERT(cvd->vdev_parent == NULL); cvd->vdev_parent = pvd; if (pvd == NULL) return; ASSERT(id >= pvd->vdev_children || pvd->vdev_child[id] == NULL); oldsize = pvd->vdev_children * sizeof (vdev_t *); pvd->vdev_children = MAX(pvd->vdev_children, id + 1); newsize = pvd->vdev_children * sizeof (vdev_t *); newchild = kmem_alloc(newsize, KM_PUSHPAGE); if (pvd->vdev_child != NULL) { bcopy(pvd->vdev_child, newchild, oldsize); kmem_free(pvd->vdev_child, oldsize); } pvd->vdev_child = newchild; pvd->vdev_child[id] = cvd; cvd->vdev_top = (pvd->vdev_top ? pvd->vdev_top: cvd); ASSERT(cvd->vdev_top->vdev_parent->vdev_parent == NULL); /* * Walk up all ancestors to update guid sum. */ for (; pvd != NULL; pvd = pvd->vdev_parent) pvd->vdev_guid_sum += cvd->vdev_guid_sum; } void vdev_remove_child(vdev_t *pvd, vdev_t *cvd) { int c; uint_t id = cvd->vdev_id; ASSERT(cvd->vdev_parent == pvd); if (pvd == NULL) return; ASSERT(id < pvd->vdev_children); ASSERT(pvd->vdev_child[id] == cvd); pvd->vdev_child[id] = NULL; cvd->vdev_parent = NULL; for (c = 0; c < pvd->vdev_children; c++) if (pvd->vdev_child[c]) break; if (c == pvd->vdev_children) { kmem_free(pvd->vdev_child, c * sizeof (vdev_t *)); pvd->vdev_child = NULL; pvd->vdev_children = 0; } /* * Walk up all ancestors to update guid sum. */ for (; pvd != NULL; pvd = pvd->vdev_parent) pvd->vdev_guid_sum -= cvd->vdev_guid_sum; } /* * Remove any holes in the child array. */ void vdev_compact_children(vdev_t *pvd) { vdev_t **newchild, *cvd; int oldc = pvd->vdev_children; int newc; int c; ASSERT(spa_config_held(pvd->vdev_spa, SCL_ALL, RW_WRITER) == SCL_ALL); for (c = newc = 0; c < oldc; c++) if (pvd->vdev_child[c]) newc++; newchild = kmem_zalloc(newc * sizeof (vdev_t *), KM_PUSHPAGE); for (c = newc = 0; c < oldc; c++) { if ((cvd = pvd->vdev_child[c]) != NULL) { newchild[newc] = cvd; cvd->vdev_id = newc++; } } kmem_free(pvd->vdev_child, oldc * sizeof (vdev_t *)); pvd->vdev_child = newchild; pvd->vdev_children = newc; } /* * Allocate and minimally initialize a vdev_t. */ vdev_t * vdev_alloc_common(spa_t *spa, uint_t id, uint64_t guid, vdev_ops_t *ops) { vdev_t *vd; int t; vd = kmem_zalloc(sizeof (vdev_t), KM_PUSHPAGE); if (spa->spa_root_vdev == NULL) { ASSERT(ops == &vdev_root_ops); spa->spa_root_vdev = vd; spa->spa_load_guid = spa_generate_guid(NULL); } if (guid == 0 && ops != &vdev_hole_ops) { if (spa->spa_root_vdev == vd) { /* * The root vdev's guid will also be the pool guid, * which must be unique among all pools. */ guid = spa_generate_guid(NULL); } else { /* * Any other vdev's guid must be unique within the pool. */ guid = spa_generate_guid(spa); } ASSERT(!spa_guid_exists(spa_guid(spa), guid)); } vd->vdev_spa = spa; vd->vdev_id = id; vd->vdev_guid = guid; vd->vdev_guid_sum = guid; vd->vdev_ops = ops; vd->vdev_state = VDEV_STATE_CLOSED; vd->vdev_ishole = (ops == &vdev_hole_ops); list_link_init(&vd->vdev_config_dirty_node); list_link_init(&vd->vdev_state_dirty_node); mutex_init(&vd->vdev_dtl_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&vd->vdev_stat_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&vd->vdev_probe_lock, NULL, MUTEX_DEFAULT, NULL); for (t = 0; t < DTL_TYPES; t++) { vd->vdev_dtl[t] = range_tree_create(NULL, NULL, &vd->vdev_dtl_lock); } txg_list_create(&vd->vdev_ms_list, offsetof(struct metaslab, ms_txg_node)); txg_list_create(&vd->vdev_dtl_list, offsetof(struct vdev, vdev_dtl_node)); vd->vdev_stat.vs_timestamp = gethrtime(); vdev_queue_init(vd); vdev_cache_init(vd); return (vd); } /* * Allocate a new vdev. The 'alloctype' is used to control whether we are * creating a new vdev or loading an existing one - the behavior is slightly * different for each case. */ int vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *nv, vdev_t *parent, uint_t id, int alloctype) { vdev_ops_t *ops; char *type; uint64_t guid = 0, islog, nparity; vdev_t *vd; ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); if (nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) != 0) return (SET_ERROR(EINVAL)); if ((ops = vdev_getops(type)) == NULL) return (SET_ERROR(EINVAL)); /* * If this is a load, get the vdev guid from the nvlist. * Otherwise, vdev_alloc_common() will generate one for us. */ if (alloctype == VDEV_ALLOC_LOAD) { uint64_t label_id; if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ID, &label_id) || label_id != id) return (SET_ERROR(EINVAL)); if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) != 0) return (SET_ERROR(EINVAL)); } else if (alloctype == VDEV_ALLOC_SPARE) { if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) != 0) return (SET_ERROR(EINVAL)); } else if (alloctype == VDEV_ALLOC_L2CACHE) { if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) != 0) return (SET_ERROR(EINVAL)); } else if (alloctype == VDEV_ALLOC_ROOTPOOL) { if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) != 0) return (SET_ERROR(EINVAL)); } /* * The first allocated vdev must be of type 'root'. */ if (ops != &vdev_root_ops && spa->spa_root_vdev == NULL) return (SET_ERROR(EINVAL)); /* * Determine whether we're a log vdev. */ islog = 0; (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_LOG, &islog); if (islog && spa_version(spa) < SPA_VERSION_SLOGS) return (SET_ERROR(ENOTSUP)); if (ops == &vdev_hole_ops && spa_version(spa) < SPA_VERSION_HOLES) return (SET_ERROR(ENOTSUP)); /* * Set the nparity property for RAID-Z vdevs. */ nparity = -1ULL; if (ops == &vdev_raidz_ops) { if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NPARITY, &nparity) == 0) { if (nparity == 0 || nparity > VDEV_RAIDZ_MAXPARITY) return (SET_ERROR(EINVAL)); /* * Previous versions could only support 1 or 2 parity * device. */ if (nparity > 1 && spa_version(spa) < SPA_VERSION_RAIDZ2) return (SET_ERROR(ENOTSUP)); if (nparity > 2 && spa_version(spa) < SPA_VERSION_RAIDZ3) return (SET_ERROR(ENOTSUP)); } else { /* * We require the parity to be specified for SPAs that * support multiple parity levels. */ if (spa_version(spa) >= SPA_VERSION_RAIDZ2) return (SET_ERROR(EINVAL)); /* * Otherwise, we default to 1 parity device for RAID-Z. */ nparity = 1; } } else { nparity = 0; } ASSERT(nparity != -1ULL); vd = vdev_alloc_common(spa, id, guid, ops); vd->vdev_islog = islog; vd->vdev_nparity = nparity; if (nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &vd->vdev_path) == 0) vd->vdev_path = spa_strdup(vd->vdev_path); if (nvlist_lookup_string(nv, ZPOOL_CONFIG_DEVID, &vd->vdev_devid) == 0) vd->vdev_devid = spa_strdup(vd->vdev_devid); if (nvlist_lookup_string(nv, ZPOOL_CONFIG_PHYS_PATH, &vd->vdev_physpath) == 0) vd->vdev_physpath = spa_strdup(vd->vdev_physpath); if (nvlist_lookup_string(nv, ZPOOL_CONFIG_FRU, &vd->vdev_fru) == 0) vd->vdev_fru = spa_strdup(vd->vdev_fru); /* * Set the whole_disk property. If it's not specified, leave the value * as -1. */ if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK, &vd->vdev_wholedisk) != 0) vd->vdev_wholedisk = -1ULL; /* * Look for the 'not present' flag. This will only be set if the device * was not present at the time of import. */ (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NOT_PRESENT, &vd->vdev_not_present); /* * Get the alignment requirement. */ (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ASHIFT, &vd->vdev_ashift); /* * Retrieve the vdev creation time. */ (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_CREATE_TXG, &vd->vdev_crtxg); /* * If we're a top-level vdev, try to load the allocation parameters. */ if (parent && !parent->vdev_parent && (alloctype == VDEV_ALLOC_LOAD || alloctype == VDEV_ALLOC_SPLIT)) { (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_METASLAB_ARRAY, &vd->vdev_ms_array); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_METASLAB_SHIFT, &vd->vdev_ms_shift); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ASIZE, &vd->vdev_asize); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_REMOVING, &vd->vdev_removing); } if (parent && !parent->vdev_parent && alloctype != VDEV_ALLOC_ATTACH) { ASSERT(alloctype == VDEV_ALLOC_LOAD || alloctype == VDEV_ALLOC_ADD || alloctype == VDEV_ALLOC_SPLIT || alloctype == VDEV_ALLOC_ROOTPOOL); vd->vdev_mg = metaslab_group_create(islog ? spa_log_class(spa) : spa_normal_class(spa), vd); } /* * If we're a leaf vdev, try to load the DTL object and other state. */ if (vd->vdev_ops->vdev_op_leaf && (alloctype == VDEV_ALLOC_LOAD || alloctype == VDEV_ALLOC_L2CACHE || alloctype == VDEV_ALLOC_ROOTPOOL)) { if (alloctype == VDEV_ALLOC_LOAD) { (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_DTL, &vd->vdev_dtl_object); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_UNSPARE, &vd->vdev_unspare); } if (alloctype == VDEV_ALLOC_ROOTPOOL) { uint64_t spare = 0; if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_SPARE, &spare) == 0 && spare) spa_spare_add(vd); } (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_OFFLINE, &vd->vdev_offline); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_RESILVER_TXG, &vd->vdev_resilver_txg); /* * When importing a pool, we want to ignore the persistent fault * state, as the diagnosis made on another system may not be * valid in the current context. Local vdevs will * remain in the faulted state. */ if (spa_load_state(spa) == SPA_LOAD_OPEN) { (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_FAULTED, &vd->vdev_faulted); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_DEGRADED, &vd->vdev_degraded); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_REMOVED, &vd->vdev_removed); if (vd->vdev_faulted || vd->vdev_degraded) { char *aux; vd->vdev_label_aux = VDEV_AUX_ERR_EXCEEDED; if (nvlist_lookup_string(nv, ZPOOL_CONFIG_AUX_STATE, &aux) == 0 && strcmp(aux, "external") == 0) vd->vdev_label_aux = VDEV_AUX_EXTERNAL; } } } /* * Add ourselves to the parent's list of children. */ vdev_add_child(parent, vd); *vdp = vd; return (0); } void vdev_free(vdev_t *vd) { int c, t; spa_t *spa = vd->vdev_spa; /* * vdev_free() implies closing the vdev first. This is simpler than * trying to ensure complicated semantics for all callers. */ vdev_close(vd); ASSERT(!list_link_active(&vd->vdev_config_dirty_node)); ASSERT(!list_link_active(&vd->vdev_state_dirty_node)); /* * Free all children. */ for (c = 0; c < vd->vdev_children; c++) vdev_free(vd->vdev_child[c]); ASSERT(vd->vdev_child == NULL); ASSERT(vd->vdev_guid_sum == vd->vdev_guid); /* * Discard allocation state. */ if (vd->vdev_mg != NULL) { vdev_metaslab_fini(vd); metaslab_group_destroy(vd->vdev_mg); } ASSERT0(vd->vdev_stat.vs_space); ASSERT0(vd->vdev_stat.vs_dspace); ASSERT0(vd->vdev_stat.vs_alloc); /* * Remove this vdev from its parent's child list. */ vdev_remove_child(vd->vdev_parent, vd); ASSERT(vd->vdev_parent == NULL); /* * Clean up vdev structure. */ vdev_queue_fini(vd); vdev_cache_fini(vd); if (vd->vdev_path) spa_strfree(vd->vdev_path); if (vd->vdev_devid) spa_strfree(vd->vdev_devid); if (vd->vdev_physpath) spa_strfree(vd->vdev_physpath); if (vd->vdev_fru) spa_strfree(vd->vdev_fru); if (vd->vdev_isspare) spa_spare_remove(vd); if (vd->vdev_isl2cache) spa_l2cache_remove(vd); txg_list_destroy(&vd->vdev_ms_list); txg_list_destroy(&vd->vdev_dtl_list); mutex_enter(&vd->vdev_dtl_lock); space_map_close(vd->vdev_dtl_sm); for (t = 0; t < DTL_TYPES; t++) { range_tree_vacate(vd->vdev_dtl[t], NULL, NULL); range_tree_destroy(vd->vdev_dtl[t]); } mutex_exit(&vd->vdev_dtl_lock); mutex_destroy(&vd->vdev_dtl_lock); mutex_destroy(&vd->vdev_stat_lock); mutex_destroy(&vd->vdev_probe_lock); if (vd == spa->spa_root_vdev) spa->spa_root_vdev = NULL; kmem_free(vd, sizeof (vdev_t)); } /* * Transfer top-level vdev state from svd to tvd. */ static void vdev_top_transfer(vdev_t *svd, vdev_t *tvd) { spa_t *spa = svd->vdev_spa; metaslab_t *msp; vdev_t *vd; int t; ASSERT(tvd == tvd->vdev_top); tvd->vdev_ms_array = svd->vdev_ms_array; tvd->vdev_ms_shift = svd->vdev_ms_shift; tvd->vdev_ms_count = svd->vdev_ms_count; svd->vdev_ms_array = 0; svd->vdev_ms_shift = 0; svd->vdev_ms_count = 0; if (tvd->vdev_mg) ASSERT3P(tvd->vdev_mg, ==, svd->vdev_mg); tvd->vdev_mg = svd->vdev_mg; tvd->vdev_ms = svd->vdev_ms; svd->vdev_mg = NULL; svd->vdev_ms = NULL; if (tvd->vdev_mg != NULL) tvd->vdev_mg->mg_vd = tvd; tvd->vdev_stat.vs_alloc = svd->vdev_stat.vs_alloc; tvd->vdev_stat.vs_space = svd->vdev_stat.vs_space; tvd->vdev_stat.vs_dspace = svd->vdev_stat.vs_dspace; svd->vdev_stat.vs_alloc = 0; svd->vdev_stat.vs_space = 0; svd->vdev_stat.vs_dspace = 0; for (t = 0; t < TXG_SIZE; t++) { while ((msp = txg_list_remove(&svd->vdev_ms_list, t)) != NULL) (void) txg_list_add(&tvd->vdev_ms_list, msp, t); while ((vd = txg_list_remove(&svd->vdev_dtl_list, t)) != NULL) (void) txg_list_add(&tvd->vdev_dtl_list, vd, t); if (txg_list_remove_this(&spa->spa_vdev_txg_list, svd, t)) (void) txg_list_add(&spa->spa_vdev_txg_list, tvd, t); } if (list_link_active(&svd->vdev_config_dirty_node)) { vdev_config_clean(svd); vdev_config_dirty(tvd); } if (list_link_active(&svd->vdev_state_dirty_node)) { vdev_state_clean(svd); vdev_state_dirty(tvd); } tvd->vdev_deflate_ratio = svd->vdev_deflate_ratio; svd->vdev_deflate_ratio = 0; tvd->vdev_islog = svd->vdev_islog; svd->vdev_islog = 0; } static void vdev_top_update(vdev_t *tvd, vdev_t *vd) { int c; if (vd == NULL) return; vd->vdev_top = tvd; for (c = 0; c < vd->vdev_children; c++) vdev_top_update(tvd, vd->vdev_child[c]); } /* * Add a mirror/replacing vdev above an existing vdev. */ vdev_t * vdev_add_parent(vdev_t *cvd, vdev_ops_t *ops) { spa_t *spa = cvd->vdev_spa; vdev_t *pvd = cvd->vdev_parent; vdev_t *mvd; ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); mvd = vdev_alloc_common(spa, cvd->vdev_id, 0, ops); mvd->vdev_asize = cvd->vdev_asize; mvd->vdev_min_asize = cvd->vdev_min_asize; mvd->vdev_max_asize = cvd->vdev_max_asize; mvd->vdev_ashift = cvd->vdev_ashift; mvd->vdev_state = cvd->vdev_state; mvd->vdev_crtxg = cvd->vdev_crtxg; vdev_remove_child(pvd, cvd); vdev_add_child(pvd, mvd); cvd->vdev_id = mvd->vdev_children; vdev_add_child(mvd, cvd); vdev_top_update(cvd->vdev_top, cvd->vdev_top); if (mvd == mvd->vdev_top) vdev_top_transfer(cvd, mvd); return (mvd); } /* * Remove a 1-way mirror/replacing vdev from the tree. */ void vdev_remove_parent(vdev_t *cvd) { vdev_t *mvd = cvd->vdev_parent; vdev_t *pvd = mvd->vdev_parent; ASSERT(spa_config_held(cvd->vdev_spa, SCL_ALL, RW_WRITER) == SCL_ALL); ASSERT(mvd->vdev_children == 1); ASSERT(mvd->vdev_ops == &vdev_mirror_ops || mvd->vdev_ops == &vdev_replacing_ops || mvd->vdev_ops == &vdev_spare_ops); cvd->vdev_ashift = mvd->vdev_ashift; vdev_remove_child(mvd, cvd); vdev_remove_child(pvd, mvd); /* * If cvd will replace mvd as a top-level vdev, preserve mvd's guid. * Otherwise, we could have detached an offline device, and when we * go to import the pool we'll think we have two top-level vdevs, * instead of a different version of the same top-level vdev. */ if (mvd->vdev_top == mvd) { uint64_t guid_delta = mvd->vdev_guid - cvd->vdev_guid; cvd->vdev_orig_guid = cvd->vdev_guid; cvd->vdev_guid += guid_delta; cvd->vdev_guid_sum += guid_delta; /* * If pool not set for autoexpand, we need to also preserve * mvd's asize to prevent automatic expansion of cvd. * Otherwise if we are adjusting the mirror by attaching and * detaching children of non-uniform sizes, the mirror could * autoexpand, unexpectedly requiring larger devices to * re-establish the mirror. */ if (!cvd->vdev_spa->spa_autoexpand) cvd->vdev_asize = mvd->vdev_asize; } cvd->vdev_id = mvd->vdev_id; vdev_add_child(pvd, cvd); vdev_top_update(cvd->vdev_top, cvd->vdev_top); if (cvd == cvd->vdev_top) vdev_top_transfer(mvd, cvd); ASSERT(mvd->vdev_children == 0); vdev_free(mvd); } int vdev_metaslab_init(vdev_t *vd, uint64_t txg) { spa_t *spa = vd->vdev_spa; objset_t *mos = spa->spa_meta_objset; uint64_t m; uint64_t oldc = vd->vdev_ms_count; uint64_t newc = vd->vdev_asize >> vd->vdev_ms_shift; metaslab_t **mspp; int error; ASSERT(txg == 0 || spa_config_held(spa, SCL_ALLOC, RW_WRITER)); /* * This vdev is not being allocated from yet or is a hole. */ if (vd->vdev_ms_shift == 0) return (0); ASSERT(!vd->vdev_ishole); /* * Compute the raidz-deflation ratio. Note, we hard-code * in 128k (1 << 17) because it is the current "typical" blocksize. * Even if SPA_MAXBLOCKSIZE changes, this algorithm must never change, * or we will inconsistently account for existing bp's. */ vd->vdev_deflate_ratio = (1 << 17) / (vdev_psize_to_asize(vd, 1 << 17) >> SPA_MINBLOCKSHIFT); ASSERT(oldc <= newc); - mspp = kmem_zalloc(newc * sizeof (*mspp), KM_PUSHPAGE | KM_NODEBUG); + mspp = kmem_zalloc(newc * sizeof (*mspp), KM_PUSHPAGE); if (oldc != 0) { bcopy(vd->vdev_ms, mspp, oldc * sizeof (*mspp)); kmem_free(vd->vdev_ms, oldc * sizeof (*mspp)); } vd->vdev_ms = mspp; vd->vdev_ms_count = newc; for (m = oldc; m < newc; m++) { uint64_t object = 0; if (txg == 0) { error = dmu_read(mos, vd->vdev_ms_array, m * sizeof (uint64_t), sizeof (uint64_t), &object, DMU_READ_PREFETCH); if (error) return (error); } error = metaslab_init(vd->vdev_mg, m, object, txg, &(vd->vdev_ms[m])); if (error) return (error); } if (txg == 0) spa_config_enter(spa, SCL_ALLOC, FTAG, RW_WRITER); /* * If the vdev is being removed we don't activate * the metaslabs since we want to ensure that no new * allocations are performed on this device. */ if (oldc == 0 && !vd->vdev_removing) metaslab_group_activate(vd->vdev_mg); if (txg == 0) spa_config_exit(spa, SCL_ALLOC, FTAG); return (0); } void vdev_metaslab_fini(vdev_t *vd) { uint64_t m; uint64_t count = vd->vdev_ms_count; if (vd->vdev_ms != NULL) { metaslab_group_passivate(vd->vdev_mg); for (m = 0; m < count; m++) { metaslab_t *msp = vd->vdev_ms[m]; if (msp != NULL) metaslab_fini(msp); } kmem_free(vd->vdev_ms, count * sizeof (metaslab_t *)); vd->vdev_ms = NULL; } ASSERT3U(vd->vdev_pending_fastwrite, ==, 0); } typedef struct vdev_probe_stats { boolean_t vps_readable; boolean_t vps_writeable; int vps_flags; } vdev_probe_stats_t; static void vdev_probe_done(zio_t *zio) { spa_t *spa = zio->io_spa; vdev_t *vd = zio->io_vd; vdev_probe_stats_t *vps = zio->io_private; ASSERT(vd->vdev_probe_zio != NULL); if (zio->io_type == ZIO_TYPE_READ) { if (zio->io_error == 0) vps->vps_readable = 1; if (zio->io_error == 0 && spa_writeable(spa)) { zio_nowait(zio_write_phys(vd->vdev_probe_zio, vd, zio->io_offset, zio->io_size, zio->io_data, ZIO_CHECKSUM_OFF, vdev_probe_done, vps, ZIO_PRIORITY_SYNC_WRITE, vps->vps_flags, B_TRUE)); } else { zio_buf_free(zio->io_data, zio->io_size); } } else if (zio->io_type == ZIO_TYPE_WRITE) { if (zio->io_error == 0) vps->vps_writeable = 1; zio_buf_free(zio->io_data, zio->io_size); } else if (zio->io_type == ZIO_TYPE_NULL) { zio_t *pio; vd->vdev_cant_read |= !vps->vps_readable; vd->vdev_cant_write |= !vps->vps_writeable; if (vdev_readable(vd) && (vdev_writeable(vd) || !spa_writeable(spa))) { zio->io_error = 0; } else { ASSERT(zio->io_error != 0); zfs_ereport_post(FM_EREPORT_ZFS_PROBE_FAILURE, spa, vd, NULL, 0, 0); zio->io_error = SET_ERROR(ENXIO); } mutex_enter(&vd->vdev_probe_lock); ASSERT(vd->vdev_probe_zio == zio); vd->vdev_probe_zio = NULL; mutex_exit(&vd->vdev_probe_lock); while ((pio = zio_walk_parents(zio)) != NULL) if (!vdev_accessible(vd, pio)) pio->io_error = SET_ERROR(ENXIO); kmem_free(vps, sizeof (*vps)); } } /* * Determine whether this device is accessible. * * Read and write to several known locations: the pad regions of each * vdev label but the first, which we leave alone in case it contains * a VTOC. */ zio_t * vdev_probe(vdev_t *vd, zio_t *zio) { spa_t *spa = vd->vdev_spa; vdev_probe_stats_t *vps = NULL; zio_t *pio; int l; ASSERT(vd->vdev_ops->vdev_op_leaf); /* * Don't probe the probe. */ if (zio && (zio->io_flags & ZIO_FLAG_PROBE)) return (NULL); /* * To prevent 'probe storms' when a device fails, we create * just one probe i/o at a time. All zios that want to probe * this vdev will become parents of the probe io. */ mutex_enter(&vd->vdev_probe_lock); if ((pio = vd->vdev_probe_zio) == NULL) { vps = kmem_zalloc(sizeof (*vps), KM_PUSHPAGE); vps->vps_flags = ZIO_FLAG_CANFAIL | ZIO_FLAG_PROBE | ZIO_FLAG_DONT_CACHE | ZIO_FLAG_DONT_AGGREGATE | ZIO_FLAG_TRYHARD; if (spa_config_held(spa, SCL_ZIO, RW_WRITER)) { /* * vdev_cant_read and vdev_cant_write can only * transition from TRUE to FALSE when we have the * SCL_ZIO lock as writer; otherwise they can only * transition from FALSE to TRUE. This ensures that * any zio looking at these values can assume that * failures persist for the life of the I/O. That's * important because when a device has intermittent * connectivity problems, we want to ensure that * they're ascribed to the device (ENXIO) and not * the zio (EIO). * * Since we hold SCL_ZIO as writer here, clear both * values so the probe can reevaluate from first * principles. */ vps->vps_flags |= ZIO_FLAG_CONFIG_WRITER; vd->vdev_cant_read = B_FALSE; vd->vdev_cant_write = B_FALSE; } vd->vdev_probe_zio = pio = zio_null(NULL, spa, vd, vdev_probe_done, vps, vps->vps_flags | ZIO_FLAG_DONT_PROPAGATE); /* * We can't change the vdev state in this context, so we * kick off an async task to do it on our behalf. */ if (zio != NULL) { vd->vdev_probe_wanted = B_TRUE; spa_async_request(spa, SPA_ASYNC_PROBE); } } if (zio != NULL) zio_add_child(zio, pio); mutex_exit(&vd->vdev_probe_lock); if (vps == NULL) { ASSERT(zio != NULL); return (NULL); } for (l = 1; l < VDEV_LABELS; l++) { zio_nowait(zio_read_phys(pio, vd, vdev_label_offset(vd->vdev_psize, l, offsetof(vdev_label_t, vl_pad2)), VDEV_PAD_SIZE, zio_buf_alloc(VDEV_PAD_SIZE), ZIO_CHECKSUM_OFF, vdev_probe_done, vps, ZIO_PRIORITY_SYNC_READ, vps->vps_flags, B_TRUE)); } if (zio == NULL) return (pio); zio_nowait(pio); return (NULL); } static void vdev_open_child(void *arg) { vdev_t *vd = arg; vd->vdev_open_thread = curthread; vd->vdev_open_error = vdev_open(vd); vd->vdev_open_thread = NULL; } static boolean_t vdev_uses_zvols(vdev_t *vd) { int c; #ifdef _KERNEL if (zvol_is_zvol(vd->vdev_path)) return (B_TRUE); #endif for (c = 0; c < vd->vdev_children; c++) if (vdev_uses_zvols(vd->vdev_child[c])) return (B_TRUE); return (B_FALSE); } void vdev_open_children(vdev_t *vd) { taskq_t *tq; int children = vd->vdev_children; int c; /* * in order to handle pools on top of zvols, do the opens * in a single thread so that the same thread holds the * spa_namespace_lock */ if (vdev_uses_zvols(vd)) { for (c = 0; c < children; c++) vd->vdev_child[c]->vdev_open_error = vdev_open(vd->vdev_child[c]); return; } tq = taskq_create("vdev_open", children, minclsyspri, children, children, TASKQ_PREPOPULATE); for (c = 0; c < children; c++) VERIFY(taskq_dispatch(tq, vdev_open_child, vd->vdev_child[c], TQ_SLEEP) != 0); taskq_destroy(tq); } /* * Prepare a virtual device for access. */ int vdev_open(vdev_t *vd) { spa_t *spa = vd->vdev_spa; int error; uint64_t osize = 0; uint64_t max_osize = 0; uint64_t asize, max_asize, psize; uint64_t ashift = 0; int c; ASSERT(vd->vdev_open_thread == curthread || spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL); ASSERT(vd->vdev_state == VDEV_STATE_CLOSED || vd->vdev_state == VDEV_STATE_CANT_OPEN || vd->vdev_state == VDEV_STATE_OFFLINE); vd->vdev_stat.vs_aux = VDEV_AUX_NONE; vd->vdev_cant_read = B_FALSE; vd->vdev_cant_write = B_FALSE; vd->vdev_min_asize = vdev_get_min_asize(vd); /* * If this vdev is not removed, check its fault status. If it's * faulted, bail out of the open. */ if (!vd->vdev_removed && vd->vdev_faulted) { ASSERT(vd->vdev_children == 0); ASSERT(vd->vdev_label_aux == VDEV_AUX_ERR_EXCEEDED || vd->vdev_label_aux == VDEV_AUX_EXTERNAL); vdev_set_state(vd, B_TRUE, VDEV_STATE_FAULTED, vd->vdev_label_aux); return (SET_ERROR(ENXIO)); } else if (vd->vdev_offline) { ASSERT(vd->vdev_children == 0); vdev_set_state(vd, B_TRUE, VDEV_STATE_OFFLINE, VDEV_AUX_NONE); return (SET_ERROR(ENXIO)); } error = vd->vdev_ops->vdev_op_open(vd, &osize, &max_osize, &ashift); /* * Reset the vdev_reopening flag so that we actually close * the vdev on error. */ vd->vdev_reopening = B_FALSE; if (zio_injection_enabled && error == 0) error = zio_handle_device_injection(vd, NULL, ENXIO); if (error) { if (vd->vdev_removed && vd->vdev_stat.vs_aux != VDEV_AUX_OPEN_FAILED) vd->vdev_removed = B_FALSE; vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, vd->vdev_stat.vs_aux); return (error); } vd->vdev_removed = B_FALSE; /* * Recheck the faulted flag now that we have confirmed that * the vdev is accessible. If we're faulted, bail. */ if (vd->vdev_faulted) { ASSERT(vd->vdev_children == 0); ASSERT(vd->vdev_label_aux == VDEV_AUX_ERR_EXCEEDED || vd->vdev_label_aux == VDEV_AUX_EXTERNAL); vdev_set_state(vd, B_TRUE, VDEV_STATE_FAULTED, vd->vdev_label_aux); return (SET_ERROR(ENXIO)); } if (vd->vdev_degraded) { ASSERT(vd->vdev_children == 0); vdev_set_state(vd, B_TRUE, VDEV_STATE_DEGRADED, VDEV_AUX_ERR_EXCEEDED); } else { vdev_set_state(vd, B_TRUE, VDEV_STATE_HEALTHY, 0); } /* * For hole or missing vdevs we just return success. */ if (vd->vdev_ishole || vd->vdev_ops == &vdev_missing_ops) return (0); for (c = 0; c < vd->vdev_children; c++) { if (vd->vdev_child[c]->vdev_state != VDEV_STATE_HEALTHY) { vdev_set_state(vd, B_TRUE, VDEV_STATE_DEGRADED, VDEV_AUX_NONE); break; } } osize = P2ALIGN(osize, (uint64_t)sizeof (vdev_label_t)); max_osize = P2ALIGN(max_osize, (uint64_t)sizeof (vdev_label_t)); if (vd->vdev_children == 0) { if (osize < SPA_MINDEVSIZE) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_TOO_SMALL); return (SET_ERROR(EOVERFLOW)); } psize = osize; asize = osize - (VDEV_LABEL_START_SIZE + VDEV_LABEL_END_SIZE); max_asize = max_osize - (VDEV_LABEL_START_SIZE + VDEV_LABEL_END_SIZE); } else { if (vd->vdev_parent != NULL && osize < SPA_MINDEVSIZE - (VDEV_LABEL_START_SIZE + VDEV_LABEL_END_SIZE)) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_TOO_SMALL); return (SET_ERROR(EOVERFLOW)); } psize = 0; asize = osize; max_asize = max_osize; } vd->vdev_psize = psize; /* * Make sure the allocatable size hasn't shrunk. */ if (asize < vd->vdev_min_asize) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_BAD_LABEL); return (SET_ERROR(EINVAL)); } if (vd->vdev_asize == 0) { /* * This is the first-ever open, so use the computed values. * For compatibility, a different ashift can be requested. */ vd->vdev_asize = asize; vd->vdev_max_asize = max_asize; if (vd->vdev_ashift == 0) vd->vdev_ashift = ashift; } else { /* * Detect if the alignment requirement has increased. * We don't want to make the pool unavailable, just * post an event instead. */ if (ashift > vd->vdev_top->vdev_ashift && vd->vdev_ops->vdev_op_leaf) { zfs_ereport_post(FM_EREPORT_ZFS_DEVICE_BAD_ASHIFT, spa, vd, NULL, 0, 0); } vd->vdev_max_asize = max_asize; } /* * If all children are healthy and the asize has increased, * then we've experienced dynamic LUN growth. If automatic * expansion is enabled then use the additional space. */ if (vd->vdev_state == VDEV_STATE_HEALTHY && asize > vd->vdev_asize && (vd->vdev_expanding || spa->spa_autoexpand)) vd->vdev_asize = asize; vdev_set_min_asize(vd); /* * Ensure we can issue some IO before declaring the * vdev open for business. */ if (vd->vdev_ops->vdev_op_leaf && (error = zio_wait(vdev_probe(vd, NULL))) != 0) { vdev_set_state(vd, B_TRUE, VDEV_STATE_FAULTED, VDEV_AUX_ERR_EXCEEDED); return (error); } /* * If a leaf vdev has a DTL, and seems healthy, then kick off a * resilver. But don't do this if we are doing a reopen for a scrub, * since this would just restart the scrub we are already doing. */ if (vd->vdev_ops->vdev_op_leaf && !spa->spa_scrub_reopen && vdev_resilver_needed(vd, NULL, NULL)) spa_async_request(spa, SPA_ASYNC_RESILVER); return (0); } /* * Called once the vdevs are all opened, this routine validates the label * contents. This needs to be done before vdev_load() so that we don't * inadvertently do repair I/Os to the wrong device. * * If 'strict' is false ignore the spa guid check. This is necessary because * if the machine crashed during a re-guid the new guid might have been written * to all of the vdev labels, but not the cached config. The strict check * will be performed when the pool is opened again using the mos config. * * This function will only return failure if one of the vdevs indicates that it * has since been destroyed or exported. This is only possible if * /etc/zfs/zpool.cache was readonly at the time. Otherwise, the vdev state * will be updated but the function will return 0. */ int vdev_validate(vdev_t *vd, boolean_t strict) { spa_t *spa = vd->vdev_spa; nvlist_t *label; uint64_t guid = 0, top_guid; uint64_t state; int c; for (c = 0; c < vd->vdev_children; c++) if (vdev_validate(vd->vdev_child[c], strict) != 0) return (SET_ERROR(EBADF)); /* * If the device has already failed, or was marked offline, don't do * any further validation. Otherwise, label I/O will fail and we will * overwrite the previous state. */ if (vd->vdev_ops->vdev_op_leaf && vdev_readable(vd)) { uint64_t aux_guid = 0; nvlist_t *nvl; uint64_t txg = spa_last_synced_txg(spa) != 0 ? spa_last_synced_txg(spa) : -1ULL; if ((label = vdev_label_read_config(vd, txg)) == NULL) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_BAD_LABEL); return (0); } /* * Determine if this vdev has been split off into another * pool. If so, then refuse to open it. */ if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_SPLIT_GUID, &aux_guid) == 0 && aux_guid == spa_guid(spa)) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_SPLIT_POOL); nvlist_free(label); return (0); } if (strict && (nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_GUID, &guid) != 0 || guid != spa_guid(spa))) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); return (0); } if (nvlist_lookup_nvlist(label, ZPOOL_CONFIG_VDEV_TREE, &nvl) != 0 || nvlist_lookup_uint64(nvl, ZPOOL_CONFIG_ORIG_GUID, &aux_guid) != 0) aux_guid = 0; /* * If this vdev just became a top-level vdev because its * sibling was detached, it will have adopted the parent's * vdev guid -- but the label may or may not be on disk yet. * Fortunately, either version of the label will have the * same top guid, so if we're a top-level vdev, we can * safely compare to that instead. * * If we split this vdev off instead, then we also check the * original pool's guid. We don't want to consider the vdev * corrupt if it is partway through a split operation. */ if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_GUID, &guid) != 0 || nvlist_lookup_uint64(label, ZPOOL_CONFIG_TOP_GUID, &top_guid) != 0 || ((vd->vdev_guid != guid && vd->vdev_guid != aux_guid) && (vd->vdev_guid != top_guid || vd != vd->vdev_top))) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); return (0); } if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_STATE, &state) != 0) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); return (0); } nvlist_free(label); /* * If this is a verbatim import, no need to check the * state of the pool. */ if (!(spa->spa_import_flags & ZFS_IMPORT_VERBATIM) && spa_load_state(spa) == SPA_LOAD_OPEN && state != POOL_STATE_ACTIVE) return (SET_ERROR(EBADF)); /* * If we were able to open and validate a vdev that was * previously marked permanently unavailable, clear that state * now. */ if (vd->vdev_not_present) vd->vdev_not_present = 0; } return (0); } /* * Close a virtual device. */ void vdev_close(vdev_t *vd) { vdev_t *pvd = vd->vdev_parent; ASSERTV(spa_t *spa = vd->vdev_spa); ASSERT(spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL); /* * If our parent is reopening, then we are as well, unless we are * going offline. */ if (pvd != NULL && pvd->vdev_reopening) vd->vdev_reopening = (pvd->vdev_reopening && !vd->vdev_offline); vd->vdev_ops->vdev_op_close(vd); vdev_cache_purge(vd); /* * We record the previous state before we close it, so that if we are * doing a reopen(), we don't generate FMA ereports if we notice that * it's still faulted. */ vd->vdev_prevstate = vd->vdev_state; if (vd->vdev_offline) vd->vdev_state = VDEV_STATE_OFFLINE; else vd->vdev_state = VDEV_STATE_CLOSED; vd->vdev_stat.vs_aux = VDEV_AUX_NONE; } void vdev_hold(vdev_t *vd) { spa_t *spa = vd->vdev_spa; int c; ASSERT(spa_is_root(spa)); if (spa->spa_state == POOL_STATE_UNINITIALIZED) return; for (c = 0; c < vd->vdev_children; c++) vdev_hold(vd->vdev_child[c]); if (vd->vdev_ops->vdev_op_leaf) vd->vdev_ops->vdev_op_hold(vd); } void vdev_rele(vdev_t *vd) { int c; ASSERT(spa_is_root(vd->vdev_spa)); for (c = 0; c < vd->vdev_children; c++) vdev_rele(vd->vdev_child[c]); if (vd->vdev_ops->vdev_op_leaf) vd->vdev_ops->vdev_op_rele(vd); } /* * Reopen all interior vdevs and any unopened leaves. We don't actually * reopen leaf vdevs which had previously been opened as they might deadlock * on the spa_config_lock. Instead we only obtain the leaf's physical size. * If the leaf has never been opened then open it, as usual. */ void vdev_reopen(vdev_t *vd) { spa_t *spa = vd->vdev_spa; ASSERT(spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL); /* set the reopening flag unless we're taking the vdev offline */ vd->vdev_reopening = !vd->vdev_offline; vdev_close(vd); (void) vdev_open(vd); /* * Call vdev_validate() here to make sure we have the same device. * Otherwise, a device with an invalid label could be successfully * opened in response to vdev_reopen(). */ if (vd->vdev_aux) { (void) vdev_validate_aux(vd); if (vdev_readable(vd) && vdev_writeable(vd) && vd->vdev_aux == &spa->spa_l2cache && !l2arc_vdev_present(vd)) l2arc_add_vdev(spa, vd); } else { (void) vdev_validate(vd, B_TRUE); } /* * Reassess parent vdev's health. */ vdev_propagate_state(vd); } int vdev_create(vdev_t *vd, uint64_t txg, boolean_t isreplacing) { int error; /* * Normally, partial opens (e.g. of a mirror) are allowed. * For a create, however, we want to fail the request if * there are any components we can't open. */ error = vdev_open(vd); if (error || vd->vdev_state != VDEV_STATE_HEALTHY) { vdev_close(vd); return (error ? error : ENXIO); } /* * Recursively load DTLs and initialize all labels. */ if ((error = vdev_dtl_load(vd)) != 0 || (error = vdev_label_init(vd, txg, isreplacing ? VDEV_LABEL_REPLACE : VDEV_LABEL_CREATE)) != 0) { vdev_close(vd); return (error); } return (0); } void vdev_metaslab_set_size(vdev_t *vd) { /* * Aim for roughly metaslabs_per_vdev (default 200) metaslabs per vdev. */ vd->vdev_ms_shift = highbit64(vd->vdev_asize / metaslabs_per_vdev); vd->vdev_ms_shift = MAX(vd->vdev_ms_shift, SPA_MAXBLOCKSHIFT); } void vdev_dirty(vdev_t *vd, int flags, void *arg, uint64_t txg) { ASSERT(vd == vd->vdev_top); ASSERT(!vd->vdev_ishole); ASSERT(ISP2(flags)); ASSERT(spa_writeable(vd->vdev_spa)); if (flags & VDD_METASLAB) (void) txg_list_add(&vd->vdev_ms_list, arg, txg); if (flags & VDD_DTL) (void) txg_list_add(&vd->vdev_dtl_list, arg, txg); (void) txg_list_add(&vd->vdev_spa->spa_vdev_txg_list, vd, txg); } void vdev_dirty_leaves(vdev_t *vd, int flags, uint64_t txg) { int c; for (c = 0; c < vd->vdev_children; c++) vdev_dirty_leaves(vd->vdev_child[c], flags, txg); if (vd->vdev_ops->vdev_op_leaf) vdev_dirty(vd->vdev_top, flags, vd, txg); } /* * DTLs. * * A vdev's DTL (dirty time log) is the set of transaction groups for which * the vdev has less than perfect replication. There are four kinds of DTL: * * DTL_MISSING: txgs for which the vdev has no valid copies of the data * * DTL_PARTIAL: txgs for which data is available, but not fully replicated * * DTL_SCRUB: the txgs that could not be repaired by the last scrub; upon * scrub completion, DTL_SCRUB replaces DTL_MISSING in the range of * txgs that was scrubbed. * * DTL_OUTAGE: txgs which cannot currently be read, whether due to * persistent errors or just some device being offline. * Unlike the other three, the DTL_OUTAGE map is not generally * maintained; it's only computed when needed, typically to * determine whether a device can be detached. * * For leaf vdevs, DTL_MISSING and DTL_PARTIAL are identical: the device * either has the data or it doesn't. * * For interior vdevs such as mirror and RAID-Z the picture is more complex. * A vdev's DTL_PARTIAL is the union of its children's DTL_PARTIALs, because * if any child is less than fully replicated, then so is its parent. * A vdev's DTL_MISSING is a modified union of its children's DTL_MISSINGs, * comprising only those txgs which appear in 'maxfaults' or more children; * those are the txgs we don't have enough replication to read. For example, * double-parity RAID-Z can tolerate up to two missing devices (maxfaults == 2); * thus, its DTL_MISSING consists of the set of txgs that appear in more than * two child DTL_MISSING maps. * * It should be clear from the above that to compute the DTLs and outage maps * for all vdevs, it suffices to know just the leaf vdevs' DTL_MISSING maps. * Therefore, that is all we keep on disk. When loading the pool, or after * a configuration change, we generate all other DTLs from first principles. */ void vdev_dtl_dirty(vdev_t *vd, vdev_dtl_type_t t, uint64_t txg, uint64_t size) { range_tree_t *rt = vd->vdev_dtl[t]; ASSERT(t < DTL_TYPES); ASSERT(vd != vd->vdev_spa->spa_root_vdev); ASSERT(spa_writeable(vd->vdev_spa)); mutex_enter(rt->rt_lock); if (!range_tree_contains(rt, txg, size)) range_tree_add(rt, txg, size); mutex_exit(rt->rt_lock); } boolean_t vdev_dtl_contains(vdev_t *vd, vdev_dtl_type_t t, uint64_t txg, uint64_t size) { range_tree_t *rt = vd->vdev_dtl[t]; boolean_t dirty = B_FALSE; ASSERT(t < DTL_TYPES); ASSERT(vd != vd->vdev_spa->spa_root_vdev); mutex_enter(rt->rt_lock); if (range_tree_space(rt) != 0) dirty = range_tree_contains(rt, txg, size); mutex_exit(rt->rt_lock); return (dirty); } boolean_t vdev_dtl_empty(vdev_t *vd, vdev_dtl_type_t t) { range_tree_t *rt = vd->vdev_dtl[t]; boolean_t empty; mutex_enter(rt->rt_lock); empty = (range_tree_space(rt) == 0); mutex_exit(rt->rt_lock); return (empty); } /* * Returns the lowest txg in the DTL range. */ static uint64_t vdev_dtl_min(vdev_t *vd) { range_seg_t *rs; ASSERT(MUTEX_HELD(&vd->vdev_dtl_lock)); ASSERT3U(range_tree_space(vd->vdev_dtl[DTL_MISSING]), !=, 0); ASSERT0(vd->vdev_children); rs = avl_first(&vd->vdev_dtl[DTL_MISSING]->rt_root); return (rs->rs_start - 1); } /* * Returns the highest txg in the DTL. */ static uint64_t vdev_dtl_max(vdev_t *vd) { range_seg_t *rs; ASSERT(MUTEX_HELD(&vd->vdev_dtl_lock)); ASSERT3U(range_tree_space(vd->vdev_dtl[DTL_MISSING]), !=, 0); ASSERT0(vd->vdev_children); rs = avl_last(&vd->vdev_dtl[DTL_MISSING]->rt_root); return (rs->rs_end); } /* * Determine if a resilvering vdev should remove any DTL entries from * its range. If the vdev was resilvering for the entire duration of the * scan then it should excise that range from its DTLs. Otherwise, this * vdev is considered partially resilvered and should leave its DTL * entries intact. The comment in vdev_dtl_reassess() describes how we * excise the DTLs. */ static boolean_t vdev_dtl_should_excise(vdev_t *vd) { spa_t *spa = vd->vdev_spa; dsl_scan_t *scn = spa->spa_dsl_pool->dp_scan; ASSERT0(scn->scn_phys.scn_errors); ASSERT0(vd->vdev_children); if (vd->vdev_resilver_txg == 0 || range_tree_space(vd->vdev_dtl[DTL_MISSING]) == 0) return (B_TRUE); /* * When a resilver is initiated the scan will assign the scn_max_txg * value to the highest txg value that exists in all DTLs. If this * device's max DTL is not part of this scan (i.e. it is not in * the range (scn_min_txg, scn_max_txg] then it is not eligible * for excision. */ if (vdev_dtl_max(vd) <= scn->scn_phys.scn_max_txg) { ASSERT3U(scn->scn_phys.scn_min_txg, <=, vdev_dtl_min(vd)); ASSERT3U(scn->scn_phys.scn_min_txg, <, vd->vdev_resilver_txg); ASSERT3U(vd->vdev_resilver_txg, <=, scn->scn_phys.scn_max_txg); return (B_TRUE); } return (B_FALSE); } /* * Reassess DTLs after a config change or scrub completion. */ void vdev_dtl_reassess(vdev_t *vd, uint64_t txg, uint64_t scrub_txg, int scrub_done) { spa_t *spa = vd->vdev_spa; avl_tree_t reftree; int c, t, minref; ASSERT(spa_config_held(spa, SCL_ALL, RW_READER) != 0); for (c = 0; c < vd->vdev_children; c++) vdev_dtl_reassess(vd->vdev_child[c], txg, scrub_txg, scrub_done); if (vd == spa->spa_root_vdev || vd->vdev_ishole || vd->vdev_aux) return; if (vd->vdev_ops->vdev_op_leaf) { dsl_scan_t *scn = spa->spa_dsl_pool->dp_scan; mutex_enter(&vd->vdev_dtl_lock); /* * If we've completed a scan cleanly then determine * if this vdev should remove any DTLs. We only want to * excise regions on vdevs that were available during * the entire duration of this scan. */ if (scrub_txg != 0 && (spa->spa_scrub_started || (scn != NULL && scn->scn_phys.scn_errors == 0)) && vdev_dtl_should_excise(vd)) { /* * We completed a scrub up to scrub_txg. If we * did it without rebooting, then the scrub dtl * will be valid, so excise the old region and * fold in the scrub dtl. Otherwise, leave the * dtl as-is if there was an error. * * There's little trick here: to excise the beginning * of the DTL_MISSING map, we put it into a reference * tree and then add a segment with refcnt -1 that * covers the range [0, scrub_txg). This means * that each txg in that range has refcnt -1 or 0. * We then add DTL_SCRUB with a refcnt of 2, so that * entries in the range [0, scrub_txg) will have a * positive refcnt -- either 1 or 2. We then convert * the reference tree into the new DTL_MISSING map. */ space_reftree_create(&reftree); space_reftree_add_map(&reftree, vd->vdev_dtl[DTL_MISSING], 1); space_reftree_add_seg(&reftree, 0, scrub_txg, -1); space_reftree_add_map(&reftree, vd->vdev_dtl[DTL_SCRUB], 2); space_reftree_generate_map(&reftree, vd->vdev_dtl[DTL_MISSING], 1); space_reftree_destroy(&reftree); } range_tree_vacate(vd->vdev_dtl[DTL_PARTIAL], NULL, NULL); range_tree_walk(vd->vdev_dtl[DTL_MISSING], range_tree_add, vd->vdev_dtl[DTL_PARTIAL]); if (scrub_done) range_tree_vacate(vd->vdev_dtl[DTL_SCRUB], NULL, NULL); range_tree_vacate(vd->vdev_dtl[DTL_OUTAGE], NULL, NULL); if (!vdev_readable(vd)) range_tree_add(vd->vdev_dtl[DTL_OUTAGE], 0, -1ULL); else range_tree_walk(vd->vdev_dtl[DTL_MISSING], range_tree_add, vd->vdev_dtl[DTL_OUTAGE]); /* * If the vdev was resilvering and no longer has any * DTLs then reset its resilvering flag. */ if (vd->vdev_resilver_txg != 0 && range_tree_space(vd->vdev_dtl[DTL_MISSING]) == 0 && range_tree_space(vd->vdev_dtl[DTL_OUTAGE]) == 0) vd->vdev_resilver_txg = 0; mutex_exit(&vd->vdev_dtl_lock); if (txg != 0) vdev_dirty(vd->vdev_top, VDD_DTL, vd, txg); return; } mutex_enter(&vd->vdev_dtl_lock); for (t = 0; t < DTL_TYPES; t++) { int c; /* account for child's outage in parent's missing map */ int s = (t == DTL_MISSING) ? DTL_OUTAGE: t; if (t == DTL_SCRUB) continue; /* leaf vdevs only */ if (t == DTL_PARTIAL) minref = 1; /* i.e. non-zero */ else if (vd->vdev_nparity != 0) minref = vd->vdev_nparity + 1; /* RAID-Z */ else minref = vd->vdev_children; /* any kind of mirror */ space_reftree_create(&reftree); for (c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; mutex_enter(&cvd->vdev_dtl_lock); space_reftree_add_map(&reftree, cvd->vdev_dtl[s], 1); mutex_exit(&cvd->vdev_dtl_lock); } space_reftree_generate_map(&reftree, vd->vdev_dtl[t], minref); space_reftree_destroy(&reftree); } mutex_exit(&vd->vdev_dtl_lock); } int vdev_dtl_load(vdev_t *vd) { spa_t *spa = vd->vdev_spa; objset_t *mos = spa->spa_meta_objset; int error = 0; int c; if (vd->vdev_ops->vdev_op_leaf && vd->vdev_dtl_object != 0) { ASSERT(!vd->vdev_ishole); error = space_map_open(&vd->vdev_dtl_sm, mos, vd->vdev_dtl_object, 0, -1ULL, 0, &vd->vdev_dtl_lock); if (error) return (error); ASSERT(vd->vdev_dtl_sm != NULL); mutex_enter(&vd->vdev_dtl_lock); /* * Now that we've opened the space_map we need to update * the in-core DTL. */ space_map_update(vd->vdev_dtl_sm); error = space_map_load(vd->vdev_dtl_sm, vd->vdev_dtl[DTL_MISSING], SM_ALLOC); mutex_exit(&vd->vdev_dtl_lock); return (error); } for (c = 0; c < vd->vdev_children; c++) { error = vdev_dtl_load(vd->vdev_child[c]); if (error != 0) break; } return (error); } void vdev_dtl_sync(vdev_t *vd, uint64_t txg) { spa_t *spa = vd->vdev_spa; range_tree_t *rt = vd->vdev_dtl[DTL_MISSING]; objset_t *mos = spa->spa_meta_objset; range_tree_t *rtsync; kmutex_t rtlock; dmu_tx_t *tx; uint64_t object = space_map_object(vd->vdev_dtl_sm); ASSERT(!vd->vdev_ishole); ASSERT(vd->vdev_ops->vdev_op_leaf); tx = dmu_tx_create_assigned(spa->spa_dsl_pool, txg); if (vd->vdev_detached || vd->vdev_top->vdev_removing) { mutex_enter(&vd->vdev_dtl_lock); space_map_free(vd->vdev_dtl_sm, tx); space_map_close(vd->vdev_dtl_sm); vd->vdev_dtl_sm = NULL; mutex_exit(&vd->vdev_dtl_lock); dmu_tx_commit(tx); return; } if (vd->vdev_dtl_sm == NULL) { uint64_t new_object; new_object = space_map_alloc(mos, tx); VERIFY3U(new_object, !=, 0); VERIFY0(space_map_open(&vd->vdev_dtl_sm, mos, new_object, 0, -1ULL, 0, &vd->vdev_dtl_lock)); ASSERT(vd->vdev_dtl_sm != NULL); } mutex_init(&rtlock, NULL, MUTEX_DEFAULT, NULL); rtsync = range_tree_create(NULL, NULL, &rtlock); mutex_enter(&rtlock); mutex_enter(&vd->vdev_dtl_lock); range_tree_walk(rt, range_tree_add, rtsync); mutex_exit(&vd->vdev_dtl_lock); space_map_truncate(vd->vdev_dtl_sm, tx); space_map_write(vd->vdev_dtl_sm, rtsync, SM_ALLOC, tx); range_tree_vacate(rtsync, NULL, NULL); range_tree_destroy(rtsync); mutex_exit(&rtlock); mutex_destroy(&rtlock); /* * If the object for the space map has changed then dirty * the top level so that we update the config. */ if (object != space_map_object(vd->vdev_dtl_sm)) { zfs_dbgmsg("txg %llu, spa %s, DTL old object %llu, " "new object %llu", txg, spa_name(spa), object, space_map_object(vd->vdev_dtl_sm)); vdev_config_dirty(vd->vdev_top); } dmu_tx_commit(tx); mutex_enter(&vd->vdev_dtl_lock); space_map_update(vd->vdev_dtl_sm); mutex_exit(&vd->vdev_dtl_lock); } /* * Determine whether the specified vdev can be offlined/detached/removed * without losing data. */ boolean_t vdev_dtl_required(vdev_t *vd) { spa_t *spa = vd->vdev_spa; vdev_t *tvd = vd->vdev_top; uint8_t cant_read = vd->vdev_cant_read; boolean_t required; ASSERT(spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL); if (vd == spa->spa_root_vdev || vd == tvd) return (B_TRUE); /* * Temporarily mark the device as unreadable, and then determine * whether this results in any DTL outages in the top-level vdev. * If not, we can safely offline/detach/remove the device. */ vd->vdev_cant_read = B_TRUE; vdev_dtl_reassess(tvd, 0, 0, B_FALSE); required = !vdev_dtl_empty(tvd, DTL_OUTAGE); vd->vdev_cant_read = cant_read; vdev_dtl_reassess(tvd, 0, 0, B_FALSE); if (!required && zio_injection_enabled) required = !!zio_handle_device_injection(vd, NULL, ECHILD); return (required); } /* * Determine if resilver is needed, and if so the txg range. */ boolean_t vdev_resilver_needed(vdev_t *vd, uint64_t *minp, uint64_t *maxp) { boolean_t needed = B_FALSE; uint64_t thismin = UINT64_MAX; uint64_t thismax = 0; int c; if (vd->vdev_children == 0) { mutex_enter(&vd->vdev_dtl_lock); if (range_tree_space(vd->vdev_dtl[DTL_MISSING]) != 0 && vdev_writeable(vd)) { thismin = vdev_dtl_min(vd); thismax = vdev_dtl_max(vd); needed = B_TRUE; } mutex_exit(&vd->vdev_dtl_lock); } else { for (c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; uint64_t cmin, cmax; if (vdev_resilver_needed(cvd, &cmin, &cmax)) { thismin = MIN(thismin, cmin); thismax = MAX(thismax, cmax); needed = B_TRUE; } } } if (needed && minp) { *minp = thismin; *maxp = thismax; } return (needed); } void vdev_load(vdev_t *vd) { int c; /* * Recursively load all children. */ for (c = 0; c < vd->vdev_children; c++) vdev_load(vd->vdev_child[c]); /* * If this is a top-level vdev, initialize its metaslabs. */ if (vd == vd->vdev_top && !vd->vdev_ishole && (vd->vdev_ashift == 0 || vd->vdev_asize == 0 || vdev_metaslab_init(vd, 0) != 0)) vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); /* * If this is a leaf vdev, load its DTL. */ if (vd->vdev_ops->vdev_op_leaf && vdev_dtl_load(vd) != 0) vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); } /* * The special vdev case is used for hot spares and l2cache devices. Its * sole purpose it to set the vdev state for the associated vdev. To do this, * we make sure that we can open the underlying device, then try to read the * label, and make sure that the label is sane and that it hasn't been * repurposed to another pool. */ int vdev_validate_aux(vdev_t *vd) { nvlist_t *label; uint64_t guid, version; uint64_t state; if (!vdev_readable(vd)) return (0); if ((label = vdev_label_read_config(vd, -1ULL)) == NULL) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); return (-1); } if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_VERSION, &version) != 0 || !SPA_VERSION_IS_SUPPORTED(version) || nvlist_lookup_uint64(label, ZPOOL_CONFIG_GUID, &guid) != 0 || guid != vd->vdev_guid || nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_STATE, &state) != 0) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); return (-1); } /* * We don't actually check the pool state here. If it's in fact in * use by another pool, we update this fact on the fly when requested. */ nvlist_free(label); return (0); } void vdev_remove(vdev_t *vd, uint64_t txg) { spa_t *spa = vd->vdev_spa; objset_t *mos = spa->spa_meta_objset; dmu_tx_t *tx; int m, i; tx = dmu_tx_create_assigned(spa_get_dsl(spa), txg); if (vd->vdev_ms != NULL) { metaslab_group_t *mg = vd->vdev_mg; metaslab_group_histogram_verify(mg); metaslab_class_histogram_verify(mg->mg_class); for (m = 0; m < vd->vdev_ms_count; m++) { metaslab_t *msp = vd->vdev_ms[m]; if (msp == NULL || msp->ms_sm == NULL) continue; mutex_enter(&msp->ms_lock); /* * If the metaslab was not loaded when the vdev * was removed then the histogram accounting may * not be accurate. Update the histogram information * here so that we ensure that the metaslab group * and metaslab class are up-to-date. */ metaslab_group_histogram_remove(mg, msp); VERIFY0(space_map_allocated(msp->ms_sm)); space_map_free(msp->ms_sm, tx); space_map_close(msp->ms_sm); msp->ms_sm = NULL; mutex_exit(&msp->ms_lock); } metaslab_group_histogram_verify(mg); metaslab_class_histogram_verify(mg->mg_class); for (i = 0; i < RANGE_TREE_HISTOGRAM_SIZE; i++) ASSERT0(mg->mg_histogram[i]); } if (vd->vdev_ms_array) { (void) dmu_object_free(mos, vd->vdev_ms_array, tx); vd->vdev_ms_array = 0; } dmu_tx_commit(tx); } void vdev_sync_done(vdev_t *vd, uint64_t txg) { metaslab_t *msp; boolean_t reassess = !txg_list_empty(&vd->vdev_ms_list, TXG_CLEAN(txg)); ASSERT(!vd->vdev_ishole); while ((msp = txg_list_remove(&vd->vdev_ms_list, TXG_CLEAN(txg)))) metaslab_sync_done(msp, txg); if (reassess) metaslab_sync_reassess(vd->vdev_mg); } void vdev_sync(vdev_t *vd, uint64_t txg) { spa_t *spa = vd->vdev_spa; vdev_t *lvd; metaslab_t *msp; dmu_tx_t *tx; ASSERT(!vd->vdev_ishole); if (vd->vdev_ms_array == 0 && vd->vdev_ms_shift != 0) { ASSERT(vd == vd->vdev_top); tx = dmu_tx_create_assigned(spa->spa_dsl_pool, txg); vd->vdev_ms_array = dmu_object_alloc(spa->spa_meta_objset, DMU_OT_OBJECT_ARRAY, 0, DMU_OT_NONE, 0, tx); ASSERT(vd->vdev_ms_array != 0); vdev_config_dirty(vd); dmu_tx_commit(tx); } /* * Remove the metadata associated with this vdev once it's empty. */ if (vd->vdev_stat.vs_alloc == 0 && vd->vdev_removing) vdev_remove(vd, txg); while ((msp = txg_list_remove(&vd->vdev_ms_list, txg)) != NULL) { metaslab_sync(msp, txg); (void) txg_list_add(&vd->vdev_ms_list, msp, TXG_CLEAN(txg)); } while ((lvd = txg_list_remove(&vd->vdev_dtl_list, txg)) != NULL) vdev_dtl_sync(lvd, txg); (void) txg_list_add(&spa->spa_vdev_txg_list, vd, TXG_CLEAN(txg)); } uint64_t vdev_psize_to_asize(vdev_t *vd, uint64_t psize) { return (vd->vdev_ops->vdev_op_asize(vd, psize)); } /* * Mark the given vdev faulted. A faulted vdev behaves as if the device could * not be opened, and no I/O is attempted. */ int vdev_fault(spa_t *spa, uint64_t guid, vdev_aux_t aux) { vdev_t *vd, *tvd; spa_vdev_state_enter(spa, SCL_NONE); if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL) return (spa_vdev_state_exit(spa, NULL, ENODEV)); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_state_exit(spa, NULL, ENOTSUP)); tvd = vd->vdev_top; /* * We don't directly use the aux state here, but if we do a * vdev_reopen(), we need this value to be present to remember why we * were faulted. */ vd->vdev_label_aux = aux; /* * Faulted state takes precedence over degraded. */ vd->vdev_delayed_close = B_FALSE; vd->vdev_faulted = 1ULL; vd->vdev_degraded = 0ULL; vdev_set_state(vd, B_FALSE, VDEV_STATE_FAULTED, aux); /* * If this device has the only valid copy of the data, then * back off and simply mark the vdev as degraded instead. */ if (!tvd->vdev_islog && vd->vdev_aux == NULL && vdev_dtl_required(vd)) { vd->vdev_degraded = 1ULL; vd->vdev_faulted = 0ULL; /* * If we reopen the device and it's not dead, only then do we * mark it degraded. */ vdev_reopen(tvd); if (vdev_readable(vd)) vdev_set_state(vd, B_FALSE, VDEV_STATE_DEGRADED, aux); } return (spa_vdev_state_exit(spa, vd, 0)); } /* * Mark the given vdev degraded. A degraded vdev is purely an indication to the * user that something is wrong. The vdev continues to operate as normal as far * as I/O is concerned. */ int vdev_degrade(spa_t *spa, uint64_t guid, vdev_aux_t aux) { vdev_t *vd; spa_vdev_state_enter(spa, SCL_NONE); if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL) return (spa_vdev_state_exit(spa, NULL, ENODEV)); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_state_exit(spa, NULL, ENOTSUP)); /* * If the vdev is already faulted, then don't do anything. */ if (vd->vdev_faulted || vd->vdev_degraded) return (spa_vdev_state_exit(spa, NULL, 0)); vd->vdev_degraded = 1ULL; if (!vdev_is_dead(vd)) vdev_set_state(vd, B_FALSE, VDEV_STATE_DEGRADED, aux); return (spa_vdev_state_exit(spa, vd, 0)); } /* * Online the given vdev. * * If 'ZFS_ONLINE_UNSPARE' is set, it implies two things. First, any attached * spare device should be detached when the device finishes resilvering. * Second, the online should be treated like a 'test' online case, so no FMA * events are generated if the device fails to open. */ int vdev_online(spa_t *spa, uint64_t guid, uint64_t flags, vdev_state_t *newstate) { vdev_t *vd, *tvd, *pvd, *rvd = spa->spa_root_vdev; spa_vdev_state_enter(spa, SCL_NONE); if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL) return (spa_vdev_state_exit(spa, NULL, ENODEV)); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_state_exit(spa, NULL, ENOTSUP)); tvd = vd->vdev_top; vd->vdev_offline = B_FALSE; vd->vdev_tmpoffline = B_FALSE; vd->vdev_checkremove = !!(flags & ZFS_ONLINE_CHECKREMOVE); vd->vdev_forcefault = !!(flags & ZFS_ONLINE_FORCEFAULT); /* XXX - L2ARC 1.0 does not support expansion */ if (!vd->vdev_aux) { for (pvd = vd; pvd != rvd; pvd = pvd->vdev_parent) pvd->vdev_expanding = !!(flags & ZFS_ONLINE_EXPAND); } vdev_reopen(tvd); vd->vdev_checkremove = vd->vdev_forcefault = B_FALSE; if (!vd->vdev_aux) { for (pvd = vd; pvd != rvd; pvd = pvd->vdev_parent) pvd->vdev_expanding = B_FALSE; } if (newstate) *newstate = vd->vdev_state; if ((flags & ZFS_ONLINE_UNSPARE) && !vdev_is_dead(vd) && vd->vdev_parent && vd->vdev_parent->vdev_ops == &vdev_spare_ops && vd->vdev_parent->vdev_child[0] == vd) vd->vdev_unspare = B_TRUE; if ((flags & ZFS_ONLINE_EXPAND) || spa->spa_autoexpand) { /* XXX - L2ARC 1.0 does not support expansion */ if (vd->vdev_aux) return (spa_vdev_state_exit(spa, vd, ENOTSUP)); spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE); } return (spa_vdev_state_exit(spa, vd, 0)); } static int vdev_offline_locked(spa_t *spa, uint64_t guid, uint64_t flags) { vdev_t *vd, *tvd; int error = 0; uint64_t generation; metaslab_group_t *mg; top: spa_vdev_state_enter(spa, SCL_ALLOC); if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL) return (spa_vdev_state_exit(spa, NULL, ENODEV)); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_state_exit(spa, NULL, ENOTSUP)); tvd = vd->vdev_top; mg = tvd->vdev_mg; generation = spa->spa_config_generation + 1; /* * If the device isn't already offline, try to offline it. */ if (!vd->vdev_offline) { /* * If this device has the only valid copy of some data, * don't allow it to be offlined. Log devices are always * expendable. */ if (!tvd->vdev_islog && vd->vdev_aux == NULL && vdev_dtl_required(vd)) return (spa_vdev_state_exit(spa, NULL, EBUSY)); /* * If the top-level is a slog and it has had allocations * then proceed. We check that the vdev's metaslab group * is not NULL since it's possible that we may have just * added this vdev but not yet initialized its metaslabs. */ if (tvd->vdev_islog && mg != NULL) { /* * Prevent any future allocations. */ metaslab_group_passivate(mg); (void) spa_vdev_state_exit(spa, vd, 0); error = spa_offline_log(spa); spa_vdev_state_enter(spa, SCL_ALLOC); /* * Check to see if the config has changed. */ if (error || generation != spa->spa_config_generation) { metaslab_group_activate(mg); if (error) return (spa_vdev_state_exit(spa, vd, error)); (void) spa_vdev_state_exit(spa, vd, 0); goto top; } ASSERT0(tvd->vdev_stat.vs_alloc); } /* * Offline this device and reopen its top-level vdev. * If the top-level vdev is a log device then just offline * it. Otherwise, if this action results in the top-level * vdev becoming unusable, undo it and fail the request. */ vd->vdev_offline = B_TRUE; vdev_reopen(tvd); if (!tvd->vdev_islog && vd->vdev_aux == NULL && vdev_is_dead(tvd)) { vd->vdev_offline = B_FALSE; vdev_reopen(tvd); return (spa_vdev_state_exit(spa, NULL, EBUSY)); } /* * Add the device back into the metaslab rotor so that * once we online the device it's open for business. */ if (tvd->vdev_islog && mg != NULL) metaslab_group_activate(mg); } vd->vdev_tmpoffline = !!(flags & ZFS_OFFLINE_TEMPORARY); return (spa_vdev_state_exit(spa, vd, 0)); } int vdev_offline(spa_t *spa, uint64_t guid, uint64_t flags) { int error; mutex_enter(&spa->spa_vdev_top_lock); error = vdev_offline_locked(spa, guid, flags); mutex_exit(&spa->spa_vdev_top_lock); return (error); } /* * Clear the error counts associated with this vdev. Unlike vdev_online() and * vdev_offline(), we assume the spa config is locked. We also clear all * children. If 'vd' is NULL, then the user wants to clear all vdevs. */ void vdev_clear(spa_t *spa, vdev_t *vd) { vdev_t *rvd = spa->spa_root_vdev; int c; ASSERT(spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL); if (vd == NULL) vd = rvd; vd->vdev_stat.vs_read_errors = 0; vd->vdev_stat.vs_write_errors = 0; vd->vdev_stat.vs_checksum_errors = 0; for (c = 0; c < vd->vdev_children; c++) vdev_clear(spa, vd->vdev_child[c]); /* * If we're in the FAULTED state or have experienced failed I/O, then * clear the persistent state and attempt to reopen the device. We * also mark the vdev config dirty, so that the new faulted state is * written out to disk. */ if (vd->vdev_faulted || vd->vdev_degraded || !vdev_readable(vd) || !vdev_writeable(vd)) { /* * When reopening in reponse to a clear event, it may be due to * a fmadm repair request. In this case, if the device is * still broken, we want to still post the ereport again. */ vd->vdev_forcefault = B_TRUE; vd->vdev_faulted = vd->vdev_degraded = 0ULL; vd->vdev_cant_read = B_FALSE; vd->vdev_cant_write = B_FALSE; vdev_reopen(vd == rvd ? rvd : vd->vdev_top); vd->vdev_forcefault = B_FALSE; if (vd != rvd && vdev_writeable(vd->vdev_top)) vdev_state_dirty(vd->vdev_top); if (vd->vdev_aux == NULL && !vdev_is_dead(vd)) spa_async_request(spa, SPA_ASYNC_RESILVER); spa_event_notify(spa, vd, FM_EREPORT_ZFS_DEVICE_CLEAR); } /* * When clearing a FMA-diagnosed fault, we always want to * unspare the device, as we assume that the original spare was * done in response to the FMA fault. */ if (!vdev_is_dead(vd) && vd->vdev_parent != NULL && vd->vdev_parent->vdev_ops == &vdev_spare_ops && vd->vdev_parent->vdev_child[0] == vd) vd->vdev_unspare = B_TRUE; } boolean_t vdev_is_dead(vdev_t *vd) { /* * Holes and missing devices are always considered "dead". * This simplifies the code since we don't have to check for * these types of devices in the various code paths. * Instead we rely on the fact that we skip over dead devices * before issuing I/O to them. */ return (vd->vdev_state < VDEV_STATE_DEGRADED || vd->vdev_ishole || vd->vdev_ops == &vdev_missing_ops); } boolean_t vdev_readable(vdev_t *vd) { return (!vdev_is_dead(vd) && !vd->vdev_cant_read); } boolean_t vdev_writeable(vdev_t *vd) { return (!vdev_is_dead(vd) && !vd->vdev_cant_write); } boolean_t vdev_allocatable(vdev_t *vd) { uint64_t state = vd->vdev_state; /* * We currently allow allocations from vdevs which may be in the * process of reopening (i.e. VDEV_STATE_CLOSED). If the device * fails to reopen then we'll catch it later when we're holding * the proper locks. Note that we have to get the vdev state * in a local variable because although it changes atomically, * we're asking two separate questions about it. */ return (!(state < VDEV_STATE_DEGRADED && state != VDEV_STATE_CLOSED) && !vd->vdev_cant_write && !vd->vdev_ishole); } boolean_t vdev_accessible(vdev_t *vd, zio_t *zio) { ASSERT(zio->io_vd == vd); if (vdev_is_dead(vd) || vd->vdev_remove_wanted) return (B_FALSE); if (zio->io_type == ZIO_TYPE_READ) return (!vd->vdev_cant_read); if (zio->io_type == ZIO_TYPE_WRITE) return (!vd->vdev_cant_write); return (B_TRUE); } /* * Get statistics for the given vdev. */ void vdev_get_stats(vdev_t *vd, vdev_stat_t *vs) { spa_t *spa = vd->vdev_spa; vdev_t *rvd = spa->spa_root_vdev; int c, t; ASSERT(spa_config_held(spa, SCL_ALL, RW_READER) != 0); mutex_enter(&vd->vdev_stat_lock); bcopy(&vd->vdev_stat, vs, sizeof (*vs)); vs->vs_timestamp = gethrtime() - vs->vs_timestamp; vs->vs_state = vd->vdev_state; vs->vs_rsize = vdev_get_min_asize(vd); if (vd->vdev_ops->vdev_op_leaf) vs->vs_rsize += VDEV_LABEL_START_SIZE + VDEV_LABEL_END_SIZE; vs->vs_esize = vd->vdev_max_asize - vd->vdev_asize; if (vd->vdev_aux == NULL && vd == vd->vdev_top && !vd->vdev_ishole) { vs->vs_fragmentation = vd->vdev_mg->mg_fragmentation; } /* * If we're getting stats on the root vdev, aggregate the I/O counts * over all top-level vdevs (i.e. the direct children of the root). */ if (vd == rvd) { for (c = 0; c < rvd->vdev_children; c++) { vdev_t *cvd = rvd->vdev_child[c]; vdev_stat_t *cvs = &cvd->vdev_stat; for (t = 0; t < ZIO_TYPES; t++) { vs->vs_ops[t] += cvs->vs_ops[t]; vs->vs_bytes[t] += cvs->vs_bytes[t]; } cvs->vs_scan_removing = cvd->vdev_removing; } } mutex_exit(&vd->vdev_stat_lock); } void vdev_clear_stats(vdev_t *vd) { mutex_enter(&vd->vdev_stat_lock); vd->vdev_stat.vs_space = 0; vd->vdev_stat.vs_dspace = 0; vd->vdev_stat.vs_alloc = 0; mutex_exit(&vd->vdev_stat_lock); } void vdev_scan_stat_init(vdev_t *vd) { vdev_stat_t *vs = &vd->vdev_stat; int c; for (c = 0; c < vd->vdev_children; c++) vdev_scan_stat_init(vd->vdev_child[c]); mutex_enter(&vd->vdev_stat_lock); vs->vs_scan_processed = 0; mutex_exit(&vd->vdev_stat_lock); } void vdev_stat_update(zio_t *zio, uint64_t psize) { spa_t *spa = zio->io_spa; vdev_t *rvd = spa->spa_root_vdev; vdev_t *vd = zio->io_vd ? zio->io_vd : rvd; vdev_t *pvd; uint64_t txg = zio->io_txg; vdev_stat_t *vs = &vd->vdev_stat; zio_type_t type = zio->io_type; int flags = zio->io_flags; /* * If this i/o is a gang leader, it didn't do any actual work. */ if (zio->io_gang_tree) return; if (zio->io_error == 0) { /* * If this is a root i/o, don't count it -- we've already * counted the top-level vdevs, and vdev_get_stats() will * aggregate them when asked. This reduces contention on * the root vdev_stat_lock and implicitly handles blocks * that compress away to holes, for which there is no i/o. * (Holes never create vdev children, so all the counters * remain zero, which is what we want.) * * Note: this only applies to successful i/o (io_error == 0) * because unlike i/o counts, errors are not additive. * When reading a ditto block, for example, failure of * one top-level vdev does not imply a root-level error. */ if (vd == rvd) return; ASSERT(vd == zio->io_vd); if (flags & ZIO_FLAG_IO_BYPASS) return; mutex_enter(&vd->vdev_stat_lock); if (flags & ZIO_FLAG_IO_REPAIR) { if (flags & ZIO_FLAG_SCAN_THREAD) { dsl_scan_phys_t *scn_phys = &spa->spa_dsl_pool->dp_scan->scn_phys; uint64_t *processed = &scn_phys->scn_processed; /* XXX cleanup? */ if (vd->vdev_ops->vdev_op_leaf) atomic_add_64(processed, psize); vs->vs_scan_processed += psize; } if (flags & ZIO_FLAG_SELF_HEAL) vs->vs_self_healed += psize; } vs->vs_ops[type]++; vs->vs_bytes[type] += psize; mutex_exit(&vd->vdev_stat_lock); return; } if (flags & ZIO_FLAG_SPECULATIVE) return; /* * If this is an I/O error that is going to be retried, then ignore the * error. Otherwise, the user may interpret B_FAILFAST I/O errors as * hard errors, when in reality they can happen for any number of * innocuous reasons (bus resets, MPxIO link failure, etc). */ if (zio->io_error == EIO && !(zio->io_flags & ZIO_FLAG_IO_RETRY)) return; /* * Intent logs writes won't propagate their error to the root * I/O so don't mark these types of failures as pool-level * errors. */ if (zio->io_vd == NULL && (zio->io_flags & ZIO_FLAG_DONT_PROPAGATE)) return; mutex_enter(&vd->vdev_stat_lock); if (type == ZIO_TYPE_READ && !vdev_is_dead(vd)) { if (zio->io_error == ECKSUM) vs->vs_checksum_errors++; else vs->vs_read_errors++; } if (type == ZIO_TYPE_WRITE && !vdev_is_dead(vd)) vs->vs_write_errors++; mutex_exit(&vd->vdev_stat_lock); if (type == ZIO_TYPE_WRITE && txg != 0 && (!(flags & ZIO_FLAG_IO_REPAIR) || (flags & ZIO_FLAG_SCAN_THREAD) || spa->spa_claiming)) { /* * This is either a normal write (not a repair), or it's * a repair induced by the scrub thread, or it's a repair * made by zil_claim() during spa_load() in the first txg. * In the normal case, we commit the DTL change in the same * txg as the block was born. In the scrub-induced repair * case, we know that scrubs run in first-pass syncing context, * so we commit the DTL change in spa_syncing_txg(spa). * In the zil_claim() case, we commit in spa_first_txg(spa). * * We currently do not make DTL entries for failed spontaneous * self-healing writes triggered by normal (non-scrubbing) * reads, because we have no transactional context in which to * do so -- and it's not clear that it'd be desirable anyway. */ if (vd->vdev_ops->vdev_op_leaf) { uint64_t commit_txg = txg; if (flags & ZIO_FLAG_SCAN_THREAD) { ASSERT(flags & ZIO_FLAG_IO_REPAIR); ASSERT(spa_sync_pass(spa) == 1); vdev_dtl_dirty(vd, DTL_SCRUB, txg, 1); commit_txg = spa_syncing_txg(spa); } else if (spa->spa_claiming) { ASSERT(flags & ZIO_FLAG_IO_REPAIR); commit_txg = spa_first_txg(spa); } ASSERT(commit_txg >= spa_syncing_txg(spa)); if (vdev_dtl_contains(vd, DTL_MISSING, txg, 1)) return; for (pvd = vd; pvd != rvd; pvd = pvd->vdev_parent) vdev_dtl_dirty(pvd, DTL_PARTIAL, txg, 1); vdev_dirty(vd->vdev_top, VDD_DTL, vd, commit_txg); } if (vd != rvd) vdev_dtl_dirty(vd, DTL_MISSING, txg, 1); } } /* * Update the in-core space usage stats for this vdev, its metaslab class, * and the root vdev. */ void vdev_space_update(vdev_t *vd, int64_t alloc_delta, int64_t defer_delta, int64_t space_delta) { int64_t dspace_delta = space_delta; spa_t *spa = vd->vdev_spa; vdev_t *rvd = spa->spa_root_vdev; metaslab_group_t *mg = vd->vdev_mg; metaslab_class_t *mc = mg ? mg->mg_class : NULL; ASSERT(vd == vd->vdev_top); /* * Apply the inverse of the psize-to-asize (ie. RAID-Z) space-expansion * factor. We must calculate this here and not at the root vdev * because the root vdev's psize-to-asize is simply the max of its * childrens', thus not accurate enough for us. */ ASSERT((dspace_delta & (SPA_MINBLOCKSIZE-1)) == 0); ASSERT(vd->vdev_deflate_ratio != 0 || vd->vdev_isl2cache); dspace_delta = (dspace_delta >> SPA_MINBLOCKSHIFT) * vd->vdev_deflate_ratio; mutex_enter(&vd->vdev_stat_lock); vd->vdev_stat.vs_alloc += alloc_delta; vd->vdev_stat.vs_space += space_delta; vd->vdev_stat.vs_dspace += dspace_delta; mutex_exit(&vd->vdev_stat_lock); if (mc == spa_normal_class(spa)) { mutex_enter(&rvd->vdev_stat_lock); rvd->vdev_stat.vs_alloc += alloc_delta; rvd->vdev_stat.vs_space += space_delta; rvd->vdev_stat.vs_dspace += dspace_delta; mutex_exit(&rvd->vdev_stat_lock); } if (mc != NULL) { ASSERT(rvd == vd->vdev_parent); ASSERT(vd->vdev_ms_count != 0); metaslab_class_space_update(mc, alloc_delta, defer_delta, space_delta, dspace_delta); } } /* * Mark a top-level vdev's config as dirty, placing it on the dirty list * so that it will be written out next time the vdev configuration is synced. * If the root vdev is specified (vdev_top == NULL), dirty all top-level vdevs. */ void vdev_config_dirty(vdev_t *vd) { spa_t *spa = vd->vdev_spa; vdev_t *rvd = spa->spa_root_vdev; int c; ASSERT(spa_writeable(spa)); /* * If this is an aux vdev (as with l2cache and spare devices), then we * update the vdev config manually and set the sync flag. */ if (vd->vdev_aux != NULL) { spa_aux_vdev_t *sav = vd->vdev_aux; nvlist_t **aux; uint_t naux; for (c = 0; c < sav->sav_count; c++) { if (sav->sav_vdevs[c] == vd) break; } if (c == sav->sav_count) { /* * We're being removed. There's nothing more to do. */ ASSERT(sav->sav_sync == B_TRUE); return; } sav->sav_sync = B_TRUE; if (nvlist_lookup_nvlist_array(sav->sav_config, ZPOOL_CONFIG_L2CACHE, &aux, &naux) != 0) { VERIFY(nvlist_lookup_nvlist_array(sav->sav_config, ZPOOL_CONFIG_SPARES, &aux, &naux) == 0); } ASSERT(c < naux); /* * Setting the nvlist in the middle if the array is a little * sketchy, but it will work. */ nvlist_free(aux[c]); aux[c] = vdev_config_generate(spa, vd, B_TRUE, 0); return; } /* * The dirty list is protected by the SCL_CONFIG lock. The caller * must either hold SCL_CONFIG as writer, or must be the sync thread * (which holds SCL_CONFIG as reader). There's only one sync thread, * so this is sufficient to ensure mutual exclusion. */ ASSERT(spa_config_held(spa, SCL_CONFIG, RW_WRITER) || (dsl_pool_sync_context(spa_get_dsl(spa)) && spa_config_held(spa, SCL_CONFIG, RW_READER))); if (vd == rvd) { for (c = 0; c < rvd->vdev_children; c++) vdev_config_dirty(rvd->vdev_child[c]); } else { ASSERT(vd == vd->vdev_top); if (!list_link_active(&vd->vdev_config_dirty_node) && !vd->vdev_ishole) list_insert_head(&spa->spa_config_dirty_list, vd); } } void vdev_config_clean(vdev_t *vd) { spa_t *spa = vd->vdev_spa; ASSERT(spa_config_held(spa, SCL_CONFIG, RW_WRITER) || (dsl_pool_sync_context(spa_get_dsl(spa)) && spa_config_held(spa, SCL_CONFIG, RW_READER))); ASSERT(list_link_active(&vd->vdev_config_dirty_node)); list_remove(&spa->spa_config_dirty_list, vd); } /* * Mark a top-level vdev's state as dirty, so that the next pass of * spa_sync() can convert this into vdev_config_dirty(). We distinguish * the state changes from larger config changes because they require * much less locking, and are often needed for administrative actions. */ void vdev_state_dirty(vdev_t *vd) { spa_t *spa = vd->vdev_spa; ASSERT(spa_writeable(spa)); ASSERT(vd == vd->vdev_top); /* * The state list is protected by the SCL_STATE lock. The caller * must either hold SCL_STATE as writer, or must be the sync thread * (which holds SCL_STATE as reader). There's only one sync thread, * so this is sufficient to ensure mutual exclusion. */ ASSERT(spa_config_held(spa, SCL_STATE, RW_WRITER) || (dsl_pool_sync_context(spa_get_dsl(spa)) && spa_config_held(spa, SCL_STATE, RW_READER))); if (!list_link_active(&vd->vdev_state_dirty_node) && !vd->vdev_ishole) list_insert_head(&spa->spa_state_dirty_list, vd); } void vdev_state_clean(vdev_t *vd) { spa_t *spa = vd->vdev_spa; ASSERT(spa_config_held(spa, SCL_STATE, RW_WRITER) || (dsl_pool_sync_context(spa_get_dsl(spa)) && spa_config_held(spa, SCL_STATE, RW_READER))); ASSERT(list_link_active(&vd->vdev_state_dirty_node)); list_remove(&spa->spa_state_dirty_list, vd); } /* * Propagate vdev state up from children to parent. */ void vdev_propagate_state(vdev_t *vd) { spa_t *spa = vd->vdev_spa; vdev_t *rvd = spa->spa_root_vdev; int degraded = 0, faulted = 0; int corrupted = 0; vdev_t *child; int c; if (vd->vdev_children > 0) { for (c = 0; c < vd->vdev_children; c++) { child = vd->vdev_child[c]; /* * Don't factor holes into the decision. */ if (child->vdev_ishole) continue; if (!vdev_readable(child) || (!vdev_writeable(child) && spa_writeable(spa))) { /* * Root special: if there is a top-level log * device, treat the root vdev as if it were * degraded. */ if (child->vdev_islog && vd == rvd) degraded++; else faulted++; } else if (child->vdev_state <= VDEV_STATE_DEGRADED) { degraded++; } if (child->vdev_stat.vs_aux == VDEV_AUX_CORRUPT_DATA) corrupted++; } vd->vdev_ops->vdev_op_state_change(vd, faulted, degraded); /* * Root special: if there is a top-level vdev that cannot be * opened due to corrupted metadata, then propagate the root * vdev's aux state as 'corrupt' rather than 'insufficient * replicas'. */ if (corrupted && vd == rvd && rvd->vdev_state == VDEV_STATE_CANT_OPEN) vdev_set_state(rvd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); } if (vd->vdev_parent) vdev_propagate_state(vd->vdev_parent); } /* * Set a vdev's state. If this is during an open, we don't update the parent * state, because we're in the process of opening children depth-first. * Otherwise, we propagate the change to the parent. * * If this routine places a device in a faulted state, an appropriate ereport is * generated. */ void vdev_set_state(vdev_t *vd, boolean_t isopen, vdev_state_t state, vdev_aux_t aux) { uint64_t save_state; spa_t *spa = vd->vdev_spa; if (state == vd->vdev_state) { vd->vdev_stat.vs_aux = aux; return; } save_state = vd->vdev_state; vd->vdev_state = state; vd->vdev_stat.vs_aux = aux; /* * If we are setting the vdev state to anything but an open state, then * always close the underlying device unless the device has requested * a delayed close (i.e. we're about to remove or fault the device). * Otherwise, we keep accessible but invalid devices open forever. * We don't call vdev_close() itself, because that implies some extra * checks (offline, etc) that we don't want here. This is limited to * leaf devices, because otherwise closing the device will affect other * children. */ if (!vd->vdev_delayed_close && vdev_is_dead(vd) && vd->vdev_ops->vdev_op_leaf) vd->vdev_ops->vdev_op_close(vd); /* * If we have brought this vdev back into service, we need * to notify fmd so that it can gracefully repair any outstanding * cases due to a missing device. We do this in all cases, even those * that probably don't correlate to a repaired fault. This is sure to * catch all cases, and we let the zfs-retire agent sort it out. If * this is a transient state it's OK, as the retire agent will * double-check the state of the vdev before repairing it. */ if (state == VDEV_STATE_HEALTHY && vd->vdev_ops->vdev_op_leaf && vd->vdev_prevstate != state) zfs_post_state_change(spa, vd); if (vd->vdev_removed && state == VDEV_STATE_CANT_OPEN && (aux == VDEV_AUX_OPEN_FAILED || vd->vdev_checkremove)) { /* * If the previous state is set to VDEV_STATE_REMOVED, then this * device was previously marked removed and someone attempted to * reopen it. If this failed due to a nonexistent device, then * keep the device in the REMOVED state. We also let this be if * it is one of our special test online cases, which is only * attempting to online the device and shouldn't generate an FMA * fault. */ vd->vdev_state = VDEV_STATE_REMOVED; vd->vdev_stat.vs_aux = VDEV_AUX_NONE; } else if (state == VDEV_STATE_REMOVED) { vd->vdev_removed = B_TRUE; } else if (state == VDEV_STATE_CANT_OPEN) { /* * If we fail to open a vdev during an import or recovery, we * mark it as "not available", which signifies that it was * never there to begin with. Failure to open such a device * is not considered an error. */ if ((spa_load_state(spa) == SPA_LOAD_IMPORT || spa_load_state(spa) == SPA_LOAD_RECOVER) && vd->vdev_ops->vdev_op_leaf) vd->vdev_not_present = 1; /* * Post the appropriate ereport. If the 'prevstate' field is * set to something other than VDEV_STATE_UNKNOWN, it indicates * that this is part of a vdev_reopen(). In this case, we don't * want to post the ereport if the device was already in the * CANT_OPEN state beforehand. * * If the 'checkremove' flag is set, then this is an attempt to * online the device in response to an insertion event. If we * hit this case, then we have detected an insertion event for a * faulted or offline device that wasn't in the removed state. * In this scenario, we don't post an ereport because we are * about to replace the device, or attempt an online with * vdev_forcefault, which will generate the fault for us. */ if ((vd->vdev_prevstate != state || vd->vdev_forcefault) && !vd->vdev_not_present && !vd->vdev_checkremove && vd != spa->spa_root_vdev) { const char *class; switch (aux) { case VDEV_AUX_OPEN_FAILED: class = FM_EREPORT_ZFS_DEVICE_OPEN_FAILED; break; case VDEV_AUX_CORRUPT_DATA: class = FM_EREPORT_ZFS_DEVICE_CORRUPT_DATA; break; case VDEV_AUX_NO_REPLICAS: class = FM_EREPORT_ZFS_DEVICE_NO_REPLICAS; break; case VDEV_AUX_BAD_GUID_SUM: class = FM_EREPORT_ZFS_DEVICE_BAD_GUID_SUM; break; case VDEV_AUX_TOO_SMALL: class = FM_EREPORT_ZFS_DEVICE_TOO_SMALL; break; case VDEV_AUX_BAD_LABEL: class = FM_EREPORT_ZFS_DEVICE_BAD_LABEL; break; default: class = FM_EREPORT_ZFS_DEVICE_UNKNOWN; } zfs_ereport_post(class, spa, vd, NULL, save_state, 0); } /* Erase any notion of persistent removed state */ vd->vdev_removed = B_FALSE; } else { vd->vdev_removed = B_FALSE; } if (!isopen && vd->vdev_parent) vdev_propagate_state(vd->vdev_parent); } /* * Check the vdev configuration to ensure that it's capable of supporting * a root pool. */ boolean_t vdev_is_bootable(vdev_t *vd) { #if defined(__sun__) || defined(__sun) /* * Currently, we do not support RAID-Z or partial configuration. * In addition, only a single top-level vdev is allowed and none of the * leaves can be wholedisks. */ int c; if (!vd->vdev_ops->vdev_op_leaf) { char *vdev_type = vd->vdev_ops->vdev_op_type; if (strcmp(vdev_type, VDEV_TYPE_ROOT) == 0 && vd->vdev_children > 1) { return (B_FALSE); } else if (strcmp(vdev_type, VDEV_TYPE_RAIDZ) == 0 || strcmp(vdev_type, VDEV_TYPE_MISSING) == 0) { return (B_FALSE); } } else if (vd->vdev_wholedisk == 1) { return (B_FALSE); } for (c = 0; c < vd->vdev_children; c++) { if (!vdev_is_bootable(vd->vdev_child[c])) return (B_FALSE); } #endif /* __sun__ || __sun */ return (B_TRUE); } /* * Load the state from the original vdev tree (ovd) which * we've retrieved from the MOS config object. If the original * vdev was offline or faulted then we transfer that state to the * device in the current vdev tree (nvd). */ void vdev_load_log_state(vdev_t *nvd, vdev_t *ovd) { int c; ASSERT(nvd->vdev_top->vdev_islog); ASSERT(spa_config_held(nvd->vdev_spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL); ASSERT3U(nvd->vdev_guid, ==, ovd->vdev_guid); for (c = 0; c < nvd->vdev_children; c++) vdev_load_log_state(nvd->vdev_child[c], ovd->vdev_child[c]); if (nvd->vdev_ops->vdev_op_leaf) { /* * Restore the persistent vdev state */ nvd->vdev_offline = ovd->vdev_offline; nvd->vdev_faulted = ovd->vdev_faulted; nvd->vdev_degraded = ovd->vdev_degraded; nvd->vdev_removed = ovd->vdev_removed; } } /* * Determine if a log device has valid content. If the vdev was * removed or faulted in the MOS config then we know that * the content on the log device has already been written to the pool. */ boolean_t vdev_log_state_valid(vdev_t *vd) { int c; if (vd->vdev_ops->vdev_op_leaf && !vd->vdev_faulted && !vd->vdev_removed) return (B_TRUE); for (c = 0; c < vd->vdev_children; c++) if (vdev_log_state_valid(vd->vdev_child[c])) return (B_TRUE); return (B_FALSE); } /* * Expand a vdev if possible. */ void vdev_expand(vdev_t *vd, uint64_t txg) { ASSERT(vd->vdev_top == vd); ASSERT(spa_config_held(vd->vdev_spa, SCL_ALL, RW_WRITER) == SCL_ALL); if ((vd->vdev_asize >> vd->vdev_ms_shift) > vd->vdev_ms_count) { VERIFY(vdev_metaslab_init(vd, txg) == 0); vdev_config_dirty(vd); } } /* * Split a vdev. */ void vdev_split(vdev_t *vd) { vdev_t *cvd, *pvd = vd->vdev_parent; vdev_remove_child(pvd, vd); vdev_compact_children(pvd); cvd = pvd->vdev_child[0]; if (pvd->vdev_children == 1) { vdev_remove_parent(cvd); cvd->vdev_splitting = B_TRUE; } vdev_propagate_state(cvd); } void vdev_deadman(vdev_t *vd) { int c; for (c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; vdev_deadman(cvd); } if (vd->vdev_ops->vdev_op_leaf) { vdev_queue_t *vq = &vd->vdev_queue; mutex_enter(&vq->vq_lock); if (avl_numnodes(&vq->vq_active_tree) > 0) { spa_t *spa = vd->vdev_spa; zio_t *fio; uint64_t delta; /* * Look at the head of all the pending queues, * if any I/O has been outstanding for longer than * the spa_deadman_synctime we log a zevent. */ fio = avl_first(&vq->vq_active_tree); delta = gethrtime() - fio->io_timestamp; if (delta > spa_deadman_synctime(spa)) { zfs_dbgmsg("SLOW IO: zio timestamp %lluns, " "delta %lluns, last io %lluns", fio->io_timestamp, delta, vq->vq_io_complete_ts); zfs_ereport_post(FM_EREPORT_ZFS_DELAY, spa, vd, fio, 0, 0); } } mutex_exit(&vq->vq_lock); } } #if defined(_KERNEL) && defined(HAVE_SPL) EXPORT_SYMBOL(vdev_fault); EXPORT_SYMBOL(vdev_degrade); EXPORT_SYMBOL(vdev_online); EXPORT_SYMBOL(vdev_offline); EXPORT_SYMBOL(vdev_clear); module_param(metaslabs_per_vdev, int, 0644); MODULE_PARM_DESC(metaslabs_per_vdev, "Divide added vdev into approximately (but no more than) this number " "of metaslabs"); #endif diff --git a/module/zfs/zfs_ioctl.c b/module/zfs/zfs_ioctl.c index 5fb407900adf..7d443ecf1854 100644 --- a/module/zfs/zfs_ioctl.c +++ b/module/zfs/zfs_ioctl.c @@ -1,6026 +1,6025 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Portions Copyright 2011 Martin Matuska * Portions Copyright 2012 Pawel Jakub Dawidek * Copyright (c) 2012, Joyent, Inc. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2012, Joyent, Inc. All rights reserved. * Copyright (c) 201i3 by Delphix. All rights reserved. * Copyright (c) 2013 by Saso Kiselkov. All rights reserved. * Copyright (c) 2013 Steven Hartland. All rights reserved. * Copyright (c) 2014, Nexenta Systems, Inc. All rights reserved. */ /* * ZFS ioctls. * * This file handles the ioctls to /dev/zfs, used for configuring ZFS storage * pools and filesystems, e.g. with /sbin/zfs and /sbin/zpool. * * There are two ways that we handle ioctls: the legacy way where almost * all of the logic is in the ioctl callback, and the new way where most * of the marshalling is handled in the common entry point, zfsdev_ioctl(). * * Non-legacy ioctls should be registered by calling * zfs_ioctl_register() from zfs_ioctl_init(). The ioctl is invoked * from userland by lzc_ioctl(). * * The registration arguments are as follows: * * const char *name * The name of the ioctl. This is used for history logging. If the * ioctl returns successfully (the callback returns 0), and allow_log * is true, then a history log entry will be recorded with the input & * output nvlists. The log entry can be printed with "zpool history -i". * * zfs_ioc_t ioc * The ioctl request number, which userland will pass to ioctl(2). * The ioctl numbers can change from release to release, because * the caller (libzfs) must be matched to the kernel. * * zfs_secpolicy_func_t *secpolicy * This function will be called before the zfs_ioc_func_t, to * determine if this operation is permitted. It should return EPERM * on failure, and 0 on success. Checks include determining if the * dataset is visible in this zone, and if the user has either all * zfs privileges in the zone (SYS_MOUNT), or has been granted permission * to do this operation on this dataset with "zfs allow". * * zfs_ioc_namecheck_t namecheck * This specifies what to expect in the zfs_cmd_t:zc_name -- a pool * name, a dataset name, or nothing. If the name is not well-formed, * the ioctl will fail and the callback will not be called. * Therefore, the callback can assume that the name is well-formed * (e.g. is null-terminated, doesn't have more than one '@' character, * doesn't have invalid characters). * * zfs_ioc_poolcheck_t pool_check * This specifies requirements on the pool state. If the pool does * not meet them (is suspended or is readonly), the ioctl will fail * and the callback will not be called. If any checks are specified * (i.e. it is not POOL_CHECK_NONE), namecheck must not be NO_NAME. * Multiple checks can be or-ed together (e.g. POOL_CHECK_SUSPENDED | * POOL_CHECK_READONLY). * * boolean_t smush_outnvlist * If smush_outnvlist is true, then the output is presumed to be a * list of errors, and it will be "smushed" down to fit into the * caller's buffer, by removing some entries and replacing them with a * single "N_MORE_ERRORS" entry indicating how many were removed. See * nvlist_smush() for details. If smush_outnvlist is false, and the * outnvlist does not fit into the userland-provided buffer, then the * ioctl will fail with ENOMEM. * * zfs_ioc_func_t *func * The callback function that will perform the operation. * * The callback should return 0 on success, or an error number on * failure. If the function fails, the userland ioctl will return -1, * and errno will be set to the callback's return value. The callback * will be called with the following arguments: * * const char *name * The name of the pool or dataset to operate on, from * zfs_cmd_t:zc_name. The 'namecheck' argument specifies the * expected type (pool, dataset, or none). * * nvlist_t *innvl * The input nvlist, deserialized from zfs_cmd_t:zc_nvlist_src. Or * NULL if no input nvlist was provided. Changes to this nvlist are * ignored. If the input nvlist could not be deserialized, the * ioctl will fail and the callback will not be called. * * nvlist_t *outnvl * The output nvlist, initially empty. The callback can fill it in, * and it will be returned to userland by serializing it into * zfs_cmd_t:zc_nvlist_dst. If it is non-empty, and serialization * fails (e.g. because the caller didn't supply a large enough * buffer), then the overall ioctl will fail. See the * 'smush_nvlist' argument above for additional behaviors. * * There are two typical uses of the output nvlist: * - To return state, e.g. property values. In this case, * smush_outnvlist should be false. If the buffer was not large * enough, the caller will reallocate a larger buffer and try * the ioctl again. * * - To return multiple errors from an ioctl which makes on-disk * changes. In this case, smush_outnvlist should be true. * Ioctls which make on-disk modifications should generally not * use the outnvl if they succeed, because the caller can not * distinguish between the operation failing, and * deserialization failing. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "zfs_namecheck.h" #include "zfs_prop.h" #include "zfs_deleg.h" #include "zfs_comutil.h" kmutex_t zfsdev_state_lock; zfsdev_state_t *zfsdev_state_list; extern void zfs_init(void); extern void zfs_fini(void); uint_t zfs_fsyncer_key; extern uint_t rrw_tsd_key; static uint_t zfs_allow_log_key; typedef int zfs_ioc_legacy_func_t(zfs_cmd_t *); typedef int zfs_ioc_func_t(const char *, nvlist_t *, nvlist_t *); typedef int zfs_secpolicy_func_t(zfs_cmd_t *, nvlist_t *, cred_t *); typedef enum { NO_NAME, POOL_NAME, DATASET_NAME } zfs_ioc_namecheck_t; typedef enum { POOL_CHECK_NONE = 1 << 0, POOL_CHECK_SUSPENDED = 1 << 1, POOL_CHECK_READONLY = 1 << 2, } zfs_ioc_poolcheck_t; typedef struct zfs_ioc_vec { zfs_ioc_legacy_func_t *zvec_legacy_func; zfs_ioc_func_t *zvec_func; zfs_secpolicy_func_t *zvec_secpolicy; zfs_ioc_namecheck_t zvec_namecheck; boolean_t zvec_allow_log; zfs_ioc_poolcheck_t zvec_pool_check; boolean_t zvec_smush_outnvlist; const char *zvec_name; } zfs_ioc_vec_t; /* This array is indexed by zfs_userquota_prop_t */ static const char *userquota_perms[] = { ZFS_DELEG_PERM_USERUSED, ZFS_DELEG_PERM_USERQUOTA, ZFS_DELEG_PERM_GROUPUSED, ZFS_DELEG_PERM_GROUPQUOTA, }; static int zfs_ioc_userspace_upgrade(zfs_cmd_t *zc); static int zfs_check_settable(const char *name, nvpair_t *property, cred_t *cr); static int zfs_check_clearable(char *dataset, nvlist_t *props, nvlist_t **errors); static int zfs_fill_zplprops_root(uint64_t, nvlist_t *, nvlist_t *, boolean_t *); int zfs_set_prop_nvlist(const char *, zprop_source_t, nvlist_t *, nvlist_t *); static int get_nvlist(uint64_t nvl, uint64_t size, int iflag, nvlist_t **nvp); #if defined(HAVE_DECLARE_EVENT_CLASS) void __dprintf(const char *file, const char *func, int line, const char *fmt, ...) { const char *newfile; size_t size = 4096; char *buf = kmem_alloc(size, KM_PUSHPAGE); char *nl; va_list adx; /* * Get rid of annoying prefix to filename. */ newfile = strrchr(file, '/'); if (newfile != NULL) { newfile = newfile + 1; /* Get rid of leading / */ } else { newfile = file; } va_start(adx, fmt); (void) vsnprintf(buf, size, fmt, adx); va_end(adx); /* * Get rid of trailing newline. */ nl = strrchr(buf, '\n'); if (nl != NULL) *nl = '\0'; /* * To get this data enable the zfs__dprintf trace point as shown: * * # Enable zfs__dprintf tracepoint, clear the tracepoint ring buffer * $ echo 1 > /sys/module/zfs/parameters/zfs_flags * $ echo 1 > /sys/kernel/debug/tracing/events/zfs/enable * $ echo 0 > /sys/kernel/debug/tracing/trace * * # Dump the ring buffer. * $ cat /sys/kernel/debug/tracing/trace */ DTRACE_PROBE4(zfs__dprintf, char *, newfile, char *, func, int, line, char *, buf); kmem_free(buf, size); } #endif /* HAVE_DECLARE_EVENT_CLASS */ static void history_str_free(char *buf) { kmem_free(buf, HIS_MAX_RECORD_LEN); } static char * history_str_get(zfs_cmd_t *zc) { char *buf; if (zc->zc_history == 0) return (NULL); - buf = kmem_alloc(HIS_MAX_RECORD_LEN, KM_SLEEP | KM_NODEBUG); + buf = kmem_alloc(HIS_MAX_RECORD_LEN, KM_SLEEP); if (copyinstr((void *)(uintptr_t)zc->zc_history, buf, HIS_MAX_RECORD_LEN, NULL) != 0) { history_str_free(buf); return (NULL); } buf[HIS_MAX_RECORD_LEN -1] = '\0'; return (buf); } /* * Check to see if the named dataset is currently defined as bootable */ static boolean_t zfs_is_bootfs(const char *name) { objset_t *os; if (dmu_objset_hold(name, FTAG, &os) == 0) { boolean_t ret; ret = (dmu_objset_id(os) == spa_bootfs(dmu_objset_spa(os))); dmu_objset_rele(os, FTAG); return (ret); } return (B_FALSE); } /* * Return non-zero if the spa version is less than requested version. */ static int zfs_earlier_version(const char *name, int version) { spa_t *spa; if (spa_open(name, &spa, FTAG) == 0) { if (spa_version(spa) < version) { spa_close(spa, FTAG); return (1); } spa_close(spa, FTAG); } return (0); } /* * Return TRUE if the ZPL version is less than requested version. */ static boolean_t zpl_earlier_version(const char *name, int version) { objset_t *os; boolean_t rc = B_TRUE; if (dmu_objset_hold(name, FTAG, &os) == 0) { uint64_t zplversion; if (dmu_objset_type(os) != DMU_OST_ZFS) { dmu_objset_rele(os, FTAG); return (B_TRUE); } /* XXX reading from non-owned objset */ if (zfs_get_zplprop(os, ZFS_PROP_VERSION, &zplversion) == 0) rc = zplversion < version; dmu_objset_rele(os, FTAG); } return (rc); } static void zfs_log_history(zfs_cmd_t *zc) { spa_t *spa; char *buf; if ((buf = history_str_get(zc)) == NULL) return; if (spa_open(zc->zc_name, &spa, FTAG) == 0) { if (spa_version(spa) >= SPA_VERSION_ZPOOL_HISTORY) (void) spa_history_log(spa, buf); spa_close(spa, FTAG); } history_str_free(buf); } /* * Policy for top-level read operations (list pools). Requires no privileges, * and can be used in the local zone, as there is no associated dataset. */ /* ARGSUSED */ static int zfs_secpolicy_none(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (0); } /* * Policy for dataset read operations (list children, get statistics). Requires * no privileges, but must be visible in the local zone. */ /* ARGSUSED */ static int zfs_secpolicy_read(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { if (INGLOBALZONE(curproc) || zone_dataset_visible(zc->zc_name, NULL)) return (0); return (SET_ERROR(ENOENT)); } static int zfs_dozonecheck_impl(const char *dataset, uint64_t zoned, cred_t *cr) { int writable = 1; /* * The dataset must be visible by this zone -- check this first * so they don't see EPERM on something they shouldn't know about. */ if (!INGLOBALZONE(curproc) && !zone_dataset_visible(dataset, &writable)) return (SET_ERROR(ENOENT)); if (INGLOBALZONE(curproc)) { /* * If the fs is zoned, only root can access it from the * global zone. */ if (secpolicy_zfs(cr) && zoned) return (SET_ERROR(EPERM)); } else { /* * If we are in a local zone, the 'zoned' property must be set. */ if (!zoned) return (SET_ERROR(EPERM)); /* must be writable by this zone */ if (!writable) return (SET_ERROR(EPERM)); } return (0); } static int zfs_dozonecheck(const char *dataset, cred_t *cr) { uint64_t zoned; if (dsl_prop_get_integer(dataset, "zoned", &zoned, NULL)) return (SET_ERROR(ENOENT)); return (zfs_dozonecheck_impl(dataset, zoned, cr)); } static int zfs_dozonecheck_ds(const char *dataset, dsl_dataset_t *ds, cred_t *cr) { uint64_t zoned; if (dsl_prop_get_int_ds(ds, "zoned", &zoned)) return (SET_ERROR(ENOENT)); return (zfs_dozonecheck_impl(dataset, zoned, cr)); } static int zfs_secpolicy_write_perms_ds(const char *name, dsl_dataset_t *ds, const char *perm, cred_t *cr) { int error; error = zfs_dozonecheck_ds(name, ds, cr); if (error == 0) { error = secpolicy_zfs(cr); if (error != 0) error = dsl_deleg_access_impl(ds, perm, cr); } return (error); } static int zfs_secpolicy_write_perms(const char *name, const char *perm, cred_t *cr) { int error; dsl_dataset_t *ds; dsl_pool_t *dp; error = dsl_pool_hold(name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, name, FTAG, &ds); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } error = zfs_secpolicy_write_perms_ds(name, ds, perm, cr); dsl_dataset_rele(ds, FTAG); dsl_pool_rele(dp, FTAG); return (error); } /* * Policy for setting the security label property. * * Returns 0 for success, non-zero for access and other errors. */ static int zfs_set_slabel_policy(const char *name, char *strval, cred_t *cr) { #ifdef HAVE_MLSLABEL char ds_hexsl[MAXNAMELEN]; bslabel_t ds_sl, new_sl; boolean_t new_default = FALSE; uint64_t zoned; int needed_priv = -1; int error; /* First get the existing dataset label. */ error = dsl_prop_get(name, zfs_prop_to_name(ZFS_PROP_MLSLABEL), 1, sizeof (ds_hexsl), &ds_hexsl, NULL); if (error != 0) return (SET_ERROR(EPERM)); if (strcasecmp(strval, ZFS_MLSLABEL_DEFAULT) == 0) new_default = TRUE; /* The label must be translatable */ if (!new_default && (hexstr_to_label(strval, &new_sl) != 0)) return (SET_ERROR(EINVAL)); /* * In a non-global zone, disallow attempts to set a label that * doesn't match that of the zone; otherwise no other checks * are needed. */ if (!INGLOBALZONE(curproc)) { if (new_default || !blequal(&new_sl, CR_SL(CRED()))) return (SET_ERROR(EPERM)); return (0); } /* * For global-zone datasets (i.e., those whose zoned property is * "off", verify that the specified new label is valid for the * global zone. */ if (dsl_prop_get_integer(name, zfs_prop_to_name(ZFS_PROP_ZONED), &zoned, NULL)) return (SET_ERROR(EPERM)); if (!zoned) { if (zfs_check_global_label(name, strval) != 0) return (SET_ERROR(EPERM)); } /* * If the existing dataset label is nondefault, check if the * dataset is mounted (label cannot be changed while mounted). * Get the zfs_sb_t; if there isn't one, then the dataset isn't * mounted (or isn't a dataset, doesn't exist, ...). */ if (strcasecmp(ds_hexsl, ZFS_MLSLABEL_DEFAULT) != 0) { objset_t *os; static char *setsl_tag = "setsl_tag"; /* * Try to own the dataset; abort if there is any error, * (e.g., already mounted, in use, or other error). */ error = dmu_objset_own(name, DMU_OST_ZFS, B_TRUE, setsl_tag, &os); if (error != 0) return (SET_ERROR(EPERM)); dmu_objset_disown(os, setsl_tag); if (new_default) { needed_priv = PRIV_FILE_DOWNGRADE_SL; goto out_check; } if (hexstr_to_label(strval, &new_sl) != 0) return (SET_ERROR(EPERM)); if (blstrictdom(&ds_sl, &new_sl)) needed_priv = PRIV_FILE_DOWNGRADE_SL; else if (blstrictdom(&new_sl, &ds_sl)) needed_priv = PRIV_FILE_UPGRADE_SL; } else { /* dataset currently has a default label */ if (!new_default) needed_priv = PRIV_FILE_UPGRADE_SL; } out_check: if (needed_priv != -1) return (PRIV_POLICY(cr, needed_priv, B_FALSE, EPERM, NULL)); return (0); #else return (ENOTSUP); #endif /* HAVE_MLSLABEL */ } static int zfs_secpolicy_setprop(const char *dsname, zfs_prop_t prop, nvpair_t *propval, cred_t *cr) { char *strval; /* * Check permissions for special properties. */ switch (prop) { default: break; case ZFS_PROP_ZONED: /* * Disallow setting of 'zoned' from within a local zone. */ if (!INGLOBALZONE(curproc)) return (SET_ERROR(EPERM)); break; case ZFS_PROP_QUOTA: if (!INGLOBALZONE(curproc)) { uint64_t zoned; char setpoint[MAXNAMELEN]; /* * Unprivileged users are allowed to modify the * quota on things *under* (ie. contained by) * the thing they own. */ if (dsl_prop_get_integer(dsname, "zoned", &zoned, setpoint)) return (SET_ERROR(EPERM)); if (!zoned || strlen(dsname) <= strlen(setpoint)) return (SET_ERROR(EPERM)); } break; case ZFS_PROP_MLSLABEL: if (!is_system_labeled()) return (SET_ERROR(EPERM)); if (nvpair_value_string(propval, &strval) == 0) { int err; err = zfs_set_slabel_policy(dsname, strval, CRED()); if (err != 0) return (err); } break; } return (zfs_secpolicy_write_perms(dsname, zfs_prop_to_name(prop), cr)); } /* ARGSUSED */ static int zfs_secpolicy_set_fsacl(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { int error; error = zfs_dozonecheck(zc->zc_name, cr); if (error != 0) return (error); /* * permission to set permissions will be evaluated later in * dsl_deleg_can_allow() */ return (0); } /* ARGSUSED */ static int zfs_secpolicy_rollback(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_ROLLBACK, cr)); } /* ARGSUSED */ static int zfs_secpolicy_send(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { dsl_pool_t *dp; dsl_dataset_t *ds; char *cp; int error; /* * Generate the current snapshot name from the given objsetid, then * use that name for the secpolicy/zone checks. */ cp = strchr(zc->zc_name, '@'); if (cp == NULL) return (SET_ERROR(EINVAL)); error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold_obj(dp, zc->zc_sendobj, FTAG, &ds); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } dsl_dataset_name(ds, zc->zc_name); error = zfs_secpolicy_write_perms_ds(zc->zc_name, ds, ZFS_DELEG_PERM_SEND, cr); dsl_dataset_rele(ds, FTAG); dsl_pool_rele(dp, FTAG); return (error); } /* ARGSUSED */ static int zfs_secpolicy_send_new(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_SEND, cr)); } #ifdef HAVE_SMB_SHARE /* ARGSUSED */ static int zfs_secpolicy_deleg_share(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { vnode_t *vp; int error; if ((error = lookupname(zc->zc_value, UIO_SYSSPACE, NO_FOLLOW, NULL, &vp)) != 0) return (error); /* Now make sure mntpnt and dataset are ZFS */ if (vp->v_vfsp->vfs_fstype != zfsfstype || (strcmp((char *)refstr_value(vp->v_vfsp->vfs_resource), zc->zc_name) != 0)) { VN_RELE(vp); return (SET_ERROR(EPERM)); } VN_RELE(vp); return (dsl_deleg_access(zc->zc_name, ZFS_DELEG_PERM_SHARE, cr)); } #endif /* HAVE_SMB_SHARE */ int zfs_secpolicy_share(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { #ifdef HAVE_SMB_SHARE if (!INGLOBALZONE(curproc)) return (SET_ERROR(EPERM)); if (secpolicy_nfs(cr) == 0) { return (0); } else { return (zfs_secpolicy_deleg_share(zc, innvl, cr)); } #else return (SET_ERROR(ENOTSUP)); #endif /* HAVE_SMB_SHARE */ } int zfs_secpolicy_smb_acl(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { #ifdef HAVE_SMB_SHARE if (!INGLOBALZONE(curproc)) return (SET_ERROR(EPERM)); if (secpolicy_smb(cr) == 0) { return (0); } else { return (zfs_secpolicy_deleg_share(zc, innvl, cr)); } #else return (SET_ERROR(ENOTSUP)); #endif /* HAVE_SMB_SHARE */ } static int zfs_get_parent(const char *datasetname, char *parent, int parentsize) { char *cp; /* * Remove the @bla or /bla from the end of the name to get the parent. */ (void) strncpy(parent, datasetname, parentsize); cp = strrchr(parent, '@'); if (cp != NULL) { cp[0] = '\0'; } else { cp = strrchr(parent, '/'); if (cp == NULL) return (SET_ERROR(ENOENT)); cp[0] = '\0'; } return (0); } int zfs_secpolicy_destroy_perms(const char *name, cred_t *cr) { int error; if ((error = zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_MOUNT, cr)) != 0) return (error); return (zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_DESTROY, cr)); } /* ARGSUSED */ static int zfs_secpolicy_destroy(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (zfs_secpolicy_destroy_perms(zc->zc_name, cr)); } /* * Destroying snapshots with delegated permissions requires * descendant mount and destroy permissions. */ /* ARGSUSED */ static int zfs_secpolicy_destroy_snaps(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { nvlist_t *snaps; nvpair_t *pair, *nextpair; int error = 0; if (nvlist_lookup_nvlist(innvl, "snaps", &snaps) != 0) return (SET_ERROR(EINVAL)); for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL; pair = nextpair) { nextpair = nvlist_next_nvpair(snaps, pair); error = zfs_secpolicy_destroy_perms(nvpair_name(pair), cr); if (error == ENOENT) { /* * Ignore any snapshots that don't exist (we consider * them "already destroyed"). Remove the name from the * nvl here in case the snapshot is created between * now and when we try to destroy it (in which case * we don't want to destroy it since we haven't * checked for permission). */ fnvlist_remove_nvpair(snaps, pair); error = 0; } if (error != 0) break; } return (error); } int zfs_secpolicy_rename_perms(const char *from, const char *to, cred_t *cr) { char parentname[MAXNAMELEN]; int error; if ((error = zfs_secpolicy_write_perms(from, ZFS_DELEG_PERM_RENAME, cr)) != 0) return (error); if ((error = zfs_secpolicy_write_perms(from, ZFS_DELEG_PERM_MOUNT, cr)) != 0) return (error); if ((error = zfs_get_parent(to, parentname, sizeof (parentname))) != 0) return (error); if ((error = zfs_secpolicy_write_perms(parentname, ZFS_DELEG_PERM_CREATE, cr)) != 0) return (error); if ((error = zfs_secpolicy_write_perms(parentname, ZFS_DELEG_PERM_MOUNT, cr)) != 0) return (error); return (error); } /* ARGSUSED */ static int zfs_secpolicy_rename(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (zfs_secpolicy_rename_perms(zc->zc_name, zc->zc_value, cr)); } /* ARGSUSED */ static int zfs_secpolicy_promote(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { dsl_pool_t *dp; dsl_dataset_t *clone; int error; error = zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_PROMOTE, cr); if (error != 0) return (error); error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, zc->zc_name, FTAG, &clone); if (error == 0) { char parentname[MAXNAMELEN]; dsl_dataset_t *origin = NULL; dsl_dir_t *dd; dd = clone->ds_dir; error = dsl_dataset_hold_obj(dd->dd_pool, dd->dd_phys->dd_origin_obj, FTAG, &origin); if (error != 0) { dsl_dataset_rele(clone, FTAG); dsl_pool_rele(dp, FTAG); return (error); } error = zfs_secpolicy_write_perms_ds(zc->zc_name, clone, ZFS_DELEG_PERM_MOUNT, cr); dsl_dataset_name(origin, parentname); if (error == 0) { error = zfs_secpolicy_write_perms_ds(parentname, origin, ZFS_DELEG_PERM_PROMOTE, cr); } dsl_dataset_rele(clone, FTAG); dsl_dataset_rele(origin, FTAG); } dsl_pool_rele(dp, FTAG); return (error); } /* ARGSUSED */ static int zfs_secpolicy_recv(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { int error; if ((error = zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_RECEIVE, cr)) != 0) return (error); if ((error = zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_MOUNT, cr)) != 0) return (error); return (zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_CREATE, cr)); } int zfs_secpolicy_snapshot_perms(const char *name, cred_t *cr) { return (zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_SNAPSHOT, cr)); } /* * Check for permission to create each snapshot in the nvlist. */ /* ARGSUSED */ static int zfs_secpolicy_snapshot(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { nvlist_t *snaps; int error = 0; nvpair_t *pair; if (nvlist_lookup_nvlist(innvl, "snaps", &snaps) != 0) return (SET_ERROR(EINVAL)); for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL; pair = nvlist_next_nvpair(snaps, pair)) { char *name = nvpair_name(pair); char *atp = strchr(name, '@'); if (atp == NULL) { error = SET_ERROR(EINVAL); break; } *atp = '\0'; error = zfs_secpolicy_snapshot_perms(name, cr); *atp = '@'; if (error != 0) break; } return (error); } /* * Check for permission to create each snapshot in the nvlist. */ /* ARGSUSED */ static int zfs_secpolicy_bookmark(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { int error = 0; nvpair_t *pair; for (pair = nvlist_next_nvpair(innvl, NULL); pair != NULL; pair = nvlist_next_nvpair(innvl, pair)) { char *name = nvpair_name(pair); char *hashp = strchr(name, '#'); if (hashp == NULL) { error = SET_ERROR(EINVAL); break; } *hashp = '\0'; error = zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_BOOKMARK, cr); *hashp = '#'; if (error != 0) break; } return (error); } /* ARGSUSED */ static int zfs_secpolicy_destroy_bookmarks(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { nvpair_t *pair, *nextpair; int error = 0; for (pair = nvlist_next_nvpair(innvl, NULL); pair != NULL; pair = nextpair) { char *name = nvpair_name(pair); char *hashp = strchr(name, '#'); nextpair = nvlist_next_nvpair(innvl, pair); if (hashp == NULL) { error = SET_ERROR(EINVAL); break; } *hashp = '\0'; error = zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_DESTROY, cr); *hashp = '#'; if (error == ENOENT) { /* * Ignore any filesystems that don't exist (we consider * their bookmarks "already destroyed"). Remove * the name from the nvl here in case the filesystem * is created between now and when we try to destroy * the bookmark (in which case we don't want to * destroy it since we haven't checked for permission). */ fnvlist_remove_nvpair(innvl, pair); error = 0; } if (error != 0) break; } return (error); } /* ARGSUSED */ static int zfs_secpolicy_log_history(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { /* * Even root must have a proper TSD so that we know what pool * to log to. */ if (tsd_get(zfs_allow_log_key) == NULL) return (SET_ERROR(EPERM)); return (0); } static int zfs_secpolicy_create_clone(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { char parentname[MAXNAMELEN]; int error; char *origin; if ((error = zfs_get_parent(zc->zc_name, parentname, sizeof (parentname))) != 0) return (error); if (nvlist_lookup_string(innvl, "origin", &origin) == 0 && (error = zfs_secpolicy_write_perms(origin, ZFS_DELEG_PERM_CLONE, cr)) != 0) return (error); if ((error = zfs_secpolicy_write_perms(parentname, ZFS_DELEG_PERM_CREATE, cr)) != 0) return (error); return (zfs_secpolicy_write_perms(parentname, ZFS_DELEG_PERM_MOUNT, cr)); } /* * Policy for pool operations - create/destroy pools, add vdevs, etc. Requires * SYS_CONFIG privilege, which is not available in a local zone. */ /* ARGSUSED */ static int zfs_secpolicy_config(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { if (secpolicy_sys_config(cr, B_FALSE) != 0) return (SET_ERROR(EPERM)); return (0); } /* * Policy for object to name lookups. */ /* ARGSUSED */ static int zfs_secpolicy_diff(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { int error; if ((error = secpolicy_sys_config(cr, B_FALSE)) == 0) return (0); error = zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_DIFF, cr); return (error); } /* * Policy for fault injection. Requires all privileges. */ /* ARGSUSED */ static int zfs_secpolicy_inject(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (secpolicy_zinject(cr)); } /* ARGSUSED */ static int zfs_secpolicy_inherit_prop(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { zfs_prop_t prop = zfs_name_to_prop(zc->zc_value); if (prop == ZPROP_INVAL) { if (!zfs_prop_user(zc->zc_value)) return (SET_ERROR(EINVAL)); return (zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_USERPROP, cr)); } else { return (zfs_secpolicy_setprop(zc->zc_name, prop, NULL, cr)); } } static int zfs_secpolicy_userspace_one(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { int err = zfs_secpolicy_read(zc, innvl, cr); if (err) return (err); if (zc->zc_objset_type >= ZFS_NUM_USERQUOTA_PROPS) return (SET_ERROR(EINVAL)); if (zc->zc_value[0] == 0) { /* * They are asking about a posix uid/gid. If it's * themself, allow it. */ if (zc->zc_objset_type == ZFS_PROP_USERUSED || zc->zc_objset_type == ZFS_PROP_USERQUOTA) { if (zc->zc_guid == crgetuid(cr)) return (0); } else { if (groupmember(zc->zc_guid, cr)) return (0); } } return (zfs_secpolicy_write_perms(zc->zc_name, userquota_perms[zc->zc_objset_type], cr)); } static int zfs_secpolicy_userspace_many(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { int err = zfs_secpolicy_read(zc, innvl, cr); if (err) return (err); if (zc->zc_objset_type >= ZFS_NUM_USERQUOTA_PROPS) return (SET_ERROR(EINVAL)); return (zfs_secpolicy_write_perms(zc->zc_name, userquota_perms[zc->zc_objset_type], cr)); } /* ARGSUSED */ static int zfs_secpolicy_userspace_upgrade(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (zfs_secpolicy_setprop(zc->zc_name, ZFS_PROP_VERSION, NULL, cr)); } /* ARGSUSED */ static int zfs_secpolicy_hold(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { nvpair_t *pair; nvlist_t *holds; int error; error = nvlist_lookup_nvlist(innvl, "holds", &holds); if (error != 0) return (SET_ERROR(EINVAL)); for (pair = nvlist_next_nvpair(holds, NULL); pair != NULL; pair = nvlist_next_nvpair(holds, pair)) { char fsname[MAXNAMELEN]; error = dmu_fsname(nvpair_name(pair), fsname); if (error != 0) return (error); error = zfs_secpolicy_write_perms(fsname, ZFS_DELEG_PERM_HOLD, cr); if (error != 0) return (error); } return (0); } /* ARGSUSED */ static int zfs_secpolicy_release(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { nvpair_t *pair; int error; for (pair = nvlist_next_nvpair(innvl, NULL); pair != NULL; pair = nvlist_next_nvpair(innvl, pair)) { char fsname[MAXNAMELEN]; error = dmu_fsname(nvpair_name(pair), fsname); if (error != 0) return (error); error = zfs_secpolicy_write_perms(fsname, ZFS_DELEG_PERM_RELEASE, cr); if (error != 0) return (error); } return (0); } /* * Policy for allowing temporary snapshots to be taken or released */ static int zfs_secpolicy_tmp_snapshot(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { /* * A temporary snapshot is the same as a snapshot, * hold, destroy and release all rolled into one. * Delegated diff alone is sufficient that we allow this. */ int error; if ((error = zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_DIFF, cr)) == 0) return (0); error = zfs_secpolicy_snapshot_perms(zc->zc_name, cr); if (error == 0) error = zfs_secpolicy_hold(zc, innvl, cr); if (error == 0) error = zfs_secpolicy_release(zc, innvl, cr); if (error == 0) error = zfs_secpolicy_destroy(zc, innvl, cr); return (error); } /* * Returns the nvlist as specified by the user in the zfs_cmd_t. */ static int get_nvlist(uint64_t nvl, uint64_t size, int iflag, nvlist_t **nvp) { char *packed; int error; nvlist_t *list = NULL; /* * Read in and unpack the user-supplied nvlist. */ if (size == 0) return (SET_ERROR(EINVAL)); - packed = kmem_alloc(size, KM_SLEEP | KM_NODEBUG); + packed = kmem_alloc(size, KM_SLEEP); if ((error = ddi_copyin((void *)(uintptr_t)nvl, packed, size, iflag)) != 0) { kmem_free(packed, size); return (error); } if ((error = nvlist_unpack(packed, size, &list, 0)) != 0) { kmem_free(packed, size); return (error); } kmem_free(packed, size); *nvp = list; return (0); } /* * Reduce the size of this nvlist until it can be serialized in 'max' bytes. * Entries will be removed from the end of the nvlist, and one int32 entry * named "N_MORE_ERRORS" will be added indicating how many entries were * removed. */ static int nvlist_smush(nvlist_t *errors, size_t max) { size_t size; size = fnvlist_size(errors); if (size > max) { nvpair_t *more_errors; int n = 0; if (max < 1024) return (SET_ERROR(ENOMEM)); fnvlist_add_int32(errors, ZPROP_N_MORE_ERRORS, 0); more_errors = nvlist_prev_nvpair(errors, NULL); do { nvpair_t *pair = nvlist_prev_nvpair(errors, more_errors); fnvlist_remove_nvpair(errors, pair); n++; size = fnvlist_size(errors); } while (size > max); fnvlist_remove_nvpair(errors, more_errors); fnvlist_add_int32(errors, ZPROP_N_MORE_ERRORS, n); ASSERT3U(fnvlist_size(errors), <=, max); } return (0); } static int put_nvlist(zfs_cmd_t *zc, nvlist_t *nvl) { char *packed = NULL; int error = 0; size_t size; size = fnvlist_size(nvl); if (size > zc->zc_nvlist_dst_size) { error = SET_ERROR(ENOMEM); } else { packed = fnvlist_pack(nvl, &size); if (ddi_copyout(packed, (void *)(uintptr_t)zc->zc_nvlist_dst, size, zc->zc_iflags) != 0) error = SET_ERROR(EFAULT); fnvlist_pack_free(packed, size); } zc->zc_nvlist_dst_size = size; zc->zc_nvlist_dst_filled = B_TRUE; return (error); } static int get_zfs_sb(const char *dsname, zfs_sb_t **zsbp) { objset_t *os; int error; error = dmu_objset_hold(dsname, FTAG, &os); if (error != 0) return (error); if (dmu_objset_type(os) != DMU_OST_ZFS) { dmu_objset_rele(os, FTAG); return (SET_ERROR(EINVAL)); } mutex_enter(&os->os_user_ptr_lock); *zsbp = dmu_objset_get_user(os); if (*zsbp && (*zsbp)->z_sb) { atomic_inc(&((*zsbp)->z_sb->s_active)); } else { error = SET_ERROR(ESRCH); } mutex_exit(&os->os_user_ptr_lock); dmu_objset_rele(os, FTAG); return (error); } /* * Find a zfs_sb_t for a mounted filesystem, or create our own, in which * case its z_sb will be NULL, and it will be opened as the owner. * If 'writer' is set, the z_teardown_lock will be held for RW_WRITER, * which prevents all inode ops from running. */ static int zfs_sb_hold(const char *name, void *tag, zfs_sb_t **zsbp, boolean_t writer) { int error = 0; if (get_zfs_sb(name, zsbp) != 0) error = zfs_sb_create(name, zsbp); if (error == 0) { rrw_enter(&(*zsbp)->z_teardown_lock, (writer) ? RW_WRITER : RW_READER, tag); if ((*zsbp)->z_unmounted) { /* * XXX we could probably try again, since the unmounting * thread should be just about to disassociate the * objset from the zsb. */ rrw_exit(&(*zsbp)->z_teardown_lock, tag); return (SET_ERROR(EBUSY)); } } return (error); } static void zfs_sb_rele(zfs_sb_t *zsb, void *tag) { rrw_exit(&zsb->z_teardown_lock, tag); if (zsb->z_sb) { deactivate_super(zsb->z_sb); } else { dmu_objset_disown(zsb->z_os, zsb); zfs_sb_free(zsb); } } static int zfs_ioc_pool_create(zfs_cmd_t *zc) { int error; nvlist_t *config, *props = NULL; nvlist_t *rootprops = NULL; nvlist_t *zplprops = NULL; if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &config))) return (error); if (zc->zc_nvlist_src_size != 0 && (error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &props))) { nvlist_free(config); return (error); } if (props) { nvlist_t *nvl = NULL; uint64_t version = SPA_VERSION; (void) nvlist_lookup_uint64(props, zpool_prop_to_name(ZPOOL_PROP_VERSION), &version); if (!SPA_VERSION_IS_SUPPORTED(version)) { error = SET_ERROR(EINVAL); goto pool_props_bad; } (void) nvlist_lookup_nvlist(props, ZPOOL_ROOTFS_PROPS, &nvl); if (nvl) { error = nvlist_dup(nvl, &rootprops, KM_SLEEP); if (error != 0) { nvlist_free(config); nvlist_free(props); return (error); } (void) nvlist_remove_all(props, ZPOOL_ROOTFS_PROPS); } VERIFY(nvlist_alloc(&zplprops, NV_UNIQUE_NAME, KM_SLEEP) == 0); error = zfs_fill_zplprops_root(version, rootprops, zplprops, NULL); if (error != 0) goto pool_props_bad; } error = spa_create(zc->zc_name, config, props, zplprops); /* * Set the remaining root properties */ if (!error && (error = zfs_set_prop_nvlist(zc->zc_name, ZPROP_SRC_LOCAL, rootprops, NULL)) != 0) (void) spa_destroy(zc->zc_name); pool_props_bad: nvlist_free(rootprops); nvlist_free(zplprops); nvlist_free(config); nvlist_free(props); return (error); } static int zfs_ioc_pool_destroy(zfs_cmd_t *zc) { int error; zfs_log_history(zc); error = spa_destroy(zc->zc_name); if (error == 0) zvol_remove_minors(zc->zc_name); return (error); } static int zfs_ioc_pool_import(zfs_cmd_t *zc) { nvlist_t *config, *props = NULL; uint64_t guid; int error; if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &config)) != 0) return (error); if (zc->zc_nvlist_src_size != 0 && (error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &props))) { nvlist_free(config); return (error); } if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &guid) != 0 || guid != zc->zc_guid) error = SET_ERROR(EINVAL); else error = spa_import(zc->zc_name, config, props, zc->zc_cookie); if (zc->zc_nvlist_dst != 0) { int err; if ((err = put_nvlist(zc, config)) != 0) error = err; } nvlist_free(config); if (props) nvlist_free(props); return (error); } static int zfs_ioc_pool_export(zfs_cmd_t *zc) { int error; boolean_t force = (boolean_t)zc->zc_cookie; boolean_t hardforce = (boolean_t)zc->zc_guid; zfs_log_history(zc); error = spa_export(zc->zc_name, NULL, force, hardforce); if (error == 0) zvol_remove_minors(zc->zc_name); return (error); } static int zfs_ioc_pool_configs(zfs_cmd_t *zc) { nvlist_t *configs; int error; if ((configs = spa_all_configs(&zc->zc_cookie)) == NULL) return (SET_ERROR(EEXIST)); error = put_nvlist(zc, configs); nvlist_free(configs); return (error); } /* * inputs: * zc_name name of the pool * * outputs: * zc_cookie real errno * zc_nvlist_dst config nvlist * zc_nvlist_dst_size size of config nvlist */ static int zfs_ioc_pool_stats(zfs_cmd_t *zc) { nvlist_t *config; int error; int ret = 0; error = spa_get_stats(zc->zc_name, &config, zc->zc_value, sizeof (zc->zc_value)); if (config != NULL) { ret = put_nvlist(zc, config); nvlist_free(config); /* * The config may be present even if 'error' is non-zero. * In this case we return success, and preserve the real errno * in 'zc_cookie'. */ zc->zc_cookie = error; } else { ret = error; } return (ret); } /* * Try to import the given pool, returning pool stats as appropriate so that * user land knows which devices are available and overall pool health. */ static int zfs_ioc_pool_tryimport(zfs_cmd_t *zc) { nvlist_t *tryconfig, *config; int error; if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &tryconfig)) != 0) return (error); config = spa_tryimport(tryconfig); nvlist_free(tryconfig); if (config == NULL) return (SET_ERROR(EINVAL)); error = put_nvlist(zc, config); nvlist_free(config); return (error); } /* * inputs: * zc_name name of the pool * zc_cookie scan func (pool_scan_func_t) */ static int zfs_ioc_pool_scan(zfs_cmd_t *zc) { spa_t *spa; int error; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); if (zc->zc_cookie == POOL_SCAN_NONE) error = spa_scan_stop(spa); else error = spa_scan(spa, zc->zc_cookie); spa_close(spa, FTAG); return (error); } static int zfs_ioc_pool_freeze(zfs_cmd_t *zc) { spa_t *spa; int error; error = spa_open(zc->zc_name, &spa, FTAG); if (error == 0) { spa_freeze(spa); spa_close(spa, FTAG); } return (error); } static int zfs_ioc_pool_upgrade(zfs_cmd_t *zc) { spa_t *spa; int error; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); if (zc->zc_cookie < spa_version(spa) || !SPA_VERSION_IS_SUPPORTED(zc->zc_cookie)) { spa_close(spa, FTAG); return (SET_ERROR(EINVAL)); } spa_upgrade(spa, zc->zc_cookie); spa_close(spa, FTAG); return (error); } static int zfs_ioc_pool_get_history(zfs_cmd_t *zc) { spa_t *spa; char *hist_buf; uint64_t size; int error; if ((size = zc->zc_history_len) == 0) return (SET_ERROR(EINVAL)); if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); if (spa_version(spa) < SPA_VERSION_ZPOOL_HISTORY) { spa_close(spa, FTAG); return (SET_ERROR(ENOTSUP)); } hist_buf = vmem_alloc(size, KM_SLEEP); if ((error = spa_history_get(spa, &zc->zc_history_offset, &zc->zc_history_len, hist_buf)) == 0) { error = ddi_copyout(hist_buf, (void *)(uintptr_t)zc->zc_history, zc->zc_history_len, zc->zc_iflags); } spa_close(spa, FTAG); vmem_free(hist_buf, size); return (error); } static int zfs_ioc_pool_reguid(zfs_cmd_t *zc) { spa_t *spa; int error; error = spa_open(zc->zc_name, &spa, FTAG); if (error == 0) { error = spa_change_guid(spa); spa_close(spa, FTAG); } return (error); } static int zfs_ioc_dsobj_to_dsname(zfs_cmd_t *zc) { return (dsl_dsobj_to_dsname(zc->zc_name, zc->zc_obj, zc->zc_value)); } /* * inputs: * zc_name name of filesystem * zc_obj object to find * * outputs: * zc_value name of object */ static int zfs_ioc_obj_to_path(zfs_cmd_t *zc) { objset_t *os; int error; /* XXX reading from objset not owned */ if ((error = dmu_objset_hold(zc->zc_name, FTAG, &os)) != 0) return (error); if (dmu_objset_type(os) != DMU_OST_ZFS) { dmu_objset_rele(os, FTAG); return (SET_ERROR(EINVAL)); } error = zfs_obj_to_path(os, zc->zc_obj, zc->zc_value, sizeof (zc->zc_value)); dmu_objset_rele(os, FTAG); return (error); } /* * inputs: * zc_name name of filesystem * zc_obj object to find * * outputs: * zc_stat stats on object * zc_value path to object */ static int zfs_ioc_obj_to_stats(zfs_cmd_t *zc) { objset_t *os; int error; /* XXX reading from objset not owned */ if ((error = dmu_objset_hold(zc->zc_name, FTAG, &os)) != 0) return (error); if (dmu_objset_type(os) != DMU_OST_ZFS) { dmu_objset_rele(os, FTAG); return (SET_ERROR(EINVAL)); } error = zfs_obj_to_stats(os, zc->zc_obj, &zc->zc_stat, zc->zc_value, sizeof (zc->zc_value)); dmu_objset_rele(os, FTAG); return (error); } static int zfs_ioc_vdev_add(zfs_cmd_t *zc) { spa_t *spa; int error; nvlist_t *config; error = spa_open(zc->zc_name, &spa, FTAG); if (error != 0) return (error); error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &config); if (error == 0) { error = spa_vdev_add(spa, config); nvlist_free(config); } spa_close(spa, FTAG); return (error); } /* * inputs: * zc_name name of the pool * zc_nvlist_conf nvlist of devices to remove * zc_cookie to stop the remove? */ static int zfs_ioc_vdev_remove(zfs_cmd_t *zc) { spa_t *spa; int error; error = spa_open(zc->zc_name, &spa, FTAG); if (error != 0) return (error); error = spa_vdev_remove(spa, zc->zc_guid, B_FALSE); spa_close(spa, FTAG); return (error); } static int zfs_ioc_vdev_set_state(zfs_cmd_t *zc) { spa_t *spa; int error; vdev_state_t newstate = VDEV_STATE_UNKNOWN; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); switch (zc->zc_cookie) { case VDEV_STATE_ONLINE: error = vdev_online(spa, zc->zc_guid, zc->zc_obj, &newstate); break; case VDEV_STATE_OFFLINE: error = vdev_offline(spa, zc->zc_guid, zc->zc_obj); break; case VDEV_STATE_FAULTED: if (zc->zc_obj != VDEV_AUX_ERR_EXCEEDED && zc->zc_obj != VDEV_AUX_EXTERNAL) zc->zc_obj = VDEV_AUX_ERR_EXCEEDED; error = vdev_fault(spa, zc->zc_guid, zc->zc_obj); break; case VDEV_STATE_DEGRADED: if (zc->zc_obj != VDEV_AUX_ERR_EXCEEDED && zc->zc_obj != VDEV_AUX_EXTERNAL) zc->zc_obj = VDEV_AUX_ERR_EXCEEDED; error = vdev_degrade(spa, zc->zc_guid, zc->zc_obj); break; default: error = SET_ERROR(EINVAL); } zc->zc_cookie = newstate; spa_close(spa, FTAG); return (error); } static int zfs_ioc_vdev_attach(zfs_cmd_t *zc) { spa_t *spa; int replacing = zc->zc_cookie; nvlist_t *config; int error; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &config)) == 0) { error = spa_vdev_attach(spa, zc->zc_guid, config, replacing); nvlist_free(config); } spa_close(spa, FTAG); return (error); } static int zfs_ioc_vdev_detach(zfs_cmd_t *zc) { spa_t *spa; int error; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); error = spa_vdev_detach(spa, zc->zc_guid, 0, B_FALSE); spa_close(spa, FTAG); return (error); } static int zfs_ioc_vdev_split(zfs_cmd_t *zc) { spa_t *spa; nvlist_t *config, *props = NULL; int error; boolean_t exp = !!(zc->zc_cookie & ZPOOL_EXPORT_AFTER_SPLIT); if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &config))) { spa_close(spa, FTAG); return (error); } if (zc->zc_nvlist_src_size != 0 && (error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &props))) { spa_close(spa, FTAG); nvlist_free(config); return (error); } error = spa_vdev_split_mirror(spa, zc->zc_string, config, props, exp); spa_close(spa, FTAG); nvlist_free(config); nvlist_free(props); return (error); } static int zfs_ioc_vdev_setpath(zfs_cmd_t *zc) { spa_t *spa; char *path = zc->zc_value; uint64_t guid = zc->zc_guid; int error; error = spa_open(zc->zc_name, &spa, FTAG); if (error != 0) return (error); error = spa_vdev_setpath(spa, guid, path); spa_close(spa, FTAG); return (error); } static int zfs_ioc_vdev_setfru(zfs_cmd_t *zc) { spa_t *spa; char *fru = zc->zc_value; uint64_t guid = zc->zc_guid; int error; error = spa_open(zc->zc_name, &spa, FTAG); if (error != 0) return (error); error = spa_vdev_setfru(spa, guid, fru); spa_close(spa, FTAG); return (error); } static int zfs_ioc_objset_stats_impl(zfs_cmd_t *zc, objset_t *os) { int error = 0; nvlist_t *nv; dmu_objset_fast_stat(os, &zc->zc_objset_stats); if (zc->zc_nvlist_dst != 0 && (error = dsl_prop_get_all(os, &nv)) == 0) { dmu_objset_stats(os, nv); /* * NB: zvol_get_stats() will read the objset contents, * which we aren't supposed to do with a * DS_MODE_USER hold, because it could be * inconsistent. So this is a bit of a workaround... * XXX reading with out owning */ if (!zc->zc_objset_stats.dds_inconsistent && dmu_objset_type(os) == DMU_OST_ZVOL) { error = zvol_get_stats(os, nv); if (error == EIO) return (error); VERIFY0(error); } if (error == 0) error = put_nvlist(zc, nv); nvlist_free(nv); } return (error); } /* * inputs: * zc_name name of filesystem * zc_nvlist_dst_size size of buffer for property nvlist * * outputs: * zc_objset_stats stats * zc_nvlist_dst property nvlist * zc_nvlist_dst_size size of property nvlist */ static int zfs_ioc_objset_stats(zfs_cmd_t *zc) { objset_t *os; int error; error = dmu_objset_hold(zc->zc_name, FTAG, &os); if (error == 0) { error = zfs_ioc_objset_stats_impl(zc, os); dmu_objset_rele(os, FTAG); } return (error); } /* * inputs: * zc_name name of filesystem * zc_nvlist_dst_size size of buffer for property nvlist * * outputs: * zc_nvlist_dst received property nvlist * zc_nvlist_dst_size size of received property nvlist * * Gets received properties (distinct from local properties on or after * SPA_VERSION_RECVD_PROPS) for callers who want to differentiate received from * local property values. */ static int zfs_ioc_objset_recvd_props(zfs_cmd_t *zc) { int error = 0; nvlist_t *nv; /* * Without this check, we would return local property values if the * caller has not already received properties on or after * SPA_VERSION_RECVD_PROPS. */ if (!dsl_prop_get_hasrecvd(zc->zc_name)) return (SET_ERROR(ENOTSUP)); if (zc->zc_nvlist_dst != 0 && (error = dsl_prop_get_received(zc->zc_name, &nv)) == 0) { error = put_nvlist(zc, nv); nvlist_free(nv); } return (error); } static int nvl_add_zplprop(objset_t *os, nvlist_t *props, zfs_prop_t prop) { uint64_t value; int error; /* * zfs_get_zplprop() will either find a value or give us * the default value (if there is one). */ if ((error = zfs_get_zplprop(os, prop, &value)) != 0) return (error); VERIFY(nvlist_add_uint64(props, zfs_prop_to_name(prop), value) == 0); return (0); } /* * inputs: * zc_name name of filesystem * zc_nvlist_dst_size size of buffer for zpl property nvlist * * outputs: * zc_nvlist_dst zpl property nvlist * zc_nvlist_dst_size size of zpl property nvlist */ static int zfs_ioc_objset_zplprops(zfs_cmd_t *zc) { objset_t *os; int err; /* XXX reading without owning */ if ((err = dmu_objset_hold(zc->zc_name, FTAG, &os))) return (err); dmu_objset_fast_stat(os, &zc->zc_objset_stats); /* * NB: nvl_add_zplprop() will read the objset contents, * which we aren't supposed to do with a DS_MODE_USER * hold, because it could be inconsistent. */ if (zc->zc_nvlist_dst != 0 && !zc->zc_objset_stats.dds_inconsistent && dmu_objset_type(os) == DMU_OST_ZFS) { nvlist_t *nv; VERIFY(nvlist_alloc(&nv, NV_UNIQUE_NAME, KM_SLEEP) == 0); if ((err = nvl_add_zplprop(os, nv, ZFS_PROP_VERSION)) == 0 && (err = nvl_add_zplprop(os, nv, ZFS_PROP_NORMALIZE)) == 0 && (err = nvl_add_zplprop(os, nv, ZFS_PROP_UTF8ONLY)) == 0 && (err = nvl_add_zplprop(os, nv, ZFS_PROP_CASE)) == 0) err = put_nvlist(zc, nv); nvlist_free(nv); } else { err = SET_ERROR(ENOENT); } dmu_objset_rele(os, FTAG); return (err); } boolean_t dataset_name_hidden(const char *name) { /* * Skip over datasets that are not visible in this zone, * internal datasets (which have a $ in their name), and * temporary datasets (which have a % in their name). */ if (strchr(name, '$') != NULL) return (B_TRUE); if (strchr(name, '%') != NULL) return (B_TRUE); if (!INGLOBALZONE(curproc) && !zone_dataset_visible(name, NULL)) return (B_TRUE); return (B_FALSE); } /* * inputs: * zc_name name of filesystem * zc_cookie zap cursor * zc_nvlist_dst_size size of buffer for property nvlist * * outputs: * zc_name name of next filesystem * zc_cookie zap cursor * zc_objset_stats stats * zc_nvlist_dst property nvlist * zc_nvlist_dst_size size of property nvlist */ static int zfs_ioc_dataset_list_next(zfs_cmd_t *zc) { objset_t *os; int error; char *p; size_t orig_len = strlen(zc->zc_name); top: if ((error = dmu_objset_hold(zc->zc_name, FTAG, &os))) { if (error == ENOENT) error = SET_ERROR(ESRCH); return (error); } p = strrchr(zc->zc_name, '/'); if (p == NULL || p[1] != '\0') (void) strlcat(zc->zc_name, "/", sizeof (zc->zc_name)); p = zc->zc_name + strlen(zc->zc_name); do { error = dmu_dir_list_next(os, sizeof (zc->zc_name) - (p - zc->zc_name), p, NULL, &zc->zc_cookie); if (error == ENOENT) error = SET_ERROR(ESRCH); } while (error == 0 && dataset_name_hidden(zc->zc_name)); dmu_objset_rele(os, FTAG); /* * If it's an internal dataset (ie. with a '$' in its name), * don't try to get stats for it, otherwise we'll return ENOENT. */ if (error == 0 && strchr(zc->zc_name, '$') == NULL) { error = zfs_ioc_objset_stats(zc); /* fill in the stats */ if (error == ENOENT) { /* We lost a race with destroy, get the next one. */ zc->zc_name[orig_len] = '\0'; goto top; } } return (error); } /* * inputs: * zc_name name of filesystem * zc_cookie zap cursor * zc_nvlist_dst_size size of buffer for property nvlist * * outputs: * zc_name name of next snapshot * zc_objset_stats stats * zc_nvlist_dst property nvlist * zc_nvlist_dst_size size of property nvlist */ static int zfs_ioc_snapshot_list_next(zfs_cmd_t *zc) { objset_t *os; int error; error = dmu_objset_hold(zc->zc_name, FTAG, &os); if (error != 0) { return (error == ENOENT ? ESRCH : error); } /* * A dataset name of maximum length cannot have any snapshots, * so exit immediately. */ if (strlcat(zc->zc_name, "@", sizeof (zc->zc_name)) >= MAXNAMELEN) { dmu_objset_rele(os, FTAG); return (SET_ERROR(ESRCH)); } error = dmu_snapshot_list_next(os, sizeof (zc->zc_name) - strlen(zc->zc_name), zc->zc_name + strlen(zc->zc_name), &zc->zc_obj, &zc->zc_cookie, NULL); if (error == 0 && !zc->zc_simple) { dsl_dataset_t *ds; dsl_pool_t *dp = os->os_dsl_dataset->ds_dir->dd_pool; error = dsl_dataset_hold_obj(dp, zc->zc_obj, FTAG, &ds); if (error == 0) { objset_t *ossnap; error = dmu_objset_from_ds(ds, &ossnap); if (error == 0) error = zfs_ioc_objset_stats_impl(zc, ossnap); dsl_dataset_rele(ds, FTAG); } } else if (error == ENOENT) { error = SET_ERROR(ESRCH); } dmu_objset_rele(os, FTAG); /* if we failed, undo the @ that we tacked on to zc_name */ if (error != 0) *strchr(zc->zc_name, '@') = '\0'; return (error); } static int zfs_prop_set_userquota(const char *dsname, nvpair_t *pair) { const char *propname = nvpair_name(pair); uint64_t *valary; unsigned int vallen; const char *domain; char *dash; zfs_userquota_prop_t type; uint64_t rid; uint64_t quota; zfs_sb_t *zsb; int err; if (nvpair_type(pair) == DATA_TYPE_NVLIST) { nvlist_t *attrs; VERIFY(nvpair_value_nvlist(pair, &attrs) == 0); if (nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &pair) != 0) return (SET_ERROR(EINVAL)); } /* * A correctly constructed propname is encoded as * userquota@-. */ if ((dash = strchr(propname, '-')) == NULL || nvpair_value_uint64_array(pair, &valary, &vallen) != 0 || vallen != 3) return (SET_ERROR(EINVAL)); domain = dash + 1; type = valary[0]; rid = valary[1]; quota = valary[2]; err = zfs_sb_hold(dsname, FTAG, &zsb, B_FALSE); if (err == 0) { err = zfs_set_userquota(zsb, type, domain, rid, quota); zfs_sb_rele(zsb, FTAG); } return (err); } /* * If the named property is one that has a special function to set its value, * return 0 on success and a positive error code on failure; otherwise if it is * not one of the special properties handled by this function, return -1. * * XXX: It would be better for callers of the property interface if we handled * these special cases in dsl_prop.c (in the dsl layer). */ static int zfs_prop_set_special(const char *dsname, zprop_source_t source, nvpair_t *pair) { const char *propname = nvpair_name(pair); zfs_prop_t prop = zfs_name_to_prop(propname); uint64_t intval; int err; if (prop == ZPROP_INVAL) { if (zfs_prop_userquota(propname)) return (zfs_prop_set_userquota(dsname, pair)); return (-1); } if (nvpair_type(pair) == DATA_TYPE_NVLIST) { nvlist_t *attrs; VERIFY(nvpair_value_nvlist(pair, &attrs) == 0); VERIFY(nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &pair) == 0); } if (zfs_prop_get_type(prop) == PROP_TYPE_STRING) return (-1); VERIFY(0 == nvpair_value_uint64(pair, &intval)); switch (prop) { case ZFS_PROP_QUOTA: err = dsl_dir_set_quota(dsname, source, intval); break; case ZFS_PROP_REFQUOTA: err = dsl_dataset_set_refquota(dsname, source, intval); break; case ZFS_PROP_RESERVATION: err = dsl_dir_set_reservation(dsname, source, intval); break; case ZFS_PROP_REFRESERVATION: err = dsl_dataset_set_refreservation(dsname, source, intval); break; case ZFS_PROP_VOLSIZE: err = zvol_set_volsize(dsname, intval); break; case ZFS_PROP_SNAPDEV: err = zvol_set_snapdev(dsname, intval); break; case ZFS_PROP_VERSION: { zfs_sb_t *zsb; if ((err = zfs_sb_hold(dsname, FTAG, &zsb, B_TRUE)) != 0) break; err = zfs_set_version(zsb, intval); zfs_sb_rele(zsb, FTAG); if (err == 0 && intval >= ZPL_VERSION_USERSPACE) { zfs_cmd_t *zc; - zc = kmem_zalloc(sizeof (zfs_cmd_t), - KM_SLEEP | KM_NODEBUG); + zc = kmem_zalloc(sizeof (zfs_cmd_t), KM_SLEEP); (void) strcpy(zc->zc_name, dsname); (void) zfs_ioc_userspace_upgrade(zc); kmem_free(zc, sizeof (zfs_cmd_t)); } break; } default: err = -1; } return (err); } /* * This function is best effort. If it fails to set any of the given properties, * it continues to set as many as it can and returns the last error * encountered. If the caller provides a non-NULL errlist, it will be filled in * with the list of names of all the properties that failed along with the * corresponding error numbers. * * If every property is set successfully, zero is returned and errlist is not * modified. */ int zfs_set_prop_nvlist(const char *dsname, zprop_source_t source, nvlist_t *nvl, nvlist_t *errlist) { nvpair_t *pair; nvpair_t *propval; int rv = 0; uint64_t intval; char *strval; nvlist_t *genericnvl = fnvlist_alloc(); nvlist_t *retrynvl = fnvlist_alloc(); retry: pair = NULL; while ((pair = nvlist_next_nvpair(nvl, pair)) != NULL) { const char *propname = nvpair_name(pair); zfs_prop_t prop = zfs_name_to_prop(propname); int err = 0; /* decode the property value */ propval = pair; if (nvpair_type(pair) == DATA_TYPE_NVLIST) { nvlist_t *attrs; attrs = fnvpair_value_nvlist(pair); if (nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &propval) != 0) err = SET_ERROR(EINVAL); } /* Validate value type */ if (err == 0 && prop == ZPROP_INVAL) { if (zfs_prop_user(propname)) { if (nvpair_type(propval) != DATA_TYPE_STRING) err = SET_ERROR(EINVAL); } else if (zfs_prop_userquota(propname)) { if (nvpair_type(propval) != DATA_TYPE_UINT64_ARRAY) err = SET_ERROR(EINVAL); } else { err = SET_ERROR(EINVAL); } } else if (err == 0) { if (nvpair_type(propval) == DATA_TYPE_STRING) { if (zfs_prop_get_type(prop) != PROP_TYPE_STRING) err = SET_ERROR(EINVAL); } else if (nvpair_type(propval) == DATA_TYPE_UINT64) { const char *unused; intval = fnvpair_value_uint64(propval); switch (zfs_prop_get_type(prop)) { case PROP_TYPE_NUMBER: break; case PROP_TYPE_STRING: err = SET_ERROR(EINVAL); break; case PROP_TYPE_INDEX: if (zfs_prop_index_to_string(prop, intval, &unused) != 0) err = SET_ERROR(EINVAL); break; default: cmn_err(CE_PANIC, "unknown property type"); } } else { err = SET_ERROR(EINVAL); } } /* Validate permissions */ if (err == 0) err = zfs_check_settable(dsname, pair, CRED()); if (err == 0) { err = zfs_prop_set_special(dsname, source, pair); if (err == -1) { /* * For better performance we build up a list of * properties to set in a single transaction. */ err = nvlist_add_nvpair(genericnvl, pair); } else if (err != 0 && nvl != retrynvl) { /* * This may be a spurious error caused by * receiving quota and reservation out of order. * Try again in a second pass. */ err = nvlist_add_nvpair(retrynvl, pair); } } if (err != 0) { if (errlist != NULL) fnvlist_add_int32(errlist, propname, err); rv = err; } } if (nvl != retrynvl && !nvlist_empty(retrynvl)) { nvl = retrynvl; goto retry; } if (!nvlist_empty(genericnvl) && dsl_props_set(dsname, source, genericnvl) != 0) { /* * If this fails, we still want to set as many properties as we * can, so try setting them individually. */ pair = NULL; while ((pair = nvlist_next_nvpair(genericnvl, pair)) != NULL) { const char *propname = nvpair_name(pair); int err = 0; propval = pair; if (nvpair_type(pair) == DATA_TYPE_NVLIST) { nvlist_t *attrs; attrs = fnvpair_value_nvlist(pair); propval = fnvlist_lookup_nvpair(attrs, ZPROP_VALUE); } if (nvpair_type(propval) == DATA_TYPE_STRING) { strval = fnvpair_value_string(propval); err = dsl_prop_set_string(dsname, propname, source, strval); } else { intval = fnvpair_value_uint64(propval); err = dsl_prop_set_int(dsname, propname, source, intval); } if (err != 0) { if (errlist != NULL) { fnvlist_add_int32(errlist, propname, err); } rv = err; } } } nvlist_free(genericnvl); nvlist_free(retrynvl); return (rv); } /* * Check that all the properties are valid user properties. */ static int zfs_check_userprops(const char *fsname, nvlist_t *nvl) { nvpair_t *pair = NULL; int error = 0; while ((pair = nvlist_next_nvpair(nvl, pair)) != NULL) { const char *propname = nvpair_name(pair); if (!zfs_prop_user(propname) || nvpair_type(pair) != DATA_TYPE_STRING) return (SET_ERROR(EINVAL)); if ((error = zfs_secpolicy_write_perms(fsname, ZFS_DELEG_PERM_USERPROP, CRED()))) return (error); if (strlen(propname) >= ZAP_MAXNAMELEN) return (SET_ERROR(ENAMETOOLONG)); if (strlen(fnvpair_value_string(pair)) >= ZAP_MAXVALUELEN) return (SET_ERROR(E2BIG)); } return (0); } static void props_skip(nvlist_t *props, nvlist_t *skipped, nvlist_t **newprops) { nvpair_t *pair; VERIFY(nvlist_alloc(newprops, NV_UNIQUE_NAME, KM_SLEEP) == 0); pair = NULL; while ((pair = nvlist_next_nvpair(props, pair)) != NULL) { if (nvlist_exists(skipped, nvpair_name(pair))) continue; VERIFY(nvlist_add_nvpair(*newprops, pair) == 0); } } static int clear_received_props(const char *dsname, nvlist_t *props, nvlist_t *skipped) { int err = 0; nvlist_t *cleared_props = NULL; props_skip(props, skipped, &cleared_props); if (!nvlist_empty(cleared_props)) { /* * Acts on local properties until the dataset has received * properties at least once on or after SPA_VERSION_RECVD_PROPS. */ zprop_source_t flags = (ZPROP_SRC_NONE | (dsl_prop_get_hasrecvd(dsname) ? ZPROP_SRC_RECEIVED : 0)); err = zfs_set_prop_nvlist(dsname, flags, cleared_props, NULL); } nvlist_free(cleared_props); return (err); } /* * inputs: * zc_name name of filesystem * zc_value name of property to set * zc_nvlist_src{_size} nvlist of properties to apply * zc_cookie received properties flag * * outputs: * zc_nvlist_dst{_size} error for each unapplied received property */ static int zfs_ioc_set_prop(zfs_cmd_t *zc) { nvlist_t *nvl; boolean_t received = zc->zc_cookie; zprop_source_t source = (received ? ZPROP_SRC_RECEIVED : ZPROP_SRC_LOCAL); nvlist_t *errors; int error; if ((error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &nvl)) != 0) return (error); if (received) { nvlist_t *origprops; if (dsl_prop_get_received(zc->zc_name, &origprops) == 0) { (void) clear_received_props(zc->zc_name, origprops, nvl); nvlist_free(origprops); } error = dsl_prop_set_hasrecvd(zc->zc_name); } errors = fnvlist_alloc(); if (error == 0) error = zfs_set_prop_nvlist(zc->zc_name, source, nvl, errors); if (zc->zc_nvlist_dst != 0 && errors != NULL) { (void) put_nvlist(zc, errors); } nvlist_free(errors); nvlist_free(nvl); return (error); } /* * inputs: * zc_name name of filesystem * zc_value name of property to inherit * zc_cookie revert to received value if TRUE * * outputs: none */ static int zfs_ioc_inherit_prop(zfs_cmd_t *zc) { const char *propname = zc->zc_value; zfs_prop_t prop = zfs_name_to_prop(propname); boolean_t received = zc->zc_cookie; zprop_source_t source = (received ? ZPROP_SRC_NONE /* revert to received value, if any */ : ZPROP_SRC_INHERITED); /* explicitly inherit */ if (received) { nvlist_t *dummy; nvpair_t *pair; zprop_type_t type; int err; /* * zfs_prop_set_special() expects properties in the form of an * nvpair with type info. */ if (prop == ZPROP_INVAL) { if (!zfs_prop_user(propname)) return (SET_ERROR(EINVAL)); type = PROP_TYPE_STRING; } else if (prop == ZFS_PROP_VOLSIZE || prop == ZFS_PROP_VERSION) { return (SET_ERROR(EINVAL)); } else { type = zfs_prop_get_type(prop); } VERIFY(nvlist_alloc(&dummy, NV_UNIQUE_NAME, KM_SLEEP) == 0); switch (type) { case PROP_TYPE_STRING: VERIFY(0 == nvlist_add_string(dummy, propname, "")); break; case PROP_TYPE_NUMBER: case PROP_TYPE_INDEX: VERIFY(0 == nvlist_add_uint64(dummy, propname, 0)); break; default: nvlist_free(dummy); return (SET_ERROR(EINVAL)); } pair = nvlist_next_nvpair(dummy, NULL); err = zfs_prop_set_special(zc->zc_name, source, pair); nvlist_free(dummy); if (err != -1) return (err); /* special property already handled */ } else { /* * Only check this in the non-received case. We want to allow * 'inherit -S' to revert non-inheritable properties like quota * and reservation to the received or default values even though * they are not considered inheritable. */ if (prop != ZPROP_INVAL && !zfs_prop_inheritable(prop)) return (SET_ERROR(EINVAL)); } /* property name has been validated by zfs_secpolicy_inherit_prop() */ return (dsl_prop_inherit(zc->zc_name, zc->zc_value, source)); } static int zfs_ioc_pool_set_props(zfs_cmd_t *zc) { nvlist_t *props; spa_t *spa; int error; nvpair_t *pair; if ((error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &props))) return (error); /* * If the only property is the configfile, then just do a spa_lookup() * to handle the faulted case. */ pair = nvlist_next_nvpair(props, NULL); if (pair != NULL && strcmp(nvpair_name(pair), zpool_prop_to_name(ZPOOL_PROP_CACHEFILE)) == 0 && nvlist_next_nvpair(props, pair) == NULL) { mutex_enter(&spa_namespace_lock); if ((spa = spa_lookup(zc->zc_name)) != NULL) { spa_configfile_set(spa, props, B_FALSE); spa_config_sync(spa, B_FALSE, B_TRUE); } mutex_exit(&spa_namespace_lock); if (spa != NULL) { nvlist_free(props); return (0); } } if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) { nvlist_free(props); return (error); } error = spa_prop_set(spa, props); nvlist_free(props); spa_close(spa, FTAG); return (error); } static int zfs_ioc_pool_get_props(zfs_cmd_t *zc) { spa_t *spa; int error; nvlist_t *nvp = NULL; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) { /* * If the pool is faulted, there may be properties we can still * get (such as altroot and cachefile), so attempt to get them * anyway. */ mutex_enter(&spa_namespace_lock); if ((spa = spa_lookup(zc->zc_name)) != NULL) error = spa_prop_get(spa, &nvp); mutex_exit(&spa_namespace_lock); } else { error = spa_prop_get(spa, &nvp); spa_close(spa, FTAG); } if (error == 0 && zc->zc_nvlist_dst != 0) error = put_nvlist(zc, nvp); else error = SET_ERROR(EFAULT); nvlist_free(nvp); return (error); } /* * inputs: * zc_name name of filesystem * zc_nvlist_src{_size} nvlist of delegated permissions * zc_perm_action allow/unallow flag * * outputs: none */ static int zfs_ioc_set_fsacl(zfs_cmd_t *zc) { int error; nvlist_t *fsaclnv = NULL; if ((error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &fsaclnv)) != 0) return (error); /* * Verify nvlist is constructed correctly */ if ((error = zfs_deleg_verify_nvlist(fsaclnv)) != 0) { nvlist_free(fsaclnv); return (SET_ERROR(EINVAL)); } /* * If we don't have PRIV_SYS_MOUNT, then validate * that user is allowed to hand out each permission in * the nvlist(s) */ error = secpolicy_zfs(CRED()); if (error != 0) { if (zc->zc_perm_action == B_FALSE) { error = dsl_deleg_can_allow(zc->zc_name, fsaclnv, CRED()); } else { error = dsl_deleg_can_unallow(zc->zc_name, fsaclnv, CRED()); } } if (error == 0) error = dsl_deleg_set(zc->zc_name, fsaclnv, zc->zc_perm_action); nvlist_free(fsaclnv); return (error); } /* * inputs: * zc_name name of filesystem * * outputs: * zc_nvlist_src{_size} nvlist of delegated permissions */ static int zfs_ioc_get_fsacl(zfs_cmd_t *zc) { nvlist_t *nvp; int error; if ((error = dsl_deleg_get(zc->zc_name, &nvp)) == 0) { error = put_nvlist(zc, nvp); nvlist_free(nvp); } return (error); } /* ARGSUSED */ static void zfs_create_cb(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx) { zfs_creat_t *zct = arg; zfs_create_fs(os, cr, zct->zct_zplprops, tx); } #define ZFS_PROP_UNDEFINED ((uint64_t)-1) /* * inputs: * os parent objset pointer (NULL if root fs) * fuids_ok fuids allowed in this version of the spa? * sa_ok SAs allowed in this version of the spa? * createprops list of properties requested by creator * * outputs: * zplprops values for the zplprops we attach to the master node object * is_ci true if requested file system will be purely case-insensitive * * Determine the settings for utf8only, normalization and * casesensitivity. Specific values may have been requested by the * creator and/or we can inherit values from the parent dataset. If * the file system is of too early a vintage, a creator can not * request settings for these properties, even if the requested * setting is the default value. We don't actually want to create dsl * properties for these, so remove them from the source nvlist after * processing. */ static int zfs_fill_zplprops_impl(objset_t *os, uint64_t zplver, boolean_t fuids_ok, boolean_t sa_ok, nvlist_t *createprops, nvlist_t *zplprops, boolean_t *is_ci) { uint64_t sense = ZFS_PROP_UNDEFINED; uint64_t norm = ZFS_PROP_UNDEFINED; uint64_t u8 = ZFS_PROP_UNDEFINED; int error; ASSERT(zplprops != NULL); /* * Pull out creator prop choices, if any. */ if (createprops) { (void) nvlist_lookup_uint64(createprops, zfs_prop_to_name(ZFS_PROP_VERSION), &zplver); (void) nvlist_lookup_uint64(createprops, zfs_prop_to_name(ZFS_PROP_NORMALIZE), &norm); (void) nvlist_remove_all(createprops, zfs_prop_to_name(ZFS_PROP_NORMALIZE)); (void) nvlist_lookup_uint64(createprops, zfs_prop_to_name(ZFS_PROP_UTF8ONLY), &u8); (void) nvlist_remove_all(createprops, zfs_prop_to_name(ZFS_PROP_UTF8ONLY)); (void) nvlist_lookup_uint64(createprops, zfs_prop_to_name(ZFS_PROP_CASE), &sense); (void) nvlist_remove_all(createprops, zfs_prop_to_name(ZFS_PROP_CASE)); } /* * If the zpl version requested is whacky or the file system * or pool is version is too "young" to support normalization * and the creator tried to set a value for one of the props, * error out. */ if ((zplver < ZPL_VERSION_INITIAL || zplver > ZPL_VERSION) || (zplver >= ZPL_VERSION_FUID && !fuids_ok) || (zplver >= ZPL_VERSION_SA && !sa_ok) || (zplver < ZPL_VERSION_NORMALIZATION && (norm != ZFS_PROP_UNDEFINED || u8 != ZFS_PROP_UNDEFINED || sense != ZFS_PROP_UNDEFINED))) return (SET_ERROR(ENOTSUP)); /* * Put the version in the zplprops */ VERIFY(nvlist_add_uint64(zplprops, zfs_prop_to_name(ZFS_PROP_VERSION), zplver) == 0); if (norm == ZFS_PROP_UNDEFINED && (error = zfs_get_zplprop(os, ZFS_PROP_NORMALIZE, &norm)) != 0) return (error); VERIFY(nvlist_add_uint64(zplprops, zfs_prop_to_name(ZFS_PROP_NORMALIZE), norm) == 0); /* * If we're normalizing, names must always be valid UTF-8 strings. */ if (norm) u8 = 1; if (u8 == ZFS_PROP_UNDEFINED && (error = zfs_get_zplprop(os, ZFS_PROP_UTF8ONLY, &u8)) != 0) return (error); VERIFY(nvlist_add_uint64(zplprops, zfs_prop_to_name(ZFS_PROP_UTF8ONLY), u8) == 0); if (sense == ZFS_PROP_UNDEFINED && (error = zfs_get_zplprop(os, ZFS_PROP_CASE, &sense)) != 0) return (error); VERIFY(nvlist_add_uint64(zplprops, zfs_prop_to_name(ZFS_PROP_CASE), sense) == 0); if (is_ci) *is_ci = (sense == ZFS_CASE_INSENSITIVE); return (0); } static int zfs_fill_zplprops(const char *dataset, nvlist_t *createprops, nvlist_t *zplprops, boolean_t *is_ci) { boolean_t fuids_ok, sa_ok; uint64_t zplver = ZPL_VERSION; objset_t *os = NULL; char parentname[MAXNAMELEN]; char *cp; spa_t *spa; uint64_t spa_vers; int error; (void) strlcpy(parentname, dataset, sizeof (parentname)); cp = strrchr(parentname, '/'); ASSERT(cp != NULL); cp[0] = '\0'; if ((error = spa_open(dataset, &spa, FTAG)) != 0) return (error); spa_vers = spa_version(spa); spa_close(spa, FTAG); zplver = zfs_zpl_version_map(spa_vers); fuids_ok = (zplver >= ZPL_VERSION_FUID); sa_ok = (zplver >= ZPL_VERSION_SA); /* * Open parent object set so we can inherit zplprop values. */ if ((error = dmu_objset_hold(parentname, FTAG, &os)) != 0) return (error); error = zfs_fill_zplprops_impl(os, zplver, fuids_ok, sa_ok, createprops, zplprops, is_ci); dmu_objset_rele(os, FTAG); return (error); } static int zfs_fill_zplprops_root(uint64_t spa_vers, nvlist_t *createprops, nvlist_t *zplprops, boolean_t *is_ci) { boolean_t fuids_ok; boolean_t sa_ok; uint64_t zplver = ZPL_VERSION; int error; zplver = zfs_zpl_version_map(spa_vers); fuids_ok = (zplver >= ZPL_VERSION_FUID); sa_ok = (zplver >= ZPL_VERSION_SA); error = zfs_fill_zplprops_impl(NULL, zplver, fuids_ok, sa_ok, createprops, zplprops, is_ci); return (error); } /* * innvl: { * "type" -> dmu_objset_type_t (int32) * (optional) "props" -> { prop -> value } * } * * outnvl: propname -> error code (int32) */ static int zfs_ioc_create(const char *fsname, nvlist_t *innvl, nvlist_t *outnvl) { int error = 0; zfs_creat_t zct = { 0 }; nvlist_t *nvprops = NULL; void (*cbfunc)(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx); int32_t type32; dmu_objset_type_t type; boolean_t is_insensitive = B_FALSE; if (nvlist_lookup_int32(innvl, "type", &type32) != 0) return (SET_ERROR(EINVAL)); type = type32; (void) nvlist_lookup_nvlist(innvl, "props", &nvprops); switch (type) { case DMU_OST_ZFS: cbfunc = zfs_create_cb; break; case DMU_OST_ZVOL: cbfunc = zvol_create_cb; break; default: cbfunc = NULL; break; } if (strchr(fsname, '@') || strchr(fsname, '%')) return (SET_ERROR(EINVAL)); zct.zct_props = nvprops; if (cbfunc == NULL) return (SET_ERROR(EINVAL)); if (type == DMU_OST_ZVOL) { uint64_t volsize, volblocksize; if (nvprops == NULL) return (SET_ERROR(EINVAL)); if (nvlist_lookup_uint64(nvprops, zfs_prop_to_name(ZFS_PROP_VOLSIZE), &volsize) != 0) return (SET_ERROR(EINVAL)); if ((error = nvlist_lookup_uint64(nvprops, zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE), &volblocksize)) != 0 && error != ENOENT) return (SET_ERROR(EINVAL)); if (error != 0) volblocksize = zfs_prop_default_numeric( ZFS_PROP_VOLBLOCKSIZE); if ((error = zvol_check_volblocksize( volblocksize)) != 0 || (error = zvol_check_volsize(volsize, volblocksize)) != 0) return (error); } else if (type == DMU_OST_ZFS) { int error; /* * We have to have normalization and * case-folding flags correct when we do the * file system creation, so go figure them out * now. */ VERIFY(nvlist_alloc(&zct.zct_zplprops, NV_UNIQUE_NAME, KM_SLEEP) == 0); error = zfs_fill_zplprops(fsname, nvprops, zct.zct_zplprops, &is_insensitive); if (error != 0) { nvlist_free(zct.zct_zplprops); return (error); } } error = dmu_objset_create(fsname, type, is_insensitive ? DS_FLAG_CI_DATASET : 0, cbfunc, &zct); nvlist_free(zct.zct_zplprops); /* * It would be nice to do this atomically. */ if (error == 0) { error = zfs_set_prop_nvlist(fsname, ZPROP_SRC_LOCAL, nvprops, outnvl); if (error != 0) (void) dsl_destroy_head(fsname); } #ifdef _KERNEL if (error == 0 && type == DMU_OST_ZVOL) zvol_create_minors(fsname); #endif return (error); } /* * innvl: { * "origin" -> name of origin snapshot * (optional) "props" -> { prop -> value } * } * * outputs: * outnvl: propname -> error code (int32) */ static int zfs_ioc_clone(const char *fsname, nvlist_t *innvl, nvlist_t *outnvl) { int error = 0; nvlist_t *nvprops = NULL; char *origin_name; if (nvlist_lookup_string(innvl, "origin", &origin_name) != 0) return (SET_ERROR(EINVAL)); (void) nvlist_lookup_nvlist(innvl, "props", &nvprops); if (strchr(fsname, '@') || strchr(fsname, '%')) return (SET_ERROR(EINVAL)); if (dataset_namecheck(origin_name, NULL, NULL) != 0) return (SET_ERROR(EINVAL)); error = dmu_objset_clone(fsname, origin_name); if (error != 0) return (error); /* * It would be nice to do this atomically. */ if (error == 0) { error = zfs_set_prop_nvlist(fsname, ZPROP_SRC_LOCAL, nvprops, outnvl); if (error != 0) (void) dsl_destroy_head(fsname); } #ifdef _KERNEL if (error == 0) zvol_create_minors(fsname); #endif return (error); } /* * innvl: { * "snaps" -> { snapshot1, snapshot2 } * (optional) "props" -> { prop -> value (string) } * } * * outnvl: snapshot -> error code (int32) */ static int zfs_ioc_snapshot(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { nvlist_t *snaps; nvlist_t *props = NULL; int error, poollen; nvpair_t *pair, *pair2; (void) nvlist_lookup_nvlist(innvl, "props", &props); if ((error = zfs_check_userprops(poolname, props)) != 0) return (error); if (!nvlist_empty(props) && zfs_earlier_version(poolname, SPA_VERSION_SNAP_PROPS)) return (SET_ERROR(ENOTSUP)); if (nvlist_lookup_nvlist(innvl, "snaps", &snaps) != 0) return (SET_ERROR(EINVAL)); poollen = strlen(poolname); for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL; pair = nvlist_next_nvpair(snaps, pair)) { const char *name = nvpair_name(pair); const char *cp = strchr(name, '@'); /* * The snap name must contain an @, and the part after it must * contain only valid characters. */ if (cp == NULL || zfs_component_namecheck(cp + 1, NULL, NULL) != 0) return (SET_ERROR(EINVAL)); /* * The snap must be in the specified pool. */ if (strncmp(name, poolname, poollen) != 0 || (name[poollen] != '/' && name[poollen] != '@')) return (SET_ERROR(EXDEV)); /* This must be the only snap of this fs. */ for (pair2 = nvlist_next_nvpair(snaps, pair); pair2 != NULL; pair2 = nvlist_next_nvpair(snaps, pair2)) { if (strncmp(name, nvpair_name(pair2), cp - name + 1) == 0) { return (SET_ERROR(EXDEV)); } } } error = dsl_dataset_snapshot(snaps, props, outnvl); #ifdef _KERNEL if (error == 0) zvol_create_minors(poolname); #endif return (error); } /* * innvl: "message" -> string */ /* ARGSUSED */ static int zfs_ioc_log_history(const char *unused, nvlist_t *innvl, nvlist_t *outnvl) { char *message; spa_t *spa; int error; char *poolname; /* * The poolname in the ioctl is not set, we get it from the TSD, * which was set at the end of the last successful ioctl that allows * logging. The secpolicy func already checked that it is set. * Only one log ioctl is allowed after each successful ioctl, so * we clear the TSD here. */ poolname = tsd_get(zfs_allow_log_key); (void) tsd_set(zfs_allow_log_key, NULL); error = spa_open(poolname, &spa, FTAG); strfree(poolname); if (error != 0) return (error); if (nvlist_lookup_string(innvl, "message", &message) != 0) { spa_close(spa, FTAG); return (SET_ERROR(EINVAL)); } if (spa_version(spa) < SPA_VERSION_ZPOOL_HISTORY) { spa_close(spa, FTAG); return (SET_ERROR(ENOTSUP)); } error = spa_history_log(spa, message); spa_close(spa, FTAG); return (error); } /* * The dp_config_rwlock must not be held when calling this, because the * unmount may need to write out data. * * This function is best-effort. Callers must deal gracefully if it * remains mounted (or is remounted after this call). * * XXX: This function should detect a failure to unmount a snapdir of a dataset * and return the appropriate error code when it is mounted. Its Illumos and * FreeBSD counterparts do this. We do not do this on Linux because there is no * clear way to access the mount information that FreeBSD and Illumos use to * distinguish between things with mounted snapshot directories, and things * without mounted snapshot directories, which include zvols. Returning a * failure for the latter causes `zfs destroy` to fail on zvol snapshots. */ int zfs_unmount_snap(const char *snapname) { zfs_sb_t *zsb = NULL; char *dsname; char *fullname; char *ptr; if ((ptr = strchr(snapname, '@')) == NULL) return (0); dsname = kmem_alloc(ptr - snapname + 1, KM_SLEEP); strlcpy(dsname, snapname, ptr - snapname + 1); fullname = strdup(snapname); if (zfs_sb_hold(dsname, FTAG, &zsb, B_FALSE) == 0) { ASSERT(!dsl_pool_config_held(dmu_objset_pool(zsb->z_os))); (void) zfsctl_unmount_snapshot(zsb, fullname, MNT_FORCE); zfs_sb_rele(zsb, FTAG); } kmem_free(dsname, ptr - snapname + 1); strfree(fullname); return (0); } /* ARGSUSED */ static int zfs_unmount_snap_cb(const char *snapname, void *arg) { return (zfs_unmount_snap(snapname)); } /* * When a clone is destroyed, its origin may also need to be destroyed, * in which case it must be unmounted. This routine will do that unmount * if necessary. */ void zfs_destroy_unmount_origin(const char *fsname) { int error; objset_t *os; dsl_dataset_t *ds; error = dmu_objset_hold(fsname, FTAG, &os); if (error != 0) return; ds = dmu_objset_ds(os); if (dsl_dir_is_clone(ds->ds_dir) && DS_IS_DEFER_DESTROY(ds->ds_prev)) { char originname[MAXNAMELEN]; dsl_dataset_name(ds->ds_prev, originname); dmu_objset_rele(os, FTAG); (void) zfs_unmount_snap(originname); } else { dmu_objset_rele(os, FTAG); } } /* * innvl: { * "snaps" -> { snapshot1, snapshot2 } * (optional boolean) "defer" * } * * outnvl: snapshot -> error code (int32) */ /* ARGSUSED */ static int zfs_ioc_destroy_snaps(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { nvlist_t *snaps; nvpair_t *pair; boolean_t defer; if (nvlist_lookup_nvlist(innvl, "snaps", &snaps) != 0) return (SET_ERROR(EINVAL)); defer = nvlist_exists(innvl, "defer"); for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL; pair = nvlist_next_nvpair(snaps, pair)) { (void) zfs_unmount_snap(nvpair_name(pair)); (void) zvol_remove_minor(nvpair_name(pair)); } return (dsl_destroy_snapshots_nvl(snaps, defer, outnvl)); } /* * Create bookmarks. Bookmark names are of the form #. * All bookmarks must be in the same pool. * * innvl: { * bookmark1 -> snapshot1, bookmark2 -> snapshot2 * } * * outnvl: bookmark -> error code (int32) * */ /* ARGSUSED */ static int zfs_ioc_bookmark(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { nvpair_t *pair, *pair2; for (pair = nvlist_next_nvpair(innvl, NULL); pair != NULL; pair = nvlist_next_nvpair(innvl, pair)) { char *snap_name; /* * Verify the snapshot argument. */ if (nvpair_value_string(pair, &snap_name) != 0) return (SET_ERROR(EINVAL)); /* Verify that the keys (bookmarks) are unique */ for (pair2 = nvlist_next_nvpair(innvl, pair); pair2 != NULL; pair2 = nvlist_next_nvpair(innvl, pair2)) { if (strcmp(nvpair_name(pair), nvpair_name(pair2)) == 0) return (SET_ERROR(EINVAL)); } } return (dsl_bookmark_create(innvl, outnvl)); } /* * innvl: { * property 1, property 2, ... * } * * outnvl: { * bookmark name 1 -> { property 1, property 2, ... }, * bookmark name 2 -> { property 1, property 2, ... } * } * */ static int zfs_ioc_get_bookmarks(const char *fsname, nvlist_t *innvl, nvlist_t *outnvl) { return (dsl_get_bookmarks(fsname, innvl, outnvl)); } /* * innvl: { * bookmark name 1, bookmark name 2 * } * * outnvl: bookmark -> error code (int32) * */ static int zfs_ioc_destroy_bookmarks(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { int error, poollen; nvpair_t *pair; poollen = strlen(poolname); for (pair = nvlist_next_nvpair(innvl, NULL); pair != NULL; pair = nvlist_next_nvpair(innvl, pair)) { const char *name = nvpair_name(pair); const char *cp = strchr(name, '#'); /* * The bookmark name must contain an #, and the part after it * must contain only valid characters. */ if (cp == NULL || zfs_component_namecheck(cp + 1, NULL, NULL) != 0) return (SET_ERROR(EINVAL)); /* * The bookmark must be in the specified pool. */ if (strncmp(name, poolname, poollen) != 0 || (name[poollen] != '/' && name[poollen] != '#')) return (SET_ERROR(EXDEV)); } error = dsl_bookmark_destroy(innvl, outnvl); return (error); } /* * inputs: * zc_name name of dataset to destroy * zc_objset_type type of objset * zc_defer_destroy mark for deferred destroy * * outputs: none */ static int zfs_ioc_destroy(zfs_cmd_t *zc) { int err; if (zc->zc_objset_type == DMU_OST_ZFS) { err = zfs_unmount_snap(zc->zc_name); if (err != 0) return (err); } if (strchr(zc->zc_name, '@')) err = dsl_destroy_snapshot(zc->zc_name, zc->zc_defer_destroy); else err = dsl_destroy_head(zc->zc_name); if (zc->zc_objset_type == DMU_OST_ZVOL && err == 0) (void) zvol_remove_minor(zc->zc_name); return (err); } /* * fsname is name of dataset to rollback (to most recent snapshot) * * innvl is not used. * * outnvl: "target" -> name of most recent snapshot * } */ /* ARGSUSED */ static int zfs_ioc_rollback(const char *fsname, nvlist_t *args, nvlist_t *outnvl) { zfs_sb_t *zsb; int error; if (get_zfs_sb(fsname, &zsb) == 0) { error = zfs_suspend_fs(zsb); if (error == 0) { int resume_err; error = dsl_dataset_rollback(fsname, zsb, outnvl); resume_err = zfs_resume_fs(zsb, fsname); error = error ? error : resume_err; } deactivate_super(zsb->z_sb); } else { error = dsl_dataset_rollback(fsname, NULL, outnvl); } return (error); } static int recursive_unmount(const char *fsname, void *arg) { const char *snapname = arg; char *fullname; int error; fullname = kmem_asprintf("%s@%s", fsname, snapname); error = zfs_unmount_snap(fullname); strfree(fullname); return (error); } /* * inputs: * zc_name old name of dataset * zc_value new name of dataset * zc_cookie recursive flag (only valid for snapshots) * * outputs: none */ static int zfs_ioc_rename(zfs_cmd_t *zc) { boolean_t recursive = zc->zc_cookie & 1; char *at; zc->zc_value[sizeof (zc->zc_value) - 1] = '\0'; if (dataset_namecheck(zc->zc_value, NULL, NULL) != 0 || strchr(zc->zc_value, '%')) return (SET_ERROR(EINVAL)); at = strchr(zc->zc_name, '@'); if (at != NULL) { /* snaps must be in same fs */ int error; if (strncmp(zc->zc_name, zc->zc_value, at - zc->zc_name + 1)) return (SET_ERROR(EXDEV)); *at = '\0'; if (zc->zc_objset_type == DMU_OST_ZFS) { error = dmu_objset_find(zc->zc_name, recursive_unmount, at + 1, recursive ? DS_FIND_CHILDREN : 0); if (error != 0) { *at = '@'; return (error); } } error = dsl_dataset_rename_snapshot(zc->zc_name, at + 1, strchr(zc->zc_value, '@') + 1, recursive); *at = '@'; return (error); } else { return (dsl_dir_rename(zc->zc_name, zc->zc_value)); } } static int zfs_check_settable(const char *dsname, nvpair_t *pair, cred_t *cr) { const char *propname = nvpair_name(pair); boolean_t issnap = (strchr(dsname, '@') != NULL); zfs_prop_t prop = zfs_name_to_prop(propname); uint64_t intval; int err; if (prop == ZPROP_INVAL) { if (zfs_prop_user(propname)) { if ((err = zfs_secpolicy_write_perms(dsname, ZFS_DELEG_PERM_USERPROP, cr))) return (err); return (0); } if (!issnap && zfs_prop_userquota(propname)) { const char *perm = NULL; const char *uq_prefix = zfs_userquota_prop_prefixes[ZFS_PROP_USERQUOTA]; const char *gq_prefix = zfs_userquota_prop_prefixes[ZFS_PROP_GROUPQUOTA]; if (strncmp(propname, uq_prefix, strlen(uq_prefix)) == 0) { perm = ZFS_DELEG_PERM_USERQUOTA; } else if (strncmp(propname, gq_prefix, strlen(gq_prefix)) == 0) { perm = ZFS_DELEG_PERM_GROUPQUOTA; } else { /* USERUSED and GROUPUSED are read-only */ return (SET_ERROR(EINVAL)); } if ((err = zfs_secpolicy_write_perms(dsname, perm, cr))) return (err); return (0); } return (SET_ERROR(EINVAL)); } if (issnap) return (SET_ERROR(EINVAL)); if (nvpair_type(pair) == DATA_TYPE_NVLIST) { /* * dsl_prop_get_all_impl() returns properties in this * format. */ nvlist_t *attrs; VERIFY(nvpair_value_nvlist(pair, &attrs) == 0); VERIFY(nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &pair) == 0); } /* * Check that this value is valid for this pool version */ switch (prop) { case ZFS_PROP_COMPRESSION: /* * If the user specified gzip compression, make sure * the SPA supports it. We ignore any errors here since * we'll catch them later. */ if (nvpair_type(pair) == DATA_TYPE_UINT64 && nvpair_value_uint64(pair, &intval) == 0) { if (intval >= ZIO_COMPRESS_GZIP_1 && intval <= ZIO_COMPRESS_GZIP_9 && zfs_earlier_version(dsname, SPA_VERSION_GZIP_COMPRESSION)) { return (SET_ERROR(ENOTSUP)); } if (intval == ZIO_COMPRESS_ZLE && zfs_earlier_version(dsname, SPA_VERSION_ZLE_COMPRESSION)) return (SET_ERROR(ENOTSUP)); if (intval == ZIO_COMPRESS_LZ4) { spa_t *spa; if ((err = spa_open(dsname, &spa, FTAG)) != 0) return (err); if (!spa_feature_is_enabled(spa, SPA_FEATURE_LZ4_COMPRESS)) { spa_close(spa, FTAG); return (SET_ERROR(ENOTSUP)); } spa_close(spa, FTAG); } /* * If this is a bootable dataset then * verify that the compression algorithm * is supported for booting. We must return * something other than ENOTSUP since it * implies a downrev pool version. */ if (zfs_is_bootfs(dsname) && !BOOTFS_COMPRESS_VALID(intval)) { return (SET_ERROR(ERANGE)); } } break; case ZFS_PROP_COPIES: if (zfs_earlier_version(dsname, SPA_VERSION_DITTO_BLOCKS)) return (SET_ERROR(ENOTSUP)); break; case ZFS_PROP_DEDUP: if (zfs_earlier_version(dsname, SPA_VERSION_DEDUP)) return (SET_ERROR(ENOTSUP)); break; case ZFS_PROP_SHARESMB: if (zpl_earlier_version(dsname, ZPL_VERSION_FUID)) return (SET_ERROR(ENOTSUP)); break; case ZFS_PROP_ACLINHERIT: if (nvpair_type(pair) == DATA_TYPE_UINT64 && nvpair_value_uint64(pair, &intval) == 0) { if (intval == ZFS_ACL_PASSTHROUGH_X && zfs_earlier_version(dsname, SPA_VERSION_PASSTHROUGH_X)) return (SET_ERROR(ENOTSUP)); } break; default: break; } return (zfs_secpolicy_setprop(dsname, prop, pair, CRED())); } /* * Removes properties from the given props list that fail permission checks * needed to clear them and to restore them in case of a receive error. For each * property, make sure we have both set and inherit permissions. * * Returns the first error encountered if any permission checks fail. If the * caller provides a non-NULL errlist, it also gives the complete list of names * of all the properties that failed a permission check along with the * corresponding error numbers. The caller is responsible for freeing the * returned errlist. * * If every property checks out successfully, zero is returned and the list * pointed at by errlist is NULL. */ static int zfs_check_clearable(char *dataset, nvlist_t *props, nvlist_t **errlist) { zfs_cmd_t *zc; nvpair_t *pair, *next_pair; nvlist_t *errors; int err, rv = 0; if (props == NULL) return (0); VERIFY(nvlist_alloc(&errors, NV_UNIQUE_NAME, KM_SLEEP) == 0); - zc = kmem_alloc(sizeof (zfs_cmd_t), KM_SLEEP | KM_NODEBUG); + zc = kmem_alloc(sizeof (zfs_cmd_t), KM_SLEEP); (void) strcpy(zc->zc_name, dataset); pair = nvlist_next_nvpair(props, NULL); while (pair != NULL) { next_pair = nvlist_next_nvpair(props, pair); (void) strcpy(zc->zc_value, nvpair_name(pair)); if ((err = zfs_check_settable(dataset, pair, CRED())) != 0 || (err = zfs_secpolicy_inherit_prop(zc, NULL, CRED())) != 0) { VERIFY(nvlist_remove_nvpair(props, pair) == 0); VERIFY(nvlist_add_int32(errors, zc->zc_value, err) == 0); } pair = next_pair; } kmem_free(zc, sizeof (zfs_cmd_t)); if ((pair = nvlist_next_nvpair(errors, NULL)) == NULL) { nvlist_free(errors); errors = NULL; } else { VERIFY(nvpair_value_int32(pair, &rv) == 0); } if (errlist == NULL) nvlist_free(errors); else *errlist = errors; return (rv); } static boolean_t propval_equals(nvpair_t *p1, nvpair_t *p2) { if (nvpair_type(p1) == DATA_TYPE_NVLIST) { /* dsl_prop_get_all_impl() format */ nvlist_t *attrs; VERIFY(nvpair_value_nvlist(p1, &attrs) == 0); VERIFY(nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &p1) == 0); } if (nvpair_type(p2) == DATA_TYPE_NVLIST) { nvlist_t *attrs; VERIFY(nvpair_value_nvlist(p2, &attrs) == 0); VERIFY(nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &p2) == 0); } if (nvpair_type(p1) != nvpair_type(p2)) return (B_FALSE); if (nvpair_type(p1) == DATA_TYPE_STRING) { char *valstr1, *valstr2; VERIFY(nvpair_value_string(p1, (char **)&valstr1) == 0); VERIFY(nvpair_value_string(p2, (char **)&valstr2) == 0); return (strcmp(valstr1, valstr2) == 0); } else { uint64_t intval1, intval2; VERIFY(nvpair_value_uint64(p1, &intval1) == 0); VERIFY(nvpair_value_uint64(p2, &intval2) == 0); return (intval1 == intval2); } } /* * Remove properties from props if they are not going to change (as determined * by comparison with origprops). Remove them from origprops as well, since we * do not need to clear or restore properties that won't change. */ static void props_reduce(nvlist_t *props, nvlist_t *origprops) { nvpair_t *pair, *next_pair; if (origprops == NULL) return; /* all props need to be received */ pair = nvlist_next_nvpair(props, NULL); while (pair != NULL) { const char *propname = nvpair_name(pair); nvpair_t *match; next_pair = nvlist_next_nvpair(props, pair); if ((nvlist_lookup_nvpair(origprops, propname, &match) != 0) || !propval_equals(pair, match)) goto next; /* need to set received value */ /* don't clear the existing received value */ (void) nvlist_remove_nvpair(origprops, match); /* don't bother receiving the property */ (void) nvlist_remove_nvpair(props, pair); next: pair = next_pair; } } #ifdef DEBUG static boolean_t zfs_ioc_recv_inject_err; #endif /* * inputs: * zc_name name of containing filesystem * zc_nvlist_src{_size} nvlist of properties to apply * zc_value name of snapshot to create * zc_string name of clone origin (if DRR_FLAG_CLONE) * zc_cookie file descriptor to recv from * zc_begin_record the BEGIN record of the stream (not byteswapped) * zc_guid force flag * zc_cleanup_fd cleanup-on-exit file descriptor * zc_action_handle handle for this guid/ds mapping (or zero on first call) * * outputs: * zc_cookie number of bytes read * zc_nvlist_dst{_size} error for each unapplied received property * zc_obj zprop_errflags_t * zc_action_handle handle for this guid/ds mapping */ static int zfs_ioc_recv(zfs_cmd_t *zc) { file_t *fp; dmu_recv_cookie_t drc; boolean_t force = (boolean_t)zc->zc_guid; int fd; int error = 0; int props_error = 0; nvlist_t *errors; offset_t off; nvlist_t *props = NULL; /* sent properties */ nvlist_t *origprops = NULL; /* existing properties */ char *origin = NULL; char *tosnap; char tofs[ZFS_MAXNAMELEN]; boolean_t first_recvd_props = B_FALSE; if (dataset_namecheck(zc->zc_value, NULL, NULL) != 0 || strchr(zc->zc_value, '@') == NULL || strchr(zc->zc_value, '%')) return (SET_ERROR(EINVAL)); (void) strcpy(tofs, zc->zc_value); tosnap = strchr(tofs, '@'); *tosnap++ = '\0'; if (zc->zc_nvlist_src != 0 && (error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &props)) != 0) return (error); fd = zc->zc_cookie; fp = getf(fd); if (fp == NULL) { nvlist_free(props); return (SET_ERROR(EBADF)); } VERIFY(nvlist_alloc(&errors, NV_UNIQUE_NAME, KM_SLEEP) == 0); if (zc->zc_string[0]) origin = zc->zc_string; error = dmu_recv_begin(tofs, tosnap, &zc->zc_begin_record, force, origin, &drc); if (error != 0) goto out; /* * Set properties before we receive the stream so that they are applied * to the new data. Note that we must call dmu_recv_stream() if * dmu_recv_begin() succeeds. */ if (props != NULL && !drc.drc_newfs) { if (spa_version(dsl_dataset_get_spa(drc.drc_ds)) >= SPA_VERSION_RECVD_PROPS && !dsl_prop_get_hasrecvd(tofs)) first_recvd_props = B_TRUE; /* * If new received properties are supplied, they are to * completely replace the existing received properties, so stash * away the existing ones. */ if (dsl_prop_get_received(tofs, &origprops) == 0) { nvlist_t *errlist = NULL; /* * Don't bother writing a property if its value won't * change (and avoid the unnecessary security checks). * * The first receive after SPA_VERSION_RECVD_PROPS is a * special case where we blow away all local properties * regardless. */ if (!first_recvd_props) props_reduce(props, origprops); if (zfs_check_clearable(tofs, origprops, &errlist) != 0) (void) nvlist_merge(errors, errlist, 0); nvlist_free(errlist); if (clear_received_props(tofs, origprops, first_recvd_props ? NULL : props) != 0) zc->zc_obj |= ZPROP_ERR_NOCLEAR; } else { zc->zc_obj |= ZPROP_ERR_NOCLEAR; } } if (props != NULL) { props_error = dsl_prop_set_hasrecvd(tofs); if (props_error == 0) { (void) zfs_set_prop_nvlist(tofs, ZPROP_SRC_RECEIVED, props, errors); } } if (zc->zc_nvlist_dst_size != 0 && (nvlist_smush(errors, zc->zc_nvlist_dst_size) != 0 || put_nvlist(zc, errors) != 0)) { /* * Caller made zc->zc_nvlist_dst less than the minimum expected * size or supplied an invalid address. */ props_error = SET_ERROR(EINVAL); } off = fp->f_offset; error = dmu_recv_stream(&drc, fp->f_vnode, &off, zc->zc_cleanup_fd, &zc->zc_action_handle); if (error == 0) { zfs_sb_t *zsb = NULL; if (get_zfs_sb(tofs, &zsb) == 0) { /* online recv */ int end_err; error = zfs_suspend_fs(zsb); /* * If the suspend fails, then the recv_end will * likely also fail, and clean up after itself. */ end_err = dmu_recv_end(&drc, zsb); if (error == 0) error = zfs_resume_fs(zsb, tofs); error = error ? error : end_err; deactivate_super(zsb->z_sb); } else { error = dmu_recv_end(&drc, NULL); } } zc->zc_cookie = off - fp->f_offset; if (VOP_SEEK(fp->f_vnode, fp->f_offset, &off, NULL) == 0) fp->f_offset = off; #ifdef DEBUG if (zfs_ioc_recv_inject_err) { zfs_ioc_recv_inject_err = B_FALSE; error = 1; } #endif #ifdef _KERNEL if (error == 0) zvol_create_minors(tofs); #endif /* * On error, restore the original props. */ if (error != 0 && props != NULL && !drc.drc_newfs) { if (clear_received_props(tofs, props, NULL) != 0) { /* * We failed to clear the received properties. * Since we may have left a $recvd value on the * system, we can't clear the $hasrecvd flag. */ zc->zc_obj |= ZPROP_ERR_NORESTORE; } else if (first_recvd_props) { dsl_prop_unset_hasrecvd(tofs); } if (origprops == NULL && !drc.drc_newfs) { /* We failed to stash the original properties. */ zc->zc_obj |= ZPROP_ERR_NORESTORE; } /* * dsl_props_set() will not convert RECEIVED to LOCAL on or * after SPA_VERSION_RECVD_PROPS, so we need to specify LOCAL * explictly if we're restoring local properties cleared in the * first new-style receive. */ if (origprops != NULL && zfs_set_prop_nvlist(tofs, (first_recvd_props ? ZPROP_SRC_LOCAL : ZPROP_SRC_RECEIVED), origprops, NULL) != 0) { /* * We stashed the original properties but failed to * restore them. */ zc->zc_obj |= ZPROP_ERR_NORESTORE; } } out: nvlist_free(props); nvlist_free(origprops); nvlist_free(errors); releasef(fd); if (error == 0) error = props_error; return (error); } /* * inputs: * zc_name name of snapshot to send * zc_cookie file descriptor to send stream to * zc_obj fromorigin flag (mutually exclusive with zc_fromobj) * zc_sendobj objsetid of snapshot to send * zc_fromobj objsetid of incremental fromsnap (may be zero) * zc_guid if set, estimate size of stream only. zc_cookie is ignored. * output size in zc_objset_type. * zc_flags if =1, WRITE_EMBEDDED records are permitted * * outputs: * zc_objset_type estimated size, if zc_guid is set */ static int zfs_ioc_send(zfs_cmd_t *zc) { int error; offset_t off; boolean_t estimate = (zc->zc_guid != 0); boolean_t embedok = (zc->zc_flags & 0x1); if (zc->zc_obj != 0) { dsl_pool_t *dp; dsl_dataset_t *tosnap; error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold_obj(dp, zc->zc_sendobj, FTAG, &tosnap); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } if (dsl_dir_is_clone(tosnap->ds_dir)) zc->zc_fromobj = tosnap->ds_dir->dd_phys->dd_origin_obj; dsl_dataset_rele(tosnap, FTAG); dsl_pool_rele(dp, FTAG); } if (estimate) { dsl_pool_t *dp; dsl_dataset_t *tosnap; dsl_dataset_t *fromsnap = NULL; error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold_obj(dp, zc->zc_sendobj, FTAG, &tosnap); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } if (zc->zc_fromobj != 0) { error = dsl_dataset_hold_obj(dp, zc->zc_fromobj, FTAG, &fromsnap); if (error != 0) { dsl_dataset_rele(tosnap, FTAG); dsl_pool_rele(dp, FTAG); return (error); } } error = dmu_send_estimate(tosnap, fromsnap, &zc->zc_objset_type); if (fromsnap != NULL) dsl_dataset_rele(fromsnap, FTAG); dsl_dataset_rele(tosnap, FTAG); dsl_pool_rele(dp, FTAG); } else { file_t *fp = getf(zc->zc_cookie); if (fp == NULL) return (SET_ERROR(EBADF)); off = fp->f_offset; error = dmu_send_obj(zc->zc_name, zc->zc_sendobj, zc->zc_fromobj, embedok, zc->zc_cookie, fp->f_vnode, &off); if (VOP_SEEK(fp->f_vnode, fp->f_offset, &off, NULL) == 0) fp->f_offset = off; releasef(zc->zc_cookie); } return (error); } /* * inputs: * zc_name name of snapshot on which to report progress * zc_cookie file descriptor of send stream * * outputs: * zc_cookie number of bytes written in send stream thus far */ static int zfs_ioc_send_progress(zfs_cmd_t *zc) { dsl_pool_t *dp; dsl_dataset_t *ds; dmu_sendarg_t *dsp = NULL; int error; error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, zc->zc_name, FTAG, &ds); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } mutex_enter(&ds->ds_sendstream_lock); /* * Iterate over all the send streams currently active on this dataset. * If there's one which matches the specified file descriptor _and_ the * stream was started by the current process, return the progress of * that stream. */ for (dsp = list_head(&ds->ds_sendstreams); dsp != NULL; dsp = list_next(&ds->ds_sendstreams, dsp)) { if (dsp->dsa_outfd == zc->zc_cookie && dsp->dsa_proc->group_leader == curproc->group_leader) break; } if (dsp != NULL) zc->zc_cookie = *(dsp->dsa_off); else error = SET_ERROR(ENOENT); mutex_exit(&ds->ds_sendstream_lock); dsl_dataset_rele(ds, FTAG); dsl_pool_rele(dp, FTAG); return (error); } static int zfs_ioc_inject_fault(zfs_cmd_t *zc) { int id, error; error = zio_inject_fault(zc->zc_name, (int)zc->zc_guid, &id, &zc->zc_inject_record); if (error == 0) zc->zc_guid = (uint64_t)id; return (error); } static int zfs_ioc_clear_fault(zfs_cmd_t *zc) { return (zio_clear_fault((int)zc->zc_guid)); } static int zfs_ioc_inject_list_next(zfs_cmd_t *zc) { int id = (int)zc->zc_guid; int error; error = zio_inject_list_next(&id, zc->zc_name, sizeof (zc->zc_name), &zc->zc_inject_record); zc->zc_guid = id; return (error); } static int zfs_ioc_error_log(zfs_cmd_t *zc) { spa_t *spa; int error; size_t count = (size_t)zc->zc_nvlist_dst_size; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); error = spa_get_errlog(spa, (void *)(uintptr_t)zc->zc_nvlist_dst, &count); if (error == 0) zc->zc_nvlist_dst_size = count; else zc->zc_nvlist_dst_size = spa_get_errlog_size(spa); spa_close(spa, FTAG); return (error); } static int zfs_ioc_clear(zfs_cmd_t *zc) { spa_t *spa; vdev_t *vd; int error; /* * On zpool clear we also fix up missing slogs */ mutex_enter(&spa_namespace_lock); spa = spa_lookup(zc->zc_name); if (spa == NULL) { mutex_exit(&spa_namespace_lock); return (SET_ERROR(EIO)); } if (spa_get_log_state(spa) == SPA_LOG_MISSING) { /* we need to let spa_open/spa_load clear the chains */ spa_set_log_state(spa, SPA_LOG_CLEAR); } spa->spa_last_open_failed = 0; mutex_exit(&spa_namespace_lock); if (zc->zc_cookie & ZPOOL_NO_REWIND) { error = spa_open(zc->zc_name, &spa, FTAG); } else { nvlist_t *policy; nvlist_t *config = NULL; if (zc->zc_nvlist_src == 0) return (SET_ERROR(EINVAL)); if ((error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &policy)) == 0) { error = spa_open_rewind(zc->zc_name, &spa, FTAG, policy, &config); if (config != NULL) { int err; if ((err = put_nvlist(zc, config)) != 0) error = err; nvlist_free(config); } nvlist_free(policy); } } if (error != 0) return (error); spa_vdev_state_enter(spa, SCL_NONE); if (zc->zc_guid == 0) { vd = NULL; } else { vd = spa_lookup_by_guid(spa, zc->zc_guid, B_TRUE); if (vd == NULL) { (void) spa_vdev_state_exit(spa, NULL, ENODEV); spa_close(spa, FTAG); return (SET_ERROR(ENODEV)); } } vdev_clear(spa, vd); (void) spa_vdev_state_exit(spa, NULL, 0); /* * Resume any suspended I/Os. */ if (zio_resume(spa) != 0) error = SET_ERROR(EIO); spa_close(spa, FTAG); return (error); } static int zfs_ioc_pool_reopen(zfs_cmd_t *zc) { spa_t *spa; int error; error = spa_open(zc->zc_name, &spa, FTAG); if (error != 0) return (error); spa_vdev_state_enter(spa, SCL_NONE); /* * If a resilver is already in progress then set the * spa_scrub_reopen flag to B_TRUE so that we don't restart * the scan as a side effect of the reopen. Otherwise, let * vdev_open() decided if a resilver is required. */ spa->spa_scrub_reopen = dsl_scan_resilvering(spa->spa_dsl_pool); vdev_reopen(spa->spa_root_vdev); spa->spa_scrub_reopen = B_FALSE; (void) spa_vdev_state_exit(spa, NULL, 0); spa_close(spa, FTAG); return (0); } /* * inputs: * zc_name name of filesystem * zc_value name of origin snapshot * * outputs: * zc_string name of conflicting snapshot, if there is one */ static int zfs_ioc_promote(zfs_cmd_t *zc) { char *cp; /* * We don't need to unmount *all* the origin fs's snapshots, but * it's easier. */ cp = strchr(zc->zc_value, '@'); if (cp) *cp = '\0'; (void) dmu_objset_find(zc->zc_value, zfs_unmount_snap_cb, NULL, DS_FIND_SNAPSHOTS); return (dsl_dataset_promote(zc->zc_name, zc->zc_string)); } /* * Retrieve a single {user|group}{used|quota}@... property. * * inputs: * zc_name name of filesystem * zc_objset_type zfs_userquota_prop_t * zc_value domain name (eg. "S-1-234-567-89") * zc_guid RID/UID/GID * * outputs: * zc_cookie property value */ static int zfs_ioc_userspace_one(zfs_cmd_t *zc) { zfs_sb_t *zsb; int error; if (zc->zc_objset_type >= ZFS_NUM_USERQUOTA_PROPS) return (SET_ERROR(EINVAL)); error = zfs_sb_hold(zc->zc_name, FTAG, &zsb, B_FALSE); if (error != 0) return (error); error = zfs_userspace_one(zsb, zc->zc_objset_type, zc->zc_value, zc->zc_guid, &zc->zc_cookie); zfs_sb_rele(zsb, FTAG); return (error); } /* * inputs: * zc_name name of filesystem * zc_cookie zap cursor * zc_objset_type zfs_userquota_prop_t * zc_nvlist_dst[_size] buffer to fill (not really an nvlist) * * outputs: * zc_nvlist_dst[_size] data buffer (array of zfs_useracct_t) * zc_cookie zap cursor */ static int zfs_ioc_userspace_many(zfs_cmd_t *zc) { zfs_sb_t *zsb; int bufsize = zc->zc_nvlist_dst_size; int error; void *buf; if (bufsize <= 0) return (SET_ERROR(ENOMEM)); error = zfs_sb_hold(zc->zc_name, FTAG, &zsb, B_FALSE); if (error != 0) return (error); buf = vmem_alloc(bufsize, KM_SLEEP); error = zfs_userspace_many(zsb, zc->zc_objset_type, &zc->zc_cookie, buf, &zc->zc_nvlist_dst_size); if (error == 0) { error = xcopyout(buf, (void *)(uintptr_t)zc->zc_nvlist_dst, zc->zc_nvlist_dst_size); } vmem_free(buf, bufsize); zfs_sb_rele(zsb, FTAG); return (error); } /* * inputs: * zc_name name of filesystem * * outputs: * none */ static int zfs_ioc_userspace_upgrade(zfs_cmd_t *zc) { objset_t *os; int error = 0; zfs_sb_t *zsb; if (get_zfs_sb(zc->zc_name, &zsb) == 0) { if (!dmu_objset_userused_enabled(zsb->z_os)) { /* * If userused is not enabled, it may be because the * objset needs to be closed & reopened (to grow the * objset_phys_t). Suspend/resume the fs will do that. */ error = zfs_suspend_fs(zsb); if (error == 0) { dmu_objset_refresh_ownership(zsb->z_os, zsb); error = zfs_resume_fs(zsb, zc->zc_name); } } if (error == 0) error = dmu_objset_userspace_upgrade(zsb->z_os); deactivate_super(zsb->z_sb); } else { /* XXX kind of reading contents without owning */ error = dmu_objset_hold(zc->zc_name, FTAG, &os); if (error != 0) return (error); error = dmu_objset_userspace_upgrade(os); dmu_objset_rele(os, FTAG); } return (error); } static int zfs_ioc_share(zfs_cmd_t *zc) { return (SET_ERROR(ENOSYS)); } ace_t full_access[] = { {(uid_t)-1, ACE_ALL_PERMS, ACE_EVERYONE, 0} }; /* * inputs: * zc_name name of containing filesystem * zc_obj object # beyond which we want next in-use object # * * outputs: * zc_obj next in-use object # */ static int zfs_ioc_next_obj(zfs_cmd_t *zc) { objset_t *os = NULL; int error; error = dmu_objset_hold(zc->zc_name, FTAG, &os); if (error != 0) return (error); error = dmu_object_next(os, &zc->zc_obj, B_FALSE, os->os_dsl_dataset->ds_phys->ds_prev_snap_txg); dmu_objset_rele(os, FTAG); return (error); } /* * inputs: * zc_name name of filesystem * zc_value prefix name for snapshot * zc_cleanup_fd cleanup-on-exit file descriptor for calling process * * outputs: * zc_value short name of new snapshot */ static int zfs_ioc_tmp_snapshot(zfs_cmd_t *zc) { char *snap_name; char *hold_name; int error; minor_t minor; error = zfs_onexit_fd_hold(zc->zc_cleanup_fd, &minor); if (error != 0) return (error); snap_name = kmem_asprintf("%s-%016llx", zc->zc_value, (u_longlong_t)ddi_get_lbolt64()); hold_name = kmem_asprintf("%%%s", zc->zc_value); error = dsl_dataset_snapshot_tmp(zc->zc_name, snap_name, minor, hold_name); if (error == 0) (void) strcpy(zc->zc_value, snap_name); strfree(snap_name); strfree(hold_name); zfs_onexit_fd_rele(zc->zc_cleanup_fd); return (error); } /* * inputs: * zc_name name of "to" snapshot * zc_value name of "from" snapshot * zc_cookie file descriptor to write diff data on * * outputs: * dmu_diff_record_t's to the file descriptor */ static int zfs_ioc_diff(zfs_cmd_t *zc) { file_t *fp; offset_t off; int error; fp = getf(zc->zc_cookie); if (fp == NULL) return (SET_ERROR(EBADF)); off = fp->f_offset; error = dmu_diff(zc->zc_name, zc->zc_value, fp->f_vnode, &off); if (VOP_SEEK(fp->f_vnode, fp->f_offset, &off, NULL) == 0) fp->f_offset = off; releasef(zc->zc_cookie); return (error); } /* * Remove all ACL files in shares dir */ #ifdef HAVE_SMB_SHARE static int zfs_smb_acl_purge(znode_t *dzp) { zap_cursor_t zc; zap_attribute_t zap; zfs_sb_t *zsb = ZTOZSB(dzp); int error; for (zap_cursor_init(&zc, zsb->z_os, dzp->z_id); (error = zap_cursor_retrieve(&zc, &zap)) == 0; zap_cursor_advance(&zc)) { if ((error = VOP_REMOVE(ZTOV(dzp), zap.za_name, kcred, NULL, 0)) != 0) break; } zap_cursor_fini(&zc); return (error); } #endif /* HAVE_SMB_SHARE */ static int zfs_ioc_smb_acl(zfs_cmd_t *zc) { #ifdef HAVE_SMB_SHARE vnode_t *vp; znode_t *dzp; vnode_t *resourcevp = NULL; znode_t *sharedir; zfs_sb_t *zsb; nvlist_t *nvlist; char *src, *target; vattr_t vattr; vsecattr_t vsec; int error = 0; if ((error = lookupname(zc->zc_value, UIO_SYSSPACE, NO_FOLLOW, NULL, &vp)) != 0) return (error); /* Now make sure mntpnt and dataset are ZFS */ if (vp->v_vfsp->vfs_fstype != zfsfstype || (strcmp((char *)refstr_value(vp->v_vfsp->vfs_resource), zc->zc_name) != 0)) { VN_RELE(vp); return (SET_ERROR(EINVAL)); } dzp = VTOZ(vp); zsb = ZTOZSB(dzp); ZFS_ENTER(zsb); /* * Create share dir if its missing. */ mutex_enter(&zsb->z_lock); if (zsb->z_shares_dir == 0) { dmu_tx_t *tx; tx = dmu_tx_create(zsb->z_os); dmu_tx_hold_zap(tx, MASTER_NODE_OBJ, TRUE, ZFS_SHARES_DIR); dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL); error = dmu_tx_assign(tx, TXG_WAIT); if (error != 0) { dmu_tx_abort(tx); } else { error = zfs_create_share_dir(zsb, tx); dmu_tx_commit(tx); } if (error != 0) { mutex_exit(&zsb->z_lock); VN_RELE(vp); ZFS_EXIT(zsb); return (error); } } mutex_exit(&zsb->z_lock); ASSERT(zsb->z_shares_dir); if ((error = zfs_zget(zsb, zsb->z_shares_dir, &sharedir)) != 0) { VN_RELE(vp); ZFS_EXIT(zsb); return (error); } switch (zc->zc_cookie) { case ZFS_SMB_ACL_ADD: vattr.va_mask = AT_MODE|AT_UID|AT_GID|AT_TYPE; vattr.va_mode = S_IFREG|0777; vattr.va_uid = 0; vattr.va_gid = 0; vsec.vsa_mask = VSA_ACE; vsec.vsa_aclentp = &full_access; vsec.vsa_aclentsz = sizeof (full_access); vsec.vsa_aclcnt = 1; error = VOP_CREATE(ZTOV(sharedir), zc->zc_string, &vattr, EXCL, 0, &resourcevp, kcred, 0, NULL, &vsec); if (resourcevp) VN_RELE(resourcevp); break; case ZFS_SMB_ACL_REMOVE: error = VOP_REMOVE(ZTOV(sharedir), zc->zc_string, kcred, NULL, 0); break; case ZFS_SMB_ACL_RENAME: if ((error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &nvlist)) != 0) { VN_RELE(vp); ZFS_EXIT(zsb); return (error); } if (nvlist_lookup_string(nvlist, ZFS_SMB_ACL_SRC, &src) || nvlist_lookup_string(nvlist, ZFS_SMB_ACL_TARGET, &target)) { VN_RELE(vp); VN_RELE(ZTOV(sharedir)); ZFS_EXIT(zsb); nvlist_free(nvlist); return (error); } error = VOP_RENAME(ZTOV(sharedir), src, ZTOV(sharedir), target, kcred, NULL, 0); nvlist_free(nvlist); break; case ZFS_SMB_ACL_PURGE: error = zfs_smb_acl_purge(sharedir); break; default: error = SET_ERROR(EINVAL); break; } VN_RELE(vp); VN_RELE(ZTOV(sharedir)); ZFS_EXIT(zsb); return (error); #else return (SET_ERROR(ENOTSUP)); #endif /* HAVE_SMB_SHARE */ } /* * innvl: { * "holds" -> { snapname -> holdname (string), ... } * (optional) "cleanup_fd" -> fd (int32) * } * * outnvl: { * snapname -> error value (int32) * ... * } */ /* ARGSUSED */ static int zfs_ioc_hold(const char *pool, nvlist_t *args, nvlist_t *errlist) { nvlist_t *holds; int cleanup_fd = -1; int error; minor_t minor = 0; error = nvlist_lookup_nvlist(args, "holds", &holds); if (error != 0) return (SET_ERROR(EINVAL)); if (nvlist_lookup_int32(args, "cleanup_fd", &cleanup_fd) == 0) { error = zfs_onexit_fd_hold(cleanup_fd, &minor); if (error != 0) return (error); } error = dsl_dataset_user_hold(holds, minor, errlist); if (minor != 0) zfs_onexit_fd_rele(cleanup_fd); return (error); } /* * innvl is not used. * * outnvl: { * holdname -> time added (uint64 seconds since epoch) * ... * } */ /* ARGSUSED */ static int zfs_ioc_get_holds(const char *snapname, nvlist_t *args, nvlist_t *outnvl) { return (dsl_dataset_get_holds(snapname, outnvl)); } /* * innvl: { * snapname -> { holdname, ... } * ... * } * * outnvl: { * snapname -> error value (int32) * ... * } */ /* ARGSUSED */ static int zfs_ioc_release(const char *pool, nvlist_t *holds, nvlist_t *errlist) { return (dsl_dataset_user_release(holds, errlist)); } /* * inputs: * zc_guid flags (ZEVENT_NONBLOCK) * zc_cleanup_fd zevent file descriptor * * outputs: * zc_nvlist_dst next nvlist event * zc_cookie dropped events since last get */ static int zfs_ioc_events_next(zfs_cmd_t *zc) { zfs_zevent_t *ze; nvlist_t *event = NULL; minor_t minor; uint64_t dropped = 0; int error; error = zfs_zevent_fd_hold(zc->zc_cleanup_fd, &minor, &ze); if (error != 0) return (error); do { error = zfs_zevent_next(ze, &event, &zc->zc_nvlist_dst_size, &dropped); if (event != NULL) { zc->zc_cookie = dropped; error = put_nvlist(zc, event); nvlist_free(event); } if (zc->zc_guid & ZEVENT_NONBLOCK) break; if ((error == 0) || (error != ENOENT)) break; error = zfs_zevent_wait(ze); if (error != 0) break; } while (1); zfs_zevent_fd_rele(zc->zc_cleanup_fd); return (error); } /* * outputs: * zc_cookie cleared events count */ static int zfs_ioc_events_clear(zfs_cmd_t *zc) { int count; zfs_zevent_drain_all(&count); zc->zc_cookie = count; return (0); } /* * inputs: * zc_guid eid | ZEVENT_SEEK_START | ZEVENT_SEEK_END * zc_cleanup zevent file descriptor */ static int zfs_ioc_events_seek(zfs_cmd_t *zc) { zfs_zevent_t *ze; minor_t minor; int error; error = zfs_zevent_fd_hold(zc->zc_cleanup_fd, &minor, &ze); if (error != 0) return (error); error = zfs_zevent_seek(ze, zc->zc_guid); zfs_zevent_fd_rele(zc->zc_cleanup_fd); return (error); } /* * inputs: * zc_name name of new filesystem or snapshot * zc_value full name of old snapshot * * outputs: * zc_cookie space in bytes * zc_objset_type compressed space in bytes * zc_perm_action uncompressed space in bytes */ static int zfs_ioc_space_written(zfs_cmd_t *zc) { int error; dsl_pool_t *dp; dsl_dataset_t *new, *old; error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, zc->zc_name, FTAG, &new); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } error = dsl_dataset_hold(dp, zc->zc_value, FTAG, &old); if (error != 0) { dsl_dataset_rele(new, FTAG); dsl_pool_rele(dp, FTAG); return (error); } error = dsl_dataset_space_written(old, new, &zc->zc_cookie, &zc->zc_objset_type, &zc->zc_perm_action); dsl_dataset_rele(old, FTAG); dsl_dataset_rele(new, FTAG); dsl_pool_rele(dp, FTAG); return (error); } /* * innvl: { * "firstsnap" -> snapshot name * } * * outnvl: { * "used" -> space in bytes * "compressed" -> compressed space in bytes * "uncompressed" -> uncompressed space in bytes * } */ static int zfs_ioc_space_snaps(const char *lastsnap, nvlist_t *innvl, nvlist_t *outnvl) { int error; dsl_pool_t *dp; dsl_dataset_t *new, *old; char *firstsnap; uint64_t used, comp, uncomp; if (nvlist_lookup_string(innvl, "firstsnap", &firstsnap) != 0) return (SET_ERROR(EINVAL)); error = dsl_pool_hold(lastsnap, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, lastsnap, FTAG, &new); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } error = dsl_dataset_hold(dp, firstsnap, FTAG, &old); if (error != 0) { dsl_dataset_rele(new, FTAG); dsl_pool_rele(dp, FTAG); return (error); } error = dsl_dataset_space_wouldfree(old, new, &used, &comp, &uncomp); dsl_dataset_rele(old, FTAG); dsl_dataset_rele(new, FTAG); dsl_pool_rele(dp, FTAG); fnvlist_add_uint64(outnvl, "used", used); fnvlist_add_uint64(outnvl, "compressed", comp); fnvlist_add_uint64(outnvl, "uncompressed", uncomp); return (error); } /* * innvl: { * "fd" -> file descriptor to write stream to (int32) * (optional) "fromsnap" -> full snap name to send an incremental from * (optional) "embedok" -> (value ignored) * presence indicates DRR_WRITE_EMBEDDED records are permitted * } * * outnvl is unused */ /* ARGSUSED */ static int zfs_ioc_send_new(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl) { int error; offset_t off; char *fromname = NULL; int fd; file_t *fp; boolean_t embedok; error = nvlist_lookup_int32(innvl, "fd", &fd); if (error != 0) return (SET_ERROR(EINVAL)); (void) nvlist_lookup_string(innvl, "fromsnap", &fromname); embedok = nvlist_exists(innvl, "embedok"); if ((fp = getf(fd)) == NULL) return (SET_ERROR(EBADF)); off = fp->f_offset; error = dmu_send(snapname, fromname, embedok, fd, fp->f_vnode, &off); if (VOP_SEEK(fp->f_vnode, fp->f_offset, &off, NULL) == 0) fp->f_offset = off; releasef(fd); return (error); } /* * Determine approximately how large a zfs send stream will be -- the number * of bytes that will be written to the fd supplied to zfs_ioc_send_new(). * * innvl: { * (optional) "fromsnap" -> full snap name to send an incremental from * } * * outnvl: { * "space" -> bytes of space (uint64) * } */ static int zfs_ioc_send_space(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl) { dsl_pool_t *dp; dsl_dataset_t *fromsnap = NULL; dsl_dataset_t *tosnap; int error; char *fromname; uint64_t space; error = dsl_pool_hold(snapname, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, snapname, FTAG, &tosnap); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } error = nvlist_lookup_string(innvl, "fromsnap", &fromname); if (error == 0) { error = dsl_dataset_hold(dp, fromname, FTAG, &fromsnap); if (error != 0) { dsl_dataset_rele(tosnap, FTAG); dsl_pool_rele(dp, FTAG); return (error); } } error = dmu_send_estimate(tosnap, fromsnap, &space); fnvlist_add_uint64(outnvl, "space", space); if (fromsnap != NULL) dsl_dataset_rele(fromsnap, FTAG); dsl_dataset_rele(tosnap, FTAG); dsl_pool_rele(dp, FTAG); return (error); } static zfs_ioc_vec_t zfs_ioc_vec[ZFS_IOC_LAST - ZFS_IOC_FIRST]; static void zfs_ioctl_register_legacy(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy, zfs_ioc_namecheck_t namecheck, boolean_t log_history, zfs_ioc_poolcheck_t pool_check) { zfs_ioc_vec_t *vec = &zfs_ioc_vec[ioc - ZFS_IOC_FIRST]; ASSERT3U(ioc, >=, ZFS_IOC_FIRST); ASSERT3U(ioc, <, ZFS_IOC_LAST); ASSERT3P(vec->zvec_legacy_func, ==, NULL); ASSERT3P(vec->zvec_func, ==, NULL); vec->zvec_legacy_func = func; vec->zvec_secpolicy = secpolicy; vec->zvec_namecheck = namecheck; vec->zvec_allow_log = log_history; vec->zvec_pool_check = pool_check; } /* * See the block comment at the beginning of this file for details on * each argument to this function. */ static void zfs_ioctl_register(const char *name, zfs_ioc_t ioc, zfs_ioc_func_t *func, zfs_secpolicy_func_t *secpolicy, zfs_ioc_namecheck_t namecheck, zfs_ioc_poolcheck_t pool_check, boolean_t smush_outnvlist, boolean_t allow_log) { zfs_ioc_vec_t *vec = &zfs_ioc_vec[ioc - ZFS_IOC_FIRST]; ASSERT3U(ioc, >=, ZFS_IOC_FIRST); ASSERT3U(ioc, <, ZFS_IOC_LAST); ASSERT3P(vec->zvec_legacy_func, ==, NULL); ASSERT3P(vec->zvec_func, ==, NULL); /* if we are logging, the name must be valid */ ASSERT(!allow_log || namecheck != NO_NAME); vec->zvec_name = name; vec->zvec_func = func; vec->zvec_secpolicy = secpolicy; vec->zvec_namecheck = namecheck; vec->zvec_pool_check = pool_check; vec->zvec_smush_outnvlist = smush_outnvlist; vec->zvec_allow_log = allow_log; } static void zfs_ioctl_register_pool(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy, boolean_t log_history, zfs_ioc_poolcheck_t pool_check) { zfs_ioctl_register_legacy(ioc, func, secpolicy, POOL_NAME, log_history, pool_check); } static void zfs_ioctl_register_dataset_nolog(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy, zfs_ioc_poolcheck_t pool_check) { zfs_ioctl_register_legacy(ioc, func, secpolicy, DATASET_NAME, B_FALSE, pool_check); } static void zfs_ioctl_register_pool_modify(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func) { zfs_ioctl_register_legacy(ioc, func, zfs_secpolicy_config, POOL_NAME, B_TRUE, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY); } static void zfs_ioctl_register_pool_meta(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy) { zfs_ioctl_register_legacy(ioc, func, secpolicy, NO_NAME, B_FALSE, POOL_CHECK_NONE); } static void zfs_ioctl_register_dataset_read_secpolicy(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy) { zfs_ioctl_register_legacy(ioc, func, secpolicy, DATASET_NAME, B_FALSE, POOL_CHECK_SUSPENDED); } static void zfs_ioctl_register_dataset_read(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func) { zfs_ioctl_register_dataset_read_secpolicy(ioc, func, zfs_secpolicy_read); } static void zfs_ioctl_register_dataset_modify(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy) { zfs_ioctl_register_legacy(ioc, func, secpolicy, DATASET_NAME, B_TRUE, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY); } static void zfs_ioctl_init(void) { zfs_ioctl_register("snapshot", ZFS_IOC_SNAPSHOT, zfs_ioc_snapshot, zfs_secpolicy_snapshot, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); zfs_ioctl_register("log_history", ZFS_IOC_LOG_HISTORY, zfs_ioc_log_history, zfs_secpolicy_log_history, NO_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_FALSE, B_FALSE); zfs_ioctl_register("space_snaps", ZFS_IOC_SPACE_SNAPS, zfs_ioc_space_snaps, zfs_secpolicy_read, DATASET_NAME, POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE); zfs_ioctl_register("send", ZFS_IOC_SEND_NEW, zfs_ioc_send_new, zfs_secpolicy_send_new, DATASET_NAME, POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE); zfs_ioctl_register("send_space", ZFS_IOC_SEND_SPACE, zfs_ioc_send_space, zfs_secpolicy_read, DATASET_NAME, POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE); zfs_ioctl_register("create", ZFS_IOC_CREATE, zfs_ioc_create, zfs_secpolicy_create_clone, DATASET_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); zfs_ioctl_register("clone", ZFS_IOC_CLONE, zfs_ioc_clone, zfs_secpolicy_create_clone, DATASET_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); zfs_ioctl_register("destroy_snaps", ZFS_IOC_DESTROY_SNAPS, zfs_ioc_destroy_snaps, zfs_secpolicy_destroy_snaps, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); zfs_ioctl_register("hold", ZFS_IOC_HOLD, zfs_ioc_hold, zfs_secpolicy_hold, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); zfs_ioctl_register("release", ZFS_IOC_RELEASE, zfs_ioc_release, zfs_secpolicy_release, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); zfs_ioctl_register("get_holds", ZFS_IOC_GET_HOLDS, zfs_ioc_get_holds, zfs_secpolicy_read, DATASET_NAME, POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE); zfs_ioctl_register("rollback", ZFS_IOC_ROLLBACK, zfs_ioc_rollback, zfs_secpolicy_rollback, DATASET_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_FALSE, B_TRUE); zfs_ioctl_register("bookmark", ZFS_IOC_BOOKMARK, zfs_ioc_bookmark, zfs_secpolicy_bookmark, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); zfs_ioctl_register("get_bookmarks", ZFS_IOC_GET_BOOKMARKS, zfs_ioc_get_bookmarks, zfs_secpolicy_read, DATASET_NAME, POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE); zfs_ioctl_register("destroy_bookmarks", ZFS_IOC_DESTROY_BOOKMARKS, zfs_ioc_destroy_bookmarks, zfs_secpolicy_destroy_bookmarks, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); /* IOCTLS that use the legacy function signature */ zfs_ioctl_register_legacy(ZFS_IOC_POOL_FREEZE, zfs_ioc_pool_freeze, zfs_secpolicy_config, NO_NAME, B_FALSE, POOL_CHECK_READONLY); zfs_ioctl_register_pool(ZFS_IOC_POOL_CREATE, zfs_ioc_pool_create, zfs_secpolicy_config, B_TRUE, POOL_CHECK_NONE); zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_SCAN, zfs_ioc_pool_scan); zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_UPGRADE, zfs_ioc_pool_upgrade); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_ADD, zfs_ioc_vdev_add); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_REMOVE, zfs_ioc_vdev_remove); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SET_STATE, zfs_ioc_vdev_set_state); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_ATTACH, zfs_ioc_vdev_attach); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_DETACH, zfs_ioc_vdev_detach); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SETPATH, zfs_ioc_vdev_setpath); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SETFRU, zfs_ioc_vdev_setfru); zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_SET_PROPS, zfs_ioc_pool_set_props); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SPLIT, zfs_ioc_vdev_split); zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_REGUID, zfs_ioc_pool_reguid); zfs_ioctl_register_pool_meta(ZFS_IOC_POOL_CONFIGS, zfs_ioc_pool_configs, zfs_secpolicy_none); zfs_ioctl_register_pool_meta(ZFS_IOC_POOL_TRYIMPORT, zfs_ioc_pool_tryimport, zfs_secpolicy_config); zfs_ioctl_register_pool_meta(ZFS_IOC_INJECT_FAULT, zfs_ioc_inject_fault, zfs_secpolicy_inject); zfs_ioctl_register_pool_meta(ZFS_IOC_CLEAR_FAULT, zfs_ioc_clear_fault, zfs_secpolicy_inject); zfs_ioctl_register_pool_meta(ZFS_IOC_INJECT_LIST_NEXT, zfs_ioc_inject_list_next, zfs_secpolicy_inject); /* * pool destroy, and export don't log the history as part of * zfsdev_ioctl, but rather zfs_ioc_pool_export * does the logging of those commands. */ zfs_ioctl_register_pool(ZFS_IOC_POOL_DESTROY, zfs_ioc_pool_destroy, zfs_secpolicy_config, B_FALSE, POOL_CHECK_NONE); zfs_ioctl_register_pool(ZFS_IOC_POOL_EXPORT, zfs_ioc_pool_export, zfs_secpolicy_config, B_FALSE, POOL_CHECK_NONE); zfs_ioctl_register_pool(ZFS_IOC_POOL_STATS, zfs_ioc_pool_stats, zfs_secpolicy_read, B_FALSE, POOL_CHECK_NONE); zfs_ioctl_register_pool(ZFS_IOC_POOL_GET_PROPS, zfs_ioc_pool_get_props, zfs_secpolicy_read, B_FALSE, POOL_CHECK_NONE); zfs_ioctl_register_pool(ZFS_IOC_ERROR_LOG, zfs_ioc_error_log, zfs_secpolicy_inject, B_FALSE, POOL_CHECK_SUSPENDED); zfs_ioctl_register_pool(ZFS_IOC_DSOBJ_TO_DSNAME, zfs_ioc_dsobj_to_dsname, zfs_secpolicy_diff, B_FALSE, POOL_CHECK_SUSPENDED); zfs_ioctl_register_pool(ZFS_IOC_POOL_GET_HISTORY, zfs_ioc_pool_get_history, zfs_secpolicy_config, B_FALSE, POOL_CHECK_SUSPENDED); zfs_ioctl_register_pool(ZFS_IOC_POOL_IMPORT, zfs_ioc_pool_import, zfs_secpolicy_config, B_TRUE, POOL_CHECK_NONE); zfs_ioctl_register_pool(ZFS_IOC_CLEAR, zfs_ioc_clear, zfs_secpolicy_config, B_TRUE, POOL_CHECK_NONE); zfs_ioctl_register_pool(ZFS_IOC_POOL_REOPEN, zfs_ioc_pool_reopen, zfs_secpolicy_config, B_TRUE, POOL_CHECK_SUSPENDED); zfs_ioctl_register_dataset_read(ZFS_IOC_SPACE_WRITTEN, zfs_ioc_space_written); zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_RECVD_PROPS, zfs_ioc_objset_recvd_props); zfs_ioctl_register_dataset_read(ZFS_IOC_NEXT_OBJ, zfs_ioc_next_obj); zfs_ioctl_register_dataset_read(ZFS_IOC_GET_FSACL, zfs_ioc_get_fsacl); zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_STATS, zfs_ioc_objset_stats); zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_ZPLPROPS, zfs_ioc_objset_zplprops); zfs_ioctl_register_dataset_read(ZFS_IOC_DATASET_LIST_NEXT, zfs_ioc_dataset_list_next); zfs_ioctl_register_dataset_read(ZFS_IOC_SNAPSHOT_LIST_NEXT, zfs_ioc_snapshot_list_next); zfs_ioctl_register_dataset_read(ZFS_IOC_SEND_PROGRESS, zfs_ioc_send_progress); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_DIFF, zfs_ioc_diff, zfs_secpolicy_diff); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_OBJ_TO_STATS, zfs_ioc_obj_to_stats, zfs_secpolicy_diff); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_OBJ_TO_PATH, zfs_ioc_obj_to_path, zfs_secpolicy_diff); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_USERSPACE_ONE, zfs_ioc_userspace_one, zfs_secpolicy_userspace_one); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_USERSPACE_MANY, zfs_ioc_userspace_many, zfs_secpolicy_userspace_many); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_SEND, zfs_ioc_send, zfs_secpolicy_send); zfs_ioctl_register_dataset_modify(ZFS_IOC_SET_PROP, zfs_ioc_set_prop, zfs_secpolicy_none); zfs_ioctl_register_dataset_modify(ZFS_IOC_DESTROY, zfs_ioc_destroy, zfs_secpolicy_destroy); zfs_ioctl_register_dataset_modify(ZFS_IOC_RENAME, zfs_ioc_rename, zfs_secpolicy_rename); zfs_ioctl_register_dataset_modify(ZFS_IOC_RECV, zfs_ioc_recv, zfs_secpolicy_recv); zfs_ioctl_register_dataset_modify(ZFS_IOC_PROMOTE, zfs_ioc_promote, zfs_secpolicy_promote); zfs_ioctl_register_dataset_modify(ZFS_IOC_INHERIT_PROP, zfs_ioc_inherit_prop, zfs_secpolicy_inherit_prop); zfs_ioctl_register_dataset_modify(ZFS_IOC_SET_FSACL, zfs_ioc_set_fsacl, zfs_secpolicy_set_fsacl); zfs_ioctl_register_dataset_nolog(ZFS_IOC_SHARE, zfs_ioc_share, zfs_secpolicy_share, POOL_CHECK_NONE); zfs_ioctl_register_dataset_nolog(ZFS_IOC_SMB_ACL, zfs_ioc_smb_acl, zfs_secpolicy_smb_acl, POOL_CHECK_NONE); zfs_ioctl_register_dataset_nolog(ZFS_IOC_USERSPACE_UPGRADE, zfs_ioc_userspace_upgrade, zfs_secpolicy_userspace_upgrade, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY); zfs_ioctl_register_dataset_nolog(ZFS_IOC_TMP_SNAPSHOT, zfs_ioc_tmp_snapshot, zfs_secpolicy_tmp_snapshot, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY); /* * ZoL functions */ zfs_ioctl_register_legacy(ZFS_IOC_EVENTS_NEXT, zfs_ioc_events_next, zfs_secpolicy_config, NO_NAME, B_FALSE, POOL_CHECK_NONE); zfs_ioctl_register_legacy(ZFS_IOC_EVENTS_CLEAR, zfs_ioc_events_clear, zfs_secpolicy_config, NO_NAME, B_FALSE, POOL_CHECK_NONE); zfs_ioctl_register_legacy(ZFS_IOC_EVENTS_SEEK, zfs_ioc_events_seek, zfs_secpolicy_config, NO_NAME, B_FALSE, POOL_CHECK_NONE); } int pool_status_check(const char *name, zfs_ioc_namecheck_t type, zfs_ioc_poolcheck_t check) { spa_t *spa; int error; ASSERT(type == POOL_NAME || type == DATASET_NAME); if (check & POOL_CHECK_NONE) return (0); error = spa_open(name, &spa, FTAG); if (error == 0) { if ((check & POOL_CHECK_SUSPENDED) && spa_suspended(spa)) error = SET_ERROR(EAGAIN); else if ((check & POOL_CHECK_READONLY) && !spa_writeable(spa)) error = SET_ERROR(EROFS); spa_close(spa, FTAG); } return (error); } static void * zfsdev_get_state_impl(minor_t minor, enum zfsdev_state_type which) { zfsdev_state_t *zs; for (zs = zfsdev_state_list; zs != NULL; zs = zs->zs_next) { if (zs->zs_minor == minor) { smp_rmb(); switch (which) { case ZST_ONEXIT: return (zs->zs_onexit); case ZST_ZEVENT: return (zs->zs_zevent); case ZST_ALL: return (zs); } } } return (NULL); } void * zfsdev_get_state(minor_t minor, enum zfsdev_state_type which) { void *ptr; ptr = zfsdev_get_state_impl(minor, which); return (ptr); } minor_t zfsdev_getminor(struct file *filp) { ASSERT(filp != NULL); ASSERT(filp->private_data != NULL); return (((zfsdev_state_t *)filp->private_data)->zs_minor); } /* * Find a free minor number. The zfsdev_state_list is expected to * be short since it is only a list of currently open file handles. */ minor_t zfsdev_minor_alloc(void) { static minor_t last_minor = 0; minor_t m; ASSERT(MUTEX_HELD(&zfsdev_state_lock)); for (m = last_minor + 1; m != last_minor; m++) { if (m > ZFSDEV_MAX_MINOR) m = 1; if (zfsdev_get_state_impl(m, ZST_ALL) == NULL) { last_minor = m; return (m); } } return (0); } static int zfsdev_state_init(struct file *filp) { zfsdev_state_t *zs, *zsprev = NULL; minor_t minor; boolean_t newzs = B_FALSE; ASSERT(MUTEX_HELD(&zfsdev_state_lock)); minor = zfsdev_minor_alloc(); if (minor == 0) return (SET_ERROR(ENXIO)); for (zs = zfsdev_state_list; zs != NULL; zs = zs->zs_next) { if (zs->zs_minor == -1) break; zsprev = zs; } if (!zs) { zs = kmem_zalloc(sizeof (zfsdev_state_t), KM_SLEEP); newzs = B_TRUE; } zs->zs_file = filp; filp->private_data = zs; zfs_onexit_init((zfs_onexit_t **)&zs->zs_onexit); zfs_zevent_init((zfs_zevent_t **)&zs->zs_zevent); /* * In order to provide for lock-free concurrent read access * to the minor list in zfsdev_get_state_impl(), new entries * must be completely written before linking them into the * list whereas existing entries are already linked; the last * operation must be updating zs_minor (from -1 to the new * value). */ if (newzs) { zs->zs_minor = minor; smp_wmb(); zsprev->zs_next = zs; } else { smp_wmb(); zs->zs_minor = minor; } return (0); } static int zfsdev_state_destroy(struct file *filp) { zfsdev_state_t *zs; ASSERT(MUTEX_HELD(&zfsdev_state_lock)); ASSERT(filp->private_data != NULL); zs = filp->private_data; zs->zs_minor = -1; zfs_onexit_destroy(zs->zs_onexit); zfs_zevent_destroy(zs->zs_zevent); return (0); } static int zfsdev_open(struct inode *ino, struct file *filp) { int error; mutex_enter(&zfsdev_state_lock); error = zfsdev_state_init(filp); mutex_exit(&zfsdev_state_lock); return (-error); } static int zfsdev_release(struct inode *ino, struct file *filp) { int error; mutex_enter(&zfsdev_state_lock); error = zfsdev_state_destroy(filp); mutex_exit(&zfsdev_state_lock); return (-error); } static long zfsdev_ioctl(struct file *filp, unsigned cmd, unsigned long arg) { zfs_cmd_t *zc; uint_t vecnum; int error, rc, flag = 0; const zfs_ioc_vec_t *vec; char *saved_poolname = NULL; nvlist_t *innvl = NULL; vecnum = cmd - ZFS_IOC_FIRST; if (vecnum >= sizeof (zfs_ioc_vec) / sizeof (zfs_ioc_vec[0])) return (-SET_ERROR(EINVAL)); vec = &zfs_ioc_vec[vecnum]; /* * The registered ioctl list may be sparse, verify that either * a normal or legacy handler are registered. */ if (vec->zvec_func == NULL && vec->zvec_legacy_func == NULL) return (-SET_ERROR(EINVAL)); - zc = kmem_zalloc(sizeof (zfs_cmd_t), KM_SLEEP | KM_NODEBUG); + zc = kmem_zalloc(sizeof (zfs_cmd_t), KM_SLEEP); error = ddi_copyin((void *)arg, zc, sizeof (zfs_cmd_t), flag); if (error != 0) { error = SET_ERROR(EFAULT); goto out; } zc->zc_iflags = flag & FKIOCTL; if (zc->zc_nvlist_src_size != 0) { error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &innvl); if (error != 0) goto out; } /* * Ensure that all pool/dataset names are valid before we pass down to * the lower layers. */ zc->zc_name[sizeof (zc->zc_name) - 1] = '\0'; switch (vec->zvec_namecheck) { case POOL_NAME: if (pool_namecheck(zc->zc_name, NULL, NULL) != 0) error = SET_ERROR(EINVAL); else error = pool_status_check(zc->zc_name, vec->zvec_namecheck, vec->zvec_pool_check); break; case DATASET_NAME: if (dataset_namecheck(zc->zc_name, NULL, NULL) != 0) error = SET_ERROR(EINVAL); else error = pool_status_check(zc->zc_name, vec->zvec_namecheck, vec->zvec_pool_check); break; case NO_NAME: break; } if (error == 0 && !(flag & FKIOCTL)) error = vec->zvec_secpolicy(zc, innvl, CRED()); if (error != 0) goto out; /* legacy ioctls can modify zc_name */ saved_poolname = strdup(zc->zc_name); if (saved_poolname == NULL) { error = SET_ERROR(ENOMEM); goto out; } else { saved_poolname[strcspn(saved_poolname, "/@#")] = '\0'; } if (vec->zvec_func != NULL) { nvlist_t *outnvl; int puterror = 0; spa_t *spa; nvlist_t *lognv = NULL; ASSERT(vec->zvec_legacy_func == NULL); /* * Add the innvl to the lognv before calling the func, * in case the func changes the innvl. */ if (vec->zvec_allow_log) { lognv = fnvlist_alloc(); fnvlist_add_string(lognv, ZPOOL_HIST_IOCTL, vec->zvec_name); if (!nvlist_empty(innvl)) { fnvlist_add_nvlist(lognv, ZPOOL_HIST_INPUT_NVL, innvl); } } VERIFY0(nvlist_alloc(&outnvl, NV_UNIQUE_NAME, KM_PUSHPAGE)); error = vec->zvec_func(zc->zc_name, innvl, outnvl); if (error == 0 && vec->zvec_allow_log && spa_open(zc->zc_name, &spa, FTAG) == 0) { if (!nvlist_empty(outnvl)) { fnvlist_add_nvlist(lognv, ZPOOL_HIST_OUTPUT_NVL, outnvl); } (void) spa_history_log_nvl(spa, lognv); spa_close(spa, FTAG); } fnvlist_free(lognv); if (!nvlist_empty(outnvl) || zc->zc_nvlist_dst_size != 0) { int smusherror = 0; if (vec->zvec_smush_outnvlist) { smusherror = nvlist_smush(outnvl, zc->zc_nvlist_dst_size); } if (smusherror == 0) puterror = put_nvlist(zc, outnvl); } if (puterror != 0) error = puterror; nvlist_free(outnvl); } else { error = vec->zvec_legacy_func(zc); } out: nvlist_free(innvl); rc = ddi_copyout(zc, (void *)arg, sizeof (zfs_cmd_t), flag); if (error == 0 && rc != 0) error = SET_ERROR(EFAULT); if (error == 0 && vec->zvec_allow_log) { char *s = tsd_get(zfs_allow_log_key); if (s != NULL) strfree(s); (void) tsd_set(zfs_allow_log_key, saved_poolname); } else { if (saved_poolname != NULL) strfree(saved_poolname); } kmem_free(zc, sizeof (zfs_cmd_t)); return (-error); } #ifdef CONFIG_COMPAT static long zfsdev_compat_ioctl(struct file *filp, unsigned cmd, unsigned long arg) { return (zfsdev_ioctl(filp, cmd, arg)); } #else #define zfsdev_compat_ioctl NULL #endif static const struct file_operations zfsdev_fops = { .open = zfsdev_open, .release = zfsdev_release, .unlocked_ioctl = zfsdev_ioctl, .compat_ioctl = zfsdev_compat_ioctl, .owner = THIS_MODULE, }; static struct miscdevice zfs_misc = { .minor = MISC_DYNAMIC_MINOR, .name = ZFS_DRIVER, .fops = &zfsdev_fops, }; static int zfs_attach(void) { int error; mutex_init(&zfsdev_state_lock, NULL, MUTEX_DEFAULT, NULL); zfsdev_state_list = kmem_zalloc(sizeof (zfsdev_state_t), KM_SLEEP); zfsdev_state_list->zs_minor = -1; error = misc_register(&zfs_misc); if (error != 0) { printk(KERN_INFO "ZFS: misc_register() failed %d\n", error); return (error); } return (0); } static void zfs_detach(void) { int error; zfsdev_state_t *zs, *zsprev = NULL; error = misc_deregister(&zfs_misc); if (error != 0) printk(KERN_INFO "ZFS: misc_deregister() failed %d\n", error); mutex_destroy(&zfsdev_state_lock); for (zs = zfsdev_state_list; zs != NULL; zs = zs->zs_next) { if (zsprev) kmem_free(zsprev, sizeof (zfsdev_state_t)); zsprev = zs; } if (zsprev) kmem_free(zsprev, sizeof (zfsdev_state_t)); } static void zfs_allow_log_destroy(void *arg) { char *poolname = arg; strfree(poolname); } #ifdef DEBUG #define ZFS_DEBUG_STR " (DEBUG mode)" #else #define ZFS_DEBUG_STR "" #endif int _init(void) { int error; spa_init(FREAD | FWRITE); zfs_init(); if ((error = zvol_init()) != 0) goto out1; zfs_ioctl_init(); if ((error = zfs_attach()) != 0) goto out2; tsd_create(&zfs_fsyncer_key, NULL); tsd_create(&rrw_tsd_key, rrw_tsd_destroy); tsd_create(&zfs_allow_log_key, zfs_allow_log_destroy); printk(KERN_NOTICE "ZFS: Loaded module v%s-%s%s, " "ZFS pool version %s, ZFS filesystem version %s\n", ZFS_META_VERSION, ZFS_META_RELEASE, ZFS_DEBUG_STR, SPA_VERSION_STRING, ZPL_VERSION_STRING); #ifndef CONFIG_FS_POSIX_ACL printk(KERN_NOTICE "ZFS: Posix ACLs disabled by kernel\n"); #endif /* CONFIG_FS_POSIX_ACL */ return (0); out2: (void) zvol_fini(); out1: zfs_fini(); spa_fini(); printk(KERN_NOTICE "ZFS: Failed to Load ZFS Filesystem v%s-%s%s" ", rc = %d\n", ZFS_META_VERSION, ZFS_META_RELEASE, ZFS_DEBUG_STR, error); return (error); } int _fini(void) { zfs_detach(); zvol_fini(); zfs_fini(); spa_fini(); tsd_destroy(&zfs_fsyncer_key); tsd_destroy(&rrw_tsd_key); tsd_destroy(&zfs_allow_log_key); printk(KERN_NOTICE "ZFS: Unloaded module v%s-%s%s\n", ZFS_META_VERSION, ZFS_META_RELEASE, ZFS_DEBUG_STR); return (0); } #ifdef HAVE_SPL spl_module_init(_init); spl_module_exit(_fini); MODULE_DESCRIPTION("ZFS"); MODULE_AUTHOR(ZFS_META_AUTHOR); MODULE_LICENSE(ZFS_META_LICENSE); MODULE_VERSION(ZFS_META_VERSION "-" ZFS_META_RELEASE); #endif /* HAVE_SPL */ diff --git a/module/zfs/zfs_vfsops.c b/module/zfs/zfs_vfsops.c index d67f11eca08b..d29fc6560a8f 100644 --- a/module/zfs/zfs_vfsops.c +++ b/module/zfs/zfs_vfsops.c @@ -1,1704 +1,1704 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2013 by Delphix. All rights reserved. */ /* Portions Copyright 2010 Robert Milkowski */ #include #include #include #include #include #include #include #include #include #include #include #include #include "fs/fs_subr.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "zfs_comutil.h" /*ARGSUSED*/ int zfs_sync(struct super_block *sb, int wait, cred_t *cr) { zfs_sb_t *zsb = sb->s_fs_info; /* * Data integrity is job one. We don't want a compromised kernel * writing to the storage pool, so we never sync during panic. */ if (unlikely(oops_in_progress)) return (0); /* * Semantically, the only requirement is that the sync be initiated. * The DMU syncs out txgs frequently, so there's nothing to do. */ if (!wait) return (0); if (zsb != NULL) { /* * Sync a specific filesystem. */ dsl_pool_t *dp; ZFS_ENTER(zsb); dp = dmu_objset_pool(zsb->z_os); /* * If the system is shutting down, then skip any * filesystems which may exist on a suspended pool. */ if (spa_suspended(dp->dp_spa)) { ZFS_EXIT(zsb); return (0); } if (zsb->z_log != NULL) zil_commit(zsb->z_log, 0); ZFS_EXIT(zsb); } else { /* * Sync all ZFS filesystems. This is what happens when you * run sync(1M). Unlike other filesystems, ZFS honors the * request by waiting for all pools to commit all dirty data. */ spa_sync_allpools(); } return (0); } EXPORT_SYMBOL(zfs_sync); boolean_t zfs_is_readonly(zfs_sb_t *zsb) { return (!!(zsb->z_sb->s_flags & MS_RDONLY)); } EXPORT_SYMBOL(zfs_is_readonly); static void atime_changed_cb(void *arg, uint64_t newval) { ((zfs_sb_t *)arg)->z_atime = newval; } static void relatime_changed_cb(void *arg, uint64_t newval) { ((zfs_sb_t *)arg)->z_relatime = newval; } static void xattr_changed_cb(void *arg, uint64_t newval) { zfs_sb_t *zsb = arg; if (newval == ZFS_XATTR_OFF) { zsb->z_flags &= ~ZSB_XATTR; } else { zsb->z_flags |= ZSB_XATTR; if (newval == ZFS_XATTR_SA) zsb->z_xattr_sa = B_TRUE; else zsb->z_xattr_sa = B_FALSE; } } static void acltype_changed_cb(void *arg, uint64_t newval) { zfs_sb_t *zsb = arg; switch (newval) { case ZFS_ACLTYPE_OFF: zsb->z_acl_type = ZFS_ACLTYPE_OFF; zsb->z_sb->s_flags &= ~MS_POSIXACL; break; case ZFS_ACLTYPE_POSIXACL: #ifdef CONFIG_FS_POSIX_ACL zsb->z_acl_type = ZFS_ACLTYPE_POSIXACL; zsb->z_sb->s_flags |= MS_POSIXACL; #else zsb->z_acl_type = ZFS_ACLTYPE_OFF; zsb->z_sb->s_flags &= ~MS_POSIXACL; #endif /* CONFIG_FS_POSIX_ACL */ break; default: break; } } static void blksz_changed_cb(void *arg, uint64_t newval) { zfs_sb_t *zsb = arg; if (newval < SPA_MINBLOCKSIZE || newval > SPA_MAXBLOCKSIZE || !ISP2(newval)) newval = SPA_MAXBLOCKSIZE; zsb->z_max_blksz = newval; } static void readonly_changed_cb(void *arg, uint64_t newval) { zfs_sb_t *zsb = arg; struct super_block *sb = zsb->z_sb; if (sb == NULL) return; if (newval) sb->s_flags |= MS_RDONLY; else sb->s_flags &= ~MS_RDONLY; } static void devices_changed_cb(void *arg, uint64_t newval) { } static void setuid_changed_cb(void *arg, uint64_t newval) { } static void exec_changed_cb(void *arg, uint64_t newval) { } static void nbmand_changed_cb(void *arg, uint64_t newval) { zfs_sb_t *zsb = arg; struct super_block *sb = zsb->z_sb; if (sb == NULL) return; if (newval == TRUE) sb->s_flags |= MS_MANDLOCK; else sb->s_flags &= ~MS_MANDLOCK; } static void snapdir_changed_cb(void *arg, uint64_t newval) { ((zfs_sb_t *)arg)->z_show_ctldir = newval; } static void vscan_changed_cb(void *arg, uint64_t newval) { ((zfs_sb_t *)arg)->z_vscan = newval; } static void acl_inherit_changed_cb(void *arg, uint64_t newval) { ((zfs_sb_t *)arg)->z_acl_inherit = newval; } int zfs_register_callbacks(zfs_sb_t *zsb) { struct dsl_dataset *ds = NULL; objset_t *os = zsb->z_os; boolean_t do_readonly = B_FALSE; int error = 0; if (zfs_is_readonly(zsb) || !spa_writeable(dmu_objset_spa(os))) do_readonly = B_TRUE; /* * Register property callbacks. * * It would probably be fine to just check for i/o error from * the first prop_register(), but I guess I like to go * overboard... */ ds = dmu_objset_ds(os); dsl_pool_config_enter(dmu_objset_pool(os), FTAG); error = dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_ATIME), atime_changed_cb, zsb); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_RELATIME), relatime_changed_cb, zsb); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_XATTR), xattr_changed_cb, zsb); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_RECORDSIZE), blksz_changed_cb, zsb); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_READONLY), readonly_changed_cb, zsb); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_DEVICES), devices_changed_cb, zsb); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_SETUID), setuid_changed_cb, zsb); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_EXEC), exec_changed_cb, zsb); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_SNAPDIR), snapdir_changed_cb, zsb); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_ACLTYPE), acltype_changed_cb, zsb); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_ACLINHERIT), acl_inherit_changed_cb, zsb); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_VSCAN), vscan_changed_cb, zsb); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_NBMAND), nbmand_changed_cb, zsb); dsl_pool_config_exit(dmu_objset_pool(os), FTAG); if (error) goto unregister; if (do_readonly) readonly_changed_cb(zsb, B_TRUE); return (0); unregister: /* * We may attempt to unregister some callbacks that are not * registered, but this is OK; it will simply return ENOMSG, * which we will ignore. */ (void) dsl_prop_unregister(ds, zfs_prop_to_name(ZFS_PROP_ATIME), atime_changed_cb, zsb); (void) dsl_prop_unregister(ds, zfs_prop_to_name(ZFS_PROP_RELATIME), relatime_changed_cb, zsb); (void) dsl_prop_unregister(ds, zfs_prop_to_name(ZFS_PROP_XATTR), xattr_changed_cb, zsb); (void) dsl_prop_unregister(ds, zfs_prop_to_name(ZFS_PROP_RECORDSIZE), blksz_changed_cb, zsb); (void) dsl_prop_unregister(ds, zfs_prop_to_name(ZFS_PROP_READONLY), readonly_changed_cb, zsb); (void) dsl_prop_unregister(ds, zfs_prop_to_name(ZFS_PROP_DEVICES), devices_changed_cb, zsb); (void) dsl_prop_unregister(ds, zfs_prop_to_name(ZFS_PROP_SETUID), setuid_changed_cb, zsb); (void) dsl_prop_unregister(ds, zfs_prop_to_name(ZFS_PROP_EXEC), exec_changed_cb, zsb); (void) dsl_prop_unregister(ds, zfs_prop_to_name(ZFS_PROP_SNAPDIR), snapdir_changed_cb, zsb); (void) dsl_prop_unregister(ds, zfs_prop_to_name(ZFS_PROP_ACLTYPE), acltype_changed_cb, zsb); (void) dsl_prop_unregister(ds, zfs_prop_to_name(ZFS_PROP_ACLINHERIT), acl_inherit_changed_cb, zsb); (void) dsl_prop_unregister(ds, zfs_prop_to_name(ZFS_PROP_VSCAN), vscan_changed_cb, zsb); (void) dsl_prop_unregister(ds, zfs_prop_to_name(ZFS_PROP_NBMAND), nbmand_changed_cb, zsb); return (error); } EXPORT_SYMBOL(zfs_register_callbacks); static int zfs_space_delta_cb(dmu_object_type_t bonustype, void *data, uint64_t *userp, uint64_t *groupp) { /* * Is it a valid type of object to track? */ if (bonustype != DMU_OT_ZNODE && bonustype != DMU_OT_SA) return (SET_ERROR(ENOENT)); /* * If we have a NULL data pointer * then assume the id's aren't changing and * return EEXIST to the dmu to let it know to * use the same ids */ if (data == NULL) return (SET_ERROR(EEXIST)); if (bonustype == DMU_OT_ZNODE) { znode_phys_t *znp = data; *userp = znp->zp_uid; *groupp = znp->zp_gid; } else { int hdrsize; sa_hdr_phys_t *sap = data; sa_hdr_phys_t sa = *sap; boolean_t swap = B_FALSE; ASSERT(bonustype == DMU_OT_SA); if (sa.sa_magic == 0) { /* * This should only happen for newly created * files that haven't had the znode data filled * in yet. */ *userp = 0; *groupp = 0; return (0); } if (sa.sa_magic == BSWAP_32(SA_MAGIC)) { sa.sa_magic = SA_MAGIC; sa.sa_layout_info = BSWAP_16(sa.sa_layout_info); swap = B_TRUE; } else { VERIFY3U(sa.sa_magic, ==, SA_MAGIC); } hdrsize = sa_hdrsize(&sa); VERIFY3U(hdrsize, >=, sizeof (sa_hdr_phys_t)); *userp = *((uint64_t *)((uintptr_t)data + hdrsize + SA_UID_OFFSET)); *groupp = *((uint64_t *)((uintptr_t)data + hdrsize + SA_GID_OFFSET)); if (swap) { *userp = BSWAP_64(*userp); *groupp = BSWAP_64(*groupp); } } return (0); } static void fuidstr_to_sid(zfs_sb_t *zsb, const char *fuidstr, char *domainbuf, int buflen, uid_t *ridp) { uint64_t fuid; const char *domain; fuid = strtonum(fuidstr, NULL); domain = zfs_fuid_find_by_idx(zsb, FUID_INDEX(fuid)); if (domain) (void) strlcpy(domainbuf, domain, buflen); else domainbuf[0] = '\0'; *ridp = FUID_RID(fuid); } static uint64_t zfs_userquota_prop_to_obj(zfs_sb_t *zsb, zfs_userquota_prop_t type) { switch (type) { case ZFS_PROP_USERUSED: return (DMU_USERUSED_OBJECT); case ZFS_PROP_GROUPUSED: return (DMU_GROUPUSED_OBJECT); case ZFS_PROP_USERQUOTA: return (zsb->z_userquota_obj); case ZFS_PROP_GROUPQUOTA: return (zsb->z_groupquota_obj); default: return (SET_ERROR(ENOTSUP)); } return (0); } int zfs_userspace_many(zfs_sb_t *zsb, zfs_userquota_prop_t type, uint64_t *cookiep, void *vbuf, uint64_t *bufsizep) { int error; zap_cursor_t zc; zap_attribute_t za; zfs_useracct_t *buf = vbuf; uint64_t obj; if (!dmu_objset_userspace_present(zsb->z_os)) return (SET_ERROR(ENOTSUP)); obj = zfs_userquota_prop_to_obj(zsb, type); if (obj == 0) { *bufsizep = 0; return (0); } for (zap_cursor_init_serialized(&zc, zsb->z_os, obj, *cookiep); (error = zap_cursor_retrieve(&zc, &za)) == 0; zap_cursor_advance(&zc)) { if ((uintptr_t)buf - (uintptr_t)vbuf + sizeof (zfs_useracct_t) > *bufsizep) break; fuidstr_to_sid(zsb, za.za_name, buf->zu_domain, sizeof (buf->zu_domain), &buf->zu_rid); buf->zu_space = za.za_first_integer; buf++; } if (error == ENOENT) error = 0; ASSERT3U((uintptr_t)buf - (uintptr_t)vbuf, <=, *bufsizep); *bufsizep = (uintptr_t)buf - (uintptr_t)vbuf; *cookiep = zap_cursor_serialize(&zc); zap_cursor_fini(&zc); return (error); } EXPORT_SYMBOL(zfs_userspace_many); /* * buf must be big enough (eg, 32 bytes) */ static int id_to_fuidstr(zfs_sb_t *zsb, const char *domain, uid_t rid, char *buf, boolean_t addok) { uint64_t fuid; int domainid = 0; if (domain && domain[0]) { domainid = zfs_fuid_find_by_domain(zsb, domain, NULL, addok); if (domainid == -1) return (SET_ERROR(ENOENT)); } fuid = FUID_ENCODE(domainid, rid); (void) sprintf(buf, "%llx", (longlong_t)fuid); return (0); } int zfs_userspace_one(zfs_sb_t *zsb, zfs_userquota_prop_t type, const char *domain, uint64_t rid, uint64_t *valp) { char buf[32]; int err; uint64_t obj; *valp = 0; if (!dmu_objset_userspace_present(zsb->z_os)) return (SET_ERROR(ENOTSUP)); obj = zfs_userquota_prop_to_obj(zsb, type); if (obj == 0) return (0); err = id_to_fuidstr(zsb, domain, rid, buf, B_FALSE); if (err) return (err); err = zap_lookup(zsb->z_os, obj, buf, 8, 1, valp); if (err == ENOENT) err = 0; return (err); } EXPORT_SYMBOL(zfs_userspace_one); int zfs_set_userquota(zfs_sb_t *zsb, zfs_userquota_prop_t type, const char *domain, uint64_t rid, uint64_t quota) { char buf[32]; int err; dmu_tx_t *tx; uint64_t *objp; boolean_t fuid_dirtied; if (type != ZFS_PROP_USERQUOTA && type != ZFS_PROP_GROUPQUOTA) return (SET_ERROR(EINVAL)); if (zsb->z_version < ZPL_VERSION_USERSPACE) return (SET_ERROR(ENOTSUP)); objp = (type == ZFS_PROP_USERQUOTA) ? &zsb->z_userquota_obj : &zsb->z_groupquota_obj; err = id_to_fuidstr(zsb, domain, rid, buf, B_TRUE); if (err) return (err); fuid_dirtied = zsb->z_fuid_dirty; tx = dmu_tx_create(zsb->z_os); dmu_tx_hold_zap(tx, *objp ? *objp : DMU_NEW_OBJECT, B_TRUE, NULL); if (*objp == 0) { dmu_tx_hold_zap(tx, MASTER_NODE_OBJ, B_TRUE, zfs_userquota_prop_prefixes[type]); } if (fuid_dirtied) zfs_fuid_txhold(zsb, tx); err = dmu_tx_assign(tx, TXG_WAIT); if (err) { dmu_tx_abort(tx); return (err); } mutex_enter(&zsb->z_lock); if (*objp == 0) { *objp = zap_create(zsb->z_os, DMU_OT_USERGROUP_QUOTA, DMU_OT_NONE, 0, tx); VERIFY(0 == zap_add(zsb->z_os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[type], 8, 1, objp, tx)); } mutex_exit(&zsb->z_lock); if (quota == 0) { err = zap_remove(zsb->z_os, *objp, buf, tx); if (err == ENOENT) err = 0; } else { err = zap_update(zsb->z_os, *objp, buf, 8, 1, "a, tx); } ASSERT(err == 0); if (fuid_dirtied) zfs_fuid_sync(zsb, tx); dmu_tx_commit(tx); return (err); } EXPORT_SYMBOL(zfs_set_userquota); boolean_t zfs_fuid_overquota(zfs_sb_t *zsb, boolean_t isgroup, uint64_t fuid) { char buf[32]; uint64_t used, quota, usedobj, quotaobj; int err; usedobj = isgroup ? DMU_GROUPUSED_OBJECT : DMU_USERUSED_OBJECT; quotaobj = isgroup ? zsb->z_groupquota_obj : zsb->z_userquota_obj; if (quotaobj == 0 || zsb->z_replay) return (B_FALSE); (void) sprintf(buf, "%llx", (longlong_t)fuid); err = zap_lookup(zsb->z_os, quotaobj, buf, 8, 1, "a); if (err != 0) return (B_FALSE); err = zap_lookup(zsb->z_os, usedobj, buf, 8, 1, &used); if (err != 0) return (B_FALSE); return (used >= quota); } EXPORT_SYMBOL(zfs_fuid_overquota); boolean_t zfs_owner_overquota(zfs_sb_t *zsb, znode_t *zp, boolean_t isgroup) { uint64_t fuid; uint64_t quotaobj; quotaobj = isgroup ? zsb->z_groupquota_obj : zsb->z_userquota_obj; fuid = isgroup ? zp->z_gid : zp->z_uid; if (quotaobj == 0 || zsb->z_replay) return (B_FALSE); return (zfs_fuid_overquota(zsb, isgroup, fuid)); } EXPORT_SYMBOL(zfs_owner_overquota); int zfs_sb_create(const char *osname, zfs_sb_t **zsbp) { objset_t *os; zfs_sb_t *zsb; uint64_t zval; int i, error; uint64_t sa_obj; - zsb = kmem_zalloc(sizeof (zfs_sb_t), KM_SLEEP | KM_NODEBUG); + zsb = kmem_zalloc(sizeof (zfs_sb_t), KM_SLEEP); /* * We claim to always be readonly so we can open snapshots; * other ZPL code will prevent us from writing to snapshots. */ error = dmu_objset_own(osname, DMU_OST_ZFS, B_TRUE, zsb, &os); if (error) { kmem_free(zsb, sizeof (zfs_sb_t)); return (error); } /* * Initialize the zfs-specific filesystem structure. * Should probably make this a kmem cache, shuffle fields, * and just bzero up to z_hold_mtx[]. */ zsb->z_sb = NULL; zsb->z_parent = zsb; zsb->z_max_blksz = SPA_MAXBLOCKSIZE; zsb->z_show_ctldir = ZFS_SNAPDIR_VISIBLE; zsb->z_os = os; error = zfs_get_zplprop(os, ZFS_PROP_VERSION, &zsb->z_version); if (error) { goto out; } else if (zsb->z_version > ZPL_VERSION) { error = SET_ERROR(ENOTSUP); goto out; } if ((error = zfs_get_zplprop(os, ZFS_PROP_NORMALIZE, &zval)) != 0) goto out; zsb->z_norm = (int)zval; if ((error = zfs_get_zplprop(os, ZFS_PROP_UTF8ONLY, &zval)) != 0) goto out; zsb->z_utf8 = (zval != 0); if ((error = zfs_get_zplprop(os, ZFS_PROP_CASE, &zval)) != 0) goto out; zsb->z_case = (uint_t)zval; if ((error = zfs_get_zplprop(os, ZFS_PROP_ACLTYPE, &zval)) != 0) goto out; zsb->z_acl_type = (uint_t)zval; /* * Fold case on file systems that are always or sometimes case * insensitive. */ if (zsb->z_case == ZFS_CASE_INSENSITIVE || zsb->z_case == ZFS_CASE_MIXED) zsb->z_norm |= U8_TEXTPREP_TOUPPER; zsb->z_use_fuids = USE_FUIDS(zsb->z_version, zsb->z_os); zsb->z_use_sa = USE_SA(zsb->z_version, zsb->z_os); if (zsb->z_use_sa) { /* should either have both of these objects or none */ error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_SA_ATTRS, 8, 1, &sa_obj); if (error) goto out; error = zfs_get_zplprop(os, ZFS_PROP_XATTR, &zval); if ((error == 0) && (zval == ZFS_XATTR_SA)) zsb->z_xattr_sa = B_TRUE; } else { /* * Pre SA versions file systems should never touch * either the attribute registration or layout objects. */ sa_obj = 0; } error = sa_setup(os, sa_obj, zfs_attr_table, ZPL_END, &zsb->z_attr_table); if (error) goto out; if (zsb->z_version >= ZPL_VERSION_SA) sa_register_update_callback(os, zfs_sa_upgrade); error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_ROOT_OBJ, 8, 1, &zsb->z_root); if (error) goto out; ASSERT(zsb->z_root != 0); error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_UNLINKED_SET, 8, 1, &zsb->z_unlinkedobj); if (error) goto out; error = zap_lookup(os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[ZFS_PROP_USERQUOTA], 8, 1, &zsb->z_userquota_obj); if (error && error != ENOENT) goto out; error = zap_lookup(os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[ZFS_PROP_GROUPQUOTA], 8, 1, &zsb->z_groupquota_obj); if (error && error != ENOENT) goto out; error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_FUID_TABLES, 8, 1, &zsb->z_fuid_obj); if (error && error != ENOENT) goto out; error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_SHARES_DIR, 8, 1, &zsb->z_shares_dir); if (error && error != ENOENT) goto out; mutex_init(&zsb->z_znodes_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&zsb->z_lock, NULL, MUTEX_DEFAULT, NULL); list_create(&zsb->z_all_znodes, sizeof (znode_t), offsetof(znode_t, z_link_node)); rrw_init(&zsb->z_teardown_lock, B_FALSE); rw_init(&zsb->z_teardown_inactive_lock, NULL, RW_DEFAULT, NULL); rw_init(&zsb->z_fuid_lock, NULL, RW_DEFAULT, NULL); for (i = 0; i != ZFS_OBJ_MTX_SZ; i++) mutex_init(&zsb->z_hold_mtx[i], NULL, MUTEX_DEFAULT, NULL); avl_create(&zsb->z_ctldir_snaps, snapentry_compare, sizeof (zfs_snapentry_t), offsetof(zfs_snapentry_t, se_node)); mutex_init(&zsb->z_ctldir_lock, NULL, MUTEX_DEFAULT, NULL); *zsbp = zsb; return (0); out: dmu_objset_disown(os, zsb); *zsbp = NULL; kmem_free(zsb, sizeof (zfs_sb_t)); return (error); } EXPORT_SYMBOL(zfs_sb_create); int zfs_sb_setup(zfs_sb_t *zsb, boolean_t mounting) { int error; error = zfs_register_callbacks(zsb); if (error) return (error); /* * Set the objset user_ptr to track its zsb. */ mutex_enter(&zsb->z_os->os_user_ptr_lock); dmu_objset_set_user(zsb->z_os, zsb); mutex_exit(&zsb->z_os->os_user_ptr_lock); zsb->z_log = zil_open(zsb->z_os, zfs_get_data); /* * If we are not mounting (ie: online recv), then we don't * have to worry about replaying the log as we blocked all * operations out since we closed the ZIL. */ if (mounting) { boolean_t readonly; /* * During replay we remove the read only flag to * allow replays to succeed. */ readonly = zfs_is_readonly(zsb); if (readonly != 0) readonly_changed_cb(zsb, B_FALSE); else zfs_unlinked_drain(zsb); /* * Parse and replay the intent log. * * Because of ziltest, this must be done after * zfs_unlinked_drain(). (Further note: ziltest * doesn't use readonly mounts, where * zfs_unlinked_drain() isn't called.) This is because * ziltest causes spa_sync() to think it's committed, * but actually it is not, so the intent log contains * many txg's worth of changes. * * In particular, if object N is in the unlinked set in * the last txg to actually sync, then it could be * actually freed in a later txg and then reallocated * in a yet later txg. This would write a "create * object N" record to the intent log. Normally, this * would be fine because the spa_sync() would have * written out the fact that object N is free, before * we could write the "create object N" intent log * record. * * But when we are in ziltest mode, we advance the "open * txg" without actually spa_sync()-ing the changes to * disk. So we would see that object N is still * allocated and in the unlinked set, and there is an * intent log record saying to allocate it. */ if (spa_writeable(dmu_objset_spa(zsb->z_os))) { if (zil_replay_disable) { zil_destroy(zsb->z_log, B_FALSE); } else { zsb->z_replay = B_TRUE; zil_replay(zsb->z_os, zsb, zfs_replay_vector); zsb->z_replay = B_FALSE; } } /* restore readonly bit */ if (readonly != 0) readonly_changed_cb(zsb, B_TRUE); } return (0); } EXPORT_SYMBOL(zfs_sb_setup); void zfs_sb_free(zfs_sb_t *zsb) { int i; zfs_fuid_destroy(zsb); mutex_destroy(&zsb->z_znodes_lock); mutex_destroy(&zsb->z_lock); list_destroy(&zsb->z_all_znodes); rrw_destroy(&zsb->z_teardown_lock); rw_destroy(&zsb->z_teardown_inactive_lock); rw_destroy(&zsb->z_fuid_lock); for (i = 0; i != ZFS_OBJ_MTX_SZ; i++) mutex_destroy(&zsb->z_hold_mtx[i]); mutex_destroy(&zsb->z_ctldir_lock); avl_destroy(&zsb->z_ctldir_snaps); kmem_free(zsb, sizeof (zfs_sb_t)); } EXPORT_SYMBOL(zfs_sb_free); static void zfs_set_fuid_feature(zfs_sb_t *zsb) { zsb->z_use_fuids = USE_FUIDS(zsb->z_version, zsb->z_os); zsb->z_use_sa = USE_SA(zsb->z_version, zsb->z_os); } void zfs_unregister_callbacks(zfs_sb_t *zsb) { objset_t *os = zsb->z_os; struct dsl_dataset *ds; /* * Unregister properties. */ if (!dmu_objset_is_snapshot(os)) { ds = dmu_objset_ds(os); VERIFY(dsl_prop_unregister(ds, "atime", atime_changed_cb, zsb) == 0); VERIFY(dsl_prop_unregister(ds, "relatime", relatime_changed_cb, zsb) == 0); VERIFY(dsl_prop_unregister(ds, "xattr", xattr_changed_cb, zsb) == 0); VERIFY(dsl_prop_unregister(ds, "recordsize", blksz_changed_cb, zsb) == 0); VERIFY(dsl_prop_unregister(ds, "readonly", readonly_changed_cb, zsb) == 0); VERIFY(dsl_prop_unregister(ds, "devices", devices_changed_cb, zsb) == 0); VERIFY(dsl_prop_unregister(ds, "setuid", setuid_changed_cb, zsb) == 0); VERIFY(dsl_prop_unregister(ds, "exec", exec_changed_cb, zsb) == 0); VERIFY(dsl_prop_unregister(ds, "snapdir", snapdir_changed_cb, zsb) == 0); VERIFY(dsl_prop_unregister(ds, "acltype", acltype_changed_cb, zsb) == 0); VERIFY(dsl_prop_unregister(ds, "aclinherit", acl_inherit_changed_cb, zsb) == 0); VERIFY(dsl_prop_unregister(ds, "vscan", vscan_changed_cb, zsb) == 0); VERIFY(dsl_prop_unregister(ds, "nbmand", nbmand_changed_cb, zsb) == 0); } } EXPORT_SYMBOL(zfs_unregister_callbacks); #ifdef HAVE_MLSLABEL /* * Check that the hex label string is appropriate for the dataset being * mounted into the global_zone proper. * * Return an error if the hex label string is not default or * admin_low/admin_high. For admin_low labels, the corresponding * dataset must be readonly. */ int zfs_check_global_label(const char *dsname, const char *hexsl) { if (strcasecmp(hexsl, ZFS_MLSLABEL_DEFAULT) == 0) return (0); if (strcasecmp(hexsl, ADMIN_HIGH) == 0) return (0); if (strcasecmp(hexsl, ADMIN_LOW) == 0) { /* must be readonly */ uint64_t rdonly; if (dsl_prop_get_integer(dsname, zfs_prop_to_name(ZFS_PROP_READONLY), &rdonly, NULL)) return (SET_ERROR(EACCES)); return (rdonly ? 0 : EACCES); } return (SET_ERROR(EACCES)); } EXPORT_SYMBOL(zfs_check_global_label); #endif /* HAVE_MLSLABEL */ int zfs_statvfs(struct dentry *dentry, struct kstatfs *statp) { zfs_sb_t *zsb = dentry->d_sb->s_fs_info; uint64_t refdbytes, availbytes, usedobjs, availobjs; uint64_t fsid; uint32_t bshift; ZFS_ENTER(zsb); dmu_objset_space(zsb->z_os, &refdbytes, &availbytes, &usedobjs, &availobjs); fsid = dmu_objset_fsid_guid(zsb->z_os); /* * The underlying storage pool actually uses multiple block * size. Under Solaris frsize (fragment size) is reported as * the smallest block size we support, and bsize (block size) * as the filesystem's maximum block size. Unfortunately, * under Linux the fragment size and block size are often used * interchangeably. Thus we are forced to report both of them * as the filesystem's maximum block size. */ statp->f_frsize = zsb->z_max_blksz; statp->f_bsize = zsb->z_max_blksz; bshift = fls(statp->f_bsize) - 1; /* * The following report "total" blocks of various kinds in * the file system, but reported in terms of f_bsize - the * "preferred" size. */ statp->f_blocks = (refdbytes + availbytes) >> bshift; statp->f_bfree = availbytes >> bshift; statp->f_bavail = statp->f_bfree; /* no root reservation */ /* * statvfs() should really be called statufs(), because it assumes * static metadata. ZFS doesn't preallocate files, so the best * we can do is report the max that could possibly fit in f_files, * and that minus the number actually used in f_ffree. * For f_ffree, report the smaller of the number of object available * and the number of blocks (each object will take at least a block). */ statp->f_ffree = MIN(availobjs, availbytes >> DNODE_SHIFT); statp->f_files = statp->f_ffree + usedobjs; statp->f_fsid.val[0] = (uint32_t)fsid; statp->f_fsid.val[1] = (uint32_t)(fsid >> 32); statp->f_type = ZFS_SUPER_MAGIC; statp->f_namelen = ZFS_MAXNAMELEN; /* * We have all of 40 characters to stuff a string here. * Is there anything useful we could/should provide? */ bzero(statp->f_spare, sizeof (statp->f_spare)); ZFS_EXIT(zsb); return (0); } EXPORT_SYMBOL(zfs_statvfs); int zfs_root(zfs_sb_t *zsb, struct inode **ipp) { znode_t *rootzp; int error; ZFS_ENTER(zsb); error = zfs_zget(zsb, zsb->z_root, &rootzp); if (error == 0) *ipp = ZTOI(rootzp); ZFS_EXIT(zsb); return (error); } EXPORT_SYMBOL(zfs_root); #ifdef HAVE_SHRINK int zfs_sb_prune(struct super_block *sb, unsigned long nr_to_scan, int *objects) { zfs_sb_t *zsb = sb->s_fs_info; struct shrinker *shrinker = &sb->s_shrink; struct shrink_control sc = { .nr_to_scan = nr_to_scan, .gfp_mask = GFP_KERNEL, }; ZFS_ENTER(zsb); *objects = (*shrinker->shrink)(shrinker, &sc); ZFS_EXIT(zsb); return (0); } EXPORT_SYMBOL(zfs_sb_prune); #endif /* HAVE_SHRINK */ /* * Teardown the zfs_sb_t. * * Note, if 'unmounting' if FALSE, we return with the 'z_teardown_lock' * and 'z_teardown_inactive_lock' held. */ int zfs_sb_teardown(zfs_sb_t *zsb, boolean_t unmounting) { znode_t *zp; /* * If someone has not already unmounted this file system, * drain the iput_taskq to ensure all active references to the * zfs_sb_t have been handled only then can it be safely destroyed. */ if (zsb->z_os) taskq_wait(dsl_pool_iput_taskq(dmu_objset_pool(zsb->z_os))); rrw_enter(&zsb->z_teardown_lock, RW_WRITER, FTAG); if (!unmounting) { /* * We purge the parent filesystem's super block as the * parent filesystem and all of its snapshots have their * inode's super block set to the parent's filesystem's * super block. Note, 'z_parent' is self referential * for non-snapshots. */ shrink_dcache_sb(zsb->z_parent->z_sb); } /* * Close the zil. NB: Can't close the zil while zfs_inactive * threads are blocked as zil_close can call zfs_inactive. */ if (zsb->z_log) { zil_close(zsb->z_log); zsb->z_log = NULL; } rw_enter(&zsb->z_teardown_inactive_lock, RW_WRITER); /* * If we are not unmounting (ie: online recv) and someone already * unmounted this file system while we were doing the switcheroo, * or a reopen of z_os failed then just bail out now. */ if (!unmounting && (zsb->z_unmounted || zsb->z_os == NULL)) { rw_exit(&zsb->z_teardown_inactive_lock); rrw_exit(&zsb->z_teardown_lock, FTAG); return (SET_ERROR(EIO)); } /* * At this point there are no VFS ops active, and any new VFS ops * will fail with EIO since we have z_teardown_lock for writer (only * relevant for forced unmount). * * Release all holds on dbufs. */ mutex_enter(&zsb->z_znodes_lock); for (zp = list_head(&zsb->z_all_znodes); zp != NULL; zp = list_next(&zsb->z_all_znodes, zp)) { if (zp->z_sa_hdl) zfs_znode_dmu_fini(zp); } mutex_exit(&zsb->z_znodes_lock); /* * If we are unmounting, set the unmounted flag and let new VFS ops * unblock. zfs_inactive will have the unmounted behavior, and all * other VFS ops will fail with EIO. */ if (unmounting) { zsb->z_unmounted = B_TRUE; rrw_exit(&zsb->z_teardown_lock, FTAG); rw_exit(&zsb->z_teardown_inactive_lock); } /* * z_os will be NULL if there was an error in attempting to reopen * zsb, so just return as the properties had already been * * unregistered and cached data had been evicted before. */ if (zsb->z_os == NULL) return (0); /* * Unregister properties. */ zfs_unregister_callbacks(zsb); /* * Evict cached data */ if (dsl_dataset_is_dirty(dmu_objset_ds(zsb->z_os)) && !zfs_is_readonly(zsb)) txg_wait_synced(dmu_objset_pool(zsb->z_os), 0); dmu_objset_evict_dbufs(zsb->z_os); return (0); } EXPORT_SYMBOL(zfs_sb_teardown); #if defined(HAVE_BDI) && !defined(HAVE_BDI_SETUP_AND_REGISTER) atomic_long_t zfs_bdi_seq = ATOMIC_LONG_INIT(0); #endif /* HAVE_BDI && !HAVE_BDI_SETUP_AND_REGISTER */ int zfs_domount(struct super_block *sb, void *data, int silent) { zpl_mount_data_t *zmd = data; const char *osname = zmd->z_osname; zfs_sb_t *zsb; struct inode *root_inode; uint64_t recordsize; int error; error = zfs_sb_create(osname, &zsb); if (error) return (error); if ((error = dsl_prop_get_integer(osname, "recordsize", &recordsize, NULL))) goto out; zsb->z_sb = sb; sb->s_fs_info = zsb; sb->s_magic = ZFS_SUPER_MAGIC; sb->s_maxbytes = MAX_LFS_FILESIZE; sb->s_time_gran = 1; sb->s_blocksize = recordsize; sb->s_blocksize_bits = ilog2(recordsize); #ifdef HAVE_BDI /* * 2.6.32 API change, * Added backing_device_info (BDI) per super block interfaces. A BDI * must be configured when using a non-device backed filesystem for * proper writeback. This is not required for older pdflush kernels. * * NOTE: Linux read-ahead is disabled in favor of zfs read-ahead. */ zsb->z_bdi.ra_pages = 0; sb->s_bdi = &zsb->z_bdi; error = -bdi_setup_and_register(&zsb->z_bdi, "zfs", BDI_CAP_MAP_COPY); if (error) goto out; #endif /* HAVE_BDI */ /* Set callback operations for the file system. */ sb->s_op = &zpl_super_operations; sb->s_xattr = zpl_xattr_handlers; sb->s_export_op = &zpl_export_operations; #ifdef HAVE_S_D_OP sb->s_d_op = &zpl_dentry_operations; #endif /* HAVE_S_D_OP */ /* Set features for file system. */ zfs_set_fuid_feature(zsb); if (dmu_objset_is_snapshot(zsb->z_os)) { uint64_t pval; atime_changed_cb(zsb, B_FALSE); readonly_changed_cb(zsb, B_TRUE); if ((error = dsl_prop_get_integer(osname, "xattr", &pval, NULL))) goto out; xattr_changed_cb(zsb, pval); if ((error = dsl_prop_get_integer(osname, "acltype", &pval, NULL))) goto out; acltype_changed_cb(zsb, pval); zsb->z_issnap = B_TRUE; zsb->z_os->os_sync = ZFS_SYNC_DISABLED; mutex_enter(&zsb->z_os->os_user_ptr_lock); dmu_objset_set_user(zsb->z_os, zsb); mutex_exit(&zsb->z_os->os_user_ptr_lock); } else { error = zfs_sb_setup(zsb, B_TRUE); } /* Allocate a root inode for the filesystem. */ error = zfs_root(zsb, &root_inode); if (error) { (void) zfs_umount(sb); goto out; } /* Allocate a root dentry for the filesystem */ sb->s_root = d_make_root(root_inode); if (sb->s_root == NULL) { (void) zfs_umount(sb); error = SET_ERROR(ENOMEM); goto out; } if (!zsb->z_issnap) zfsctl_create(zsb); out: if (error) { dmu_objset_disown(zsb->z_os, zsb); zfs_sb_free(zsb); } return (error); } EXPORT_SYMBOL(zfs_domount); /* * Called when an unmount is requested and certain sanity checks have * already passed. At this point no dentries or inodes have been reclaimed * from their respective caches. We drop the extra reference on the .zfs * control directory to allow everything to be reclaimed. All snapshots * must already have been unmounted to reach this point. */ void zfs_preumount(struct super_block *sb) { zfs_sb_t *zsb = sb->s_fs_info; if (zsb != NULL && zsb->z_ctldir != NULL) zfsctl_destroy(zsb); } EXPORT_SYMBOL(zfs_preumount); /* * Called once all other unmount released tear down has occurred. * It is our responsibility to release any remaining infrastructure. */ /*ARGSUSED*/ int zfs_umount(struct super_block *sb) { zfs_sb_t *zsb = sb->s_fs_info; objset_t *os; VERIFY(zfs_sb_teardown(zsb, B_TRUE) == 0); os = zsb->z_os; #ifdef HAVE_BDI bdi_destroy(sb->s_bdi); #endif /* HAVE_BDI */ /* * z_os will be NULL if there was an error in * attempting to reopen zsb. */ if (os != NULL) { /* * Unset the objset user_ptr. */ mutex_enter(&os->os_user_ptr_lock); dmu_objset_set_user(os, NULL); mutex_exit(&os->os_user_ptr_lock); /* * Finally release the objset */ dmu_objset_disown(os, zsb); } zfs_sb_free(zsb); return (0); } EXPORT_SYMBOL(zfs_umount); int zfs_remount(struct super_block *sb, int *flags, char *data) { /* * All namespace flags (MNT_*) and super block flags (MS_*) will * be handled by the Linux VFS. Only handle custom options here. */ return (0); } EXPORT_SYMBOL(zfs_remount); int zfs_vget(struct super_block *sb, struct inode **ipp, fid_t *fidp) { zfs_sb_t *zsb = sb->s_fs_info; znode_t *zp; uint64_t object = 0; uint64_t fid_gen = 0; uint64_t gen_mask; uint64_t zp_gen; int i, err; *ipp = NULL; ZFS_ENTER(zsb); if (fidp->fid_len == LONG_FID_LEN) { zfid_long_t *zlfid = (zfid_long_t *)fidp; uint64_t objsetid = 0; uint64_t setgen = 0; for (i = 0; i < sizeof (zlfid->zf_setid); i++) objsetid |= ((uint64_t)zlfid->zf_setid[i]) << (8 * i); for (i = 0; i < sizeof (zlfid->zf_setgen); i++) setgen |= ((uint64_t)zlfid->zf_setgen[i]) << (8 * i); ZFS_EXIT(zsb); err = zfsctl_lookup_objset(sb, objsetid, &zsb); if (err) return (SET_ERROR(EINVAL)); ZFS_ENTER(zsb); } if (fidp->fid_len == SHORT_FID_LEN || fidp->fid_len == LONG_FID_LEN) { zfid_short_t *zfid = (zfid_short_t *)fidp; for (i = 0; i < sizeof (zfid->zf_object); i++) object |= ((uint64_t)zfid->zf_object[i]) << (8 * i); for (i = 0; i < sizeof (zfid->zf_gen); i++) fid_gen |= ((uint64_t)zfid->zf_gen[i]) << (8 * i); } else { ZFS_EXIT(zsb); return (SET_ERROR(EINVAL)); } /* A zero fid_gen means we are in the .zfs control directories */ if (fid_gen == 0 && (object == ZFSCTL_INO_ROOT || object == ZFSCTL_INO_SNAPDIR)) { *ipp = zsb->z_ctldir; ASSERT(*ipp != NULL); if (object == ZFSCTL_INO_SNAPDIR) { VERIFY(zfsctl_root_lookup(*ipp, "snapshot", ipp, 0, kcred, NULL, NULL) == 0); } else { igrab(*ipp); } ZFS_EXIT(zsb); return (0); } gen_mask = -1ULL >> (64 - 8 * i); dprintf("getting %llu [%llu mask %llx]\n", object, fid_gen, gen_mask); if ((err = zfs_zget(zsb, object, &zp))) { ZFS_EXIT(zsb); return (err); } (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(zsb), &zp_gen, sizeof (uint64_t)); zp_gen = zp_gen & gen_mask; if (zp_gen == 0) zp_gen = 1; if (zp->z_unlinked || zp_gen != fid_gen) { dprintf("znode gen (%llu) != fid gen (%llu)\n", zp_gen, fid_gen); iput(ZTOI(zp)); ZFS_EXIT(zsb); return (SET_ERROR(EINVAL)); } *ipp = ZTOI(zp); if (*ipp) zfs_inode_update(ITOZ(*ipp)); ZFS_EXIT(zsb); return (0); } EXPORT_SYMBOL(zfs_vget); /* * Block out VFS ops and close zfs_sb_t * * Note, if successful, then we return with the 'z_teardown_lock' and * 'z_teardown_inactive_lock' write held. We leave ownership of the underlying * dataset and objset intact so that they can be atomically handed off during * a subsequent rollback or recv operation and the resume thereafter. */ int zfs_suspend_fs(zfs_sb_t *zsb) { int error; if ((error = zfs_sb_teardown(zsb, B_FALSE)) != 0) return (error); return (0); } EXPORT_SYMBOL(zfs_suspend_fs); /* * Reopen zfs_sb_t and release VFS ops. */ int zfs_resume_fs(zfs_sb_t *zsb, const char *osname) { int err, err2; znode_t *zp; uint64_t sa_obj = 0; ASSERT(RRW_WRITE_HELD(&zsb->z_teardown_lock)); ASSERT(RW_WRITE_HELD(&zsb->z_teardown_inactive_lock)); /* * We already own this, so just hold and rele it to update the * objset_t, as the one we had before may have been evicted. */ VERIFY0(dmu_objset_hold(osname, zsb, &zsb->z_os)); VERIFY3P(zsb->z_os->os_dsl_dataset->ds_owner, ==, zsb); VERIFY(dsl_dataset_long_held(zsb->z_os->os_dsl_dataset)); dmu_objset_rele(zsb->z_os, zsb); /* * Make sure version hasn't changed */ err = zfs_get_zplprop(zsb->z_os, ZFS_PROP_VERSION, &zsb->z_version); if (err) goto bail; err = zap_lookup(zsb->z_os, MASTER_NODE_OBJ, ZFS_SA_ATTRS, 8, 1, &sa_obj); if (err && zsb->z_version >= ZPL_VERSION_SA) goto bail; if ((err = sa_setup(zsb->z_os, sa_obj, zfs_attr_table, ZPL_END, &zsb->z_attr_table)) != 0) goto bail; if (zsb->z_version >= ZPL_VERSION_SA) sa_register_update_callback(zsb->z_os, zfs_sa_upgrade); VERIFY(zfs_sb_setup(zsb, B_FALSE) == 0); zfs_set_fuid_feature(zsb); zsb->z_rollback_time = jiffies; /* * Attempt to re-establish all the active inodes with their * dbufs. If a zfs_rezget() fails, then we unhash the inode * and mark it stale. This prevents a collision if a new * inode/object is created which must use the same inode * number. The stale inode will be be released when the * VFS prunes the dentry holding the remaining references * on the stale inode. */ mutex_enter(&zsb->z_znodes_lock); for (zp = list_head(&zsb->z_all_znodes); zp; zp = list_next(&zsb->z_all_znodes, zp)) { err2 = zfs_rezget(zp); if (err2) { remove_inode_hash(ZTOI(zp)); zp->z_is_stale = B_TRUE; } } mutex_exit(&zsb->z_znodes_lock); bail: /* release the VFS ops */ rw_exit(&zsb->z_teardown_inactive_lock); rrw_exit(&zsb->z_teardown_lock, FTAG); if (err) { /* * Since we couldn't setup the sa framework, try to force * unmount this file system. */ if (zsb->z_os) (void) zfs_umount(zsb->z_sb); } return (err); } EXPORT_SYMBOL(zfs_resume_fs); int zfs_set_version(zfs_sb_t *zsb, uint64_t newvers) { int error; objset_t *os = zsb->z_os; dmu_tx_t *tx; if (newvers < ZPL_VERSION_INITIAL || newvers > ZPL_VERSION) return (SET_ERROR(EINVAL)); if (newvers < zsb->z_version) return (SET_ERROR(EINVAL)); if (zfs_spa_version_map(newvers) > spa_version(dmu_objset_spa(zsb->z_os))) return (SET_ERROR(ENOTSUP)); tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, MASTER_NODE_OBJ, B_FALSE, ZPL_VERSION_STR); if (newvers >= ZPL_VERSION_SA && !zsb->z_use_sa) { dmu_tx_hold_zap(tx, MASTER_NODE_OBJ, B_TRUE, ZFS_SA_ATTRS); dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL); } error = dmu_tx_assign(tx, TXG_WAIT); if (error) { dmu_tx_abort(tx); return (error); } error = zap_update(os, MASTER_NODE_OBJ, ZPL_VERSION_STR, 8, 1, &newvers, tx); if (error) { dmu_tx_commit(tx); return (error); } if (newvers >= ZPL_VERSION_SA && !zsb->z_use_sa) { uint64_t sa_obj; ASSERT3U(spa_version(dmu_objset_spa(zsb->z_os)), >=, SPA_VERSION_SA); sa_obj = zap_create(os, DMU_OT_SA_MASTER_NODE, DMU_OT_NONE, 0, tx); error = zap_add(os, MASTER_NODE_OBJ, ZFS_SA_ATTRS, 8, 1, &sa_obj, tx); ASSERT0(error); VERIFY(0 == sa_set_sa_object(os, sa_obj)); sa_register_update_callback(os, zfs_sa_upgrade); } spa_history_log_internal_ds(dmu_objset_ds(os), "upgrade", tx, "from %llu to %llu", zsb->z_version, newvers); dmu_tx_commit(tx); zsb->z_version = newvers; zfs_set_fuid_feature(zsb); return (0); } EXPORT_SYMBOL(zfs_set_version); /* * Read a property stored within the master node. */ int zfs_get_zplprop(objset_t *os, zfs_prop_t prop, uint64_t *value) { const char *pname; int error = SET_ERROR(ENOENT); /* * Look up the file system's value for the property. For the * version property, we look up a slightly different string. */ if (prop == ZFS_PROP_VERSION) pname = ZPL_VERSION_STR; else pname = zfs_prop_to_name(prop); if (os != NULL) error = zap_lookup(os, MASTER_NODE_OBJ, pname, 8, 1, value); if (error == ENOENT) { /* No value set, use the default value */ switch (prop) { case ZFS_PROP_VERSION: *value = ZPL_VERSION; break; case ZFS_PROP_NORMALIZE: case ZFS_PROP_UTF8ONLY: *value = 0; break; case ZFS_PROP_CASE: *value = ZFS_CASE_SENSITIVE; break; case ZFS_PROP_ACLTYPE: *value = ZFS_ACLTYPE_OFF; break; default: return (error); } error = 0; } return (error); } EXPORT_SYMBOL(zfs_get_zplprop); void zfs_init(void) { zfsctl_init(); zfs_znode_init(); dmu_objset_register_type(DMU_OST_ZFS, zfs_space_delta_cb); register_filesystem(&zpl_fs_type); (void) arc_add_prune_callback(zpl_prune_sbs, NULL); } void zfs_fini(void) { taskq_wait(system_taskq); unregister_filesystem(&zpl_fs_type); zfs_znode_fini(); zfsctl_fini(); } diff --git a/module/zfs/zfs_znode.c b/module/zfs/zfs_znode.c index f02c02184b92..b51d780375c4 100644 --- a/module/zfs/zfs_znode.c +++ b/module/zfs/zfs_znode.c @@ -1,1994 +1,1994 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2013 by Delphix. All rights reserved. */ /* Portions Copyright 2007 Jeremy Teo */ #ifdef _KERNEL #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "fs/fs_subr.h" #include #include #include #include #include #include #include #include #include #include #include #endif /* _KERNEL */ #include #include #include #include #include #include #include #include #include "zfs_prop.h" #include "zfs_comutil.h" /* * Define ZNODE_STATS to turn on statistic gathering. By default, it is only * turned on when DEBUG is also defined. */ #ifdef DEBUG #define ZNODE_STATS #endif /* DEBUG */ #ifdef ZNODE_STATS #define ZNODE_STAT_ADD(stat) ((stat)++) #else #define ZNODE_STAT_ADD(stat) /* nothing */ #endif /* ZNODE_STATS */ /* * Functions needed for userland (ie: libzpool) are not put under * #ifdef_KERNEL; the rest of the functions have dependencies * (such as VFS logic) that will not compile easily in userland. */ #ifdef _KERNEL static kmem_cache_t *znode_cache = NULL; /*ARGSUSED*/ static int zfs_znode_cache_constructor(void *buf, void *arg, int kmflags) { znode_t *zp = buf; inode_init_once(ZTOI(zp)); list_link_init(&zp->z_link_node); mutex_init(&zp->z_lock, NULL, MUTEX_DEFAULT, NULL); rw_init(&zp->z_parent_lock, NULL, RW_DEFAULT, NULL); rw_init(&zp->z_name_lock, NULL, RW_DEFAULT, NULL); mutex_init(&zp->z_acl_lock, NULL, MUTEX_DEFAULT, NULL); rw_init(&zp->z_xattr_lock, NULL, RW_DEFAULT, NULL); mutex_init(&zp->z_range_lock, NULL, MUTEX_DEFAULT, NULL); avl_create(&zp->z_range_avl, zfs_range_compare, sizeof (rl_t), offsetof(rl_t, r_node)); zp->z_dirlocks = NULL; zp->z_acl_cached = NULL; zp->z_xattr_cached = NULL; zp->z_xattr_parent = NULL; zp->z_moved = 0; return (0); } /*ARGSUSED*/ static void zfs_znode_cache_destructor(void *buf, void *arg) { znode_t *zp = buf; ASSERT(!list_link_active(&zp->z_link_node)); mutex_destroy(&zp->z_lock); rw_destroy(&zp->z_parent_lock); rw_destroy(&zp->z_name_lock); mutex_destroy(&zp->z_acl_lock); rw_destroy(&zp->z_xattr_lock); avl_destroy(&zp->z_range_avl); mutex_destroy(&zp->z_range_lock); ASSERT(zp->z_dirlocks == NULL); ASSERT(zp->z_acl_cached == NULL); ASSERT(zp->z_xattr_cached == NULL); ASSERT(zp->z_xattr_parent == NULL); } void zfs_znode_init(void) { /* * Initialize zcache */ ASSERT(znode_cache == NULL); znode_cache = kmem_cache_create("zfs_znode_cache", sizeof (znode_t), 0, zfs_znode_cache_constructor, zfs_znode_cache_destructor, NULL, NULL, NULL, KMC_KMEM); } void zfs_znode_fini(void) { /* * Cleanup zcache */ if (znode_cache) kmem_cache_destroy(znode_cache); znode_cache = NULL; } int zfs_create_share_dir(zfs_sb_t *zsb, dmu_tx_t *tx) { #ifdef HAVE_SMB_SHARE zfs_acl_ids_t acl_ids; vattr_t vattr; znode_t *sharezp; vnode_t *vp; znode_t *zp; int error; vattr.va_mask = AT_MODE|AT_UID|AT_GID|AT_TYPE; vattr.va_mode = S_IFDIR | 0555; vattr.va_uid = crgetuid(kcred); vattr.va_gid = crgetgid(kcred); sharezp = kmem_cache_alloc(znode_cache, KM_PUSHPAGE); sharezp->z_moved = 0; sharezp->z_unlinked = 0; sharezp->z_atime_dirty = 0; sharezp->z_zfsvfs = zfsvfs; sharezp->z_is_sa = zfsvfs->z_use_sa; vp = ZTOV(sharezp); vn_reinit(vp); vp->v_type = VDIR; VERIFY(0 == zfs_acl_ids_create(sharezp, IS_ROOT_NODE, &vattr, kcred, NULL, &acl_ids)); zfs_mknode(sharezp, &vattr, tx, kcred, IS_ROOT_NODE, &zp, &acl_ids); ASSERT3P(zp, ==, sharezp); ASSERT(!vn_in_dnlc(ZTOV(sharezp))); /* not valid to move */ POINTER_INVALIDATE(&sharezp->z_zfsvfs); error = zap_add(zfsvfs->z_os, MASTER_NODE_OBJ, ZFS_SHARES_DIR, 8, 1, &sharezp->z_id, tx); zfsvfs->z_shares_dir = sharezp->z_id; zfs_acl_ids_free(&acl_ids); // ZTOV(sharezp)->v_count = 0; sa_handle_destroy(sharezp->z_sa_hdl); kmem_cache_free(znode_cache, sharezp); return (error); #else return (0); #endif /* HAVE_SMB_SHARE */ } static void zfs_znode_sa_init(zfs_sb_t *zsb, znode_t *zp, dmu_buf_t *db, dmu_object_type_t obj_type, sa_handle_t *sa_hdl) { ASSERT(MUTEX_HELD(ZFS_OBJ_MUTEX(zsb, zp->z_id))); mutex_enter(&zp->z_lock); ASSERT(zp->z_sa_hdl == NULL); ASSERT(zp->z_acl_cached == NULL); if (sa_hdl == NULL) { VERIFY(0 == sa_handle_get_from_db(zsb->z_os, db, zp, SA_HDL_SHARED, &zp->z_sa_hdl)); } else { zp->z_sa_hdl = sa_hdl; sa_set_userp(sa_hdl, zp); } zp->z_is_sa = (obj_type == DMU_OT_SA) ? B_TRUE : B_FALSE; mutex_exit(&zp->z_lock); } void zfs_znode_dmu_fini(znode_t *zp) { ASSERT(MUTEX_HELD(ZFS_OBJ_MUTEX(ZTOZSB(zp), zp->z_id)) || zp->z_unlinked || RW_WRITE_HELD(&ZTOZSB(zp)->z_teardown_inactive_lock)); sa_handle_destroy(zp->z_sa_hdl); zp->z_sa_hdl = NULL; } /* * Called by new_inode() to allocate a new inode. */ int zfs_inode_alloc(struct super_block *sb, struct inode **ip) { znode_t *zp; zp = kmem_cache_alloc(znode_cache, KM_PUSHPAGE); *ip = ZTOI(zp); return (0); } /* * Called in multiple places when an inode should be destroyed. */ void zfs_inode_destroy(struct inode *ip) { znode_t *zp = ITOZ(ip); zfs_sb_t *zsb = ZTOZSB(zp); if (zfsctl_is_node(ip)) zfsctl_inode_destroy(ip); mutex_enter(&zsb->z_znodes_lock); if (list_link_active(&zp->z_link_node)) { list_remove(&zsb->z_all_znodes, zp); zsb->z_nr_znodes--; } mutex_exit(&zsb->z_znodes_lock); if (zp->z_acl_cached) { zfs_acl_free(zp->z_acl_cached); zp->z_acl_cached = NULL; } if (zp->z_xattr_cached) { nvlist_free(zp->z_xattr_cached); zp->z_xattr_cached = NULL; } if (zp->z_xattr_parent) { zfs_iput_async(ZTOI(zp->z_xattr_parent)); zp->z_xattr_parent = NULL; } kmem_cache_free(znode_cache, zp); } static void zfs_inode_set_ops(zfs_sb_t *zsb, struct inode *ip) { uint64_t rdev = 0; switch (ip->i_mode & S_IFMT) { case S_IFREG: ip->i_op = &zpl_inode_operations; ip->i_fop = &zpl_file_operations; ip->i_mapping->a_ops = &zpl_address_space_operations; break; case S_IFDIR: ip->i_op = &zpl_dir_inode_operations; ip->i_fop = &zpl_dir_file_operations; ITOZ(ip)->z_zn_prefetch = B_TRUE; break; case S_IFLNK: ip->i_op = &zpl_symlink_inode_operations; break; /* * rdev is only stored in a SA only for device files. */ case S_IFCHR: case S_IFBLK: VERIFY(sa_lookup(ITOZ(ip)->z_sa_hdl, SA_ZPL_RDEV(zsb), &rdev, sizeof (rdev)) == 0); /*FALLTHROUGH*/ case S_IFIFO: case S_IFSOCK: init_special_inode(ip, ip->i_mode, rdev); ip->i_op = &zpl_special_inode_operations; break; default: printk("ZFS: Invalid mode: 0x%x\n", ip->i_mode); VERIFY(0); } } /* * Construct a znode+inode and initialize. * * This does not do a call to dmu_set_user() that is * up to the caller to do, in case you don't want to * return the znode */ static znode_t * zfs_znode_alloc(zfs_sb_t *zsb, dmu_buf_t *db, int blksz, dmu_object_type_t obj_type, uint64_t obj, sa_handle_t *hdl, struct inode *dip) { znode_t *zp; struct inode *ip; uint64_t mode; uint64_t parent; sa_bulk_attr_t bulk[9]; int count = 0; ASSERT(zsb != NULL); ip = new_inode(zsb->z_sb); if (ip == NULL) return (NULL); zp = ITOZ(ip); ASSERT(zp->z_dirlocks == NULL); ASSERT3P(zp->z_acl_cached, ==, NULL); ASSERT3P(zp->z_xattr_cached, ==, NULL); ASSERT3P(zp->z_xattr_parent, ==, NULL); zp->z_moved = 0; zp->z_sa_hdl = NULL; zp->z_unlinked = 0; zp->z_atime_dirty = 0; zp->z_mapcnt = 0; zp->z_id = db->db_object; zp->z_blksz = blksz; zp->z_seq = 0x7A4653; zp->z_sync_cnt = 0; zp->z_is_zvol = B_FALSE; zp->z_is_mapped = B_FALSE; zp->z_is_ctldir = B_FALSE; zp->z_is_stale = B_FALSE; zfs_znode_sa_init(zsb, zp, db, obj_type, hdl); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zsb), NULL, &mode, 8); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GEN(zsb), NULL, &zp->z_gen, 8); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zsb), NULL, &zp->z_size, 8); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_LINKS(zsb), NULL, &zp->z_links, 8); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zsb), NULL, &zp->z_pflags, 8); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_PARENT(zsb), NULL, &parent, 8); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zsb), NULL, &zp->z_atime, 16); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zsb), NULL, &zp->z_uid, 8); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zsb), NULL, &zp->z_gid, 8); if (sa_bulk_lookup(zp->z_sa_hdl, bulk, count) != 0 || zp->z_gen == 0) { if (hdl == NULL) sa_handle_destroy(zp->z_sa_hdl); goto error; } zp->z_mode = mode; /* * xattr znodes hold a reference on their unique parent */ if (dip && zp->z_pflags & ZFS_XATTR) { igrab(dip); zp->z_xattr_parent = ITOZ(dip); } ip->i_ino = obj; zfs_inode_update(zp); zfs_inode_set_ops(zsb, ip); /* * The only way insert_inode_locked() can fail is if the ip->i_ino * number is already hashed for this super block. This can never * happen because the inode numbers map 1:1 with the object numbers. * * The one exception is rolling back a mounted file system, but in * this case all the active inode are unhashed during the rollback. */ VERIFY3S(insert_inode_locked(ip), ==, 0); mutex_enter(&zsb->z_znodes_lock); list_insert_tail(&zsb->z_all_znodes, zp); zsb->z_nr_znodes++; membar_producer(); mutex_exit(&zsb->z_znodes_lock); unlock_new_inode(ip); return (zp); error: unlock_new_inode(ip); iput(ip); return (NULL); } void zfs_set_inode_flags(znode_t *zp, struct inode *ip) { /* * Linux and Solaris have different sets of file attributes, so we * restrict this conversion to the intersection of the two. */ if (zp->z_pflags & ZFS_IMMUTABLE) ip->i_flags |= S_IMMUTABLE; else ip->i_flags &= ~S_IMMUTABLE; if (zp->z_pflags & ZFS_APPENDONLY) ip->i_flags |= S_APPEND; else ip->i_flags &= ~S_APPEND; } /* * Update the embedded inode given the znode. We should work toward * eliminating this function as soon as possible by removing values * which are duplicated between the znode and inode. If the generic * inode has the correct field it should be used, and the ZFS code * updated to access the inode. This can be done incrementally. */ void zfs_inode_update(znode_t *zp) { zfs_sb_t *zsb; struct inode *ip; uint32_t blksize; uint64_t atime[2], mtime[2], ctime[2]; ASSERT(zp != NULL); zsb = ZTOZSB(zp); ip = ZTOI(zp); /* Skip .zfs control nodes which do not exist on disk. */ if (zfsctl_is_node(ip)) return; sa_lookup(zp->z_sa_hdl, SA_ZPL_ATIME(zsb), &atime, 16); sa_lookup(zp->z_sa_hdl, SA_ZPL_MTIME(zsb), &mtime, 16); sa_lookup(zp->z_sa_hdl, SA_ZPL_CTIME(zsb), &ctime, 16); spin_lock(&ip->i_lock); ip->i_generation = zp->z_gen; ip->i_uid = SUID_TO_KUID(zp->z_uid); ip->i_gid = SGID_TO_KGID(zp->z_gid); set_nlink(ip, zp->z_links); ip->i_mode = zp->z_mode; zfs_set_inode_flags(zp, ip); ip->i_blkbits = SPA_MINBLOCKSHIFT; dmu_object_size_from_db(sa_get_db(zp->z_sa_hdl), &blksize, (u_longlong_t *)&ip->i_blocks); ZFS_TIME_DECODE(&ip->i_atime, atime); ZFS_TIME_DECODE(&ip->i_mtime, mtime); ZFS_TIME_DECODE(&ip->i_ctime, ctime); i_size_write(ip, zp->z_size); spin_unlock(&ip->i_lock); } /* * Safely mark an inode dirty. Inodes which are part of a read-only * file system or snapshot may not be dirtied. */ void zfs_mark_inode_dirty(struct inode *ip) { zfs_sb_t *zsb = ITOZSB(ip); if (zfs_is_readonly(zsb) || dmu_objset_is_snapshot(zsb->z_os)) return; mark_inode_dirty(ip); } static uint64_t empty_xattr; static uint64_t pad[4]; static zfs_acl_phys_t acl_phys; /* * Create a new DMU object to hold a zfs znode. * * IN: dzp - parent directory for new znode * vap - file attributes for new znode * tx - dmu transaction id for zap operations * cr - credentials of caller * flag - flags: * IS_ROOT_NODE - new object will be root * IS_XATTR - new object is an attribute * bonuslen - length of bonus buffer * setaclp - File/Dir initial ACL * fuidp - Tracks fuid allocation. * * OUT: zpp - allocated znode * */ void zfs_mknode(znode_t *dzp, vattr_t *vap, dmu_tx_t *tx, cred_t *cr, uint_t flag, znode_t **zpp, zfs_acl_ids_t *acl_ids) { uint64_t crtime[2], atime[2], mtime[2], ctime[2]; uint64_t mode, size, links, parent, pflags; uint64_t dzp_pflags = 0; uint64_t rdev = 0; zfs_sb_t *zsb = ZTOZSB(dzp); dmu_buf_t *db; timestruc_t now; uint64_t gen, obj; int bonuslen; sa_handle_t *sa_hdl; dmu_object_type_t obj_type; sa_bulk_attr_t *sa_attrs; int cnt = 0; zfs_acl_locator_cb_t locate = { 0 }; if (zsb->z_replay) { obj = vap->va_nodeid; now = vap->va_ctime; /* see zfs_replay_create() */ gen = vap->va_nblocks; /* ditto */ } else { obj = 0; gethrestime(&now); gen = dmu_tx_get_txg(tx); } obj_type = zsb->z_use_sa ? DMU_OT_SA : DMU_OT_ZNODE; bonuslen = (obj_type == DMU_OT_SA) ? DN_MAX_BONUSLEN : ZFS_OLD_ZNODE_PHYS_SIZE; /* * Create a new DMU object. */ /* * There's currently no mechanism for pre-reading the blocks that will * be needed to allocate a new object, so we accept the small chance * that there will be an i/o error and we will fail one of the * assertions below. */ if (S_ISDIR(vap->va_mode)) { if (zsb->z_replay) { VERIFY0(zap_create_claim_norm(zsb->z_os, obj, zsb->z_norm, DMU_OT_DIRECTORY_CONTENTS, obj_type, bonuslen, tx)); } else { obj = zap_create_norm(zsb->z_os, zsb->z_norm, DMU_OT_DIRECTORY_CONTENTS, obj_type, bonuslen, tx); } } else { if (zsb->z_replay) { VERIFY0(dmu_object_claim(zsb->z_os, obj, DMU_OT_PLAIN_FILE_CONTENTS, 0, obj_type, bonuslen, tx)); } else { obj = dmu_object_alloc(zsb->z_os, DMU_OT_PLAIN_FILE_CONTENTS, 0, obj_type, bonuslen, tx); } } ZFS_OBJ_HOLD_ENTER(zsb, obj); VERIFY(0 == sa_buf_hold(zsb->z_os, obj, NULL, &db)); /* * If this is the root, fix up the half-initialized parent pointer * to reference the just-allocated physical data area. */ if (flag & IS_ROOT_NODE) { dzp->z_id = obj; } else { dzp_pflags = dzp->z_pflags; } /* * If parent is an xattr, so am I. */ if (dzp_pflags & ZFS_XATTR) { flag |= IS_XATTR; } if (zsb->z_use_fuids) pflags = ZFS_ARCHIVE | ZFS_AV_MODIFIED; else pflags = 0; if (S_ISDIR(vap->va_mode)) { size = 2; /* contents ("." and "..") */ links = (flag & (IS_ROOT_NODE | IS_XATTR)) ? 2 : 1; } else { size = links = 0; } if (S_ISBLK(vap->va_mode) || S_ISCHR(vap->va_mode)) rdev = vap->va_rdev; parent = dzp->z_id; mode = acl_ids->z_mode; if (flag & IS_XATTR) pflags |= ZFS_XATTR; /* * No execs denied will be deterimed when zfs_mode_compute() is called. */ pflags |= acl_ids->z_aclp->z_hints & (ZFS_ACL_TRIVIAL|ZFS_INHERIT_ACE|ZFS_ACL_AUTO_INHERIT| ZFS_ACL_DEFAULTED|ZFS_ACL_PROTECTED); ZFS_TIME_ENCODE(&now, crtime); ZFS_TIME_ENCODE(&now, ctime); if (vap->va_mask & ATTR_ATIME) { ZFS_TIME_ENCODE(&vap->va_atime, atime); } else { ZFS_TIME_ENCODE(&now, atime); } if (vap->va_mask & ATTR_MTIME) { ZFS_TIME_ENCODE(&vap->va_mtime, mtime); } else { ZFS_TIME_ENCODE(&now, mtime); } /* Now add in all of the "SA" attributes */ VERIFY(0 == sa_handle_get_from_db(zsb->z_os, db, NULL, SA_HDL_SHARED, &sa_hdl)); /* * Setup the array of attributes to be replaced/set on the new file * * order for DMU_OT_ZNODE is critical since it needs to be constructed * in the old znode_phys_t format. Don't change this ordering */ sa_attrs = kmem_alloc(sizeof (sa_bulk_attr_t) * ZPL_END, KM_PUSHPAGE); if (obj_type == DMU_OT_ZNODE) { SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ATIME(zsb), NULL, &atime, 16); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MTIME(zsb), NULL, &mtime, 16); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CTIME(zsb), NULL, &ctime, 16); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CRTIME(zsb), NULL, &crtime, 16); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GEN(zsb), NULL, &gen, 8); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MODE(zsb), NULL, &mode, 8); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_SIZE(zsb), NULL, &size, 8); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_PARENT(zsb), NULL, &parent, 8); } else { SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MODE(zsb), NULL, &mode, 8); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_SIZE(zsb), NULL, &size, 8); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GEN(zsb), NULL, &gen, 8); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_UID(zsb), NULL, &acl_ids->z_fuid, 8); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GID(zsb), NULL, &acl_ids->z_fgid, 8); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_PARENT(zsb), NULL, &parent, 8); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_FLAGS(zsb), NULL, &pflags, 8); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ATIME(zsb), NULL, &atime, 16); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MTIME(zsb), NULL, &mtime, 16); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CTIME(zsb), NULL, &ctime, 16); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CRTIME(zsb), NULL, &crtime, 16); } SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_LINKS(zsb), NULL, &links, 8); if (obj_type == DMU_OT_ZNODE) { SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_XATTR(zsb), NULL, &empty_xattr, 8); } if (obj_type == DMU_OT_ZNODE || (S_ISBLK(vap->va_mode) || S_ISCHR(vap->va_mode))) { SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_RDEV(zsb), NULL, &rdev, 8); } if (obj_type == DMU_OT_ZNODE) { SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_FLAGS(zsb), NULL, &pflags, 8); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_UID(zsb), NULL, &acl_ids->z_fuid, 8); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GID(zsb), NULL, &acl_ids->z_fgid, 8); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_PAD(zsb), NULL, pad, sizeof (uint64_t) * 4); SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ZNODE_ACL(zsb), NULL, &acl_phys, sizeof (zfs_acl_phys_t)); } else if (acl_ids->z_aclp->z_version >= ZFS_ACL_VERSION_FUID) { SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_DACL_COUNT(zsb), NULL, &acl_ids->z_aclp->z_acl_count, 8); locate.cb_aclp = acl_ids->z_aclp; SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_DACL_ACES(zsb), zfs_acl_data_locator, &locate, acl_ids->z_aclp->z_acl_bytes); mode = zfs_mode_compute(mode, acl_ids->z_aclp, &pflags, acl_ids->z_fuid, acl_ids->z_fgid); } VERIFY(sa_replace_all_by_template(sa_hdl, sa_attrs, cnt, tx) == 0); if (!(flag & IS_ROOT_NODE)) { *zpp = zfs_znode_alloc(zsb, db, 0, obj_type, obj, sa_hdl, ZTOI(dzp)); VERIFY(*zpp != NULL); VERIFY(dzp != NULL); } else { /* * If we are creating the root node, the "parent" we * passed in is the znode for the root. */ *zpp = dzp; (*zpp)->z_sa_hdl = sa_hdl; } (*zpp)->z_pflags = pflags; (*zpp)->z_mode = mode; if (obj_type == DMU_OT_ZNODE || acl_ids->z_aclp->z_version < ZFS_ACL_VERSION_FUID) { VERIFY0(zfs_aclset_common(*zpp, acl_ids->z_aclp, cr, tx)); } kmem_free(sa_attrs, sizeof (sa_bulk_attr_t) * ZPL_END); ZFS_OBJ_HOLD_EXIT(zsb, obj); } /* * Update in-core attributes. It is assumed the caller will be doing an * sa_bulk_update to push the changes out. */ void zfs_xvattr_set(znode_t *zp, xvattr_t *xvap, dmu_tx_t *tx) { xoptattr_t *xoap; xoap = xva_getxoptattr(xvap); ASSERT(xoap); if (XVA_ISSET_REQ(xvap, XAT_CREATETIME)) { uint64_t times[2]; ZFS_TIME_ENCODE(&xoap->xoa_createtime, times); (void) sa_update(zp->z_sa_hdl, SA_ZPL_CRTIME(ZTOZSB(zp)), ×, sizeof (times), tx); XVA_SET_RTN(xvap, XAT_CREATETIME); } if (XVA_ISSET_REQ(xvap, XAT_READONLY)) { ZFS_ATTR_SET(zp, ZFS_READONLY, xoap->xoa_readonly, zp->z_pflags, tx); XVA_SET_RTN(xvap, XAT_READONLY); } if (XVA_ISSET_REQ(xvap, XAT_HIDDEN)) { ZFS_ATTR_SET(zp, ZFS_HIDDEN, xoap->xoa_hidden, zp->z_pflags, tx); XVA_SET_RTN(xvap, XAT_HIDDEN); } if (XVA_ISSET_REQ(xvap, XAT_SYSTEM)) { ZFS_ATTR_SET(zp, ZFS_SYSTEM, xoap->xoa_system, zp->z_pflags, tx); XVA_SET_RTN(xvap, XAT_SYSTEM); } if (XVA_ISSET_REQ(xvap, XAT_ARCHIVE)) { ZFS_ATTR_SET(zp, ZFS_ARCHIVE, xoap->xoa_archive, zp->z_pflags, tx); XVA_SET_RTN(xvap, XAT_ARCHIVE); } if (XVA_ISSET_REQ(xvap, XAT_IMMUTABLE)) { ZFS_ATTR_SET(zp, ZFS_IMMUTABLE, xoap->xoa_immutable, zp->z_pflags, tx); XVA_SET_RTN(xvap, XAT_IMMUTABLE); } if (XVA_ISSET_REQ(xvap, XAT_NOUNLINK)) { ZFS_ATTR_SET(zp, ZFS_NOUNLINK, xoap->xoa_nounlink, zp->z_pflags, tx); XVA_SET_RTN(xvap, XAT_NOUNLINK); } if (XVA_ISSET_REQ(xvap, XAT_APPENDONLY)) { ZFS_ATTR_SET(zp, ZFS_APPENDONLY, xoap->xoa_appendonly, zp->z_pflags, tx); XVA_SET_RTN(xvap, XAT_APPENDONLY); } if (XVA_ISSET_REQ(xvap, XAT_NODUMP)) { ZFS_ATTR_SET(zp, ZFS_NODUMP, xoap->xoa_nodump, zp->z_pflags, tx); XVA_SET_RTN(xvap, XAT_NODUMP); } if (XVA_ISSET_REQ(xvap, XAT_OPAQUE)) { ZFS_ATTR_SET(zp, ZFS_OPAQUE, xoap->xoa_opaque, zp->z_pflags, tx); XVA_SET_RTN(xvap, XAT_OPAQUE); } if (XVA_ISSET_REQ(xvap, XAT_AV_QUARANTINED)) { ZFS_ATTR_SET(zp, ZFS_AV_QUARANTINED, xoap->xoa_av_quarantined, zp->z_pflags, tx); XVA_SET_RTN(xvap, XAT_AV_QUARANTINED); } if (XVA_ISSET_REQ(xvap, XAT_AV_MODIFIED)) { ZFS_ATTR_SET(zp, ZFS_AV_MODIFIED, xoap->xoa_av_modified, zp->z_pflags, tx); XVA_SET_RTN(xvap, XAT_AV_MODIFIED); } if (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP)) { zfs_sa_set_scanstamp(zp, xvap, tx); XVA_SET_RTN(xvap, XAT_AV_SCANSTAMP); } if (XVA_ISSET_REQ(xvap, XAT_REPARSE)) { ZFS_ATTR_SET(zp, ZFS_REPARSE, xoap->xoa_reparse, zp->z_pflags, tx); XVA_SET_RTN(xvap, XAT_REPARSE); } if (XVA_ISSET_REQ(xvap, XAT_OFFLINE)) { ZFS_ATTR_SET(zp, ZFS_OFFLINE, xoap->xoa_offline, zp->z_pflags, tx); XVA_SET_RTN(xvap, XAT_OFFLINE); } if (XVA_ISSET_REQ(xvap, XAT_SPARSE)) { ZFS_ATTR_SET(zp, ZFS_SPARSE, xoap->xoa_sparse, zp->z_pflags, tx); XVA_SET_RTN(xvap, XAT_SPARSE); } } int zfs_zget(zfs_sb_t *zsb, uint64_t obj_num, znode_t **zpp) { dmu_object_info_t doi; dmu_buf_t *db; znode_t *zp; int err; sa_handle_t *hdl; *zpp = NULL; again: ZFS_OBJ_HOLD_ENTER(zsb, obj_num); err = sa_buf_hold(zsb->z_os, obj_num, NULL, &db); if (err) { ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (err); } dmu_object_info_from_db(db, &doi); if (doi.doi_bonus_type != DMU_OT_SA && (doi.doi_bonus_type != DMU_OT_ZNODE || (doi.doi_bonus_type == DMU_OT_ZNODE && doi.doi_bonus_size < sizeof (znode_phys_t)))) { sa_buf_rele(db, NULL); ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (SET_ERROR(EINVAL)); } hdl = dmu_buf_get_user(db); if (hdl != NULL) { zp = sa_get_userdata(hdl); /* * Since "SA" does immediate eviction we * should never find a sa handle that doesn't * know about the znode. */ ASSERT3P(zp, !=, NULL); mutex_enter(&zp->z_lock); ASSERT3U(zp->z_id, ==, obj_num); if (zp->z_unlinked) { err = SET_ERROR(ENOENT); } else { /* * If igrab() returns NULL the VFS has independently * determined the inode should be evicted and has * called iput_final() to start the eviction process. * The SA handle is still valid but because the VFS * requires that the eviction succeed we must drop * our locks and references to allow the eviction to * complete. The zfs_zget() may then be retried. * * This unlikely case could be optimized by registering * a sops->drop_inode() callback. The callback would * need to detect the active SA hold thereby informing * the VFS that this inode should not be evicted. */ if (igrab(ZTOI(zp)) == NULL) { mutex_exit(&zp->z_lock); sa_buf_rele(db, NULL); ZFS_OBJ_HOLD_EXIT(zsb, obj_num); goto again; } *zpp = zp; err = 0; } mutex_exit(&zp->z_lock); sa_buf_rele(db, NULL); ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (err); } /* * Not found create new znode/vnode but only if file exists. * * There is a small window where zfs_vget() could * find this object while a file create is still in * progress. This is checked for in zfs_znode_alloc() * * if zfs_znode_alloc() fails it will drop the hold on the * bonus buffer. */ zp = zfs_znode_alloc(zsb, db, doi.doi_data_block_size, doi.doi_bonus_type, obj_num, NULL, NULL); if (zp == NULL) { err = SET_ERROR(ENOENT); } else { *zpp = zp; } ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (err); } int zfs_rezget(znode_t *zp) { zfs_sb_t *zsb = ZTOZSB(zp); dmu_object_info_t doi; dmu_buf_t *db; uint64_t obj_num = zp->z_id; uint64_t mode; sa_bulk_attr_t bulk[8]; int err; int count = 0; uint64_t gen; ZFS_OBJ_HOLD_ENTER(zsb, obj_num); mutex_enter(&zp->z_acl_lock); if (zp->z_acl_cached) { zfs_acl_free(zp->z_acl_cached); zp->z_acl_cached = NULL; } mutex_exit(&zp->z_acl_lock); rw_enter(&zp->z_xattr_lock, RW_WRITER); if (zp->z_xattr_cached) { nvlist_free(zp->z_xattr_cached); zp->z_xattr_cached = NULL; } if (zp->z_xattr_parent) { zfs_iput_async(ZTOI(zp->z_xattr_parent)); zp->z_xattr_parent = NULL; } rw_exit(&zp->z_xattr_lock); ASSERT(zp->z_sa_hdl == NULL); err = sa_buf_hold(zsb->z_os, obj_num, NULL, &db); if (err) { ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (err); } dmu_object_info_from_db(db, &doi); if (doi.doi_bonus_type != DMU_OT_SA && (doi.doi_bonus_type != DMU_OT_ZNODE || (doi.doi_bonus_type == DMU_OT_ZNODE && doi.doi_bonus_size < sizeof (znode_phys_t)))) { sa_buf_rele(db, NULL); ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (SET_ERROR(EINVAL)); } zfs_znode_sa_init(zsb, zp, db, doi.doi_bonus_type, NULL); /* reload cached values */ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GEN(zsb), NULL, &gen, sizeof (gen)); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zsb), NULL, &zp->z_size, sizeof (zp->z_size)); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_LINKS(zsb), NULL, &zp->z_links, sizeof (zp->z_links)); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zsb), NULL, &zp->z_pflags, sizeof (zp->z_pflags)); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zsb), NULL, &zp->z_atime, sizeof (zp->z_atime)); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zsb), NULL, &zp->z_uid, sizeof (zp->z_uid)); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zsb), NULL, &zp->z_gid, sizeof (zp->z_gid)); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zsb), NULL, &mode, sizeof (mode)); if (sa_bulk_lookup(zp->z_sa_hdl, bulk, count)) { zfs_znode_dmu_fini(zp); ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (SET_ERROR(EIO)); } zp->z_mode = mode; if (gen != zp->z_gen) { zfs_znode_dmu_fini(zp); ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (SET_ERROR(EIO)); } zp->z_unlinked = (zp->z_links == 0); zp->z_blksz = doi.doi_data_block_size; zfs_inode_update(zp); ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (0); } void zfs_znode_delete(znode_t *zp, dmu_tx_t *tx) { zfs_sb_t *zsb = ZTOZSB(zp); objset_t *os = zsb->z_os; uint64_t obj = zp->z_id; uint64_t acl_obj = zfs_external_acl(zp); ZFS_OBJ_HOLD_ENTER(zsb, obj); if (acl_obj) { VERIFY(!zp->z_is_sa); VERIFY(0 == dmu_object_free(os, acl_obj, tx)); } VERIFY(0 == dmu_object_free(os, obj, tx)); zfs_znode_dmu_fini(zp); ZFS_OBJ_HOLD_EXIT(zsb, obj); } void zfs_zinactive(znode_t *zp) { zfs_sb_t *zsb = ZTOZSB(zp); uint64_t z_id = zp->z_id; boolean_t drop_mutex = 0; ASSERT(zp->z_sa_hdl); /* * Don't allow a zfs_zget() while were trying to release this znode. * * Linux allows direct memory reclaim which means that any KM_SLEEP * allocation may trigger inode eviction. This can lead to a deadlock * through the ->shrink_icache_memory()->evict()->zfs_inactive()-> * zfs_zinactive() call path. To avoid this deadlock the process * must not reacquire the mutex when it is already holding it. */ if (!ZFS_OBJ_HOLD_OWNED(zsb, z_id)) { ZFS_OBJ_HOLD_ENTER(zsb, z_id); drop_mutex = 1; } mutex_enter(&zp->z_lock); /* * If this was the last reference to a file with no links, * remove the file from the file system. */ if (zp->z_unlinked) { mutex_exit(&zp->z_lock); if (drop_mutex) ZFS_OBJ_HOLD_EXIT(zsb, z_id); zfs_rmnode(zp); return; } mutex_exit(&zp->z_lock); zfs_znode_dmu_fini(zp); if (drop_mutex) ZFS_OBJ_HOLD_EXIT(zsb, z_id); } static inline int zfs_compare_timespec(struct timespec *t1, struct timespec *t2) { if (t1->tv_sec < t2->tv_sec) return (-1); if (t1->tv_sec > t2->tv_sec) return (1); return (t1->tv_nsec - t2->tv_nsec); } /* * Determine whether the znode's atime must be updated. The logic mostly * duplicates the Linux kernel's relatime_need_update() functionality. * This function is only called if the underlying filesystem actually has * atime updates enabled. */ static inline boolean_t zfs_atime_need_update(znode_t *zp, timestruc_t *now) { if (!ZTOZSB(zp)->z_relatime) return (B_TRUE); /* * In relatime mode, only update the atime if the previous atime * is earlier than either the ctime or mtime or if at least a day * has passed since the last update of atime. */ if (zfs_compare_timespec(&ZTOI(zp)->i_mtime, &ZTOI(zp)->i_atime) >= 0) return (B_TRUE); if (zfs_compare_timespec(&ZTOI(zp)->i_ctime, &ZTOI(zp)->i_atime) >= 0) return (B_TRUE); if ((long)now->tv_sec - ZTOI(zp)->i_atime.tv_sec >= 24*60*60) return (B_TRUE); return (B_FALSE); } /* * Prepare to update znode time stamps. * * IN: zp - znode requiring timestamp update * flag - ATTR_MTIME, ATTR_CTIME, ATTR_ATIME flags * have_tx - true of caller is creating a new txg * * OUT: zp - new atime (via underlying inode's i_atime) * mtime - new mtime * ctime - new ctime * * NOTE: The arguments are somewhat redundant. The following condition * is always true: * * have_tx == !(flag & ATTR_ATIME) */ void zfs_tstamp_update_setup(znode_t *zp, uint_t flag, uint64_t mtime[2], uint64_t ctime[2], boolean_t have_tx) { timestruc_t now; ASSERT(have_tx == !(flag & ATTR_ATIME)); gethrestime(&now); /* * NOTE: The following test intentionally does not update z_atime_dirty * in the case where an ATIME update has been requested but for which * the update is omitted due to relatime logic. The rationale being * that if the flag was set somewhere else, we should leave it alone * here. */ if (flag & ATTR_ATIME) { if (zfs_atime_need_update(zp, &now)) { ZFS_TIME_ENCODE(&now, zp->z_atime); ZTOI(zp)->i_atime.tv_sec = zp->z_atime[0]; ZTOI(zp)->i_atime.tv_nsec = zp->z_atime[1]; zp->z_atime_dirty = 1; } } else { zp->z_atime_dirty = 0; zp->z_seq++; } if (flag & ATTR_MTIME) { ZFS_TIME_ENCODE(&now, mtime); if (ZTOZSB(zp)->z_use_fuids) { zp->z_pflags |= (ZFS_ARCHIVE | ZFS_AV_MODIFIED); } } if (flag & ATTR_CTIME) { ZFS_TIME_ENCODE(&now, ctime); if (ZTOZSB(zp)->z_use_fuids) zp->z_pflags |= ZFS_ARCHIVE; } } /* * Grow the block size for a file. * * IN: zp - znode of file to free data in. * size - requested block size * tx - open transaction. * * NOTE: this function assumes that the znode is write locked. */ void zfs_grow_blocksize(znode_t *zp, uint64_t size, dmu_tx_t *tx) { int error; u_longlong_t dummy; if (size <= zp->z_blksz) return; /* * If the file size is already greater than the current blocksize, * we will not grow. If there is more than one block in a file, * the blocksize cannot change. */ if (zp->z_blksz && zp->z_size > zp->z_blksz) return; error = dmu_object_set_blocksize(ZTOZSB(zp)->z_os, zp->z_id, size, 0, tx); if (error == ENOTSUP) return; ASSERT0(error); /* What blocksize did we actually get? */ dmu_object_size_from_db(sa_get_db(zp->z_sa_hdl), &zp->z_blksz, &dummy); } /* * Increase the file length * * IN: zp - znode of file to free data in. * end - new end-of-file * * RETURN: 0 on success, error code on failure */ static int zfs_extend(znode_t *zp, uint64_t end) { zfs_sb_t *zsb = ZTOZSB(zp); dmu_tx_t *tx; rl_t *rl; uint64_t newblksz; int error; /* * We will change zp_size, lock the whole file. */ rl = zfs_range_lock(zp, 0, UINT64_MAX, RL_WRITER); /* * Nothing to do if file already at desired length. */ if (end <= zp->z_size) { zfs_range_unlock(rl); return (0); } tx = dmu_tx_create(zsb->z_os); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); zfs_sa_upgrade_txholds(tx, zp); if (end > zp->z_blksz && (!ISP2(zp->z_blksz) || zp->z_blksz < zsb->z_max_blksz)) { /* * We are growing the file past the current block size. */ if (zp->z_blksz > ZTOZSB(zp)->z_max_blksz) { ASSERT(!ISP2(zp->z_blksz)); newblksz = MIN(end, SPA_MAXBLOCKSIZE); } else { newblksz = MIN(end, ZTOZSB(zp)->z_max_blksz); } dmu_tx_hold_write(tx, zp->z_id, 0, newblksz); } else { newblksz = 0; } error = dmu_tx_assign(tx, TXG_WAIT); if (error) { dmu_tx_abort(tx); zfs_range_unlock(rl); return (error); } if (newblksz) zfs_grow_blocksize(zp, newblksz, tx); zp->z_size = end; VERIFY(0 == sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(ZTOZSB(zp)), &zp->z_size, sizeof (zp->z_size), tx)); zfs_range_unlock(rl); dmu_tx_commit(tx); return (0); } /* * zfs_zero_partial_page - Modeled after update_pages() but * with different arguments and semantics for use by zfs_freesp(). * * Zeroes a piece of a single page cache entry for zp at offset * start and length len. * * Caller must acquire a range lock on the file for the region * being zeroed in order that the ARC and page cache stay in sync. */ static void zfs_zero_partial_page(znode_t *zp, uint64_t start, uint64_t len) { struct address_space *mp = ZTOI(zp)->i_mapping; struct page *pp; int64_t off; void *pb; ASSERT((start & PAGE_CACHE_MASK) == ((start + len - 1) & PAGE_CACHE_MASK)); off = start & (PAGE_CACHE_SIZE - 1); start &= PAGE_CACHE_MASK; pp = find_lock_page(mp, start >> PAGE_CACHE_SHIFT); if (pp) { if (mapping_writably_mapped(mp)) flush_dcache_page(pp); pb = kmap(pp); bzero(pb + off, len); kunmap(pp); if (mapping_writably_mapped(mp)) flush_dcache_page(pp); mark_page_accessed(pp); SetPageUptodate(pp); ClearPageError(pp); unlock_page(pp); page_cache_release(pp); } } /* * Free space in a file. * * IN: zp - znode of file to free data in. * off - start of section to free. * len - length of section to free. * * RETURN: 0 on success, error code on failure */ static int zfs_free_range(znode_t *zp, uint64_t off, uint64_t len) { zfs_sb_t *zsb = ZTOZSB(zp); rl_t *rl; int error; /* * Lock the range being freed. */ rl = zfs_range_lock(zp, off, len, RL_WRITER); /* * Nothing to do if file already at desired length. */ if (off >= zp->z_size) { zfs_range_unlock(rl); return (0); } if (off + len > zp->z_size) len = zp->z_size - off; error = dmu_free_long_range(zsb->z_os, zp->z_id, off, len); /* * Zero partial page cache entries. This must be done under a * range lock in order to keep the ARC and page cache in sync. */ if (zp->z_is_mapped) { loff_t first_page, last_page, page_len; loff_t first_page_offset, last_page_offset; /* first possible full page in hole */ first_page = (off + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; /* last page of hole */ last_page = (off + len) >> PAGE_CACHE_SHIFT; /* offset of first_page */ first_page_offset = first_page << PAGE_CACHE_SHIFT; /* offset of last_page */ last_page_offset = last_page << PAGE_CACHE_SHIFT; /* truncate whole pages */ if (last_page_offset > first_page_offset) { truncate_inode_pages_range(ZTOI(zp)->i_mapping, first_page_offset, last_page_offset - 1); } /* truncate sub-page ranges */ if (first_page > last_page) { /* entire punched area within a single page */ zfs_zero_partial_page(zp, off, len); } else { /* beginning of punched area at the end of a page */ page_len = first_page_offset - off; if (page_len > 0) zfs_zero_partial_page(zp, off, page_len); /* end of punched area at the beginning of a page */ page_len = off + len - last_page_offset; if (page_len > 0) zfs_zero_partial_page(zp, last_page_offset, page_len); } } zfs_range_unlock(rl); return (error); } /* * Truncate a file * * IN: zp - znode of file to free data in. * end - new end-of-file. * * RETURN: 0 on success, error code on failure */ static int zfs_trunc(znode_t *zp, uint64_t end) { zfs_sb_t *zsb = ZTOZSB(zp); dmu_tx_t *tx; rl_t *rl; int error; sa_bulk_attr_t bulk[2]; int count = 0; /* * We will change zp_size, lock the whole file. */ rl = zfs_range_lock(zp, 0, UINT64_MAX, RL_WRITER); /* * Nothing to do if file already at desired length. */ if (end >= zp->z_size) { zfs_range_unlock(rl); return (0); } error = dmu_free_long_range(zsb->z_os, zp->z_id, end, -1); if (error) { zfs_range_unlock(rl); return (error); } tx = dmu_tx_create(zsb->z_os); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); zfs_sa_upgrade_txholds(tx, zp); error = dmu_tx_assign(tx, TXG_WAIT); if (error) { dmu_tx_abort(tx); zfs_range_unlock(rl); return (error); } zp->z_size = end; SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zsb), NULL, &zp->z_size, sizeof (zp->z_size)); if (end == 0) { zp->z_pflags &= ~ZFS_SPARSE; SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zsb), NULL, &zp->z_pflags, 8); } VERIFY(sa_bulk_update(zp->z_sa_hdl, bulk, count, tx) == 0); dmu_tx_commit(tx); zfs_range_unlock(rl); return (0); } /* * Free space in a file * * IN: zp - znode of file to free data in. * off - start of range * len - end of range (0 => EOF) * flag - current file open mode flags. * log - TRUE if this action should be logged * * RETURN: 0 on success, error code on failure */ int zfs_freesp(znode_t *zp, uint64_t off, uint64_t len, int flag, boolean_t log) { dmu_tx_t *tx; zfs_sb_t *zsb = ZTOZSB(zp); zilog_t *zilog = zsb->z_log; uint64_t mode; uint64_t mtime[2], ctime[2]; sa_bulk_attr_t bulk[3]; int count = 0; int error; if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_MODE(zsb), &mode, sizeof (mode))) != 0) return (error); if (off > zp->z_size) { error = zfs_extend(zp, off+len); if (error == 0 && log) goto log; goto out; } if (len == 0) { error = zfs_trunc(zp, off); } else { if ((error = zfs_free_range(zp, off, len)) == 0 && off + len > zp->z_size) error = zfs_extend(zp, off+len); } if (error || !log) goto out; log: tx = dmu_tx_create(zsb->z_os); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); zfs_sa_upgrade_txholds(tx, zp); error = dmu_tx_assign(tx, TXG_WAIT); if (error) { dmu_tx_abort(tx); goto out; } SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zsb), NULL, mtime, 16); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zsb), NULL, ctime, 16); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zsb), NULL, &zp->z_pflags, 8); zfs_tstamp_update_setup(zp, CONTENT_MODIFIED, mtime, ctime, B_TRUE); error = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx); ASSERT(error == 0); zfs_log_truncate(zilog, tx, TX_TRUNCATE, zp, off, len); dmu_tx_commit(tx); zfs_inode_update(zp); error = 0; out: /* * Truncate the page cache - for file truncate operations, use * the purpose-built API for truncations. For punching operations, * the truncation is handled under a range lock in zfs_free_range. */ if (len == 0) truncate_setsize(ZTOI(zp), off); return (error); } void zfs_create_fs(objset_t *os, cred_t *cr, nvlist_t *zplprops, dmu_tx_t *tx) { struct super_block *sb; zfs_sb_t *zsb; uint64_t moid, obj, sa_obj, version; uint64_t sense = ZFS_CASE_SENSITIVE; uint64_t norm = 0; nvpair_t *elem; int error; int i; znode_t *rootzp = NULL; vattr_t vattr; znode_t *zp; zfs_acl_ids_t acl_ids; /* * First attempt to create master node. */ /* * In an empty objset, there are no blocks to read and thus * there can be no i/o errors (which we assert below). */ moid = MASTER_NODE_OBJ; error = zap_create_claim(os, moid, DMU_OT_MASTER_NODE, DMU_OT_NONE, 0, tx); ASSERT(error == 0); /* * Set starting attributes. */ version = zfs_zpl_version_map(spa_version(dmu_objset_spa(os))); elem = NULL; while ((elem = nvlist_next_nvpair(zplprops, elem)) != NULL) { /* For the moment we expect all zpl props to be uint64_ts */ uint64_t val; char *name; ASSERT(nvpair_type(elem) == DATA_TYPE_UINT64); VERIFY(nvpair_value_uint64(elem, &val) == 0); name = nvpair_name(elem); if (strcmp(name, zfs_prop_to_name(ZFS_PROP_VERSION)) == 0) { if (val < version) version = val; } else { error = zap_update(os, moid, name, 8, 1, &val, tx); } ASSERT(error == 0); if (strcmp(name, zfs_prop_to_name(ZFS_PROP_NORMALIZE)) == 0) norm = val; else if (strcmp(name, zfs_prop_to_name(ZFS_PROP_CASE)) == 0) sense = val; } ASSERT(version != 0); error = zap_update(os, moid, ZPL_VERSION_STR, 8, 1, &version, tx); /* * Create zap object used for SA attribute registration */ if (version >= ZPL_VERSION_SA) { sa_obj = zap_create(os, DMU_OT_SA_MASTER_NODE, DMU_OT_NONE, 0, tx); error = zap_add(os, moid, ZFS_SA_ATTRS, 8, 1, &sa_obj, tx); ASSERT(error == 0); } else { sa_obj = 0; } /* * Create a delete queue. */ obj = zap_create(os, DMU_OT_UNLINKED_SET, DMU_OT_NONE, 0, tx); error = zap_add(os, moid, ZFS_UNLINKED_SET, 8, 1, &obj, tx); ASSERT(error == 0); /* * Create root znode. Create minimal znode/inode/zsb/sb * to allow zfs_mknode to work. */ vattr.va_mask = ATTR_MODE|ATTR_UID|ATTR_GID; vattr.va_mode = S_IFDIR|0755; vattr.va_uid = crgetuid(cr); vattr.va_gid = crgetgid(cr); rootzp = kmem_cache_alloc(znode_cache, KM_PUSHPAGE); rootzp->z_moved = 0; rootzp->z_unlinked = 0; rootzp->z_atime_dirty = 0; rootzp->z_is_sa = USE_SA(version, os); - zsb = kmem_zalloc(sizeof (zfs_sb_t), KM_PUSHPAGE | KM_NODEBUG); + zsb = kmem_zalloc(sizeof (zfs_sb_t), KM_PUSHPAGE); zsb->z_os = os; zsb->z_parent = zsb; zsb->z_version = version; zsb->z_use_fuids = USE_FUIDS(version, os); zsb->z_use_sa = USE_SA(version, os); zsb->z_norm = norm; sb = kmem_zalloc(sizeof (struct super_block), KM_PUSHPAGE); sb->s_fs_info = zsb; ZTOI(rootzp)->i_sb = sb; error = sa_setup(os, sa_obj, zfs_attr_table, ZPL_END, &zsb->z_attr_table); ASSERT(error == 0); /* * Fold case on file systems that are always or sometimes case * insensitive. */ if (sense == ZFS_CASE_INSENSITIVE || sense == ZFS_CASE_MIXED) zsb->z_norm |= U8_TEXTPREP_TOUPPER; mutex_init(&zsb->z_znodes_lock, NULL, MUTEX_DEFAULT, NULL); list_create(&zsb->z_all_znodes, sizeof (znode_t), offsetof(znode_t, z_link_node)); for (i = 0; i != ZFS_OBJ_MTX_SZ; i++) mutex_init(&zsb->z_hold_mtx[i], NULL, MUTEX_DEFAULT, NULL); VERIFY(0 == zfs_acl_ids_create(rootzp, IS_ROOT_NODE, &vattr, cr, NULL, &acl_ids)); zfs_mknode(rootzp, &vattr, tx, cr, IS_ROOT_NODE, &zp, &acl_ids); ASSERT3P(zp, ==, rootzp); error = zap_add(os, moid, ZFS_ROOT_OBJ, 8, 1, &rootzp->z_id, tx); ASSERT(error == 0); zfs_acl_ids_free(&acl_ids); atomic_set(&ZTOI(rootzp)->i_count, 0); sa_handle_destroy(rootzp->z_sa_hdl); kmem_cache_free(znode_cache, rootzp); /* * Create shares directory */ error = zfs_create_share_dir(zsb, tx); ASSERT(error == 0); for (i = 0; i != ZFS_OBJ_MTX_SZ; i++) mutex_destroy(&zsb->z_hold_mtx[i]); kmem_free(sb, sizeof (struct super_block)); kmem_free(zsb, sizeof (zfs_sb_t)); } #endif /* _KERNEL */ static int zfs_sa_setup(objset_t *osp, sa_attr_type_t **sa_table) { uint64_t sa_obj = 0; int error; error = zap_lookup(osp, MASTER_NODE_OBJ, ZFS_SA_ATTRS, 8, 1, &sa_obj); if (error != 0 && error != ENOENT) return (error); error = sa_setup(osp, sa_obj, zfs_attr_table, ZPL_END, sa_table); return (error); } static int zfs_grab_sa_handle(objset_t *osp, uint64_t obj, sa_handle_t **hdlp, dmu_buf_t **db, void *tag) { dmu_object_info_t doi; int error; if ((error = sa_buf_hold(osp, obj, tag, db)) != 0) return (error); dmu_object_info_from_db(*db, &doi); if ((doi.doi_bonus_type != DMU_OT_SA && doi.doi_bonus_type != DMU_OT_ZNODE) || (doi.doi_bonus_type == DMU_OT_ZNODE && doi.doi_bonus_size < sizeof (znode_phys_t))) { sa_buf_rele(*db, tag); return (SET_ERROR(ENOTSUP)); } error = sa_handle_get(osp, obj, NULL, SA_HDL_PRIVATE, hdlp); if (error != 0) { sa_buf_rele(*db, tag); return (error); } return (0); } void zfs_release_sa_handle(sa_handle_t *hdl, dmu_buf_t *db, void *tag) { sa_handle_destroy(hdl); sa_buf_rele(db, tag); } /* * Given an object number, return its parent object number and whether * or not the object is an extended attribute directory. */ static int zfs_obj_to_pobj(sa_handle_t *hdl, sa_attr_type_t *sa_table, uint64_t *pobjp, int *is_xattrdir) { uint64_t parent; uint64_t pflags; uint64_t mode; sa_bulk_attr_t bulk[3]; int count = 0; int error; SA_ADD_BULK_ATTR(bulk, count, sa_table[ZPL_PARENT], NULL, &parent, sizeof (parent)); SA_ADD_BULK_ATTR(bulk, count, sa_table[ZPL_FLAGS], NULL, &pflags, sizeof (pflags)); SA_ADD_BULK_ATTR(bulk, count, sa_table[ZPL_MODE], NULL, &mode, sizeof (mode)); if ((error = sa_bulk_lookup(hdl, bulk, count)) != 0) return (error); *pobjp = parent; *is_xattrdir = ((pflags & ZFS_XATTR) != 0) && S_ISDIR(mode); return (0); } /* * Given an object number, return some zpl level statistics */ static int zfs_obj_to_stats_impl(sa_handle_t *hdl, sa_attr_type_t *sa_table, zfs_stat_t *sb) { sa_bulk_attr_t bulk[4]; int count = 0; SA_ADD_BULK_ATTR(bulk, count, sa_table[ZPL_MODE], NULL, &sb->zs_mode, sizeof (sb->zs_mode)); SA_ADD_BULK_ATTR(bulk, count, sa_table[ZPL_GEN], NULL, &sb->zs_gen, sizeof (sb->zs_gen)); SA_ADD_BULK_ATTR(bulk, count, sa_table[ZPL_LINKS], NULL, &sb->zs_links, sizeof (sb->zs_links)); SA_ADD_BULK_ATTR(bulk, count, sa_table[ZPL_CTIME], NULL, &sb->zs_ctime, sizeof (sb->zs_ctime)); return (sa_bulk_lookup(hdl, bulk, count)); } static int zfs_obj_to_path_impl(objset_t *osp, uint64_t obj, sa_handle_t *hdl, sa_attr_type_t *sa_table, char *buf, int len) { sa_handle_t *sa_hdl; sa_handle_t *prevhdl = NULL; dmu_buf_t *prevdb = NULL; dmu_buf_t *sa_db = NULL; char *path = buf + len - 1; int error; *path = '\0'; sa_hdl = hdl; for (;;) { uint64_t pobj = 0; char component[MAXNAMELEN + 2]; size_t complen; int is_xattrdir = 0; if (prevdb) zfs_release_sa_handle(prevhdl, prevdb, FTAG); if ((error = zfs_obj_to_pobj(sa_hdl, sa_table, &pobj, &is_xattrdir)) != 0) break; if (pobj == obj) { if (path[0] != '/') *--path = '/'; break; } component[0] = '/'; if (is_xattrdir) { (void) sprintf(component + 1, ""); } else { error = zap_value_search(osp, pobj, obj, ZFS_DIRENT_OBJ(-1ULL), component + 1); if (error != 0) break; } complen = strlen(component); path -= complen; ASSERT(path >= buf); bcopy(component, path, complen); obj = pobj; if (sa_hdl != hdl) { prevhdl = sa_hdl; prevdb = sa_db; } error = zfs_grab_sa_handle(osp, obj, &sa_hdl, &sa_db, FTAG); if (error != 0) { sa_hdl = prevhdl; sa_db = prevdb; break; } } if (sa_hdl != NULL && sa_hdl != hdl) { ASSERT(sa_db != NULL); zfs_release_sa_handle(sa_hdl, sa_db, FTAG); } if (error == 0) (void) memmove(buf, path, buf + len - path); return (error); } int zfs_obj_to_path(objset_t *osp, uint64_t obj, char *buf, int len) { sa_attr_type_t *sa_table; sa_handle_t *hdl; dmu_buf_t *db; int error; error = zfs_sa_setup(osp, &sa_table); if (error != 0) return (error); error = zfs_grab_sa_handle(osp, obj, &hdl, &db, FTAG); if (error != 0) return (error); error = zfs_obj_to_path_impl(osp, obj, hdl, sa_table, buf, len); zfs_release_sa_handle(hdl, db, FTAG); return (error); } int zfs_obj_to_stats(objset_t *osp, uint64_t obj, zfs_stat_t *sb, char *buf, int len) { char *path = buf + len - 1; sa_attr_type_t *sa_table; sa_handle_t *hdl; dmu_buf_t *db; int error; *path = '\0'; error = zfs_sa_setup(osp, &sa_table); if (error != 0) return (error); error = zfs_grab_sa_handle(osp, obj, &hdl, &db, FTAG); if (error != 0) return (error); error = zfs_obj_to_stats_impl(hdl, sa_table, sb); if (error != 0) { zfs_release_sa_handle(hdl, db, FTAG); return (error); } error = zfs_obj_to_path_impl(osp, obj, hdl, sa_table, buf, len); zfs_release_sa_handle(hdl, db, FTAG); return (error); } #if defined(_KERNEL) && defined(HAVE_SPL) EXPORT_SYMBOL(zfs_create_fs); EXPORT_SYMBOL(zfs_obj_to_path); #endif diff --git a/module/zfs/zil.c b/module/zfs/zil.c index 6ee6c9868e69..1d8e7cb4d151 100644 --- a/module/zfs/zil.c +++ b/module/zfs/zil.c @@ -1,2288 +1,2287 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2014 by Delphix. All rights reserved. */ /* Portions Copyright 2010 Robert Milkowski */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * The zfs intent log (ZIL) saves transaction records of system calls * that change the file system in memory with enough information * to be able to replay them. These are stored in memory until * either the DMU transaction group (txg) commits them to the stable pool * and they can be discarded, or they are flushed to the stable log * (also in the pool) due to a fsync, O_DSYNC or other synchronous * requirement. In the event of a panic or power fail then those log * records (transactions) are replayed. * * There is one ZIL per file system. Its on-disk (pool) format consists * of 3 parts: * * - ZIL header * - ZIL blocks * - ZIL records * * A log record holds a system call transaction. Log blocks can * hold many log records and the blocks are chained together. * Each ZIL block contains a block pointer (blkptr_t) to the next * ZIL block in the chain. The ZIL header points to the first * block in the chain. Note there is not a fixed place in the pool * to hold blocks. They are dynamically allocated and freed as * needed from the blocks available. Figure X shows the ZIL structure: */ /* * See zil.h for more information about these fields. */ zil_stats_t zil_stats = { { "zil_commit_count", KSTAT_DATA_UINT64 }, { "zil_commit_writer_count", KSTAT_DATA_UINT64 }, { "zil_itx_count", KSTAT_DATA_UINT64 }, { "zil_itx_indirect_count", KSTAT_DATA_UINT64 }, { "zil_itx_indirect_bytes", KSTAT_DATA_UINT64 }, { "zil_itx_copied_count", KSTAT_DATA_UINT64 }, { "zil_itx_copied_bytes", KSTAT_DATA_UINT64 }, { "zil_itx_needcopy_count", KSTAT_DATA_UINT64 }, { "zil_itx_needcopy_bytes", KSTAT_DATA_UINT64 }, { "zil_itx_metaslab_normal_count", KSTAT_DATA_UINT64 }, { "zil_itx_metaslab_normal_bytes", KSTAT_DATA_UINT64 }, { "zil_itx_metaslab_slog_count", KSTAT_DATA_UINT64 }, { "zil_itx_metaslab_slog_bytes", KSTAT_DATA_UINT64 }, }; static kstat_t *zil_ksp; /* * Disable intent logging replay. This global ZIL switch affects all pools. */ int zil_replay_disable = 0; /* * Tunable parameter for debugging or performance analysis. Setting * zfs_nocacheflush will cause corruption on power loss if a volatile * out-of-order write cache is enabled. */ int zfs_nocacheflush = 0; static kmem_cache_t *zil_lwb_cache; static void zil_async_to_sync(zilog_t *zilog, uint64_t foid); #define LWB_EMPTY(lwb) ((BP_GET_LSIZE(&lwb->lwb_blk) - \ sizeof (zil_chain_t)) == (lwb->lwb_sz - lwb->lwb_nused)) /* * ziltest is by and large an ugly hack, but very useful in * checking replay without tedious work. * When running ziltest we want to keep all itx's and so maintain * a single list in the zl_itxg[] that uses a high txg: ZILTEST_TXG * We subtract TXG_CONCURRENT_STATES to allow for common code. */ #define ZILTEST_TXG (UINT64_MAX - TXG_CONCURRENT_STATES) static int zil_bp_compare(const void *x1, const void *x2) { const dva_t *dva1 = &((zil_bp_node_t *)x1)->zn_dva; const dva_t *dva2 = &((zil_bp_node_t *)x2)->zn_dva; if (DVA_GET_VDEV(dva1) < DVA_GET_VDEV(dva2)) return (-1); if (DVA_GET_VDEV(dva1) > DVA_GET_VDEV(dva2)) return (1); if (DVA_GET_OFFSET(dva1) < DVA_GET_OFFSET(dva2)) return (-1); if (DVA_GET_OFFSET(dva1) > DVA_GET_OFFSET(dva2)) return (1); return (0); } static void zil_bp_tree_init(zilog_t *zilog) { avl_create(&zilog->zl_bp_tree, zil_bp_compare, sizeof (zil_bp_node_t), offsetof(zil_bp_node_t, zn_node)); } static void zil_bp_tree_fini(zilog_t *zilog) { avl_tree_t *t = &zilog->zl_bp_tree; zil_bp_node_t *zn; void *cookie = NULL; while ((zn = avl_destroy_nodes(t, &cookie)) != NULL) kmem_free(zn, sizeof (zil_bp_node_t)); avl_destroy(t); } int zil_bp_tree_add(zilog_t *zilog, const blkptr_t *bp) { avl_tree_t *t = &zilog->zl_bp_tree; const dva_t *dva; zil_bp_node_t *zn; avl_index_t where; if (BP_IS_EMBEDDED(bp)) return (0); dva = BP_IDENTITY(bp); if (avl_find(t, dva, &where) != NULL) return (SET_ERROR(EEXIST)); zn = kmem_alloc(sizeof (zil_bp_node_t), KM_PUSHPAGE); zn->zn_dva = *dva; avl_insert(t, zn, where); return (0); } static zil_header_t * zil_header_in_syncing_context(zilog_t *zilog) { return ((zil_header_t *)zilog->zl_header); } static void zil_init_log_chain(zilog_t *zilog, blkptr_t *bp) { zio_cksum_t *zc = &bp->blk_cksum; zc->zc_word[ZIL_ZC_GUID_0] = spa_get_random(-1ULL); zc->zc_word[ZIL_ZC_GUID_1] = spa_get_random(-1ULL); zc->zc_word[ZIL_ZC_OBJSET] = dmu_objset_id(zilog->zl_os); zc->zc_word[ZIL_ZC_SEQ] = 1ULL; } /* * Read a log block and make sure it's valid. */ static int zil_read_log_block(zilog_t *zilog, const blkptr_t *bp, blkptr_t *nbp, void *dst, char **end) { enum zio_flag zio_flags = ZIO_FLAG_CANFAIL; uint32_t aflags = ARC_WAIT; arc_buf_t *abuf = NULL; zbookmark_phys_t zb; int error; if (zilog->zl_header->zh_claim_txg == 0) zio_flags |= ZIO_FLAG_SPECULATIVE | ZIO_FLAG_SCRUB; if (!(zilog->zl_header->zh_flags & ZIL_CLAIM_LR_SEQ_VALID)) zio_flags |= ZIO_FLAG_SPECULATIVE; SET_BOOKMARK(&zb, bp->blk_cksum.zc_word[ZIL_ZC_OBJSET], ZB_ZIL_OBJECT, ZB_ZIL_LEVEL, bp->blk_cksum.zc_word[ZIL_ZC_SEQ]); error = arc_read(NULL, zilog->zl_spa, bp, arc_getbuf_func, &abuf, ZIO_PRIORITY_SYNC_READ, zio_flags, &aflags, &zb); if (error == 0) { zio_cksum_t cksum = bp->blk_cksum; /* * Validate the checksummed log block. * * Sequence numbers should be... sequential. The checksum * verifier for the next block should be bp's checksum plus 1. * * Also check the log chain linkage and size used. */ cksum.zc_word[ZIL_ZC_SEQ]++; if (BP_GET_CHECKSUM(bp) == ZIO_CHECKSUM_ZILOG2) { zil_chain_t *zilc = abuf->b_data; char *lr = (char *)(zilc + 1); uint64_t len = zilc->zc_nused - sizeof (zil_chain_t); if (bcmp(&cksum, &zilc->zc_next_blk.blk_cksum, sizeof (cksum)) || BP_IS_HOLE(&zilc->zc_next_blk)) { error = SET_ERROR(ECKSUM); } else { bcopy(lr, dst, len); *end = (char *)dst + len; *nbp = zilc->zc_next_blk; } } else { char *lr = abuf->b_data; uint64_t size = BP_GET_LSIZE(bp); zil_chain_t *zilc = (zil_chain_t *)(lr + size) - 1; if (bcmp(&cksum, &zilc->zc_next_blk.blk_cksum, sizeof (cksum)) || BP_IS_HOLE(&zilc->zc_next_blk) || (zilc->zc_nused > (size - sizeof (*zilc)))) { error = SET_ERROR(ECKSUM); } else { bcopy(lr, dst, zilc->zc_nused); *end = (char *)dst + zilc->zc_nused; *nbp = zilc->zc_next_blk; } } VERIFY(arc_buf_remove_ref(abuf, &abuf)); } return (error); } /* * Read a TX_WRITE log data block. */ static int zil_read_log_data(zilog_t *zilog, const lr_write_t *lr, void *wbuf) { enum zio_flag zio_flags = ZIO_FLAG_CANFAIL; const blkptr_t *bp = &lr->lr_blkptr; uint32_t aflags = ARC_WAIT; arc_buf_t *abuf = NULL; zbookmark_phys_t zb; int error; if (BP_IS_HOLE(bp)) { if (wbuf != NULL) bzero(wbuf, MAX(BP_GET_LSIZE(bp), lr->lr_length)); return (0); } if (zilog->zl_header->zh_claim_txg == 0) zio_flags |= ZIO_FLAG_SPECULATIVE | ZIO_FLAG_SCRUB; SET_BOOKMARK(&zb, dmu_objset_id(zilog->zl_os), lr->lr_foid, ZB_ZIL_LEVEL, lr->lr_offset / BP_GET_LSIZE(bp)); error = arc_read(NULL, zilog->zl_spa, bp, arc_getbuf_func, &abuf, ZIO_PRIORITY_SYNC_READ, zio_flags, &aflags, &zb); if (error == 0) { if (wbuf != NULL) bcopy(abuf->b_data, wbuf, arc_buf_size(abuf)); (void) arc_buf_remove_ref(abuf, &abuf); } return (error); } /* * Parse the intent log, and call parse_func for each valid record within. */ int zil_parse(zilog_t *zilog, zil_parse_blk_func_t *parse_blk_func, zil_parse_lr_func_t *parse_lr_func, void *arg, uint64_t txg) { const zil_header_t *zh = zilog->zl_header; boolean_t claimed = !!zh->zh_claim_txg; uint64_t claim_blk_seq = claimed ? zh->zh_claim_blk_seq : UINT64_MAX; uint64_t claim_lr_seq = claimed ? zh->zh_claim_lr_seq : UINT64_MAX; uint64_t max_blk_seq = 0; uint64_t max_lr_seq = 0; uint64_t blk_count = 0; uint64_t lr_count = 0; blkptr_t blk, next_blk; char *lrbuf, *lrp; int error = 0; bzero(&next_blk, sizeof (blkptr_t)); /* * Old logs didn't record the maximum zh_claim_lr_seq. */ if (!(zh->zh_flags & ZIL_CLAIM_LR_SEQ_VALID)) claim_lr_seq = UINT64_MAX; /* * Starting at the block pointed to by zh_log we read the log chain. * For each block in the chain we strongly check that block to * ensure its validity. We stop when an invalid block is found. * For each block pointer in the chain we call parse_blk_func(). * For each record in each valid block we call parse_lr_func(). * If the log has been claimed, stop if we encounter a sequence * number greater than the highest claimed sequence number. */ lrbuf = zio_buf_alloc(SPA_MAXBLOCKSIZE); zil_bp_tree_init(zilog); for (blk = zh->zh_log; !BP_IS_HOLE(&blk); blk = next_blk) { uint64_t blk_seq = blk.blk_cksum.zc_word[ZIL_ZC_SEQ]; int reclen; char *end = NULL; if (blk_seq > claim_blk_seq) break; if ((error = parse_blk_func(zilog, &blk, arg, txg)) != 0) break; ASSERT3U(max_blk_seq, <, blk_seq); max_blk_seq = blk_seq; blk_count++; if (max_lr_seq == claim_lr_seq && max_blk_seq == claim_blk_seq) break; error = zil_read_log_block(zilog, &blk, &next_blk, lrbuf, &end); if (error != 0) break; for (lrp = lrbuf; lrp < end; lrp += reclen) { lr_t *lr = (lr_t *)lrp; reclen = lr->lrc_reclen; ASSERT3U(reclen, >=, sizeof (lr_t)); if (lr->lrc_seq > claim_lr_seq) goto done; if ((error = parse_lr_func(zilog, lr, arg, txg)) != 0) goto done; ASSERT3U(max_lr_seq, <, lr->lrc_seq); max_lr_seq = lr->lrc_seq; lr_count++; } } done: zilog->zl_parse_error = error; zilog->zl_parse_blk_seq = max_blk_seq; zilog->zl_parse_lr_seq = max_lr_seq; zilog->zl_parse_blk_count = blk_count; zilog->zl_parse_lr_count = lr_count; ASSERT(!claimed || !(zh->zh_flags & ZIL_CLAIM_LR_SEQ_VALID) || (max_blk_seq == claim_blk_seq && max_lr_seq == claim_lr_seq)); zil_bp_tree_fini(zilog); zio_buf_free(lrbuf, SPA_MAXBLOCKSIZE); return (error); } static int zil_claim_log_block(zilog_t *zilog, blkptr_t *bp, void *tx, uint64_t first_txg) { /* * Claim log block if not already committed and not already claimed. * If tx == NULL, just verify that the block is claimable. */ if (BP_IS_HOLE(bp) || bp->blk_birth < first_txg || zil_bp_tree_add(zilog, bp) != 0) return (0); return (zio_wait(zio_claim(NULL, zilog->zl_spa, tx == NULL ? 0 : first_txg, bp, spa_claim_notify, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE | ZIO_FLAG_SCRUB))); } static int zil_claim_log_record(zilog_t *zilog, lr_t *lrc, void *tx, uint64_t first_txg) { lr_write_t *lr = (lr_write_t *)lrc; int error; if (lrc->lrc_txtype != TX_WRITE) return (0); /* * If the block is not readable, don't claim it. This can happen * in normal operation when a log block is written to disk before * some of the dmu_sync() blocks it points to. In this case, the * transaction cannot have been committed to anyone (we would have * waited for all writes to be stable first), so it is semantically * correct to declare this the end of the log. */ if (lr->lr_blkptr.blk_birth >= first_txg && (error = zil_read_log_data(zilog, lr, NULL)) != 0) return (error); return (zil_claim_log_block(zilog, &lr->lr_blkptr, tx, first_txg)); } /* ARGSUSED */ static int zil_free_log_block(zilog_t *zilog, blkptr_t *bp, void *tx, uint64_t claim_txg) { zio_free_zil(zilog->zl_spa, dmu_tx_get_txg(tx), bp); return (0); } static int zil_free_log_record(zilog_t *zilog, lr_t *lrc, void *tx, uint64_t claim_txg) { lr_write_t *lr = (lr_write_t *)lrc; blkptr_t *bp = &lr->lr_blkptr; /* * If we previously claimed it, we need to free it. */ if (claim_txg != 0 && lrc->lrc_txtype == TX_WRITE && bp->blk_birth >= claim_txg && zil_bp_tree_add(zilog, bp) == 0 && !BP_IS_HOLE(bp)) zio_free(zilog->zl_spa, dmu_tx_get_txg(tx), bp); return (0); } static lwb_t * zil_alloc_lwb(zilog_t *zilog, blkptr_t *bp, uint64_t txg, boolean_t fastwrite) { lwb_t *lwb; lwb = kmem_cache_alloc(zil_lwb_cache, KM_PUSHPAGE); lwb->lwb_zilog = zilog; lwb->lwb_blk = *bp; lwb->lwb_fastwrite = fastwrite; lwb->lwb_buf = zio_buf_alloc(BP_GET_LSIZE(bp)); lwb->lwb_max_txg = txg; lwb->lwb_zio = NULL; lwb->lwb_tx = NULL; if (BP_GET_CHECKSUM(bp) == ZIO_CHECKSUM_ZILOG2) { lwb->lwb_nused = sizeof (zil_chain_t); lwb->lwb_sz = BP_GET_LSIZE(bp); } else { lwb->lwb_nused = 0; lwb->lwb_sz = BP_GET_LSIZE(bp) - sizeof (zil_chain_t); } mutex_enter(&zilog->zl_lock); list_insert_tail(&zilog->zl_lwb_list, lwb); mutex_exit(&zilog->zl_lock); return (lwb); } /* * Called when we create in-memory log transactions so that we know * to cleanup the itxs at the end of spa_sync(). */ void zilog_dirty(zilog_t *zilog, uint64_t txg) { dsl_pool_t *dp = zilog->zl_dmu_pool; dsl_dataset_t *ds = dmu_objset_ds(zilog->zl_os); if (dsl_dataset_is_snapshot(ds)) panic("dirtying snapshot!"); if (txg_list_add(&dp->dp_dirty_zilogs, zilog, txg)) { /* up the hold count until we can be written out */ dmu_buf_add_ref(ds->ds_dbuf, zilog); } } boolean_t zilog_is_dirty(zilog_t *zilog) { dsl_pool_t *dp = zilog->zl_dmu_pool; int t; for (t = 0; t < TXG_SIZE; t++) { if (txg_list_member(&dp->dp_dirty_zilogs, zilog, t)) return (B_TRUE); } return (B_FALSE); } /* * Create an on-disk intent log. */ static lwb_t * zil_create(zilog_t *zilog) { const zil_header_t *zh = zilog->zl_header; lwb_t *lwb = NULL; uint64_t txg = 0; dmu_tx_t *tx = NULL; blkptr_t blk; int error = 0; boolean_t fastwrite = FALSE; /* * Wait for any previous destroy to complete. */ txg_wait_synced(zilog->zl_dmu_pool, zilog->zl_destroy_txg); ASSERT(zh->zh_claim_txg == 0); ASSERT(zh->zh_replay_seq == 0); blk = zh->zh_log; /* * Allocate an initial log block if: * - there isn't one already * - the existing block is the wrong endianess */ if (BP_IS_HOLE(&blk) || BP_SHOULD_BYTESWAP(&blk)) { tx = dmu_tx_create(zilog->zl_os); VERIFY(dmu_tx_assign(tx, TXG_WAIT) == 0); dsl_dataset_dirty(dmu_objset_ds(zilog->zl_os), tx); txg = dmu_tx_get_txg(tx); if (!BP_IS_HOLE(&blk)) { zio_free_zil(zilog->zl_spa, txg, &blk); BP_ZERO(&blk); } error = zio_alloc_zil(zilog->zl_spa, txg, &blk, ZIL_MIN_BLKSZ, B_TRUE); fastwrite = TRUE; if (error == 0) zil_init_log_chain(zilog, &blk); } /* * Allocate a log write buffer (lwb) for the first log block. */ if (error == 0) lwb = zil_alloc_lwb(zilog, &blk, txg, fastwrite); /* * If we just allocated the first log block, commit our transaction * and wait for zil_sync() to stuff the block poiner into zh_log. * (zh is part of the MOS, so we cannot modify it in open context.) */ if (tx != NULL) { dmu_tx_commit(tx); txg_wait_synced(zilog->zl_dmu_pool, txg); } ASSERT(bcmp(&blk, &zh->zh_log, sizeof (blk)) == 0); return (lwb); } /* * In one tx, free all log blocks and clear the log header. * If keep_first is set, then we're replaying a log with no content. * We want to keep the first block, however, so that the first * synchronous transaction doesn't require a txg_wait_synced() * in zil_create(). We don't need to txg_wait_synced() here either * when keep_first is set, because both zil_create() and zil_destroy() * will wait for any in-progress destroys to complete. */ void zil_destroy(zilog_t *zilog, boolean_t keep_first) { const zil_header_t *zh = zilog->zl_header; lwb_t *lwb; dmu_tx_t *tx; uint64_t txg; /* * Wait for any previous destroy to complete. */ txg_wait_synced(zilog->zl_dmu_pool, zilog->zl_destroy_txg); zilog->zl_old_header = *zh; /* debugging aid */ if (BP_IS_HOLE(&zh->zh_log)) return; tx = dmu_tx_create(zilog->zl_os); VERIFY(dmu_tx_assign(tx, TXG_WAIT) == 0); dsl_dataset_dirty(dmu_objset_ds(zilog->zl_os), tx); txg = dmu_tx_get_txg(tx); mutex_enter(&zilog->zl_lock); ASSERT3U(zilog->zl_destroy_txg, <, txg); zilog->zl_destroy_txg = txg; zilog->zl_keep_first = keep_first; if (!list_is_empty(&zilog->zl_lwb_list)) { ASSERT(zh->zh_claim_txg == 0); VERIFY(!keep_first); while ((lwb = list_head(&zilog->zl_lwb_list)) != NULL) { ASSERT(lwb->lwb_zio == NULL); if (lwb->lwb_fastwrite) metaslab_fastwrite_unmark(zilog->zl_spa, &lwb->lwb_blk); list_remove(&zilog->zl_lwb_list, lwb); if (lwb->lwb_buf != NULL) zio_buf_free(lwb->lwb_buf, lwb->lwb_sz); zio_free_zil(zilog->zl_spa, txg, &lwb->lwb_blk); kmem_cache_free(zil_lwb_cache, lwb); } } else if (!keep_first) { zil_destroy_sync(zilog, tx); } mutex_exit(&zilog->zl_lock); dmu_tx_commit(tx); } void zil_destroy_sync(zilog_t *zilog, dmu_tx_t *tx) { ASSERT(list_is_empty(&zilog->zl_lwb_list)); (void) zil_parse(zilog, zil_free_log_block, zil_free_log_record, tx, zilog->zl_header->zh_claim_txg); } int zil_claim(const char *osname, void *txarg) { dmu_tx_t *tx = txarg; uint64_t first_txg = dmu_tx_get_txg(tx); zilog_t *zilog; zil_header_t *zh; objset_t *os; int error; error = dmu_objset_own(osname, DMU_OST_ANY, B_FALSE, FTAG, &os); if (error != 0) { /* * EBUSY indicates that the objset is inconsistent, in which * case it can not have a ZIL. */ if (error != EBUSY) { cmn_err(CE_WARN, "can't open objset for %s, error %u", osname, error); } return (0); } zilog = dmu_objset_zil(os); zh = zil_header_in_syncing_context(zilog); if (spa_get_log_state(zilog->zl_spa) == SPA_LOG_CLEAR) { if (!BP_IS_HOLE(&zh->zh_log)) zio_free_zil(zilog->zl_spa, first_txg, &zh->zh_log); BP_ZERO(&zh->zh_log); dsl_dataset_dirty(dmu_objset_ds(os), tx); dmu_objset_disown(os, FTAG); return (0); } /* * Claim all log blocks if we haven't already done so, and remember * the highest claimed sequence number. This ensures that if we can * read only part of the log now (e.g. due to a missing device), * but we can read the entire log later, we will not try to replay * or destroy beyond the last block we successfully claimed. */ ASSERT3U(zh->zh_claim_txg, <=, first_txg); if (zh->zh_claim_txg == 0 && !BP_IS_HOLE(&zh->zh_log)) { (void) zil_parse(zilog, zil_claim_log_block, zil_claim_log_record, tx, first_txg); zh->zh_claim_txg = first_txg; zh->zh_claim_blk_seq = zilog->zl_parse_blk_seq; zh->zh_claim_lr_seq = zilog->zl_parse_lr_seq; if (zilog->zl_parse_lr_count || zilog->zl_parse_blk_count > 1) zh->zh_flags |= ZIL_REPLAY_NEEDED; zh->zh_flags |= ZIL_CLAIM_LR_SEQ_VALID; dsl_dataset_dirty(dmu_objset_ds(os), tx); } ASSERT3U(first_txg, ==, (spa_last_synced_txg(zilog->zl_spa) + 1)); dmu_objset_disown(os, FTAG); return (0); } /* * Check the log by walking the log chain. * Checksum errors are ok as they indicate the end of the chain. * Any other error (no device or read failure) returns an error. */ int zil_check_log_chain(const char *osname, void *tx) { zilog_t *zilog; objset_t *os; blkptr_t *bp; int error; ASSERT(tx == NULL); error = dmu_objset_hold(osname, FTAG, &os); if (error != 0) { cmn_err(CE_WARN, "can't open objset for %s", osname); return (0); } zilog = dmu_objset_zil(os); bp = (blkptr_t *)&zilog->zl_header->zh_log; /* * Check the first block and determine if it's on a log device * which may have been removed or faulted prior to loading this * pool. If so, there's no point in checking the rest of the log * as its content should have already been synced to the pool. */ if (!BP_IS_HOLE(bp)) { vdev_t *vd; boolean_t valid = B_TRUE; spa_config_enter(os->os_spa, SCL_STATE, FTAG, RW_READER); vd = vdev_lookup_top(os->os_spa, DVA_GET_VDEV(&bp->blk_dva[0])); if (vd->vdev_islog && vdev_is_dead(vd)) valid = vdev_log_state_valid(vd); spa_config_exit(os->os_spa, SCL_STATE, FTAG); if (!valid) { dmu_objset_rele(os, FTAG); return (0); } } /* * Because tx == NULL, zil_claim_log_block() will not actually claim * any blocks, but just determine whether it is possible to do so. * In addition to checking the log chain, zil_claim_log_block() * will invoke zio_claim() with a done func of spa_claim_notify(), * which will update spa_max_claim_txg. See spa_load() for details. */ error = zil_parse(zilog, zil_claim_log_block, zil_claim_log_record, tx, zilog->zl_header->zh_claim_txg ? -1ULL : spa_first_txg(os->os_spa)); dmu_objset_rele(os, FTAG); return ((error == ECKSUM || error == ENOENT) ? 0 : error); } static int zil_vdev_compare(const void *x1, const void *x2) { const uint64_t v1 = ((zil_vdev_node_t *)x1)->zv_vdev; const uint64_t v2 = ((zil_vdev_node_t *)x2)->zv_vdev; if (v1 < v2) return (-1); if (v1 > v2) return (1); return (0); } void zil_add_block(zilog_t *zilog, const blkptr_t *bp) { avl_tree_t *t = &zilog->zl_vdev_tree; avl_index_t where; zil_vdev_node_t *zv, zvsearch; int ndvas = BP_GET_NDVAS(bp); int i; if (zfs_nocacheflush) return; ASSERT(zilog->zl_writer); /* * Even though we're zl_writer, we still need a lock because the * zl_get_data() callbacks may have dmu_sync() done callbacks * that will run concurrently. */ mutex_enter(&zilog->zl_vdev_lock); for (i = 0; i < ndvas; i++) { zvsearch.zv_vdev = DVA_GET_VDEV(&bp->blk_dva[i]); if (avl_find(t, &zvsearch, &where) == NULL) { zv = kmem_alloc(sizeof (*zv), KM_PUSHPAGE); zv->zv_vdev = zvsearch.zv_vdev; avl_insert(t, zv, where); } } mutex_exit(&zilog->zl_vdev_lock); } static void zil_flush_vdevs(zilog_t *zilog) { spa_t *spa = zilog->zl_spa; avl_tree_t *t = &zilog->zl_vdev_tree; void *cookie = NULL; zil_vdev_node_t *zv; zio_t *zio; ASSERT(zilog->zl_writer); /* * We don't need zl_vdev_lock here because we're the zl_writer, * and all zl_get_data() callbacks are done. */ if (avl_numnodes(t) == 0) return; spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); zio = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL); while ((zv = avl_destroy_nodes(t, &cookie)) != NULL) { vdev_t *vd = vdev_lookup_top(spa, zv->zv_vdev); if (vd != NULL) zio_flush(zio, vd); kmem_free(zv, sizeof (*zv)); } /* * Wait for all the flushes to complete. Not all devices actually * support the DKIOCFLUSHWRITECACHE ioctl, so it's OK if it fails. */ (void) zio_wait(zio); spa_config_exit(spa, SCL_STATE, FTAG); } /* * Function called when a log block write completes */ static void zil_lwb_write_done(zio_t *zio) { lwb_t *lwb = zio->io_private; zilog_t *zilog = lwb->lwb_zilog; dmu_tx_t *tx = lwb->lwb_tx; ASSERT(BP_GET_COMPRESS(zio->io_bp) == ZIO_COMPRESS_OFF); ASSERT(BP_GET_TYPE(zio->io_bp) == DMU_OT_INTENT_LOG); ASSERT(BP_GET_LEVEL(zio->io_bp) == 0); ASSERT(BP_GET_BYTEORDER(zio->io_bp) == ZFS_HOST_BYTEORDER); ASSERT(!BP_IS_GANG(zio->io_bp)); ASSERT(!BP_IS_HOLE(zio->io_bp)); ASSERT(BP_GET_FILL(zio->io_bp) == 0); /* * Ensure the lwb buffer pointer is cleared before releasing * the txg. If we have had an allocation failure and * the txg is waiting to sync then we want want zil_sync() * to remove the lwb so that it's not picked up as the next new * one in zil_commit_writer(). zil_sync() will only remove * the lwb if lwb_buf is null. */ zio_buf_free(lwb->lwb_buf, lwb->lwb_sz); mutex_enter(&zilog->zl_lock); lwb->lwb_zio = NULL; lwb->lwb_fastwrite = FALSE; lwb->lwb_buf = NULL; lwb->lwb_tx = NULL; mutex_exit(&zilog->zl_lock); /* * Now that we've written this log block, we have a stable pointer * to the next block in the chain, so it's OK to let the txg in * which we allocated the next block sync. */ dmu_tx_commit(tx); } /* * Initialize the io for a log block. */ static void zil_lwb_write_init(zilog_t *zilog, lwb_t *lwb) { zbookmark_phys_t zb; SET_BOOKMARK(&zb, lwb->lwb_blk.blk_cksum.zc_word[ZIL_ZC_OBJSET], ZB_ZIL_OBJECT, ZB_ZIL_LEVEL, lwb->lwb_blk.blk_cksum.zc_word[ZIL_ZC_SEQ]); if (zilog->zl_root_zio == NULL) { zilog->zl_root_zio = zio_root(zilog->zl_spa, NULL, NULL, ZIO_FLAG_CANFAIL); } /* Lock so zil_sync() doesn't fastwrite_unmark after zio is created */ mutex_enter(&zilog->zl_lock); if (lwb->lwb_zio == NULL) { if (!lwb->lwb_fastwrite) { metaslab_fastwrite_mark(zilog->zl_spa, &lwb->lwb_blk); lwb->lwb_fastwrite = 1; } lwb->lwb_zio = zio_rewrite(zilog->zl_root_zio, zilog->zl_spa, 0, &lwb->lwb_blk, lwb->lwb_buf, BP_GET_LSIZE(&lwb->lwb_blk), zil_lwb_write_done, lwb, ZIO_PRIORITY_SYNC_WRITE, ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_PROPAGATE | ZIO_FLAG_FASTWRITE, &zb); } mutex_exit(&zilog->zl_lock); } /* * Define a limited set of intent log block sizes. * * These must be a multiple of 4KB. Note only the amount used (again * aligned to 4KB) actually gets written. However, we can't always just * allocate SPA_MAXBLOCKSIZE as the slog space could be exhausted. */ uint64_t zil_block_buckets[] = { 4096, /* non TX_WRITE */ 8192+4096, /* data base */ 32*1024 + 4096, /* NFS writes */ UINT64_MAX }; /* * Use the slog as long as the current commit size is less than the * limit or the total list size is less than 2X the limit. Limit * checking is disabled by setting zil_slog_limit to UINT64_MAX. */ unsigned long zil_slog_limit = 1024 * 1024; #define USE_SLOG(zilog) (((zilog)->zl_cur_used < zil_slog_limit) || \ ((zilog)->zl_itx_list_sz < (zil_slog_limit << 1))) /* * Start a log block write and advance to the next log block. * Calls are serialized. */ static lwb_t * zil_lwb_write_start(zilog_t *zilog, lwb_t *lwb) { lwb_t *nlwb = NULL; zil_chain_t *zilc; spa_t *spa = zilog->zl_spa; blkptr_t *bp; dmu_tx_t *tx; uint64_t txg; uint64_t zil_blksz, wsz; int i, error; boolean_t use_slog; if (BP_GET_CHECKSUM(&lwb->lwb_blk) == ZIO_CHECKSUM_ZILOG2) { zilc = (zil_chain_t *)lwb->lwb_buf; bp = &zilc->zc_next_blk; } else { zilc = (zil_chain_t *)(lwb->lwb_buf + lwb->lwb_sz); bp = &zilc->zc_next_blk; } ASSERT(lwb->lwb_nused <= lwb->lwb_sz); /* * Allocate the next block and save its address in this block * before writing it in order to establish the log chain. * Note that if the allocation of nlwb synced before we wrote * the block that points at it (lwb), we'd leak it if we crashed. * Therefore, we don't do dmu_tx_commit() until zil_lwb_write_done(). * We dirty the dataset to ensure that zil_sync() will be called * to clean up in the event of allocation failure or I/O failure. */ tx = dmu_tx_create(zilog->zl_os); VERIFY(dmu_tx_assign(tx, TXG_WAIT) == 0); dsl_dataset_dirty(dmu_objset_ds(zilog->zl_os), tx); txg = dmu_tx_get_txg(tx); lwb->lwb_tx = tx; /* * Log blocks are pre-allocated. Here we select the size of the next * block, based on size used in the last block. * - first find the smallest bucket that will fit the block from a * limited set of block sizes. This is because it's faster to write * blocks allocated from the same metaslab as they are adjacent or * close. * - next find the maximum from the new suggested size and an array of * previous sizes. This lessens a picket fence effect of wrongly * guesssing the size if we have a stream of say 2k, 64k, 2k, 64k * requests. * * Note we only write what is used, but we can't just allocate * the maximum block size because we can exhaust the available * pool log space. */ zil_blksz = zilog->zl_cur_used + sizeof (zil_chain_t); for (i = 0; zil_blksz > zil_block_buckets[i]; i++) continue; zil_blksz = zil_block_buckets[i]; if (zil_blksz == UINT64_MAX) zil_blksz = SPA_MAXBLOCKSIZE; zilog->zl_prev_blks[zilog->zl_prev_rotor] = zil_blksz; for (i = 0; i < ZIL_PREV_BLKS; i++) zil_blksz = MAX(zil_blksz, zilog->zl_prev_blks[i]); zilog->zl_prev_rotor = (zilog->zl_prev_rotor + 1) & (ZIL_PREV_BLKS - 1); BP_ZERO(bp); use_slog = USE_SLOG(zilog); error = zio_alloc_zil(spa, txg, bp, zil_blksz, USE_SLOG(zilog)); if (use_slog) { ZIL_STAT_BUMP(zil_itx_metaslab_slog_count); ZIL_STAT_INCR(zil_itx_metaslab_slog_bytes, lwb->lwb_nused); } else { ZIL_STAT_BUMP(zil_itx_metaslab_normal_count); ZIL_STAT_INCR(zil_itx_metaslab_normal_bytes, lwb->lwb_nused); } if (error == 0) { ASSERT3U(bp->blk_birth, ==, txg); bp->blk_cksum = lwb->lwb_blk.blk_cksum; bp->blk_cksum.zc_word[ZIL_ZC_SEQ]++; /* * Allocate a new log write buffer (lwb). */ nlwb = zil_alloc_lwb(zilog, bp, txg, TRUE); /* Record the block for later vdev flushing */ zil_add_block(zilog, &lwb->lwb_blk); } if (BP_GET_CHECKSUM(&lwb->lwb_blk) == ZIO_CHECKSUM_ZILOG2) { /* For Slim ZIL only write what is used. */ wsz = P2ROUNDUP_TYPED(lwb->lwb_nused, ZIL_MIN_BLKSZ, uint64_t); ASSERT3U(wsz, <=, lwb->lwb_sz); zio_shrink(lwb->lwb_zio, wsz); } else { wsz = lwb->lwb_sz; } zilc->zc_pad = 0; zilc->zc_nused = lwb->lwb_nused; zilc->zc_eck.zec_cksum = lwb->lwb_blk.blk_cksum; /* * clear unused data for security */ bzero(lwb->lwb_buf + lwb->lwb_nused, wsz - lwb->lwb_nused); zio_nowait(lwb->lwb_zio); /* Kick off the write for the old log block */ /* * If there was an allocation failure then nlwb will be null which * forces a txg_wait_synced(). */ return (nlwb); } static lwb_t * zil_lwb_commit(zilog_t *zilog, itx_t *itx, lwb_t *lwb) { lr_t *lrc = &itx->itx_lr; /* common log record */ lr_write_t *lrw = (lr_write_t *)lrc; char *lr_buf; uint64_t txg = lrc->lrc_txg; uint64_t reclen = lrc->lrc_reclen; uint64_t dlen = 0; if (lwb == NULL) return (NULL); ASSERT(lwb->lwb_buf != NULL); ASSERT(zilog_is_dirty(zilog) || spa_freeze_txg(zilog->zl_spa) != UINT64_MAX); if (lrc->lrc_txtype == TX_WRITE && itx->itx_wr_state == WR_NEED_COPY) dlen = P2ROUNDUP_TYPED( lrw->lr_length, sizeof (uint64_t), uint64_t); zilog->zl_cur_used += (reclen + dlen); zil_lwb_write_init(zilog, lwb); /* * If this record won't fit in the current log block, start a new one. */ if (lwb->lwb_nused + reclen + dlen > lwb->lwb_sz) { lwb = zil_lwb_write_start(zilog, lwb); if (lwb == NULL) return (NULL); zil_lwb_write_init(zilog, lwb); ASSERT(LWB_EMPTY(lwb)); if (lwb->lwb_nused + reclen + dlen > lwb->lwb_sz) { txg_wait_synced(zilog->zl_dmu_pool, txg); return (lwb); } } lr_buf = lwb->lwb_buf + lwb->lwb_nused; bcopy(lrc, lr_buf, reclen); lrc = (lr_t *)lr_buf; lrw = (lr_write_t *)lrc; ZIL_STAT_BUMP(zil_itx_count); /* * If it's a write, fetch the data or get its blkptr as appropriate. */ if (lrc->lrc_txtype == TX_WRITE) { if (txg > spa_freeze_txg(zilog->zl_spa)) txg_wait_synced(zilog->zl_dmu_pool, txg); if (itx->itx_wr_state == WR_COPIED) { ZIL_STAT_BUMP(zil_itx_copied_count); ZIL_STAT_INCR(zil_itx_copied_bytes, lrw->lr_length); } else { char *dbuf; int error; if (dlen) { ASSERT(itx->itx_wr_state == WR_NEED_COPY); dbuf = lr_buf + reclen; lrw->lr_common.lrc_reclen += dlen; ZIL_STAT_BUMP(zil_itx_needcopy_count); ZIL_STAT_INCR(zil_itx_needcopy_bytes, lrw->lr_length); } else { ASSERT(itx->itx_wr_state == WR_INDIRECT); dbuf = NULL; ZIL_STAT_BUMP(zil_itx_indirect_count); ZIL_STAT_INCR(zil_itx_indirect_bytes, lrw->lr_length); } error = zilog->zl_get_data( itx->itx_private, lrw, dbuf, lwb->lwb_zio); if (error == EIO) { txg_wait_synced(zilog->zl_dmu_pool, txg); return (lwb); } if (error != 0) { ASSERT(error == ENOENT || error == EEXIST || error == EALREADY); return (lwb); } } } /* * We're actually making an entry, so update lrc_seq to be the * log record sequence number. Note that this is generally not * equal to the itx sequence number because not all transactions * are synchronous, and sometimes spa_sync() gets there first. */ lrc->lrc_seq = ++zilog->zl_lr_seq; /* we are single threaded */ lwb->lwb_nused += reclen + dlen; lwb->lwb_max_txg = MAX(lwb->lwb_max_txg, txg); ASSERT3U(lwb->lwb_nused, <=, lwb->lwb_sz); ASSERT0(P2PHASE(lwb->lwb_nused, sizeof (uint64_t))); return (lwb); } itx_t * zil_itx_create(uint64_t txtype, size_t lrsize) { itx_t *itx; lrsize = P2ROUNDUP_TYPED(lrsize, sizeof (uint64_t), size_t); - itx = kmem_alloc(offsetof(itx_t, itx_lr) + lrsize, - KM_PUSHPAGE | KM_NODEBUG); + itx = vmem_alloc(offsetof(itx_t, itx_lr) + lrsize, KM_PUSHPAGE); itx->itx_lr.lrc_txtype = txtype; itx->itx_lr.lrc_reclen = lrsize; itx->itx_sod = lrsize; /* if write & WR_NEED_COPY will be increased */ itx->itx_lr.lrc_seq = 0; /* defensive */ itx->itx_sync = B_TRUE; /* default is synchronous */ itx->itx_callback = NULL; itx->itx_callback_data = NULL; return (itx); } void zil_itx_destroy(itx_t *itx) { - kmem_free(itx, offsetof(itx_t, itx_lr) + itx->itx_lr.lrc_reclen); + vmem_free(itx, offsetof(itx_t, itx_lr) + itx->itx_lr.lrc_reclen); } /* * Free up the sync and async itxs. The itxs_t has already been detached * so no locks are needed. */ static void zil_itxg_clean(itxs_t *itxs) { itx_t *itx; list_t *list; avl_tree_t *t; void *cookie; itx_async_node_t *ian; list = &itxs->i_sync_list; while ((itx = list_head(list)) != NULL) { if (itx->itx_callback != NULL) itx->itx_callback(itx->itx_callback_data); list_remove(list, itx); kmem_free(itx, offsetof(itx_t, itx_lr) + itx->itx_lr.lrc_reclen); } cookie = NULL; t = &itxs->i_async_tree; while ((ian = avl_destroy_nodes(t, &cookie)) != NULL) { list = &ian->ia_list; while ((itx = list_head(list)) != NULL) { if (itx->itx_callback != NULL) itx->itx_callback(itx->itx_callback_data); list_remove(list, itx); kmem_free(itx, offsetof(itx_t, itx_lr) + itx->itx_lr.lrc_reclen); } list_destroy(list); kmem_free(ian, sizeof (itx_async_node_t)); } avl_destroy(t); kmem_free(itxs, sizeof (itxs_t)); } static int zil_aitx_compare(const void *x1, const void *x2) { const uint64_t o1 = ((itx_async_node_t *)x1)->ia_foid; const uint64_t o2 = ((itx_async_node_t *)x2)->ia_foid; if (o1 < o2) return (-1); if (o1 > o2) return (1); return (0); } /* * Remove all async itx with the given oid. */ static void zil_remove_async(zilog_t *zilog, uint64_t oid) { uint64_t otxg, txg; itx_async_node_t *ian; avl_tree_t *t; avl_index_t where; list_t clean_list; itx_t *itx; ASSERT(oid != 0); list_create(&clean_list, sizeof (itx_t), offsetof(itx_t, itx_node)); if (spa_freeze_txg(zilog->zl_spa) != UINT64_MAX) /* ziltest support */ otxg = ZILTEST_TXG; else otxg = spa_last_synced_txg(zilog->zl_spa) + 1; for (txg = otxg; txg < (otxg + TXG_CONCURRENT_STATES); txg++) { itxg_t *itxg = &zilog->zl_itxg[txg & TXG_MASK]; mutex_enter(&itxg->itxg_lock); if (itxg->itxg_txg != txg) { mutex_exit(&itxg->itxg_lock); continue; } /* * Locate the object node and append its list. */ t = &itxg->itxg_itxs->i_async_tree; ian = avl_find(t, &oid, &where); if (ian != NULL) list_move_tail(&clean_list, &ian->ia_list); mutex_exit(&itxg->itxg_lock); } while ((itx = list_head(&clean_list)) != NULL) { if (itx->itx_callback != NULL) itx->itx_callback(itx->itx_callback_data); list_remove(&clean_list, itx); kmem_free(itx, offsetof(itx_t, itx_lr) + itx->itx_lr.lrc_reclen); } list_destroy(&clean_list); } void zil_itx_assign(zilog_t *zilog, itx_t *itx, dmu_tx_t *tx) { uint64_t txg; itxg_t *itxg; itxs_t *itxs, *clean = NULL; /* * Object ids can be re-instantiated in the next txg so * remove any async transactions to avoid future leaks. * This can happen if a fsync occurs on the re-instantiated * object for a WR_INDIRECT or WR_NEED_COPY write, which gets * the new file data and flushes a write record for the old object. */ if ((itx->itx_lr.lrc_txtype & ~TX_CI) == TX_REMOVE) zil_remove_async(zilog, itx->itx_oid); /* * Ensure the data of a renamed file is committed before the rename. */ if ((itx->itx_lr.lrc_txtype & ~TX_CI) == TX_RENAME) zil_async_to_sync(zilog, itx->itx_oid); if (spa_freeze_txg(zilog->zl_spa) != UINT64_MAX) txg = ZILTEST_TXG; else txg = dmu_tx_get_txg(tx); itxg = &zilog->zl_itxg[txg & TXG_MASK]; mutex_enter(&itxg->itxg_lock); itxs = itxg->itxg_itxs; if (itxg->itxg_txg != txg) { if (itxs != NULL) { /* * The zil_clean callback hasn't got around to cleaning * this itxg. Save the itxs for release below. * This should be rare. */ atomic_add_64(&zilog->zl_itx_list_sz, -itxg->itxg_sod); itxg->itxg_sod = 0; clean = itxg->itxg_itxs; } ASSERT(itxg->itxg_sod == 0); itxg->itxg_txg = txg; itxs = itxg->itxg_itxs = kmem_zalloc(sizeof (itxs_t), KM_PUSHPAGE); list_create(&itxs->i_sync_list, sizeof (itx_t), offsetof(itx_t, itx_node)); avl_create(&itxs->i_async_tree, zil_aitx_compare, sizeof (itx_async_node_t), offsetof(itx_async_node_t, ia_node)); } if (itx->itx_sync) { list_insert_tail(&itxs->i_sync_list, itx); atomic_add_64(&zilog->zl_itx_list_sz, itx->itx_sod); itxg->itxg_sod += itx->itx_sod; } else { avl_tree_t *t = &itxs->i_async_tree; uint64_t foid = ((lr_ooo_t *)&itx->itx_lr)->lr_foid; itx_async_node_t *ian; avl_index_t where; ian = avl_find(t, &foid, &where); if (ian == NULL) { ian = kmem_alloc(sizeof (itx_async_node_t), KM_PUSHPAGE); list_create(&ian->ia_list, sizeof (itx_t), offsetof(itx_t, itx_node)); ian->ia_foid = foid; avl_insert(t, ian, where); } list_insert_tail(&ian->ia_list, itx); } itx->itx_lr.lrc_txg = dmu_tx_get_txg(tx); zilog_dirty(zilog, txg); mutex_exit(&itxg->itxg_lock); /* Release the old itxs now we've dropped the lock */ if (clean != NULL) zil_itxg_clean(clean); } /* * If there are any in-memory intent log transactions which have now been * synced then start up a taskq to free them. We should only do this after we * have written out the uberblocks (i.e. txg has been comitted) so that * don't inadvertently clean out in-memory log records that would be required * by zil_commit(). */ void zil_clean(zilog_t *zilog, uint64_t synced_txg) { itxg_t *itxg = &zilog->zl_itxg[synced_txg & TXG_MASK]; itxs_t *clean_me; mutex_enter(&itxg->itxg_lock); if (itxg->itxg_itxs == NULL || itxg->itxg_txg == ZILTEST_TXG) { mutex_exit(&itxg->itxg_lock); return; } ASSERT3U(itxg->itxg_txg, <=, synced_txg); ASSERT(itxg->itxg_txg != 0); ASSERT(zilog->zl_clean_taskq != NULL); atomic_add_64(&zilog->zl_itx_list_sz, -itxg->itxg_sod); itxg->itxg_sod = 0; clean_me = itxg->itxg_itxs; itxg->itxg_itxs = NULL; itxg->itxg_txg = 0; mutex_exit(&itxg->itxg_lock); /* * Preferably start a task queue to free up the old itxs but * if taskq_dispatch can't allocate resources to do that then * free it in-line. This should be rare. Note, using TQ_SLEEP * created a bad performance problem. */ if (taskq_dispatch(zilog->zl_clean_taskq, (void (*)(void *))zil_itxg_clean, clean_me, TQ_NOSLEEP) == 0) zil_itxg_clean(clean_me); } /* * Get the list of itxs to commit into zl_itx_commit_list. */ static void zil_get_commit_list(zilog_t *zilog) { uint64_t otxg, txg; list_t *commit_list = &zilog->zl_itx_commit_list; uint64_t push_sod = 0; if (spa_freeze_txg(zilog->zl_spa) != UINT64_MAX) /* ziltest support */ otxg = ZILTEST_TXG; else otxg = spa_last_synced_txg(zilog->zl_spa) + 1; for (txg = otxg; txg < (otxg + TXG_CONCURRENT_STATES); txg++) { itxg_t *itxg = &zilog->zl_itxg[txg & TXG_MASK]; mutex_enter(&itxg->itxg_lock); if (itxg->itxg_txg != txg) { mutex_exit(&itxg->itxg_lock); continue; } list_move_tail(commit_list, &itxg->itxg_itxs->i_sync_list); push_sod += itxg->itxg_sod; itxg->itxg_sod = 0; mutex_exit(&itxg->itxg_lock); } atomic_add_64(&zilog->zl_itx_list_sz, -push_sod); } /* * Move the async itxs for a specified object to commit into sync lists. */ static void zil_async_to_sync(zilog_t *zilog, uint64_t foid) { uint64_t otxg, txg; itx_async_node_t *ian; avl_tree_t *t; avl_index_t where; if (spa_freeze_txg(zilog->zl_spa) != UINT64_MAX) /* ziltest support */ otxg = ZILTEST_TXG; else otxg = spa_last_synced_txg(zilog->zl_spa) + 1; for (txg = otxg; txg < (otxg + TXG_CONCURRENT_STATES); txg++) { itxg_t *itxg = &zilog->zl_itxg[txg & TXG_MASK]; mutex_enter(&itxg->itxg_lock); if (itxg->itxg_txg != txg) { mutex_exit(&itxg->itxg_lock); continue; } /* * If a foid is specified then find that node and append its * list. Otherwise walk the tree appending all the lists * to the sync list. We add to the end rather than the * beginning to ensure the create has happened. */ t = &itxg->itxg_itxs->i_async_tree; if (foid != 0) { ian = avl_find(t, &foid, &where); if (ian != NULL) { list_move_tail(&itxg->itxg_itxs->i_sync_list, &ian->ia_list); } } else { void *cookie = NULL; while ((ian = avl_destroy_nodes(t, &cookie)) != NULL) { list_move_tail(&itxg->itxg_itxs->i_sync_list, &ian->ia_list); list_destroy(&ian->ia_list); kmem_free(ian, sizeof (itx_async_node_t)); } } mutex_exit(&itxg->itxg_lock); } } static void zil_commit_writer(zilog_t *zilog) { uint64_t txg; itx_t *itx; lwb_t *lwb; spa_t *spa = zilog->zl_spa; int error = 0; ASSERT(zilog->zl_root_zio == NULL); mutex_exit(&zilog->zl_lock); zil_get_commit_list(zilog); /* * Return if there's nothing to commit before we dirty the fs by * calling zil_create(). */ if (list_head(&zilog->zl_itx_commit_list) == NULL) { mutex_enter(&zilog->zl_lock); return; } if (zilog->zl_suspend) { lwb = NULL; } else { lwb = list_tail(&zilog->zl_lwb_list); if (lwb == NULL) lwb = zil_create(zilog); } DTRACE_PROBE1(zil__cw1, zilog_t *, zilog); for (itx = list_head(&zilog->zl_itx_commit_list); itx != NULL; itx = list_next(&zilog->zl_itx_commit_list, itx)) { txg = itx->itx_lr.lrc_txg; ASSERT(txg); if (txg > spa_last_synced_txg(spa) || txg > spa_freeze_txg(spa)) lwb = zil_lwb_commit(zilog, itx, lwb); } DTRACE_PROBE1(zil__cw2, zilog_t *, zilog); /* write the last block out */ if (lwb != NULL && lwb->lwb_zio != NULL) lwb = zil_lwb_write_start(zilog, lwb); zilog->zl_cur_used = 0; /* * Wait if necessary for the log blocks to be on stable storage. */ if (zilog->zl_root_zio) { error = zio_wait(zilog->zl_root_zio); zilog->zl_root_zio = NULL; zil_flush_vdevs(zilog); } if (error || lwb == NULL) txg_wait_synced(zilog->zl_dmu_pool, 0); while ((itx = list_head(&zilog->zl_itx_commit_list))) { txg = itx->itx_lr.lrc_txg; ASSERT(txg); if (itx->itx_callback != NULL) itx->itx_callback(itx->itx_callback_data); list_remove(&zilog->zl_itx_commit_list, itx); kmem_free(itx, offsetof(itx_t, itx_lr) + itx->itx_lr.lrc_reclen); } mutex_enter(&zilog->zl_lock); /* * Remember the highest committed log sequence number for ztest. * We only update this value when all the log writes succeeded, * because ztest wants to ASSERT that it got the whole log chain. */ if (error == 0 && lwb != NULL) zilog->zl_commit_lr_seq = zilog->zl_lr_seq; } /* * Commit zfs transactions to stable storage. * If foid is 0 push out all transactions, otherwise push only those * for that object or might reference that object. * * itxs are committed in batches. In a heavily stressed zil there will be * a commit writer thread who is writing out a bunch of itxs to the log * for a set of committing threads (cthreads) in the same batch as the writer. * Those cthreads are all waiting on the same cv for that batch. * * There will also be a different and growing batch of threads that are * waiting to commit (qthreads). When the committing batch completes * a transition occurs such that the cthreads exit and the qthreads become * cthreads. One of the new cthreads becomes the writer thread for the * batch. Any new threads arriving become new qthreads. * * Only 2 condition variables are needed and there's no transition * between the two cvs needed. They just flip-flop between qthreads * and cthreads. * * Using this scheme we can efficiently wakeup up only those threads * that have been committed. */ void zil_commit(zilog_t *zilog, uint64_t foid) { uint64_t mybatch; if (zilog->zl_sync == ZFS_SYNC_DISABLED) return; ZIL_STAT_BUMP(zil_commit_count); /* move the async itxs for the foid to the sync queues */ zil_async_to_sync(zilog, foid); mutex_enter(&zilog->zl_lock); mybatch = zilog->zl_next_batch; while (zilog->zl_writer) { cv_wait(&zilog->zl_cv_batch[mybatch & 1], &zilog->zl_lock); if (mybatch <= zilog->zl_com_batch) { mutex_exit(&zilog->zl_lock); return; } } zilog->zl_next_batch++; zilog->zl_writer = B_TRUE; ZIL_STAT_BUMP(zil_commit_writer_count); zil_commit_writer(zilog); zilog->zl_com_batch = mybatch; zilog->zl_writer = B_FALSE; /* wake up one thread to become the next writer */ cv_signal(&zilog->zl_cv_batch[(mybatch+1) & 1]); /* wake up all threads waiting for this batch to be committed */ cv_broadcast(&zilog->zl_cv_batch[mybatch & 1]); mutex_exit(&zilog->zl_lock); } /* * Called in syncing context to free committed log blocks and update log header. */ void zil_sync(zilog_t *zilog, dmu_tx_t *tx) { zil_header_t *zh = zil_header_in_syncing_context(zilog); uint64_t txg = dmu_tx_get_txg(tx); spa_t *spa = zilog->zl_spa; uint64_t *replayed_seq = &zilog->zl_replayed_seq[txg & TXG_MASK]; lwb_t *lwb; /* * We don't zero out zl_destroy_txg, so make sure we don't try * to destroy it twice. */ if (spa_sync_pass(spa) != 1) return; mutex_enter(&zilog->zl_lock); ASSERT(zilog->zl_stop_sync == 0); if (*replayed_seq != 0) { ASSERT(zh->zh_replay_seq < *replayed_seq); zh->zh_replay_seq = *replayed_seq; *replayed_seq = 0; } if (zilog->zl_destroy_txg == txg) { blkptr_t blk = zh->zh_log; ASSERT(list_head(&zilog->zl_lwb_list) == NULL); bzero(zh, sizeof (zil_header_t)); bzero(zilog->zl_replayed_seq, sizeof (zilog->zl_replayed_seq)); if (zilog->zl_keep_first) { /* * If this block was part of log chain that couldn't * be claimed because a device was missing during * zil_claim(), but that device later returns, * then this block could erroneously appear valid. * To guard against this, assign a new GUID to the new * log chain so it doesn't matter what blk points to. */ zil_init_log_chain(zilog, &blk); zh->zh_log = blk; } } while ((lwb = list_head(&zilog->zl_lwb_list)) != NULL) { zh->zh_log = lwb->lwb_blk; if (lwb->lwb_buf != NULL || lwb->lwb_max_txg > txg) break; ASSERT(lwb->lwb_zio == NULL); list_remove(&zilog->zl_lwb_list, lwb); zio_free_zil(spa, txg, &lwb->lwb_blk); kmem_cache_free(zil_lwb_cache, lwb); /* * If we don't have anything left in the lwb list then * we've had an allocation failure and we need to zero * out the zil_header blkptr so that we don't end * up freeing the same block twice. */ if (list_head(&zilog->zl_lwb_list) == NULL) BP_ZERO(&zh->zh_log); } /* * Remove fastwrite on any blocks that have been pre-allocated for * the next commit. This prevents fastwrite counter pollution by * unused, long-lived LWBs. */ for (; lwb != NULL; lwb = list_next(&zilog->zl_lwb_list, lwb)) { if (lwb->lwb_fastwrite && !lwb->lwb_zio) { metaslab_fastwrite_unmark(zilog->zl_spa, &lwb->lwb_blk); lwb->lwb_fastwrite = 0; } } mutex_exit(&zilog->zl_lock); } void zil_init(void) { zil_lwb_cache = kmem_cache_create("zil_lwb_cache", sizeof (struct lwb), 0, NULL, NULL, NULL, NULL, NULL, 0); zil_ksp = kstat_create("zfs", 0, "zil", "misc", KSTAT_TYPE_NAMED, sizeof (zil_stats) / sizeof (kstat_named_t), KSTAT_FLAG_VIRTUAL); if (zil_ksp != NULL) { zil_ksp->ks_data = &zil_stats; kstat_install(zil_ksp); } } void zil_fini(void) { kmem_cache_destroy(zil_lwb_cache); if (zil_ksp != NULL) { kstat_delete(zil_ksp); zil_ksp = NULL; } } void zil_set_sync(zilog_t *zilog, uint64_t sync) { zilog->zl_sync = sync; } void zil_set_logbias(zilog_t *zilog, uint64_t logbias) { zilog->zl_logbias = logbias; } zilog_t * zil_alloc(objset_t *os, zil_header_t *zh_phys) { zilog_t *zilog; int i; zilog = kmem_zalloc(sizeof (zilog_t), KM_PUSHPAGE); zilog->zl_header = zh_phys; zilog->zl_os = os; zilog->zl_spa = dmu_objset_spa(os); zilog->zl_dmu_pool = dmu_objset_pool(os); zilog->zl_destroy_txg = TXG_INITIAL - 1; zilog->zl_logbias = dmu_objset_logbias(os); zilog->zl_sync = dmu_objset_syncprop(os); zilog->zl_next_batch = 1; mutex_init(&zilog->zl_lock, NULL, MUTEX_DEFAULT, NULL); for (i = 0; i < TXG_SIZE; i++) { mutex_init(&zilog->zl_itxg[i].itxg_lock, NULL, MUTEX_DEFAULT, NULL); } list_create(&zilog->zl_lwb_list, sizeof (lwb_t), offsetof(lwb_t, lwb_node)); list_create(&zilog->zl_itx_commit_list, sizeof (itx_t), offsetof(itx_t, itx_node)); mutex_init(&zilog->zl_vdev_lock, NULL, MUTEX_DEFAULT, NULL); avl_create(&zilog->zl_vdev_tree, zil_vdev_compare, sizeof (zil_vdev_node_t), offsetof(zil_vdev_node_t, zv_node)); cv_init(&zilog->zl_cv_writer, NULL, CV_DEFAULT, NULL); cv_init(&zilog->zl_cv_suspend, NULL, CV_DEFAULT, NULL); cv_init(&zilog->zl_cv_batch[0], NULL, CV_DEFAULT, NULL); cv_init(&zilog->zl_cv_batch[1], NULL, CV_DEFAULT, NULL); return (zilog); } void zil_free(zilog_t *zilog) { int i; zilog->zl_stop_sync = 1; ASSERT0(zilog->zl_suspend); ASSERT0(zilog->zl_suspending); ASSERT(list_is_empty(&zilog->zl_lwb_list)); list_destroy(&zilog->zl_lwb_list); avl_destroy(&zilog->zl_vdev_tree); mutex_destroy(&zilog->zl_vdev_lock); ASSERT(list_is_empty(&zilog->zl_itx_commit_list)); list_destroy(&zilog->zl_itx_commit_list); for (i = 0; i < TXG_SIZE; i++) { /* * It's possible for an itx to be generated that doesn't dirty * a txg (e.g. ztest TX_TRUNCATE). So there's no zil_clean() * callback to remove the entry. We remove those here. * * Also free up the ziltest itxs. */ if (zilog->zl_itxg[i].itxg_itxs) zil_itxg_clean(zilog->zl_itxg[i].itxg_itxs); mutex_destroy(&zilog->zl_itxg[i].itxg_lock); } mutex_destroy(&zilog->zl_lock); cv_destroy(&zilog->zl_cv_writer); cv_destroy(&zilog->zl_cv_suspend); cv_destroy(&zilog->zl_cv_batch[0]); cv_destroy(&zilog->zl_cv_batch[1]); kmem_free(zilog, sizeof (zilog_t)); } /* * Open an intent log. */ zilog_t * zil_open(objset_t *os, zil_get_data_t *get_data) { zilog_t *zilog = dmu_objset_zil(os); ASSERT(zilog->zl_clean_taskq == NULL); ASSERT(zilog->zl_get_data == NULL); ASSERT(list_is_empty(&zilog->zl_lwb_list)); zilog->zl_get_data = get_data; zilog->zl_clean_taskq = taskq_create("zil_clean", 1, minclsyspri, 2, 2, TASKQ_PREPOPULATE); return (zilog); } /* * Close an intent log. */ void zil_close(zilog_t *zilog) { lwb_t *lwb; uint64_t txg = 0; zil_commit(zilog, 0); /* commit all itx */ /* * The lwb_max_txg for the stubby lwb will reflect the last activity * for the zil. After a txg_wait_synced() on the txg we know all the * callbacks have occurred that may clean the zil. Only then can we * destroy the zl_clean_taskq. */ mutex_enter(&zilog->zl_lock); lwb = list_tail(&zilog->zl_lwb_list); if (lwb != NULL) txg = lwb->lwb_max_txg; mutex_exit(&zilog->zl_lock); if (txg) txg_wait_synced(zilog->zl_dmu_pool, txg); ASSERT(!zilog_is_dirty(zilog)); taskq_destroy(zilog->zl_clean_taskq); zilog->zl_clean_taskq = NULL; zilog->zl_get_data = NULL; /* * We should have only one LWB left on the list; remove it now. */ mutex_enter(&zilog->zl_lock); lwb = list_head(&zilog->zl_lwb_list); if (lwb != NULL) { ASSERT(lwb == list_tail(&zilog->zl_lwb_list)); ASSERT(lwb->lwb_zio == NULL); if (lwb->lwb_fastwrite) metaslab_fastwrite_unmark(zilog->zl_spa, &lwb->lwb_blk); list_remove(&zilog->zl_lwb_list, lwb); zio_buf_free(lwb->lwb_buf, lwb->lwb_sz); kmem_cache_free(zil_lwb_cache, lwb); } mutex_exit(&zilog->zl_lock); } static char *suspend_tag = "zil suspending"; /* * Suspend an intent log. While in suspended mode, we still honor * synchronous semantics, but we rely on txg_wait_synced() to do it. * On old version pools, we suspend the log briefly when taking a * snapshot so that it will have an empty intent log. * * Long holds are not really intended to be used the way we do here -- * held for such a short time. A concurrent caller of dsl_dataset_long_held() * could fail. Therefore we take pains to only put a long hold if it is * actually necessary. Fortunately, it will only be necessary if the * objset is currently mounted (or the ZVOL equivalent). In that case it * will already have a long hold, so we are not really making things any worse. * * Ideally, we would locate the existing long-holder (i.e. the zfsvfs_t or * zvol_state_t), and use their mechanism to prevent their hold from being * dropped (e.g. VFS_HOLD()). However, that would be even more pain for * very little gain. * * if cookiep == NULL, this does both the suspend & resume. * Otherwise, it returns with the dataset "long held", and the cookie * should be passed into zil_resume(). */ int zil_suspend(const char *osname, void **cookiep) { objset_t *os; zilog_t *zilog; const zil_header_t *zh; int error; error = dmu_objset_hold(osname, suspend_tag, &os); if (error != 0) return (error); zilog = dmu_objset_zil(os); mutex_enter(&zilog->zl_lock); zh = zilog->zl_header; if (zh->zh_flags & ZIL_REPLAY_NEEDED) { /* unplayed log */ mutex_exit(&zilog->zl_lock); dmu_objset_rele(os, suspend_tag); return (SET_ERROR(EBUSY)); } /* * Don't put a long hold in the cases where we can avoid it. This * is when there is no cookie so we are doing a suspend & resume * (i.e. called from zil_vdev_offline()), and there's nothing to do * for the suspend because it's already suspended, or there's no ZIL. */ if (cookiep == NULL && !zilog->zl_suspending && (zilog->zl_suspend > 0 || BP_IS_HOLE(&zh->zh_log))) { mutex_exit(&zilog->zl_lock); dmu_objset_rele(os, suspend_tag); return (0); } dsl_dataset_long_hold(dmu_objset_ds(os), suspend_tag); dsl_pool_rele(dmu_objset_pool(os), suspend_tag); zilog->zl_suspend++; if (zilog->zl_suspend > 1) { /* * Someone else is already suspending it. * Just wait for them to finish. */ while (zilog->zl_suspending) cv_wait(&zilog->zl_cv_suspend, &zilog->zl_lock); mutex_exit(&zilog->zl_lock); if (cookiep == NULL) zil_resume(os); else *cookiep = os; return (0); } /* * If there is no pointer to an on-disk block, this ZIL must not * be active (e.g. filesystem not mounted), so there's nothing * to clean up. */ if (BP_IS_HOLE(&zh->zh_log)) { ASSERT(cookiep != NULL); /* fast path already handled */ *cookiep = os; mutex_exit(&zilog->zl_lock); return (0); } zilog->zl_suspending = B_TRUE; mutex_exit(&zilog->zl_lock); zil_commit(zilog, 0); zil_destroy(zilog, B_FALSE); mutex_enter(&zilog->zl_lock); zilog->zl_suspending = B_FALSE; cv_broadcast(&zilog->zl_cv_suspend); mutex_exit(&zilog->zl_lock); if (cookiep == NULL) zil_resume(os); else *cookiep = os; return (0); } void zil_resume(void *cookie) { objset_t *os = cookie; zilog_t *zilog = dmu_objset_zil(os); mutex_enter(&zilog->zl_lock); ASSERT(zilog->zl_suspend != 0); zilog->zl_suspend--; mutex_exit(&zilog->zl_lock); dsl_dataset_long_rele(dmu_objset_ds(os), suspend_tag); dsl_dataset_rele(dmu_objset_ds(os), suspend_tag); } typedef struct zil_replay_arg { zil_replay_func_t *zr_replay; void *zr_arg; boolean_t zr_byteswap; char *zr_lr; } zil_replay_arg_t; static int zil_replay_error(zilog_t *zilog, lr_t *lr, int error) { char name[MAXNAMELEN]; zilog->zl_replaying_seq--; /* didn't actually replay this one */ dmu_objset_name(zilog->zl_os, name); cmn_err(CE_WARN, "ZFS replay transaction error %d, " "dataset %s, seq 0x%llx, txtype %llu %s\n", error, name, (u_longlong_t)lr->lrc_seq, (u_longlong_t)(lr->lrc_txtype & ~TX_CI), (lr->lrc_txtype & TX_CI) ? "CI" : ""); return (error); } static int zil_replay_log_record(zilog_t *zilog, lr_t *lr, void *zra, uint64_t claim_txg) { zil_replay_arg_t *zr = zra; const zil_header_t *zh = zilog->zl_header; uint64_t reclen = lr->lrc_reclen; uint64_t txtype = lr->lrc_txtype; int error = 0; zilog->zl_replaying_seq = lr->lrc_seq; if (lr->lrc_seq <= zh->zh_replay_seq) /* already replayed */ return (0); if (lr->lrc_txg < claim_txg) /* already committed */ return (0); /* Strip case-insensitive bit, still present in log record */ txtype &= ~TX_CI; if (txtype == 0 || txtype >= TX_MAX_TYPE) return (zil_replay_error(zilog, lr, EINVAL)); /* * If this record type can be logged out of order, the object * (lr_foid) may no longer exist. That's legitimate, not an error. */ if (TX_OOO(txtype)) { error = dmu_object_info(zilog->zl_os, ((lr_ooo_t *)lr)->lr_foid, NULL); if (error == ENOENT || error == EEXIST) return (0); } /* * Make a copy of the data so we can revise and extend it. */ bcopy(lr, zr->zr_lr, reclen); /* * If this is a TX_WRITE with a blkptr, suck in the data. */ if (txtype == TX_WRITE && reclen == sizeof (lr_write_t)) { error = zil_read_log_data(zilog, (lr_write_t *)lr, zr->zr_lr + reclen); if (error != 0) return (zil_replay_error(zilog, lr, error)); } /* * The log block containing this lr may have been byteswapped * so that we can easily examine common fields like lrc_txtype. * However, the log is a mix of different record types, and only the * replay vectors know how to byteswap their records. Therefore, if * the lr was byteswapped, undo it before invoking the replay vector. */ if (zr->zr_byteswap) byteswap_uint64_array(zr->zr_lr, reclen); /* * We must now do two things atomically: replay this log record, * and update the log header sequence number to reflect the fact that * we did so. At the end of each replay function the sequence number * is updated if we are in replay mode. */ error = zr->zr_replay[txtype](zr->zr_arg, zr->zr_lr, zr->zr_byteswap); if (error != 0) { /* * The DMU's dnode layer doesn't see removes until the txg * commits, so a subsequent claim can spuriously fail with * EEXIST. So if we receive any error we try syncing out * any removes then retry the transaction. Note that we * specify B_FALSE for byteswap now, so we don't do it twice. */ txg_wait_synced(spa_get_dsl(zilog->zl_spa), 0); error = zr->zr_replay[txtype](zr->zr_arg, zr->zr_lr, B_FALSE); if (error != 0) return (zil_replay_error(zilog, lr, error)); } return (0); } /* ARGSUSED */ static int zil_incr_blks(zilog_t *zilog, blkptr_t *bp, void *arg, uint64_t claim_txg) { zilog->zl_replay_blks++; return (0); } /* * If this dataset has a non-empty intent log, replay it and destroy it. */ void zil_replay(objset_t *os, void *arg, zil_replay_func_t replay_func[TX_MAX_TYPE]) { zilog_t *zilog = dmu_objset_zil(os); const zil_header_t *zh = zilog->zl_header; zil_replay_arg_t zr; if ((zh->zh_flags & ZIL_REPLAY_NEEDED) == 0) { zil_destroy(zilog, B_TRUE); return; } zr.zr_replay = replay_func; zr.zr_arg = arg; zr.zr_byteswap = BP_SHOULD_BYTESWAP(&zh->zh_log); zr.zr_lr = vmem_alloc(2 * SPA_MAXBLOCKSIZE, KM_PUSHPAGE); /* * Wait for in-progress removes to sync before starting replay. */ txg_wait_synced(zilog->zl_dmu_pool, 0); zilog->zl_replay = B_TRUE; zilog->zl_replay_time = ddi_get_lbolt(); ASSERT(zilog->zl_replay_blks == 0); (void) zil_parse(zilog, zil_incr_blks, zil_replay_log_record, &zr, zh->zh_claim_txg); vmem_free(zr.zr_lr, 2 * SPA_MAXBLOCKSIZE); zil_destroy(zilog, B_FALSE); txg_wait_synced(zilog->zl_dmu_pool, zilog->zl_destroy_txg); zilog->zl_replay = B_FALSE; } boolean_t zil_replaying(zilog_t *zilog, dmu_tx_t *tx) { if (zilog->zl_sync == ZFS_SYNC_DISABLED) return (B_TRUE); if (zilog->zl_replay) { dsl_dataset_dirty(dmu_objset_ds(zilog->zl_os), tx); zilog->zl_replayed_seq[dmu_tx_get_txg(tx) & TXG_MASK] = zilog->zl_replaying_seq; return (B_TRUE); } return (B_FALSE); } /* ARGSUSED */ int zil_vdev_offline(const char *osname, void *arg) { int error; error = zil_suspend(osname, NULL); if (error != 0) return (SET_ERROR(EEXIST)); return (0); } #if defined(_KERNEL) && defined(HAVE_SPL) EXPORT_SYMBOL(zil_alloc); EXPORT_SYMBOL(zil_free); EXPORT_SYMBOL(zil_open); EXPORT_SYMBOL(zil_close); EXPORT_SYMBOL(zil_replay); EXPORT_SYMBOL(zil_replaying); EXPORT_SYMBOL(zil_destroy); EXPORT_SYMBOL(zil_destroy_sync); EXPORT_SYMBOL(zil_itx_create); EXPORT_SYMBOL(zil_itx_destroy); EXPORT_SYMBOL(zil_itx_assign); EXPORT_SYMBOL(zil_commit); EXPORT_SYMBOL(zil_vdev_offline); EXPORT_SYMBOL(zil_claim); EXPORT_SYMBOL(zil_check_log_chain); EXPORT_SYMBOL(zil_sync); EXPORT_SYMBOL(zil_clean); EXPORT_SYMBOL(zil_suspend); EXPORT_SYMBOL(zil_resume); EXPORT_SYMBOL(zil_add_block); EXPORT_SYMBOL(zil_bp_tree_add); EXPORT_SYMBOL(zil_set_sync); EXPORT_SYMBOL(zil_set_logbias); module_param(zil_replay_disable, int, 0644); MODULE_PARM_DESC(zil_replay_disable, "Disable intent logging replay"); module_param(zfs_nocacheflush, int, 0644); MODULE_PARM_DESC(zfs_nocacheflush, "Disable cache flushes"); module_param(zil_slog_limit, ulong, 0644); MODULE_PARM_DESC(zil_slog_limit, "Max commit bytes to separate log device"); #endif diff --git a/module/zfs/zio.c b/module/zfs/zio.c index 7c0e6bf7e8ee..d7c8458f473d 100644 --- a/module/zfs/zio.c +++ b/module/zfs/zio.c @@ -1,3440 +1,3439 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2014 by Delphix. All rights reserved. * Copyright (c) 2011 Nexenta Systems, Inc. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * ========================================================================== * I/O type descriptions * ========================================================================== */ const char *zio_type_name[ZIO_TYPES] = { "z_null", "z_rd", "z_wr", "z_fr", "z_cl", "z_ioctl" }; /* * ========================================================================== * I/O kmem caches * ========================================================================== */ kmem_cache_t *zio_cache; kmem_cache_t *zio_link_cache; kmem_cache_t *zio_vdev_cache; kmem_cache_t *zio_buf_cache[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; kmem_cache_t *zio_data_buf_cache[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; int zio_bulk_flags = 0; int zio_delay_max = ZIO_DELAY_MAX; /* * The following actions directly effect the spa's sync-to-convergence logic. * The values below define the sync pass when we start performing the action. * Care should be taken when changing these values as they directly impact * spa_sync() performance. Tuning these values may introduce subtle performance * pathologies and should only be done in the context of performance analysis. * These tunables will eventually be removed and replaced with #defines once * enough analysis has been done to determine optimal values. * * The 'zfs_sync_pass_deferred_free' pass must be greater than 1 to ensure that * regular blocks are not deferred. */ int zfs_sync_pass_deferred_free = 2; /* defer frees starting in this pass */ int zfs_sync_pass_dont_compress = 5; /* don't compress starting in this pass */ int zfs_sync_pass_rewrite = 2; /* rewrite new bps starting in this pass */ /* * An allocating zio is one that either currently has the DVA allocate * stage set or will have it later in its lifetime. */ #define IO_IS_ALLOCATING(zio) ((zio)->io_orig_pipeline & ZIO_STAGE_DVA_ALLOCATE) int zio_requeue_io_start_cut_in_line = 1; #ifdef ZFS_DEBUG int zio_buf_debug_limit = 16384; #else int zio_buf_debug_limit = 0; #endif static inline void __zio_execute(zio_t *zio); static int zio_cons(void *arg, void *unused, int kmflag) { zio_t *zio = arg; bzero(zio, sizeof (zio_t)); mutex_init(&zio->io_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&zio->io_cv, NULL, CV_DEFAULT, NULL); list_create(&zio->io_parent_list, sizeof (zio_link_t), offsetof(zio_link_t, zl_parent_node)); list_create(&zio->io_child_list, sizeof (zio_link_t), offsetof(zio_link_t, zl_child_node)); return (0); } static void zio_dest(void *arg, void *unused) { zio_t *zio = arg; mutex_destroy(&zio->io_lock); cv_destroy(&zio->io_cv); list_destroy(&zio->io_parent_list); list_destroy(&zio->io_child_list); } void zio_init(void) { size_t c; vmem_t *data_alloc_arena = NULL; zio_cache = kmem_cache_create("zio_cache", sizeof (zio_t), 0, zio_cons, zio_dest, NULL, NULL, NULL, 0); zio_link_cache = kmem_cache_create("zio_link_cache", sizeof (zio_link_t), 0, NULL, NULL, NULL, NULL, NULL, 0); zio_vdev_cache = kmem_cache_create("zio_vdev_cache", sizeof (vdev_io_t), PAGESIZE, NULL, NULL, NULL, NULL, NULL, 0); /* * For small buffers, we want a cache for each multiple of * SPA_MINBLOCKSIZE. For medium-size buffers, we want a cache * for each quarter-power of 2. For large buffers, we want * a cache for each multiple of PAGESIZE. */ for (c = 0; c < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT; c++) { size_t size = (c + 1) << SPA_MINBLOCKSHIFT; size_t p2 = size; size_t align = 0; while (p2 & (p2 - 1)) p2 &= p2 - 1; #ifndef _KERNEL /* * If we are using watchpoints, put each buffer on its own page, * to eliminate the performance overhead of trapping to the * kernel when modifying a non-watched buffer that shares the * page with a watched buffer. */ if (arc_watch && !IS_P2ALIGNED(size, PAGESIZE)) continue; #endif if (size <= 4 * SPA_MINBLOCKSIZE) { align = SPA_MINBLOCKSIZE; } else if (IS_P2ALIGNED(size, PAGESIZE)) { align = PAGESIZE; } else if (IS_P2ALIGNED(size, p2 >> 2)) { align = p2 >> 2; } if (align != 0) { char name[36]; int flags = zio_bulk_flags; (void) sprintf(name, "zio_buf_%lu", (ulong_t)size); zio_buf_cache[c] = kmem_cache_create(name, size, align, NULL, NULL, NULL, NULL, NULL, flags); (void) sprintf(name, "zio_data_buf_%lu", (ulong_t)size); zio_data_buf_cache[c] = kmem_cache_create(name, size, align, NULL, NULL, NULL, NULL, data_alloc_arena, flags); } } while (--c != 0) { ASSERT(zio_buf_cache[c] != NULL); if (zio_buf_cache[c - 1] == NULL) zio_buf_cache[c - 1] = zio_buf_cache[c]; ASSERT(zio_data_buf_cache[c] != NULL); if (zio_data_buf_cache[c - 1] == NULL) zio_data_buf_cache[c - 1] = zio_data_buf_cache[c]; } zio_inject_init(); lz4_init(); } void zio_fini(void) { size_t c; kmem_cache_t *last_cache = NULL; kmem_cache_t *last_data_cache = NULL; for (c = 0; c < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT; c++) { if (zio_buf_cache[c] != last_cache) { last_cache = zio_buf_cache[c]; kmem_cache_destroy(zio_buf_cache[c]); } zio_buf_cache[c] = NULL; if (zio_data_buf_cache[c] != last_data_cache) { last_data_cache = zio_data_buf_cache[c]; kmem_cache_destroy(zio_data_buf_cache[c]); } zio_data_buf_cache[c] = NULL; } kmem_cache_destroy(zio_vdev_cache); kmem_cache_destroy(zio_link_cache); kmem_cache_destroy(zio_cache); zio_inject_fini(); lz4_fini(); } /* * ========================================================================== * Allocate and free I/O buffers * ========================================================================== */ /* * Use zio_buf_alloc to allocate ZFS metadata. This data will appear in a * crashdump if the kernel panics, so use it judiciously. Obviously, it's * useful to inspect ZFS metadata, but if possible, we should avoid keeping * excess / transient data in-core during a crashdump. */ void * zio_buf_alloc(size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; ASSERT3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); - return (kmem_cache_alloc(zio_buf_cache[c], KM_PUSHPAGE | KM_NODEBUG)); + return (kmem_cache_alloc(zio_buf_cache[c], KM_PUSHPAGE)); } /* * Use zio_data_buf_alloc to allocate data. The data will not appear in a * crashdump if the kernel panics. This exists so that we will limit the amount * of ZFS data that shows up in a kernel crashdump. (Thus reducing the amount * of kernel heap dumped to disk when the kernel panics) */ void * zio_data_buf_alloc(size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; ASSERT(c < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); - return (kmem_cache_alloc(zio_data_buf_cache[c], - KM_PUSHPAGE | KM_NODEBUG)); + return (kmem_cache_alloc(zio_data_buf_cache[c], KM_PUSHPAGE)); } void zio_buf_free(void *buf, size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; ASSERT(c < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); kmem_cache_free(zio_buf_cache[c], buf); } void zio_data_buf_free(void *buf, size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; ASSERT(c < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); kmem_cache_free(zio_data_buf_cache[c], buf); } /* * Dedicated I/O buffers to ensure that memory fragmentation never prevents * or significantly delays the issuing of a zio. These buffers are used * to aggregate I/O and could be used for raidz stripes. */ void * zio_vdev_alloc(void) { return (kmem_cache_alloc(zio_vdev_cache, KM_PUSHPAGE)); } void zio_vdev_free(void *buf) { kmem_cache_free(zio_vdev_cache, buf); } /* * ========================================================================== * Push and pop I/O transform buffers * ========================================================================== */ static void zio_push_transform(zio_t *zio, void *data, uint64_t size, uint64_t bufsize, zio_transform_func_t *transform) { zio_transform_t *zt = kmem_alloc(sizeof (zio_transform_t), KM_PUSHPAGE); zt->zt_orig_data = zio->io_data; zt->zt_orig_size = zio->io_size; zt->zt_bufsize = bufsize; zt->zt_transform = transform; zt->zt_next = zio->io_transform_stack; zio->io_transform_stack = zt; zio->io_data = data; zio->io_size = size; } static void zio_pop_transforms(zio_t *zio) { zio_transform_t *zt; while ((zt = zio->io_transform_stack) != NULL) { if (zt->zt_transform != NULL) zt->zt_transform(zio, zt->zt_orig_data, zt->zt_orig_size); if (zt->zt_bufsize != 0) zio_buf_free(zio->io_data, zt->zt_bufsize); zio->io_data = zt->zt_orig_data; zio->io_size = zt->zt_orig_size; zio->io_transform_stack = zt->zt_next; kmem_free(zt, sizeof (zio_transform_t)); } } /* * ========================================================================== * I/O transform callbacks for subblocks and decompression * ========================================================================== */ static void zio_subblock(zio_t *zio, void *data, uint64_t size) { ASSERT(zio->io_size > size); if (zio->io_type == ZIO_TYPE_READ) bcopy(zio->io_data, data, size); } static void zio_decompress(zio_t *zio, void *data, uint64_t size) { if (zio->io_error == 0 && zio_decompress_data(BP_GET_COMPRESS(zio->io_bp), zio->io_data, data, zio->io_size, size) != 0) zio->io_error = SET_ERROR(EIO); } /* * ========================================================================== * I/O parent/child relationships and pipeline interlocks * ========================================================================== */ /* * NOTE - Callers to zio_walk_parents() and zio_walk_children must * continue calling these functions until they return NULL. * Otherwise, the next caller will pick up the list walk in * some indeterminate state. (Otherwise every caller would * have to pass in a cookie to keep the state represented by * io_walk_link, which gets annoying.) */ zio_t * zio_walk_parents(zio_t *cio) { zio_link_t *zl = cio->io_walk_link; list_t *pl = &cio->io_parent_list; zl = (zl == NULL) ? list_head(pl) : list_next(pl, zl); cio->io_walk_link = zl; if (zl == NULL) return (NULL); ASSERT(zl->zl_child == cio); return (zl->zl_parent); } zio_t * zio_walk_children(zio_t *pio) { zio_link_t *zl = pio->io_walk_link; list_t *cl = &pio->io_child_list; zl = (zl == NULL) ? list_head(cl) : list_next(cl, zl); pio->io_walk_link = zl; if (zl == NULL) return (NULL); ASSERT(zl->zl_parent == pio); return (zl->zl_child); } zio_t * zio_unique_parent(zio_t *cio) { zio_t *pio = zio_walk_parents(cio); VERIFY(zio_walk_parents(cio) == NULL); return (pio); } void zio_add_child(zio_t *pio, zio_t *cio) { zio_link_t *zl = kmem_cache_alloc(zio_link_cache, KM_PUSHPAGE); int w; /* * Logical I/Os can have logical, gang, or vdev children. * Gang I/Os can have gang or vdev children. * Vdev I/Os can only have vdev children. * The following ASSERT captures all of these constraints. */ ASSERT(cio->io_child_type <= pio->io_child_type); zl->zl_parent = pio; zl->zl_child = cio; mutex_enter(&cio->io_lock); mutex_enter(&pio->io_lock); ASSERT(pio->io_state[ZIO_WAIT_DONE] == 0); for (w = 0; w < ZIO_WAIT_TYPES; w++) pio->io_children[cio->io_child_type][w] += !cio->io_state[w]; list_insert_head(&pio->io_child_list, zl); list_insert_head(&cio->io_parent_list, zl); pio->io_child_count++; cio->io_parent_count++; mutex_exit(&pio->io_lock); mutex_exit(&cio->io_lock); } static void zio_remove_child(zio_t *pio, zio_t *cio, zio_link_t *zl) { ASSERT(zl->zl_parent == pio); ASSERT(zl->zl_child == cio); mutex_enter(&cio->io_lock); mutex_enter(&pio->io_lock); list_remove(&pio->io_child_list, zl); list_remove(&cio->io_parent_list, zl); pio->io_child_count--; cio->io_parent_count--; mutex_exit(&pio->io_lock); mutex_exit(&cio->io_lock); kmem_cache_free(zio_link_cache, zl); } static boolean_t zio_wait_for_children(zio_t *zio, enum zio_child child, enum zio_wait_type wait) { uint64_t *countp = &zio->io_children[child][wait]; boolean_t waiting = B_FALSE; mutex_enter(&zio->io_lock); ASSERT(zio->io_stall == NULL); if (*countp != 0) { zio->io_stage >>= 1; zio->io_stall = countp; waiting = B_TRUE; } mutex_exit(&zio->io_lock); return (waiting); } __attribute__((always_inline)) static inline void zio_notify_parent(zio_t *pio, zio_t *zio, enum zio_wait_type wait) { uint64_t *countp = &pio->io_children[zio->io_child_type][wait]; int *errorp = &pio->io_child_error[zio->io_child_type]; mutex_enter(&pio->io_lock); if (zio->io_error && !(zio->io_flags & ZIO_FLAG_DONT_PROPAGATE)) *errorp = zio_worst_error(*errorp, zio->io_error); pio->io_reexecute |= zio->io_reexecute; ASSERT3U(*countp, >, 0); (*countp)--; if (*countp == 0 && pio->io_stall == countp) { pio->io_stall = NULL; mutex_exit(&pio->io_lock); __zio_execute(pio); } else { mutex_exit(&pio->io_lock); } } static void zio_inherit_child_errors(zio_t *zio, enum zio_child c) { if (zio->io_child_error[c] != 0 && zio->io_error == 0) zio->io_error = zio->io_child_error[c]; } /* * ========================================================================== * Create the various types of I/O (read, write, free, etc) * ========================================================================== */ static zio_t * zio_create(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, void *data, uint64_t size, zio_done_func_t *done, void *private, zio_type_t type, zio_priority_t priority, enum zio_flag flags, vdev_t *vd, uint64_t offset, const zbookmark_phys_t *zb, enum zio_stage stage, enum zio_stage pipeline) { zio_t *zio; ASSERT3U(size, <=, SPA_MAXBLOCKSIZE); ASSERT(P2PHASE(size, SPA_MINBLOCKSIZE) == 0); ASSERT(P2PHASE(offset, SPA_MINBLOCKSIZE) == 0); ASSERT(!vd || spa_config_held(spa, SCL_STATE_ALL, RW_READER)); ASSERT(!bp || !(flags & ZIO_FLAG_CONFIG_WRITER)); ASSERT(vd || stage == ZIO_STAGE_OPEN); zio = kmem_cache_alloc(zio_cache, KM_PUSHPAGE); if (vd != NULL) zio->io_child_type = ZIO_CHILD_VDEV; else if (flags & ZIO_FLAG_GANG_CHILD) zio->io_child_type = ZIO_CHILD_GANG; else if (flags & ZIO_FLAG_DDT_CHILD) zio->io_child_type = ZIO_CHILD_DDT; else zio->io_child_type = ZIO_CHILD_LOGICAL; if (bp != NULL) { zio->io_logical = NULL; zio->io_bp = (blkptr_t *)bp; zio->io_bp_copy = *bp; zio->io_bp_orig = *bp; if (type != ZIO_TYPE_WRITE || zio->io_child_type == ZIO_CHILD_DDT) zio->io_bp = &zio->io_bp_copy; /* so caller can free */ if (zio->io_child_type == ZIO_CHILD_LOGICAL) zio->io_logical = zio; if (zio->io_child_type > ZIO_CHILD_GANG && BP_IS_GANG(bp)) pipeline |= ZIO_GANG_STAGES; } else { zio->io_logical = NULL; zio->io_bp = NULL; bzero(&zio->io_bp_copy, sizeof (blkptr_t)); bzero(&zio->io_bp_orig, sizeof (blkptr_t)); } zio->io_spa = spa; zio->io_txg = txg; zio->io_ready = NULL; zio->io_physdone = NULL; zio->io_done = done; zio->io_private = private; zio->io_prev_space_delta = 0; zio->io_type = type; zio->io_priority = priority; zio->io_vd = vd; zio->io_vsd = NULL; zio->io_vsd_ops = NULL; zio->io_offset = offset; zio->io_timestamp = 0; zio->io_delta = 0; zio->io_delay = 0; zio->io_orig_data = zio->io_data = data; zio->io_orig_size = zio->io_size = size; zio->io_orig_flags = zio->io_flags = flags; zio->io_orig_stage = zio->io_stage = stage; zio->io_orig_pipeline = zio->io_pipeline = pipeline; bzero(&zio->io_prop, sizeof (zio_prop_t)); zio->io_cmd = 0; zio->io_reexecute = 0; zio->io_bp_override = NULL; zio->io_walk_link = NULL; zio->io_transform_stack = NULL; zio->io_error = 0; zio->io_child_count = 0; zio->io_phys_children = 0; zio->io_parent_count = 0; zio->io_stall = NULL; zio->io_gang_leader = NULL; zio->io_gang_tree = NULL; zio->io_executor = NULL; zio->io_waiter = NULL; zio->io_cksum_report = NULL; zio->io_ena = 0; bzero(zio->io_child_error, sizeof (int) * ZIO_CHILD_TYPES); bzero(zio->io_children, sizeof (uint64_t) * ZIO_CHILD_TYPES * ZIO_WAIT_TYPES); bzero(&zio->io_bookmark, sizeof (zbookmark_phys_t)); zio->io_state[ZIO_WAIT_READY] = (stage >= ZIO_STAGE_READY); zio->io_state[ZIO_WAIT_DONE] = (stage >= ZIO_STAGE_DONE); if (zb != NULL) zio->io_bookmark = *zb; if (pio != NULL) { if (zio->io_logical == NULL) zio->io_logical = pio->io_logical; if (zio->io_child_type == ZIO_CHILD_GANG) zio->io_gang_leader = pio->io_gang_leader; zio_add_child(pio, zio); } taskq_init_ent(&zio->io_tqent); return (zio); } static void zio_destroy(zio_t *zio) { kmem_cache_free(zio_cache, zio); } zio_t * zio_null(zio_t *pio, spa_t *spa, vdev_t *vd, zio_done_func_t *done, void *private, enum zio_flag flags) { zio_t *zio; zio = zio_create(pio, spa, 0, NULL, NULL, 0, done, private, ZIO_TYPE_NULL, ZIO_PRIORITY_NOW, flags, vd, 0, NULL, ZIO_STAGE_OPEN, ZIO_INTERLOCK_PIPELINE); return (zio); } zio_t * zio_root(spa_t *spa, zio_done_func_t *done, void *private, enum zio_flag flags) { return (zio_null(NULL, spa, NULL, done, private, flags)); } zio_t * zio_read(zio_t *pio, spa_t *spa, const blkptr_t *bp, void *data, uint64_t size, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, const zbookmark_phys_t *zb) { zio_t *zio; zio = zio_create(pio, spa, BP_PHYSICAL_BIRTH(bp), bp, data, size, done, private, ZIO_TYPE_READ, priority, flags, NULL, 0, zb, ZIO_STAGE_OPEN, (flags & ZIO_FLAG_DDT_CHILD) ? ZIO_DDT_CHILD_READ_PIPELINE : ZIO_READ_PIPELINE); return (zio); } zio_t * zio_write(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, void *data, uint64_t size, const zio_prop_t *zp, zio_done_func_t *ready, zio_done_func_t *physdone, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, const zbookmark_phys_t *zb) { zio_t *zio; ASSERT(zp->zp_checksum >= ZIO_CHECKSUM_OFF && zp->zp_checksum < ZIO_CHECKSUM_FUNCTIONS && zp->zp_compress >= ZIO_COMPRESS_OFF && zp->zp_compress < ZIO_COMPRESS_FUNCTIONS && DMU_OT_IS_VALID(zp->zp_type) && zp->zp_level < 32 && zp->zp_copies > 0 && zp->zp_copies <= spa_max_replication(spa)); zio = zio_create(pio, spa, txg, bp, data, size, done, private, ZIO_TYPE_WRITE, priority, flags, NULL, 0, zb, ZIO_STAGE_OPEN, (flags & ZIO_FLAG_DDT_CHILD) ? ZIO_DDT_CHILD_WRITE_PIPELINE : ZIO_WRITE_PIPELINE); zio->io_ready = ready; zio->io_physdone = physdone; zio->io_prop = *zp; /* * Data can be NULL if we are going to call zio_write_override() to * provide the already-allocated BP. But we may need the data to * verify a dedup hit (if requested). In this case, don't try to * dedup (just take the already-allocated BP verbatim). */ if (data == NULL && zio->io_prop.zp_dedup_verify) { zio->io_prop.zp_dedup = zio->io_prop.zp_dedup_verify = B_FALSE; } return (zio); } zio_t * zio_rewrite(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, void *data, uint64_t size, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, zbookmark_phys_t *zb) { zio_t *zio; zio = zio_create(pio, spa, txg, bp, data, size, done, private, ZIO_TYPE_WRITE, priority, flags, NULL, 0, zb, ZIO_STAGE_OPEN, ZIO_REWRITE_PIPELINE); return (zio); } void zio_write_override(zio_t *zio, blkptr_t *bp, int copies, boolean_t nopwrite) { ASSERT(zio->io_type == ZIO_TYPE_WRITE); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(zio->io_stage == ZIO_STAGE_OPEN); ASSERT(zio->io_txg == spa_syncing_txg(zio->io_spa)); /* * We must reset the io_prop to match the values that existed * when the bp was first written by dmu_sync() keeping in mind * that nopwrite and dedup are mutually exclusive. */ zio->io_prop.zp_dedup = nopwrite ? B_FALSE : zio->io_prop.zp_dedup; zio->io_prop.zp_nopwrite = nopwrite; zio->io_prop.zp_copies = copies; zio->io_bp_override = bp; } void zio_free(spa_t *spa, uint64_t txg, const blkptr_t *bp) { /* * The check for EMBEDDED is a performance optimization. We * process the free here (by ignoring it) rather than * putting it on the list and then processing it in zio_free_sync(). */ if (BP_IS_EMBEDDED(bp)) return; metaslab_check_free(spa, bp); /* * Frees that are for the currently-syncing txg, are not going to be * deferred, and which will not need to do a read (i.e. not GANG or * DEDUP), can be processed immediately. Otherwise, put them on the * in-memory list for later processing. */ if (BP_IS_GANG(bp) || BP_GET_DEDUP(bp) || txg != spa->spa_syncing_txg || spa_sync_pass(spa) >= zfs_sync_pass_deferred_free) { bplist_append(&spa->spa_free_bplist[txg & TXG_MASK], bp); } else { VERIFY0(zio_wait(zio_free_sync(NULL, spa, txg, bp, 0))); } } zio_t * zio_free_sync(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, enum zio_flag flags) { zio_t *zio; enum zio_stage stage = ZIO_FREE_PIPELINE; ASSERT(!BP_IS_HOLE(bp)); ASSERT(spa_syncing_txg(spa) == txg); ASSERT(spa_sync_pass(spa) < zfs_sync_pass_deferred_free); if (BP_IS_EMBEDDED(bp)) return (zio_null(pio, spa, NULL, NULL, NULL, 0)); metaslab_check_free(spa, bp); arc_freed(spa, bp); /* * GANG and DEDUP blocks can induce a read (for the gang block header, * or the DDT), so issue them asynchronously so that this thread is * not tied up. */ if (BP_IS_GANG(bp) || BP_GET_DEDUP(bp)) stage |= ZIO_STAGE_ISSUE_ASYNC; zio = zio_create(pio, spa, txg, bp, NULL, BP_GET_PSIZE(bp), NULL, NULL, ZIO_TYPE_FREE, ZIO_PRIORITY_NOW, flags, NULL, 0, NULL, ZIO_STAGE_OPEN, stage); return (zio); } zio_t * zio_claim(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, zio_done_func_t *done, void *private, enum zio_flag flags) { zio_t *zio; dprintf_bp(bp, "claiming in txg %llu", txg); if (BP_IS_EMBEDDED(bp)) return (zio_null(pio, spa, NULL, NULL, NULL, 0)); /* * A claim is an allocation of a specific block. Claims are needed * to support immediate writes in the intent log. The issue is that * immediate writes contain committed data, but in a txg that was * *not* committed. Upon opening the pool after an unclean shutdown, * the intent log claims all blocks that contain immediate write data * so that the SPA knows they're in use. * * All claims *must* be resolved in the first txg -- before the SPA * starts allocating blocks -- so that nothing is allocated twice. * If txg == 0 we just verify that the block is claimable. */ ASSERT3U(spa->spa_uberblock.ub_rootbp.blk_birth, <, spa_first_txg(spa)); ASSERT(txg == spa_first_txg(spa) || txg == 0); ASSERT(!BP_GET_DEDUP(bp) || !spa_writeable(spa)); /* zdb(1M) */ zio = zio_create(pio, spa, txg, bp, NULL, BP_GET_PSIZE(bp), done, private, ZIO_TYPE_CLAIM, ZIO_PRIORITY_NOW, flags, NULL, 0, NULL, ZIO_STAGE_OPEN, ZIO_CLAIM_PIPELINE); return (zio); } zio_t * zio_ioctl(zio_t *pio, spa_t *spa, vdev_t *vd, int cmd, zio_done_func_t *done, void *private, enum zio_flag flags) { zio_t *zio; int c; if (vd->vdev_children == 0) { zio = zio_create(pio, spa, 0, NULL, NULL, 0, done, private, ZIO_TYPE_IOCTL, ZIO_PRIORITY_NOW, flags, vd, 0, NULL, ZIO_STAGE_OPEN, ZIO_IOCTL_PIPELINE); zio->io_cmd = cmd; } else { zio = zio_null(pio, spa, NULL, NULL, NULL, flags); for (c = 0; c < vd->vdev_children; c++) zio_nowait(zio_ioctl(zio, spa, vd->vdev_child[c], cmd, done, private, flags)); } return (zio); } zio_t * zio_read_phys(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size, void *data, int checksum, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, boolean_t labels) { zio_t *zio; ASSERT(vd->vdev_children == 0); ASSERT(!labels || offset + size <= VDEV_LABEL_START_SIZE || offset >= vd->vdev_psize - VDEV_LABEL_END_SIZE); ASSERT3U(offset + size, <=, vd->vdev_psize); zio = zio_create(pio, vd->vdev_spa, 0, NULL, data, size, done, private, ZIO_TYPE_READ, priority, flags | ZIO_FLAG_PHYSICAL, vd, offset, NULL, ZIO_STAGE_OPEN, ZIO_READ_PHYS_PIPELINE); zio->io_prop.zp_checksum = checksum; return (zio); } zio_t * zio_write_phys(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size, void *data, int checksum, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, boolean_t labels) { zio_t *zio; ASSERT(vd->vdev_children == 0); ASSERT(!labels || offset + size <= VDEV_LABEL_START_SIZE || offset >= vd->vdev_psize - VDEV_LABEL_END_SIZE); ASSERT3U(offset + size, <=, vd->vdev_psize); zio = zio_create(pio, vd->vdev_spa, 0, NULL, data, size, done, private, ZIO_TYPE_WRITE, priority, flags | ZIO_FLAG_PHYSICAL, vd, offset, NULL, ZIO_STAGE_OPEN, ZIO_WRITE_PHYS_PIPELINE); zio->io_prop.zp_checksum = checksum; if (zio_checksum_table[checksum].ci_eck) { /* * zec checksums are necessarily destructive -- they modify * the end of the write buffer to hold the verifier/checksum. * Therefore, we must make a local copy in case the data is * being written to multiple places in parallel. */ void *wbuf = zio_buf_alloc(size); bcopy(data, wbuf, size); zio_push_transform(zio, wbuf, size, size, NULL); } return (zio); } /* * Create a child I/O to do some work for us. */ zio_t * zio_vdev_child_io(zio_t *pio, blkptr_t *bp, vdev_t *vd, uint64_t offset, void *data, uint64_t size, int type, zio_priority_t priority, enum zio_flag flags, zio_done_func_t *done, void *private) { enum zio_stage pipeline = ZIO_VDEV_CHILD_PIPELINE; zio_t *zio; ASSERT(vd->vdev_parent == (pio->io_vd ? pio->io_vd : pio->io_spa->spa_root_vdev)); if (type == ZIO_TYPE_READ && bp != NULL) { /* * If we have the bp, then the child should perform the * checksum and the parent need not. This pushes error * detection as close to the leaves as possible and * eliminates redundant checksums in the interior nodes. */ pipeline |= ZIO_STAGE_CHECKSUM_VERIFY; pio->io_pipeline &= ~ZIO_STAGE_CHECKSUM_VERIFY; } if (vd->vdev_children == 0) offset += VDEV_LABEL_START_SIZE; flags |= ZIO_VDEV_CHILD_FLAGS(pio) | ZIO_FLAG_DONT_PROPAGATE; /* * If we've decided to do a repair, the write is not speculative -- * even if the original read was. */ if (flags & ZIO_FLAG_IO_REPAIR) flags &= ~ZIO_FLAG_SPECULATIVE; zio = zio_create(pio, pio->io_spa, pio->io_txg, bp, data, size, done, private, type, priority, flags, vd, offset, &pio->io_bookmark, ZIO_STAGE_VDEV_IO_START >> 1, pipeline); zio->io_physdone = pio->io_physdone; if (vd->vdev_ops->vdev_op_leaf && zio->io_logical != NULL) zio->io_logical->io_phys_children++; return (zio); } zio_t * zio_vdev_delegated_io(vdev_t *vd, uint64_t offset, void *data, uint64_t size, int type, zio_priority_t priority, enum zio_flag flags, zio_done_func_t *done, void *private) { zio_t *zio; ASSERT(vd->vdev_ops->vdev_op_leaf); zio = zio_create(NULL, vd->vdev_spa, 0, NULL, data, size, done, private, type, priority, flags | ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_RETRY | ZIO_FLAG_DELEGATED, vd, offset, NULL, ZIO_STAGE_VDEV_IO_START >> 1, ZIO_VDEV_CHILD_PIPELINE); return (zio); } void zio_flush(zio_t *zio, vdev_t *vd) { zio_nowait(zio_ioctl(zio, zio->io_spa, vd, DKIOCFLUSHWRITECACHE, NULL, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_PROPAGATE | ZIO_FLAG_DONT_RETRY)); } void zio_shrink(zio_t *zio, uint64_t size) { ASSERT(zio->io_executor == NULL); ASSERT(zio->io_orig_size == zio->io_size); ASSERT(size <= zio->io_size); /* * We don't shrink for raidz because of problems with the * reconstruction when reading back less than the block size. * Note, BP_IS_RAIDZ() assumes no compression. */ ASSERT(BP_GET_COMPRESS(zio->io_bp) == ZIO_COMPRESS_OFF); if (!BP_IS_RAIDZ(zio->io_bp)) zio->io_orig_size = zio->io_size = size; } /* * ========================================================================== * Prepare to read and write logical blocks * ========================================================================== */ static int zio_read_bp_init(zio_t *zio) { blkptr_t *bp = zio->io_bp; if (BP_GET_COMPRESS(bp) != ZIO_COMPRESS_OFF && zio->io_child_type == ZIO_CHILD_LOGICAL && !(zio->io_flags & ZIO_FLAG_RAW)) { uint64_t psize = BP_IS_EMBEDDED(bp) ? BPE_GET_PSIZE(bp) : BP_GET_PSIZE(bp); void *cbuf = zio_buf_alloc(psize); zio_push_transform(zio, cbuf, psize, psize, zio_decompress); } if (BP_IS_EMBEDDED(bp) && BPE_GET_ETYPE(bp) == BP_EMBEDDED_TYPE_DATA) { zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; decode_embedded_bp_compressed(bp, zio->io_data); } else { ASSERT(!BP_IS_EMBEDDED(bp)); } if (!DMU_OT_IS_METADATA(BP_GET_TYPE(bp)) && BP_GET_LEVEL(bp) == 0) zio->io_flags |= ZIO_FLAG_DONT_CACHE; if (BP_GET_TYPE(bp) == DMU_OT_DDT_ZAP) zio->io_flags |= ZIO_FLAG_DONT_CACHE; if (BP_GET_DEDUP(bp) && zio->io_child_type == ZIO_CHILD_LOGICAL) zio->io_pipeline = ZIO_DDT_READ_PIPELINE; return (ZIO_PIPELINE_CONTINUE); } static int zio_write_bp_init(zio_t *zio) { spa_t *spa = zio->io_spa; zio_prop_t *zp = &zio->io_prop; enum zio_compress compress = zp->zp_compress; blkptr_t *bp = zio->io_bp; uint64_t lsize = zio->io_size; uint64_t psize = lsize; int pass = 1; /* * If our children haven't all reached the ready stage, * wait for them and then repeat this pipeline stage. */ if (zio_wait_for_children(zio, ZIO_CHILD_GANG, ZIO_WAIT_READY) || zio_wait_for_children(zio, ZIO_CHILD_LOGICAL, ZIO_WAIT_READY)) return (ZIO_PIPELINE_STOP); if (!IO_IS_ALLOCATING(zio)) return (ZIO_PIPELINE_CONTINUE); ASSERT(zio->io_child_type != ZIO_CHILD_DDT); if (zio->io_bp_override) { ASSERT(bp->blk_birth != zio->io_txg); ASSERT(BP_GET_DEDUP(zio->io_bp_override) == 0); *bp = *zio->io_bp_override; zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; if (BP_IS_EMBEDDED(bp)) return (ZIO_PIPELINE_CONTINUE); /* * If we've been overridden and nopwrite is set then * set the flag accordingly to indicate that a nopwrite * has already occurred. */ if (!BP_IS_HOLE(bp) && zp->zp_nopwrite) { ASSERT(!zp->zp_dedup); zio->io_flags |= ZIO_FLAG_NOPWRITE; return (ZIO_PIPELINE_CONTINUE); } ASSERT(!zp->zp_nopwrite); if (BP_IS_HOLE(bp) || !zp->zp_dedup) return (ZIO_PIPELINE_CONTINUE); ASSERT(zio_checksum_table[zp->zp_checksum].ci_dedup || zp->zp_dedup_verify); if (BP_GET_CHECKSUM(bp) == zp->zp_checksum) { BP_SET_DEDUP(bp, 1); zio->io_pipeline |= ZIO_STAGE_DDT_WRITE; return (ZIO_PIPELINE_CONTINUE); } zio->io_bp_override = NULL; BP_ZERO(bp); } if (!BP_IS_HOLE(bp) && bp->blk_birth == zio->io_txg) { /* * We're rewriting an existing block, which means we're * working on behalf of spa_sync(). For spa_sync() to * converge, it must eventually be the case that we don't * have to allocate new blocks. But compression changes * the blocksize, which forces a reallocate, and makes * convergence take longer. Therefore, after the first * few passes, stop compressing to ensure convergence. */ pass = spa_sync_pass(spa); ASSERT(zio->io_txg == spa_syncing_txg(spa)); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(!BP_GET_DEDUP(bp)); if (pass >= zfs_sync_pass_dont_compress) compress = ZIO_COMPRESS_OFF; /* Make sure someone doesn't change their mind on overwrites */ ASSERT(BP_IS_EMBEDDED(bp) || MIN(zp->zp_copies + BP_IS_GANG(bp), spa_max_replication(spa)) == BP_GET_NDVAS(bp)); } if (compress != ZIO_COMPRESS_OFF) { void *cbuf = zio_buf_alloc(lsize); psize = zio_compress_data(compress, zio->io_data, cbuf, lsize); if (psize == 0 || psize == lsize) { compress = ZIO_COMPRESS_OFF; zio_buf_free(cbuf, lsize); } else if (!zp->zp_dedup && psize <= BPE_PAYLOAD_SIZE && zp->zp_level == 0 && !DMU_OT_HAS_FILL(zp->zp_type) && spa_feature_is_enabled(spa, SPA_FEATURE_EMBEDDED_DATA)) { encode_embedded_bp_compressed(bp, cbuf, compress, lsize, psize); BPE_SET_ETYPE(bp, BP_EMBEDDED_TYPE_DATA); BP_SET_TYPE(bp, zio->io_prop.zp_type); BP_SET_LEVEL(bp, zio->io_prop.zp_level); zio_buf_free(cbuf, lsize); bp->blk_birth = zio->io_txg; zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; ASSERT(spa_feature_is_active(spa, SPA_FEATURE_EMBEDDED_DATA)); return (ZIO_PIPELINE_CONTINUE); } else { /* * Round up compressed size to MINBLOCKSIZE and * zero the tail. */ size_t rounded = P2ROUNDUP(psize, (size_t)SPA_MINBLOCKSIZE); if (rounded > psize) { bzero((char *)cbuf + psize, rounded - psize); psize = rounded; } if (psize == lsize) { compress = ZIO_COMPRESS_OFF; zio_buf_free(cbuf, lsize); } else { zio_push_transform(zio, cbuf, psize, lsize, NULL); } } } /* * The final pass of spa_sync() must be all rewrites, but the first * few passes offer a trade-off: allocating blocks defers convergence, * but newly allocated blocks are sequential, so they can be written * to disk faster. Therefore, we allow the first few passes of * spa_sync() to allocate new blocks, but force rewrites after that. * There should only be a handful of blocks after pass 1 in any case. */ if (!BP_IS_HOLE(bp) && bp->blk_birth == zio->io_txg && BP_GET_PSIZE(bp) == psize && pass >= zfs_sync_pass_rewrite) { enum zio_stage gang_stages = zio->io_pipeline & ZIO_GANG_STAGES; ASSERT(psize != 0); zio->io_pipeline = ZIO_REWRITE_PIPELINE | gang_stages; zio->io_flags |= ZIO_FLAG_IO_REWRITE; } else { BP_ZERO(bp); zio->io_pipeline = ZIO_WRITE_PIPELINE; } if (psize == 0) { if (zio->io_bp_orig.blk_birth != 0 && spa_feature_is_active(spa, SPA_FEATURE_HOLE_BIRTH)) { BP_SET_LSIZE(bp, lsize); BP_SET_TYPE(bp, zp->zp_type); BP_SET_LEVEL(bp, zp->zp_level); BP_SET_BIRTH(bp, zio->io_txg, 0); } zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; } else { ASSERT(zp->zp_checksum != ZIO_CHECKSUM_GANG_HEADER); BP_SET_LSIZE(bp, lsize); BP_SET_TYPE(bp, zp->zp_type); BP_SET_LEVEL(bp, zp->zp_level); BP_SET_PSIZE(bp, psize); BP_SET_COMPRESS(bp, compress); BP_SET_CHECKSUM(bp, zp->zp_checksum); BP_SET_DEDUP(bp, zp->zp_dedup); BP_SET_BYTEORDER(bp, ZFS_HOST_BYTEORDER); if (zp->zp_dedup) { ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE)); zio->io_pipeline = ZIO_DDT_WRITE_PIPELINE; } if (zp->zp_nopwrite) { ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE)); zio->io_pipeline |= ZIO_STAGE_NOP_WRITE; } } return (ZIO_PIPELINE_CONTINUE); } static int zio_free_bp_init(zio_t *zio) { blkptr_t *bp = zio->io_bp; if (zio->io_child_type == ZIO_CHILD_LOGICAL) { if (BP_GET_DEDUP(bp)) zio->io_pipeline = ZIO_DDT_FREE_PIPELINE; } return (ZIO_PIPELINE_CONTINUE); } /* * ========================================================================== * Execute the I/O pipeline * ========================================================================== */ static void zio_taskq_dispatch(zio_t *zio, zio_taskq_type_t q, boolean_t cutinline) { spa_t *spa = zio->io_spa; zio_type_t t = zio->io_type; int flags = (cutinline ? TQ_FRONT : 0); /* * If we're a config writer or a probe, the normal issue and * interrupt threads may all be blocked waiting for the config lock. * In this case, select the otherwise-unused taskq for ZIO_TYPE_NULL. */ if (zio->io_flags & (ZIO_FLAG_CONFIG_WRITER | ZIO_FLAG_PROBE)) t = ZIO_TYPE_NULL; /* * A similar issue exists for the L2ARC write thread until L2ARC 2.0. */ if (t == ZIO_TYPE_WRITE && zio->io_vd && zio->io_vd->vdev_aux) t = ZIO_TYPE_NULL; /* * If this is a high priority I/O, then use the high priority taskq if * available. */ if (zio->io_priority == ZIO_PRIORITY_NOW && spa->spa_zio_taskq[t][q + 1].stqs_count != 0) q++; ASSERT3U(q, <, ZIO_TASKQ_TYPES); /* * NB: We are assuming that the zio can only be dispatched * to a single taskq at a time. It would be a grievous error * to dispatch the zio to another taskq at the same time. */ ASSERT(taskq_empty_ent(&zio->io_tqent)); spa_taskq_dispatch_ent(spa, t, q, (task_func_t *)zio_execute, zio, flags, &zio->io_tqent); } static boolean_t zio_taskq_member(zio_t *zio, zio_taskq_type_t q) { kthread_t *executor = zio->io_executor; spa_t *spa = zio->io_spa; zio_type_t t; for (t = 0; t < ZIO_TYPES; t++) { spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q]; uint_t i; for (i = 0; i < tqs->stqs_count; i++) { if (taskq_member(tqs->stqs_taskq[i], executor)) return (B_TRUE); } } return (B_FALSE); } static int zio_issue_async(zio_t *zio) { zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, B_FALSE); return (ZIO_PIPELINE_STOP); } void zio_interrupt(zio_t *zio) { zio_taskq_dispatch(zio, ZIO_TASKQ_INTERRUPT, B_FALSE); } /* * Execute the I/O pipeline until one of the following occurs: * (1) the I/O completes; (2) the pipeline stalls waiting for * dependent child I/Os; (3) the I/O issues, so we're waiting * for an I/O completion interrupt; (4) the I/O is delegated by * vdev-level caching or aggregation; (5) the I/O is deferred * due to vdev-level queueing; (6) the I/O is handed off to * another thread. In all cases, the pipeline stops whenever * there's no CPU work; it never burns a thread in cv_wait_io(). * * There's no locking on io_stage because there's no legitimate way * for multiple threads to be attempting to process the same I/O. */ static zio_pipe_stage_t *zio_pipeline[]; /* * zio_execute() is a wrapper around the static function * __zio_execute() so that we can force __zio_execute() to be * inlined. This reduces stack overhead which is important * because __zio_execute() is called recursively in several zio * code paths. zio_execute() itself cannot be inlined because * it is externally visible. */ void zio_execute(zio_t *zio) { fstrans_cookie_t cookie; cookie = spl_fstrans_mark(); __zio_execute(zio); spl_fstrans_unmark(cookie); } __attribute__((always_inline)) static inline void __zio_execute(zio_t *zio) { zio->io_executor = curthread; while (zio->io_stage < ZIO_STAGE_DONE) { enum zio_stage pipeline = zio->io_pipeline; enum zio_stage stage = zio->io_stage; dsl_pool_t *dp; boolean_t cut; int rv; ASSERT(!MUTEX_HELD(&zio->io_lock)); ASSERT(ISP2(stage)); ASSERT(zio->io_stall == NULL); do { stage <<= 1; } while ((stage & pipeline) == 0); ASSERT(stage <= ZIO_STAGE_DONE); dp = spa_get_dsl(zio->io_spa); cut = (stage == ZIO_STAGE_VDEV_IO_START) ? zio_requeue_io_start_cut_in_line : B_FALSE; /* * If we are in interrupt context and this pipeline stage * will grab a config lock that is held across I/O, * or may wait for an I/O that needs an interrupt thread * to complete, issue async to avoid deadlock. * * For VDEV_IO_START, we cut in line so that the io will * be sent to disk promptly. */ if ((stage & ZIO_BLOCKING_STAGES) && zio->io_vd == NULL && zio_taskq_member(zio, ZIO_TASKQ_INTERRUPT)) { zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, cut); return; } /* * If we executing in the context of the tx_sync_thread, * or we are performing pool initialization outside of a * zio_taskq[ZIO_TASKQ_ISSUE|ZIO_TASKQ_ISSUE_HIGH] context. * Then issue the zio asynchronously to minimize stack usage * for these deep call paths. */ if ((dp && curthread == dp->dp_tx.tx_sync_thread) || (dp && spa_is_initializing(dp->dp_spa) && !zio_taskq_member(zio, ZIO_TASKQ_ISSUE) && !zio_taskq_member(zio, ZIO_TASKQ_ISSUE_HIGH))) { zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, cut); return; } zio->io_stage = stage; rv = zio_pipeline[highbit64(stage) - 1](zio); if (rv == ZIO_PIPELINE_STOP) return; ASSERT(rv == ZIO_PIPELINE_CONTINUE); } } /* * ========================================================================== * Initiate I/O, either sync or async * ========================================================================== */ int zio_wait(zio_t *zio) { int error; ASSERT(zio->io_stage == ZIO_STAGE_OPEN); ASSERT(zio->io_executor == NULL); zio->io_waiter = curthread; __zio_execute(zio); mutex_enter(&zio->io_lock); while (zio->io_executor != NULL) cv_wait_io(&zio->io_cv, &zio->io_lock); mutex_exit(&zio->io_lock); error = zio->io_error; zio_destroy(zio); return (error); } void zio_nowait(zio_t *zio) { ASSERT(zio->io_executor == NULL); if (zio->io_child_type == ZIO_CHILD_LOGICAL && zio_unique_parent(zio) == NULL) { zio_t *pio; /* * This is a logical async I/O with no parent to wait for it. * We add it to the spa_async_root_zio "Godfather" I/O which * will ensure they complete prior to unloading the pool. */ spa_t *spa = zio->io_spa; kpreempt_disable(); pio = spa->spa_async_zio_root[CPU_SEQID]; kpreempt_enable(); zio_add_child(pio, zio); } __zio_execute(zio); } /* * ========================================================================== * Reexecute or suspend/resume failed I/O * ========================================================================== */ static void zio_reexecute(zio_t *pio) { zio_t *cio, *cio_next; int c, w; ASSERT(pio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(pio->io_orig_stage == ZIO_STAGE_OPEN); ASSERT(pio->io_gang_leader == NULL); ASSERT(pio->io_gang_tree == NULL); pio->io_flags = pio->io_orig_flags; pio->io_stage = pio->io_orig_stage; pio->io_pipeline = pio->io_orig_pipeline; pio->io_reexecute = 0; pio->io_flags |= ZIO_FLAG_REEXECUTED; pio->io_error = 0; for (w = 0; w < ZIO_WAIT_TYPES; w++) pio->io_state[w] = 0; for (c = 0; c < ZIO_CHILD_TYPES; c++) pio->io_child_error[c] = 0; if (IO_IS_ALLOCATING(pio)) BP_ZERO(pio->io_bp); /* * As we reexecute pio's children, new children could be created. * New children go to the head of pio's io_child_list, however, * so we will (correctly) not reexecute them. The key is that * the remainder of pio's io_child_list, from 'cio_next' onward, * cannot be affected by any side effects of reexecuting 'cio'. */ for (cio = zio_walk_children(pio); cio != NULL; cio = cio_next) { cio_next = zio_walk_children(pio); mutex_enter(&pio->io_lock); for (w = 0; w < ZIO_WAIT_TYPES; w++) pio->io_children[cio->io_child_type][w]++; mutex_exit(&pio->io_lock); zio_reexecute(cio); } /* * Now that all children have been reexecuted, execute the parent. * We don't reexecute "The Godfather" I/O here as it's the * responsibility of the caller to wait on him. */ if (!(pio->io_flags & ZIO_FLAG_GODFATHER)) __zio_execute(pio); } void zio_suspend(spa_t *spa, zio_t *zio) { if (spa_get_failmode(spa) == ZIO_FAILURE_MODE_PANIC) fm_panic("Pool '%s' has encountered an uncorrectable I/O " "failure and the failure mode property for this pool " "is set to panic.", spa_name(spa)); cmn_err(CE_WARN, "Pool '%s' has encountered an uncorrectable I/O " "failure and has been suspended.\n", spa_name(spa)); zfs_ereport_post(FM_EREPORT_ZFS_IO_FAILURE, spa, NULL, NULL, 0, 0); mutex_enter(&spa->spa_suspend_lock); if (spa->spa_suspend_zio_root == NULL) spa->spa_suspend_zio_root = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE | ZIO_FLAG_GODFATHER); spa->spa_suspended = B_TRUE; if (zio != NULL) { ASSERT(!(zio->io_flags & ZIO_FLAG_GODFATHER)); ASSERT(zio != spa->spa_suspend_zio_root); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(zio_unique_parent(zio) == NULL); ASSERT(zio->io_stage == ZIO_STAGE_DONE); zio_add_child(spa->spa_suspend_zio_root, zio); } mutex_exit(&spa->spa_suspend_lock); } int zio_resume(spa_t *spa) { zio_t *pio; /* * Reexecute all previously suspended i/o. */ mutex_enter(&spa->spa_suspend_lock); spa->spa_suspended = B_FALSE; cv_broadcast(&spa->spa_suspend_cv); pio = spa->spa_suspend_zio_root; spa->spa_suspend_zio_root = NULL; mutex_exit(&spa->spa_suspend_lock); if (pio == NULL) return (0); zio_reexecute(pio); return (zio_wait(pio)); } void zio_resume_wait(spa_t *spa) { mutex_enter(&spa->spa_suspend_lock); while (spa_suspended(spa)) cv_wait(&spa->spa_suspend_cv, &spa->spa_suspend_lock); mutex_exit(&spa->spa_suspend_lock); } /* * ========================================================================== * Gang blocks. * * A gang block is a collection of small blocks that looks to the DMU * like one large block. When zio_dva_allocate() cannot find a block * of the requested size, due to either severe fragmentation or the pool * being nearly full, it calls zio_write_gang_block() to construct the * block from smaller fragments. * * A gang block consists of a gang header (zio_gbh_phys_t) and up to * three (SPA_GBH_NBLKPTRS) gang members. The gang header is just like * an indirect block: it's an array of block pointers. It consumes * only one sector and hence is allocatable regardless of fragmentation. * The gang header's bps point to its gang members, which hold the data. * * Gang blocks are self-checksumming, using the bp's * as the verifier to ensure uniqueness of the SHA256 checksum. * Critically, the gang block bp's blk_cksum is the checksum of the data, * not the gang header. This ensures that data block signatures (needed for * deduplication) are independent of how the block is physically stored. * * Gang blocks can be nested: a gang member may itself be a gang block. * Thus every gang block is a tree in which root and all interior nodes are * gang headers, and the leaves are normal blocks that contain user data. * The root of the gang tree is called the gang leader. * * To perform any operation (read, rewrite, free, claim) on a gang block, * zio_gang_assemble() first assembles the gang tree (minus data leaves) * in the io_gang_tree field of the original logical i/o by recursively * reading the gang leader and all gang headers below it. This yields * an in-core tree containing the contents of every gang header and the * bps for every constituent of the gang block. * * With the gang tree now assembled, zio_gang_issue() just walks the gang tree * and invokes a callback on each bp. To free a gang block, zio_gang_issue() * calls zio_free_gang() -- a trivial wrapper around zio_free() -- for each bp. * zio_claim_gang() provides a similarly trivial wrapper for zio_claim(). * zio_read_gang() is a wrapper around zio_read() that omits reading gang * headers, since we already have those in io_gang_tree. zio_rewrite_gang() * performs a zio_rewrite() of the data or, for gang headers, a zio_rewrite() * of the gang header plus zio_checksum_compute() of the data to update the * gang header's blk_cksum as described above. * * The two-phase assemble/issue model solves the problem of partial failure -- * what if you'd freed part of a gang block but then couldn't read the * gang header for another part? Assembling the entire gang tree first * ensures that all the necessary gang header I/O has succeeded before * starting the actual work of free, claim, or write. Once the gang tree * is assembled, free and claim are in-memory operations that cannot fail. * * In the event that a gang write fails, zio_dva_unallocate() walks the * gang tree to immediately free (i.e. insert back into the space map) * everything we've allocated. This ensures that we don't get ENOSPC * errors during repeated suspend/resume cycles due to a flaky device. * * Gang rewrites only happen during sync-to-convergence. If we can't assemble * the gang tree, we won't modify the block, so we can safely defer the free * (knowing that the block is still intact). If we *can* assemble the gang * tree, then even if some of the rewrites fail, zio_dva_unallocate() will free * each constituent bp and we can allocate a new block on the next sync pass. * * In all cases, the gang tree allows complete recovery from partial failure. * ========================================================================== */ static zio_t * zio_read_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, void *data) { if (gn != NULL) return (pio); return (zio_read(pio, pio->io_spa, bp, data, BP_GET_PSIZE(bp), NULL, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark)); } zio_t * zio_rewrite_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, void *data) { zio_t *zio; if (gn != NULL) { zio = zio_rewrite(pio, pio->io_spa, pio->io_txg, bp, gn->gn_gbh, SPA_GANGBLOCKSIZE, NULL, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); /* * As we rewrite each gang header, the pipeline will compute * a new gang block header checksum for it; but no one will * compute a new data checksum, so we do that here. The one * exception is the gang leader: the pipeline already computed * its data checksum because that stage precedes gang assembly. * (Presently, nothing actually uses interior data checksums; * this is just good hygiene.) */ if (gn != pio->io_gang_leader->io_gang_tree) { zio_checksum_compute(zio, BP_GET_CHECKSUM(bp), data, BP_GET_PSIZE(bp)); } /* * If we are here to damage data for testing purposes, * leave the GBH alone so that we can detect the damage. */ if (pio->io_gang_leader->io_flags & ZIO_FLAG_INDUCE_DAMAGE) zio->io_pipeline &= ~ZIO_VDEV_IO_STAGES; } else { zio = zio_rewrite(pio, pio->io_spa, pio->io_txg, bp, data, BP_GET_PSIZE(bp), NULL, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); } return (zio); } /* ARGSUSED */ zio_t * zio_free_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, void *data) { return (zio_free_sync(pio, pio->io_spa, pio->io_txg, bp, ZIO_GANG_CHILD_FLAGS(pio))); } /* ARGSUSED */ zio_t * zio_claim_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, void *data) { return (zio_claim(pio, pio->io_spa, pio->io_txg, bp, NULL, NULL, ZIO_GANG_CHILD_FLAGS(pio))); } static zio_gang_issue_func_t *zio_gang_issue_func[ZIO_TYPES] = { NULL, zio_read_gang, zio_rewrite_gang, zio_free_gang, zio_claim_gang, NULL }; static void zio_gang_tree_assemble_done(zio_t *zio); static zio_gang_node_t * zio_gang_node_alloc(zio_gang_node_t **gnpp) { zio_gang_node_t *gn; ASSERT(*gnpp == NULL); gn = kmem_zalloc(sizeof (*gn), KM_PUSHPAGE); gn->gn_gbh = zio_buf_alloc(SPA_GANGBLOCKSIZE); *gnpp = gn; return (gn); } static void zio_gang_node_free(zio_gang_node_t **gnpp) { zio_gang_node_t *gn = *gnpp; int g; for (g = 0; g < SPA_GBH_NBLKPTRS; g++) ASSERT(gn->gn_child[g] == NULL); zio_buf_free(gn->gn_gbh, SPA_GANGBLOCKSIZE); kmem_free(gn, sizeof (*gn)); *gnpp = NULL; } static void zio_gang_tree_free(zio_gang_node_t **gnpp) { zio_gang_node_t *gn = *gnpp; int g; if (gn == NULL) return; for (g = 0; g < SPA_GBH_NBLKPTRS; g++) zio_gang_tree_free(&gn->gn_child[g]); zio_gang_node_free(gnpp); } static void zio_gang_tree_assemble(zio_t *gio, blkptr_t *bp, zio_gang_node_t **gnpp) { zio_gang_node_t *gn = zio_gang_node_alloc(gnpp); ASSERT(gio->io_gang_leader == gio); ASSERT(BP_IS_GANG(bp)); zio_nowait(zio_read(gio, gio->io_spa, bp, gn->gn_gbh, SPA_GANGBLOCKSIZE, zio_gang_tree_assemble_done, gn, gio->io_priority, ZIO_GANG_CHILD_FLAGS(gio), &gio->io_bookmark)); } static void zio_gang_tree_assemble_done(zio_t *zio) { zio_t *gio = zio->io_gang_leader; zio_gang_node_t *gn = zio->io_private; blkptr_t *bp = zio->io_bp; int g; ASSERT(gio == zio_unique_parent(zio)); ASSERT(zio->io_child_count == 0); if (zio->io_error) return; if (BP_SHOULD_BYTESWAP(bp)) byteswap_uint64_array(zio->io_data, zio->io_size); ASSERT(zio->io_data == gn->gn_gbh); ASSERT(zio->io_size == SPA_GANGBLOCKSIZE); ASSERT(gn->gn_gbh->zg_tail.zec_magic == ZEC_MAGIC); for (g = 0; g < SPA_GBH_NBLKPTRS; g++) { blkptr_t *gbp = &gn->gn_gbh->zg_blkptr[g]; if (!BP_IS_GANG(gbp)) continue; zio_gang_tree_assemble(gio, gbp, &gn->gn_child[g]); } } static void zio_gang_tree_issue(zio_t *pio, zio_gang_node_t *gn, blkptr_t *bp, void *data) { zio_t *gio = pio->io_gang_leader; zio_t *zio; int g; ASSERT(BP_IS_GANG(bp) == !!gn); ASSERT(BP_GET_CHECKSUM(bp) == BP_GET_CHECKSUM(gio->io_bp)); ASSERT(BP_GET_LSIZE(bp) == BP_GET_PSIZE(bp) || gn == gio->io_gang_tree); /* * If you're a gang header, your data is in gn->gn_gbh. * If you're a gang member, your data is in 'data' and gn == NULL. */ zio = zio_gang_issue_func[gio->io_type](pio, bp, gn, data); if (gn != NULL) { ASSERT(gn->gn_gbh->zg_tail.zec_magic == ZEC_MAGIC); for (g = 0; g < SPA_GBH_NBLKPTRS; g++) { blkptr_t *gbp = &gn->gn_gbh->zg_blkptr[g]; if (BP_IS_HOLE(gbp)) continue; zio_gang_tree_issue(zio, gn->gn_child[g], gbp, data); data = (char *)data + BP_GET_PSIZE(gbp); } } if (gn == gio->io_gang_tree) ASSERT3P((char *)gio->io_data + gio->io_size, ==, data); if (zio != pio) zio_nowait(zio); } static int zio_gang_assemble(zio_t *zio) { blkptr_t *bp = zio->io_bp; ASSERT(BP_IS_GANG(bp) && zio->io_gang_leader == NULL); ASSERT(zio->io_child_type > ZIO_CHILD_GANG); zio->io_gang_leader = zio; zio_gang_tree_assemble(zio, bp, &zio->io_gang_tree); return (ZIO_PIPELINE_CONTINUE); } static int zio_gang_issue(zio_t *zio) { blkptr_t *bp = zio->io_bp; if (zio_wait_for_children(zio, ZIO_CHILD_GANG, ZIO_WAIT_DONE)) return (ZIO_PIPELINE_STOP); ASSERT(BP_IS_GANG(bp) && zio->io_gang_leader == zio); ASSERT(zio->io_child_type > ZIO_CHILD_GANG); if (zio->io_child_error[ZIO_CHILD_GANG] == 0) zio_gang_tree_issue(zio, zio->io_gang_tree, bp, zio->io_data); else zio_gang_tree_free(&zio->io_gang_tree); zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; return (ZIO_PIPELINE_CONTINUE); } static void zio_write_gang_member_ready(zio_t *zio) { zio_t *pio = zio_unique_parent(zio); dva_t *cdva = zio->io_bp->blk_dva; dva_t *pdva = pio->io_bp->blk_dva; uint64_t asize; int d; ASSERTV(zio_t *gio = zio->io_gang_leader); if (BP_IS_HOLE(zio->io_bp)) return; ASSERT(BP_IS_HOLE(&zio->io_bp_orig)); ASSERT(zio->io_child_type == ZIO_CHILD_GANG); ASSERT3U(zio->io_prop.zp_copies, ==, gio->io_prop.zp_copies); ASSERT3U(zio->io_prop.zp_copies, <=, BP_GET_NDVAS(zio->io_bp)); ASSERT3U(pio->io_prop.zp_copies, <=, BP_GET_NDVAS(pio->io_bp)); ASSERT3U(BP_GET_NDVAS(zio->io_bp), <=, BP_GET_NDVAS(pio->io_bp)); mutex_enter(&pio->io_lock); for (d = 0; d < BP_GET_NDVAS(zio->io_bp); d++) { ASSERT(DVA_GET_GANG(&pdva[d])); asize = DVA_GET_ASIZE(&pdva[d]); asize += DVA_GET_ASIZE(&cdva[d]); DVA_SET_ASIZE(&pdva[d], asize); } mutex_exit(&pio->io_lock); } static int zio_write_gang_block(zio_t *pio) { spa_t *spa = pio->io_spa; blkptr_t *bp = pio->io_bp; zio_t *gio = pio->io_gang_leader; zio_t *zio; zio_gang_node_t *gn, **gnpp; zio_gbh_phys_t *gbh; uint64_t txg = pio->io_txg; uint64_t resid = pio->io_size; uint64_t lsize; int copies = gio->io_prop.zp_copies; int gbh_copies = MIN(copies + 1, spa_max_replication(spa)); zio_prop_t zp; int g, error; error = metaslab_alloc(spa, spa_normal_class(spa), SPA_GANGBLOCKSIZE, bp, gbh_copies, txg, pio == gio ? NULL : gio->io_bp, METASLAB_HINTBP_FAVOR | METASLAB_GANG_HEADER); if (error) { pio->io_error = error; return (ZIO_PIPELINE_CONTINUE); } if (pio == gio) { gnpp = &gio->io_gang_tree; } else { gnpp = pio->io_private; ASSERT(pio->io_ready == zio_write_gang_member_ready); } gn = zio_gang_node_alloc(gnpp); gbh = gn->gn_gbh; bzero(gbh, SPA_GANGBLOCKSIZE); /* * Create the gang header. */ zio = zio_rewrite(pio, spa, txg, bp, gbh, SPA_GANGBLOCKSIZE, NULL, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); /* * Create and nowait the gang children. */ for (g = 0; resid != 0; resid -= lsize, g++) { lsize = P2ROUNDUP(resid / (SPA_GBH_NBLKPTRS - g), SPA_MINBLOCKSIZE); ASSERT(lsize >= SPA_MINBLOCKSIZE && lsize <= resid); zp.zp_checksum = gio->io_prop.zp_checksum; zp.zp_compress = ZIO_COMPRESS_OFF; zp.zp_type = DMU_OT_NONE; zp.zp_level = 0; zp.zp_copies = gio->io_prop.zp_copies; zp.zp_dedup = B_FALSE; zp.zp_dedup_verify = B_FALSE; zp.zp_nopwrite = B_FALSE; zio_nowait(zio_write(zio, spa, txg, &gbh->zg_blkptr[g], (char *)pio->io_data + (pio->io_size - resid), lsize, &zp, zio_write_gang_member_ready, NULL, NULL, &gn->gn_child[g], pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark)); } /* * Set pio's pipeline to just wait for zio to finish. */ pio->io_pipeline = ZIO_INTERLOCK_PIPELINE; /* * We didn't allocate this bp, so make sure it doesn't get unmarked. */ pio->io_flags &= ~ZIO_FLAG_FASTWRITE; zio_nowait(zio); return (ZIO_PIPELINE_CONTINUE); } /* * The zio_nop_write stage in the pipeline determines if allocating * a new bp is necessary. By leveraging a cryptographically secure checksum, * such as SHA256, we can compare the checksums of the new data and the old * to determine if allocating a new block is required. The nopwrite * feature can handle writes in either syncing or open context (i.e. zil * writes) and as a result is mutually exclusive with dedup. */ static int zio_nop_write(zio_t *zio) { blkptr_t *bp = zio->io_bp; blkptr_t *bp_orig = &zio->io_bp_orig; zio_prop_t *zp = &zio->io_prop; ASSERT(BP_GET_LEVEL(bp) == 0); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE)); ASSERT(zp->zp_nopwrite); ASSERT(!zp->zp_dedup); ASSERT(zio->io_bp_override == NULL); ASSERT(IO_IS_ALLOCATING(zio)); /* * Check to see if the original bp and the new bp have matching * characteristics (i.e. same checksum, compression algorithms, etc). * If they don't then just continue with the pipeline which will * allocate a new bp. */ if (BP_IS_HOLE(bp_orig) || !zio_checksum_table[BP_GET_CHECKSUM(bp)].ci_dedup || BP_GET_CHECKSUM(bp) != BP_GET_CHECKSUM(bp_orig) || BP_GET_COMPRESS(bp) != BP_GET_COMPRESS(bp_orig) || BP_GET_DEDUP(bp) != BP_GET_DEDUP(bp_orig) || zp->zp_copies != BP_GET_NDVAS(bp_orig)) return (ZIO_PIPELINE_CONTINUE); /* * If the checksums match then reset the pipeline so that we * avoid allocating a new bp and issuing any I/O. */ if (ZIO_CHECKSUM_EQUAL(bp->blk_cksum, bp_orig->blk_cksum)) { ASSERT(zio_checksum_table[zp->zp_checksum].ci_dedup); ASSERT3U(BP_GET_PSIZE(bp), ==, BP_GET_PSIZE(bp_orig)); ASSERT3U(BP_GET_LSIZE(bp), ==, BP_GET_LSIZE(bp_orig)); ASSERT(zp->zp_compress != ZIO_COMPRESS_OFF); ASSERT(bcmp(&bp->blk_prop, &bp_orig->blk_prop, sizeof (uint64_t)) == 0); *bp = *bp_orig; zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; zio->io_flags |= ZIO_FLAG_NOPWRITE; } return (ZIO_PIPELINE_CONTINUE); } /* * ========================================================================== * Dedup * ========================================================================== */ static void zio_ddt_child_read_done(zio_t *zio) { blkptr_t *bp = zio->io_bp; ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp; zio_t *pio = zio_unique_parent(zio); mutex_enter(&pio->io_lock); ddp = ddt_phys_select(dde, bp); if (zio->io_error == 0) ddt_phys_clear(ddp); /* this ddp doesn't need repair */ if (zio->io_error == 0 && dde->dde_repair_data == NULL) dde->dde_repair_data = zio->io_data; else zio_buf_free(zio->io_data, zio->io_size); mutex_exit(&pio->io_lock); } static int zio_ddt_read_start(zio_t *zio) { blkptr_t *bp = zio->io_bp; int p; ASSERT(BP_GET_DEDUP(bp)); ASSERT(BP_GET_PSIZE(bp) == zio->io_size); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); if (zio->io_child_error[ZIO_CHILD_DDT]) { ddt_t *ddt = ddt_select(zio->io_spa, bp); ddt_entry_t *dde = ddt_repair_start(ddt, bp); ddt_phys_t *ddp = dde->dde_phys; ddt_phys_t *ddp_self = ddt_phys_select(dde, bp); blkptr_t blk; ASSERT(zio->io_vsd == NULL); zio->io_vsd = dde; if (ddp_self == NULL) return (ZIO_PIPELINE_CONTINUE); for (p = 0; p < DDT_PHYS_TYPES; p++, ddp++) { if (ddp->ddp_phys_birth == 0 || ddp == ddp_self) continue; ddt_bp_create(ddt->ddt_checksum, &dde->dde_key, ddp, &blk); zio_nowait(zio_read(zio, zio->io_spa, &blk, zio_buf_alloc(zio->io_size), zio->io_size, zio_ddt_child_read_done, dde, zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio) | ZIO_FLAG_DONT_PROPAGATE, &zio->io_bookmark)); } return (ZIO_PIPELINE_CONTINUE); } zio_nowait(zio_read(zio, zio->io_spa, bp, zio->io_data, zio->io_size, NULL, NULL, zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio), &zio->io_bookmark)); return (ZIO_PIPELINE_CONTINUE); } static int zio_ddt_read_done(zio_t *zio) { blkptr_t *bp = zio->io_bp; if (zio_wait_for_children(zio, ZIO_CHILD_DDT, ZIO_WAIT_DONE)) return (ZIO_PIPELINE_STOP); ASSERT(BP_GET_DEDUP(bp)); ASSERT(BP_GET_PSIZE(bp) == zio->io_size); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); if (zio->io_child_error[ZIO_CHILD_DDT]) { ddt_t *ddt = ddt_select(zio->io_spa, bp); ddt_entry_t *dde = zio->io_vsd; if (ddt == NULL) { ASSERT(spa_load_state(zio->io_spa) != SPA_LOAD_NONE); return (ZIO_PIPELINE_CONTINUE); } if (dde == NULL) { zio->io_stage = ZIO_STAGE_DDT_READ_START >> 1; zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, B_FALSE); return (ZIO_PIPELINE_STOP); } if (dde->dde_repair_data != NULL) { bcopy(dde->dde_repair_data, zio->io_data, zio->io_size); zio->io_child_error[ZIO_CHILD_DDT] = 0; } ddt_repair_done(ddt, dde); zio->io_vsd = NULL; } ASSERT(zio->io_vsd == NULL); return (ZIO_PIPELINE_CONTINUE); } static boolean_t zio_ddt_collision(zio_t *zio, ddt_t *ddt, ddt_entry_t *dde) { spa_t *spa = zio->io_spa; int p; /* * Note: we compare the original data, not the transformed data, * because when zio->io_bp is an override bp, we will not have * pushed the I/O transforms. That's an important optimization * because otherwise we'd compress/encrypt all dmu_sync() data twice. */ for (p = DDT_PHYS_SINGLE; p <= DDT_PHYS_TRIPLE; p++) { zio_t *lio = dde->dde_lead_zio[p]; if (lio != NULL) { return (lio->io_orig_size != zio->io_orig_size || bcmp(zio->io_orig_data, lio->io_orig_data, zio->io_orig_size) != 0); } } for (p = DDT_PHYS_SINGLE; p <= DDT_PHYS_TRIPLE; p++) { ddt_phys_t *ddp = &dde->dde_phys[p]; if (ddp->ddp_phys_birth != 0) { arc_buf_t *abuf = NULL; uint32_t aflags = ARC_WAIT; blkptr_t blk = *zio->io_bp; int error; ddt_bp_fill(ddp, &blk, ddp->ddp_phys_birth); ddt_exit(ddt); error = arc_read(NULL, spa, &blk, arc_getbuf_func, &abuf, ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE, &aflags, &zio->io_bookmark); if (error == 0) { if (arc_buf_size(abuf) != zio->io_orig_size || bcmp(abuf->b_data, zio->io_orig_data, zio->io_orig_size) != 0) error = SET_ERROR(EEXIST); VERIFY(arc_buf_remove_ref(abuf, &abuf)); } ddt_enter(ddt); return (error != 0); } } return (B_FALSE); } static void zio_ddt_child_write_ready(zio_t *zio) { int p = zio->io_prop.zp_copies; ddt_t *ddt = ddt_select(zio->io_spa, zio->io_bp); ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp = &dde->dde_phys[p]; zio_t *pio; if (zio->io_error) return; ddt_enter(ddt); ASSERT(dde->dde_lead_zio[p] == zio); ddt_phys_fill(ddp, zio->io_bp); while ((pio = zio_walk_parents(zio)) != NULL) ddt_bp_fill(ddp, pio->io_bp, zio->io_txg); ddt_exit(ddt); } static void zio_ddt_child_write_done(zio_t *zio) { int p = zio->io_prop.zp_copies; ddt_t *ddt = ddt_select(zio->io_spa, zio->io_bp); ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp = &dde->dde_phys[p]; ddt_enter(ddt); ASSERT(ddp->ddp_refcnt == 0); ASSERT(dde->dde_lead_zio[p] == zio); dde->dde_lead_zio[p] = NULL; if (zio->io_error == 0) { while (zio_walk_parents(zio) != NULL) ddt_phys_addref(ddp); } else { ddt_phys_clear(ddp); } ddt_exit(ddt); } static void zio_ddt_ditto_write_done(zio_t *zio) { int p = DDT_PHYS_DITTO; blkptr_t *bp = zio->io_bp; ddt_t *ddt = ddt_select(zio->io_spa, bp); ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp = &dde->dde_phys[p]; ddt_key_t *ddk = &dde->dde_key; ASSERTV(zio_prop_t *zp = &zio->io_prop); ddt_enter(ddt); ASSERT(ddp->ddp_refcnt == 0); ASSERT(dde->dde_lead_zio[p] == zio); dde->dde_lead_zio[p] = NULL; if (zio->io_error == 0) { ASSERT(ZIO_CHECKSUM_EQUAL(bp->blk_cksum, ddk->ddk_cksum)); ASSERT(zp->zp_copies < SPA_DVAS_PER_BP); ASSERT(zp->zp_copies == BP_GET_NDVAS(bp) - BP_IS_GANG(bp)); if (ddp->ddp_phys_birth != 0) ddt_phys_free(ddt, ddk, ddp, zio->io_txg); ddt_phys_fill(ddp, bp); } ddt_exit(ddt); } static int zio_ddt_write(zio_t *zio) { spa_t *spa = zio->io_spa; blkptr_t *bp = zio->io_bp; uint64_t txg = zio->io_txg; zio_prop_t *zp = &zio->io_prop; int p = zp->zp_copies; int ditto_copies; zio_t *cio = NULL; zio_t *dio = NULL; ddt_t *ddt = ddt_select(spa, bp); ddt_entry_t *dde; ddt_phys_t *ddp; ASSERT(BP_GET_DEDUP(bp)); ASSERT(BP_GET_CHECKSUM(bp) == zp->zp_checksum); ASSERT(BP_IS_HOLE(bp) || zio->io_bp_override); ddt_enter(ddt); dde = ddt_lookup(ddt, bp, B_TRUE); ddp = &dde->dde_phys[p]; if (zp->zp_dedup_verify && zio_ddt_collision(zio, ddt, dde)) { /* * If we're using a weak checksum, upgrade to a strong checksum * and try again. If we're already using a strong checksum, * we can't resolve it, so just convert to an ordinary write. * (And automatically e-mail a paper to Nature?) */ if (!zio_checksum_table[zp->zp_checksum].ci_dedup) { zp->zp_checksum = spa_dedup_checksum(spa); zio_pop_transforms(zio); zio->io_stage = ZIO_STAGE_OPEN; BP_ZERO(bp); } else { zp->zp_dedup = B_FALSE; } zio->io_pipeline = ZIO_WRITE_PIPELINE; ddt_exit(ddt); return (ZIO_PIPELINE_CONTINUE); } ditto_copies = ddt_ditto_copies_needed(ddt, dde, ddp); ASSERT(ditto_copies < SPA_DVAS_PER_BP); if (ditto_copies > ddt_ditto_copies_present(dde) && dde->dde_lead_zio[DDT_PHYS_DITTO] == NULL) { zio_prop_t czp = *zp; czp.zp_copies = ditto_copies; /* * If we arrived here with an override bp, we won't have run * the transform stack, so we won't have the data we need to * generate a child i/o. So, toss the override bp and restart. * This is safe, because using the override bp is just an * optimization; and it's rare, so the cost doesn't matter. */ if (zio->io_bp_override) { zio_pop_transforms(zio); zio->io_stage = ZIO_STAGE_OPEN; zio->io_pipeline = ZIO_WRITE_PIPELINE; zio->io_bp_override = NULL; BP_ZERO(bp); ddt_exit(ddt); return (ZIO_PIPELINE_CONTINUE); } dio = zio_write(zio, spa, txg, bp, zio->io_orig_data, zio->io_orig_size, &czp, NULL, NULL, zio_ddt_ditto_write_done, dde, zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio), &zio->io_bookmark); zio_push_transform(dio, zio->io_data, zio->io_size, 0, NULL); dde->dde_lead_zio[DDT_PHYS_DITTO] = dio; } if (ddp->ddp_phys_birth != 0 || dde->dde_lead_zio[p] != NULL) { if (ddp->ddp_phys_birth != 0) ddt_bp_fill(ddp, bp, txg); if (dde->dde_lead_zio[p] != NULL) zio_add_child(zio, dde->dde_lead_zio[p]); else ddt_phys_addref(ddp); } else if (zio->io_bp_override) { ASSERT(bp->blk_birth == txg); ASSERT(BP_EQUAL(bp, zio->io_bp_override)); ddt_phys_fill(ddp, bp); ddt_phys_addref(ddp); } else { cio = zio_write(zio, spa, txg, bp, zio->io_orig_data, zio->io_orig_size, zp, zio_ddt_child_write_ready, NULL, zio_ddt_child_write_done, dde, zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio), &zio->io_bookmark); zio_push_transform(cio, zio->io_data, zio->io_size, 0, NULL); dde->dde_lead_zio[p] = cio; } ddt_exit(ddt); if (cio) zio_nowait(cio); if (dio) zio_nowait(dio); return (ZIO_PIPELINE_CONTINUE); } ddt_entry_t *freedde; /* for debugging */ static int zio_ddt_free(zio_t *zio) { spa_t *spa = zio->io_spa; blkptr_t *bp = zio->io_bp; ddt_t *ddt = ddt_select(spa, bp); ddt_entry_t *dde; ddt_phys_t *ddp; ASSERT(BP_GET_DEDUP(bp)); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ddt_enter(ddt); freedde = dde = ddt_lookup(ddt, bp, B_TRUE); if (dde) { ddp = ddt_phys_select(dde, bp); if (ddp) ddt_phys_decref(ddp); } ddt_exit(ddt); return (ZIO_PIPELINE_CONTINUE); } /* * ========================================================================== * Allocate and free blocks * ========================================================================== */ static int zio_dva_allocate(zio_t *zio) { spa_t *spa = zio->io_spa; metaslab_class_t *mc = spa_normal_class(spa); blkptr_t *bp = zio->io_bp; int error; int flags = 0; if (zio->io_gang_leader == NULL) { ASSERT(zio->io_child_type > ZIO_CHILD_GANG); zio->io_gang_leader = zio; } ASSERT(BP_IS_HOLE(bp)); ASSERT0(BP_GET_NDVAS(bp)); ASSERT3U(zio->io_prop.zp_copies, >, 0); ASSERT3U(zio->io_prop.zp_copies, <=, spa_max_replication(spa)); ASSERT3U(zio->io_size, ==, BP_GET_PSIZE(bp)); /* * The dump device does not support gang blocks so allocation on * behalf of the dump device (i.e. ZIO_FLAG_NODATA) must avoid * the "fast" gang feature. */ flags |= (zio->io_flags & ZIO_FLAG_NODATA) ? METASLAB_GANG_AVOID : 0; flags |= (zio->io_flags & ZIO_FLAG_GANG_CHILD) ? METASLAB_GANG_CHILD : 0; flags |= (zio->io_flags & ZIO_FLAG_FASTWRITE) ? METASLAB_FASTWRITE : 0; error = metaslab_alloc(spa, mc, zio->io_size, bp, zio->io_prop.zp_copies, zio->io_txg, NULL, flags); if (error) { spa_dbgmsg(spa, "%s: metaslab allocation failure: zio %p, " "size %llu, error %d", spa_name(spa), zio, zio->io_size, error); if (error == ENOSPC && zio->io_size > SPA_MINBLOCKSIZE) return (zio_write_gang_block(zio)); zio->io_error = error; } return (ZIO_PIPELINE_CONTINUE); } static int zio_dva_free(zio_t *zio) { metaslab_free(zio->io_spa, zio->io_bp, zio->io_txg, B_FALSE); return (ZIO_PIPELINE_CONTINUE); } static int zio_dva_claim(zio_t *zio) { int error; error = metaslab_claim(zio->io_spa, zio->io_bp, zio->io_txg); if (error) zio->io_error = error; return (ZIO_PIPELINE_CONTINUE); } /* * Undo an allocation. This is used by zio_done() when an I/O fails * and we want to give back the block we just allocated. * This handles both normal blocks and gang blocks. */ static void zio_dva_unallocate(zio_t *zio, zio_gang_node_t *gn, blkptr_t *bp) { int g; ASSERT(bp->blk_birth == zio->io_txg || BP_IS_HOLE(bp)); ASSERT(zio->io_bp_override == NULL); if (!BP_IS_HOLE(bp)) metaslab_free(zio->io_spa, bp, bp->blk_birth, B_TRUE); if (gn != NULL) { for (g = 0; g < SPA_GBH_NBLKPTRS; g++) { zio_dva_unallocate(zio, gn->gn_child[g], &gn->gn_gbh->zg_blkptr[g]); } } } /* * Try to allocate an intent log block. Return 0 on success, errno on failure. */ int zio_alloc_zil(spa_t *spa, uint64_t txg, blkptr_t *new_bp, uint64_t size, boolean_t use_slog) { int error = 1; ASSERT(txg > spa_syncing_txg(spa)); /* * ZIL blocks are always contiguous (i.e. not gang blocks) so we * set the METASLAB_GANG_AVOID flag so that they don't "fast gang" * when allocating them. */ if (use_slog) { error = metaslab_alloc(spa, spa_log_class(spa), size, new_bp, 1, txg, NULL, METASLAB_FASTWRITE | METASLAB_GANG_AVOID); } if (error) { error = metaslab_alloc(spa, spa_normal_class(spa), size, new_bp, 1, txg, NULL, METASLAB_FASTWRITE); } if (error == 0) { BP_SET_LSIZE(new_bp, size); BP_SET_PSIZE(new_bp, size); BP_SET_COMPRESS(new_bp, ZIO_COMPRESS_OFF); BP_SET_CHECKSUM(new_bp, spa_version(spa) >= SPA_VERSION_SLIM_ZIL ? ZIO_CHECKSUM_ZILOG2 : ZIO_CHECKSUM_ZILOG); BP_SET_TYPE(new_bp, DMU_OT_INTENT_LOG); BP_SET_LEVEL(new_bp, 0); BP_SET_DEDUP(new_bp, 0); BP_SET_BYTEORDER(new_bp, ZFS_HOST_BYTEORDER); } return (error); } /* * Free an intent log block. */ void zio_free_zil(spa_t *spa, uint64_t txg, blkptr_t *bp) { ASSERT(BP_GET_TYPE(bp) == DMU_OT_INTENT_LOG); ASSERT(!BP_IS_GANG(bp)); zio_free(spa, txg, bp); } /* * ========================================================================== * Read and write to physical devices * ========================================================================== */ static int zio_vdev_io_start(zio_t *zio) { vdev_t *vd = zio->io_vd; uint64_t align; spa_t *spa = zio->io_spa; ASSERT(zio->io_error == 0); ASSERT(zio->io_child_error[ZIO_CHILD_VDEV] == 0); if (vd == NULL) { if (!(zio->io_flags & ZIO_FLAG_CONFIG_WRITER)) spa_config_enter(spa, SCL_ZIO, zio, RW_READER); /* * The mirror_ops handle multiple DVAs in a single BP. */ return (vdev_mirror_ops.vdev_op_io_start(zio)); } /* * We keep track of time-sensitive I/Os so that the scan thread * can quickly react to certain workloads. In particular, we care * about non-scrubbing, top-level reads and writes with the following * characteristics: * - synchronous writes of user data to non-slog devices * - any reads of user data * When these conditions are met, adjust the timestamp of spa_last_io * which allows the scan thread to adjust its workload accordingly. */ if (!(zio->io_flags & ZIO_FLAG_SCAN_THREAD) && zio->io_bp != NULL && vd == vd->vdev_top && !vd->vdev_islog && zio->io_bookmark.zb_objset != DMU_META_OBJSET && zio->io_txg != spa_syncing_txg(spa)) { uint64_t old = spa->spa_last_io; uint64_t new = ddi_get_lbolt64(); if (old != new) (void) atomic_cas_64(&spa->spa_last_io, old, new); } align = 1ULL << vd->vdev_top->vdev_ashift; if (!(zio->io_flags & ZIO_FLAG_PHYSICAL) && P2PHASE(zio->io_size, align) != 0) { /* Transform logical writes to be a full physical block size. */ uint64_t asize = P2ROUNDUP(zio->io_size, align); char *abuf = zio_buf_alloc(asize); ASSERT(vd == vd->vdev_top); if (zio->io_type == ZIO_TYPE_WRITE) { bcopy(zio->io_data, abuf, zio->io_size); bzero(abuf + zio->io_size, asize - zio->io_size); } zio_push_transform(zio, abuf, asize, asize, zio_subblock); } /* * If this is not a physical io, make sure that it is properly aligned * before proceeding. */ if (!(zio->io_flags & ZIO_FLAG_PHYSICAL)) { ASSERT0(P2PHASE(zio->io_offset, align)); ASSERT0(P2PHASE(zio->io_size, align)); } else { /* * For physical writes, we allow 512b aligned writes and assume * the device will perform a read-modify-write as necessary. */ ASSERT0(P2PHASE(zio->io_offset, SPA_MINBLOCKSIZE)); ASSERT0(P2PHASE(zio->io_size, SPA_MINBLOCKSIZE)); } VERIFY(zio->io_type != ZIO_TYPE_WRITE || spa_writeable(spa)); /* * If this is a repair I/O, and there's no self-healing involved -- * that is, we're just resilvering what we expect to resilver -- * then don't do the I/O unless zio's txg is actually in vd's DTL. * This prevents spurious resilvering with nested replication. * For example, given a mirror of mirrors, (A+B)+(C+D), if only * A is out of date, we'll read from C+D, then use the data to * resilver A+B -- but we don't actually want to resilver B, just A. * The top-level mirror has no way to know this, so instead we just * discard unnecessary repairs as we work our way down the vdev tree. * The same logic applies to any form of nested replication: * ditto + mirror, RAID-Z + replacing, etc. This covers them all. */ if ((zio->io_flags & ZIO_FLAG_IO_REPAIR) && !(zio->io_flags & ZIO_FLAG_SELF_HEAL) && zio->io_txg != 0 && /* not a delegated i/o */ !vdev_dtl_contains(vd, DTL_PARTIAL, zio->io_txg, 1)) { ASSERT(zio->io_type == ZIO_TYPE_WRITE); zio_vdev_io_bypass(zio); return (ZIO_PIPELINE_CONTINUE); } if (vd->vdev_ops->vdev_op_leaf && (zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_WRITE)) { if (zio->io_type == ZIO_TYPE_READ && vdev_cache_read(zio)) return (ZIO_PIPELINE_CONTINUE); if ((zio = vdev_queue_io(zio)) == NULL) return (ZIO_PIPELINE_STOP); if (!vdev_accessible(vd, zio)) { zio->io_error = SET_ERROR(ENXIO); zio_interrupt(zio); return (ZIO_PIPELINE_STOP); } } return (vd->vdev_ops->vdev_op_io_start(zio)); } static int zio_vdev_io_done(zio_t *zio) { vdev_t *vd = zio->io_vd; vdev_ops_t *ops = vd ? vd->vdev_ops : &vdev_mirror_ops; boolean_t unexpected_error = B_FALSE; if (zio_wait_for_children(zio, ZIO_CHILD_VDEV, ZIO_WAIT_DONE)) return (ZIO_PIPELINE_STOP); ASSERT(zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_WRITE); if (vd != NULL && vd->vdev_ops->vdev_op_leaf) { vdev_queue_io_done(zio); if (zio->io_type == ZIO_TYPE_WRITE) vdev_cache_write(zio); if (zio_injection_enabled && zio->io_error == 0) zio->io_error = zio_handle_device_injection(vd, zio, EIO); if (zio_injection_enabled && zio->io_error == 0) zio->io_error = zio_handle_label_injection(zio, EIO); if (zio->io_error) { if (!vdev_accessible(vd, zio)) { zio->io_error = SET_ERROR(ENXIO); } else { unexpected_error = B_TRUE; } } } ops->vdev_op_io_done(zio); if (unexpected_error) VERIFY(vdev_probe(vd, zio) == NULL); return (ZIO_PIPELINE_CONTINUE); } /* * For non-raidz ZIOs, we can just copy aside the bad data read from the * disk, and use that to finish the checksum ereport later. */ static void zio_vsd_default_cksum_finish(zio_cksum_report_t *zcr, const void *good_buf) { /* no processing needed */ zfs_ereport_finish_checksum(zcr, good_buf, zcr->zcr_cbdata, B_FALSE); } /*ARGSUSED*/ void zio_vsd_default_cksum_report(zio_t *zio, zio_cksum_report_t *zcr, void *ignored) { void *buf = zio_buf_alloc(zio->io_size); bcopy(zio->io_data, buf, zio->io_size); zcr->zcr_cbinfo = zio->io_size; zcr->zcr_cbdata = buf; zcr->zcr_finish = zio_vsd_default_cksum_finish; zcr->zcr_free = zio_buf_free; } static int zio_vdev_io_assess(zio_t *zio) { vdev_t *vd = zio->io_vd; if (zio_wait_for_children(zio, ZIO_CHILD_VDEV, ZIO_WAIT_DONE)) return (ZIO_PIPELINE_STOP); if (vd == NULL && !(zio->io_flags & ZIO_FLAG_CONFIG_WRITER)) spa_config_exit(zio->io_spa, SCL_ZIO, zio); if (zio->io_vsd != NULL) { zio->io_vsd_ops->vsd_free(zio); zio->io_vsd = NULL; } if (zio_injection_enabled && zio->io_error == 0) zio->io_error = zio_handle_fault_injection(zio, EIO); /* * If the I/O failed, determine whether we should attempt to retry it. * * On retry, we cut in line in the issue queue, since we don't want * compression/checksumming/etc. work to prevent our (cheap) IO reissue. */ if (zio->io_error && vd == NULL && !(zio->io_flags & (ZIO_FLAG_DONT_RETRY | ZIO_FLAG_IO_RETRY))) { ASSERT(!(zio->io_flags & ZIO_FLAG_DONT_QUEUE)); /* not a leaf */ ASSERT(!(zio->io_flags & ZIO_FLAG_IO_BYPASS)); /* not a leaf */ zio->io_error = 0; zio->io_flags |= ZIO_FLAG_IO_RETRY | ZIO_FLAG_DONT_CACHE | ZIO_FLAG_DONT_AGGREGATE; zio->io_stage = ZIO_STAGE_VDEV_IO_START >> 1; zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, zio_requeue_io_start_cut_in_line); return (ZIO_PIPELINE_STOP); } /* * If we got an error on a leaf device, convert it to ENXIO * if the device is not accessible at all. */ if (zio->io_error && vd != NULL && vd->vdev_ops->vdev_op_leaf && !vdev_accessible(vd, zio)) zio->io_error = SET_ERROR(ENXIO); /* * If we can't write to an interior vdev (mirror or RAID-Z), * set vdev_cant_write so that we stop trying to allocate from it. */ if (zio->io_error == ENXIO && zio->io_type == ZIO_TYPE_WRITE && vd != NULL && !vd->vdev_ops->vdev_op_leaf) { vd->vdev_cant_write = B_TRUE; } if (zio->io_error) zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; if (vd != NULL && vd->vdev_ops->vdev_op_leaf && zio->io_physdone != NULL) { ASSERT(!(zio->io_flags & ZIO_FLAG_DELEGATED)); ASSERT(zio->io_child_type == ZIO_CHILD_VDEV); zio->io_physdone(zio->io_logical); } return (ZIO_PIPELINE_CONTINUE); } void zio_vdev_io_reissue(zio_t *zio) { ASSERT(zio->io_stage == ZIO_STAGE_VDEV_IO_START); ASSERT(zio->io_error == 0); zio->io_stage >>= 1; } void zio_vdev_io_redone(zio_t *zio) { ASSERT(zio->io_stage == ZIO_STAGE_VDEV_IO_DONE); zio->io_stage >>= 1; } void zio_vdev_io_bypass(zio_t *zio) { ASSERT(zio->io_stage == ZIO_STAGE_VDEV_IO_START); ASSERT(zio->io_error == 0); zio->io_flags |= ZIO_FLAG_IO_BYPASS; zio->io_stage = ZIO_STAGE_VDEV_IO_ASSESS >> 1; } /* * ========================================================================== * Generate and verify checksums * ========================================================================== */ static int zio_checksum_generate(zio_t *zio) { blkptr_t *bp = zio->io_bp; enum zio_checksum checksum; if (bp == NULL) { /* * This is zio_write_phys(). * We're either generating a label checksum, or none at all. */ checksum = zio->io_prop.zp_checksum; if (checksum == ZIO_CHECKSUM_OFF) return (ZIO_PIPELINE_CONTINUE); ASSERT(checksum == ZIO_CHECKSUM_LABEL); } else { if (BP_IS_GANG(bp) && zio->io_child_type == ZIO_CHILD_GANG) { ASSERT(!IO_IS_ALLOCATING(zio)); checksum = ZIO_CHECKSUM_GANG_HEADER; } else { checksum = BP_GET_CHECKSUM(bp); } } zio_checksum_compute(zio, checksum, zio->io_data, zio->io_size); return (ZIO_PIPELINE_CONTINUE); } static int zio_checksum_verify(zio_t *zio) { zio_bad_cksum_t info; blkptr_t *bp = zio->io_bp; int error; ASSERT(zio->io_vd != NULL); if (bp == NULL) { /* * This is zio_read_phys(). * We're either verifying a label checksum, or nothing at all. */ if (zio->io_prop.zp_checksum == ZIO_CHECKSUM_OFF) return (ZIO_PIPELINE_CONTINUE); ASSERT(zio->io_prop.zp_checksum == ZIO_CHECKSUM_LABEL); } if ((error = zio_checksum_error(zio, &info)) != 0) { zio->io_error = error; if (!(zio->io_flags & ZIO_FLAG_SPECULATIVE)) { zfs_ereport_start_checksum(zio->io_spa, zio->io_vd, zio, zio->io_offset, zio->io_size, NULL, &info); } } return (ZIO_PIPELINE_CONTINUE); } /* * Called by RAID-Z to ensure we don't compute the checksum twice. */ void zio_checksum_verified(zio_t *zio) { zio->io_pipeline &= ~ZIO_STAGE_CHECKSUM_VERIFY; } /* * ========================================================================== * Error rank. Error are ranked in the order 0, ENXIO, ECKSUM, EIO, other. * An error of 0 indicates success. ENXIO indicates whole-device failure, * which may be transient (e.g. unplugged) or permament. ECKSUM and EIO * indicate errors that are specific to one I/O, and most likely permanent. * Any other error is presumed to be worse because we weren't expecting it. * ========================================================================== */ int zio_worst_error(int e1, int e2) { static int zio_error_rank[] = { 0, ENXIO, ECKSUM, EIO }; int r1, r2; for (r1 = 0; r1 < sizeof (zio_error_rank) / sizeof (int); r1++) if (e1 == zio_error_rank[r1]) break; for (r2 = 0; r2 < sizeof (zio_error_rank) / sizeof (int); r2++) if (e2 == zio_error_rank[r2]) break; return (r1 > r2 ? e1 : e2); } /* * ========================================================================== * I/O completion * ========================================================================== */ static int zio_ready(zio_t *zio) { blkptr_t *bp = zio->io_bp; zio_t *pio, *pio_next; if (zio_wait_for_children(zio, ZIO_CHILD_GANG, ZIO_WAIT_READY) || zio_wait_for_children(zio, ZIO_CHILD_DDT, ZIO_WAIT_READY)) return (ZIO_PIPELINE_STOP); if (zio->io_ready) { ASSERT(IO_IS_ALLOCATING(zio)); ASSERT(bp->blk_birth == zio->io_txg || BP_IS_HOLE(bp) || (zio->io_flags & ZIO_FLAG_NOPWRITE)); ASSERT(zio->io_children[ZIO_CHILD_GANG][ZIO_WAIT_READY] == 0); zio->io_ready(zio); } if (bp != NULL && bp != &zio->io_bp_copy) zio->io_bp_copy = *bp; if (zio->io_error) zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; mutex_enter(&zio->io_lock); zio->io_state[ZIO_WAIT_READY] = 1; pio = zio_walk_parents(zio); mutex_exit(&zio->io_lock); /* * As we notify zio's parents, new parents could be added. * New parents go to the head of zio's io_parent_list, however, * so we will (correctly) not notify them. The remainder of zio's * io_parent_list, from 'pio_next' onward, cannot change because * all parents must wait for us to be done before they can be done. */ for (; pio != NULL; pio = pio_next) { pio_next = zio_walk_parents(zio); zio_notify_parent(pio, zio, ZIO_WAIT_READY); } if (zio->io_flags & ZIO_FLAG_NODATA) { if (BP_IS_GANG(bp)) { zio->io_flags &= ~ZIO_FLAG_NODATA; } else { ASSERT((uintptr_t)zio->io_data < SPA_MAXBLOCKSIZE); zio->io_pipeline &= ~ZIO_VDEV_IO_STAGES; } } if (zio_injection_enabled && zio->io_spa->spa_syncing_txg == zio->io_txg) zio_handle_ignored_writes(zio); return (ZIO_PIPELINE_CONTINUE); } static int zio_done(zio_t *zio) { zio_t *pio, *pio_next; int c, w; /* * If our children haven't all completed, * wait for them and then repeat this pipeline stage. */ if (zio_wait_for_children(zio, ZIO_CHILD_VDEV, ZIO_WAIT_DONE) || zio_wait_for_children(zio, ZIO_CHILD_GANG, ZIO_WAIT_DONE) || zio_wait_for_children(zio, ZIO_CHILD_DDT, ZIO_WAIT_DONE) || zio_wait_for_children(zio, ZIO_CHILD_LOGICAL, ZIO_WAIT_DONE)) return (ZIO_PIPELINE_STOP); for (c = 0; c < ZIO_CHILD_TYPES; c++) for (w = 0; w < ZIO_WAIT_TYPES; w++) ASSERT(zio->io_children[c][w] == 0); if (zio->io_bp != NULL && !BP_IS_EMBEDDED(zio->io_bp)) { ASSERT(zio->io_bp->blk_pad[0] == 0); ASSERT(zio->io_bp->blk_pad[1] == 0); ASSERT(bcmp(zio->io_bp, &zio->io_bp_copy, sizeof (blkptr_t)) == 0 || (zio->io_bp == zio_unique_parent(zio)->io_bp)); if (zio->io_type == ZIO_TYPE_WRITE && !BP_IS_HOLE(zio->io_bp) && zio->io_bp_override == NULL && !(zio->io_flags & ZIO_FLAG_IO_REPAIR)) { ASSERT(!BP_SHOULD_BYTESWAP(zio->io_bp)); ASSERT3U(zio->io_prop.zp_copies, <=, BP_GET_NDVAS(zio->io_bp)); ASSERT(BP_COUNT_GANG(zio->io_bp) == 0 || (BP_COUNT_GANG(zio->io_bp) == BP_GET_NDVAS(zio->io_bp))); } if (zio->io_flags & ZIO_FLAG_NOPWRITE) VERIFY(BP_EQUAL(zio->io_bp, &zio->io_bp_orig)); } /* * If there were child vdev/gang/ddt errors, they apply to us now. */ zio_inherit_child_errors(zio, ZIO_CHILD_VDEV); zio_inherit_child_errors(zio, ZIO_CHILD_GANG); zio_inherit_child_errors(zio, ZIO_CHILD_DDT); /* * If the I/O on the transformed data was successful, generate any * checksum reports now while we still have the transformed data. */ if (zio->io_error == 0) { while (zio->io_cksum_report != NULL) { zio_cksum_report_t *zcr = zio->io_cksum_report; uint64_t align = zcr->zcr_align; uint64_t asize = P2ROUNDUP(zio->io_size, align); char *abuf = zio->io_data; if (asize != zio->io_size) { abuf = zio_buf_alloc(asize); bcopy(zio->io_data, abuf, zio->io_size); bzero(abuf+zio->io_size, asize-zio->io_size); } zio->io_cksum_report = zcr->zcr_next; zcr->zcr_next = NULL; zcr->zcr_finish(zcr, abuf); zfs_ereport_free_checksum(zcr); if (asize != zio->io_size) zio_buf_free(abuf, asize); } } zio_pop_transforms(zio); /* note: may set zio->io_error */ vdev_stat_update(zio, zio->io_size); /* * If this I/O is attached to a particular vdev is slow, exceeding * 30 seconds to complete, post an error described the I/O delay. * We ignore these errors if the device is currently unavailable. */ if (zio->io_delay >= MSEC_TO_TICK(zio_delay_max)) { if (zio->io_vd != NULL && !vdev_is_dead(zio->io_vd)) zfs_ereport_post(FM_EREPORT_ZFS_DELAY, zio->io_spa, zio->io_vd, zio, 0, 0); } if (zio->io_error) { /* * If this I/O is attached to a particular vdev, * generate an error message describing the I/O failure * at the block level. We ignore these errors if the * device is currently unavailable. */ if (zio->io_error != ECKSUM && zio->io_vd != NULL && !vdev_is_dead(zio->io_vd)) zfs_ereport_post(FM_EREPORT_ZFS_IO, zio->io_spa, zio->io_vd, zio, 0, 0); if ((zio->io_error == EIO || !(zio->io_flags & (ZIO_FLAG_SPECULATIVE | ZIO_FLAG_DONT_PROPAGATE))) && zio == zio->io_logical) { /* * For logical I/O requests, tell the SPA to log the * error and generate a logical data ereport. */ spa_log_error(zio->io_spa, zio); zfs_ereport_post(FM_EREPORT_ZFS_DATA, zio->io_spa, NULL, zio, 0, 0); } } if (zio->io_error && zio == zio->io_logical) { /* * Determine whether zio should be reexecuted. This will * propagate all the way to the root via zio_notify_parent(). */ ASSERT(zio->io_vd == NULL && zio->io_bp != NULL); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); if (IO_IS_ALLOCATING(zio) && !(zio->io_flags & ZIO_FLAG_CANFAIL)) { if (zio->io_error != ENOSPC) zio->io_reexecute |= ZIO_REEXECUTE_NOW; else zio->io_reexecute |= ZIO_REEXECUTE_SUSPEND; } if ((zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_FREE) && !(zio->io_flags & ZIO_FLAG_SCAN_THREAD) && zio->io_error == ENXIO && spa_load_state(zio->io_spa) == SPA_LOAD_NONE && spa_get_failmode(zio->io_spa) != ZIO_FAILURE_MODE_CONTINUE) zio->io_reexecute |= ZIO_REEXECUTE_SUSPEND; if (!(zio->io_flags & ZIO_FLAG_CANFAIL) && !zio->io_reexecute) zio->io_reexecute |= ZIO_REEXECUTE_SUSPEND; /* * Here is a possibly good place to attempt to do * either combinatorial reconstruction or error correction * based on checksums. It also might be a good place * to send out preliminary ereports before we suspend * processing. */ } /* * If there were logical child errors, they apply to us now. * We defer this until now to avoid conflating logical child * errors with errors that happened to the zio itself when * updating vdev stats and reporting FMA events above. */ zio_inherit_child_errors(zio, ZIO_CHILD_LOGICAL); if ((zio->io_error || zio->io_reexecute) && IO_IS_ALLOCATING(zio) && zio->io_gang_leader == zio && !(zio->io_flags & (ZIO_FLAG_IO_REWRITE | ZIO_FLAG_NOPWRITE))) zio_dva_unallocate(zio, zio->io_gang_tree, zio->io_bp); zio_gang_tree_free(&zio->io_gang_tree); /* * Godfather I/Os should never suspend. */ if ((zio->io_flags & ZIO_FLAG_GODFATHER) && (zio->io_reexecute & ZIO_REEXECUTE_SUSPEND)) zio->io_reexecute = 0; if (zio->io_reexecute) { /* * This is a logical I/O that wants to reexecute. * * Reexecute is top-down. When an i/o fails, if it's not * the root, it simply notifies its parent and sticks around. * The parent, seeing that it still has children in zio_done(), * does the same. This percolates all the way up to the root. * The root i/o will reexecute or suspend the entire tree. * * This approach ensures that zio_reexecute() honors * all the original i/o dependency relationships, e.g. * parents not executing until children are ready. */ ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); zio->io_gang_leader = NULL; mutex_enter(&zio->io_lock); zio->io_state[ZIO_WAIT_DONE] = 1; mutex_exit(&zio->io_lock); /* * "The Godfather" I/O monitors its children but is * not a true parent to them. It will track them through * the pipeline but severs its ties whenever they get into * trouble (e.g. suspended). This allows "The Godfather" * I/O to return status without blocking. */ for (pio = zio_walk_parents(zio); pio != NULL; pio = pio_next) { zio_link_t *zl = zio->io_walk_link; pio_next = zio_walk_parents(zio); if ((pio->io_flags & ZIO_FLAG_GODFATHER) && (zio->io_reexecute & ZIO_REEXECUTE_SUSPEND)) { zio_remove_child(pio, zio, zl); zio_notify_parent(pio, zio, ZIO_WAIT_DONE); } } if ((pio = zio_unique_parent(zio)) != NULL) { /* * We're not a root i/o, so there's nothing to do * but notify our parent. Don't propagate errors * upward since we haven't permanently failed yet. */ ASSERT(!(zio->io_flags & ZIO_FLAG_GODFATHER)); zio->io_flags |= ZIO_FLAG_DONT_PROPAGATE; zio_notify_parent(pio, zio, ZIO_WAIT_DONE); } else if (zio->io_reexecute & ZIO_REEXECUTE_SUSPEND) { /* * We'd fail again if we reexecuted now, so suspend * until conditions improve (e.g. device comes online). */ zio_suspend(zio->io_spa, zio); } else { /* * Reexecution is potentially a huge amount of work. * Hand it off to the otherwise-unused claim taskq. */ ASSERT(taskq_empty_ent(&zio->io_tqent)); spa_taskq_dispatch_ent(zio->io_spa, ZIO_TYPE_CLAIM, ZIO_TASKQ_ISSUE, (task_func_t *)zio_reexecute, zio, 0, &zio->io_tqent); } return (ZIO_PIPELINE_STOP); } ASSERT(zio->io_child_count == 0); ASSERT(zio->io_reexecute == 0); ASSERT(zio->io_error == 0 || (zio->io_flags & ZIO_FLAG_CANFAIL)); /* * Report any checksum errors, since the I/O is complete. */ while (zio->io_cksum_report != NULL) { zio_cksum_report_t *zcr = zio->io_cksum_report; zio->io_cksum_report = zcr->zcr_next; zcr->zcr_next = NULL; zcr->zcr_finish(zcr, NULL); zfs_ereport_free_checksum(zcr); } if (zio->io_flags & ZIO_FLAG_FASTWRITE && zio->io_bp && !BP_IS_HOLE(zio->io_bp) && !BP_IS_EMBEDDED(zio->io_bp) && !(zio->io_flags & ZIO_FLAG_NOPWRITE)) { metaslab_fastwrite_unmark(zio->io_spa, zio->io_bp); } /* * It is the responsibility of the done callback to ensure that this * particular zio is no longer discoverable for adoption, and as * such, cannot acquire any new parents. */ if (zio->io_done) zio->io_done(zio); mutex_enter(&zio->io_lock); zio->io_state[ZIO_WAIT_DONE] = 1; mutex_exit(&zio->io_lock); for (pio = zio_walk_parents(zio); pio != NULL; pio = pio_next) { zio_link_t *zl = zio->io_walk_link; pio_next = zio_walk_parents(zio); zio_remove_child(pio, zio, zl); zio_notify_parent(pio, zio, ZIO_WAIT_DONE); } if (zio->io_waiter != NULL) { mutex_enter(&zio->io_lock); zio->io_executor = NULL; cv_broadcast(&zio->io_cv); mutex_exit(&zio->io_lock); } else { zio_destroy(zio); } return (ZIO_PIPELINE_STOP); } /* * ========================================================================== * I/O pipeline definition * ========================================================================== */ static zio_pipe_stage_t *zio_pipeline[] = { NULL, zio_read_bp_init, zio_free_bp_init, zio_issue_async, zio_write_bp_init, zio_checksum_generate, zio_nop_write, zio_ddt_read_start, zio_ddt_read_done, zio_ddt_write, zio_ddt_free, zio_gang_assemble, zio_gang_issue, zio_dva_allocate, zio_dva_free, zio_dva_claim, zio_ready, zio_vdev_io_start, zio_vdev_io_done, zio_vdev_io_assess, zio_checksum_verify, zio_done }; /* dnp is the dnode for zb1->zb_object */ boolean_t zbookmark_is_before(const dnode_phys_t *dnp, const zbookmark_phys_t *zb1, const zbookmark_phys_t *zb2) { uint64_t zb1nextL0, zb2thisobj; ASSERT(zb1->zb_objset == zb2->zb_objset); ASSERT(zb2->zb_level == 0); /* The objset_phys_t isn't before anything. */ if (dnp == NULL) return (B_FALSE); zb1nextL0 = (zb1->zb_blkid + 1) << ((zb1->zb_level) * (dnp->dn_indblkshift - SPA_BLKPTRSHIFT)); zb2thisobj = zb2->zb_object ? zb2->zb_object : zb2->zb_blkid << (DNODE_BLOCK_SHIFT - DNODE_SHIFT); if (zb1->zb_object == DMU_META_DNODE_OBJECT) { uint64_t nextobj = zb1nextL0 * (dnp->dn_datablkszsec << SPA_MINBLOCKSHIFT) >> DNODE_SHIFT; return (nextobj <= zb2thisobj); } if (zb1->zb_object < zb2thisobj) return (B_TRUE); if (zb1->zb_object > zb2thisobj) return (B_FALSE); if (zb2->zb_object == DMU_META_DNODE_OBJECT) return (B_FALSE); return (zb1nextL0 <= zb2->zb_blkid); } #if defined(_KERNEL) && defined(HAVE_SPL) /* Fault injection */ EXPORT_SYMBOL(zio_injection_enabled); EXPORT_SYMBOL(zio_inject_fault); EXPORT_SYMBOL(zio_inject_list_next); EXPORT_SYMBOL(zio_clear_fault); EXPORT_SYMBOL(zio_handle_fault_injection); EXPORT_SYMBOL(zio_handle_device_injection); EXPORT_SYMBOL(zio_handle_label_injection); EXPORT_SYMBOL(zio_type_name); module_param(zio_bulk_flags, int, 0644); MODULE_PARM_DESC(zio_bulk_flags, "Additional flags to pass to bulk buffers"); module_param(zio_delay_max, int, 0644); MODULE_PARM_DESC(zio_delay_max, "Max zio millisec delay before posting event"); module_param(zio_requeue_io_start_cut_in_line, int, 0644); MODULE_PARM_DESC(zio_requeue_io_start_cut_in_line, "Prioritize requeued I/O"); module_param(zfs_sync_pass_deferred_free, int, 0644); MODULE_PARM_DESC(zfs_sync_pass_deferred_free, "Defer frees starting in this pass"); module_param(zfs_sync_pass_dont_compress, int, 0644); MODULE_PARM_DESC(zfs_sync_pass_dont_compress, "Don't compress starting in this pass"); module_param(zfs_sync_pass_rewrite, int, 0644); MODULE_PARM_DESC(zfs_sync_pass_rewrite, "Rewrite new bps starting in this pass"); #endif diff --git a/module/zfs/zpl_file.c b/module/zfs/zpl_file.c index 61005dcd4ff7..1f4f219ed58c 100644 --- a/module/zfs/zpl_file.c +++ b/module/zfs/zpl_file.c @@ -1,765 +1,765 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2011, Lawrence Livermore National Security, LLC. */ #include #include #include #include #include static int zpl_open(struct inode *ip, struct file *filp) { cred_t *cr = CRED(); int error; error = generic_file_open(ip, filp); if (error) return (error); crhold(cr); error = -zfs_open(ip, filp->f_mode, filp->f_flags, cr); crfree(cr); ASSERT3S(error, <=, 0); return (error); } static int zpl_release(struct inode *ip, struct file *filp) { cred_t *cr = CRED(); int error; if (ITOZ(ip)->z_atime_dirty) zfs_mark_inode_dirty(ip); crhold(cr); error = -zfs_close(ip, filp->f_flags, cr); crfree(cr); ASSERT3S(error, <=, 0); return (error); } static int zpl_iterate(struct file *filp, struct dir_context *ctx) { struct dentry *dentry = filp->f_path.dentry; cred_t *cr = CRED(); int error; crhold(cr); error = -zfs_readdir(dentry->d_inode, ctx, cr); crfree(cr); ASSERT3S(error, <=, 0); return (error); } #if !defined(HAVE_VFS_ITERATE) static int zpl_readdir(struct file *filp, void *dirent, filldir_t filldir) { struct dir_context ctx = DIR_CONTEXT_INIT(dirent, filldir, filp->f_pos); int error; error = zpl_iterate(filp, &ctx); filp->f_pos = ctx.pos; return (error); } #endif /* HAVE_VFS_ITERATE */ #if defined(HAVE_FSYNC_WITH_DENTRY) /* * Linux 2.6.x - 2.6.34 API, * Through 2.6.34 the nfsd kernel server would pass a NULL 'file struct *' * to the fops->fsync() hook. For this reason, we must be careful not to * use filp unconditionally. */ static int zpl_fsync(struct file *filp, struct dentry *dentry, int datasync) { cred_t *cr = CRED(); int error; crhold(cr); error = -zfs_fsync(dentry->d_inode, datasync, cr); crfree(cr); ASSERT3S(error, <=, 0); return (error); } static int zpl_aio_fsync(struct kiocb *kiocb, int datasync) { struct file *filp = kiocb->ki_filp; return (zpl_fsync(filp, filp->f_path.dentry, datasync)); } #elif defined(HAVE_FSYNC_WITHOUT_DENTRY) /* * Linux 2.6.35 - 3.0 API, * As of 2.6.35 the dentry argument to the fops->fsync() hook was deemed * redundant. The dentry is still accessible via filp->f_path.dentry, * and we are guaranteed that filp will never be NULL. */ static int zpl_fsync(struct file *filp, int datasync) { struct inode *inode = filp->f_mapping->host; cred_t *cr = CRED(); int error; crhold(cr); error = -zfs_fsync(inode, datasync, cr); crfree(cr); ASSERT3S(error, <=, 0); return (error); } static int zpl_aio_fsync(struct kiocb *kiocb, int datasync) { return (zpl_fsync(kiocb->ki_filp, datasync)); } #elif defined(HAVE_FSYNC_RANGE) /* * Linux 3.1 - 3.x API, * As of 3.1 the responsibility to call filemap_write_and_wait_range() has * been pushed down in to the .fsync() vfs hook. Additionally, the i_mutex * lock is no longer held by the caller, for zfs we don't require the lock * to be held so we don't acquire it. */ static int zpl_fsync(struct file *filp, loff_t start, loff_t end, int datasync) { struct inode *inode = filp->f_mapping->host; cred_t *cr = CRED(); int error; error = filemap_write_and_wait_range(inode->i_mapping, start, end); if (error) return (error); crhold(cr); error = -zfs_fsync(inode, datasync, cr); crfree(cr); ASSERT3S(error, <=, 0); return (error); } static int zpl_aio_fsync(struct kiocb *kiocb, int datasync) { return (zpl_fsync(kiocb->ki_filp, kiocb->ki_pos, kiocb->ki_pos + kiocb->ki_nbytes, datasync)); } #else #error "Unsupported fops->fsync() implementation" #endif static inline ssize_t zpl_read_common_iovec(struct inode *ip, const struct iovec *iovp, size_t count, unsigned long nr_segs, loff_t *ppos, uio_seg_t segment, int flags, cred_t *cr) { ssize_t read; uio_t uio; int error; uio.uio_iov = (struct iovec *)iovp; uio.uio_resid = count; uio.uio_iovcnt = nr_segs; uio.uio_loffset = *ppos; uio.uio_limit = MAXOFFSET_T; uio.uio_segflg = segment; error = -zfs_read(ip, &uio, flags, cr); if (error < 0) return (error); read = count - uio.uio_resid; *ppos += read; task_io_account_read(read); return (read); } inline ssize_t zpl_read_common(struct inode *ip, const char *buf, size_t len, loff_t *ppos, uio_seg_t segment, int flags, cred_t *cr) { struct iovec iov; iov.iov_base = (void *)buf; iov.iov_len = len; return (zpl_read_common_iovec(ip, &iov, len, 1, ppos, segment, flags, cr)); } static ssize_t zpl_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos) { cred_t *cr = CRED(); ssize_t read; crhold(cr); read = zpl_read_common(filp->f_mapping->host, buf, len, ppos, UIO_USERSPACE, filp->f_flags, cr); crfree(cr); return (read); } static ssize_t zpl_aio_read(struct kiocb *kiocb, const struct iovec *iovp, unsigned long nr_segs, loff_t pos) { cred_t *cr = CRED(); struct file *filp = kiocb->ki_filp; size_t count = kiocb->ki_nbytes; ssize_t read; size_t alloc_size = sizeof (struct iovec) * nr_segs; - struct iovec *iov_tmp = kmem_alloc(alloc_size, KM_SLEEP | KM_NODEBUG); + struct iovec *iov_tmp = kmem_alloc(alloc_size, KM_SLEEP); bcopy(iovp, iov_tmp, alloc_size); ASSERT(iovp); crhold(cr); read = zpl_read_common_iovec(filp->f_mapping->host, iov_tmp, count, nr_segs, &kiocb->ki_pos, UIO_USERSPACE, filp->f_flags, cr); crfree(cr); kmem_free(iov_tmp, alloc_size); return (read); } static inline ssize_t zpl_write_common_iovec(struct inode *ip, const struct iovec *iovp, size_t count, unsigned long nr_segs, loff_t *ppos, uio_seg_t segment, int flags, cred_t *cr) { ssize_t wrote; uio_t uio; int error; uio.uio_iov = (struct iovec *)iovp; uio.uio_resid = count; uio.uio_iovcnt = nr_segs; uio.uio_loffset = *ppos; uio.uio_limit = MAXOFFSET_T; uio.uio_segflg = segment; error = -zfs_write(ip, &uio, flags, cr); if (error < 0) return (error); wrote = count - uio.uio_resid; *ppos += wrote; task_io_account_write(wrote); return (wrote); } inline ssize_t zpl_write_common(struct inode *ip, const char *buf, size_t len, loff_t *ppos, uio_seg_t segment, int flags, cred_t *cr) { struct iovec iov; iov.iov_base = (void *)buf; iov.iov_len = len; return (zpl_write_common_iovec(ip, &iov, len, 1, ppos, segment, flags, cr)); } static ssize_t zpl_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos) { cred_t *cr = CRED(); ssize_t wrote; crhold(cr); wrote = zpl_write_common(filp->f_mapping->host, buf, len, ppos, UIO_USERSPACE, filp->f_flags, cr); crfree(cr); return (wrote); } static ssize_t zpl_aio_write(struct kiocb *kiocb, const struct iovec *iovp, unsigned long nr_segs, loff_t pos) { cred_t *cr = CRED(); struct file *filp = kiocb->ki_filp; size_t count = kiocb->ki_nbytes; ssize_t wrote; size_t alloc_size = sizeof (struct iovec) * nr_segs; - struct iovec *iov_tmp = kmem_alloc(alloc_size, KM_SLEEP | KM_NODEBUG); + struct iovec *iov_tmp = kmem_alloc(alloc_size, KM_SLEEP); bcopy(iovp, iov_tmp, alloc_size); ASSERT(iovp); crhold(cr); wrote = zpl_write_common_iovec(filp->f_mapping->host, iov_tmp, count, nr_segs, &kiocb->ki_pos, UIO_USERSPACE, filp->f_flags, cr); crfree(cr); kmem_free(iov_tmp, alloc_size); return (wrote); } static loff_t zpl_llseek(struct file *filp, loff_t offset, int whence) { #if defined(SEEK_HOLE) && defined(SEEK_DATA) if (whence == SEEK_DATA || whence == SEEK_HOLE) { struct inode *ip = filp->f_mapping->host; loff_t maxbytes = ip->i_sb->s_maxbytes; loff_t error; spl_inode_lock(ip); error = -zfs_holey(ip, whence, &offset); if (error == 0) error = lseek_execute(filp, ip, offset, maxbytes); spl_inode_unlock(ip); return (error); } #endif /* SEEK_HOLE && SEEK_DATA */ return (generic_file_llseek(filp, offset, whence)); } /* * It's worth taking a moment to describe how mmap is implemented * for zfs because it differs considerably from other Linux filesystems. * However, this issue is handled the same way under OpenSolaris. * * The issue is that by design zfs bypasses the Linux page cache and * leaves all caching up to the ARC. This has been shown to work * well for the common read(2)/write(2) case. However, mmap(2) * is problem because it relies on being tightly integrated with the * page cache. To handle this we cache mmap'ed files twice, once in * the ARC and a second time in the page cache. The code is careful * to keep both copies synchronized. * * When a file with an mmap'ed region is written to using write(2) * both the data in the ARC and existing pages in the page cache * are updated. For a read(2) data will be read first from the page * cache then the ARC if needed. Neither a write(2) or read(2) will * will ever result in new pages being added to the page cache. * * New pages are added to the page cache only via .readpage() which * is called when the vfs needs to read a page off disk to back the * virtual memory region. These pages may be modified without * notifying the ARC and will be written out periodically via * .writepage(). This will occur due to either a sync or the usual * page aging behavior. Note because a read(2) of a mmap'ed file * will always check the page cache first even when the ARC is out * of date correct data will still be returned. * * While this implementation ensures correct behavior it does have * have some drawbacks. The most obvious of which is that it * increases the required memory footprint when access mmap'ed * files. It also adds additional complexity to the code keeping * both caches synchronized. * * Longer term it may be possible to cleanly resolve this wart by * mapping page cache pages directly on to the ARC buffers. The * Linux address space operations are flexible enough to allow * selection of which pages back a particular index. The trick * would be working out the details of which subsystem is in * charge, the ARC, the page cache, or both. It may also prove * helpful to move the ARC buffers to a scatter-gather lists * rather than a vmalloc'ed region. */ static int zpl_mmap(struct file *filp, struct vm_area_struct *vma) { struct inode *ip = filp->f_mapping->host; znode_t *zp = ITOZ(ip); int error; error = -zfs_map(ip, vma->vm_pgoff, (caddr_t *)vma->vm_start, (size_t)(vma->vm_end - vma->vm_start), vma->vm_flags); if (error) return (error); error = generic_file_mmap(filp, vma); if (error) return (error); mutex_enter(&zp->z_lock); zp->z_is_mapped = 1; mutex_exit(&zp->z_lock); return (error); } /* * Populate a page with data for the Linux page cache. This function is * only used to support mmap(2). There will be an identical copy of the * data in the ARC which is kept up to date via .write() and .writepage(). * * Current this function relies on zpl_read_common() and the O_DIRECT * flag to read in a page. This works but the more correct way is to * update zfs_fillpage() to be Linux friendly and use that interface. */ static int zpl_readpage(struct file *filp, struct page *pp) { struct inode *ip; struct page *pl[1]; int error = 0; ASSERT(PageLocked(pp)); ip = pp->mapping->host; pl[0] = pp; error = -zfs_getpage(ip, pl, 1); if (error) { SetPageError(pp); ClearPageUptodate(pp); } else { ClearPageError(pp); SetPageUptodate(pp); flush_dcache_page(pp); } unlock_page(pp); return (error); } /* * Populate a set of pages with data for the Linux page cache. This * function will only be called for read ahead and never for demand * paging. For simplicity, the code relies on read_cache_pages() to * correctly lock each page for IO and call zpl_readpage(). */ static int zpl_readpages(struct file *filp, struct address_space *mapping, struct list_head *pages, unsigned nr_pages) { return (read_cache_pages(mapping, pages, (filler_t *)zpl_readpage, filp)); } int zpl_putpage(struct page *pp, struct writeback_control *wbc, void *data) { struct address_space *mapping = data; fstrans_cookie_t cookie; ASSERT(PageLocked(pp)); ASSERT(!PageWriteback(pp)); cookie = spl_fstrans_mark(); (void) zfs_putpage(mapping->host, pp, wbc); spl_fstrans_unmark(cookie); return (0); } static int zpl_writepages(struct address_space *mapping, struct writeback_control *wbc) { znode_t *zp = ITOZ(mapping->host); zfs_sb_t *zsb = ITOZSB(mapping->host); enum writeback_sync_modes sync_mode; int result; ZFS_ENTER(zsb); if (zsb->z_os->os_sync == ZFS_SYNC_ALWAYS) wbc->sync_mode = WB_SYNC_ALL; ZFS_EXIT(zsb); sync_mode = wbc->sync_mode; /* * We don't want to run write_cache_pages() in SYNC mode here, because * that would make putpage() wait for a single page to be committed to * disk every single time, resulting in atrocious performance. Instead * we run it once in non-SYNC mode so that the ZIL gets all the data, * and then we commit it all in one go. */ wbc->sync_mode = WB_SYNC_NONE; result = write_cache_pages(mapping, wbc, zpl_putpage, mapping); if (sync_mode != wbc->sync_mode) { ZFS_ENTER(zsb); ZFS_VERIFY_ZP(zp); if (zsb->z_log != NULL) zil_commit(zsb->z_log, zp->z_id); ZFS_EXIT(zsb); /* * We need to call write_cache_pages() again (we can't just * return after the commit) because the previous call in * non-SYNC mode does not guarantee that we got all the dirty * pages (see the implementation of write_cache_pages() for * details). That being said, this is a no-op in most cases. */ wbc->sync_mode = sync_mode; result = write_cache_pages(mapping, wbc, zpl_putpage, mapping); } return (result); } /* * Write out dirty pages to the ARC, this function is only required to * support mmap(2). Mapped pages may be dirtied by memory operations * which never call .write(). These dirty pages are kept in sync with * the ARC buffers via this hook. */ static int zpl_writepage(struct page *pp, struct writeback_control *wbc) { if (ITOZSB(pp->mapping->host)->z_os->os_sync == ZFS_SYNC_ALWAYS) wbc->sync_mode = WB_SYNC_ALL; return (zpl_putpage(pp, wbc, pp->mapping)); } /* * The only flag combination which matches the behavior of zfs_space() * is FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE. The FALLOC_FL_PUNCH_HOLE * flag was introduced in the 2.6.38 kernel. */ #if defined(HAVE_FILE_FALLOCATE) || defined(HAVE_INODE_FALLOCATE) long zpl_fallocate_common(struct inode *ip, int mode, loff_t offset, loff_t len) { int error = -EOPNOTSUPP; #if defined(FALLOC_FL_PUNCH_HOLE) && defined(FALLOC_FL_KEEP_SIZE) cred_t *cr = CRED(); flock64_t bf; loff_t olen; if (mode != (FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE)) return (error); crhold(cr); if (offset < 0 || len <= 0) return (-EINVAL); spl_inode_lock(ip); olen = i_size_read(ip); if (offset > olen) { spl_inode_unlock(ip); return (0); } if (offset + len > olen) len = olen - offset; bf.l_type = F_WRLCK; bf.l_whence = 0; bf.l_start = offset; bf.l_len = len; bf.l_pid = 0; error = -zfs_space(ip, F_FREESP, &bf, FWRITE, offset, cr); spl_inode_unlock(ip); crfree(cr); #endif /* defined(FALLOC_FL_PUNCH_HOLE) && defined(FALLOC_FL_KEEP_SIZE) */ ASSERT3S(error, <=, 0); return (error); } #endif /* defined(HAVE_FILE_FALLOCATE) || defined(HAVE_INODE_FALLOCATE) */ #ifdef HAVE_FILE_FALLOCATE static long zpl_fallocate(struct file *filp, int mode, loff_t offset, loff_t len) { return zpl_fallocate_common(filp->f_path.dentry->d_inode, mode, offset, len); } #endif /* HAVE_FILE_FALLOCATE */ /* * Map zfs file z_pflags (xvattr_t) to linux file attributes. Only file * attributes common to both Linux and Solaris are mapped. */ static int zpl_ioctl_getflags(struct file *filp, void __user *arg) { struct inode *ip = filp->f_dentry->d_inode; unsigned int ioctl_flags = 0; uint64_t zfs_flags = ITOZ(ip)->z_pflags; int error; if (zfs_flags & ZFS_IMMUTABLE) ioctl_flags |= FS_IMMUTABLE_FL; if (zfs_flags & ZFS_APPENDONLY) ioctl_flags |= FS_APPEND_FL; if (zfs_flags & ZFS_NODUMP) ioctl_flags |= FS_NODUMP_FL; ioctl_flags &= FS_FL_USER_VISIBLE; error = copy_to_user(arg, &ioctl_flags, sizeof (ioctl_flags)); return (error); } /* * fchange() is a helper macro to detect if we have been asked to change a * flag. This is ugly, but the requirement that we do this is a consequence of * how the Linux file attribute interface was designed. Another consequence is * that concurrent modification of files suffers from a TOCTOU race. Neither * are things we can fix without modifying the kernel-userland interface, which * is outside of our jurisdiction. */ #define fchange(f0, f1, b0, b1) ((((f0) & (b0)) == (b0)) != \ (((b1) & (f1)) == (f1))) static int zpl_ioctl_setflags(struct file *filp, void __user *arg) { struct inode *ip = filp->f_dentry->d_inode; uint64_t zfs_flags = ITOZ(ip)->z_pflags; unsigned int ioctl_flags; cred_t *cr = CRED(); xvattr_t xva; xoptattr_t *xoap; int error; if (copy_from_user(&ioctl_flags, arg, sizeof (ioctl_flags))) return (-EFAULT); if ((ioctl_flags & ~(FS_IMMUTABLE_FL | FS_APPEND_FL | FS_NODUMP_FL))) return (-EOPNOTSUPP); if ((ioctl_flags & ~(FS_FL_USER_MODIFIABLE))) return (-EACCES); if ((fchange(ioctl_flags, zfs_flags, FS_IMMUTABLE_FL, ZFS_IMMUTABLE) || fchange(ioctl_flags, zfs_flags, FS_APPEND_FL, ZFS_APPENDONLY)) && !capable(CAP_LINUX_IMMUTABLE)) return (-EACCES); if (!zpl_inode_owner_or_capable(ip)) return (-EACCES); xva_init(&xva); xoap = xva_getxoptattr(&xva); XVA_SET_REQ(&xva, XAT_IMMUTABLE); if (ioctl_flags & FS_IMMUTABLE_FL) xoap->xoa_immutable = B_TRUE; XVA_SET_REQ(&xva, XAT_APPENDONLY); if (ioctl_flags & FS_APPEND_FL) xoap->xoa_appendonly = B_TRUE; XVA_SET_REQ(&xva, XAT_NODUMP); if (ioctl_flags & FS_NODUMP_FL) xoap->xoa_nodump = B_TRUE; crhold(cr); error = -zfs_setattr(ip, (vattr_t *)&xva, 0, cr); crfree(cr); return (error); } static long zpl_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) { switch (cmd) { case FS_IOC_GETFLAGS: return (zpl_ioctl_getflags(filp, (void *)arg)); case FS_IOC_SETFLAGS: return (zpl_ioctl_setflags(filp, (void *)arg)); default: return (-ENOTTY); } } #ifdef CONFIG_COMPAT static long zpl_compat_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) { return (zpl_ioctl(filp, cmd, arg)); } #endif /* CONFIG_COMPAT */ const struct address_space_operations zpl_address_space_operations = { .readpages = zpl_readpages, .readpage = zpl_readpage, .writepage = zpl_writepage, .writepages = zpl_writepages, }; const struct file_operations zpl_file_operations = { .open = zpl_open, .release = zpl_release, .llseek = zpl_llseek, .read = zpl_read, .write = zpl_write, .aio_read = zpl_aio_read, .aio_write = zpl_aio_write, .mmap = zpl_mmap, .fsync = zpl_fsync, .aio_fsync = zpl_aio_fsync, #ifdef HAVE_FILE_FALLOCATE .fallocate = zpl_fallocate, #endif /* HAVE_FILE_FALLOCATE */ .unlocked_ioctl = zpl_ioctl, #ifdef CONFIG_COMPAT .compat_ioctl = zpl_compat_ioctl, #endif }; const struct file_operations zpl_dir_file_operations = { .llseek = generic_file_llseek, .read = generic_read_dir, #ifdef HAVE_VFS_ITERATE .iterate = zpl_iterate, #else .readdir = zpl_readdir, #endif .fsync = zpl_fsync, .unlocked_ioctl = zpl_ioctl, #ifdef CONFIG_COMPAT .compat_ioctl = zpl_compat_ioctl, #endif };