Index: stable/11/sys/boot/efi/boot1/Makefile =================================================================== --- stable/11/sys/boot/efi/boot1/Makefile (revision 328865) +++ stable/11/sys/boot/efi/boot1/Makefile (revision 328866) @@ -1,143 +1,154 @@ # $FreeBSD$ MAN= .include MK_SSP= no PROG= boot1.sym INTERNALPROG= WARNS?= 6 .if ${MK_ZFS} != "no" # Disable warnings that are currently incompatible with the zfs boot code CWARNFLAGS.zfs_module.c += -Wno-array-bounds CWARNFLAGS.zfs_module.c += -Wno-cast-align CWARNFLAGS.zfs_module.c += -Wno-cast-qual CWARNFLAGS.zfs_module.c += -Wno-missing-prototypes CWARNFLAGS.zfs_module.c += -Wno-sign-compare CWARNFLAGS.zfs_module.c += -Wno-unused-parameter CWARNFLAGS.zfs_module.c += -Wno-unused-function +CWARNFLAGS.skein.c += -Wno-cast-align +.if ${COMPILER_TYPE} == "clang" +CWARNFLAGS.skein.c += -Wno-missing-variable-declarations +.else if ${COMPILER_TYPE} == "gcc" +CWARNFLAGS.skein.c += -Wno-missing-declarations .endif +.endif # architecture-specific loader code SRCS= boot1.c self_reloc.c start.S ufs_module.c .if ${MK_ZFS} != "no" SRCS+= zfs_module.c +SRCS+= skein.c skein_block.c +# Do not unroll skein loops, reduce code size +CFLAGS+= -DSKEIN_LOOP=111 +.PATH: ${.CURDIR}/../../../crypto/skein .endif .if ${COMPILER_TYPE} == "gcc" && ${COMPILER_VERSION} > 40201 CWARNFLAGS.self_reloc.c+= -Wno-error=maybe-uninitialized .endif CFLAGS+= -I. CFLAGS+= -I${.CURDIR}/../include CFLAGS+= -I${.CURDIR}/../include/${MACHINE} CFLAGS+= -I${.CURDIR}/../../../contrib/dev/acpica/include CFLAGS+= -I${.CURDIR}/../../.. CFLAGS+= -DEFI_UFS_BOOT .ifdef(EFI_DEBUG) CFLAGS+= -DEFI_DEBUG .endif .if ${MK_ZFS} != "no" CFLAGS+= -I${.CURDIR}/../../zfs/ CFLAGS+= -I${.CURDIR}/../../../cddl/boot/zfs/ +CFLAGS+= -I${.CURDIR}/../../../crypto/skein CFLAGS+= -DEFI_ZFS_BOOT .endif # Always add MI sources and REGULAR efi loader bits .PATH: ${.CURDIR}/../loader/arch/${MACHINE} .PATH: ${.CURDIR}/../loader .PATH: ${.CURDIR}/../../common CFLAGS+= -I${.CURDIR}/../../common FILES= boot1.efi boot1.efifat FILESMODE_boot1.efi= ${BINMODE} LDSCRIPT= ${.CURDIR}/../loader/arch/${MACHINE}/ldscript.${MACHINE} LDFLAGS+= -Wl,-T${LDSCRIPT} -Wl,-Bsymbolic -shared .if ${MACHINE_CPUARCH} == "aarch64" CFLAGS+= -msoft-float -mgeneral-regs-only .endif .if ${MACHINE_CPUARCH} == "amd64" || ${MACHINE_CPUARCH} == "i386" CFLAGS+= -fPIC LDFLAGS+= -Wl,-znocombreloc .endif # # Add libstand for the runtime functions used by the compiler - for example # __aeabi_* (arm) or __divdi3 (i386). # as well as required string and memory functions for all platforms. # DPADD+= ${LIBSTAND} LDADD+= -lstand DPADD+= ${LDSCRIPT} NM?= nm OBJCOPY?= objcopy .if ${MACHINE_CPUARCH} == "amd64" EFI_TARGET= efi-app-x86_64 .elif ${MACHINE_CPUARCH} == "i386" EFI_TARGET= efi-app-ia32 .else EFI_TARGET= binary .endif # Arbitrarily set the PE/COFF header timestamps to 1 Jan 2016 00:00:00 # for build reproducibility. SOURCE_DATE_EPOCH?=1451606400 boot1.efi: ${PROG} if ${NM} ${.ALLSRC} | grep ' U '; then \ echo "Undefined symbols in ${.ALLSRC}"; \ exit 1; \ fi SOURCE_DATE_EPOCH=${SOURCE_DATE_EPOCH} \ ${OBJCOPY} -j .peheader -j .text -j .sdata -j .data \ -j .dynamic -j .dynsym -j .rel.dyn \ -j .rela.dyn -j .reloc -j .eh_frame \ --output-target=${EFI_TARGET} ${.ALLSRC} ${.TARGET} boot1.o: ${.CURDIR}/../../common/ufsread.c # The following inserts our objects into a template FAT file system # created by generate-fat.sh .include "${.CURDIR}/Makefile.fat" BOOT1_MAXSIZE?= 131072 boot1.efifat: boot1.efi @set -- `ls -l ${.ALLSRC}`; \ x=$$(($$5-${BOOT1_MAXSIZE})); \ if [ $$x -ge 0 ]; then \ echo "boot1 $$x bytes too large; regenerate FAT templates?" >&2 ;\ exit 1; \ fi echo ${.OBJDIR} uudecode ${.CURDIR}/fat-${MACHINE}.tmpl.bz2.uu mv fat-${MACHINE}.tmpl.bz2 ${.TARGET}.bz2 bzip2 -f -d ${.TARGET}.bz2 ${DD} if=${.ALLSRC} of=${.TARGET} seek=${BOOT1_OFFSET} conv=notrunc CLEANFILES= boot1.efi boot1.efifat .include beforedepend ${OBJS}: machine CLEANFILES+= machine machine: .NOMETA ln -sf ${.CURDIR}/../../../${MACHINE}/include machine .if ${MACHINE_CPUARCH} == "amd64" || ${MACHINE_CPUARCH} == "i386" beforedepend ${OBJS}: x86 CLEANFILES+= x86 x86: .NOMETA ln -sf ${.CURDIR}/../../../x86/include x86 .endif Index: stable/11/sys/boot/efi/loader/Makefile =================================================================== --- stable/11/sys/boot/efi/loader/Makefile (revision 328865) +++ stable/11/sys/boot/efi/loader/Makefile (revision 328866) @@ -1,167 +1,172 @@ # $FreeBSD$ MAN= .include MK_SSP= no PROG= loader.sym INTERNALPROG= WARNS?= 3 # architecture-specific loader code SRCS= autoload.c \ bootinfo.c \ conf.c \ copy.c \ devicename.c \ main.c \ self_reloc.c \ smbios.c \ vers.c .if ${MK_ZFS} != "no" SRCS+= zfs.c .PATH: ${.CURDIR}/../../zfs +SRCS+= skein.c skein_block.c +# Do not unroll skein loops, reduce code size +CFLAGS+= -DSKEIN_LOOP=111 +.PATH: ${.CURDIR}/../../../crypto/skein # Disable warnings that are currently incompatible with the zfs boot code CWARNFLAGS.zfs.c+= -Wno-sign-compare CWARNFLAGS.zfs.c+= -Wno-array-bounds CWARNFLAGS.zfs.c+= -Wno-missing-prototypes .endif .if ${COMPILER_TYPE} == "gcc" && ${COMPILER_VERSION} > 40201 CWARNFLAGS.self_reloc.c+= -Wno-error=maybe-uninitialized .endif # We implement a slightly non-standard %S in that it always takes a # CHAR16 that's common in UEFI-land instead of a wchar_t. This only # seems to matter on arm64 where wchar_t defaults to an int instead # of a short. There's no good cast to use here so just ignore the # warnings for now. CWARNFLAGS.main.c+= -Wno-format .PATH: ${.CURDIR}/arch/${MACHINE} # For smbios.c .PATH: ${.CURDIR}/../../i386/libi386 .include "${.CURDIR}/arch/${MACHINE}/Makefile.inc" CFLAGS+= -I${.CURDIR} CFLAGS+= -I${.CURDIR}/arch/${MACHINE} CFLAGS+= -I${.CURDIR}/../include CFLAGS+= -I${.CURDIR}/../include/${MACHINE} CFLAGS+= -I${.CURDIR}/../../../contrib/dev/acpica/include CFLAGS+= -I${.CURDIR}/../../.. CFLAGS+= -I${.CURDIR}/../../i386/libi386 .if ${MK_ZFS} != "no" CFLAGS+= -I${.CURDIR}/../../zfs CFLAGS+= -I${.CURDIR}/../../../cddl/boot/zfs +CFLAGS+= -I${.CURDIR}/../../../crypto/skein CFLAGS+= -DEFI_ZFS_BOOT .endif CFLAGS+= -DNO_PCI -DEFI # make buildenv doesn't set DESTDIR, this means LIBSTAND # will be wrong when crossbuilding. .if exists(${.OBJDIR}/../../../../lib/libstand/libstand.a) LIBSTAND= ${.OBJDIR}/../../../../lib/libstand/libstand.a .endif .if !defined(BOOT_HIDE_SERIAL_NUMBERS) # Export serial numbers, UUID, and asset tag from loader. CFLAGS+= -DSMBIOS_SERIAL_NUMBERS .if defined(BOOT_LITTLE_ENDIAN_UUID) # Use little-endian UUID format as defined in SMBIOS 2.6. CFLAGS+= -DSMBIOS_LITTLE_ENDIAN_UUID .elif defined(BOOT_NETWORK_ENDIAN_UUID) # Use network-endian UUID format for backward compatibility. CFLAGS+= -DSMBIOS_NETWORK_ENDIAN_UUID .endif .endif .if ${MK_FORTH} != "no" BOOT_FORTH= yes CFLAGS+= -DBOOT_FORTH CFLAGS+= -I${.CURDIR}/../../ficl CFLAGS+= -I${.CURDIR}/../../ficl/${MACHINE_CPUARCH} LIBFICL= ${.OBJDIR}/../../ficl/libficl.a .endif LOADER_FDT_SUPPORT?= no .if ${MK_FDT} != "no" && ${LOADER_FDT_SUPPORT} != "no" CFLAGS+= -I${.CURDIR}/../../fdt CFLAGS+= -I${.OBJDIR}/../../fdt CFLAGS+= -DLOADER_FDT_SUPPORT LIBEFI_FDT= ${.OBJDIR}/../../efi/fdt/libefi_fdt.a LIBFDT= ${.OBJDIR}/../../fdt/libfdt.a .endif # Include bcache code. HAVE_BCACHE= yes .if defined(EFI_STAGING_SIZE) CFLAGS+= -DEFI_STAGING_SIZE=${EFI_STAGING_SIZE} .endif # Always add MI sources .PATH: ${.CURDIR}/../../common .include "${.CURDIR}/../../common/Makefile.inc" CFLAGS+= -I${.CURDIR}/../../common FILES+= loader.efi FILESMODE_loader.efi= ${BINMODE} LDSCRIPT= ${.CURDIR}/arch/${MACHINE}/ldscript.${MACHINE} LDFLAGS+= -Wl,-T${LDSCRIPT} -Wl,-Bsymbolic -shared CLEANFILES+= loader.efi NEWVERSWHAT= "EFI loader" ${MACHINE} NM?= nm OBJCOPY?= objcopy .if ${MACHINE_CPUARCH} == "amd64" EFI_TARGET= efi-app-x86_64 .elif ${MACHINE_CPUARCH} == "i386" EFI_TARGET= efi-app-ia32 .else EFI_TARGET= binary .endif # Arbitrarily set the PE/COFF header timestamps to 1 Jan 2016 00:00:00 # for build reproducibility. SOURCE_DATE_EPOCH?=1451606400 loader.efi: ${PROG} if ${NM} ${.ALLSRC} | grep ' U '; then \ echo "Undefined symbols in ${.ALLSRC}"; \ exit 1; \ fi SOURCE_DATE_EPOCH=${SOURCE_DATE_EPOCH} \ ${OBJCOPY} -j .peheader -j .text -j .sdata -j .data \ -j .dynamic -j .dynsym -j .rel.dyn \ -j .rela.dyn -j .reloc -j .eh_frame -j set_Xcommand_set \ --output-target=${EFI_TARGET} ${.ALLSRC} ${.TARGET} LIBEFI= ${.OBJDIR}/../libefi/libefi.a DPADD= ${LIBFICL} ${LIBEFI} ${LIBFDT} ${LIBEFI_FDT} ${LIBSTAND} \ ${LDSCRIPT} LDADD= ${LIBFICL} ${LIBEFI} ${LIBFDT} ${LIBEFI_FDT} ${LIBSTAND} .include beforedepend ${OBJS}: machine CLEANFILES+= machine machine: .NOMETA ln -sf ${.CURDIR}/../../../${MACHINE}/include machine .if ${MACHINE_CPUARCH} == "amd64" || ${MACHINE_CPUARCH} == "i386" beforedepend ${OBJS}: x86 CLEANFILES+= x86 x86: .NOMETA ln -sf ${.CURDIR}/../../../x86/include x86 .endif Index: stable/11/sys/boot/i386/boot2/Makefile =================================================================== --- stable/11/sys/boot/i386/boot2/Makefile (revision 328865) +++ stable/11/sys/boot/i386/boot2/Makefile (revision 328866) @@ -1,120 +1,120 @@ # $FreeBSD$ .include FILES= boot boot1 boot2 NM?= nm # A value of 0x80 enables LBA support. BOOT_BOOT1_FLAGS?= 0x80 BOOT_COMCONSOLE_PORT?= 0x3f8 BOOT_COMCONSOLE_SPEED?= 9600 B2SIOFMT?= 0x3 REL1= 0x700 ORG1= 0x7c00 ORG2= 0x2000 # Decide level of UFS support. BOOT2_UFS?= UFS1_AND_UFS2 #BOOT2_UFS?= UFS2_ONLY #BOOT2_UFS?= UFS1_ONLY CFLAGS= -fomit-frame-pointer \ -mrtd \ -mregparm=3 \ -DUSE_XREAD \ -D${BOOT2_UFS} \ -DFLAGS=${BOOT_BOOT1_FLAGS} \ -DSIOPRT=${BOOT_COMCONSOLE_PORT} \ -DSIOFMT=${B2SIOFMT} \ -DSIOSPD=${BOOT_COMCONSOLE_SPEED} \ -I${.CURDIR}/../../common \ -I${.CURDIR}/../btx/lib -I. \ - -Wall -Waggregate-return -Wbad-function-cast -Wcast-align \ + -Wall -Waggregate-return -Wbad-function-cast -Wno-cast-align \ -Wmissing-declarations -Wmissing-prototypes -Wnested-externs \ -Wpointer-arith -Wshadow -Wstrict-prototypes -Wwrite-strings \ -Winline CFLAGS.gcc+= -Os \ -fno-asynchronous-unwind-tables \ -fno-guess-branch-probability \ -fno-unit-at-a-time \ --param max-inline-insns-single=100 .if ${COMPILER_TYPE} == "gcc" && ${COMPILER_VERSION} <= 40201 CFLAGS.gcc+= -mno-align-long-strings .endif CFLAGS.clang+= -Oz ${CLANG_OPT_SMALL} LD_FLAGS=-static -N --gc-sections # Pick up ../Makefile.inc early. .include CLEANFILES= boot boot: boot1 boot2 cat boot1 boot2 > boot CLEANFILES+= boot1 boot1.out boot1.o boot1: boot1.out ${OBJCOPY} -S -O binary boot1.out ${.TARGET} boot1.out: boot1.o ${LD} ${LD_FLAGS} -e start -Ttext ${ORG1} -o ${.TARGET} boot1.o CLEANFILES+= boot2 boot2.ld boot2.ldr boot2.bin boot2.out boot2.o \ boot2.s boot2.s.tmp boot2.h sio.o BOOT2SIZE= 7680 boot2: boot2.ld @set -- `ls -l ${.ALLSRC}`; x=$$((${BOOT2SIZE}-$$5)); \ echo "$$x bytes available"; test $$x -ge 0 ${DD} if=${.ALLSRC} of=${.TARGET} obs=${BOOT2SIZE} conv=osync boot2.ld: boot2.ldr boot2.bin ${BTXKERN} btxld -v -E ${ORG2} -f bin -b ${BTXKERN} -l boot2.ldr \ -o ${.TARGET} -P 1 boot2.bin boot2.ldr: ${DD} if=/dev/zero of=${.TARGET} bs=512 count=1 boot2.bin: boot2.out ${OBJCOPY} -S -O binary boot2.out ${.TARGET} boot2.out: ${BTXCRT} boot2.o sio.o ${LD} ${LD_FLAGS} -Ttext ${ORG2} -o ${.TARGET} ${.ALLSRC} boot2.o: boot2.s ${CC} ${ACFLAGS} -c boot2.s SRCS= boot2.c boot2.h boot2.s: boot2.c boot2.h ${.CURDIR}/../../common/ufsread.c ${CC} ${CFLAGS} -S -o boot2.s.tmp ${.CURDIR}/boot2.c sed -e '/align/d' -e '/nop/d' < boot2.s.tmp > boot2.s rm -f boot2.s.tmp boot2.h: boot1.out ${NM} -t d ${.ALLSRC} | awk '/([0-9])+ T xread/ \ { x = $$1 - ORG1; \ printf("#define XREADORG %#x\n", REL1 + x) }' \ ORG1=`printf "%d" ${ORG1}` \ REL1=`printf "%d" ${REL1}` > ${.TARGET} .if ${MACHINE_CPUARCH} == "amd64" beforedepend boot2.s: machine CLEANFILES+= machine machine: ${.CURDIR}/../../../i386/include .NOMETA ln -sf ${.ALLSRC} ${.TARGET} .endif .include # XXX: clang integrated-as doesn't grok .codeNN directives yet CFLAGS.boot1.S= ${CLANG_NO_IAS} Index: stable/11/sys/boot/i386/gptboot/Makefile =================================================================== --- stable/11/sys/boot/i386/gptboot/Makefile (revision 328865) +++ stable/11/sys/boot/i386/gptboot/Makefile (revision 328866) @@ -1,92 +1,92 @@ # $FreeBSD$ .PATH: ${.CURDIR}/../boot2 ${.CURDIR}/../common ${.CURDIR}/../../common FILES= gptboot MAN= gptboot.8 NM?= nm BOOT_COMCONSOLE_PORT?= 0x3f8 BOOT_COMCONSOLE_SPEED?= 9600 B2SIOFMT?= 0x3 REL1= 0x700 ORG1= 0x7c00 ORG2= 0x0 # Decide level of UFS support. GPTBOOT_UFS?= UFS1_AND_UFS2 #GPTBOOT_UFS?= UFS2_ONLY #GPTBOOT_UFS?= UFS1_ONLY CFLAGS= -DBOOTPROG=\"gptboot\" \ -O1 \ -DGPT \ -D${GPTBOOT_UFS} \ -DSIOPRT=${BOOT_COMCONSOLE_PORT} \ -DSIOFMT=${B2SIOFMT} \ -DSIOSPD=${BOOT_COMCONSOLE_SPEED} \ -I${.CURDIR}/../../common \ -I${.CURDIR}/../common \ -I${.CURDIR}/../btx/lib -I. \ -I${.CURDIR}/../boot2 \ -I${.CURDIR}/../../.. \ - -Wall -Waggregate-return -Wbad-function-cast -Wcast-align \ + -Wall -Waggregate-return -Wbad-function-cast -Wno-cast-align \ -Wmissing-declarations -Wmissing-prototypes -Wnested-externs \ -Wpointer-arith -Wshadow -Wstrict-prototypes -Wwrite-strings \ - -Winline + -Winline -Wno-pointer-sign CFLAGS.gcc+= --param max-inline-insns-single=100 .if !defined(LOADER_NO_GELI_SUPPORT) CFLAGS+= -DLOADER_GELI_SUPPORT CFLAGS+= -I${.CURDIR}/../../geli LIBGELIBOOT= ${.OBJDIR}/../../geli/libgeliboot.a .PATH: ${.CURDIR}/../../../opencrypto OPENCRYPTO_XTS= xform_aes_xts.o .endif LD_FLAGS=-static -N --gc-sections LIBSTAND= ${.OBJDIR}/../../libstand32/libstand.a # Pick up ../Makefile.inc early. .include CLEANFILES= gptboot gptboot: gptldr.bin gptboot.bin ${BTXKERN} btxld -v -E ${ORG2} -f bin -b ${BTXKERN} -l gptldr.bin \ -o ${.TARGET} gptboot.bin CLEANFILES+= gptldr.bin gptldr.out gptldr.o gptldr.bin: gptldr.out ${OBJCOPY} -S -O binary gptldr.out ${.TARGET} gptldr.out: gptldr.o ${LD} ${LD_FLAGS} -e start -Ttext ${ORG1} -o ${.TARGET} gptldr.o CLEANFILES+= gptboot.bin gptboot.out gptboot.o sio.o crc32.o drv.o \ cons.o util.o ${OPENCRYPTO_XTS} gptboot.bin: gptboot.out ${OBJCOPY} -S -O binary gptboot.out ${.TARGET} gptboot.out: ${BTXCRT} gptboot.o sio.o crc32.o drv.o cons.o util.o ${OPENCRYPTO_XTS} ${LD} ${LD_FLAGS} -Ttext ${ORG2} -o ${.TARGET} ${.ALLSRC} ${LIBSTAND} ${LIBGELIBOOT} gptboot.o: ${.CURDIR}/../../common/ufsread.c .if ${MACHINE_CPUARCH} == "amd64" beforedepend gptboot.o: machine CLEANFILES+= machine machine: .NOMETA ln -sf ${.CURDIR}/../../../i386/include machine .endif .include # XXX: clang integrated-as doesn't grok .codeNN directives yet CFLAGS.gptldr.S= ${CLANG_NO_IAS} Index: stable/11/sys/boot/i386/gptboot/gptldr.S =================================================================== --- stable/11/sys/boot/i386/gptboot/gptldr.S (revision 328865) +++ stable/11/sys/boot/i386/gptboot/gptldr.S (revision 328866) @@ -1,142 +1,142 @@ /*- * Copyright (c) 2007 Yahoo!, Inc. * All rights reserved. * Written by: John Baldwin * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the author nor the names of any co-contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ * * Partly from: src/sys/boot/i386/boot2/boot1.S 1.31 */ /* Memory Locations */ .set MEM_REL,0x700 # Relocation address .set MEM_ARG,0x900 # Arguments .set MEM_ORG,0x7c00 # Origin .set MEM_BUF,0x8cec # Load area .set MEM_BTX,0x9000 # BTX start .set MEM_JMP,0x9010 # BTX entry point .set MEM_USR,0xa000 # Client start .set BDA_BOOT,0x472 # Boot howto flag /* Misc. Constants */ .set SIZ_PAG,0x1000 # Page size .set SIZ_SEC,0x200 # Sector size - .set COPY_BLKS,0x4 # Number of blocks + .set COPY_BLKS,0x8 # Number of blocks # to copy for boot2 .set COPY_BLK_SZ,0x8000 # Copy in 32k blocks; must be # a multiple of 16 bytes .globl start .code16 /* * Copy BTX and boot2 to the right locations and start it all up. */ /* * Setup the segment registers to flat addressing (segment 0) and setup the * stack to end just below the start of our code. */ start: xor %cx,%cx # Zero mov %cx,%es # Address mov %cx,%ds # data mov %cx,%ss # Set up mov $start,%sp # stack /* * BTX is right after us at 'end'. We read the length of BTX out of * its header to find boot2. We need to copy boot2 to MEM_USR and BTX * to MEM_BTX. Since those might overlap, we have to copy boot2 * backwards first and then copy BTX. We aren't sure exactly how long * boot2 is, but it's currently under 128kB so we'll copy 4 blocks of 32kB * each; this can be adjusted via COPY_BLK and COPY_BLK_SZ above. */ mov $end,%bx # BTX mov 0xa(%bx),%si # Get BTX length and set add %bx,%si # %si to start of boot2 dec %si # Set %ds:%si to point at the mov %si,%ax # last byte we want to copy shr $4,%ax # from boot2, with %si made as add $(COPY_BLKS*COPY_BLK_SZ/16),%ax # small as possible. and $0xf,%si # mov %ax,%ds # mov $MEM_USR/16,%ax # Set %es:(-1) to point at add $(COPY_BLKS*COPY_BLK_SZ/16),%ax # the last byte we mov %ax,%es # want to copy boot2 into. mov $COPY_BLKS,%bx # Copy COPY_BLKS 32k blocks copyloop: add $COPY_BLK_SZ,%si # Adjust %ds:%si to point at mov %ds,%ax # the end of the next 32k to sub $COPY_BLK_SZ/16,%ax # copy from boot2 mov %ax,%ds mov $COPY_BLK_SZ-1,%di # Adjust %es:%di to point at mov %es,%ax # the end of the next 32k into sub $COPY_BLK_SZ/16,%ax # which we want boot2 copied mov %ax,%es mov $COPY_BLK_SZ,%cx # Copy 32k std rep movsb dec %bx jnz copyloop mov %cx,%ds # Reset %ds and %es mov %cx,%es mov $end,%bx # BTX mov 0xa(%bx),%cx # Get BTX length and set mov %bx,%si # %si to end of BTX mov $MEM_BTX,%di # %di -> end of BTX at add %cx,%si # MEM_BTX add %cx,%di dec %si dec %di rep movsb # Move BTX cld # String ops inc /* * Enable A20 so we can access memory above 1 meg. * Use the zero-valued %cx as a timeout for embedded hardware which do not * have a keyboard controller. */ seta20: cli # Disable interrupts seta20.1: dec %cx # Timeout? jz seta20.3 # Yes inb $0x64,%al # Get status testb $0x2,%al # Busy? jnz seta20.1 # Yes movb $0xd1,%al # Command: Write outb %al,$0x64 # output port seta20.2: inb $0x64,%al # Get status testb $0x2,%al # Busy? jnz seta20.2 # Yes movb $0xdf,%al # Enable outb %al,$0x60 # A20 seta20.3: sti # Enable interrupts /* * Save drive number from BIOS so boot2 can see it and start BTX. */ movb %dl,MEM_ARG jmp MEM_JMP # Start BTX end: Index: stable/11/sys/boot/i386/gptzfsboot/Makefile =================================================================== --- stable/11/sys/boot/i386/gptzfsboot/Makefile (revision 328865) +++ stable/11/sys/boot/i386/gptzfsboot/Makefile (revision 328866) @@ -1,90 +1,102 @@ # $FreeBSD$ +.include + .PATH: ${.CURDIR}/../boot2 ${.CURDIR}/../gptboot \ ${.CURDIR}/../zfsboot ${.CURDIR}/../common \ - ${.CURDIR}/../../common + ${.CURDIR}/../../common ${.CURDIR}/../../../crypto/skein FILES= gptzfsboot MAN= gptzfsboot.8 NM?= nm BOOT_COMCONSOLE_PORT?= 0x3f8 BOOT_COMCONSOLE_SPEED?= 9600 B2SIOFMT?= 0x3 REL1= 0x700 ORG1= 0x7c00 ORG2= 0x0 CFLAGS= -DBOOTPROG=\"gptzfsboot\" \ -O1 \ -DGPT -DZFS -DBOOT2 \ -DSIOPRT=${BOOT_COMCONSOLE_PORT} \ -DSIOFMT=${B2SIOFMT} \ -DSIOSPD=${BOOT_COMCONSOLE_SPEED} \ -I${.CURDIR}/../../common \ -I${.CURDIR}/../common \ -I${.CURDIR}/../../zfs \ -I${.CURDIR}/../../../cddl/boot/zfs \ + -I${.CURDIR}/../../../crypto/skein \ -I${.CURDIR}/../btx/lib -I. \ -I${.CURDIR}/../boot2 \ -I${.CURDIR}/../../.. \ - -Wall -Waggregate-return -Wbad-function-cast -Wcast-align \ + -Wall -Waggregate-return -Wbad-function-cast -Wno-cast-align \ -Wmissing-declarations -Wmissing-prototypes -Wnested-externs \ -Wpointer-arith -Wshadow -Wstrict-prototypes -Wwrite-strings \ - -Winline + -Winline -Wno-pointer-sign +.if ${COMPILER_TYPE} == "clang" || \ + (${COMPILER_TYPE} == "gcc" && ${COMPILER_VERSION} > 40201) +CFLAGS+= -Wno-tentative-definition-incomplete-type +.endif + +# Do not unroll skein loops, reduce code size +CFLAGS+= -DSKEIN_LOOP=111 + .if !defined(LOADER_NO_GELI_SUPPORT) CFLAGS+= -DLOADER_GELI_SUPPORT CFLAGS+= -I${.CURDIR}/../../geli LIBGELIBOOT= ${.OBJDIR}/../../geli/libgeliboot.a .PATH: ${.CURDIR}/../../../opencrypto OPENCRYPTO_XTS= xform_aes_xts.o .endif CFLAGS.gcc+= --param max-inline-insns-single=100 LD_FLAGS=-static -N --gc-sections LIBSTAND= ${.OBJDIR}/../../libstand32/libstand.a # Pick up ../Makefile.inc early. .include CLEANFILES= gptzfsboot gptzfsboot: gptldr.bin gptzfsboot.bin ${BTXKERN} btxld -v -E ${ORG2} -f bin -b ${BTXKERN} -l gptldr.bin \ -o ${.TARGET} gptzfsboot.bin CLEANFILES+= gptldr.bin gptldr.out gptldr.o gptldr.bin: gptldr.out ${OBJCOPY} -S -O binary gptldr.out ${.TARGET} gptldr.out: gptldr.o ${LD} ${LD_FLAGS} -e start -Ttext ${ORG1} -o ${.TARGET} gptldr.o CLEANFILES+= gptzfsboot.bin gptzfsboot.out zfsboot.o sio.o cons.o \ - drv.o gpt.o util.o ${OPENCRYPTO_XTS} + drv.o gpt.o util.o skein.o skein_block.o ${OPENCRYPTO_XTS} gptzfsboot.bin: gptzfsboot.out ${OBJCOPY} -S -O binary gptzfsboot.out ${.TARGET} -gptzfsboot.out: ${BTXCRT} zfsboot.o sio.o gpt.o drv.o cons.o util.o ${OPENCRYPTO_XTS} +gptzfsboot.out: ${BTXCRT} zfsboot.o sio.o gpt.o drv.o cons.o util.o \ + skein.o skein_block.o ${OPENCRYPTO_XTS} ${LD} ${LD_FLAGS} -Ttext ${ORG2} -o ${.TARGET} ${.ALLSRC} ${LIBSTAND} ${LIBGELIBOOT} zfsboot.o: ${.CURDIR}/../../zfs/zfsimpl.c .if ${MACHINE_CPUARCH} == "amd64" beforedepend zfsboot.o: machine CLEANFILES+= machine machine: .NOMETA ln -sf ${.CURDIR}/../../../i386/include machine .endif .include # XXX: clang integrated-as doesn't grok .codeNN directives yet CFLAGS.gptldr.S= ${CLANG_NO_IAS} Index: stable/11/sys/boot/i386/zfsboot/Makefile =================================================================== --- stable/11/sys/boot/i386/zfsboot/Makefile (revision 328865) +++ stable/11/sys/boot/i386/zfsboot/Makefile (revision 328866) @@ -1,95 +1,100 @@ # $FreeBSD$ -.PATH: ${.CURDIR}/../boot2 ${.CURDIR}/../common ${.CURDIR}/../../common +.PATH: ${.CURDIR}/../boot2 ${.CURDIR}/../common \ + ${.CURDIR}/../../common ${.CURDIR}/../../../crypto/skein FILES= zfsboot MAN= zfsboot.8 NM?= nm BOOT_COMCONSOLE_PORT?= 0x3f8 BOOT_COMCONSOLE_SPEED?= 9600 B2SIOFMT?= 0x3 REL1= 0x700 ORG1= 0x7c00 ORG2= 0x2000 CFLAGS= -DBOOTPROG=\"zfsboot\" \ -O1 \ -DZFS -DBOOT2 \ -DSIOPRT=${BOOT_COMCONSOLE_PORT} \ -DSIOFMT=${B2SIOFMT} \ -DSIOSPD=${BOOT_COMCONSOLE_SPEED} \ -I${.CURDIR}/../../common \ -I${.CURDIR}/../common \ -I${.CURDIR}/../../zfs \ -I${.CURDIR}/../../../cddl/boot/zfs \ + -I${.CURDIR}/../../../crypto/skein \ -I${.CURDIR}/../btx/lib -I. \ -I${.CURDIR}/../boot2 \ - -Wall -Waggregate-return -Wbad-function-cast -Wcast-align \ + -Wall -Waggregate-return -Wbad-function-cast -Wno-cast-align \ -Wmissing-declarations -Wmissing-prototypes -Wnested-externs \ -Wpointer-arith -Wshadow -Wstrict-prototypes -Wwrite-strings \ -Winline CFLAGS.gcc+= --param max-inline-insns-single=100 +# Do not unroll skein loops, reduce code size +CFLAGS+= -DSKEIN_LOOP=111 LD_FLAGS=-static -N --gc-sections LIBSTAND= ${.OBJDIR}/../../libstand32/libstand.a # Pick up ../Makefile.inc early. .include CLEANFILES= zfsboot zfsboot: zfsboot1 zfsboot2 cat zfsboot1 zfsboot2 > zfsboot CLEANFILES+= zfsboot1 zfsldr.out zfsldr.o zfsboot1: zfsldr.out ${OBJCOPY} -S -O binary zfsldr.out ${.TARGET} zfsldr.out: zfsldr.o ${LD} ${LD_FLAGS} -e start -Ttext ${ORG1} -o ${.TARGET} zfsldr.o CLEANFILES+= zfsboot2 zfsboot.ld zfsboot.ldr zfsboot.bin zfsboot.out \ - zfsboot.o zfsboot.s zfsboot.s.tmp sio.o cons.o drv.o util.o + zfsboot.o zfsboot.s zfsboot.s.tmp sio.o cons.o drv.o util.o \ + skein.o skein_block.o -# We currently allow 65536 bytes for zfsboot - in practice it could be +# We currently allow 128k bytes for zfsboot - in practice it could be # any size up to 3.5Mb but keeping it fixed size simplifies zfsldr. # -BOOT2SIZE= 65536 +BOOT2SIZE= 131072 zfsboot2: zfsboot.ld @set -- `ls -l ${.ALLSRC}`; x=$$((${BOOT2SIZE}-$$5)); \ echo "$$x bytes available"; test $$x -ge 0 ${DD} if=${.ALLSRC} of=${.TARGET} obs=${BOOT2SIZE} conv=osync zfsboot.ld: zfsboot.ldr zfsboot.bin ${BTXKERN} btxld -v -E ${ORG2} -f bin -b ${BTXKERN} -l zfsboot.ldr \ -o ${.TARGET} -P 1 zfsboot.bin zfsboot.ldr: cp /dev/null ${.TARGET} zfsboot.bin: zfsboot.out ${OBJCOPY} -S -O binary zfsboot.out ${.TARGET} -zfsboot.out: ${BTXCRT} zfsboot.o sio.o drv.o cons.o util.o +zfsboot.out: ${BTXCRT} zfsboot.o sio.o drv.o cons.o util.o skein.o skein_block.o ${LD} ${LD_FLAGS} -Ttext ${ORG2} -o ${.TARGET} ${.ALLSRC} ${LIBSTAND} SRCS= zfsboot.c .if ${MACHINE_CPUARCH} == "amd64" beforedepend zfsboot.o: machine CLEANFILES+= machine machine: .NOMETA ln -sf ${.CURDIR}/../../../i386/include machine .endif .include # XXX: clang integrated-as doesn't grok .codeNN directives yet CFLAGS.zfsldr.S= ${CLANG_NO_IAS} Index: stable/11/sys/boot/i386/zfsboot/zfsboot.c =================================================================== --- stable/11/sys/boot/i386/zfsboot/zfsboot.c (revision 328865) +++ stable/11/sys/boot/i386/zfsboot/zfsboot.c (revision 328866) @@ -1,1072 +1,1072 @@ /*- * Copyright (c) 1998 Robert Nordier * All rights reserved. * * Redistribution and use in source and binary forms are freely * permitted provided that the above copyright notice and this * paragraph and the following disclaimer are duplicated in all * such forms. * * This software is provided "AS IS" and without any express or * implied warranties, including, without limitation, the implied * warranties of merchantability and fitness for a particular * purpose. */ #include __FBSDID("$FreeBSD$"); #include #include #include #ifdef GPT #include #endif #include #include #include #include #include #include #include #include #include #include "lib.h" #include "rbx.h" #include "drv.h" #include "util.h" #include "cons.h" #include "bootargs.h" #include "paths.h" #include "libzfs.h" #define ARGS 0x900 #define NOPT 14 #define NDEV 3 #define BIOS_NUMDRIVES 0x475 #define DRV_HARD 0x80 #define DRV_MASK 0x7f #define TYPE_AD 0 #define TYPE_DA 1 #define TYPE_MAXHARD TYPE_DA #define TYPE_FD 2 #define DEV_GELIBOOT_BSIZE 4096 extern uint32_t _end; #ifdef GPT static const uuid_t freebsd_zfs_uuid = GPT_ENT_TYPE_FREEBSD_ZFS; #endif static const char optstr[NOPT] = "DhaCcdgmnpqrsv"; /* Also 'P', 'S' */ static const unsigned char flags[NOPT] = { RBX_DUAL, RBX_SERIAL, RBX_ASKNAME, RBX_CDROM, RBX_CONFIG, RBX_KDB, RBX_GDB, RBX_MUTE, RBX_NOINTR, RBX_PAUSE, RBX_QUIET, RBX_DFLTROOT, RBX_SINGLE, RBX_VERBOSE }; uint32_t opts; static const unsigned char dev_maj[NDEV] = {30, 4, 2}; static char cmd[512]; static char cmddup[512]; static char kname[1024]; static char rootname[256]; static int comspeed = SIOSPD; static struct bootinfo bootinfo; static uint32_t bootdev; static struct zfs_boot_args zfsargs; static struct zfsmount zfsmount; vm_offset_t high_heap_base; uint32_t bios_basemem, bios_extmem, high_heap_size; static struct bios_smap smap; /* * The minimum amount of memory to reserve in bios_extmem for the heap. */ -#define HEAP_MIN (3 * 1024 * 1024) +#define HEAP_MIN (64 * 1024 * 1024) static char *heap_next; static char *heap_end; /* Buffers that must not span a 64k boundary. */ #define READ_BUF_SIZE 8192 struct dmadat { char rdbuf[READ_BUF_SIZE]; /* for reading large things */ char secbuf[READ_BUF_SIZE]; /* for MBR/disklabel */ }; static struct dmadat *dmadat; void exit(int); void reboot(void); static void load(void); static int parse(void); static void bios_getmem(void); void *malloc(size_t n); void free(void *ptr); void * malloc(size_t n) { char *p = heap_next; if (p + n > heap_end) { printf("malloc failure\n"); for (;;) ; /* NOTREACHED */ return (0); } heap_next += n; return (p); } void free(void *ptr) { return; } static char * strdup(const char *s) { char *p = malloc(strlen(s) + 1); strcpy(p, s); return (p); } #ifdef LOADER_GELI_SUPPORT #include "geliboot.c" static char gelipw[GELI_PW_MAXLEN]; #endif #include "zfsimpl.c" /* * Read from a dnode (which must be from a ZPL filesystem). */ static int zfs_read(spa_t *spa, const dnode_phys_t *dnode, off_t *offp, void *start, size_t size) { const znode_phys_t *zp = (const znode_phys_t *) dnode->dn_bonus; size_t n; int rc; n = size; if (*offp + n > zp->zp_size) n = zp->zp_size - *offp; rc = dnode_read(spa, dnode, *offp, start, n); if (rc) return (-1); *offp += n; return (n); } /* * Current ZFS pool */ static spa_t *spa; static spa_t *primary_spa; static vdev_t *primary_vdev; /* * A wrapper for dskread that doesn't have to worry about whether the * buffer pointer crosses a 64k boundary. */ static int vdev_read(vdev_t *vdev, void *priv, off_t off, void *buf, size_t bytes) { char *p; daddr_t lba, alignlba; off_t diff; unsigned int nb, alignnb; struct dsk *dsk = (struct dsk *) priv; if ((off & (DEV_BSIZE - 1)) || (bytes & (DEV_BSIZE - 1))) return -1; p = buf; lba = off / DEV_BSIZE; lba += dsk->start; /* * Align reads to 4k else 4k sector GELIs will not decrypt. * Round LBA down to nearest multiple of DEV_GELIBOOT_BSIZE bytes. */ alignlba = rounddown2(off, DEV_GELIBOOT_BSIZE) / DEV_BSIZE; /* * The read must be aligned to DEV_GELIBOOT_BSIZE bytes relative to the * start of the GELI partition, not the start of the actual disk. */ alignlba += dsk->start; diff = (lba - alignlba) * DEV_BSIZE; while (bytes > 0) { nb = bytes / DEV_BSIZE; /* * Ensure that the read size plus the leading offset does not * exceed the size of the read buffer. */ if (nb > (READ_BUF_SIZE - diff) / DEV_BSIZE) nb = (READ_BUF_SIZE - diff) / DEV_BSIZE; /* * Round the number of blocks to read up to the nearest multiple * of DEV_GELIBOOT_BSIZE. */ alignnb = roundup2(nb * DEV_BSIZE + diff, DEV_GELIBOOT_BSIZE) / DEV_BSIZE; if (drvread(dsk, dmadat->rdbuf, alignlba, alignnb)) return -1; #ifdef LOADER_GELI_SUPPORT /* decrypt */ if (is_geli(dsk) == 0) { if (geli_read(dsk, ((alignlba - dsk->start) * DEV_BSIZE), dmadat->rdbuf, alignnb * DEV_BSIZE)) return (-1); } #endif memcpy(p, dmadat->rdbuf + diff, nb * DEV_BSIZE); p += nb * DEV_BSIZE; lba += nb; alignlba += alignnb; bytes -= nb * DEV_BSIZE; /* Don't need the leading offset after the first block. */ diff = 0; } return 0; } static int vdev_write(vdev_t *vdev, void *priv, off_t off, void *buf, size_t bytes) { char *p; daddr_t lba; unsigned int nb; struct dsk *dsk = (struct dsk *) priv; if ((off & (DEV_BSIZE - 1)) || (bytes & (DEV_BSIZE - 1))) return -1; p = buf; lba = off / DEV_BSIZE; lba += dsk->start; while (bytes > 0) { nb = bytes / DEV_BSIZE; if (nb > READ_BUF_SIZE / DEV_BSIZE) nb = READ_BUF_SIZE / DEV_BSIZE; memcpy(dmadat->rdbuf, p, nb * DEV_BSIZE); if (drvwrite(dsk, dmadat->rdbuf, lba, nb)) return -1; p += nb * DEV_BSIZE; lba += nb; bytes -= nb * DEV_BSIZE; } return 0; } static int xfsread(const dnode_phys_t *dnode, off_t *offp, void *buf, size_t nbyte) { if ((size_t)zfs_read(spa, dnode, offp, buf, nbyte) != nbyte) { printf("Invalid format\n"); return -1; } return 0; } /* * Read Pad2 (formerly "Boot Block Header") area of the first * vdev label of the given vdev. */ static int vdev_read_pad2(vdev_t *vdev, char *buf, size_t size) { blkptr_t bp; char *tmp = zap_scratch; off_t off = offsetof(vdev_label_t, vl_pad2); if (size > VDEV_PAD_SIZE) size = VDEV_PAD_SIZE; BP_ZERO(&bp); BP_SET_LSIZE(&bp, VDEV_PAD_SIZE); BP_SET_PSIZE(&bp, VDEV_PAD_SIZE); BP_SET_CHECKSUM(&bp, ZIO_CHECKSUM_LABEL); BP_SET_COMPRESS(&bp, ZIO_COMPRESS_OFF); DVA_SET_OFFSET(BP_IDENTITY(&bp), off); if (vdev_read_phys(vdev, &bp, tmp, off, 0)) return (EIO); memcpy(buf, tmp, size); return (0); } static int vdev_clear_pad2(vdev_t *vdev) { char *zeroes = zap_scratch; uint64_t *end; off_t off = offsetof(vdev_label_t, vl_pad2); memset(zeroes, 0, VDEV_PAD_SIZE); end = (uint64_t *)(zeroes + VDEV_PAD_SIZE); /* ZIO_CHECKSUM_LABEL magic and pre-calcualted checksum for all zeros */ end[-5] = 0x0210da7ab10c7a11; end[-4] = 0x97f48f807f6e2a3f; end[-3] = 0xaf909f1658aacefc; end[-2] = 0xcbd1ea57ff6db48b; end[-1] = 0x6ec692db0d465fab; if (vdev_write(vdev, vdev->v_read_priv, off, zeroes, VDEV_PAD_SIZE)) return (EIO); return (0); } static void bios_getmem(void) { uint64_t size; /* Parse system memory map */ v86.ebx = 0; do { v86.ctl = V86_FLAGS; v86.addr = 0x15; /* int 0x15 function 0xe820*/ v86.eax = 0xe820; v86.ecx = sizeof(struct bios_smap); v86.edx = SMAP_SIG; v86.es = VTOPSEG(&smap); v86.edi = VTOPOFF(&smap); v86int(); if (V86_CY(v86.efl) || (v86.eax != SMAP_SIG)) break; /* look for a low-memory segment that's large enough */ if ((smap.type == SMAP_TYPE_MEMORY) && (smap.base == 0) && (smap.length >= (512 * 1024))) bios_basemem = smap.length; /* look for the first segment in 'extended' memory */ if ((smap.type == SMAP_TYPE_MEMORY) && (smap.base == 0x100000)) { bios_extmem = smap.length; } /* * Look for the largest segment in 'extended' memory beyond * 1MB but below 4GB. */ if ((smap.type == SMAP_TYPE_MEMORY) && (smap.base > 0x100000) && (smap.base < 0x100000000ull)) { size = smap.length; /* * If this segment crosses the 4GB boundary, truncate it. */ if (smap.base + size > 0x100000000ull) size = 0x100000000ull - smap.base; if (size > high_heap_size) { high_heap_size = size; high_heap_base = smap.base; } } } while (v86.ebx != 0); /* Fall back to the old compatibility function for base memory */ if (bios_basemem == 0) { v86.ctl = 0; v86.addr = 0x12; /* int 0x12 */ v86int(); bios_basemem = (v86.eax & 0xffff) * 1024; } /* Fall back through several compatibility functions for extended memory */ if (bios_extmem == 0) { v86.ctl = V86_FLAGS; v86.addr = 0x15; /* int 0x15 function 0xe801*/ v86.eax = 0xe801; v86int(); if (!V86_CY(v86.efl)) { bios_extmem = ((v86.ecx & 0xffff) + ((v86.edx & 0xffff) * 64)) * 1024; } } if (bios_extmem == 0) { v86.ctl = 0; v86.addr = 0x15; /* int 0x15 function 0x88*/ v86.eax = 0x8800; v86int(); bios_extmem = (v86.eax & 0xffff) * 1024; } /* * If we have extended memory and did not find a suitable heap * region in the SMAP, use the last 3MB of 'extended' memory as a * high heap candidate. */ if (bios_extmem >= HEAP_MIN && high_heap_size < HEAP_MIN) { high_heap_size = HEAP_MIN; high_heap_base = bios_extmem + 0x100000 - HEAP_MIN; } } /* * Try to detect a device supported by the legacy int13 BIOS */ static int int13probe(int drive) { v86.ctl = V86_FLAGS; v86.addr = 0x13; v86.eax = 0x800; v86.edx = drive; v86int(); if (!V86_CY(v86.efl) && /* carry clear */ ((v86.edx & 0xff) != (drive & DRV_MASK))) { /* unit # OK */ if ((v86.ecx & 0x3f) == 0) { /* absurd sector size */ return(0); /* skip device */ } return (1); } return(0); } /* * We call this when we find a ZFS vdev - ZFS consumes the dsk * structure so we must make a new one. */ static struct dsk * copy_dsk(struct dsk *dsk) { struct dsk *newdsk; newdsk = malloc(sizeof(struct dsk)); *newdsk = *dsk; return (newdsk); } static void probe_drive(struct dsk *dsk) { #ifdef GPT struct gpt_hdr hdr; struct gpt_ent *ent; unsigned part, entries_per_sec; daddr_t slba; #endif #if defined(GPT) || defined(LOADER_GELI_SUPPORT) daddr_t elba; #endif struct dos_partition *dp; char *sec; unsigned i; /* * If we find a vdev on the whole disk, stop here. */ if (vdev_probe(vdev_read, dsk, NULL) == 0) return; #ifdef LOADER_GELI_SUPPORT /* * Taste the disk, if it is GELI encrypted, decrypt it and check to see if * it is a usable vdev then. Otherwise dig * out the partition table and probe each slice/partition * in turn for a vdev or GELI encrypted vdev. */ elba = drvsize(dsk); if (elba > 0) { elba--; } if (geli_taste(vdev_read, dsk, elba) == 0) { if (geli_passphrase(&gelipw, dsk->unit, ':', 0, dsk) == 0) { if (vdev_probe(vdev_read, dsk, NULL) == 0) { return; } } } #endif /* LOADER_GELI_SUPPORT */ sec = dmadat->secbuf; dsk->start = 0; #ifdef GPT /* * First check for GPT. */ if (drvread(dsk, sec, 1, 1)) { return; } memcpy(&hdr, sec, sizeof(hdr)); if (memcmp(hdr.hdr_sig, GPT_HDR_SIG, sizeof(hdr.hdr_sig)) != 0 || hdr.hdr_lba_self != 1 || hdr.hdr_revision < 0x00010000 || hdr.hdr_entsz < sizeof(*ent) || DEV_BSIZE % hdr.hdr_entsz != 0) { goto trymbr; } /* * Probe all GPT partitions for the presence of ZFS pools. We * return the spa_t for the first we find (if requested). This * will have the effect of booting from the first pool on the * disk. * * If no vdev is found, GELI decrypting the device and try again */ entries_per_sec = DEV_BSIZE / hdr.hdr_entsz; slba = hdr.hdr_lba_table; elba = slba + hdr.hdr_entries / entries_per_sec; while (slba < elba) { dsk->start = 0; if (drvread(dsk, sec, slba, 1)) return; for (part = 0; part < entries_per_sec; part++) { ent = (struct gpt_ent *)(sec + part * hdr.hdr_entsz); if (memcmp(&ent->ent_type, &freebsd_zfs_uuid, sizeof(uuid_t)) == 0) { dsk->start = ent->ent_lba_start; dsk->slice = part + 1; dsk->part = 255; if (vdev_probe(vdev_read, dsk, NULL) == 0) { /* * This slice had a vdev. We need a new dsk * structure now since the vdev now owns this one. */ dsk = copy_dsk(dsk); } #ifdef LOADER_GELI_SUPPORT else if (geli_taste(vdev_read, dsk, ent->ent_lba_end - ent->ent_lba_start) == 0) { if (geli_passphrase(&gelipw, dsk->unit, 'p', dsk->slice, dsk) == 0) { /* * This slice has GELI, check it for ZFS. */ if (vdev_probe(vdev_read, dsk, NULL) == 0) { /* * This slice had a vdev. We need a new dsk * structure now since the vdev now owns this one. */ dsk = copy_dsk(dsk); } break; } } #endif /* LOADER_GELI_SUPPORT */ } } slba++; } return; trymbr: #endif /* GPT */ if (drvread(dsk, sec, DOSBBSECTOR, 1)) return; dp = (void *)(sec + DOSPARTOFF); for (i = 0; i < NDOSPART; i++) { if (!dp[i].dp_typ) continue; dsk->start = dp[i].dp_start; dsk->slice = i + 1; if (vdev_probe(vdev_read, dsk, NULL) == 0) { dsk = copy_dsk(dsk); } #ifdef LOADER_GELI_SUPPORT else if (geli_taste(vdev_read, dsk, dp[i].dp_size - dp[i].dp_start) == 0) { if (geli_passphrase(&gelipw, dsk->unit, 's', i, dsk) == 0) { /* * This slice has GELI, check it for ZFS. */ if (vdev_probe(vdev_read, dsk, NULL) == 0) { /* * This slice had a vdev. We need a new dsk * structure now since the vdev now owns this one. */ dsk = copy_dsk(dsk); } break; } } #endif /* LOADER_GELI_SUPPORT */ } } int main(void) { dnode_phys_t dn; off_t off; struct dsk *dsk; int autoboot, i; int nextboot; int rc; dmadat = (void *)(roundup2(__base + (int32_t)&_end, 0x10000) - __base); bios_getmem(); if (high_heap_size > 0) { heap_end = PTOV(high_heap_base + high_heap_size); heap_next = PTOV(high_heap_base); } else { heap_next = (char *)dmadat + sizeof(*dmadat); heap_end = (char *)PTOV(bios_basemem); } dsk = malloc(sizeof(struct dsk)); dsk->drive = *(uint8_t *)PTOV(ARGS); dsk->type = dsk->drive & DRV_HARD ? TYPE_AD : TYPE_FD; dsk->unit = dsk->drive & DRV_MASK; dsk->slice = *(uint8_t *)PTOV(ARGS + 1) + 1; dsk->part = 0; dsk->start = 0; dsk->init = 0; bootinfo.bi_version = BOOTINFO_VERSION; bootinfo.bi_size = sizeof(bootinfo); bootinfo.bi_basemem = bios_basemem / 1024; bootinfo.bi_extmem = bios_extmem / 1024; bootinfo.bi_memsizes_valid++; bootinfo.bi_bios_dev = dsk->drive; bootdev = MAKEBOOTDEV(dev_maj[dsk->type], dsk->slice, dsk->unit, dsk->part); /* Process configuration file */ autoboot = 1; #ifdef LOADER_GELI_SUPPORT geli_init(); #endif zfs_init(); /* * Probe the boot drive first - we will try to boot from whatever * pool we find on that drive. */ probe_drive(dsk); /* * Probe the rest of the drives that the bios knows about. This * will find any other available pools and it may fill in missing * vdevs for the boot pool. */ #ifndef VIRTUALBOX for (i = 0; i < *(unsigned char *)PTOV(BIOS_NUMDRIVES); i++) #else for (i = 0; i < MAXBDDEV; i++) #endif { if ((i | DRV_HARD) == *(uint8_t *)PTOV(ARGS)) continue; if (!int13probe(i | DRV_HARD)) break; dsk = malloc(sizeof(struct dsk)); dsk->drive = i | DRV_HARD; dsk->type = dsk->drive & TYPE_AD; dsk->unit = i; dsk->slice = 0; dsk->part = 0; dsk->start = 0; dsk->init = 0; probe_drive(dsk); } /* * The first discovered pool, if any, is the pool. */ spa = spa_get_primary(); if (!spa) { printf("%s: No ZFS pools located, can't boot\n", BOOTPROG); for (;;) ; } primary_spa = spa; primary_vdev = spa_get_primary_vdev(spa); nextboot = 0; rc = vdev_read_pad2(primary_vdev, cmd, sizeof(cmd)); if (vdev_clear_pad2(primary_vdev)) printf("failed to clear pad2 area of primary vdev\n"); if (rc == 0) { if (*cmd) { /* * We could find an old-style ZFS Boot Block header here. * Simply ignore it. */ if (*(uint64_t *)cmd != 0x2f5b007b10c) { /* * Note that parse() is destructive to cmd[] and we also want * to honor RBX_QUIET option that could be present in cmd[]. */ nextboot = 1; memcpy(cmddup, cmd, sizeof(cmd)); if (parse()) { printf("failed to parse pad2 area of primary vdev\n"); reboot(); } if (!OPT_CHECK(RBX_QUIET)) printf("zfs nextboot: %s\n", cmddup); } /* Do not process this command twice */ *cmd = 0; } } else printf("failed to read pad2 area of primary vdev\n"); /* Mount ZFS only if it's not already mounted via nextboot parsing. */ if (zfsmount.spa == NULL && (zfs_spa_init(spa) != 0 || zfs_mount(spa, 0, &zfsmount) != 0)) { printf("%s: failed to mount default pool %s\n", BOOTPROG, spa->spa_name); autoboot = 0; } else if (zfs_lookup(&zfsmount, PATH_CONFIG, &dn) == 0 || zfs_lookup(&zfsmount, PATH_DOTCONFIG, &dn) == 0) { off = 0; zfs_read(spa, &dn, &off, cmd, sizeof(cmd)); } if (*cmd) { /* * Note that parse() is destructive to cmd[] and we also want * to honor RBX_QUIET option that could be present in cmd[]. */ memcpy(cmddup, cmd, sizeof(cmd)); if (parse()) autoboot = 0; if (!OPT_CHECK(RBX_QUIET)) printf("%s: %s\n", PATH_CONFIG, cmddup); /* Do not process this command twice */ *cmd = 0; } /* Do not risk waiting at the prompt forever. */ if (nextboot && !autoboot) reboot(); /* * Try to exec /boot/loader. If interrupted by a keypress, * or in case of failure, try to load a kernel directly instead. */ if (autoboot && !*kname) { memcpy(kname, PATH_LOADER_ZFS, sizeof(PATH_LOADER_ZFS)); if (!keyhit(3)) { load(); memcpy(kname, PATH_KERNEL, sizeof(PATH_KERNEL)); } } /* Present the user with the boot2 prompt. */ for (;;) { if (!autoboot || !OPT_CHECK(RBX_QUIET)) { printf("\nFreeBSD/x86 boot\n"); if (zfs_rlookup(spa, zfsmount.rootobj, rootname) != 0) printf("Default: %s/<0x%llx>:%s\n" "boot: ", spa->spa_name, zfsmount.rootobj, kname); else if (rootname[0] != '\0') printf("Default: %s/%s:%s\n" "boot: ", spa->spa_name, rootname, kname); else printf("Default: %s:%s\n" "boot: ", spa->spa_name, kname); } if (ioctrl & IO_SERIAL) sio_flush(); if (!autoboot || keyhit(5)) getstr(cmd, sizeof(cmd)); else if (!autoboot || !OPT_CHECK(RBX_QUIET)) putchar('\n'); autoboot = 0; if (parse()) putchar('\a'); else load(); } } /* XXX - Needed for btxld to link the boot2 binary; do not remove. */ void exit(int x) { __exit(x); } void reboot(void) { __exit(0); } static void load(void) { union { struct exec ex; Elf32_Ehdr eh; } hdr; static Elf32_Phdr ep[2]; static Elf32_Shdr es[2]; caddr_t p; dnode_phys_t dn; off_t off; uint32_t addr, x; int fmt, i, j; if (zfs_lookup(&zfsmount, kname, &dn)) { printf("\nCan't find %s\n", kname); return; } off = 0; if (xfsread(&dn, &off, &hdr, sizeof(hdr))) return; if (N_GETMAGIC(hdr.ex) == ZMAGIC) fmt = 0; else if (IS_ELF(hdr.eh)) fmt = 1; else { printf("Invalid %s\n", "format"); return; } if (fmt == 0) { addr = hdr.ex.a_entry & 0xffffff; p = PTOV(addr); off = PAGE_SIZE; if (xfsread(&dn, &off, p, hdr.ex.a_text)) return; p += roundup2(hdr.ex.a_text, PAGE_SIZE); if (xfsread(&dn, &off, p, hdr.ex.a_data)) return; p += hdr.ex.a_data + roundup2(hdr.ex.a_bss, PAGE_SIZE); bootinfo.bi_symtab = VTOP(p); memcpy(p, &hdr.ex.a_syms, sizeof(hdr.ex.a_syms)); p += sizeof(hdr.ex.a_syms); if (hdr.ex.a_syms) { if (xfsread(&dn, &off, p, hdr.ex.a_syms)) return; p += hdr.ex.a_syms; if (xfsread(&dn, &off, p, sizeof(int))) return; x = *(uint32_t *)p; p += sizeof(int); x -= sizeof(int); if (xfsread(&dn, &off, p, x)) return; p += x; } } else { off = hdr.eh.e_phoff; for (j = i = 0; i < hdr.eh.e_phnum && j < 2; i++) { if (xfsread(&dn, &off, ep + j, sizeof(ep[0]))) return; if (ep[j].p_type == PT_LOAD) j++; } for (i = 0; i < 2; i++) { p = PTOV(ep[i].p_paddr & 0xffffff); off = ep[i].p_offset; if (xfsread(&dn, &off, p, ep[i].p_filesz)) return; } p += roundup2(ep[1].p_memsz, PAGE_SIZE); bootinfo.bi_symtab = VTOP(p); if (hdr.eh.e_shnum == hdr.eh.e_shstrndx + 3) { off = hdr.eh.e_shoff + sizeof(es[0]) * (hdr.eh.e_shstrndx + 1); if (xfsread(&dn, &off, &es, sizeof(es))) return; for (i = 0; i < 2; i++) { memcpy(p, &es[i].sh_size, sizeof(es[i].sh_size)); p += sizeof(es[i].sh_size); off = es[i].sh_offset; if (xfsread(&dn, &off, p, es[i].sh_size)) return; p += es[i].sh_size; } } addr = hdr.eh.e_entry & 0xffffff; } bootinfo.bi_esymtab = VTOP(p); bootinfo.bi_kernelname = VTOP(kname); zfsargs.size = sizeof(zfsargs); zfsargs.pool = zfsmount.spa->spa_guid; zfsargs.root = zfsmount.rootobj; zfsargs.primary_pool = primary_spa->spa_guid; #ifdef LOADER_GELI_SUPPORT bcopy(gelipw, zfsargs.gelipw, sizeof(zfsargs.gelipw)); bzero(gelipw, sizeof(gelipw)); #else zfsargs.gelipw[0] = '\0'; #endif if (primary_vdev != NULL) zfsargs.primary_vdev = primary_vdev->v_guid; else printf("failed to detect primary vdev\n"); __exec((caddr_t)addr, RB_BOOTINFO | (opts & RBX_MASK), bootdev, KARGS_FLAGS_ZFS | KARGS_FLAGS_EXTARG, (uint32_t) spa->spa_guid, (uint32_t) (spa->spa_guid >> 32), VTOP(&bootinfo), zfsargs); } static int zfs_mount_ds(char *dsname) { uint64_t newroot; spa_t *newspa; char *q; q = strchr(dsname, '/'); if (q) *q++ = '\0'; newspa = spa_find_by_name(dsname); if (newspa == NULL) { printf("\nCan't find ZFS pool %s\n", dsname); return -1; } if (zfs_spa_init(newspa)) return -1; newroot = 0; if (q) { if (zfs_lookup_dataset(newspa, q, &newroot)) { printf("\nCan't find dataset %s in ZFS pool %s\n", q, newspa->spa_name); return -1; } } if (zfs_mount(newspa, newroot, &zfsmount)) { printf("\nCan't mount ZFS dataset\n"); return -1; } spa = newspa; return (0); } static int parse(void) { char *arg = cmd; char *ep, *p, *q; const char *cp; int c, i, j; while ((c = *arg++)) { if (c == ' ' || c == '\t' || c == '\n') continue; for (p = arg; *p && *p != '\n' && *p != ' ' && *p != '\t'; p++); ep = p; if (*p) *p++ = 0; if (c == '-') { while ((c = *arg++)) { if (c == 'P') { if (*(uint8_t *)PTOV(0x496) & 0x10) { cp = "yes"; } else { opts |= OPT_SET(RBX_DUAL) | OPT_SET(RBX_SERIAL); cp = "no"; } printf("Keyboard: %s\n", cp); continue; } else if (c == 'S') { j = 0; while ((unsigned int)(i = *arg++ - '0') <= 9) j = j * 10 + i; if (j > 0 && i == -'0') { comspeed = j; break; } /* Fall through to error below ('S' not in optstr[]). */ } for (i = 0; c != optstr[i]; i++) if (i == NOPT - 1) return -1; opts ^= OPT_SET(flags[i]); } ioctrl = OPT_CHECK(RBX_DUAL) ? (IO_SERIAL|IO_KEYBOARD) : OPT_CHECK(RBX_SERIAL) ? IO_SERIAL : IO_KEYBOARD; if (ioctrl & IO_SERIAL) { if (sio_init(115200 / comspeed) != 0) ioctrl &= ~IO_SERIAL; } } if (c == '?') { dnode_phys_t dn; if (zfs_lookup(&zfsmount, arg, &dn) == 0) { zap_list(spa, &dn); } return -1; } else { arg--; /* * Report pool status if the comment is 'status'. Lets * hope no-one wants to load /status as a kernel. */ if (!strcmp(arg, "status")) { spa_all_status(); return -1; } /* * If there is "zfs:" prefix simply ignore it. */ if (strncmp(arg, "zfs:", 4) == 0) arg += 4; /* * If there is a colon, switch pools. */ q = strchr(arg, ':'); if (q) { *q++ = '\0'; if (zfs_mount_ds(arg) != 0) return -1; arg = q; } if ((i = ep - arg)) { if ((size_t)i >= sizeof(kname)) return -1; memcpy(kname, arg, i + 1); } } arg = p; } return 0; } Index: stable/11/sys/boot/i386/zfsboot/zfsldr.S =================================================================== --- stable/11/sys/boot/i386/zfsboot/zfsldr.S (revision 328865) +++ stable/11/sys/boot/i386/zfsboot/zfsldr.S (revision 328866) @@ -1,266 +1,283 @@ /* * Copyright (c) 1998 Robert Nordier * All rights reserved. * * Redistribution and use in source and binary forms are freely * permitted provided that the above copyright notice and this * paragraph and the following disclaimer are duplicated in all * such forms. * * This software is provided "AS IS" and without any express or * implied warranties, including, without limitation, the implied * warranties of merchantability and fitness for a particular * purpose. * * $FreeBSD$ */ /* Memory Locations */ .set MEM_ARG,0x900 # Arguments .set MEM_ORG,0x7c00 # Origin .set MEM_BUF,0x8000 # Load area .set MEM_BTX,0x9000 # BTX start .set MEM_JMP,0x9010 # BTX entry point .set MEM_USR,0xa000 # Client start .set BDA_BOOT,0x472 # Boot howto flag /* Partition Constants */ .set PRT_OFF,0x1be # Partition offset .set PRT_NUM,0x4 # Partitions .set PRT_BSD,0xa5 # Partition type /* Misc. Constants */ .set SIZ_PAG,0x1000 # Page size .set SIZ_SEC,0x200 # Sector size - - .set NSECT,0x80 + .set COPY_BLKS,0x8 # Number of blocks + # to copy for boot2 + .set COPY_BLK_SZ,0x8000 # Copy in 32k blocks; must be + # a multiple of 16 bytes + .set NSECT,(COPY_BLK_SZ / SIZ_SEC * COPY_BLKS) .globl start .code16 /* * Load the rest of zfsboot2 and BTX up, copy the parts to the right locations, * and start it all up. */ /* * Setup the segment registers to flat addressing (segment 0) and setup the * stack to end just below the start of our code. */ start: cld # String ops inc xor %cx,%cx # Zero mov %cx,%es # Address mov %cx,%ds # data mov %cx,%ss # Set up mov $start,%sp # stack /* * Load the MBR and look for the first FreeBSD slice. We use the fake * partition entry below that points to the MBR when we call read. * The first pass looks for the first active FreeBSD slice. The * second pass looks for the first non-active FreeBSD slice if the * first one fails. */ call check_edd # Make sure EDD works mov $part4,%si # Dummy partition xor %eax,%eax # Read MBR movl $MEM_BUF,%ebx # from first call read # sector mov $0x1,%cx # Two passes main.1: mov $MEM_BUF+PRT_OFF,%si # Partition table movb $0x1,%dh # Partition main.2: cmpb $PRT_BSD,0x4(%si) # Our partition type? jne main.3 # No jcxz main.5 # If second pass testb $0x80,(%si) # Active? jnz main.5 # Yes main.3: add $0x10,%si # Next entry incb %dh # Partition cmpb $0x1+PRT_NUM,%dh # In table? jb main.2 # Yes dec %cx # Do two jcxz main.1 # passes /* * If we get here, we didn't find any FreeBSD slices at all, so print an * error message and die. */ mov $msg_part,%si # Message jmp error # Error /* * Ok, we have a slice and drive in %dx now, so use that to locate and * load boot2. %si references the start of the slice we are looking - * for, so go ahead and load up the 128 sectors starting at sector 1024 - * (i.e. after the two vdev labels). We don't have do anything fancy - * here to allow for an extra copy of boot1 and a partition table - * (compare to this section of the UFS bootstrap) so we just load it - * all at 0x9000. The first part of boot2 is BTX, which wants to run - * at 0x9000. The boot2.bin binary starts right after the end of BTX, + * for, so go ahead and load up the COPY_BLKS*COPY_BLK_SZ/SIZ_SEC sectors + * starting at sector 1024 (i.e. after the two vdev labels). We don't + * have do anything fancy here to allow for an extra copy of boot1 and + * a partition table (compare to this section of the UFS bootstrap) so we + * just load it all at 0x9000. The first part of boot2 is BTX, which wants + * to run at 0x9000. The boot2.bin binary starts right after the end of BTX, * so we have to figure out where the start of it is and then move the * binary to 0xc000. Normally, BTX clients start at MEM_USR, or 0xa000, * but when we use btxld to create zfsboot2, we use an entry point of * 0x2000. That entry point is relative to MEM_USR; thus boot2.bin * starts at 0xc000. * * The load area and the target area for the client overlap so we have * to use a decrementing string move. We also play segment register * games with the destination address for the move so that the client * can be larger than 16k (which would overflow the zero segment since * the client starts at 0xc000). */ main.5: mov %dx,MEM_ARG # Save args mov $NSECT,%cx # Sector count movl $1024,%eax # Offset to boot2 mov $MEM_BTX,%ebx # Destination buffer main.6: pushal # Save params call read # Read disk popal # Restore incl %eax # Advance to add $SIZ_SEC,%ebx # next sector loop main.6 # If not last, read another - mov MEM_BTX+0xa,%bx # Get BTX length - mov $NSECT*SIZ_SEC-1,%di # Size of load area (less one) - mov %di,%si # End of load area, 0x9000 rel - sub %bx,%di # End of client, 0xc000 rel - mov %di,%cx # Size of - inc %cx # client - mov $(MEM_BTX)>>4,%dx # Segment - mov %dx,%ds # addressing 0x9000 - mov $(MEM_USR+2*SIZ_PAG)>>4,%dx # Segment - mov %dx,%es # addressing 0xc000 - std # Move with decrement - rep # Relocate - movsb # client + + mov $MEM_BTX,%bx # BTX + mov 0xa(%bx),%si # Get BTX length and set + add %bx,%si # %si to start of boot2 + dec %si # Set %ds:%si to point at the + mov %si,%ax # last byte we want to copy + shr $4,%ax # from boot2, with %si made as + add $(COPY_BLKS*COPY_BLK_SZ/16),%ax # small as possible. + and $0xf,%si # + mov %ax,%ds # + mov $(MEM_USR+2*SIZ_PAG)/16,%ax # Set %es:(-1) to point at + add $(COPY_BLKS*COPY_BLK_SZ/16),%ax # the last byte we + mov %ax,%es # want to copy boot2 into. + mov $COPY_BLKS,%bx # Copy COPY_BLKS 32k blocks +copyloop: + add $COPY_BLK_SZ,%si # Adjust %ds:%si to point at + mov %ds,%ax # the end of the next 32k to + sub $COPY_BLK_SZ/16,%ax # copy from boot2 + mov %ax,%ds + mov $COPY_BLK_SZ-1,%di # Adjust %es:%di to point at + mov %es,%ax # the end of the next 32k into + sub $COPY_BLK_SZ/16,%ax # which we want boot2 copied + mov %ax,%es + mov $COPY_BLK_SZ,%cx # Copy 32k + std + rep movsb + dec %bx + jnz copyloop + mov %cx,%ds # Reset %ds and %es + mov %cx,%es cld # Back to increment - xor %dx,%dx # Back - mov %ds,%dx # to zero - mov %dx,%es # segment /* * Enable A20 so we can access memory above 1 meg. * Use the zero-valued %cx as a timeout for embedded hardware which do not * have a keyboard controller. */ seta20: cli # Disable interrupts seta20.1: dec %cx # Timeout? jz seta20.3 # Yes inb $0x64,%al # Get status testb $0x2,%al # Busy? jnz seta20.1 # Yes movb $0xd1,%al # Command: Write outb %al,$0x64 # output port seta20.2: inb $0x64,%al # Get status testb $0x2,%al # Busy? jnz seta20.2 # Yes movb $0xdf,%al # Enable outb %al,$0x60 # A20 seta20.3: sti # Enable interrupts jmp start+MEM_JMP-MEM_ORG # Start BTX /* * Read a sector from the disk. Sets up an EDD packet on the stack * and passes it to read. We assume that the destination address is * always segment-aligned. * * %eax - int - LBA to read in relative to partition start * %ebx - ptr - destination address * %dl - byte - drive to read from * %si - ptr - MBR partition entry */ read: xor %ecx,%ecx # Get addl 0x8(%si),%eax # LBA adc $0,%ecx pushl %ecx # Starting absolute block pushl %eax # block number shr $4,%ebx # Convert to segment push %bx # Address of push $0 # transfer buffer push $0x1 # Read 1 sector push $0x10 # Size of packet mov %sp,%si # Packet pointer mov $0x42,%ah # BIOS: Extended int $0x13 # read jc read.1 # If error, fail lea 0x10(%si),%sp # Clear stack ret # If success, return read.1: mov %ah,%al # Format mov $read_err,%di # error call hex8 # code mov $msg_read,%si # Set the error message and # fall through to the error # routine /* * Print out the error message pointed to by %ds:(%si) followed * by a prompt, wait for a keypress, and then reboot the machine. */ error: callw putstr # Display message mov $prompt,%si # Display callw putstr # prompt xorb %ah,%ah # BIOS: Get int $0x16 # keypress movw $0x1234, BDA_BOOT # Do a warm boot ljmp $0xffff,$0x0 # reboot the machine /* * Display a null-terminated string using the BIOS output. */ putstr.0: mov $0x7,%bx # Page:attribute movb $0xe,%ah # BIOS: Display int $0x10 # character putstr: lodsb # Get char testb %al,%al # End of string? jne putstr.0 # No ret # To caller /* * Check to see if the disk supports EDD. zfsboot requires EDD and does not * support older C/H/S disk I/O. */ check_edd: cmpb $0x80,%dl # Hard drive? jb check_edd.1 # No, fail to boot mov $0x55aa,%bx # Magic push %dx # Save movb $0x41,%ah # BIOS: Check int $0x13 # extensions present pop %dx # Restore jc check_edd.1 # If error, fail cmp $0xaa55,%bx # Magic? jne check_edd.1 # No, so fail testb $0x1,%cl # Packet interface? jz check_edd.1 # No, so fail ret # EDD ok, keep booting check_edd.1: mov $msg_chs,%si # Warn that CHS is jmp error # unsupported and fail /* * AL to hex, saving the result to [EDI]. */ hex8: push %ax # Save shrb $0x4,%al # Do upper call hex8.1 # 4 pop %ax # Restore hex8.1: andb $0xf,%al # Get lower 4 cmpb $0xa,%al # Convert sbbb $0x69,%al # to hex das # digit orb $0x20,%al # To lower case stosb # Save char ret # (Recursive) /* Messages */ msg_chs: .asciz "CHS not supported" msg_read: .ascii "Read error: " read_err: .asciz "XX" msg_part: .asciz "Boot error" prompt: .asciz "\r\n" .org PRT_OFF,0x90 /* Partition table */ .fill 0x30,0x1,0x0 part4: .byte 0x80, 0x00, 0x01, 0x00 .byte 0xa5, 0xfe, 0xff, 0xff .byte 0x00, 0x00, 0x00, 0x00 .byte 0x50, 0xc3, 0x00, 0x00 # 50000 sectors long, bleh .word 0xaa55 # Magic number Index: stable/11/sys/boot/userboot/ficl/Makefile =================================================================== --- stable/11/sys/boot/userboot/ficl/Makefile (revision 328865) +++ stable/11/sys/boot/userboot/ficl/Makefile (revision 328866) @@ -1,60 +1,63 @@ # $FreeBSD$ # .include MK_SSP= no .PATH: ${.CURDIR}/../../ficl .PATH: ${.CURDIR}/../../ficl/${MACHINE_CPUARCH} BASE_SRCS= dict.c ficl.c fileaccess.c float.c loader.c math64.c \ prefix.c search.c stack.c tools.c vm.c words.c SRCS= ${BASE_SRCS} sysdep.c softcore.c CLEANFILES= softcore.c testmain testmain.o + +CWARNFLAGS.loader.c.c += -Wno-implicit-function-declaration + .if HAVE_PNP CFLAGS+= -DHAVE_PNP .endif .include .ifmake testmain CFLAGS+= -DTESTMAIN -D_TESTMAIN SRCS+= testmain.c PROG= testmain .include .else LIB= ficl INTERNALLIB= .include .endif # Standard softwords .PATH: ${.CURDIR}/../../ficl/softwords SOFTWORDS= softcore.fr jhlocal.fr marker.fr freebsd.fr ficllocal.fr \ ifbrack.fr # Optional OO extension softwords #SOFTWORDS+= oo.fr classes.fr #.if ${MACHINE_CPUARCH} == "amd64" #CFLAGS+= -m32 -I. #.endif .if ${MACHINE_ARCH} == "powerpc64" CFLAGS+= -m32 -mcpu=powerpc -I. .endif CFLAGS+= -I${.CURDIR}/../../ficl CFLAGS+= -I${.CURDIR}/../../ficl/${MACHINE_CPUARCH} CFLAGS+= -I${.CURDIR}/../../common softcore.c: ${SOFTWORDS} softcore.awk (cd ${.CURDIR}/../../ficl/softwords; cat ${SOFTWORDS} \ | awk -f softcore.awk -v datestamp="`LC_ALL=C date`") > ${.TARGET} #.if ${MACHINE_CPUARCH} == "amd64" #${SRCS:M*.c:R:S/$/.o/g}: machine # #beforedepend ${OBJS}: machine # #machine: .NOMETA # ln -sf ${.CURDIR}/../../i386/include machine # #CLEANFILES+= machine #.endif Index: stable/11/sys/boot/userboot/userboot/Makefile =================================================================== --- stable/11/sys/boot/userboot/userboot/Makefile (revision 328865) +++ stable/11/sys/boot/userboot/userboot/Makefile (revision 328866) @@ -1,65 +1,67 @@ # $FreeBSD$ MAN= .include MK_SSP= no SHLIB_NAME= userboot.so MK_CTF= no STRIP= LIBDIR= /boot SRCS= autoload.c SRCS+= bcache.c SRCS+= biossmap.c SRCS+= bootinfo.c SRCS+= bootinfo32.c SRCS+= bootinfo64.c SRCS+= conf.c SRCS+= console.c SRCS+= copy.c SRCS+= devicename.c SRCS+= elf32_freebsd.c SRCS+= elf64_freebsd.c SRCS+= host.c SRCS+= main.c SRCS+= userboot_cons.c SRCS+= userboot_disk.c SRCS+= vers.c CFLAGS+= -Wall CFLAGS+= -I${.CURDIR}/.. CFLAGS+= -I${.CURDIR}/../../common CFLAGS+= -I${.CURDIR}/../../.. CFLAGS+= -I${.CURDIR}/../../../../lib/libstand CFLAGS+= -ffreestanding -I. +CWARNFLAGS.main.c += -Wno-implicit-function-declaration + LDFLAGS+= -nostdlib -Wl,-Bsymbolic NEWVERSWHAT= "User boot" ${MACHINE_CPUARCH} .if ${MK_FORTH} != "no" BOOT_FORTH= yes CFLAGS+= -DBOOT_FORTH -I${.CURDIR}/../../ficl CFLAGS+= -I${.CURDIR}/../../ficl/i386 CFLAGS+= -DBF_DICTSIZE=15000 LIBFICL= ${.OBJDIR}/../ficl/libficl.a .endif LIBSTAND= ${.OBJDIR}/../libstand/libstand.a .if ${MK_ZFS} != "no" CFLAGS+= -DUSERBOOT_ZFS_SUPPORT LIBZFSBOOT= ${.OBJDIR}/../zfs/libzfsboot.a .endif # Always add MI sources .PATH: ${.CURDIR}/../../common .include "${.CURDIR}/../../common/Makefile.inc" CFLAGS+= -I${.CURDIR}/../../common CFLAGS+= -I. DPADD+= ${LIBFICL} ${LIBZFSBOOT} ${LIBSTAND} LDADD+= ${LIBFICL} ${LIBZFSBOOT} ${LIBSTAND} .include Index: stable/11/sys/boot/userboot/userboot/main.c =================================================================== --- stable/11/sys/boot/userboot/userboot/main.c (revision 328865) +++ stable/11/sys/boot/userboot/userboot/main.c (revision 328866) @@ -1,293 +1,293 @@ /*- * Copyright (c) 1998 Michael Smith * Copyright (c) 1998,2000 Doug Rabson * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include "bootstrap.h" #include "disk.h" #include "libuserboot.h" #if defined(USERBOOT_ZFS_SUPPORT) #include "../zfs/libzfs.h" static void userboot_zfs_probe(void); static int userboot_zfs_found; #endif /* Minimum version required */ #define USERBOOT_VERSION USERBOOT_VERSION_3 -#define MALLOCSZ (10*1024*1024) +#define MALLOCSZ (64*1024*1024) struct loader_callbacks *callbacks; void *callbacks_arg; extern char bootprog_info[]; static jmp_buf jb; struct arch_switch archsw; /* MI/MD interface boundary */ static void extract_currdev(void); void delay(int usec) { CALLBACK(delay, usec); } void exit(int v) { CALLBACK(exit, v); longjmp(jb, 1); } void loader_main(struct loader_callbacks *cb, void *arg, int version, int ndisks) { static char mallocbuf[MALLOCSZ]; const char *var; int i; if (version < USERBOOT_VERSION) abort(); callbacks = cb; callbacks_arg = arg; userboot_disk_maxunit = ndisks; /* * initialise the heap as early as possible. Once this is done, * alloc() is usable. */ setheap((void *)mallocbuf, (void *)(mallocbuf + sizeof(mallocbuf))); /* * Hook up the console */ cons_probe(); printf("\n%s", bootprog_info); #if 0 printf("Memory: %ld k\n", memsize() / 1024); #endif setenv("LINES", "24", 1); /* optional */ /* * Set custom environment variables */ i = 0; while (1) { var = CALLBACK(getenv, i++); if (var == NULL) break; putenv(var); } archsw.arch_autoload = userboot_autoload; archsw.arch_getdev = userboot_getdev; archsw.arch_copyin = userboot_copyin; archsw.arch_copyout = userboot_copyout; archsw.arch_readin = userboot_readin; #if defined(USERBOOT_ZFS_SUPPORT) archsw.arch_zfs_probe = userboot_zfs_probe; #endif /* * Initialise the block cache. Set the upper limit. */ bcache_init(32768, 512); /* * March through the device switch probing for things. */ for (i = 0; devsw[i] != NULL; i++) if (devsw[i]->dv_init != NULL) (devsw[i]->dv_init)(); extract_currdev(); if (setjmp(jb)) return; interact(NULL); /* doesn't return */ exit(0); } /* * Set the 'current device' by (if possible) recovering the boot device as * supplied by the initial bootstrap. */ static void extract_currdev(void) { struct disk_devdesc dev; //bzero(&dev, sizeof(dev)); #if defined(USERBOOT_ZFS_SUPPORT) if (userboot_zfs_found) { struct zfs_devdesc zdev; /* Leave the pool/root guid's unassigned */ bzero(&zdev, sizeof(zdev)); zdev.d_dev = &zfs_dev; zdev.d_type = zdev.d_dev->dv_type; dev = *(struct disk_devdesc *)&zdev; init_zfs_bootenv(zfs_fmtdev(&dev)); } else #endif if (userboot_disk_maxunit > 0) { dev.d_dev = &userboot_disk; dev.d_type = dev.d_dev->dv_type; dev.d_unit = 0; dev.d_slice = 0; dev.d_partition = 0; /* * If we cannot auto-detect the partition type then * access the disk as a raw device. */ if (dev.d_dev->dv_open(NULL, &dev)) { dev.d_slice = -1; dev.d_partition = -1; } } else { dev.d_dev = &host_dev; dev.d_type = dev.d_dev->dv_type; dev.d_unit = 0; } env_setenv("currdev", EV_VOLATILE, userboot_fmtdev(&dev), userboot_setcurrdev, env_nounset); env_setenv("loaddev", EV_VOLATILE, userboot_fmtdev(&dev), env_noset, env_nounset); } #if defined(USERBOOT_ZFS_SUPPORT) static void userboot_zfs_probe(void) { char devname[32]; uint64_t pool_guid; int unit; /* * Open all the disks we can find and see if we can reconstruct * ZFS pools from them. Record if any were found. */ for (unit = 0; unit < userboot_disk_maxunit; unit++) { sprintf(devname, "disk%d:", unit); pool_guid = 0; zfs_probe_dev(devname, &pool_guid); if (pool_guid != 0) userboot_zfs_found = 1; } } COMMAND_SET(lszfs, "lszfs", "list child datasets of a zfs dataset", command_lszfs); static int command_lszfs(int argc, char *argv[]) { int err; if (argc != 2) { command_errmsg = "a single dataset must be supplied"; return (CMD_ERROR); } err = zfs_list(argv[1]); if (err != 0) { command_errmsg = strerror(err); return (CMD_ERROR); } return (CMD_OK); } COMMAND_SET(reloadbe, "reloadbe", "refresh the list of ZFS Boot Environments", command_reloadbe); static int command_reloadbe(int argc, char *argv[]) { int err; char *root; if (argc > 2) { command_errmsg = "wrong number of arguments"; return (CMD_ERROR); } if (argc == 2) { err = zfs_bootenv(argv[1]); } else { root = getenv("zfs_be_root"); if (root == NULL) { return (CMD_OK); } err = zfs_bootenv(root); } if (err != 0) { command_errmsg = strerror(err); return (CMD_ERROR); } return (CMD_OK); } #endif /* USERBOOT_ZFS_SUPPORT */ COMMAND_SET(quit, "quit", "exit the loader", command_quit); static int command_quit(int argc, char *argv[]) { exit(USERBOOT_EXIT_QUIT); return (CMD_OK); } COMMAND_SET(reboot, "reboot", "reboot the system", command_reboot); static int command_reboot(int argc, char *argv[]) { exit(USERBOOT_EXIT_REBOOT); return (CMD_OK); } Index: stable/11/sys/boot/userboot/zfs/Makefile =================================================================== --- stable/11/sys/boot/userboot/zfs/Makefile (revision 328865) +++ stable/11/sys/boot/userboot/zfs/Makefile (revision 328866) @@ -1,18 +1,22 @@ # $FreeBSD$ S= ${.CURDIR}/../../zfs -.PATH: ${S} +.PATH: ${S} ${.CURDIR}/../../../crypto/skein LIB= zfsboot INTERNALLIB= -SRCS+= zfs.c +SRCS+= zfs.c skein.c skein_block.c +# Do not unroll skein loops, reduce code size +CFLAGS+= -DSKEIN_LOOP=111 + CFLAGS+= -I${.CURDIR}/../../common -I${.CURDIR}/../../.. -I. CFLAGS+= -I${.CURDIR}/../../../../lib/libstand CFLAGS+= -I${.CURDIR}/../../../cddl/boot/zfs +CFLAGS+= -I${.CURDIR}/../../../crypto/skein CFLAGS+= -ffreestanding -fPIC CFLAGS+= -Wformat -Wall .include Index: stable/11/sys/boot/zfs/Makefile =================================================================== --- stable/11/sys/boot/zfs/Makefile (revision 328865) +++ stable/11/sys/boot/zfs/Makefile (revision 328866) @@ -1,35 +1,41 @@ # $FreeBSD$ LIB= zfsboot INTERNALLIB= SRCS+= zfs.c +SRCS+= skein.c skein_block.c +# Do not unroll skein loops, reduce code size +CFLAGS+= -DSKEIN_LOOP=111 +.PATH: ${.CURDIR}/../../crypto/skein + CFLAGS+= -DBOOTPROG=\"zfsloader\" CFLAGS+= -I${.CURDIR}/../common -I${.CURDIR}/../.. -I. CFLAGS+= -I${.CURDIR}/../../../lib/libstand CFLAGS+= -I${.CURDIR}/../../cddl/boot/zfs +CFLAGS+= -I${.CURDIR}/../../crypto/skein .if ${MACHINE_CPUARCH} == "i386" || ${MACHINE_CPUARCH} == "amd64" CFLAGS+= -march=i386 .endif .if ${MACHINE_CPUARCH} == "amd64" CFLAGS+= -m32 .endif CFLAGS+= -Wformat -Wall .if ${MACHINE_CPUARCH} == "amd64" CLEANFILES+= machine machine: .NOMETA ln -sf ${.CURDIR}/../../i386/include machine .endif .include .include .if ${MACHINE_CPUARCH} == "amd64" .if !exists(machine) beforedepend ${OBJS}: machine .endif .endif Index: stable/11/sys/boot/zfs/zfsimpl.c =================================================================== --- stable/11/sys/boot/zfs/zfsimpl.c (revision 328865) +++ stable/11/sys/boot/zfs/zfsimpl.c (revision 328866) @@ -1,2371 +1,2458 @@ /*- * Copyright (c) 2007 Doug Rabson * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); /* * Stand-alone ZFS file reader. */ #include #include #include "zfsimpl.h" #include "zfssubr.c" struct zfsmount { const spa_t *spa; objset_phys_t objset; uint64_t rootobj; }; /* * List of all vdevs, chained through v_alllink. */ static vdev_list_t zfs_vdevs; /* * List of ZFS features supported for read */ static const char *features_for_read[] = { "org.illumos:lz4_compress", "com.delphix:hole_birth", "com.delphix:extensible_dataset", "com.delphix:embedded_data", "org.open-zfs:large_blocks", + "org.illumos:sha512", + "org.illumos:skein", NULL }; /* * List of all pools, chained through spa_link. */ static spa_list_t zfs_pools; static const dnode_phys_t *dnode_cache_obj = NULL; static uint64_t dnode_cache_bn; static char *dnode_cache_buf; static char *zap_scratch; static char *zfs_temp_buf, *zfs_temp_end, *zfs_temp_ptr; #define TEMP_SIZE (1024 * 1024) static int zio_read(const spa_t *spa, const blkptr_t *bp, void *buf); static int zfs_get_root(const spa_t *spa, uint64_t *objid); static int zfs_rlookup(const spa_t *spa, uint64_t objnum, char *result); +static int zap_lookup(const spa_t *spa, const dnode_phys_t *dnode, + const char *name, uint64_t integer_size, uint64_t num_integers, + void *value); static void zfs_init(void) { STAILQ_INIT(&zfs_vdevs); STAILQ_INIT(&zfs_pools); zfs_temp_buf = malloc(TEMP_SIZE); zfs_temp_end = zfs_temp_buf + TEMP_SIZE; zfs_temp_ptr = zfs_temp_buf; dnode_cache_buf = malloc(SPA_MAXBLOCKSIZE); zap_scratch = malloc(SPA_MAXBLOCKSIZE); zfs_init_crc(); } static void * zfs_alloc(size_t size) { char *ptr; if (zfs_temp_ptr + size > zfs_temp_end) { printf("ZFS: out of temporary buffer space\n"); for (;;) ; } ptr = zfs_temp_ptr; zfs_temp_ptr += size; return (ptr); } static void zfs_free(void *ptr, size_t size) { zfs_temp_ptr -= size; if (zfs_temp_ptr != ptr) { printf("ZFS: zfs_alloc()/zfs_free() mismatch\n"); for (;;) ; } } static int xdr_int(const unsigned char **xdr, int *ip) { *ip = ((*xdr)[0] << 24) | ((*xdr)[1] << 16) | ((*xdr)[2] << 8) | ((*xdr)[3] << 0); (*xdr) += 4; return (0); } static int xdr_u_int(const unsigned char **xdr, u_int *ip) { *ip = ((*xdr)[0] << 24) | ((*xdr)[1] << 16) | ((*xdr)[2] << 8) | ((*xdr)[3] << 0); (*xdr) += 4; return (0); } static int xdr_uint64_t(const unsigned char **xdr, uint64_t *lp) { u_int hi, lo; xdr_u_int(xdr, &hi); xdr_u_int(xdr, &lo); *lp = (((uint64_t) hi) << 32) | lo; return (0); } static int nvlist_find(const unsigned char *nvlist, const char *name, int type, int* elementsp, void *valuep) { const unsigned char *p, *pair; int junk; int encoded_size, decoded_size; p = nvlist; xdr_int(&p, &junk); xdr_int(&p, &junk); pair = p; xdr_int(&p, &encoded_size); xdr_int(&p, &decoded_size); while (encoded_size && decoded_size) { int namelen, pairtype, elements; const char *pairname; xdr_int(&p, &namelen); pairname = (const char*) p; p += roundup(namelen, 4); xdr_int(&p, &pairtype); if (!memcmp(name, pairname, namelen) && type == pairtype) { xdr_int(&p, &elements); if (elementsp) *elementsp = elements; if (type == DATA_TYPE_UINT64) { xdr_uint64_t(&p, (uint64_t *) valuep); return (0); } else if (type == DATA_TYPE_STRING) { int len; xdr_int(&p, &len); (*(const char**) valuep) = (const char*) p; return (0); } else if (type == DATA_TYPE_NVLIST || type == DATA_TYPE_NVLIST_ARRAY) { (*(const unsigned char**) valuep) = (const unsigned char*) p; return (0); } else { return (EIO); } } else { /* * Not the pair we are looking for, skip to the next one. */ p = pair + encoded_size; } pair = p; xdr_int(&p, &encoded_size); xdr_int(&p, &decoded_size); } return (EIO); } static int nvlist_check_features_for_read(const unsigned char *nvlist) { const unsigned char *p, *pair; int junk; int encoded_size, decoded_size; int rc; rc = 0; p = nvlist; xdr_int(&p, &junk); xdr_int(&p, &junk); pair = p; xdr_int(&p, &encoded_size); xdr_int(&p, &decoded_size); while (encoded_size && decoded_size) { int namelen, pairtype; const char *pairname; int i, found; found = 0; xdr_int(&p, &namelen); pairname = (const char*) p; p += roundup(namelen, 4); xdr_int(&p, &pairtype); for (i = 0; features_for_read[i] != NULL; i++) { if (!memcmp(pairname, features_for_read[i], namelen)) { found = 1; break; } } if (!found) { printf("ZFS: unsupported feature: %s\n", pairname); rc = EIO; } p = pair + encoded_size; pair = p; xdr_int(&p, &encoded_size); xdr_int(&p, &decoded_size); } return (rc); } /* * Return the next nvlist in an nvlist array. */ static const unsigned char * nvlist_next(const unsigned char *nvlist) { const unsigned char *p, *pair; int junk; int encoded_size, decoded_size; p = nvlist; xdr_int(&p, &junk); xdr_int(&p, &junk); pair = p; xdr_int(&p, &encoded_size); xdr_int(&p, &decoded_size); while (encoded_size && decoded_size) { p = pair + encoded_size; pair = p; xdr_int(&p, &encoded_size); xdr_int(&p, &decoded_size); } return p; } #ifdef TEST static const unsigned char * nvlist_print(const unsigned char *nvlist, unsigned int indent) { static const char* typenames[] = { "DATA_TYPE_UNKNOWN", "DATA_TYPE_BOOLEAN", "DATA_TYPE_BYTE", "DATA_TYPE_INT16", "DATA_TYPE_UINT16", "DATA_TYPE_INT32", "DATA_TYPE_UINT32", "DATA_TYPE_INT64", "DATA_TYPE_UINT64", "DATA_TYPE_STRING", "DATA_TYPE_BYTE_ARRAY", "DATA_TYPE_INT16_ARRAY", "DATA_TYPE_UINT16_ARRAY", "DATA_TYPE_INT32_ARRAY", "DATA_TYPE_UINT32_ARRAY", "DATA_TYPE_INT64_ARRAY", "DATA_TYPE_UINT64_ARRAY", "DATA_TYPE_STRING_ARRAY", "DATA_TYPE_HRTIME", "DATA_TYPE_NVLIST", "DATA_TYPE_NVLIST_ARRAY", "DATA_TYPE_BOOLEAN_VALUE", "DATA_TYPE_INT8", "DATA_TYPE_UINT8", "DATA_TYPE_BOOLEAN_ARRAY", "DATA_TYPE_INT8_ARRAY", "DATA_TYPE_UINT8_ARRAY" }; unsigned int i, j; const unsigned char *p, *pair; int junk; int encoded_size, decoded_size; p = nvlist; xdr_int(&p, &junk); xdr_int(&p, &junk); pair = p; xdr_int(&p, &encoded_size); xdr_int(&p, &decoded_size); while (encoded_size && decoded_size) { int namelen, pairtype, elements; const char *pairname; xdr_int(&p, &namelen); pairname = (const char*) p; p += roundup(namelen, 4); xdr_int(&p, &pairtype); for (i = 0; i < indent; i++) printf(" "); printf("%s %s", typenames[pairtype], pairname); xdr_int(&p, &elements); switch (pairtype) { case DATA_TYPE_UINT64: { uint64_t val; xdr_uint64_t(&p, &val); printf(" = 0x%jx\n", (uintmax_t)val); break; } case DATA_TYPE_STRING: { int len; xdr_int(&p, &len); printf(" = \"%s\"\n", p); break; } case DATA_TYPE_NVLIST: printf("\n"); nvlist_print(p, indent + 1); break; case DATA_TYPE_NVLIST_ARRAY: for (j = 0; j < elements; j++) { printf("[%d]\n", j); p = nvlist_print(p, indent + 1); if (j != elements - 1) { for (i = 0; i < indent; i++) printf(" "); printf("%s %s", typenames[pairtype], pairname); } } break; default: printf("\n"); } p = pair + encoded_size; pair = p; xdr_int(&p, &encoded_size); xdr_int(&p, &decoded_size); } return p; } #endif static int vdev_read_phys(vdev_t *vdev, const blkptr_t *bp, void *buf, off_t offset, size_t size) { size_t psize; int rc; if (!vdev->v_phys_read) return (EIO); if (bp) { psize = BP_GET_PSIZE(bp); } else { psize = size; } /*printf("ZFS: reading %d bytes at 0x%jx to %p\n", psize, (uintmax_t)offset, buf);*/ rc = vdev->v_phys_read(vdev, vdev->v_read_priv, offset, buf, psize); if (rc) return (rc); - if (bp && zio_checksum_verify(bp, buf)) + if (bp && zio_checksum_verify(vdev->spa, bp, buf)) return (EIO); return (0); } static int vdev_disk_read(vdev_t *vdev, const blkptr_t *bp, void *buf, off_t offset, size_t bytes) { return (vdev_read_phys(vdev, bp, buf, offset + VDEV_LABEL_START_SIZE, bytes)); } static int vdev_mirror_read(vdev_t *vdev, const blkptr_t *bp, void *buf, off_t offset, size_t bytes) { vdev_t *kid; int rc; rc = EIO; STAILQ_FOREACH(kid, &vdev->v_children, v_childlink) { if (kid->v_state != VDEV_STATE_HEALTHY) continue; rc = kid->v_read(kid, bp, buf, offset, bytes); if (!rc) return (0); } return (rc); } static int vdev_replacing_read(vdev_t *vdev, const blkptr_t *bp, void *buf, off_t offset, size_t bytes) { vdev_t *kid; /* * Here we should have two kids: * First one which is the one we are replacing and we can trust * only this one to have valid data, but it might not be present. * Second one is that one we are replacing with. It is most likely * healthy, but we can't trust it has needed data, so we won't use it. */ kid = STAILQ_FIRST(&vdev->v_children); if (kid == NULL) return (EIO); if (kid->v_state != VDEV_STATE_HEALTHY) return (EIO); return (kid->v_read(kid, bp, buf, offset, bytes)); } static vdev_t * vdev_find(uint64_t guid) { vdev_t *vdev; STAILQ_FOREACH(vdev, &zfs_vdevs, v_alllink) if (vdev->v_guid == guid) return (vdev); return (0); } static vdev_t * vdev_create(uint64_t guid, vdev_read_t *_read) { vdev_t *vdev; vdev = malloc(sizeof(vdev_t)); memset(vdev, 0, sizeof(vdev_t)); STAILQ_INIT(&vdev->v_children); vdev->v_guid = guid; vdev->v_state = VDEV_STATE_OFFLINE; vdev->v_read = _read; vdev->v_phys_read = 0; vdev->v_read_priv = 0; STAILQ_INSERT_TAIL(&zfs_vdevs, vdev, v_alllink); return (vdev); } static int vdev_init_from_nvlist(const unsigned char *nvlist, vdev_t *pvdev, vdev_t **vdevp, int is_newer) { int rc; uint64_t guid, id, ashift, nparity; const char *type; const char *path; vdev_t *vdev, *kid; const unsigned char *kids; int nkids, i, is_new; uint64_t is_offline, is_faulted, is_degraded, is_removed, isnt_present; if (nvlist_find(nvlist, ZPOOL_CONFIG_GUID, DATA_TYPE_UINT64, 0, &guid) || nvlist_find(nvlist, ZPOOL_CONFIG_ID, DATA_TYPE_UINT64, 0, &id) || nvlist_find(nvlist, ZPOOL_CONFIG_TYPE, DATA_TYPE_STRING, 0, &type)) { printf("ZFS: can't find vdev details\n"); return (ENOENT); } if (strcmp(type, VDEV_TYPE_MIRROR) && strcmp(type, VDEV_TYPE_DISK) #ifdef ZFS_TEST && strcmp(type, VDEV_TYPE_FILE) #endif && strcmp(type, VDEV_TYPE_RAIDZ) && strcmp(type, VDEV_TYPE_REPLACING)) { printf("ZFS: can only boot from disk, mirror, raidz1, raidz2 and raidz3 vdevs\n"); return (EIO); } is_offline = is_removed = is_faulted = is_degraded = isnt_present = 0; nvlist_find(nvlist, ZPOOL_CONFIG_OFFLINE, DATA_TYPE_UINT64, 0, &is_offline); nvlist_find(nvlist, ZPOOL_CONFIG_REMOVED, DATA_TYPE_UINT64, 0, &is_removed); nvlist_find(nvlist, ZPOOL_CONFIG_FAULTED, DATA_TYPE_UINT64, 0, &is_faulted); nvlist_find(nvlist, ZPOOL_CONFIG_DEGRADED, DATA_TYPE_UINT64, 0, &is_degraded); nvlist_find(nvlist, ZPOOL_CONFIG_NOT_PRESENT, DATA_TYPE_UINT64, 0, &isnt_present); vdev = vdev_find(guid); if (!vdev) { is_new = 1; if (!strcmp(type, VDEV_TYPE_MIRROR)) vdev = vdev_create(guid, vdev_mirror_read); else if (!strcmp(type, VDEV_TYPE_RAIDZ)) vdev = vdev_create(guid, vdev_raidz_read); else if (!strcmp(type, VDEV_TYPE_REPLACING)) vdev = vdev_create(guid, vdev_replacing_read); else vdev = vdev_create(guid, vdev_disk_read); vdev->v_id = id; vdev->v_top = pvdev != NULL ? pvdev : vdev; if (nvlist_find(nvlist, ZPOOL_CONFIG_ASHIFT, DATA_TYPE_UINT64, 0, &ashift) == 0) vdev->v_ashift = ashift; else vdev->v_ashift = 0; if (nvlist_find(nvlist, ZPOOL_CONFIG_NPARITY, DATA_TYPE_UINT64, 0, &nparity) == 0) vdev->v_nparity = nparity; else vdev->v_nparity = 0; if (nvlist_find(nvlist, ZPOOL_CONFIG_PATH, DATA_TYPE_STRING, 0, &path) == 0) { if (strncmp(path, "/dev/", 5) == 0) path += 5; vdev->v_name = strdup(path); } else { if (!strcmp(type, "raidz")) { if (vdev->v_nparity == 1) vdev->v_name = "raidz1"; else if (vdev->v_nparity == 2) vdev->v_name = "raidz2"; else if (vdev->v_nparity == 3) vdev->v_name = "raidz3"; else { printf("ZFS: can only boot from disk, mirror, raidz1, raidz2 and raidz3 vdevs\n"); return (EIO); } } else { vdev->v_name = strdup(type); } } } else { is_new = 0; } if (is_new || is_newer) { /* * This is either new vdev or we've already seen this vdev, * but from an older vdev label, so let's refresh its state * from the newer label. */ if (is_offline) vdev->v_state = VDEV_STATE_OFFLINE; else if (is_removed) vdev->v_state = VDEV_STATE_REMOVED; else if (is_faulted) vdev->v_state = VDEV_STATE_FAULTED; else if (is_degraded) vdev->v_state = VDEV_STATE_DEGRADED; else if (isnt_present) vdev->v_state = VDEV_STATE_CANT_OPEN; } rc = nvlist_find(nvlist, ZPOOL_CONFIG_CHILDREN, DATA_TYPE_NVLIST_ARRAY, &nkids, &kids); /* * Its ok if we don't have any kids. */ if (rc == 0) { vdev->v_nchildren = nkids; for (i = 0; i < nkids; i++) { rc = vdev_init_from_nvlist(kids, vdev, &kid, is_newer); if (rc) return (rc); if (is_new) STAILQ_INSERT_TAIL(&vdev->v_children, kid, v_childlink); kids = nvlist_next(kids); } } else { vdev->v_nchildren = 0; } if (vdevp) *vdevp = vdev; return (0); } static void vdev_set_state(vdev_t *vdev) { vdev_t *kid; int good_kids; int bad_kids; /* * A mirror or raidz is healthy if all its kids are healthy. A * mirror is degraded if any of its kids is healthy; a raidz * is degraded if at most nparity kids are offline. */ if (STAILQ_FIRST(&vdev->v_children)) { good_kids = 0; bad_kids = 0; STAILQ_FOREACH(kid, &vdev->v_children, v_childlink) { if (kid->v_state == VDEV_STATE_HEALTHY) good_kids++; else bad_kids++; } if (bad_kids == 0) { vdev->v_state = VDEV_STATE_HEALTHY; } else { if (vdev->v_read == vdev_mirror_read) { if (good_kids) { vdev->v_state = VDEV_STATE_DEGRADED; } else { vdev->v_state = VDEV_STATE_OFFLINE; } } else if (vdev->v_read == vdev_raidz_read) { if (bad_kids > vdev->v_nparity) { vdev->v_state = VDEV_STATE_OFFLINE; } else { vdev->v_state = VDEV_STATE_DEGRADED; } } } } } static spa_t * spa_find_by_guid(uint64_t guid) { spa_t *spa; STAILQ_FOREACH(spa, &zfs_pools, spa_link) if (spa->spa_guid == guid) return (spa); return (0); } static spa_t * spa_find_by_name(const char *name) { spa_t *spa; STAILQ_FOREACH(spa, &zfs_pools, spa_link) if (!strcmp(spa->spa_name, name)) return (spa); return (0); } #ifdef BOOT2 static spa_t * spa_get_primary(void) { return (STAILQ_FIRST(&zfs_pools)); } static vdev_t * spa_get_primary_vdev(const spa_t *spa) { vdev_t *vdev; vdev_t *kid; if (spa == NULL) spa = spa_get_primary(); if (spa == NULL) return (NULL); vdev = STAILQ_FIRST(&spa->spa_vdevs); if (vdev == NULL) return (NULL); for (kid = STAILQ_FIRST(&vdev->v_children); kid != NULL; kid = STAILQ_FIRST(&vdev->v_children)) vdev = kid; return (vdev); } #endif static spa_t * spa_create(uint64_t guid) { spa_t *spa; spa = malloc(sizeof(spa_t)); memset(spa, 0, sizeof(spa_t)); STAILQ_INIT(&spa->spa_vdevs); spa->spa_guid = guid; STAILQ_INSERT_TAIL(&zfs_pools, spa, spa_link); return (spa); } static const char * state_name(vdev_state_t state) { static const char* names[] = { "UNKNOWN", "CLOSED", "OFFLINE", "REMOVED", "CANT_OPEN", "FAULTED", "DEGRADED", "ONLINE" }; return names[state]; } #ifdef BOOT2 #define pager_printf printf #else static void pager_printf(const char *fmt, ...) { char line[80]; va_list args; va_start(args, fmt); vsprintf(line, fmt, args); va_end(args); pager_output(line); } #endif #define STATUS_FORMAT " %s %s\n" static void print_state(int indent, const char *name, vdev_state_t state) { int i; char buf[512]; buf[0] = 0; for (i = 0; i < indent; i++) strcat(buf, " "); strcat(buf, name); pager_printf(STATUS_FORMAT, buf, state_name(state)); } static void vdev_status(vdev_t *vdev, int indent) { vdev_t *kid; print_state(indent, vdev->v_name, vdev->v_state); STAILQ_FOREACH(kid, &vdev->v_children, v_childlink) { vdev_status(kid, indent + 1); } } static void spa_status(spa_t *spa) { static char bootfs[ZFS_MAXNAMELEN]; uint64_t rootid; vdev_t *vdev; int good_kids, bad_kids, degraded_kids; vdev_state_t state; pager_printf(" pool: %s\n", spa->spa_name); if (zfs_get_root(spa, &rootid) == 0 && zfs_rlookup(spa, rootid, bootfs) == 0) { if (bootfs[0] == '\0') pager_printf("bootfs: %s\n", spa->spa_name); else pager_printf("bootfs: %s/%s\n", spa->spa_name, bootfs); } pager_printf("config:\n\n"); pager_printf(STATUS_FORMAT, "NAME", "STATE"); good_kids = 0; degraded_kids = 0; bad_kids = 0; STAILQ_FOREACH(vdev, &spa->spa_vdevs, v_childlink) { if (vdev->v_state == VDEV_STATE_HEALTHY) good_kids++; else if (vdev->v_state == VDEV_STATE_DEGRADED) degraded_kids++; else bad_kids++; } state = VDEV_STATE_CLOSED; if (good_kids > 0 && (degraded_kids + bad_kids) == 0) state = VDEV_STATE_HEALTHY; else if ((good_kids + degraded_kids) > 0) state = VDEV_STATE_DEGRADED; print_state(0, spa->spa_name, state); STAILQ_FOREACH(vdev, &spa->spa_vdevs, v_childlink) { vdev_status(vdev, 1); } } static void spa_all_status(void) { spa_t *spa; int first = 1; STAILQ_FOREACH(spa, &zfs_pools, spa_link) { if (!first) pager_printf("\n"); first = 0; spa_status(spa); } } static int vdev_probe(vdev_phys_read_t *_read, void *read_priv, spa_t **spap) { vdev_t vtmp; vdev_phys_t *vdev_label = (vdev_phys_t *) zap_scratch; spa_t *spa; vdev_t *vdev, *top_vdev, *pool_vdev; off_t off; blkptr_t bp; const unsigned char *nvlist; uint64_t val; uint64_t guid; uint64_t pool_txg, pool_guid; uint64_t is_log; const char *pool_name; const unsigned char *vdevs; const unsigned char *features; int i, rc, is_newer; char *upbuf; const struct uberblock *up; /* * Load the vdev label and figure out which * uberblock is most current. */ memset(&vtmp, 0, sizeof(vtmp)); vtmp.v_phys_read = _read; vtmp.v_read_priv = read_priv; off = offsetof(vdev_label_t, vl_vdev_phys); BP_ZERO(&bp); BP_SET_LSIZE(&bp, sizeof(vdev_phys_t)); BP_SET_PSIZE(&bp, sizeof(vdev_phys_t)); BP_SET_CHECKSUM(&bp, ZIO_CHECKSUM_LABEL); BP_SET_COMPRESS(&bp, ZIO_COMPRESS_OFF); DVA_SET_OFFSET(BP_IDENTITY(&bp), off); ZIO_SET_CHECKSUM(&bp.blk_cksum, off, 0, 0, 0); if (vdev_read_phys(&vtmp, &bp, vdev_label, off, 0)) return (EIO); if (vdev_label->vp_nvlist[0] != NV_ENCODE_XDR) { return (EIO); } nvlist = (const unsigned char *) vdev_label->vp_nvlist + 4; if (nvlist_find(nvlist, ZPOOL_CONFIG_VERSION, DATA_TYPE_UINT64, 0, &val)) { return (EIO); } if (!SPA_VERSION_IS_SUPPORTED(val)) { printf("ZFS: unsupported ZFS version %u (should be %u)\n", (unsigned) val, (unsigned) SPA_VERSION); return (EIO); } /* Check ZFS features for read */ if (nvlist_find(nvlist, ZPOOL_CONFIG_FEATURES_FOR_READ, DATA_TYPE_NVLIST, 0, &features) == 0 && nvlist_check_features_for_read(features) != 0) return (EIO); if (nvlist_find(nvlist, ZPOOL_CONFIG_POOL_STATE, DATA_TYPE_UINT64, 0, &val)) { return (EIO); } if (val == POOL_STATE_DESTROYED) { /* We don't boot only from destroyed pools. */ return (EIO); } if (nvlist_find(nvlist, ZPOOL_CONFIG_POOL_TXG, DATA_TYPE_UINT64, 0, &pool_txg) || nvlist_find(nvlist, ZPOOL_CONFIG_POOL_GUID, DATA_TYPE_UINT64, 0, &pool_guid) || nvlist_find(nvlist, ZPOOL_CONFIG_POOL_NAME, DATA_TYPE_STRING, 0, &pool_name)) { /* * Cache and spare devices end up here - just ignore * them. */ /*printf("ZFS: can't find pool details\n");*/ return (EIO); } is_log = 0; (void) nvlist_find(nvlist, ZPOOL_CONFIG_IS_LOG, DATA_TYPE_UINT64, 0, &is_log); if (is_log) return (EIO); /* * Create the pool if this is the first time we've seen it. */ spa = spa_find_by_guid(pool_guid); if (!spa) { spa = spa_create(pool_guid); spa->spa_name = strdup(pool_name); } if (pool_txg > spa->spa_txg) { spa->spa_txg = pool_txg; is_newer = 1; } else is_newer = 0; /* * Get the vdev tree and create our in-core copy of it. * If we already have a vdev with this guid, this must * be some kind of alias (overlapping slices, dangerously dedicated * disks etc). */ if (nvlist_find(nvlist, ZPOOL_CONFIG_GUID, DATA_TYPE_UINT64, 0, &guid)) { return (EIO); } vdev = vdev_find(guid); if (vdev && vdev->v_phys_read) /* Has this vdev already been inited? */ return (EIO); if (nvlist_find(nvlist, ZPOOL_CONFIG_VDEV_TREE, DATA_TYPE_NVLIST, 0, &vdevs)) { return (EIO); } rc = vdev_init_from_nvlist(vdevs, NULL, &top_vdev, is_newer); if (rc) return (rc); /* * Add the toplevel vdev to the pool if its not already there. */ STAILQ_FOREACH(pool_vdev, &spa->spa_vdevs, v_childlink) if (top_vdev == pool_vdev) break; if (!pool_vdev && top_vdev) STAILQ_INSERT_TAIL(&spa->spa_vdevs, top_vdev, v_childlink); /* * We should already have created an incomplete vdev for this * vdev. Find it and initialise it with our read proc. */ vdev = vdev_find(guid); if (vdev) { vdev->v_phys_read = _read; vdev->v_read_priv = read_priv; vdev->v_state = VDEV_STATE_HEALTHY; } else { printf("ZFS: inconsistent nvlist contents\n"); return (EIO); } /* * Re-evaluate top-level vdev state. */ vdev_set_state(top_vdev); /* * Ok, we are happy with the pool so far. Lets find * the best uberblock and then we can actually access * the contents of the pool. */ upbuf = zfs_alloc(VDEV_UBERBLOCK_SIZE(vdev)); up = (const struct uberblock *)upbuf; for (i = 0; i < VDEV_UBERBLOCK_COUNT(vdev); i++) { off = VDEV_UBERBLOCK_OFFSET(vdev, i); BP_ZERO(&bp); DVA_SET_OFFSET(&bp.blk_dva[0], off); BP_SET_LSIZE(&bp, VDEV_UBERBLOCK_SIZE(vdev)); BP_SET_PSIZE(&bp, VDEV_UBERBLOCK_SIZE(vdev)); BP_SET_CHECKSUM(&bp, ZIO_CHECKSUM_LABEL); BP_SET_COMPRESS(&bp, ZIO_COMPRESS_OFF); ZIO_SET_CHECKSUM(&bp.blk_cksum, off, 0, 0, 0); if (vdev_read_phys(vdev, &bp, upbuf, off, 0)) continue; if (up->ub_magic != UBERBLOCK_MAGIC) continue; if (up->ub_txg < spa->spa_txg) continue; if (up->ub_txg > spa->spa_uberblock.ub_txg) { spa->spa_uberblock = *up; } else if (up->ub_txg == spa->spa_uberblock.ub_txg) { if (up->ub_timestamp > spa->spa_uberblock.ub_timestamp) spa->spa_uberblock = *up; } } zfs_free(upbuf, VDEV_UBERBLOCK_SIZE(vdev)); + vdev->spa = spa; if (spap) *spap = spa; return (0); } static int ilog2(int n) { int v; for (v = 0; v < 32; v++) if (n == (1 << v)) return v; return -1; } static int zio_read_gang(const spa_t *spa, const blkptr_t *bp, void *buf) { blkptr_t gbh_bp; zio_gbh_phys_t zio_gb; char *pbuf; int i; /* Artificial BP for gang block header. */ gbh_bp = *bp; BP_SET_PSIZE(&gbh_bp, SPA_GANGBLOCKSIZE); BP_SET_LSIZE(&gbh_bp, SPA_GANGBLOCKSIZE); BP_SET_CHECKSUM(&gbh_bp, ZIO_CHECKSUM_GANG_HEADER); BP_SET_COMPRESS(&gbh_bp, ZIO_COMPRESS_OFF); for (i = 0; i < SPA_DVAS_PER_BP; i++) DVA_SET_GANG(&gbh_bp.blk_dva[i], 0); /* Read gang header block using the artificial BP. */ if (zio_read(spa, &gbh_bp, &zio_gb)) return (EIO); pbuf = buf; for (i = 0; i < SPA_GBH_NBLKPTRS; i++) { blkptr_t *gbp = &zio_gb.zg_blkptr[i]; if (BP_IS_HOLE(gbp)) continue; if (zio_read(spa, gbp, pbuf)) return (EIO); pbuf += BP_GET_PSIZE(gbp); } - if (zio_checksum_verify(bp, buf)) + if (zio_checksum_verify(spa, bp, buf)) return (EIO); return (0); } static int zio_read(const spa_t *spa, const blkptr_t *bp, void *buf) { int cpfunc = BP_GET_COMPRESS(bp); uint64_t align, size; void *pbuf; int i, error; /* * Process data embedded in block pointer */ if (BP_IS_EMBEDDED(bp)) { ASSERT(BPE_GET_ETYPE(bp) == BP_EMBEDDED_TYPE_DATA); size = BPE_GET_PSIZE(bp); ASSERT(size <= BPE_PAYLOAD_SIZE); if (cpfunc != ZIO_COMPRESS_OFF) pbuf = zfs_alloc(size); else pbuf = buf; decode_embedded_bp_compressed(bp, pbuf); error = 0; if (cpfunc != ZIO_COMPRESS_OFF) { error = zio_decompress_data(cpfunc, pbuf, size, buf, BP_GET_LSIZE(bp)); zfs_free(pbuf, size); } if (error != 0) printf("ZFS: i/o error - unable to decompress block pointer data, error %d\n", error); return (error); } error = EIO; for (i = 0; i < SPA_DVAS_PER_BP; i++) { const dva_t *dva = &bp->blk_dva[i]; vdev_t *vdev; int vdevid; off_t offset; if (!dva->dva_word[0] && !dva->dva_word[1]) continue; vdevid = DVA_GET_VDEV(dva); offset = DVA_GET_OFFSET(dva); STAILQ_FOREACH(vdev, &spa->spa_vdevs, v_childlink) { if (vdev->v_id == vdevid) break; } if (!vdev || !vdev->v_read) continue; size = BP_GET_PSIZE(bp); if (vdev->v_read == vdev_raidz_read) { align = 1ULL << vdev->v_top->v_ashift; if (P2PHASE(size, align) != 0) size = P2ROUNDUP(size, align); } if (size != BP_GET_PSIZE(bp) || cpfunc != ZIO_COMPRESS_OFF) pbuf = zfs_alloc(size); else pbuf = buf; if (DVA_GET_GANG(dva)) error = zio_read_gang(spa, bp, pbuf); else error = vdev->v_read(vdev, bp, pbuf, offset, size); if (error == 0) { if (cpfunc != ZIO_COMPRESS_OFF) error = zio_decompress_data(cpfunc, pbuf, BP_GET_PSIZE(bp), buf, BP_GET_LSIZE(bp)); else if (size != BP_GET_PSIZE(bp)) bcopy(pbuf, buf, BP_GET_PSIZE(bp)); } if (buf != pbuf) zfs_free(pbuf, size); if (error == 0) break; } if (error != 0) printf("ZFS: i/o error - all block copies unavailable\n"); return (error); } static int dnode_read(const spa_t *spa, const dnode_phys_t *dnode, off_t offset, void *buf, size_t buflen) { int ibshift = dnode->dn_indblkshift - SPA_BLKPTRSHIFT; int bsize = dnode->dn_datablkszsec << SPA_MINBLOCKSHIFT; int nlevels = dnode->dn_nlevels; int i, rc; if (bsize > SPA_MAXBLOCKSIZE) { - printf("ZFS: I/O error - blocks larger than 128K are not supported\n"); + printf("ZFS: I/O error - blocks larger than %llu are not " + "supported\n", SPA_MAXBLOCKSIZE); return (EIO); } /* * Note: bsize may not be a power of two here so we need to do an * actual divide rather than a bitshift. */ while (buflen > 0) { uint64_t bn = offset / bsize; int boff = offset % bsize; int ibn; const blkptr_t *indbp; blkptr_t bp; if (bn > dnode->dn_maxblkid) return (EIO); if (dnode == dnode_cache_obj && bn == dnode_cache_bn) goto cached; indbp = dnode->dn_blkptr; for (i = 0; i < nlevels; i++) { /* * Copy the bp from the indirect array so that * we can re-use the scratch buffer for multi-level * objects. */ ibn = bn >> ((nlevels - i - 1) * ibshift); ibn &= ((1 << ibshift) - 1); bp = indbp[ibn]; if (BP_IS_HOLE(&bp)) { memset(dnode_cache_buf, 0, bsize); break; } rc = zio_read(spa, &bp, dnode_cache_buf); if (rc) return (rc); indbp = (const blkptr_t *) dnode_cache_buf; } dnode_cache_obj = dnode; dnode_cache_bn = bn; cached: /* * The buffer contains our data block. Copy what we * need from it and loop. */ i = bsize - boff; if (i > buflen) i = buflen; memcpy(buf, &dnode_cache_buf[boff], i); buf = ((char*) buf) + i; offset += i; buflen -= i; } return (0); } /* * Lookup a value in a microzap directory. Assumes that the zap * scratch buffer contains the directory contents. */ static int mzap_lookup(const dnode_phys_t *dnode, const char *name, uint64_t *value) { const mzap_phys_t *mz; const mzap_ent_phys_t *mze; size_t size; int chunks, i; /* * Microzap objects use exactly one block. Read the whole * thing. */ size = dnode->dn_datablkszsec * 512; mz = (const mzap_phys_t *) zap_scratch; chunks = size / MZAP_ENT_LEN - 1; for (i = 0; i < chunks; i++) { mze = &mz->mz_chunk[i]; if (!strcmp(mze->mze_name, name)) { *value = mze->mze_value; return (0); } } return (ENOENT); } /* * Compare a name with a zap leaf entry. Return non-zero if the name * matches. */ static int fzap_name_equal(const zap_leaf_t *zl, const zap_leaf_chunk_t *zc, const char *name) { size_t namelen; const zap_leaf_chunk_t *nc; const char *p; namelen = zc->l_entry.le_name_numints; nc = &ZAP_LEAF_CHUNK(zl, zc->l_entry.le_name_chunk); p = name; while (namelen > 0) { size_t len; len = namelen; if (len > ZAP_LEAF_ARRAY_BYTES) len = ZAP_LEAF_ARRAY_BYTES; if (memcmp(p, nc->l_array.la_array, len)) return (0); p += len; namelen -= len; nc = &ZAP_LEAF_CHUNK(zl, nc->l_array.la_next); } return 1; } /* * Extract a uint64_t value from a zap leaf entry. */ static uint64_t fzap_leaf_value(const zap_leaf_t *zl, const zap_leaf_chunk_t *zc) { const zap_leaf_chunk_t *vc; int i; uint64_t value; const uint8_t *p; vc = &ZAP_LEAF_CHUNK(zl, zc->l_entry.le_value_chunk); for (i = 0, value = 0, p = vc->l_array.la_array; i < 8; i++) { value = (value << 8) | p[i]; } return value; } +static void +stv(int len, void *addr, uint64_t value) +{ + switch (len) { + case 1: + *(uint8_t *)addr = value; + return; + case 2: + *(uint16_t *)addr = value; + return; + case 4: + *(uint32_t *)addr = value; + return; + case 8: + *(uint64_t *)addr = value; + return; + } +} + /* + * Extract a array from a zap leaf entry. + */ +static void +fzap_leaf_array(const zap_leaf_t *zl, const zap_leaf_chunk_t *zc, + uint64_t integer_size, uint64_t num_integers, void *buf) +{ + uint64_t array_int_len = zc->l_entry.le_value_intlen; + uint64_t value = 0; + uint64_t *u64 = buf; + char *p = buf; + int len = MIN(zc->l_entry.le_value_numints, num_integers); + int chunk = zc->l_entry.le_value_chunk; + int byten = 0; + + if (integer_size == 8 && len == 1) { + *u64 = fzap_leaf_value(zl, zc); + return; + } + + while (len > 0) { + struct zap_leaf_array *la = &ZAP_LEAF_CHUNK(zl, chunk).l_array; + int i; + + ASSERT3U(chunk, <, ZAP_LEAF_NUMCHUNKS(zl)); + for (i = 0; i < ZAP_LEAF_ARRAY_BYTES && len > 0; i++) { + value = (value << 8) | la->la_array[i]; + byten++; + if (byten == array_int_len) { + stv(integer_size, p, value); + byten = 0; + len--; + if (len == 0) + return; + p += integer_size; + } + } + chunk = la->la_next; + } +} + +/* * Lookup a value in a fatzap directory. Assumes that the zap scratch * buffer contains the directory header. */ static int -fzap_lookup(const spa_t *spa, const dnode_phys_t *dnode, const char *name, uint64_t *value) +fzap_lookup(const spa_t *spa, const dnode_phys_t *dnode, const char *name, + uint64_t integer_size, uint64_t num_integers, void *value) { int bsize = dnode->dn_datablkszsec << SPA_MINBLOCKSHIFT; zap_phys_t zh = *(zap_phys_t *) zap_scratch; fat_zap_t z; uint64_t *ptrtbl; uint64_t hash; int rc; if (zh.zap_magic != ZAP_MAGIC) return (EIO); z.zap_block_shift = ilog2(bsize); z.zap_phys = (zap_phys_t *) zap_scratch; /* * Figure out where the pointer table is and read it in if necessary. */ if (zh.zap_ptrtbl.zt_blk) { rc = dnode_read(spa, dnode, zh.zap_ptrtbl.zt_blk * bsize, zap_scratch, bsize); if (rc) return (rc); ptrtbl = (uint64_t *) zap_scratch; } else { ptrtbl = &ZAP_EMBEDDED_PTRTBL_ENT(&z, 0); } hash = zap_hash(zh.zap_salt, name); zap_leaf_t zl; zl.l_bs = z.zap_block_shift; off_t off = ptrtbl[hash >> (64 - zh.zap_ptrtbl.zt_shift)] << zl.l_bs; zap_leaf_chunk_t *zc; rc = dnode_read(spa, dnode, off, zap_scratch, bsize); if (rc) return (rc); zl.l_phys = (zap_leaf_phys_t *) zap_scratch; /* * Make sure this chunk matches our hash. */ if (zl.l_phys->l_hdr.lh_prefix_len > 0 && zl.l_phys->l_hdr.lh_prefix != hash >> (64 - zl.l_phys->l_hdr.lh_prefix_len)) return (ENOENT); /* * Hash within the chunk to find our entry. */ int shift = (64 - ZAP_LEAF_HASH_SHIFT(&zl) - zl.l_phys->l_hdr.lh_prefix_len); int h = (hash >> shift) & ((1 << ZAP_LEAF_HASH_SHIFT(&zl)) - 1); h = zl.l_phys->l_hash[h]; if (h == 0xffff) return (ENOENT); zc = &ZAP_LEAF_CHUNK(&zl, h); while (zc->l_entry.le_hash != hash) { if (zc->l_entry.le_next == 0xffff) { zc = NULL; break; } zc = &ZAP_LEAF_CHUNK(&zl, zc->l_entry.le_next); } if (fzap_name_equal(&zl, zc, name)) { - if (zc->l_entry.le_value_intlen * zc->l_entry.le_value_numints > 8) + if (zc->l_entry.le_value_intlen * zc->l_entry.le_value_numints > + integer_size * num_integers) return (E2BIG); - *value = fzap_leaf_value(&zl, zc); + fzap_leaf_array(&zl, zc, integer_size, num_integers, value); return (0); } return (ENOENT); } /* * Lookup a name in a zap object and return its value as a uint64_t. */ static int -zap_lookup(const spa_t *spa, const dnode_phys_t *dnode, const char *name, uint64_t *value) +zap_lookup(const spa_t *spa, const dnode_phys_t *dnode, const char *name, + uint64_t integer_size, uint64_t num_integers, void *value) { int rc; uint64_t zap_type; size_t size = dnode->dn_datablkszsec << SPA_MINBLOCKSHIFT; rc = dnode_read(spa, dnode, 0, zap_scratch, size); if (rc) return (rc); zap_type = *(uint64_t *) zap_scratch; if (zap_type == ZBT_MICRO) return mzap_lookup(dnode, name, value); - else if (zap_type == ZBT_HEADER) - return fzap_lookup(spa, dnode, name, value); + else if (zap_type == ZBT_HEADER) { + return fzap_lookup(spa, dnode, name, integer_size, + num_integers, value); + } printf("ZFS: invalid zap_type=%d\n", (int)zap_type); return (EIO); } /* * List a microzap directory. Assumes that the zap scratch buffer contains * the directory contents. */ static int mzap_list(const dnode_phys_t *dnode, int (*callback)(const char *, uint64_t)) { const mzap_phys_t *mz; const mzap_ent_phys_t *mze; size_t size; int chunks, i, rc; /* * Microzap objects use exactly one block. Read the whole * thing. */ size = dnode->dn_datablkszsec * 512; mz = (const mzap_phys_t *) zap_scratch; chunks = size / MZAP_ENT_LEN - 1; for (i = 0; i < chunks; i++) { mze = &mz->mz_chunk[i]; if (mze->mze_name[0]) { rc = callback(mze->mze_name, mze->mze_value); if (rc != 0) return (rc); } } return (0); } /* * List a fatzap directory. Assumes that the zap scratch buffer contains * the directory header. */ static int fzap_list(const spa_t *spa, const dnode_phys_t *dnode, int (*callback)(const char *, uint64_t)) { int bsize = dnode->dn_datablkszsec << SPA_MINBLOCKSHIFT; zap_phys_t zh = *(zap_phys_t *) zap_scratch; fat_zap_t z; int i, j, rc; if (zh.zap_magic != ZAP_MAGIC) return (EIO); z.zap_block_shift = ilog2(bsize); z.zap_phys = (zap_phys_t *) zap_scratch; /* * This assumes that the leaf blocks start at block 1. The * documentation isn't exactly clear on this. */ zap_leaf_t zl; zl.l_bs = z.zap_block_shift; for (i = 0; i < zh.zap_num_leafs; i++) { off_t off = (i + 1) << zl.l_bs; char name[256], *p; uint64_t value; if (dnode_read(spa, dnode, off, zap_scratch, bsize)) return (EIO); zl.l_phys = (zap_leaf_phys_t *) zap_scratch; for (j = 0; j < ZAP_LEAF_NUMCHUNKS(&zl); j++) { zap_leaf_chunk_t *zc, *nc; int namelen; zc = &ZAP_LEAF_CHUNK(&zl, j); if (zc->l_entry.le_type != ZAP_CHUNK_ENTRY) continue; namelen = zc->l_entry.le_name_numints; if (namelen > sizeof(name)) namelen = sizeof(name); /* * Paste the name back together. */ nc = &ZAP_LEAF_CHUNK(&zl, zc->l_entry.le_name_chunk); p = name; while (namelen > 0) { int len; len = namelen; if (len > ZAP_LEAF_ARRAY_BYTES) len = ZAP_LEAF_ARRAY_BYTES; memcpy(p, nc->l_array.la_array, len); p += len; namelen -= len; nc = &ZAP_LEAF_CHUNK(&zl, nc->l_array.la_next); } /* * Assume the first eight bytes of the value are * a uint64_t. */ value = fzap_leaf_value(&zl, zc); //printf("%s 0x%jx\n", name, (uintmax_t)value); rc = callback((const char *)name, value); if (rc != 0) return (rc); } } return (0); } static int zfs_printf(const char *name, uint64_t value __unused) { printf("%s\n", name); return (0); } /* * List a zap directory. */ static int zap_list(const spa_t *spa, const dnode_phys_t *dnode) { uint64_t zap_type; size_t size = dnode->dn_datablkszsec * 512; if (dnode_read(spa, dnode, 0, zap_scratch, size)) return (EIO); zap_type = *(uint64_t *) zap_scratch; if (zap_type == ZBT_MICRO) return mzap_list(dnode, zfs_printf); else return fzap_list(spa, dnode, zfs_printf); } static int objset_get_dnode(const spa_t *spa, const objset_phys_t *os, uint64_t objnum, dnode_phys_t *dnode) { off_t offset; offset = objnum * sizeof(dnode_phys_t); return dnode_read(spa, &os->os_meta_dnode, offset, dnode, sizeof(dnode_phys_t)); } static int mzap_rlookup(const spa_t *spa, const dnode_phys_t *dnode, char *name, uint64_t value) { const mzap_phys_t *mz; const mzap_ent_phys_t *mze; size_t size; int chunks, i; /* * Microzap objects use exactly one block. Read the whole * thing. */ size = dnode->dn_datablkszsec * 512; mz = (const mzap_phys_t *) zap_scratch; chunks = size / MZAP_ENT_LEN - 1; for (i = 0; i < chunks; i++) { mze = &mz->mz_chunk[i]; if (value == mze->mze_value) { strcpy(name, mze->mze_name); return (0); } } return (ENOENT); } static void fzap_name_copy(const zap_leaf_t *zl, const zap_leaf_chunk_t *zc, char *name) { size_t namelen; const zap_leaf_chunk_t *nc; char *p; namelen = zc->l_entry.le_name_numints; nc = &ZAP_LEAF_CHUNK(zl, zc->l_entry.le_name_chunk); p = name; while (namelen > 0) { size_t len; len = namelen; if (len > ZAP_LEAF_ARRAY_BYTES) len = ZAP_LEAF_ARRAY_BYTES; memcpy(p, nc->l_array.la_array, len); p += len; namelen -= len; nc = &ZAP_LEAF_CHUNK(zl, nc->l_array.la_next); } *p = '\0'; } static int fzap_rlookup(const spa_t *spa, const dnode_phys_t *dnode, char *name, uint64_t value) { int bsize = dnode->dn_datablkszsec << SPA_MINBLOCKSHIFT; zap_phys_t zh = *(zap_phys_t *) zap_scratch; fat_zap_t z; int i, j; if (zh.zap_magic != ZAP_MAGIC) return (EIO); z.zap_block_shift = ilog2(bsize); z.zap_phys = (zap_phys_t *) zap_scratch; /* * This assumes that the leaf blocks start at block 1. The * documentation isn't exactly clear on this. */ zap_leaf_t zl; zl.l_bs = z.zap_block_shift; for (i = 0; i < zh.zap_num_leafs; i++) { off_t off = (i + 1) << zl.l_bs; if (dnode_read(spa, dnode, off, zap_scratch, bsize)) return (EIO); zl.l_phys = (zap_leaf_phys_t *) zap_scratch; for (j = 0; j < ZAP_LEAF_NUMCHUNKS(&zl); j++) { zap_leaf_chunk_t *zc; zc = &ZAP_LEAF_CHUNK(&zl, j); if (zc->l_entry.le_type != ZAP_CHUNK_ENTRY) continue; if (zc->l_entry.le_value_intlen != 8 || zc->l_entry.le_value_numints != 1) continue; if (fzap_leaf_value(&zl, zc) == value) { fzap_name_copy(&zl, zc, name); return (0); } } } return (ENOENT); } static int zap_rlookup(const spa_t *spa, const dnode_phys_t *dnode, char *name, uint64_t value) { int rc; uint64_t zap_type; size_t size = dnode->dn_datablkszsec * 512; rc = dnode_read(spa, dnode, 0, zap_scratch, size); if (rc) return (rc); zap_type = *(uint64_t *) zap_scratch; if (zap_type == ZBT_MICRO) return mzap_rlookup(spa, dnode, name, value); else return fzap_rlookup(spa, dnode, name, value); } static int zfs_rlookup(const spa_t *spa, uint64_t objnum, char *result) { char name[256]; char component[256]; uint64_t dir_obj, parent_obj, child_dir_zapobj; dnode_phys_t child_dir_zap, dataset, dir, parent; dsl_dir_phys_t *dd; dsl_dataset_phys_t *ds; char *p; int len; p = &name[sizeof(name) - 1]; *p = '\0'; if (objset_get_dnode(spa, &spa->spa_mos, objnum, &dataset)) { printf("ZFS: can't find dataset %ju\n", (uintmax_t)objnum); return (EIO); } ds = (dsl_dataset_phys_t *)&dataset.dn_bonus; dir_obj = ds->ds_dir_obj; for (;;) { if (objset_get_dnode(spa, &spa->spa_mos, dir_obj, &dir) != 0) return (EIO); dd = (dsl_dir_phys_t *)&dir.dn_bonus; /* Actual loop condition. */ parent_obj = dd->dd_parent_obj; if (parent_obj == 0) break; if (objset_get_dnode(spa, &spa->spa_mos, parent_obj, &parent) != 0) return (EIO); dd = (dsl_dir_phys_t *)&parent.dn_bonus; child_dir_zapobj = dd->dd_child_dir_zapobj; if (objset_get_dnode(spa, &spa->spa_mos, child_dir_zapobj, &child_dir_zap) != 0) return (EIO); if (zap_rlookup(spa, &child_dir_zap, component, dir_obj) != 0) return (EIO); len = strlen(component); p -= len; memcpy(p, component, len); --p; *p = '/'; /* Actual loop iteration. */ dir_obj = parent_obj; } if (*p != '\0') ++p; strcpy(result, p); return (0); } static int zfs_lookup_dataset(const spa_t *spa, const char *name, uint64_t *objnum) { char element[256]; uint64_t dir_obj, child_dir_zapobj; dnode_phys_t child_dir_zap, dir; dsl_dir_phys_t *dd; const char *p, *q; if (objset_get_dnode(spa, &spa->spa_mos, DMU_POOL_DIRECTORY_OBJECT, &dir)) return (EIO); - if (zap_lookup(spa, &dir, DMU_POOL_ROOT_DATASET, &dir_obj)) + if (zap_lookup(spa, &dir, DMU_POOL_ROOT_DATASET, sizeof (dir_obj), + 1, &dir_obj)) return (EIO); p = name; for (;;) { if (objset_get_dnode(spa, &spa->spa_mos, dir_obj, &dir)) return (EIO); dd = (dsl_dir_phys_t *)&dir.dn_bonus; while (*p == '/') p++; /* Actual loop condition #1. */ if (*p == '\0') break; q = strchr(p, '/'); if (q) { memcpy(element, p, q - p); element[q - p] = '\0'; p = q + 1; } else { strcpy(element, p); p += strlen(p); } child_dir_zapobj = dd->dd_child_dir_zapobj; if (objset_get_dnode(spa, &spa->spa_mos, child_dir_zapobj, &child_dir_zap) != 0) return (EIO); /* Actual loop condition #2. */ - if (zap_lookup(spa, &child_dir_zap, element, &dir_obj) != 0) + if (zap_lookup(spa, &child_dir_zap, element, sizeof (dir_obj), + 1, &dir_obj) != 0) return (ENOENT); } *objnum = dd->dd_head_dataset_obj; return (0); } #ifndef BOOT2 static int zfs_list_dataset(const spa_t *spa, uint64_t objnum/*, int pos, char *entry*/) { uint64_t dir_obj, child_dir_zapobj; dnode_phys_t child_dir_zap, dir, dataset; dsl_dataset_phys_t *ds; dsl_dir_phys_t *dd; if (objset_get_dnode(spa, &spa->spa_mos, objnum, &dataset)) { printf("ZFS: can't find dataset %ju\n", (uintmax_t)objnum); return (EIO); } ds = (dsl_dataset_phys_t *) &dataset.dn_bonus; dir_obj = ds->ds_dir_obj; if (objset_get_dnode(spa, &spa->spa_mos, dir_obj, &dir)) { printf("ZFS: can't find dirobj %ju\n", (uintmax_t)dir_obj); return (EIO); } dd = (dsl_dir_phys_t *)&dir.dn_bonus; child_dir_zapobj = dd->dd_child_dir_zapobj; if (objset_get_dnode(spa, &spa->spa_mos, child_dir_zapobj, &child_dir_zap) != 0) { printf("ZFS: can't find child zap %ju\n", (uintmax_t)dir_obj); return (EIO); } return (zap_list(spa, &child_dir_zap) != 0); } int zfs_callback_dataset(const spa_t *spa, uint64_t objnum, int (*callback)(const char *, uint64_t)) { uint64_t dir_obj, child_dir_zapobj, zap_type; dnode_phys_t child_dir_zap, dir, dataset; dsl_dataset_phys_t *ds; dsl_dir_phys_t *dd; int err; err = objset_get_dnode(spa, &spa->spa_mos, objnum, &dataset); if (err != 0) { printf("ZFS: can't find dataset %ju\n", (uintmax_t)objnum); return (err); } ds = (dsl_dataset_phys_t *) &dataset.dn_bonus; dir_obj = ds->ds_dir_obj; err = objset_get_dnode(spa, &spa->spa_mos, dir_obj, &dir); if (err != 0) { printf("ZFS: can't find dirobj %ju\n", (uintmax_t)dir_obj); return (err); } dd = (dsl_dir_phys_t *)&dir.dn_bonus; child_dir_zapobj = dd->dd_child_dir_zapobj; err = objset_get_dnode(spa, &spa->spa_mos, child_dir_zapobj, &child_dir_zap); if (err != 0) { printf("ZFS: can't find child zap %ju\n", (uintmax_t)dir_obj); return (err); } err = dnode_read(spa, &child_dir_zap, 0, zap_scratch, child_dir_zap.dn_datablkszsec * 512); if (err != 0) return (err); zap_type = *(uint64_t *) zap_scratch; if (zap_type == ZBT_MICRO) return mzap_list(&child_dir_zap, callback); else return fzap_list(spa, &child_dir_zap, callback); } #endif /* * Find the object set given the object number of its dataset object * and return its details in *objset */ static int zfs_mount_dataset(const spa_t *spa, uint64_t objnum, objset_phys_t *objset) { dnode_phys_t dataset; dsl_dataset_phys_t *ds; if (objset_get_dnode(spa, &spa->spa_mos, objnum, &dataset)) { printf("ZFS: can't find dataset %ju\n", (uintmax_t)objnum); return (EIO); } ds = (dsl_dataset_phys_t *) &dataset.dn_bonus; if (zio_read(spa, &ds->ds_bp, objset)) { printf("ZFS: can't read object set for dataset %ju\n", (uintmax_t)objnum); return (EIO); } return (0); } /* * Find the object set pointed to by the BOOTFS property or the root * dataset if there is none and return its details in *objset */ static int zfs_get_root(const spa_t *spa, uint64_t *objid) { dnode_phys_t dir, propdir; uint64_t props, bootfs, root; *objid = 0; /* * Start with the MOS directory object. */ if (objset_get_dnode(spa, &spa->spa_mos, DMU_POOL_DIRECTORY_OBJECT, &dir)) { printf("ZFS: can't read MOS object directory\n"); return (EIO); } /* * Lookup the pool_props and see if we can find a bootfs. */ - if (zap_lookup(spa, &dir, DMU_POOL_PROPS, &props) == 0 + if (zap_lookup(spa, &dir, DMU_POOL_PROPS, sizeof (props), 1, &props) == 0 && objset_get_dnode(spa, &spa->spa_mos, props, &propdir) == 0 - && zap_lookup(spa, &propdir, "bootfs", &bootfs) == 0 + && zap_lookup(spa, &propdir, "bootfs", sizeof (bootfs), 1, &bootfs) == 0 && bootfs != 0) { *objid = bootfs; return (0); } /* * Lookup the root dataset directory */ - if (zap_lookup(spa, &dir, DMU_POOL_ROOT_DATASET, &root) + if (zap_lookup(spa, &dir, DMU_POOL_ROOT_DATASET, sizeof (root), 1, &root) || objset_get_dnode(spa, &spa->spa_mos, root, &dir)) { printf("ZFS: can't find root dsl_dir\n"); return (EIO); } /* * Use the information from the dataset directory's bonus buffer * to find the dataset object and from that the object set itself. */ dsl_dir_phys_t *dd = (dsl_dir_phys_t *) &dir.dn_bonus; *objid = dd->dd_head_dataset_obj; return (0); } static int zfs_mount(const spa_t *spa, uint64_t rootobj, struct zfsmount *mount) { mount->spa = spa; /* * Find the root object set if not explicitly provided */ if (rootobj == 0 && zfs_get_root(spa, &rootobj)) { printf("ZFS: can't find root filesystem\n"); return (EIO); } if (zfs_mount_dataset(spa, rootobj, &mount->objset)) { printf("ZFS: can't open root filesystem\n"); return (EIO); } mount->rootobj = rootobj; return (0); } /* * callback function for feature name checks. */ static int check_feature(const char *name, uint64_t value) { int i; if (value == 0) return (0); if (name[0] == '\0') return (0); for (i = 0; features_for_read[i] != NULL; i++) { if (strcmp(name, features_for_read[i]) == 0) return (0); } printf("ZFS: unsupported feature: %s\n", name); return (EIO); } /* * Checks whether the MOS features that are active are supported. */ static int check_mos_features(const spa_t *spa) { dnode_phys_t dir; uint64_t objnum, zap_type; size_t size; int rc; if ((rc = objset_get_dnode(spa, &spa->spa_mos, DMU_OT_OBJECT_DIRECTORY, &dir)) != 0) return (rc); - if ((rc = zap_lookup(spa, &dir, DMU_POOL_FEATURES_FOR_READ, &objnum)) != 0) { + if ((rc = zap_lookup(spa, &dir, DMU_POOL_FEATURES_FOR_READ, + sizeof (objnum), 1, &objnum)) != 0) { /* * It is older pool without features. As we have already * tested the label, just return without raising the error. */ return (0); } if ((rc = objset_get_dnode(spa, &spa->spa_mos, objnum, &dir)) != 0) return (rc); if (dir.dn_type != DMU_OTN_ZAP_METADATA) return (EIO); size = dir.dn_datablkszsec * 512; if (dnode_read(spa, &dir, 0, zap_scratch, size)) return (EIO); zap_type = *(uint64_t *) zap_scratch; if (zap_type == ZBT_MICRO) rc = mzap_list(&dir, check_feature); else rc = fzap_list(spa, &dir, check_feature); return (rc); } static int zfs_spa_init(spa_t *spa) { + dnode_phys_t dir; int rc; if (zio_read(spa, &spa->spa_uberblock.ub_rootbp, &spa->spa_mos)) { printf("ZFS: can't read MOS of pool %s\n", spa->spa_name); return (EIO); } if (spa->spa_mos.os_type != DMU_OST_META) { printf("ZFS: corrupted MOS of pool %s\n", spa->spa_name); return (EIO); } + if (objset_get_dnode(spa, &spa->spa_mos, DMU_POOL_DIRECTORY_OBJECT, + &dir)) { + printf("ZFS: failed to read pool %s directory object\n", + spa->spa_name); + return (EIO); + } + /* this is allowed to fail, older pools do not have salt */ + rc = zap_lookup(spa, &dir, DMU_POOL_CHECKSUM_SALT, 1, + sizeof (spa->spa_cksum_salt.zcs_bytes), + spa->spa_cksum_salt.zcs_bytes); + rc = check_mos_features(spa); if (rc != 0) { printf("ZFS: pool %s is not supported\n", spa->spa_name); } return (rc); } static int zfs_dnode_stat(const spa_t *spa, dnode_phys_t *dn, struct stat *sb) { if (dn->dn_bonustype != DMU_OT_SA) { znode_phys_t *zp = (znode_phys_t *)dn->dn_bonus; sb->st_mode = zp->zp_mode; sb->st_uid = zp->zp_uid; sb->st_gid = zp->zp_gid; sb->st_size = zp->zp_size; } else { sa_hdr_phys_t *sahdrp; int hdrsize; size_t size = 0; void *buf = NULL; if (dn->dn_bonuslen != 0) sahdrp = (sa_hdr_phys_t *)DN_BONUS(dn); else { if ((dn->dn_flags & DNODE_FLAG_SPILL_BLKPTR) != 0) { blkptr_t *bp = &dn->dn_spill; int error; size = BP_GET_LSIZE(bp); buf = zfs_alloc(size); error = zio_read(spa, bp, buf); if (error != 0) { zfs_free(buf, size); return (error); } sahdrp = buf; } else { return (EIO); } } hdrsize = SA_HDR_SIZE(sahdrp); sb->st_mode = *(uint64_t *)((char *)sahdrp + hdrsize + SA_MODE_OFFSET); sb->st_uid = *(uint64_t *)((char *)sahdrp + hdrsize + SA_UID_OFFSET); sb->st_gid = *(uint64_t *)((char *)sahdrp + hdrsize + SA_GID_OFFSET); sb->st_size = *(uint64_t *)((char *)sahdrp + hdrsize + SA_SIZE_OFFSET); if (buf != NULL) zfs_free(buf, size); } return (0); } static int zfs_dnode_readlink(const spa_t *spa, dnode_phys_t *dn, char *path, size_t psize) { int rc = 0; if (dn->dn_bonustype == DMU_OT_SA) { sa_hdr_phys_t *sahdrp = NULL; size_t size = 0; void *buf = NULL; int hdrsize; char *p; if (dn->dn_bonuslen != 0) sahdrp = (sa_hdr_phys_t *)DN_BONUS(dn); else { blkptr_t *bp; if ((dn->dn_flags & DNODE_FLAG_SPILL_BLKPTR) == 0) return (EIO); bp = &dn->dn_spill; size = BP_GET_LSIZE(bp); buf = zfs_alloc(size); rc = zio_read(spa, bp, buf); if (rc != 0) { zfs_free(buf, size); return (rc); } sahdrp = buf; } hdrsize = SA_HDR_SIZE(sahdrp); p = (char *)((uintptr_t)sahdrp + hdrsize + SA_SYMLINK_OFFSET); memcpy(path, p, psize); if (buf != NULL) zfs_free(buf, size); return (0); } /* * Second test is purely to silence bogus compiler * warning about accessing past the end of dn_bonus. */ if (psize + sizeof(znode_phys_t) <= dn->dn_bonuslen && sizeof(znode_phys_t) <= sizeof(dn->dn_bonus)) { memcpy(path, &dn->dn_bonus[sizeof(znode_phys_t)], psize); } else { rc = dnode_read(spa, dn, 0, path, psize); } return (rc); } struct obj_list { uint64_t objnum; STAILQ_ENTRY(obj_list) entry; }; /* * Lookup a file and return its dnode. */ static int zfs_lookup(const struct zfsmount *mount, const char *upath, dnode_phys_t *dnode) { int rc; uint64_t objnum; const spa_t *spa; dnode_phys_t dn; const char *p, *q; char element[256]; char path[1024]; int symlinks_followed = 0; struct stat sb; struct obj_list *entry, *tentry; STAILQ_HEAD(, obj_list) on_cache = STAILQ_HEAD_INITIALIZER(on_cache); spa = mount->spa; if (mount->objset.os_type != DMU_OST_ZFS) { printf("ZFS: unexpected object set type %ju\n", (uintmax_t)mount->objset.os_type); return (EIO); } if ((entry = malloc(sizeof(struct obj_list))) == NULL) return (ENOMEM); /* * Get the root directory dnode. */ rc = objset_get_dnode(spa, &mount->objset, MASTER_NODE_OBJ, &dn); if (rc) { free(entry); return (rc); } - rc = zap_lookup(spa, &dn, ZFS_ROOT_OBJ, &objnum); + rc = zap_lookup(spa, &dn, ZFS_ROOT_OBJ, sizeof (objnum), 1, &objnum); if (rc) { free(entry); return (rc); } entry->objnum = objnum; STAILQ_INSERT_HEAD(&on_cache, entry, entry); rc = objset_get_dnode(spa, &mount->objset, objnum, &dn); if (rc != 0) goto done; p = upath; while (p && *p) { rc = objset_get_dnode(spa, &mount->objset, objnum, &dn); if (rc != 0) goto done; while (*p == '/') p++; if (*p == '\0') break; q = p; while (*q != '\0' && *q != '/') q++; /* skip dot */ if (p + 1 == q && p[0] == '.') { p++; continue; } /* double dot */ if (p + 2 == q && p[0] == '.' && p[1] == '.') { p += 2; if (STAILQ_FIRST(&on_cache) == STAILQ_LAST(&on_cache, obj_list, entry)) { rc = ENOENT; goto done; } entry = STAILQ_FIRST(&on_cache); STAILQ_REMOVE_HEAD(&on_cache, entry); free(entry); objnum = (STAILQ_FIRST(&on_cache))->objnum; continue; } if (q - p + 1 > sizeof(element)) { rc = ENAMETOOLONG; goto done; } memcpy(element, p, q - p); element[q - p] = 0; p = q; if ((rc = zfs_dnode_stat(spa, &dn, &sb)) != 0) goto done; if (!S_ISDIR(sb.st_mode)) { rc = ENOTDIR; goto done; } - rc = zap_lookup(spa, &dn, element, &objnum); + rc = zap_lookup(spa, &dn, element, sizeof (objnum), 1, &objnum); if (rc) goto done; objnum = ZFS_DIRENT_OBJ(objnum); if ((entry = malloc(sizeof(struct obj_list))) == NULL) { rc = ENOMEM; goto done; } entry->objnum = objnum; STAILQ_INSERT_HEAD(&on_cache, entry, entry); rc = objset_get_dnode(spa, &mount->objset, objnum, &dn); if (rc) goto done; /* * Check for symlink. */ rc = zfs_dnode_stat(spa, &dn, &sb); if (rc) goto done; if (S_ISLNK(sb.st_mode)) { if (symlinks_followed > 10) { rc = EMLINK; goto done; } symlinks_followed++; /* * Read the link value and copy the tail of our * current path onto the end. */ if (sb.st_size + strlen(p) + 1 > sizeof(path)) { rc = ENAMETOOLONG; goto done; } strcpy(&path[sb.st_size], p); rc = zfs_dnode_readlink(spa, &dn, path, sb.st_size); if (rc != 0) goto done; /* * Restart with the new path, starting either at * the root or at the parent depending whether or * not the link is relative. */ p = path; if (*p == '/') { while (STAILQ_FIRST(&on_cache) != STAILQ_LAST(&on_cache, obj_list, entry)) { entry = STAILQ_FIRST(&on_cache); STAILQ_REMOVE_HEAD(&on_cache, entry); free(entry); } } else { entry = STAILQ_FIRST(&on_cache); STAILQ_REMOVE_HEAD(&on_cache, entry); free(entry); } objnum = (STAILQ_FIRST(&on_cache))->objnum; } } *dnode = dn; done: STAILQ_FOREACH_SAFE(entry, &on_cache, entry, tentry) free(entry); return (rc); } Index: stable/11/sys/cddl/boot/zfs/fletcher.c =================================================================== --- stable/11/sys/cddl/boot/zfs/fletcher.c (revision 328865) +++ stable/11/sys/cddl/boot/zfs/fletcher.c (revision 328866) @@ -1,94 +1,96 @@ /* * 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 2006 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ -/*#pragma ident "%Z%%M% %I% %E% SMI"*/ - static void -fletcher_2_native(const void *buf, uint64_t size, zio_cksum_t *zcp) +fletcher_2_native(const void *buf, uint64_t size, + const void *ctx_template __unused, zio_cksum_t *zcp) { const uint64_t *ip = buf; const uint64_t *ipend = ip + (size / sizeof (uint64_t)); uint64_t a0, b0, a1, b1; for (a0 = b0 = a1 = b1 = 0; ip < ipend; ip += 2) { a0 += ip[0]; a1 += ip[1]; b0 += a0; b1 += a1; } ZIO_SET_CHECKSUM(zcp, a0, a1, b0, b1); } static void -fletcher_2_byteswap(const void *buf, uint64_t size, zio_cksum_t *zcp) +fletcher_2_byteswap(const void *buf, uint64_t size, + const void *ctx_template __unused, zio_cksum_t *zcp) { const uint64_t *ip = buf; const uint64_t *ipend = ip + (size / sizeof (uint64_t)); uint64_t a0, b0, a1, b1; for (a0 = b0 = a1 = b1 = 0; ip < ipend; ip += 2) { a0 += BSWAP_64(ip[0]); a1 += BSWAP_64(ip[1]); b0 += a0; b1 += a1; } ZIO_SET_CHECKSUM(zcp, a0, a1, b0, b1); } static void -fletcher_4_native(const void *buf, uint64_t size, zio_cksum_t *zcp) +fletcher_4_native(const void *buf, uint64_t size, + const void *ctx_template __unused, zio_cksum_t *zcp) { const uint32_t *ip = buf; const uint32_t *ipend = ip + (size / sizeof (uint32_t)); uint64_t a, b, c, d; for (a = b = c = d = 0; ip < ipend; ip++) { a += ip[0]; b += a; c += b; d += c; } ZIO_SET_CHECKSUM(zcp, a, b, c, d); } static void -fletcher_4_byteswap(const void *buf, uint64_t size, zio_cksum_t *zcp) +fletcher_4_byteswap(const void *buf, uint64_t size, + const void *ctx_template __unused, zio_cksum_t *zcp) { const uint32_t *ip = buf; const uint32_t *ipend = ip + (size / sizeof (uint32_t)); uint64_t a, b, c, d; for (a = b = c = d = 0; ip < ipend; ip++) { a += BSWAP_32(ip[0]); b += a; c += b; d += c; } ZIO_SET_CHECKSUM(zcp, a, b, c, d); } Index: stable/11/sys/cddl/boot/zfs/sha256.c =================================================================== --- stable/11/sys/cddl/boot/zfs/sha256.c (revision 328865) +++ stable/11/sys/cddl/boot/zfs/sha256.c (revision 328866) @@ -1,127 +1,325 @@ /* * 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 2005 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ +/* + * Copyright 2013 Saso Kiselkov. All rights reserved. + * Copyright 2015 Toomas Soome + */ -/*#pragma ident "%Z%%M% %I% %E% SMI"*/ - /* - * SHA-256 checksum, as specified in FIPS 180-2, available at: + * SHA-256 and SHA-512/256 hashes, as specified in FIPS 180-4, available at: * http://csrc.nist.gov/cryptval * - * This is a very compact implementation of SHA-256. + * This is a very compact implementation of SHA-256 and SHA-512/256. * It is designed to be simple and portable, not to be fast. */ /* - * The literal definitions according to FIPS180-2 would be: + * The literal definitions according to FIPS180-4 would be: * * Ch(x, y, z) (((x) & (y)) ^ ((~(x)) & (z))) * Maj(x, y, z) (((x) & (y)) | ((x) & (z)) | ((y) & (z))) * * We use logical equivalents which require one less op. */ #define Ch(x, y, z) ((z) ^ ((x) & ((y) ^ (z)))) #define Maj(x, y, z) (((x) & (y)) ^ ((z) & ((x) ^ (y)))) -#define Rot32(x, s) (((x) >> s) | ((x) << (32 - s))) -#define SIGMA0(x) (Rot32(x, 2) ^ Rot32(x, 13) ^ Rot32(x, 22)) -#define SIGMA1(x) (Rot32(x, 6) ^ Rot32(x, 11) ^ Rot32(x, 25)) -#define sigma0(x) (Rot32(x, 7) ^ Rot32(x, 18) ^ ((x) >> 3)) -#define sigma1(x) (Rot32(x, 17) ^ Rot32(x, 19) ^ ((x) >> 10)) +#define ROTR(x, n) (((x) >> (n)) | ((x) << ((sizeof (x) * NBBY)-(n)))) +/* SHA-224/256 operations */ +#define BIGSIGMA0_256(x) (ROTR(x, 2) ^ ROTR(x, 13) ^ ROTR(x, 22)) +#define BIGSIGMA1_256(x) (ROTR(x, 6) ^ ROTR(x, 11) ^ ROTR(x, 25)) +#define SIGMA0_256(x) (ROTR(x, 7) ^ ROTR(x, 18) ^ ((x) >> 3)) +#define SIGMA1_256(x) (ROTR(x, 17) ^ ROTR(x, 19) ^ ((x) >> 10)) + +/* SHA-384/512 operations */ +#define BIGSIGMA0_512(x) (ROTR((x), 28) ^ ROTR((x), 34) ^ ROTR((x), 39)) +#define BIGSIGMA1_512(x) (ROTR((x), 14) ^ ROTR((x), 18) ^ ROTR((x), 41)) +#define SIGMA0_512(x) (ROTR((x), 1) ^ ROTR((x), 8) ^ ((x) >> 7)) +#define SIGMA1_512(x) (ROTR((x), 19) ^ ROTR((x), 61) ^ ((x) >> 6)) + +/* SHA-256 round constants */ static const uint32_t SHA256_K[64] = { 0x428a2f98, 0x71374491, 0xb5c0fbcf, 0xe9b5dba5, 0x3956c25b, 0x59f111f1, 0x923f82a4, 0xab1c5ed5, 0xd807aa98, 0x12835b01, 0x243185be, 0x550c7dc3, 0x72be5d74, 0x80deb1fe, 0x9bdc06a7, 0xc19bf174, 0xe49b69c1, 0xefbe4786, 0x0fc19dc6, 0x240ca1cc, 0x2de92c6f, 0x4a7484aa, 0x5cb0a9dc, 0x76f988da, 0x983e5152, 0xa831c66d, 0xb00327c8, 0xbf597fc7, 0xc6e00bf3, 0xd5a79147, 0x06ca6351, 0x14292967, 0x27b70a85, 0x2e1b2138, 0x4d2c6dfc, 0x53380d13, 0x650a7354, 0x766a0abb, 0x81c2c92e, 0x92722c85, 0xa2bfe8a1, 0xa81a664b, 0xc24b8b70, 0xc76c51a3, 0xd192e819, 0xd6990624, 0xf40e3585, 0x106aa070, 0x19a4c116, 0x1e376c08, 0x2748774c, 0x34b0bcb5, 0x391c0cb3, 0x4ed8aa4a, 0x5b9cca4f, 0x682e6ff3, 0x748f82ee, 0x78a5636f, 0x84c87814, 0x8cc70208, 0x90befffa, 0xa4506ceb, 0xbef9a3f7, 0xc67178f2 }; +/* SHA-512 round constants */ +static const uint64_t SHA512_K[80] = { + 0x428A2F98D728AE22ULL, 0x7137449123EF65CDULL, + 0xB5C0FBCFEC4D3B2FULL, 0xE9B5DBA58189DBBCULL, + 0x3956C25BF348B538ULL, 0x59F111F1B605D019ULL, + 0x923F82A4AF194F9BULL, 0xAB1C5ED5DA6D8118ULL, + 0xD807AA98A3030242ULL, 0x12835B0145706FBEULL, + 0x243185BE4EE4B28CULL, 0x550C7DC3D5FFB4E2ULL, + 0x72BE5D74F27B896FULL, 0x80DEB1FE3B1696B1ULL, + 0x9BDC06A725C71235ULL, 0xC19BF174CF692694ULL, + 0xE49B69C19EF14AD2ULL, 0xEFBE4786384F25E3ULL, + 0x0FC19DC68B8CD5B5ULL, 0x240CA1CC77AC9C65ULL, + 0x2DE92C6F592B0275ULL, 0x4A7484AA6EA6E483ULL, + 0x5CB0A9DCBD41FBD4ULL, 0x76F988DA831153B5ULL, + 0x983E5152EE66DFABULL, 0xA831C66D2DB43210ULL, + 0xB00327C898FB213FULL, 0xBF597FC7BEEF0EE4ULL, + 0xC6E00BF33DA88FC2ULL, 0xD5A79147930AA725ULL, + 0x06CA6351E003826FULL, 0x142929670A0E6E70ULL, + 0x27B70A8546D22FFCULL, 0x2E1B21385C26C926ULL, + 0x4D2C6DFC5AC42AEDULL, 0x53380D139D95B3DFULL, + 0x650A73548BAF63DEULL, 0x766A0ABB3C77B2A8ULL, + 0x81C2C92E47EDAEE6ULL, 0x92722C851482353BULL, + 0xA2BFE8A14CF10364ULL, 0xA81A664BBC423001ULL, + 0xC24B8B70D0F89791ULL, 0xC76C51A30654BE30ULL, + 0xD192E819D6EF5218ULL, 0xD69906245565A910ULL, + 0xF40E35855771202AULL, 0x106AA07032BBD1B8ULL, + 0x19A4C116B8D2D0C8ULL, 0x1E376C085141AB53ULL, + 0x2748774CDF8EEB99ULL, 0x34B0BCB5E19B48A8ULL, + 0x391C0CB3C5C95A63ULL, 0x4ED8AA4AE3418ACBULL, + 0x5B9CCA4F7763E373ULL, 0x682E6FF3D6B2B8A3ULL, + 0x748F82EE5DEFB2FCULL, 0x78A5636F43172F60ULL, + 0x84C87814A1F0AB72ULL, 0x8CC702081A6439ECULL, + 0x90BEFFFA23631E28ULL, 0xA4506CEBDE82BDE9ULL, + 0xBEF9A3F7B2C67915ULL, 0xC67178F2E372532BULL, + 0xCA273ECEEA26619CULL, 0xD186B8C721C0C207ULL, + 0xEADA7DD6CDE0EB1EULL, 0xF57D4F7FEE6ED178ULL, + 0x06F067AA72176FBAULL, 0x0A637DC5A2C898A6ULL, + 0x113F9804BEF90DAEULL, 0x1B710B35131C471BULL, + 0x28DB77F523047D84ULL, 0x32CAAB7B40C72493ULL, + 0x3C9EBE0A15C9BEBCULL, 0x431D67C49C100D4CULL, + 0x4CC5D4BECB3E42B6ULL, 0x597F299CFC657E2AULL, + 0x5FCB6FAB3AD6FAECULL, 0x6C44198C4A475817ULL +}; + static void SHA256Transform(uint32_t *H, const uint8_t *cp) { uint32_t a, b, c, d, e, f, g, h, t, T1, T2, W[64]; - for (t = 0; t < 16; t++, cp += 4) + /* copy chunk into the first 16 words of the message schedule */ + for (t = 0; t < 16; t++, cp += sizeof (uint32_t)) W[t] = (cp[0] << 24) | (cp[1] << 16) | (cp[2] << 8) | cp[3]; + /* extend the first 16 words into the remaining 48 words */ for (t = 16; t < 64; t++) - W[t] = sigma1(W[t - 2]) + W[t - 7] + - sigma0(W[t - 15]) + W[t - 16]; + W[t] = SIGMA1_256(W[t - 2]) + W[t - 7] + + SIGMA0_256(W[t - 15]) + W[t - 16]; + /* init working variables to the current hash value */ a = H[0]; b = H[1]; c = H[2]; d = H[3]; e = H[4]; f = H[5]; g = H[6]; h = H[7]; + /* iterate the compression function for all rounds of the hash */ for (t = 0; t < 64; t++) { - T1 = h + SIGMA1(e) + Ch(e, f, g) + SHA256_K[t] + W[t]; - T2 = SIGMA0(a) + Maj(a, b, c); + T1 = h + BIGSIGMA1_256(e) + Ch(e, f, g) + SHA256_K[t] + W[t]; + T2 = BIGSIGMA0_256(a) + Maj(a, b, c); h = g; g = f; f = e; e = d + T1; d = c; c = b; b = a; a = T1 + T2; } + /* add the compressed chunk to the current hash value */ H[0] += a; H[1] += b; H[2] += c; H[3] += d; H[4] += e; H[5] += f; H[6] += g; H[7] += h; } static void -zio_checksum_SHA256(const void *buf, uint64_t size, zio_cksum_t *zcp) +SHA512Transform(uint64_t *H, const uint8_t *cp) { - uint32_t H[8] = { 0x6a09e667, 0xbb67ae85, 0x3c6ef372, 0xa54ff53a, - 0x510e527f, 0x9b05688c, 0x1f83d9ab, 0x5be0cd19 }; + uint64_t a, b, c, d, e, f, g, h, t, T1, T2, W[80]; + + /* copy chunk into the first 16 words of the message schedule */ + for (t = 0; t < 16; t++, cp += sizeof (uint64_t)) + W[t] = ((uint64_t)cp[0] << 56) | ((uint64_t)cp[1] << 48) | + ((uint64_t)cp[2] << 40) | ((uint64_t)cp[3] << 32) | + ((uint64_t)cp[4] << 24) | ((uint64_t)cp[5] << 16) | + ((uint64_t)cp[6] << 8) | (uint64_t)cp[7]; + + /* extend the first 16 words into the remaining 64 words */ + for (t = 16; t < 80; t++) + W[t] = SIGMA1_512(W[t - 2]) + W[t - 7] + + SIGMA0_512(W[t - 15]) + W[t - 16]; + + /* init working variables to the current hash value */ + a = H[0]; b = H[1]; c = H[2]; d = H[3]; + e = H[4]; f = H[5]; g = H[6]; h = H[7]; + + /* iterate the compression function for all rounds of the hash */ + for (t = 0; t < 80; t++) { + T1 = h + BIGSIGMA1_512(e) + Ch(e, f, g) + SHA512_K[t] + W[t]; + T2 = BIGSIGMA0_512(a) + Maj(a, b, c); + h = g; g = f; f = e; e = d + T1; + d = c; c = b; b = a; a = T1 + T2; + } + + /* add the compressed chunk to the current hash value */ + H[0] += a; H[1] += b; H[2] += c; H[3] += d; + H[4] += e; H[5] += f; H[6] += g; H[7] += h; +} + +/* + * Implements the SHA-224 and SHA-256 hash algos - to select between them + * pass the appropriate initial values of 'H' and truncate the last 32 bits + * in case of SHA-224. + */ +static void +SHA256(uint32_t *H, const void *buf, uint64_t size, zio_cksum_t *zcp) +{ uint8_t pad[128]; - int padsize = size & 63; - int i; + unsigned padsize = size & 63; + unsigned i, k; + /* process all blocks up to the last one */ for (i = 0; i < size - padsize; i += 64) SHA256Transform(H, (uint8_t *)buf + i); - for (i = 0; i < padsize; i++) - pad[i] = ((uint8_t *)buf)[i]; + /* process the last block and padding */ + for (k = 0; k < padsize; k++) + pad[k] = ((uint8_t *)buf)[k+i]; for (pad[padsize++] = 0x80; (padsize & 63) != 56; padsize++) pad[padsize] = 0; for (i = 0; i < 8; i++) pad[padsize++] = (size << 3) >> (56 - 8 * i); for (i = 0; i < padsize; i += 64) SHA256Transform(H, pad + i); ZIO_SET_CHECKSUM(zcp, (uint64_t)H[0] << 32 | H[1], (uint64_t)H[2] << 32 | H[3], (uint64_t)H[4] << 32 | H[5], (uint64_t)H[6] << 32 | H[7]); +} + +/* + * encode 64bit data in big-endian format. + */ +static void +Encode64(uint8_t *output, uint64_t *input, size_t len) +{ + size_t i, j; + for (i = 0, j = 0; j < len; i++, j += 8) { + output[j] = (input[i] >> 56) & 0xff; + output[j + 1] = (input[i] >> 48) & 0xff; + output[j + 2] = (input[i] >> 40) & 0xff; + output[j + 3] = (input[i] >> 32) & 0xff; + output[j + 4] = (input[i] >> 24) & 0xff; + output[j + 5] = (input[i] >> 16) & 0xff; + output[j + 6] = (input[i] >> 8) & 0xff; + output[j + 7] = input[i] & 0xff; + } +} + +/* + * Implements the SHA-384, SHA-512 and SHA-512/t hash algos - to select + * between them pass the appropriate initial values for 'H'. The output + * of this function is truncated to the first 256 bits that fit into 'zcp'. + */ +static void +SHA512(uint64_t *H, const void *buf, uint64_t size, zio_cksum_t *zcp) +{ + uint64_t c64[2]; + uint8_t pad[256]; + unsigned padsize = size & 127; + unsigned i, k; + + /* process all blocks up to the last one */ + for (i = 0; i < size - padsize; i += 128) + SHA512Transform(H, (uint8_t *)buf + i); + + /* process the last block and padding */ + for (k = 0; k < padsize; k++) + pad[k] = ((uint8_t *)buf)[k+i]; + + if (padsize < 112) { + for (pad[padsize++] = 0x80; padsize < 112; padsize++) + pad[padsize] = 0; + } else { + for (pad[padsize++] = 0x80; padsize < 240; padsize++) + pad[padsize] = 0; + } + + c64[0] = 0; + c64[1] = size << 3; + Encode64(pad+padsize, c64, sizeof (c64)); + padsize += sizeof (c64); + + for (i = 0; i < padsize; i += 128) + SHA512Transform(H, pad + i); + + /* truncate the output to the first 256 bits which fit into 'zcp' */ + Encode64((uint8_t *)zcp, H, sizeof (uint64_t) * 4); +} + +static void +zio_checksum_SHA256(const void *buf, uint64_t size, + const void *ctx_template __unused, zio_cksum_t *zcp) +{ + /* SHA-256 as per FIPS 180-4. */ + uint32_t H[] = { + 0x6a09e667, 0xbb67ae85, 0x3c6ef372, 0xa54ff53a, + 0x510e527f, 0x9b05688c, 0x1f83d9ab, 0x5be0cd19 + }; + SHA256(H, buf, size, zcp); +} + +static void +zio_checksum_SHA512_native(const void *buf, uint64_t size, + const void *ctx_template __unused, zio_cksum_t *zcp) +{ + /* SHA-512/256 as per FIPS 180-4. */ + uint64_t H[] = { + 0x22312194FC2BF72CULL, 0x9F555FA3C84C64C2ULL, + 0x2393B86B6F53B151ULL, 0x963877195940EABDULL, + 0x96283EE2A88EFFE3ULL, 0xBE5E1E2553863992ULL, + 0x2B0199FC2C85B8AAULL, 0x0EB72DDC81C52CA2ULL + }; + SHA512(H, buf, size, zcp); +} + +static void +zio_checksum_SHA512_byteswap(const void *buf, uint64_t size, + const void *ctx_template, zio_cksum_t *zcp) +{ + zio_cksum_t tmp; + + zio_checksum_SHA512_native(buf, size, ctx_template, &tmp); + zcp->zc_word[0] = BSWAP_64(tmp.zc_word[0]); + zcp->zc_word[1] = BSWAP_64(tmp.zc_word[1]); + zcp->zc_word[2] = BSWAP_64(tmp.zc_word[2]); + zcp->zc_word[3] = BSWAP_64(tmp.zc_word[3]); } Index: stable/11/sys/cddl/boot/zfs/skein_zfs.c =================================================================== --- stable/11/sys/cddl/boot/zfs/skein_zfs.c (nonexistent) +++ stable/11/sys/cddl/boot/zfs/skein_zfs.c (revision 328866) @@ -0,0 +1,92 @@ +/* + * 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://opensource.org/licenses/CDDL-1.0. + * 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 + * + * $FreeBSD$ + */ +/* + * Copyright 2013 Saso Kiselkov. All rights reserved. + */ +#include + +/* + * Computes a native 256-bit skein MAC checksum. Please note that this + * function requires the presence of a ctx_template that should be allocated + * using zio_checksum_skein_tmpl_init. + */ +/*ARGSUSED*/ +static void +zio_checksum_skein_native(const void *buf, uint64_t size, + const void *ctx_template, zio_cksum_t *zcp) +{ + Skein_512_Ctxt_t ctx; + + ASSERT(ctx_template != NULL); + bcopy(ctx_template, &ctx, sizeof (ctx)); + (void) Skein_512_Update(&ctx, buf, size); + (void) Skein_512_Final(&ctx, (uint8_t *)zcp); + bzero(&ctx, sizeof (ctx)); +} + +/* + * Byteswapped version of zio_checksum_skein_native. This just invokes + * the native checksum function and byteswaps the resulting checksum (since + * skein is internally endian-insensitive). + */ +static void +zio_checksum_skein_byteswap(const void *buf, uint64_t size, + const void *ctx_template, zio_cksum_t *zcp) +{ + zio_cksum_t tmp; + + zio_checksum_skein_native(buf, size, ctx_template, &tmp); + zcp->zc_word[0] = BSWAP_64(tmp.zc_word[0]); + zcp->zc_word[1] = BSWAP_64(tmp.zc_word[1]); + zcp->zc_word[2] = BSWAP_64(tmp.zc_word[2]); + zcp->zc_word[3] = BSWAP_64(tmp.zc_word[3]); +} + +/* + * Allocates a skein MAC template suitable for using in skein MAC checksum + * computations and returns a pointer to it. + */ +static void * +zio_checksum_skein_tmpl_init(const zio_cksum_salt_t *salt) +{ + Skein_512_Ctxt_t *ctx; + + ctx = malloc(sizeof (*ctx)); + bzero(ctx, sizeof (*ctx)); + (void) Skein_512_InitExt(ctx, sizeof (zio_cksum_t) * 8, 0, + salt->zcs_bytes, sizeof (salt->zcs_bytes)); + return (ctx); +} + +/* + * Frees a skein context template previously allocated using + * zio_checksum_skein_tmpl_init. + */ +static void +zio_checksum_skein_tmpl_free(void *ctx_template) +{ + Skein_512_Ctxt_t *ctx = ctx_template; + + bzero(ctx, sizeof (*ctx)); + free(ctx); +} Property changes on: stable/11/sys/cddl/boot/zfs/skein_zfs.c ___________________________________________________________________ Added: svn:eol-style ## -0,0 +1 ## +native \ No newline at end of property Added: svn:keywords ## -0,0 +1 ## +FreeBSD=%H \ No newline at end of property Added: svn:mime-type ## -0,0 +1 ## +text/plain \ No newline at end of property Index: stable/11/sys/cddl/boot/zfs/zfsimpl.h =================================================================== --- stable/11/sys/cddl/boot/zfs/zfsimpl.h (revision 328865) +++ stable/11/sys/cddl/boot/zfs/zfsimpl.h (revision 328866) @@ -1,1506 +1,1521 @@ /*- * Copyright (c) 2002 McAfee, Inc. * All rights reserved. * * This software was developed for the FreeBSD Project by Marshall * Kirk McKusick and McAfee Research,, the Security Research Division of * McAfee, Inc. under DARPA/SPAWAR contract N66001-01-C-8035 ("CBOSS"), as * part of the DARPA CHATS research program * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ /* * 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 2013 by Saso Kiselkov. All rights reserved. */ /* * Copyright (c) 2013 by Delphix. All rights reserved. */ #define MAXNAMELEN 256 #define _NOTE(s) typedef enum { B_FALSE, B_TRUE } boolean_t; /* CRC64 table */ #define ZFS_CRC64_POLY 0xC96C5795D7870F42ULL /* ECMA-182, reflected form */ /* * Macros for various sorts of alignment and rounding when the alignment * is known to be a power of 2. */ #define P2ALIGN(x, align) ((x) & -(align)) #define P2PHASE(x, align) ((x) & ((align) - 1)) #define P2NPHASE(x, align) (-(x) & ((align) - 1)) #define P2ROUNDUP(x, align) (-(-(x) & -(align))) #define P2END(x, align) (-(~(x) & -(align))) #define P2PHASEUP(x, align, phase) ((phase) - (((phase) - (x)) & -(align))) #define P2BOUNDARY(off, len, align) (((off) ^ ((off) + (len) - 1)) > (align) - 1) /* * 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) \ ((x) ^= BF32_ENCODE((x >> low) ^ (val), low, len)) #define BF64_SET(x, low, len, val) \ ((x) ^= BF64_ENCODE((x >> low) ^ (val), low, len)) #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) \ BF32_SET(x, low, len, ((val) >> (shift)) - (bias)) #define BF64_SET_SB(x, low, len, shift, bias, val) \ BF64_SET(x, low, len, ((val) >> (shift)) - (bias)) /* * Macros to reverse byte order */ #define BSWAP_8(x) ((x) & 0xff) #define BSWAP_16(x) ((BSWAP_8(x) << 8) | BSWAP_8((x) >> 8)) #define BSWAP_32(x) ((BSWAP_16(x) << 16) | BSWAP_16((x) >> 16)) #define BSWAP_64(x) ((BSWAP_32(x) << 32) | BSWAP_32((x) >> 32)) -/* - * Note: the boot loader can't actually read blocks larger than 128KB, - * due to lack of memory. Therefore its SPA_MAXBLOCKSIZE is still 128KB. - */ #define SPA_MINBLOCKSHIFT 9 -#define SPA_MAXBLOCKSHIFT 17 +#define SPA_OLDMAXBLOCKSHIFT 17 +#define SPA_MAXBLOCKSHIFT 24 #define SPA_MINBLOCKSIZE (1ULL << SPA_MINBLOCKSHIFT) +#define SPA_OLDMAXBLOCKSIZE (1ULL << SPA_OLDMAXBLOCKSHIFT) #define SPA_MAXBLOCKSIZE (1ULL << SPA_MAXBLOCKSHIFT) /* * 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; /* + * Some checksums/hashes need a 256-bit initialization salt. This salt is kept + * secret and is suitable for use in MAC algorithms as the key. + */ +typedef struct zio_cksum_salt { + uint8_t zcs_bytes[32]; +} zio_cksum_salt_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 */ 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) \ BF64_GET_SB((bp)->blk_prop, 16, SPA_LSIZEBITS, SPA_MINBLOCKSHIFT, 1) #define BP_SET_PSIZE(bp, x) \ BF64_SET_SB((bp)->blk_prop, 16, SPA_LSIZEBITS, SPA_MINBLOCKSHIFT, 1, x) #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_GET_CHECKSUM(bp) BF64_GET((bp)->blk_prop, 40, 8) #define BP_SET_CHECKSUM(bp, x) BF64_SET((bp)->blk_prop, 40, 8, x) #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_IS_EMBEDDED(bp) BF64_GET((bp)->blk_prop, 39, 1) #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)->blk_phys_birth ? (bp)->blk_phys_birth : (bp)->blk_birth) #define BP_GET_ASIZE(bp) \ (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[BP_GET_TYPE(bp)].ot_metadata) ? \ BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp)); #define BP_GET_NDVAS(bp) \ (!!DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \ !!DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \ !!DVA_GET_ASIZE(&(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 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) (&(bp)->blk_dva[0]) #define BP_IS_GANG(bp) 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) DVA_IS_EMPTY(BP_IDENTITY(bp)) #define BP_IS_OLDER(bp, txg) (!BP_IS_HOLE(bp) && (bp)->blk_birth < (txg)) #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); \ } #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) /* * Embedded checksum */ #define ZEC_MAGIC 0x210da7ab10c7a11ULL typedef struct zio_eck { uint64_t zec_magic; /* for validation, endianness */ zio_cksum_t zec_cksum; /* 256-bit checksum */ } zio_eck_t; /* * Gang block headers are self-checksumming and contain an array * of block pointers. */ #define SPA_GANGBLOCKSIZE SPA_MINBLOCKSIZE #define SPA_GBH_NBLKPTRS ((SPA_GANGBLOCKSIZE - \ sizeof (zio_eck_t)) / sizeof (blkptr_t)) #define SPA_GBH_FILLER ((SPA_GANGBLOCKSIZE - \ sizeof (zio_eck_t) - \ (SPA_GBH_NBLKPTRS * sizeof (blkptr_t))) /\ sizeof (uint64_t)) typedef struct zio_gbh { blkptr_t zg_blkptr[SPA_GBH_NBLKPTRS]; uint64_t zg_filler[SPA_GBH_FILLER]; zio_eck_t zg_tail; } zio_gbh_phys_t; #define VDEV_RAIDZ_MAXPARITY 3 #define VDEV_PAD_SIZE (8 << 10) /* 2 padding areas (vl_pad1 and vl_pad2) to skip */ #define VDEV_SKIP_SIZE VDEV_PAD_SIZE * 2 #define VDEV_PHYS_SIZE (112 << 10) #define VDEV_UBERBLOCK_RING (128 << 10) #define VDEV_UBERBLOCK_SHIFT(vd) \ MAX((vd)->v_top->v_ashift, UBERBLOCK_SHIFT) #define VDEV_UBERBLOCK_COUNT(vd) \ (VDEV_UBERBLOCK_RING >> VDEV_UBERBLOCK_SHIFT(vd)) #define VDEV_UBERBLOCK_OFFSET(vd, n) \ offsetof(vdev_label_t, vl_uberblock[(n) << VDEV_UBERBLOCK_SHIFT(vd)]) #define VDEV_UBERBLOCK_SIZE(vd) (1ULL << VDEV_UBERBLOCK_SHIFT(vd)) typedef struct vdev_phys { char vp_nvlist[VDEV_PHYS_SIZE - sizeof (zio_eck_t)]; zio_eck_t vp_zbt; } vdev_phys_t; typedef struct vdev_label { char vl_pad1[VDEV_PAD_SIZE]; /* 8K */ char vl_pad2[VDEV_PAD_SIZE]; /* 8K */ vdev_phys_t vl_vdev_phys; /* 112K */ char vl_uberblock[VDEV_UBERBLOCK_RING]; /* 128K */ } vdev_label_t; /* 256K total */ /* * vdev_dirty() flags */ #define VDD_METASLAB 0x01 #define VDD_DTL 0x02 /* * Size and offset of embedded boot loader region on each label. * The total size of the first two labels plus the boot area is 4MB. */ #define VDEV_BOOT_OFFSET (2 * sizeof (vdev_label_t)) #define VDEV_BOOT_SIZE (7ULL << 19) /* 3.5M */ /* * Size of label regions at the start and end of each leaf device. */ #define VDEV_LABEL_START_SIZE (2 * sizeof (vdev_label_t) + VDEV_BOOT_SIZE) #define VDEV_LABEL_END_SIZE (2 * sizeof (vdev_label_t)) #define VDEV_LABELS 4 enum zio_checksum { ZIO_CHECKSUM_INHERIT = 0, ZIO_CHECKSUM_ON, ZIO_CHECKSUM_OFF, ZIO_CHECKSUM_LABEL, ZIO_CHECKSUM_GANG_HEADER, ZIO_CHECKSUM_ZILOG, ZIO_CHECKSUM_FLETCHER_2, ZIO_CHECKSUM_FLETCHER_4, ZIO_CHECKSUM_SHA256, ZIO_CHECKSUM_ZILOG2, + ZIO_CHECKSUM_NOPARITY, + ZIO_CHECKSUM_SHA512, + ZIO_CHECKSUM_SKEIN, + ZIO_CHECKSUM_EDONR, ZIO_CHECKSUM_FUNCTIONS }; #define ZIO_CHECKSUM_ON_VALUE ZIO_CHECKSUM_FLETCHER_4 #define ZIO_CHECKSUM_DEFAULT ZIO_CHECKSUM_ON enum zio_compress { ZIO_COMPRESS_INHERIT = 0, ZIO_COMPRESS_ON, ZIO_COMPRESS_OFF, ZIO_COMPRESS_LZJB, ZIO_COMPRESS_EMPTY, ZIO_COMPRESS_GZIP_1, ZIO_COMPRESS_GZIP_2, ZIO_COMPRESS_GZIP_3, ZIO_COMPRESS_GZIP_4, ZIO_COMPRESS_GZIP_5, ZIO_COMPRESS_GZIP_6, ZIO_COMPRESS_GZIP_7, ZIO_COMPRESS_GZIP_8, ZIO_COMPRESS_GZIP_9, ZIO_COMPRESS_ZLE, ZIO_COMPRESS_LZ4, ZIO_COMPRESS_FUNCTIONS }; #define ZIO_COMPRESS_ON_VALUE ZIO_COMPRESS_LZJB #define ZIO_COMPRESS_DEFAULT ZIO_COMPRESS_OFF /* nvlist pack encoding */ #define NV_ENCODE_NATIVE 0 #define NV_ENCODE_XDR 1 typedef enum { DATA_TYPE_UNKNOWN = 0, DATA_TYPE_BOOLEAN, DATA_TYPE_BYTE, DATA_TYPE_INT16, DATA_TYPE_UINT16, DATA_TYPE_INT32, DATA_TYPE_UINT32, DATA_TYPE_INT64, DATA_TYPE_UINT64, DATA_TYPE_STRING, DATA_TYPE_BYTE_ARRAY, DATA_TYPE_INT16_ARRAY, DATA_TYPE_UINT16_ARRAY, DATA_TYPE_INT32_ARRAY, DATA_TYPE_UINT32_ARRAY, DATA_TYPE_INT64_ARRAY, DATA_TYPE_UINT64_ARRAY, DATA_TYPE_STRING_ARRAY, DATA_TYPE_HRTIME, DATA_TYPE_NVLIST, DATA_TYPE_NVLIST_ARRAY, DATA_TYPE_BOOLEAN_VALUE, DATA_TYPE_INT8, DATA_TYPE_UINT8, DATA_TYPE_BOOLEAN_ARRAY, DATA_TYPE_INT8_ARRAY, DATA_TYPE_UINT8_ARRAY } data_type_t; /* * On-disk version number. */ #define SPA_VERSION_1 1ULL #define SPA_VERSION_2 2ULL #define SPA_VERSION_3 3ULL #define SPA_VERSION_4 4ULL #define SPA_VERSION_5 5ULL #define SPA_VERSION_6 6ULL #define SPA_VERSION_7 7ULL #define SPA_VERSION_8 8ULL #define SPA_VERSION_9 9ULL #define SPA_VERSION_10 10ULL #define SPA_VERSION_11 11ULL #define SPA_VERSION_12 12ULL #define SPA_VERSION_13 13ULL #define SPA_VERSION_14 14ULL #define SPA_VERSION_15 15ULL #define SPA_VERSION_16 16ULL #define SPA_VERSION_17 17ULL #define SPA_VERSION_18 18ULL #define SPA_VERSION_19 19ULL #define SPA_VERSION_20 20ULL #define SPA_VERSION_21 21ULL #define SPA_VERSION_22 22ULL #define SPA_VERSION_23 23ULL #define SPA_VERSION_24 24ULL #define SPA_VERSION_25 25ULL #define SPA_VERSION_26 26ULL #define SPA_VERSION_27 27ULL #define SPA_VERSION_28 28ULL #define SPA_VERSION_5000 5000ULL /* * When bumping up SPA_VERSION, make sure GRUB ZFS understands the on-disk * format change. Go to usr/src/grub/grub-0.97/stage2/{zfs-include/, fsys_zfs*}, * and do the appropriate changes. Also bump the version number in * usr/src/grub/capability. */ #define SPA_VERSION SPA_VERSION_5000 #define SPA_VERSION_STRING "5000" /* * Symbolic names for the changes that caused a SPA_VERSION switch. * Used in the code when checking for presence or absence of a feature. * Feel free to define multiple symbolic names for each version if there * were multiple changes to on-disk structures during that version. * * NOTE: When checking the current SPA_VERSION in your code, be sure * to use spa_version() since it reports the version of the * last synced uberblock. Checking the in-flight version can * be dangerous in some cases. */ #define SPA_VERSION_INITIAL SPA_VERSION_1 #define SPA_VERSION_DITTO_BLOCKS SPA_VERSION_2 #define SPA_VERSION_SPARES SPA_VERSION_3 #define SPA_VERSION_RAID6 SPA_VERSION_3 #define SPA_VERSION_BPLIST_ACCOUNT SPA_VERSION_3 #define SPA_VERSION_RAIDZ_DEFLATE SPA_VERSION_3 #define SPA_VERSION_DNODE_BYTES SPA_VERSION_3 #define SPA_VERSION_ZPOOL_HISTORY SPA_VERSION_4 #define SPA_VERSION_GZIP_COMPRESSION SPA_VERSION_5 #define SPA_VERSION_BOOTFS SPA_VERSION_6 #define SPA_VERSION_SLOGS SPA_VERSION_7 #define SPA_VERSION_DELEGATED_PERMS SPA_VERSION_8 #define SPA_VERSION_FUID SPA_VERSION_9 #define SPA_VERSION_REFRESERVATION SPA_VERSION_9 #define SPA_VERSION_REFQUOTA SPA_VERSION_9 #define SPA_VERSION_UNIQUE_ACCURATE SPA_VERSION_9 #define SPA_VERSION_L2CACHE SPA_VERSION_10 #define SPA_VERSION_NEXT_CLONES SPA_VERSION_11 #define SPA_VERSION_ORIGIN SPA_VERSION_11 #define SPA_VERSION_DSL_SCRUB SPA_VERSION_11 #define SPA_VERSION_SNAP_PROPS SPA_VERSION_12 #define SPA_VERSION_USED_BREAKDOWN SPA_VERSION_13 #define SPA_VERSION_PASSTHROUGH_X SPA_VERSION_14 #define SPA_VERSION_USERSPACE SPA_VERSION_15 #define SPA_VERSION_STMF_PROP SPA_VERSION_16 #define SPA_VERSION_RAIDZ3 SPA_VERSION_17 #define SPA_VERSION_USERREFS SPA_VERSION_18 #define SPA_VERSION_HOLES SPA_VERSION_19 #define SPA_VERSION_ZLE_COMPRESSION SPA_VERSION_20 #define SPA_VERSION_DEDUP SPA_VERSION_21 #define SPA_VERSION_RECVD_PROPS SPA_VERSION_22 #define SPA_VERSION_SLIM_ZIL SPA_VERSION_23 #define SPA_VERSION_SA SPA_VERSION_24 #define SPA_VERSION_SCAN SPA_VERSION_25 #define SPA_VERSION_DIR_CLONES SPA_VERSION_26 #define SPA_VERSION_DEADLISTS SPA_VERSION_26 #define SPA_VERSION_FAST_SNAP SPA_VERSION_27 #define SPA_VERSION_MULTI_REPLACE SPA_VERSION_28 #define SPA_VERSION_BEFORE_FEATURES SPA_VERSION_28 #define SPA_VERSION_FEATURES SPA_VERSION_5000 #define SPA_VERSION_IS_SUPPORTED(v) \ (((v) >= SPA_VERSION_INITIAL && (v) <= SPA_VERSION_BEFORE_FEATURES) || \ ((v) >= SPA_VERSION_FEATURES && (v) <= SPA_VERSION)) /* * The following are configuration names used in the nvlist describing a pool's * configuration. */ #define ZPOOL_CONFIG_VERSION "version" #define ZPOOL_CONFIG_POOL_NAME "name" #define ZPOOL_CONFIG_POOL_STATE "state" #define ZPOOL_CONFIG_POOL_TXG "txg" #define ZPOOL_CONFIG_POOL_GUID "pool_guid" #define ZPOOL_CONFIG_CREATE_TXG "create_txg" #define ZPOOL_CONFIG_TOP_GUID "top_guid" #define ZPOOL_CONFIG_VDEV_TREE "vdev_tree" #define ZPOOL_CONFIG_TYPE "type" #define ZPOOL_CONFIG_CHILDREN "children" #define ZPOOL_CONFIG_ID "id" #define ZPOOL_CONFIG_GUID "guid" #define ZPOOL_CONFIG_PATH "path" #define ZPOOL_CONFIG_DEVID "devid" #define ZPOOL_CONFIG_METASLAB_ARRAY "metaslab_array" #define ZPOOL_CONFIG_METASLAB_SHIFT "metaslab_shift" #define ZPOOL_CONFIG_ASHIFT "ashift" #define ZPOOL_CONFIG_ASIZE "asize" #define ZPOOL_CONFIG_DTL "DTL" #define ZPOOL_CONFIG_STATS "stats" #define ZPOOL_CONFIG_WHOLE_DISK "whole_disk" #define ZPOOL_CONFIG_ERRCOUNT "error_count" #define ZPOOL_CONFIG_NOT_PRESENT "not_present" #define ZPOOL_CONFIG_SPARES "spares" #define ZPOOL_CONFIG_IS_SPARE "is_spare" #define ZPOOL_CONFIG_NPARITY "nparity" #define ZPOOL_CONFIG_HOSTID "hostid" #define ZPOOL_CONFIG_HOSTNAME "hostname" #define ZPOOL_CONFIG_IS_LOG "is_log" #define ZPOOL_CONFIG_TIMESTAMP "timestamp" /* not stored on disk */ #define ZPOOL_CONFIG_FEATURES_FOR_READ "features_for_read" /* * The persistent vdev state is stored as separate values rather than a single * 'vdev_state' entry. This is because a device can be in multiple states, such * as offline and degraded. */ #define ZPOOL_CONFIG_OFFLINE "offline" #define ZPOOL_CONFIG_FAULTED "faulted" #define ZPOOL_CONFIG_DEGRADED "degraded" #define ZPOOL_CONFIG_REMOVED "removed" #define ZPOOL_CONFIG_FRU "fru" #define ZPOOL_CONFIG_AUX_STATE "aux_state" #define VDEV_TYPE_ROOT "root" #define VDEV_TYPE_MIRROR "mirror" #define VDEV_TYPE_REPLACING "replacing" #define VDEV_TYPE_RAIDZ "raidz" #define VDEV_TYPE_DISK "disk" #define VDEV_TYPE_FILE "file" #define VDEV_TYPE_MISSING "missing" #define VDEV_TYPE_HOLE "hole" #define VDEV_TYPE_SPARE "spare" #define VDEV_TYPE_LOG "log" #define VDEV_TYPE_L2CACHE "l2cache" /* * This is needed in userland to report the minimum necessary device size. */ #define SPA_MINDEVSIZE (64ULL << 20) /* * The location of the pool configuration repository, shared between kernel and * userland. */ #define ZPOOL_CACHE "/boot/zfs/zpool.cache" /* * vdev states are ordered from least to most healthy. * A vdev that's CANT_OPEN or below is considered unusable. */ typedef enum vdev_state { VDEV_STATE_UNKNOWN = 0, /* Uninitialized vdev */ VDEV_STATE_CLOSED, /* Not currently open */ VDEV_STATE_OFFLINE, /* Not allowed to open */ VDEV_STATE_REMOVED, /* Explicitly removed from system */ VDEV_STATE_CANT_OPEN, /* Tried to open, but failed */ VDEV_STATE_FAULTED, /* External request to fault device */ VDEV_STATE_DEGRADED, /* Replicated vdev with unhealthy kids */ VDEV_STATE_HEALTHY /* Presumed good */ } vdev_state_t; /* * vdev aux states. When a vdev is in the CANT_OPEN state, the aux field * of the vdev stats structure uses these constants to distinguish why. */ typedef enum vdev_aux { VDEV_AUX_NONE, /* no error */ VDEV_AUX_OPEN_FAILED, /* ldi_open_*() or vn_open() failed */ VDEV_AUX_CORRUPT_DATA, /* bad label or disk contents */ VDEV_AUX_NO_REPLICAS, /* insufficient number of replicas */ VDEV_AUX_BAD_GUID_SUM, /* vdev guid sum doesn't match */ VDEV_AUX_TOO_SMALL, /* vdev size is too small */ VDEV_AUX_BAD_LABEL, /* the label is OK but invalid */ VDEV_AUX_VERSION_NEWER, /* on-disk version is too new */ VDEV_AUX_VERSION_OLDER, /* on-disk version is too old */ VDEV_AUX_SPARED /* hot spare used in another pool */ } vdev_aux_t; /* * pool state. The following states are written to disk as part of the normal * SPA lifecycle: ACTIVE, EXPORTED, DESTROYED, SPARE. The remaining states are * software abstractions used at various levels to communicate pool state. */ typedef enum pool_state { POOL_STATE_ACTIVE = 0, /* In active use */ POOL_STATE_EXPORTED, /* Explicitly exported */ POOL_STATE_DESTROYED, /* Explicitly destroyed */ POOL_STATE_SPARE, /* Reserved for hot spare use */ POOL_STATE_UNINITIALIZED, /* Internal spa_t state */ POOL_STATE_UNAVAIL, /* Internal libzfs state */ POOL_STATE_POTENTIALLY_ACTIVE /* Internal libzfs state */ } pool_state_t; /* * The uberblock version is incremented whenever an incompatible on-disk * format change is made to the SPA, DMU, or ZAP. * * Note: the first two fields should never be moved. When a storage pool * is opened, the uberblock must be read off the disk before the version * can be checked. If the ub_version field is moved, we may not detect * version mismatch. If the ub_magic field is moved, applications that * expect the magic number in the first word won't work. */ #define UBERBLOCK_MAGIC 0x00bab10c /* oo-ba-bloc! */ #define UBERBLOCK_SHIFT 10 /* up to 1K */ struct uberblock { uint64_t ub_magic; /* UBERBLOCK_MAGIC */ uint64_t ub_version; /* SPA_VERSION */ uint64_t ub_txg; /* txg of last sync */ uint64_t ub_guid_sum; /* sum of all vdev guids */ uint64_t ub_timestamp; /* UTC time of last sync */ blkptr_t ub_rootbp; /* MOS objset_phys_t */ }; /* * Flags. */ #define DNODE_MUST_BE_ALLOCATED 1 #define DNODE_MUST_BE_FREE 2 /* * Fixed constants. */ #define DNODE_SHIFT 9 /* 512 bytes */ #define DN_MIN_INDBLKSHIFT 12 /* 4k */ #define DN_MAX_INDBLKSHIFT 14 /* 16k */ #define DNODE_BLOCK_SHIFT 14 /* 16k */ #define DNODE_CORE_SIZE 64 /* 64 bytes for dnode sans blkptrs */ #define DN_MAX_OBJECT_SHIFT 48 /* 256 trillion (zfs_fid_t limit) */ #define DN_MAX_OFFSET_SHIFT 64 /* 2^64 bytes in a dnode */ /* * Derived constants. */ #define DNODE_SIZE (1 << DNODE_SHIFT) #define DN_MAX_NBLKPTR ((DNODE_SIZE - DNODE_CORE_SIZE) >> SPA_BLKPTRSHIFT) #define DN_MAX_BONUSLEN (DNODE_SIZE - DNODE_CORE_SIZE - (1 << SPA_BLKPTRSHIFT)) #define DN_MAX_OBJECT (1ULL << DN_MAX_OBJECT_SHIFT) #define DNODES_PER_BLOCK_SHIFT (DNODE_BLOCK_SHIFT - DNODE_SHIFT) #define DNODES_PER_BLOCK (1ULL << DNODES_PER_BLOCK_SHIFT) #define DNODES_PER_LEVEL_SHIFT (DN_MAX_INDBLKSHIFT - SPA_BLKPTRSHIFT) /* The +2 here is a cheesy way to round up */ #define DN_MAX_LEVELS (2 + ((DN_MAX_OFFSET_SHIFT - SPA_MINBLOCKSHIFT) / \ (DN_MIN_INDBLKSHIFT - SPA_BLKPTRSHIFT))) #define DN_BONUS(dnp) ((void*)((dnp)->dn_bonus + \ (((dnp)->dn_nblkptr - 1) * sizeof (blkptr_t)))) #define DN_USED_BYTES(dnp) (((dnp)->dn_flags & DNODE_FLAG_USED_BYTES) ? \ (dnp)->dn_used : (dnp)->dn_used << SPA_MINBLOCKSHIFT) #define EPB(blkshift, typeshift) (1 << (blkshift - typeshift)) /* Is dn_used in bytes? if not, it's in multiples of SPA_MINBLOCKSIZE */ #define DNODE_FLAG_USED_BYTES (1<<0) #define DNODE_FLAG_USERUSED_ACCOUNTED (1<<1) /* Does dnode have a SA spill blkptr in bonus? */ #define DNODE_FLAG_SPILL_BLKPTR (1<<2) typedef struct dnode_phys { uint8_t dn_type; /* dmu_object_type_t */ uint8_t dn_indblkshift; /* ln2(indirect block size) */ uint8_t dn_nlevels; /* 1=dn_blkptr->data blocks */ uint8_t dn_nblkptr; /* length of dn_blkptr */ uint8_t dn_bonustype; /* type of data in bonus buffer */ uint8_t dn_checksum; /* ZIO_CHECKSUM type */ uint8_t dn_compress; /* ZIO_COMPRESS type */ uint8_t dn_flags; /* DNODE_FLAG_* */ uint16_t dn_datablkszsec; /* data block size in 512b sectors */ uint16_t dn_bonuslen; /* length of dn_bonus */ uint8_t dn_pad2[4]; /* accounting is protected by dn_dirty_mtx */ uint64_t dn_maxblkid; /* largest allocated block ID */ uint64_t dn_used; /* bytes (or sectors) of disk space */ uint64_t dn_pad3[4]; blkptr_t dn_blkptr[1]; uint8_t dn_bonus[DN_MAX_BONUSLEN - sizeof (blkptr_t)]; blkptr_t dn_spill; } dnode_phys_t; typedef enum dmu_object_byteswap { DMU_BSWAP_UINT8, DMU_BSWAP_UINT16, DMU_BSWAP_UINT32, DMU_BSWAP_UINT64, DMU_BSWAP_ZAP, DMU_BSWAP_DNODE, DMU_BSWAP_OBJSET, DMU_BSWAP_ZNODE, DMU_BSWAP_OLDACL, DMU_BSWAP_ACL, /* * Allocating a new byteswap type number makes the on-disk format * incompatible with any other format that uses the same number. * * Data can usually be structured to work with one of the * DMU_BSWAP_UINT* or DMU_BSWAP_ZAP types. */ DMU_BSWAP_NUMFUNCS } dmu_object_byteswap_t; #define DMU_OT_NEWTYPE 0x80 #define DMU_OT_METADATA 0x40 #define DMU_OT_BYTESWAP_MASK 0x3f /* * Defines a uint8_t object type. Object types specify if the data * in the object is metadata (boolean) and how to byteswap the data * (dmu_object_byteswap_t). */ #define DMU_OT(byteswap, metadata) \ (DMU_OT_NEWTYPE | \ ((metadata) ? DMU_OT_METADATA : 0) | \ ((byteswap) & DMU_OT_BYTESWAP_MASK)) typedef enum dmu_object_type { DMU_OT_NONE, /* general: */ DMU_OT_OBJECT_DIRECTORY, /* ZAP */ DMU_OT_OBJECT_ARRAY, /* UINT64 */ DMU_OT_PACKED_NVLIST, /* UINT8 (XDR by nvlist_pack/unpack) */ DMU_OT_PACKED_NVLIST_SIZE, /* UINT64 */ DMU_OT_BPLIST, /* UINT64 */ DMU_OT_BPLIST_HDR, /* UINT64 */ /* spa: */ DMU_OT_SPACE_MAP_HEADER, /* UINT64 */ DMU_OT_SPACE_MAP, /* UINT64 */ /* zil: */ DMU_OT_INTENT_LOG, /* UINT64 */ /* dmu: */ DMU_OT_DNODE, /* DNODE */ DMU_OT_OBJSET, /* OBJSET */ /* dsl: */ DMU_OT_DSL_DIR, /* UINT64 */ DMU_OT_DSL_DIR_CHILD_MAP, /* ZAP */ DMU_OT_DSL_DS_SNAP_MAP, /* ZAP */ DMU_OT_DSL_PROPS, /* ZAP */ DMU_OT_DSL_DATASET, /* UINT64 */ /* zpl: */ DMU_OT_ZNODE, /* ZNODE */ DMU_OT_OLDACL, /* Old ACL */ DMU_OT_PLAIN_FILE_CONTENTS, /* UINT8 */ DMU_OT_DIRECTORY_CONTENTS, /* ZAP */ DMU_OT_MASTER_NODE, /* ZAP */ DMU_OT_UNLINKED_SET, /* ZAP */ /* zvol: */ DMU_OT_ZVOL, /* UINT8 */ DMU_OT_ZVOL_PROP, /* ZAP */ /* other; for testing only! */ DMU_OT_PLAIN_OTHER, /* UINT8 */ DMU_OT_UINT64_OTHER, /* UINT64 */ DMU_OT_ZAP_OTHER, /* ZAP */ /* new object types: */ DMU_OT_ERROR_LOG, /* ZAP */ DMU_OT_SPA_HISTORY, /* UINT8 */ DMU_OT_SPA_HISTORY_OFFSETS, /* spa_his_phys_t */ DMU_OT_POOL_PROPS, /* ZAP */ DMU_OT_DSL_PERMS, /* ZAP */ DMU_OT_ACL, /* ACL */ DMU_OT_SYSACL, /* SYSACL */ DMU_OT_FUID, /* FUID table (Packed NVLIST UINT8) */ DMU_OT_FUID_SIZE, /* FUID table size UINT64 */ DMU_OT_NEXT_CLONES, /* ZAP */ DMU_OT_SCAN_QUEUE, /* ZAP */ DMU_OT_USERGROUP_USED, /* ZAP */ DMU_OT_USERGROUP_QUOTA, /* ZAP */ DMU_OT_USERREFS, /* ZAP */ DMU_OT_DDT_ZAP, /* ZAP */ DMU_OT_DDT_STATS, /* ZAP */ DMU_OT_SA, /* System attr */ DMU_OT_SA_MASTER_NODE, /* ZAP */ DMU_OT_SA_ATTR_REGISTRATION, /* ZAP */ DMU_OT_SA_ATTR_LAYOUTS, /* ZAP */ DMU_OT_SCAN_XLATE, /* ZAP */ DMU_OT_DEDUP, /* fake dedup BP from ddt_bp_create() */ DMU_OT_NUMTYPES, /* * Names for valid types declared with DMU_OT(). */ DMU_OTN_UINT8_DATA = DMU_OT(DMU_BSWAP_UINT8, B_FALSE), DMU_OTN_UINT8_METADATA = DMU_OT(DMU_BSWAP_UINT8, B_TRUE), DMU_OTN_UINT16_DATA = DMU_OT(DMU_BSWAP_UINT16, B_FALSE), DMU_OTN_UINT16_METADATA = DMU_OT(DMU_BSWAP_UINT16, B_TRUE), DMU_OTN_UINT32_DATA = DMU_OT(DMU_BSWAP_UINT32, B_FALSE), DMU_OTN_UINT32_METADATA = DMU_OT(DMU_BSWAP_UINT32, B_TRUE), DMU_OTN_UINT64_DATA = DMU_OT(DMU_BSWAP_UINT64, B_FALSE), DMU_OTN_UINT64_METADATA = DMU_OT(DMU_BSWAP_UINT64, B_TRUE), DMU_OTN_ZAP_DATA = DMU_OT(DMU_BSWAP_ZAP, B_FALSE), DMU_OTN_ZAP_METADATA = DMU_OT(DMU_BSWAP_ZAP, B_TRUE) } dmu_object_type_t; typedef enum dmu_objset_type { DMU_OST_NONE, DMU_OST_META, DMU_OST_ZFS, DMU_OST_ZVOL, DMU_OST_OTHER, /* For testing only! */ DMU_OST_ANY, /* Be careful! */ DMU_OST_NUMTYPES } dmu_objset_type_t; /* * header for all bonus and spill buffers. * The header has a fixed portion with a variable number * of "lengths" depending on the number of variable sized * attribues which are determined by the "layout number" */ #define SA_MAGIC 0x2F505A /* ZFS SA */ typedef struct sa_hdr_phys { uint32_t sa_magic; uint16_t sa_layout_info; /* Encoded with hdrsize and layout number */ uint16_t sa_lengths[1]; /* optional sizes for variable length attrs */ /* ... Data follows the lengths. */ } sa_hdr_phys_t; /* * sa_hdr_phys -> sa_layout_info * * 16 10 0 * +--------+-------+ * | hdrsz |layout | * +--------+-------+ * * Bits 0-10 are the layout number * Bits 11-16 are the size of the header. * The hdrsize is the number * 8 * * For example. * hdrsz of 1 ==> 8 byte header * 2 ==> 16 byte header * */ #define SA_HDR_LAYOUT_NUM(hdr) BF32_GET(hdr->sa_layout_info, 0, 10) #define SA_HDR_SIZE(hdr) BF32_GET_SB(hdr->sa_layout_info, 10, 16, 3, 0) #define SA_HDR_LAYOUT_INFO_ENCODE(x, num, size) \ { \ BF32_SET_SB(x, 10, 6, 3, 0, size); \ BF32_SET(x, 0, 10, num); \ } #define SA_MODE_OFFSET 0 #define SA_SIZE_OFFSET 8 #define SA_GEN_OFFSET 16 #define SA_UID_OFFSET 24 #define SA_GID_OFFSET 32 #define SA_PARENT_OFFSET 40 #define SA_SYMLINK_OFFSET 160 /* * Intent log header - this on disk structure holds fields to manage * the log. All fields are 64 bit to easily handle cross architectures. */ typedef struct zil_header { uint64_t zh_claim_txg; /* txg in which log blocks were claimed */ uint64_t zh_replay_seq; /* highest replayed sequence number */ blkptr_t zh_log; /* log chain */ uint64_t zh_claim_seq; /* highest claimed sequence number */ uint64_t zh_pad[5]; } zil_header_t; #define OBJSET_PHYS_SIZE 2048 typedef struct objset_phys { dnode_phys_t os_meta_dnode; zil_header_t os_zil_header; uint64_t os_type; uint64_t os_flags; char os_pad[OBJSET_PHYS_SIZE - sizeof (dnode_phys_t)*3 - sizeof (zil_header_t) - sizeof (uint64_t)*2]; dnode_phys_t os_userused_dnode; dnode_phys_t os_groupused_dnode; } objset_phys_t; typedef struct dsl_dir_phys { uint64_t dd_creation_time; /* not actually used */ uint64_t dd_head_dataset_obj; uint64_t dd_parent_obj; uint64_t dd_clone_parent_obj; uint64_t dd_child_dir_zapobj; /* * how much space our children are accounting for; for leaf * datasets, == physical space used by fs + snaps */ uint64_t dd_used_bytes; uint64_t dd_compressed_bytes; uint64_t dd_uncompressed_bytes; /* Administrative quota setting */ uint64_t dd_quota; /* Administrative reservation setting */ uint64_t dd_reserved; uint64_t dd_props_zapobj; uint64_t dd_pad[21]; /* pad out to 256 bytes for good measure */ } dsl_dir_phys_t; typedef struct dsl_dataset_phys { uint64_t ds_dir_obj; uint64_t ds_prev_snap_obj; uint64_t ds_prev_snap_txg; uint64_t ds_next_snap_obj; uint64_t ds_snapnames_zapobj; /* zap obj of snaps; ==0 for snaps */ uint64_t ds_num_children; /* clone/snap children; ==0 for head */ uint64_t ds_creation_time; /* seconds since 1970 */ uint64_t ds_creation_txg; uint64_t ds_deadlist_obj; uint64_t ds_used_bytes; uint64_t ds_compressed_bytes; uint64_t ds_uncompressed_bytes; uint64_t ds_unique_bytes; /* only relevant to snapshots */ /* * The ds_fsid_guid is a 56-bit ID that can change to avoid * collisions. The ds_guid is a 64-bit ID that will never * change, so there is a small probability that it will collide. */ uint64_t ds_fsid_guid; uint64_t ds_guid; uint64_t ds_flags; blkptr_t ds_bp; uint64_t ds_pad[8]; /* pad out to 320 bytes for good measure */ } dsl_dataset_phys_t; /* * The names of zap entries in the DIRECTORY_OBJECT of the MOS. */ #define DMU_POOL_DIRECTORY_OBJECT 1 #define DMU_POOL_CONFIG "config" #define DMU_POOL_FEATURES_FOR_READ "features_for_read" #define DMU_POOL_ROOT_DATASET "root_dataset" #define DMU_POOL_SYNC_BPLIST "sync_bplist" #define DMU_POOL_ERRLOG_SCRUB "errlog_scrub" #define DMU_POOL_ERRLOG_LAST "errlog_last" #define DMU_POOL_SPARES "spares" #define DMU_POOL_DEFLATE "deflate" #define DMU_POOL_HISTORY "history" #define DMU_POOL_PROPS "pool_props" +#define DMU_POOL_CHECKSUM_SALT "org.illumos:checksum_salt" #define ZAP_MAGIC 0x2F52AB2ABULL #define FZAP_BLOCK_SHIFT(zap) ((zap)->zap_block_shift) #define ZAP_MAXCD (uint32_t)(-1) #define ZAP_HASHBITS 28 #define MZAP_ENT_LEN 64 #define MZAP_NAME_LEN (MZAP_ENT_LEN - 8 - 4 - 2) #define MZAP_MAX_BLKSHIFT SPA_MAXBLOCKSHIFT #define MZAP_MAX_BLKSZ (1 << MZAP_MAX_BLKSHIFT) typedef struct mzap_ent_phys { uint64_t mze_value; uint32_t mze_cd; uint16_t mze_pad; /* in case we want to chain them someday */ char mze_name[MZAP_NAME_LEN]; } mzap_ent_phys_t; typedef struct mzap_phys { uint64_t mz_block_type; /* ZBT_MICRO */ uint64_t mz_salt; uint64_t mz_pad[6]; mzap_ent_phys_t mz_chunk[1]; /* actually variable size depending on block size */ } mzap_phys_t; /* * The (fat) zap is stored in one object. It is an array of * 1<= 6] [zap_leaf_t] [ptrtbl] ... * */ #define ZBT_LEAF ((1ULL << 63) + 0) #define ZBT_HEADER ((1ULL << 63) + 1) #define ZBT_MICRO ((1ULL << 63) + 3) /* any other values are ptrtbl blocks */ /* * the embedded pointer table takes up half a block: * block size / entry size (2^3) / 2 */ #define ZAP_EMBEDDED_PTRTBL_SHIFT(zap) (FZAP_BLOCK_SHIFT(zap) - 3 - 1) /* * The embedded pointer table starts half-way through the block. Since * the pointer table itself is half the block, it starts at (64-bit) * word number (1<zap_phys) \ [(idx) + (1<l_bs) - hash entry size (2) * number of hash * entries - header space (2*chunksize) */ #define ZAP_LEAF_NUMCHUNKS(l) \ (((1<<(l)->l_bs) - 2*ZAP_LEAF_HASH_NUMENTRIES(l)) / \ ZAP_LEAF_CHUNKSIZE - 2) /* * The amount of space within the chunk available for the array is: * chunk size - space for type (1) - space for next pointer (2) */ #define ZAP_LEAF_ARRAY_BYTES (ZAP_LEAF_CHUNKSIZE - 3) #define ZAP_LEAF_ARRAY_NCHUNKS(bytes) \ (((bytes)+ZAP_LEAF_ARRAY_BYTES-1)/ZAP_LEAF_ARRAY_BYTES) /* * Low water mark: when there are only this many chunks free, start * growing the ptrtbl. Ideally, this should be larger than a * "reasonably-sized" entry. 20 chunks is more than enough for the * largest directory entry (MAXNAMELEN (256) byte name, 8-byte value), * while still being only around 3% for 16k blocks. */ #define ZAP_LEAF_LOW_WATER (20) /* * The leaf hash table has block size / 2^5 (32) number of entries, * which should be more than enough for the maximum number of entries, * which is less than block size / CHUNKSIZE (24) / minimum number of * chunks per entry (3). */ #define ZAP_LEAF_HASH_SHIFT(l) ((l)->l_bs - 5) #define ZAP_LEAF_HASH_NUMENTRIES(l) (1 << ZAP_LEAF_HASH_SHIFT(l)) /* * The chunks start immediately after the hash table. The end of the * hash table is at l_hash + HASH_NUMENTRIES, which we simply cast to a * chunk_t. */ #define ZAP_LEAF_CHUNK(l, idx) \ ((zap_leaf_chunk_t *) \ ((l)->l_phys->l_hash + ZAP_LEAF_HASH_NUMENTRIES(l)))[idx] #define ZAP_LEAF_ENTRY(l, idx) (&ZAP_LEAF_CHUNK(l, idx).l_entry) typedef enum zap_chunk_type { ZAP_CHUNK_FREE = 253, ZAP_CHUNK_ENTRY = 252, ZAP_CHUNK_ARRAY = 251, ZAP_CHUNK_TYPE_MAX = 250 } zap_chunk_type_t; /* * TAKE NOTE: * If zap_leaf_phys_t is modified, zap_leaf_byteswap() must be modified. */ typedef struct zap_leaf_phys { struct zap_leaf_header { uint64_t lh_block_type; /* ZBT_LEAF */ uint64_t lh_pad1; uint64_t lh_prefix; /* hash prefix of this leaf */ uint32_t lh_magic; /* ZAP_LEAF_MAGIC */ uint16_t lh_nfree; /* number free chunks */ uint16_t lh_nentries; /* number of entries */ uint16_t lh_prefix_len; /* num bits used to id this */ /* above is accessable to zap, below is zap_leaf private */ uint16_t lh_freelist; /* chunk head of free list */ uint8_t lh_pad2[12]; } l_hdr; /* 2 24-byte chunks */ /* * The header is followed by a hash table with * ZAP_LEAF_HASH_NUMENTRIES(zap) entries. The hash table is * followed by an array of ZAP_LEAF_NUMCHUNKS(zap) * zap_leaf_chunk structures. These structures are accessed * with the ZAP_LEAF_CHUNK() macro. */ uint16_t l_hash[1]; } zap_leaf_phys_t; typedef union zap_leaf_chunk { struct zap_leaf_entry { uint8_t le_type; /* always ZAP_CHUNK_ENTRY */ uint8_t le_value_intlen; /* size of ints */ uint16_t le_next; /* next entry in hash chain */ uint16_t le_name_chunk; /* first chunk of the name */ uint16_t le_name_numints; /* bytes in name, incl null */ uint16_t le_value_chunk; /* first chunk of the value */ uint16_t le_value_numints; /* value length in ints */ uint32_t le_cd; /* collision differentiator */ uint64_t le_hash; /* hash value of the name */ } l_entry; struct zap_leaf_array { uint8_t la_type; /* always ZAP_CHUNK_ARRAY */ uint8_t la_array[ZAP_LEAF_ARRAY_BYTES]; uint16_t la_next; /* next blk or CHAIN_END */ } l_array; struct zap_leaf_free { uint8_t lf_type; /* always ZAP_CHUNK_FREE */ uint8_t lf_pad[ZAP_LEAF_ARRAY_BYTES]; uint16_t lf_next; /* next in free list, or CHAIN_END */ } l_free; } zap_leaf_chunk_t; typedef struct zap_leaf { int l_bs; /* block size shift */ zap_leaf_phys_t *l_phys; } zap_leaf_t; /* * Define special zfs pflags */ #define ZFS_XATTR 0x1 /* is an extended attribute */ #define ZFS_INHERIT_ACE 0x2 /* ace has inheritable ACEs */ #define ZFS_ACL_TRIVIAL 0x4 /* files ACL is trivial */ #define MASTER_NODE_OBJ 1 /* * special attributes for master node. */ #define ZFS_FSID "FSID" #define ZFS_UNLINKED_SET "DELETE_QUEUE" #define ZFS_ROOT_OBJ "ROOT" #define ZPL_VERSION_OBJ "VERSION" #define ZFS_PROP_BLOCKPERPAGE "BLOCKPERPAGE" #define ZFS_PROP_NOGROWBLOCKS "NOGROWBLOCKS" #define ZFS_FLAG_BLOCKPERPAGE 0x1 #define ZFS_FLAG_NOGROWBLOCKS 0x2 /* * ZPL version - rev'd whenever an incompatible on-disk format change * occurs. Independent of SPA/DMU/ZAP versioning. */ #define ZPL_VERSION 1ULL /* * The directory entry has the type (currently unused on Solaris) in the * top 4 bits, and the object number in the low 48 bits. The "middle" * 12 bits are unused. */ #define ZFS_DIRENT_TYPE(de) BF64_GET(de, 60, 4) #define ZFS_DIRENT_OBJ(de) BF64_GET(de, 0, 48) #define ZFS_DIRENT_MAKE(type, obj) (((uint64_t)type << 60) | obj) typedef struct ace { uid_t a_who; /* uid or gid */ uint32_t a_access_mask; /* read,write,... */ uint16_t a_flags; /* see below */ uint16_t a_type; /* allow or deny */ } ace_t; #define ACE_SLOT_CNT 6 typedef struct zfs_znode_acl { uint64_t z_acl_extern_obj; /* ext acl pieces */ uint32_t z_acl_count; /* Number of ACEs */ uint16_t z_acl_version; /* acl version */ uint16_t z_acl_pad; /* pad */ ace_t z_ace_data[ACE_SLOT_CNT]; /* 6 standard ACEs */ } zfs_znode_acl_t; /* * This is the persistent portion of the znode. It is stored * in the "bonus buffer" of the file. Short symbolic links * are also stored in the bonus buffer. */ typedef struct znode_phys { uint64_t zp_atime[2]; /* 0 - last file access time */ uint64_t zp_mtime[2]; /* 16 - last file modification time */ uint64_t zp_ctime[2]; /* 32 - last file change time */ uint64_t zp_crtime[2]; /* 48 - creation time */ uint64_t zp_gen; /* 64 - generation (txg of creation) */ uint64_t zp_mode; /* 72 - file mode bits */ uint64_t zp_size; /* 80 - size of file */ uint64_t zp_parent; /* 88 - directory parent (`..') */ uint64_t zp_links; /* 96 - number of links to file */ uint64_t zp_xattr; /* 104 - DMU object for xattrs */ uint64_t zp_rdev; /* 112 - dev_t for VBLK & VCHR files */ uint64_t zp_flags; /* 120 - persistent flags */ uint64_t zp_uid; /* 128 - file owner */ uint64_t zp_gid; /* 136 - owning group */ uint64_t zp_pad[4]; /* 144 - future */ zfs_znode_acl_t zp_acl; /* 176 - 263 ACL */ /* * Data may pad out any remaining bytes in the znode buffer, eg: * * |<---------------------- dnode_phys (512) ------------------------>| * |<-- dnode (192) --->|<----------- "bonus" buffer (320) ---------->| * |<---- znode (264) ---->|<---- data (56) ---->| * * At present, we only use this space to store symbolic links. */ } znode_phys_t; /* * In-core vdev representation. */ struct vdev; +struct spa; typedef int vdev_phys_read_t(struct vdev *vdev, void *priv, off_t offset, void *buf, size_t bytes); typedef int vdev_read_t(struct vdev *vdev, const blkptr_t *bp, void *buf, off_t offset, size_t bytes); typedef STAILQ_HEAD(vdev_list, vdev) vdev_list_t; typedef struct vdev { STAILQ_ENTRY(vdev) v_childlink; /* link in parent's child list */ STAILQ_ENTRY(vdev) v_alllink; /* link in global vdev list */ vdev_list_t v_children; /* children of this vdev */ const char *v_name; /* vdev name */ uint64_t v_guid; /* vdev guid */ int v_id; /* index in parent */ int v_ashift; /* offset to block shift */ int v_nparity; /* # parity for raidz */ struct vdev *v_top; /* parent vdev */ int v_nchildren; /* # children */ vdev_state_t v_state; /* current state */ vdev_phys_read_t *v_phys_read; /* read from raw leaf vdev */ vdev_read_t *v_read; /* read from vdev */ void *v_read_priv; /* private data for read function */ + struct spa *spa; /* link to spa */ } vdev_t; /* * In-core pool representation. */ typedef STAILQ_HEAD(spa_list, spa) spa_list_t; typedef struct spa { STAILQ_ENTRY(spa) spa_link; /* link in global pool list */ char *spa_name; /* pool name */ uint64_t spa_guid; /* pool guid */ uint64_t spa_txg; /* most recent transaction */ struct uberblock spa_uberblock; /* best uberblock so far */ vdev_list_t spa_vdevs; /* list of all toplevel vdevs */ objset_phys_t spa_mos; /* MOS for this pool */ + zio_cksum_salt_t spa_cksum_salt; /* secret salt for cksum */ + void *spa_cksum_tmpls[ZIO_CHECKSUM_FUNCTIONS]; int spa_inited; /* initialized */ } spa_t; static void decode_embedded_bp_compressed(const blkptr_t *, void *); Index: stable/11/sys/cddl/boot/zfs/zfssubr.c =================================================================== --- stable/11/sys/cddl/boot/zfs/zfssubr.c (revision 328865) +++ stable/11/sys/cddl/boot/zfs/zfssubr.c (revision 328866) @@ -1,1703 +1,1787 @@ /* * 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 2007 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ #include __FBSDID("$FreeBSD$"); static uint64_t zfs_crc64_table[256]; #define ECKSUM 666 #define ASSERT3S(x, y, z) ((void)0) #define ASSERT3U(x, y, z) ((void)0) #define ASSERT3P(x, y, z) ((void)0) #define ASSERT0(x) ((void)0) #define ASSERT(x) ((void)0) #define panic(...) do { \ printf(__VA_ARGS__); \ for (;;) ; \ } while (0) #define kmem_alloc(size, flag) zfs_alloc((size)) #define kmem_free(ptr, size) zfs_free((ptr), (size)) static void zfs_init_crc(void) { int i, j; uint64_t *ct; /* * Calculate the crc64 table (used for the zap hash * function). */ if (zfs_crc64_table[128] != ZFS_CRC64_POLY) { memset(zfs_crc64_table, 0, sizeof(zfs_crc64_table)); for (i = 0; i < 256; i++) for (ct = zfs_crc64_table + i, *ct = i, j = 8; j > 0; j--) *ct = (*ct >> 1) ^ (-(*ct & 1) & ZFS_CRC64_POLY); } } static void -zio_checksum_off(const void *buf, uint64_t size, zio_cksum_t *zcp) +zio_checksum_off(const void *buf, uint64_t size, + const void *ctx_template, zio_cksum_t *zcp) { ZIO_SET_CHECKSUM(zcp, 0, 0, 0, 0); } /* * Signature for checksum functions. */ -typedef void zio_checksum_t(const void *data, uint64_t size, zio_cksum_t *zcp); +typedef void zio_checksum_t(const void *data, uint64_t size, + const void *ctx_template, zio_cksum_t *zcp); +typedef void *zio_checksum_tmpl_init_t(const zio_cksum_salt_t *salt); +typedef void zio_checksum_tmpl_free_t(void *ctx_template); +typedef enum zio_checksum_flags { + /* Strong enough for metadata? */ + ZCHECKSUM_FLAG_METADATA = (1 << 1), + /* ZIO embedded checksum */ + ZCHECKSUM_FLAG_EMBEDDED = (1 << 2), + /* Strong enough for dedup (without verification)? */ + ZCHECKSUM_FLAG_DEDUP = (1 << 3), + /* Uses salt value */ + ZCHECKSUM_FLAG_SALTED = (1 << 4), + /* Strong enough for nopwrite? */ + ZCHECKSUM_FLAG_NOPWRITE = (1 << 5) +} zio_checksum_flags_t; + /* * Information about each checksum function. */ typedef struct zio_checksum_info { - zio_checksum_t *ci_func[2]; /* checksum function for each byteorder */ - int ci_correctable; /* number of correctable bits */ - int ci_eck; /* uses zio embedded checksum? */ - int ci_dedup; /* strong enough for dedup? */ - const char *ci_name; /* descriptive name */ + /* checksum function for each byteorder */ + zio_checksum_t *ci_func[2]; + zio_checksum_tmpl_init_t *ci_tmpl_init; + zio_checksum_tmpl_free_t *ci_tmpl_free; + zio_checksum_flags_t ci_flags; + const char *ci_name; /* descriptive name */ } zio_checksum_info_t; #include "blkptr.c" #include "fletcher.c" #include "sha256.c" +#include "skein_zfs.c" static zio_checksum_info_t zio_checksum_table[ZIO_CHECKSUM_FUNCTIONS] = { - {{NULL, NULL}, 0, 0, 0, "inherit"}, - {{NULL, NULL}, 0, 0, 0, "on"}, - {{zio_checksum_off, zio_checksum_off}, 0, 0, 0, "off"}, - {{zio_checksum_SHA256, zio_checksum_SHA256}, 1, 1, 0, "label"}, - {{zio_checksum_SHA256, zio_checksum_SHA256}, 1, 1, 0, "gang_header"}, - {{fletcher_2_native, fletcher_2_byteswap}, 0, 1, 0, "zilog"}, - {{fletcher_2_native, fletcher_2_byteswap}, 0, 0, 0, "fletcher2"}, - {{fletcher_4_native, fletcher_4_byteswap}, 1, 0, 0, "fletcher4"}, - {{zio_checksum_SHA256, zio_checksum_SHA256}, 1, 0, 1, "SHA256"}, - {{fletcher_4_native, fletcher_4_byteswap}, 0, 1, 0, "zillog2"}, + {{NULL, NULL}, NULL, NULL, 0, "inherit"}, + {{NULL, NULL}, NULL, NULL, 0, "on"}, + {{zio_checksum_off, zio_checksum_off}, NULL, NULL, 0, "off"}, + {{zio_checksum_SHA256, zio_checksum_SHA256}, NULL, NULL, + ZCHECKSUM_FLAG_METADATA | ZCHECKSUM_FLAG_EMBEDDED, "label"}, + {{zio_checksum_SHA256, zio_checksum_SHA256}, NULL, NULL, + ZCHECKSUM_FLAG_METADATA | ZCHECKSUM_FLAG_EMBEDDED, "gang_header"}, + {{fletcher_2_native, fletcher_2_byteswap}, NULL, NULL, + ZCHECKSUM_FLAG_EMBEDDED, "zilog"}, + {{fletcher_2_native, fletcher_2_byteswap}, NULL, NULL, + 0, "fletcher2"}, + {{fletcher_4_native, fletcher_4_byteswap}, NULL, NULL, + ZCHECKSUM_FLAG_METADATA, "fletcher4"}, + {{zio_checksum_SHA256, zio_checksum_SHA256}, NULL, NULL, + ZCHECKSUM_FLAG_METADATA | ZCHECKSUM_FLAG_DEDUP | + ZCHECKSUM_FLAG_NOPWRITE, "SHA256"}, + {{fletcher_4_native, fletcher_4_byteswap}, NULL, NULL, + ZCHECKSUM_FLAG_EMBEDDED, "zillog2"}, + {{zio_checksum_off, zio_checksum_off}, NULL, NULL, + 0, "noparity"}, + {{zio_checksum_SHA512_native, zio_checksum_SHA512_byteswap}, + NULL, NULL, ZCHECKSUM_FLAG_METADATA | ZCHECKSUM_FLAG_DEDUP | + ZCHECKSUM_FLAG_NOPWRITE, "SHA512"}, + {{zio_checksum_skein_native, zio_checksum_skein_byteswap}, + zio_checksum_skein_tmpl_init, zio_checksum_skein_tmpl_free, + ZCHECKSUM_FLAG_METADATA | ZCHECKSUM_FLAG_DEDUP | + ZCHECKSUM_FLAG_SALTED | ZCHECKSUM_FLAG_NOPWRITE, "skein"}, + /* no edonr for now */ + {{NULL, NULL}, NULL, NULL, ZCHECKSUM_FLAG_METADATA | + ZCHECKSUM_FLAG_SALTED | ZCHECKSUM_FLAG_NOPWRITE, "edonr"} }; - /* * Common signature for all zio compress/decompress functions. */ typedef size_t zio_compress_func_t(void *src, void *dst, size_t s_len, size_t d_len, int); typedef int zio_decompress_func_t(void *src, void *dst, size_t s_len, size_t d_len, int); /* * Information about each compression function. */ typedef struct zio_compress_info { zio_compress_func_t *ci_compress; /* compression function */ zio_decompress_func_t *ci_decompress; /* decompression function */ int ci_level; /* level parameter */ const char *ci_name; /* algorithm name */ } zio_compress_info_t; #include "lzjb.c" #include "zle.c" #include "lz4.c" /* * Compression vectors. */ static zio_compress_info_t zio_compress_table[ZIO_COMPRESS_FUNCTIONS] = { {NULL, NULL, 0, "inherit"}, {NULL, NULL, 0, "on"}, {NULL, NULL, 0, "uncompressed"}, {NULL, lzjb_decompress, 0, "lzjb"}, {NULL, NULL, 0, "empty"}, {NULL, NULL, 1, "gzip-1"}, {NULL, NULL, 2, "gzip-2"}, {NULL, NULL, 3, "gzip-3"}, {NULL, NULL, 4, "gzip-4"}, {NULL, NULL, 5, "gzip-5"}, {NULL, NULL, 6, "gzip-6"}, {NULL, NULL, 7, "gzip-7"}, {NULL, NULL, 8, "gzip-8"}, {NULL, NULL, 9, "gzip-9"}, {NULL, zle_decompress, 64, "zle"}, {NULL, lz4_decompress, 0, "lz4"}, }; static 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]); } /* * Set the external verifier for a gang block based on , * a tuple which is guaranteed to be unique for the life of the pool. */ static void zio_checksum_gang_verifier(zio_cksum_t *zcp, const blkptr_t *bp) { const dva_t *dva = BP_IDENTITY(bp); uint64_t txg = BP_PHYSICAL_BIRTH(bp); ASSERT(BP_IS_GANG(bp)); ZIO_SET_CHECKSUM(zcp, DVA_GET_VDEV(dva), DVA_GET_OFFSET(dva), txg, 0); } /* * Set the external verifier for a label block based on its offset. * The vdev is implicit, and the txg is unknowable at pool open time -- * hence the logic in vdev_uberblock_load() to find the most recent copy. */ static void zio_checksum_label_verifier(zio_cksum_t *zcp, uint64_t offset) { ZIO_SET_CHECKSUM(zcp, offset, 0, 0, 0); } +/* + * Calls the template init function of a checksum which supports context + * templates and installs the template into the spa_t. + */ +static void +zio_checksum_template_init(enum zio_checksum checksum, spa_t *spa) +{ + zio_checksum_info_t *ci = &zio_checksum_table[checksum]; + + if (ci->ci_tmpl_init == NULL) + return; + + if (spa->spa_cksum_tmpls[checksum] != NULL) + return; + + if (spa->spa_cksum_tmpls[checksum] == NULL) { + spa->spa_cksum_tmpls[checksum] = + ci->ci_tmpl_init(&spa->spa_cksum_salt); + } +} + +/* + * Called by a spa_t that's about to be deallocated. This steps through + * all of the checksum context templates and deallocates any that were + * initialized using the algorithm-specific template init function. + */ +void +zio_checksum_templates_free(spa_t *spa) +{ + for (enum zio_checksum checksum = 0; + checksum < ZIO_CHECKSUM_FUNCTIONS; checksum++) { + if (spa->spa_cksum_tmpls[checksum] != NULL) { + zio_checksum_info_t *ci = &zio_checksum_table[checksum]; + + ci->ci_tmpl_free(spa->spa_cksum_tmpls[checksum]); + spa->spa_cksum_tmpls[checksum] = NULL; + } + } +} + static int -zio_checksum_verify(const blkptr_t *bp, void *data) +zio_checksum_verify(const spa_t *spa, const blkptr_t *bp, void *data) { uint64_t size; unsigned int checksum; zio_checksum_info_t *ci; + void *ctx = NULL; zio_cksum_t actual_cksum, expected_cksum, verifier; int byteswap; checksum = BP_GET_CHECKSUM(bp); size = BP_GET_PSIZE(bp); if (checksum >= ZIO_CHECKSUM_FUNCTIONS) return (EINVAL); ci = &zio_checksum_table[checksum]; if (ci->ci_func[0] == NULL || ci->ci_func[1] == NULL) return (EINVAL); - if (ci->ci_eck) { + if (spa != NULL) { + zio_checksum_template_init(checksum, (spa_t *) spa); + ctx = spa->spa_cksum_tmpls[checksum]; + } + + if (ci->ci_flags & ZCHECKSUM_FLAG_EMBEDDED) { zio_eck_t *eck; ASSERT(checksum == ZIO_CHECKSUM_GANG_HEADER || checksum == ZIO_CHECKSUM_LABEL); eck = (zio_eck_t *)((char *)data + size) - 1; if (checksum == ZIO_CHECKSUM_GANG_HEADER) zio_checksum_gang_verifier(&verifier, bp); else if (checksum == ZIO_CHECKSUM_LABEL) zio_checksum_label_verifier(&verifier, DVA_GET_OFFSET(BP_IDENTITY(bp))); else verifier = bp->blk_cksum; byteswap = (eck->zec_magic == BSWAP_64(ZEC_MAGIC)); if (byteswap) byteswap_uint64_array(&verifier, sizeof (zio_cksum_t)); expected_cksum = eck->zec_cksum; eck->zec_cksum = verifier; - ci->ci_func[byteswap](data, size, &actual_cksum); + ci->ci_func[byteswap](data, size, ctx, &actual_cksum); eck->zec_cksum = expected_cksum; if (byteswap) byteswap_uint64_array(&expected_cksum, sizeof (zio_cksum_t)); } else { expected_cksum = bp->blk_cksum; - ci->ci_func[0](data, size, &actual_cksum); + ci->ci_func[0](data, size, ctx, &actual_cksum); } if (!ZIO_CHECKSUM_EQUAL(actual_cksum, expected_cksum)) { - /*printf("ZFS: read checksum failed\n");*/ + /*printf("ZFS: read checksum %s failed\n", ci->ci_name);*/ return (EIO); } return (0); } static int zio_decompress_data(int cpfunc, void *src, uint64_t srcsize, void *dest, uint64_t destsize) { zio_compress_info_t *ci; if (cpfunc >= ZIO_COMPRESS_FUNCTIONS) { printf("ZFS: unsupported compression algorithm %u\n", cpfunc); return (EIO); } ci = &zio_compress_table[cpfunc]; if (!ci->ci_decompress) { printf("ZFS: unsupported compression algorithm %s\n", ci->ci_name); return (EIO); } return (ci->ci_decompress(src, dest, srcsize, destsize, ci->ci_level)); } static uint64_t zap_hash(uint64_t salt, const char *name) { const uint8_t *cp; uint8_t c; uint64_t crc = salt; ASSERT(crc != 0); ASSERT(zfs_crc64_table[128] == ZFS_CRC64_POLY); for (cp = (const uint8_t *)name; (c = *cp) != '\0'; cp++) crc = (crc >> 8) ^ zfs_crc64_table[(crc ^ c) & 0xFF]; /* * Only use 28 bits, since we need 4 bits in the cookie for the * collision differentiator. We MUST use the high bits, since * those are the onces that we first pay attention to when * chosing the bucket. */ crc &= ~((1ULL << (64 - ZAP_HASHBITS)) - 1); return (crc); } static void *zfs_alloc(size_t size); static void zfs_free(void *ptr, size_t size); typedef struct raidz_col { uint64_t rc_devidx; /* child device index for I/O */ uint64_t rc_offset; /* device offset */ uint64_t rc_size; /* I/O size */ void *rc_data; /* I/O data */ int rc_error; /* I/O error for this device */ uint8_t rc_tried; /* Did we attempt this I/O column? */ uint8_t rc_skipped; /* Did we skip this I/O column? */ } raidz_col_t; typedef struct raidz_map { uint64_t rm_cols; /* Regular column count */ uint64_t rm_scols; /* Count including skipped columns */ uint64_t rm_bigcols; /* Number of oversized columns */ uint64_t rm_asize; /* Actual total I/O size */ uint64_t rm_missingdata; /* Count of missing data devices */ uint64_t rm_missingparity; /* Count of missing parity devices */ uint64_t rm_firstdatacol; /* First data column/parity count */ uint64_t rm_nskip; /* Skipped sectors for padding */ uint64_t rm_skipstart; /* Column index of padding start */ uintptr_t rm_reports; /* # of referencing checksum reports */ uint8_t rm_freed; /* map no longer has referencing ZIO */ uint8_t rm_ecksuminjected; /* checksum error was injected */ raidz_col_t rm_col[1]; /* Flexible array of I/O columns */ } raidz_map_t; #define VDEV_RAIDZ_P 0 #define VDEV_RAIDZ_Q 1 #define VDEV_RAIDZ_R 2 #define VDEV_RAIDZ_MUL_2(x) (((x) << 1) ^ (((x) & 0x80) ? 0x1d : 0)) #define VDEV_RAIDZ_MUL_4(x) (VDEV_RAIDZ_MUL_2(VDEV_RAIDZ_MUL_2(x))) /* * We provide a mechanism to perform the field multiplication operation on a * 64-bit value all at once rather than a byte at a time. This works by * creating a mask from the top bit in each byte and using that to * conditionally apply the XOR of 0x1d. */ #define VDEV_RAIDZ_64MUL_2(x, mask) \ { \ (mask) = (x) & 0x8080808080808080ULL; \ (mask) = ((mask) << 1) - ((mask) >> 7); \ (x) = (((x) << 1) & 0xfefefefefefefefeULL) ^ \ ((mask) & 0x1d1d1d1d1d1d1d1dULL); \ } #define VDEV_RAIDZ_64MUL_4(x, mask) \ { \ VDEV_RAIDZ_64MUL_2((x), mask); \ VDEV_RAIDZ_64MUL_2((x), mask); \ } /* * These two tables represent powers and logs of 2 in the Galois field defined * above. These values were computed by repeatedly multiplying by 2 as above. */ static const uint8_t vdev_raidz_pow2[256] = { 0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80, 0x1d, 0x3a, 0x74, 0xe8, 0xcd, 0x87, 0x13, 0x26, 0x4c, 0x98, 0x2d, 0x5a, 0xb4, 0x75, 0xea, 0xc9, 0x8f, 0x03, 0x06, 0x0c, 0x18, 0x30, 0x60, 0xc0, 0x9d, 0x27, 0x4e, 0x9c, 0x25, 0x4a, 0x94, 0x35, 0x6a, 0xd4, 0xb5, 0x77, 0xee, 0xc1, 0x9f, 0x23, 0x46, 0x8c, 0x05, 0x0a, 0x14, 0x28, 0x50, 0xa0, 0x5d, 0xba, 0x69, 0xd2, 0xb9, 0x6f, 0xde, 0xa1, 0x5f, 0xbe, 0x61, 0xc2, 0x99, 0x2f, 0x5e, 0xbc, 0x65, 0xca, 0x89, 0x0f, 0x1e, 0x3c, 0x78, 0xf0, 0xfd, 0xe7, 0xd3, 0xbb, 0x6b, 0xd6, 0xb1, 0x7f, 0xfe, 0xe1, 0xdf, 0xa3, 0x5b, 0xb6, 0x71, 0xe2, 0xd9, 0xaf, 0x43, 0x86, 0x11, 0x22, 0x44, 0x88, 0x0d, 0x1a, 0x34, 0x68, 0xd0, 0xbd, 0x67, 0xce, 0x81, 0x1f, 0x3e, 0x7c, 0xf8, 0xed, 0xc7, 0x93, 0x3b, 0x76, 0xec, 0xc5, 0x97, 0x33, 0x66, 0xcc, 0x85, 0x17, 0x2e, 0x5c, 0xb8, 0x6d, 0xda, 0xa9, 0x4f, 0x9e, 0x21, 0x42, 0x84, 0x15, 0x2a, 0x54, 0xa8, 0x4d, 0x9a, 0x29, 0x52, 0xa4, 0x55, 0xaa, 0x49, 0x92, 0x39, 0x72, 0xe4, 0xd5, 0xb7, 0x73, 0xe6, 0xd1, 0xbf, 0x63, 0xc6, 0x91, 0x3f, 0x7e, 0xfc, 0xe5, 0xd7, 0xb3, 0x7b, 0xf6, 0xf1, 0xff, 0xe3, 0xdb, 0xab, 0x4b, 0x96, 0x31, 0x62, 0xc4, 0x95, 0x37, 0x6e, 0xdc, 0xa5, 0x57, 0xae, 0x41, 0x82, 0x19, 0x32, 0x64, 0xc8, 0x8d, 0x07, 0x0e, 0x1c, 0x38, 0x70, 0xe0, 0xdd, 0xa7, 0x53, 0xa6, 0x51, 0xa2, 0x59, 0xb2, 0x79, 0xf2, 0xf9, 0xef, 0xc3, 0x9b, 0x2b, 0x56, 0xac, 0x45, 0x8a, 0x09, 0x12, 0x24, 0x48, 0x90, 0x3d, 0x7a, 0xf4, 0xf5, 0xf7, 0xf3, 0xfb, 0xeb, 0xcb, 0x8b, 0x0b, 0x16, 0x2c, 0x58, 0xb0, 0x7d, 0xfa, 0xe9, 0xcf, 0x83, 0x1b, 0x36, 0x6c, 0xd8, 0xad, 0x47, 0x8e, 0x01 }; static const uint8_t vdev_raidz_log2[256] = { 0x00, 0x00, 0x01, 0x19, 0x02, 0x32, 0x1a, 0xc6, 0x03, 0xdf, 0x33, 0xee, 0x1b, 0x68, 0xc7, 0x4b, 0x04, 0x64, 0xe0, 0x0e, 0x34, 0x8d, 0xef, 0x81, 0x1c, 0xc1, 0x69, 0xf8, 0xc8, 0x08, 0x4c, 0x71, 0x05, 0x8a, 0x65, 0x2f, 0xe1, 0x24, 0x0f, 0x21, 0x35, 0x93, 0x8e, 0xda, 0xf0, 0x12, 0x82, 0x45, 0x1d, 0xb5, 0xc2, 0x7d, 0x6a, 0x27, 0xf9, 0xb9, 0xc9, 0x9a, 0x09, 0x78, 0x4d, 0xe4, 0x72, 0xa6, 0x06, 0xbf, 0x8b, 0x62, 0x66, 0xdd, 0x30, 0xfd, 0xe2, 0x98, 0x25, 0xb3, 0x10, 0x91, 0x22, 0x88, 0x36, 0xd0, 0x94, 0xce, 0x8f, 0x96, 0xdb, 0xbd, 0xf1, 0xd2, 0x13, 0x5c, 0x83, 0x38, 0x46, 0x40, 0x1e, 0x42, 0xb6, 0xa3, 0xc3, 0x48, 0x7e, 0x6e, 0x6b, 0x3a, 0x28, 0x54, 0xfa, 0x85, 0xba, 0x3d, 0xca, 0x5e, 0x9b, 0x9f, 0x0a, 0x15, 0x79, 0x2b, 0x4e, 0xd4, 0xe5, 0xac, 0x73, 0xf3, 0xa7, 0x57, 0x07, 0x70, 0xc0, 0xf7, 0x8c, 0x80, 0x63, 0x0d, 0x67, 0x4a, 0xde, 0xed, 0x31, 0xc5, 0xfe, 0x18, 0xe3, 0xa5, 0x99, 0x77, 0x26, 0xb8, 0xb4, 0x7c, 0x11, 0x44, 0x92, 0xd9, 0x23, 0x20, 0x89, 0x2e, 0x37, 0x3f, 0xd1, 0x5b, 0x95, 0xbc, 0xcf, 0xcd, 0x90, 0x87, 0x97, 0xb2, 0xdc, 0xfc, 0xbe, 0x61, 0xf2, 0x56, 0xd3, 0xab, 0x14, 0x2a, 0x5d, 0x9e, 0x84, 0x3c, 0x39, 0x53, 0x47, 0x6d, 0x41, 0xa2, 0x1f, 0x2d, 0x43, 0xd8, 0xb7, 0x7b, 0xa4, 0x76, 0xc4, 0x17, 0x49, 0xec, 0x7f, 0x0c, 0x6f, 0xf6, 0x6c, 0xa1, 0x3b, 0x52, 0x29, 0x9d, 0x55, 0xaa, 0xfb, 0x60, 0x86, 0xb1, 0xbb, 0xcc, 0x3e, 0x5a, 0xcb, 0x59, 0x5f, 0xb0, 0x9c, 0xa9, 0xa0, 0x51, 0x0b, 0xf5, 0x16, 0xeb, 0x7a, 0x75, 0x2c, 0xd7, 0x4f, 0xae, 0xd5, 0xe9, 0xe6, 0xe7, 0xad, 0xe8, 0x74, 0xd6, 0xf4, 0xea, 0xa8, 0x50, 0x58, 0xaf, }; /* * Multiply a given number by 2 raised to the given power. */ static uint8_t vdev_raidz_exp2(uint8_t a, int exp) { if (a == 0) return (0); ASSERT(exp >= 0); ASSERT(vdev_raidz_log2[a] > 0 || a == 1); exp += vdev_raidz_log2[a]; if (exp > 255) exp -= 255; return (vdev_raidz_pow2[exp]); } static void vdev_raidz_generate_parity_p(raidz_map_t *rm) { uint64_t *p, *src, pcount, ccount, i; int c; pcount = rm->rm_col[VDEV_RAIDZ_P].rc_size / sizeof (src[0]); for (c = rm->rm_firstdatacol; c < rm->rm_cols; c++) { src = rm->rm_col[c].rc_data; p = rm->rm_col[VDEV_RAIDZ_P].rc_data; ccount = rm->rm_col[c].rc_size / sizeof (src[0]); if (c == rm->rm_firstdatacol) { ASSERT(ccount == pcount); for (i = 0; i < ccount; i++, src++, p++) { *p = *src; } } else { ASSERT(ccount <= pcount); for (i = 0; i < ccount; i++, src++, p++) { *p ^= *src; } } } } static void vdev_raidz_generate_parity_pq(raidz_map_t *rm) { uint64_t *p, *q, *src, pcnt, ccnt, mask, i; int c; pcnt = rm->rm_col[VDEV_RAIDZ_P].rc_size / sizeof (src[0]); ASSERT(rm->rm_col[VDEV_RAIDZ_P].rc_size == rm->rm_col[VDEV_RAIDZ_Q].rc_size); for (c = rm->rm_firstdatacol; c < rm->rm_cols; c++) { src = rm->rm_col[c].rc_data; p = rm->rm_col[VDEV_RAIDZ_P].rc_data; q = rm->rm_col[VDEV_RAIDZ_Q].rc_data; ccnt = rm->rm_col[c].rc_size / sizeof (src[0]); if (c == rm->rm_firstdatacol) { ASSERT(ccnt == pcnt || ccnt == 0); for (i = 0; i < ccnt; i++, src++, p++, q++) { *p = *src; *q = *src; } for (; i < pcnt; i++, src++, p++, q++) { *p = 0; *q = 0; } } else { ASSERT(ccnt <= pcnt); /* * Apply the algorithm described above by multiplying * the previous result and adding in the new value. */ for (i = 0; i < ccnt; i++, src++, p++, q++) { *p ^= *src; VDEV_RAIDZ_64MUL_2(*q, mask); *q ^= *src; } /* * Treat short columns as though they are full of 0s. * Note that there's therefore nothing needed for P. */ for (; i < pcnt; i++, q++) { VDEV_RAIDZ_64MUL_2(*q, mask); } } } } static void vdev_raidz_generate_parity_pqr(raidz_map_t *rm) { uint64_t *p, *q, *r, *src, pcnt, ccnt, mask, i; int c; pcnt = rm->rm_col[VDEV_RAIDZ_P].rc_size / sizeof (src[0]); ASSERT(rm->rm_col[VDEV_RAIDZ_P].rc_size == rm->rm_col[VDEV_RAIDZ_Q].rc_size); ASSERT(rm->rm_col[VDEV_RAIDZ_P].rc_size == rm->rm_col[VDEV_RAIDZ_R].rc_size); for (c = rm->rm_firstdatacol; c < rm->rm_cols; c++) { src = rm->rm_col[c].rc_data; p = rm->rm_col[VDEV_RAIDZ_P].rc_data; q = rm->rm_col[VDEV_RAIDZ_Q].rc_data; r = rm->rm_col[VDEV_RAIDZ_R].rc_data; ccnt = rm->rm_col[c].rc_size / sizeof (src[0]); if (c == rm->rm_firstdatacol) { ASSERT(ccnt == pcnt || ccnt == 0); for (i = 0; i < ccnt; i++, src++, p++, q++, r++) { *p = *src; *q = *src; *r = *src; } for (; i < pcnt; i++, src++, p++, q++, r++) { *p = 0; *q = 0; *r = 0; } } else { ASSERT(ccnt <= pcnt); /* * Apply the algorithm described above by multiplying * the previous result and adding in the new value. */ for (i = 0; i < ccnt; i++, src++, p++, q++, r++) { *p ^= *src; VDEV_RAIDZ_64MUL_2(*q, mask); *q ^= *src; VDEV_RAIDZ_64MUL_4(*r, mask); *r ^= *src; } /* * Treat short columns as though they are full of 0s. * Note that there's therefore nothing needed for P. */ for (; i < pcnt; i++, q++, r++) { VDEV_RAIDZ_64MUL_2(*q, mask); VDEV_RAIDZ_64MUL_4(*r, mask); } } } } /* * Generate RAID parity in the first virtual columns according to the number of * parity columns available. */ static void vdev_raidz_generate_parity(raidz_map_t *rm) { switch (rm->rm_firstdatacol) { case 1: vdev_raidz_generate_parity_p(rm); break; case 2: vdev_raidz_generate_parity_pq(rm); break; case 3: vdev_raidz_generate_parity_pqr(rm); break; default: panic("invalid RAID-Z configuration"); } } /* BEGIN CSTYLED */ /* * In the general case of reconstruction, we must solve the system of linear * equations defined by the coeffecients used to generate parity as well as * the contents of the data and parity disks. This can be expressed with * vectors for the original data (D) and the actual data (d) and parity (p) * and a matrix composed of the identity matrix (I) and a dispersal matrix (V): * * __ __ __ __ * | | __ __ | p_0 | * | V | | D_0 | | p_m-1 | * | | x | : | = | d_0 | * | I | | D_n-1 | | : | * | | ~~ ~~ | d_n-1 | * ~~ ~~ ~~ ~~ * * I is simply a square identity matrix of size n, and V is a vandermonde * matrix defined by the coeffecients we chose for the various parity columns * (1, 2, 4). Note that these values were chosen both for simplicity, speedy * computation as well as linear separability. * * __ __ __ __ * | 1 .. 1 1 1 | | p_0 | * | 2^n-1 .. 4 2 1 | __ __ | : | * | 4^n-1 .. 16 4 1 | | D_0 | | p_m-1 | * | 1 .. 0 0 0 | | D_1 | | d_0 | * | 0 .. 0 0 0 | x | D_2 | = | d_1 | * | : : : : | | : | | d_2 | * | 0 .. 1 0 0 | | D_n-1 | | : | * | 0 .. 0 1 0 | ~~ ~~ | : | * | 0 .. 0 0 1 | | d_n-1 | * ~~ ~~ ~~ ~~ * * Note that I, V, d, and p are known. To compute D, we must invert the * matrix and use the known data and parity values to reconstruct the unknown * data values. We begin by removing the rows in V|I and d|p that correspond * to failed or missing columns; we then make V|I square (n x n) and d|p * sized n by removing rows corresponding to unused parity from the bottom up * to generate (V|I)' and (d|p)'. We can then generate the inverse of (V|I)' * using Gauss-Jordan elimination. In the example below we use m=3 parity * columns, n=8 data columns, with errors in d_1, d_2, and p_1: * __ __ * | 1 1 1 1 1 1 1 1 | * | 128 64 32 16 8 4 2 1 | <-----+-+-- missing disks * | 19 205 116 29 64 16 4 1 | / / * | 1 0 0 0 0 0 0 0 | / / * | 0 1 0 0 0 0 0 0 | <--' / * (V|I) = | 0 0 1 0 0 0 0 0 | <---' * | 0 0 0 1 0 0 0 0 | * | 0 0 0 0 1 0 0 0 | * | 0 0 0 0 0 1 0 0 | * | 0 0 0 0 0 0 1 0 | * | 0 0 0 0 0 0 0 1 | * ~~ ~~ * __ __ * | 1 1 1 1 1 1 1 1 | * | 128 64 32 16 8 4 2 1 | * | 19 205 116 29 64 16 4 1 | * | 1 0 0 0 0 0 0 0 | * | 0 1 0 0 0 0 0 0 | * (V|I)' = | 0 0 1 0 0 0 0 0 | * | 0 0 0 1 0 0 0 0 | * | 0 0 0 0 1 0 0 0 | * | 0 0 0 0 0 1 0 0 | * | 0 0 0 0 0 0 1 0 | * | 0 0 0 0 0 0 0 1 | * ~~ ~~ * * Here we employ Gauss-Jordan elimination to find the inverse of (V|I)'. We * have carefully chosen the seed values 1, 2, and 4 to ensure that this * matrix is not singular. * __ __ * | 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 | * | 19 205 116 29 64 16 4 1 0 1 0 0 0 0 0 0 | * | 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 | * | 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 | * | 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 | * | 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 | * | 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 | * | 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 | * ~~ ~~ * __ __ * | 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 | * | 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 | * | 19 205 116 29 64 16 4 1 0 1 0 0 0 0 0 0 | * | 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 | * | 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 | * | 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 | * | 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 | * | 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 | * ~~ ~~ * __ __ * | 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 | * | 0 1 1 0 0 0 0 0 1 0 1 1 1 1 1 1 | * | 0 205 116 0 0 0 0 0 0 1 19 29 64 16 4 1 | * | 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 | * | 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 | * | 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 | * | 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 | * | 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 | * ~~ ~~ * __ __ * | 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 | * | 0 1 1 0 0 0 0 0 1 0 1 1 1 1 1 1 | * | 0 0 185 0 0 0 0 0 205 1 222 208 141 221 201 204 | * | 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 | * | 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 | * | 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 | * | 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 | * | 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 | * ~~ ~~ * __ __ * | 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 | * | 0 1 1 0 0 0 0 0 1 0 1 1 1 1 1 1 | * | 0 0 1 0 0 0 0 0 166 100 4 40 158 168 216 209 | * | 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 | * | 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 | * | 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 | * | 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 | * | 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 | * ~~ ~~ * __ __ * | 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 | * | 0 1 0 0 0 0 0 0 167 100 5 41 159 169 217 208 | * | 0 0 1 0 0 0 0 0 166 100 4 40 158 168 216 209 | * | 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 | * | 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 | * | 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 | * | 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 | * | 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 | * ~~ ~~ * __ __ * | 0 0 1 0 0 0 0 0 | * | 167 100 5 41 159 169 217 208 | * | 166 100 4 40 158 168 216 209 | * (V|I)'^-1 = | 0 0 0 1 0 0 0 0 | * | 0 0 0 0 1 0 0 0 | * | 0 0 0 0 0 1 0 0 | * | 0 0 0 0 0 0 1 0 | * | 0 0 0 0 0 0 0 1 | * ~~ ~~ * * We can then simply compute D = (V|I)'^-1 x (d|p)' to discover the values * of the missing data. * * As is apparent from the example above, the only non-trivial rows in the * inverse matrix correspond to the data disks that we're trying to * reconstruct. Indeed, those are the only rows we need as the others would * only be useful for reconstructing data known or assumed to be valid. For * that reason, we only build the coefficients in the rows that correspond to * targeted columns. */ /* END CSTYLED */ static void vdev_raidz_matrix_init(raidz_map_t *rm, int n, int nmap, int *map, uint8_t **rows) { int i, j; int pow; ASSERT(n == rm->rm_cols - rm->rm_firstdatacol); /* * Fill in the missing rows of interest. */ for (i = 0; i < nmap; i++) { ASSERT3S(0, <=, map[i]); ASSERT3S(map[i], <=, 2); pow = map[i] * n; if (pow > 255) pow -= 255; ASSERT(pow <= 255); for (j = 0; j < n; j++) { pow -= map[i]; if (pow < 0) pow += 255; rows[i][j] = vdev_raidz_pow2[pow]; } } } static void vdev_raidz_matrix_invert(raidz_map_t *rm, int n, int nmissing, int *missing, uint8_t **rows, uint8_t **invrows, const uint8_t *used) { int i, j, ii, jj; uint8_t log; /* * Assert that the first nmissing entries from the array of used * columns correspond to parity columns and that subsequent entries * correspond to data columns. */ for (i = 0; i < nmissing; i++) { ASSERT3S(used[i], <, rm->rm_firstdatacol); } for (; i < n; i++) { ASSERT3S(used[i], >=, rm->rm_firstdatacol); } /* * First initialize the storage where we'll compute the inverse rows. */ for (i = 0; i < nmissing; i++) { for (j = 0; j < n; j++) { invrows[i][j] = (i == j) ? 1 : 0; } } /* * Subtract all trivial rows from the rows of consequence. */ for (i = 0; i < nmissing; i++) { for (j = nmissing; j < n; j++) { ASSERT3U(used[j], >=, rm->rm_firstdatacol); jj = used[j] - rm->rm_firstdatacol; ASSERT3S(jj, <, n); invrows[i][j] = rows[i][jj]; rows[i][jj] = 0; } } /* * For each of the rows of interest, we must normalize it and subtract * a multiple of it from the other rows. */ for (i = 0; i < nmissing; i++) { for (j = 0; j < missing[i]; j++) { ASSERT3U(rows[i][j], ==, 0); } ASSERT3U(rows[i][missing[i]], !=, 0); /* * Compute the inverse of the first element and multiply each * element in the row by that value. */ log = 255 - vdev_raidz_log2[rows[i][missing[i]]]; for (j = 0; j < n; j++) { rows[i][j] = vdev_raidz_exp2(rows[i][j], log); invrows[i][j] = vdev_raidz_exp2(invrows[i][j], log); } for (ii = 0; ii < nmissing; ii++) { if (i == ii) continue; ASSERT3U(rows[ii][missing[i]], !=, 0); log = vdev_raidz_log2[rows[ii][missing[i]]]; for (j = 0; j < n; j++) { rows[ii][j] ^= vdev_raidz_exp2(rows[i][j], log); invrows[ii][j] ^= vdev_raidz_exp2(invrows[i][j], log); } } } /* * Verify that the data that is left in the rows are properly part of * an identity matrix. */ for (i = 0; i < nmissing; i++) { for (j = 0; j < n; j++) { if (j == missing[i]) { ASSERT3U(rows[i][j], ==, 1); } else { ASSERT3U(rows[i][j], ==, 0); } } } } static void vdev_raidz_matrix_reconstruct(raidz_map_t *rm, int n, int nmissing, int *missing, uint8_t **invrows, const uint8_t *used) { int i, j, x, cc, c; uint8_t *src; uint64_t ccount; uint8_t *dst[VDEV_RAIDZ_MAXPARITY]; uint64_t dcount[VDEV_RAIDZ_MAXPARITY]; uint8_t log, val; int ll; uint8_t *invlog[VDEV_RAIDZ_MAXPARITY]; uint8_t *p, *pp; size_t psize; log = 0; /* gcc */ psize = sizeof (invlog[0][0]) * n * nmissing; p = zfs_alloc(psize); for (pp = p, i = 0; i < nmissing; i++) { invlog[i] = pp; pp += n; } for (i = 0; i < nmissing; i++) { for (j = 0; j < n; j++) { ASSERT3U(invrows[i][j], !=, 0); invlog[i][j] = vdev_raidz_log2[invrows[i][j]]; } } for (i = 0; i < n; i++) { c = used[i]; ASSERT3U(c, <, rm->rm_cols); src = rm->rm_col[c].rc_data; ccount = rm->rm_col[c].rc_size; for (j = 0; j < nmissing; j++) { cc = missing[j] + rm->rm_firstdatacol; ASSERT3U(cc, >=, rm->rm_firstdatacol); ASSERT3U(cc, <, rm->rm_cols); ASSERT3U(cc, !=, c); dst[j] = rm->rm_col[cc].rc_data; dcount[j] = rm->rm_col[cc].rc_size; } ASSERT(ccount >= rm->rm_col[missing[0]].rc_size || i > 0); for (x = 0; x < ccount; x++, src++) { if (*src != 0) log = vdev_raidz_log2[*src]; for (cc = 0; cc < nmissing; cc++) { if (x >= dcount[cc]) continue; if (*src == 0) { val = 0; } else { if ((ll = log + invlog[cc][i]) >= 255) ll -= 255; val = vdev_raidz_pow2[ll]; } if (i == 0) dst[cc][x] = val; else dst[cc][x] ^= val; } } } zfs_free(p, psize); } static int vdev_raidz_reconstruct_general(raidz_map_t *rm, int *tgts, int ntgts) { int n, i, c, t, tt; int nmissing_rows; int missing_rows[VDEV_RAIDZ_MAXPARITY]; int parity_map[VDEV_RAIDZ_MAXPARITY]; uint8_t *p, *pp; size_t psize; uint8_t *rows[VDEV_RAIDZ_MAXPARITY]; uint8_t *invrows[VDEV_RAIDZ_MAXPARITY]; uint8_t *used; int code = 0; n = rm->rm_cols - rm->rm_firstdatacol; /* * Figure out which data columns are missing. */ nmissing_rows = 0; for (t = 0; t < ntgts; t++) { if (tgts[t] >= rm->rm_firstdatacol) { missing_rows[nmissing_rows++] = tgts[t] - rm->rm_firstdatacol; } } /* * Figure out which parity columns to use to help generate the missing * data columns. */ for (tt = 0, c = 0, i = 0; i < nmissing_rows; c++) { ASSERT(tt < ntgts); ASSERT(c < rm->rm_firstdatacol); /* * Skip any targeted parity columns. */ if (c == tgts[tt]) { tt++; continue; } code |= 1 << c; parity_map[i] = c; i++; } ASSERT(code != 0); ASSERT3U(code, <, 1 << VDEV_RAIDZ_MAXPARITY); psize = (sizeof (rows[0][0]) + sizeof (invrows[0][0])) * nmissing_rows * n + sizeof (used[0]) * n; p = kmem_alloc(psize, KM_SLEEP); for (pp = p, i = 0; i < nmissing_rows; i++) { rows[i] = pp; pp += n; invrows[i] = pp; pp += n; } used = pp; for (i = 0; i < nmissing_rows; i++) { used[i] = parity_map[i]; } for (tt = 0, c = rm->rm_firstdatacol; c < rm->rm_cols; c++) { if (tt < nmissing_rows && c == missing_rows[tt] + rm->rm_firstdatacol) { tt++; continue; } ASSERT3S(i, <, n); used[i] = c; i++; } /* * Initialize the interesting rows of the matrix. */ vdev_raidz_matrix_init(rm, n, nmissing_rows, parity_map, rows); /* * Invert the matrix. */ vdev_raidz_matrix_invert(rm, n, nmissing_rows, missing_rows, rows, invrows, used); /* * Reconstruct the missing data using the generated matrix. */ vdev_raidz_matrix_reconstruct(rm, n, nmissing_rows, missing_rows, invrows, used); kmem_free(p, psize); return (code); } static int vdev_raidz_reconstruct(raidz_map_t *rm, int *t, int nt) { int tgts[VDEV_RAIDZ_MAXPARITY]; int ntgts; int i, c; int code; int nbadparity, nbaddata; /* * The tgts list must already be sorted. */ for (i = 1; i < nt; i++) { ASSERT(t[i] > t[i - 1]); } nbadparity = rm->rm_firstdatacol; nbaddata = rm->rm_cols - nbadparity; ntgts = 0; for (i = 0, c = 0; c < rm->rm_cols; c++) { if (i < nt && c == t[i]) { tgts[ntgts++] = c; i++; } else if (rm->rm_col[c].rc_error != 0) { tgts[ntgts++] = c; } else if (c >= rm->rm_firstdatacol) { nbaddata--; } else { nbadparity--; } } ASSERT(ntgts >= nt); ASSERT(nbaddata >= 0); ASSERT(nbaddata + nbadparity == ntgts); code = vdev_raidz_reconstruct_general(rm, tgts, ntgts); ASSERT(code < (1 << VDEV_RAIDZ_MAXPARITY)); ASSERT(code > 0); return (code); } static raidz_map_t * vdev_raidz_map_alloc(void *data, off_t offset, size_t size, uint64_t unit_shift, uint64_t dcols, uint64_t nparity) { raidz_map_t *rm; uint64_t b = offset >> unit_shift; uint64_t s = size >> unit_shift; uint64_t f = b % dcols; uint64_t o = (b / dcols) << unit_shift; uint64_t q, r, c, bc, col, acols, scols, coff, devidx, asize, tot; q = s / (dcols - nparity); r = s - q * (dcols - nparity); bc = (r == 0 ? 0 : r + nparity); tot = s + nparity * (q + (r == 0 ? 0 : 1)); if (q == 0) { acols = bc; scols = MIN(dcols, roundup(bc, nparity + 1)); } else { acols = dcols; scols = dcols; } ASSERT3U(acols, <=, scols); rm = zfs_alloc(offsetof(raidz_map_t, rm_col[scols])); rm->rm_cols = acols; rm->rm_scols = scols; rm->rm_bigcols = bc; rm->rm_skipstart = bc; rm->rm_missingdata = 0; rm->rm_missingparity = 0; rm->rm_firstdatacol = nparity; rm->rm_reports = 0; rm->rm_freed = 0; rm->rm_ecksuminjected = 0; asize = 0; for (c = 0; c < scols; c++) { col = f + c; coff = o; if (col >= dcols) { col -= dcols; coff += 1ULL << unit_shift; } rm->rm_col[c].rc_devidx = col; rm->rm_col[c].rc_offset = coff; rm->rm_col[c].rc_data = NULL; rm->rm_col[c].rc_error = 0; rm->rm_col[c].rc_tried = 0; rm->rm_col[c].rc_skipped = 0; if (c >= acols) rm->rm_col[c].rc_size = 0; else if (c < bc) rm->rm_col[c].rc_size = (q + 1) << unit_shift; else rm->rm_col[c].rc_size = q << unit_shift; asize += rm->rm_col[c].rc_size; } ASSERT3U(asize, ==, tot << unit_shift); rm->rm_asize = roundup(asize, (nparity + 1) << unit_shift); rm->rm_nskip = roundup(tot, nparity + 1) - tot; ASSERT3U(rm->rm_asize - asize, ==, rm->rm_nskip << unit_shift); ASSERT3U(rm->rm_nskip, <=, nparity); for (c = 0; c < rm->rm_firstdatacol; c++) rm->rm_col[c].rc_data = zfs_alloc(rm->rm_col[c].rc_size); rm->rm_col[c].rc_data = data; for (c = c + 1; c < acols; c++) rm->rm_col[c].rc_data = (char *)rm->rm_col[c - 1].rc_data + rm->rm_col[c - 1].rc_size; /* * If all data stored spans all columns, there's a danger that parity * will always be on the same device and, since parity isn't read * during normal operation, that that device's I/O bandwidth won't be * used effectively. We therefore switch the parity every 1MB. * * ... at least that was, ostensibly, the theory. As a practical * matter unless we juggle the parity between all devices evenly, we * won't see any benefit. Further, occasional writes that aren't a * multiple of the LCM of the number of children and the minimum * stripe width are sufficient to avoid pessimal behavior. * Unfortunately, this decision created an implicit on-disk format * requirement that we need to support for all eternity, but only * for single-parity RAID-Z. * * If we intend to skip a sector in the zeroth column for padding * we must make sure to note this swap. We will never intend to * skip the first column since at least one data and one parity * column must appear in each row. */ ASSERT(rm->rm_cols >= 2); ASSERT(rm->rm_col[0].rc_size == rm->rm_col[1].rc_size); if (rm->rm_firstdatacol == 1 && (offset & (1ULL << 20))) { devidx = rm->rm_col[0].rc_devidx; o = rm->rm_col[0].rc_offset; rm->rm_col[0].rc_devidx = rm->rm_col[1].rc_devidx; rm->rm_col[0].rc_offset = rm->rm_col[1].rc_offset; rm->rm_col[1].rc_devidx = devidx; rm->rm_col[1].rc_offset = o; if (rm->rm_skipstart == 0) rm->rm_skipstart = 1; } return (rm); } static void vdev_raidz_map_free(raidz_map_t *rm) { int c; for (c = rm->rm_firstdatacol - 1; c >= 0; c--) zfs_free(rm->rm_col[c].rc_data, rm->rm_col[c].rc_size); zfs_free(rm, offsetof(raidz_map_t, rm_col[rm->rm_scols])); } static vdev_t * vdev_child(vdev_t *pvd, uint64_t devidx) { vdev_t *cvd; STAILQ_FOREACH(cvd, &pvd->v_children, v_childlink) { if (cvd->v_id == devidx) break; } return (cvd); } /* * We keep track of whether or not there were any injected errors, so that * any ereports we generate can note it. */ static int -raidz_checksum_verify(const blkptr_t *bp, void *data, uint64_t size) +raidz_checksum_verify(const spa_t *spa, const blkptr_t *bp, void *data, + uint64_t size) { - - return (zio_checksum_verify(bp, data)); + return (zio_checksum_verify(spa, bp, data)); } /* * Generate the parity from the data columns. If we tried and were able to * read the parity without error, verify that the generated parity matches the * data we read. If it doesn't, we fire off a checksum error. Return the * number such failures. */ static int raidz_parity_verify(raidz_map_t *rm) { void *orig[VDEV_RAIDZ_MAXPARITY]; int c, ret = 0; raidz_col_t *rc; for (c = 0; c < rm->rm_firstdatacol; c++) { rc = &rm->rm_col[c]; if (!rc->rc_tried || rc->rc_error != 0) continue; orig[c] = zfs_alloc(rc->rc_size); bcopy(rc->rc_data, orig[c], rc->rc_size); } vdev_raidz_generate_parity(rm); for (c = rm->rm_firstdatacol - 1; c >= 0; c--) { rc = &rm->rm_col[c]; if (!rc->rc_tried || rc->rc_error != 0) continue; if (bcmp(orig[c], rc->rc_data, rc->rc_size) != 0) { rc->rc_error = ECKSUM; ret++; } zfs_free(orig[c], rc->rc_size); } return (ret); } /* * Iterate over all combinations of bad data and attempt a reconstruction. * Note that the algorithm below is non-optimal because it doesn't take into * account how reconstruction is actually performed. For example, with * triple-parity RAID-Z the reconstruction procedure is the same if column 4 * is targeted as invalid as if columns 1 and 4 are targeted since in both * cases we'd only use parity information in column 0. */ static int -vdev_raidz_combrec(raidz_map_t *rm, const blkptr_t *bp, void *data, - off_t offset, uint64_t bytes, int total_errors, int data_errors) +vdev_raidz_combrec(const spa_t *spa, raidz_map_t *rm, const blkptr_t *bp, + void *data, off_t offset, uint64_t bytes, int total_errors, int data_errors) { raidz_col_t *rc; void *orig[VDEV_RAIDZ_MAXPARITY]; int tstore[VDEV_RAIDZ_MAXPARITY + 2]; int *tgts = &tstore[1]; int current, next, i, c, n; int code, ret = 0; ASSERT(total_errors < rm->rm_firstdatacol); /* * This simplifies one edge condition. */ tgts[-1] = -1; for (n = 1; n <= rm->rm_firstdatacol - total_errors; n++) { /* * Initialize the targets array by finding the first n columns * that contain no error. * * If there were no data errors, we need to ensure that we're * always explicitly attempting to reconstruct at least one * data column. To do this, we simply push the highest target * up into the data columns. */ for (c = 0, i = 0; i < n; i++) { if (i == n - 1 && data_errors == 0 && c < rm->rm_firstdatacol) { c = rm->rm_firstdatacol; } while (rm->rm_col[c].rc_error != 0) { c++; ASSERT3S(c, <, rm->rm_cols); } tgts[i] = c++; } /* * Setting tgts[n] simplifies the other edge condition. */ tgts[n] = rm->rm_cols; /* * These buffers were allocated in previous iterations. */ for (i = 0; i < n - 1; i++) { ASSERT(orig[i] != NULL); } orig[n - 1] = zfs_alloc(rm->rm_col[0].rc_size); current = 0; next = tgts[current]; while (current != n) { tgts[current] = next; current = 0; /* * Save off the original data that we're going to * attempt to reconstruct. */ for (i = 0; i < n; i++) { ASSERT(orig[i] != NULL); c = tgts[i]; ASSERT3S(c, >=, 0); ASSERT3S(c, <, rm->rm_cols); rc = &rm->rm_col[c]; bcopy(rc->rc_data, orig[i], rc->rc_size); } /* * Attempt a reconstruction and exit the outer loop on * success. */ code = vdev_raidz_reconstruct(rm, tgts, n); - if (raidz_checksum_verify(bp, data, bytes) == 0) { + if (raidz_checksum_verify(spa, bp, data, bytes) == 0) { for (i = 0; i < n; i++) { c = tgts[i]; rc = &rm->rm_col[c]; ASSERT(rc->rc_error == 0); rc->rc_error = ECKSUM; } ret = code; goto done; } /* * Restore the original data. */ for (i = 0; i < n; i++) { c = tgts[i]; rc = &rm->rm_col[c]; bcopy(orig[i], rc->rc_data, rc->rc_size); } do { /* * Find the next valid column after the current * position.. */ for (next = tgts[current] + 1; next < rm->rm_cols && rm->rm_col[next].rc_error != 0; next++) continue; ASSERT(next <= tgts[current + 1]); /* * If that spot is available, we're done here. */ if (next != tgts[current + 1]) break; /* * Otherwise, find the next valid column after * the previous position. */ for (c = tgts[current - 1] + 1; rm->rm_col[c].rc_error != 0; c++) continue; tgts[current] = c; current++; } while (current != n); } } n--; done: for (i = n - 1; i >= 0; i--) { zfs_free(orig[i], rm->rm_col[0].rc_size); } return (ret); } static int vdev_raidz_read(vdev_t *vd, const blkptr_t *bp, void *data, off_t offset, size_t bytes) { vdev_t *tvd = vd->v_top; vdev_t *cvd; raidz_map_t *rm; raidz_col_t *rc; int c, error; int unexpected_errors; int parity_errors; int parity_untried; int data_errors; int total_errors; int n; int tgts[VDEV_RAIDZ_MAXPARITY]; int code; rc = NULL; /* gcc */ error = 0; rm = vdev_raidz_map_alloc(data, offset, bytes, tvd->v_ashift, vd->v_nchildren, vd->v_nparity); /* * Iterate over the columns in reverse order so that we hit the parity * last -- any errors along the way will force us to read the parity. */ for (c = rm->rm_cols - 1; c >= 0; c--) { rc = &rm->rm_col[c]; cvd = vdev_child(vd, rc->rc_devidx); if (cvd == NULL || cvd->v_state != VDEV_STATE_HEALTHY) { if (c >= rm->rm_firstdatacol) rm->rm_missingdata++; else rm->rm_missingparity++; rc->rc_error = ENXIO; rc->rc_tried = 1; /* don't even try */ rc->rc_skipped = 1; continue; } #if 0 /* XXX: Too hard for the boot code. */ if (vdev_dtl_contains(cvd, DTL_MISSING, zio->io_txg, 1)) { if (c >= rm->rm_firstdatacol) rm->rm_missingdata++; else rm->rm_missingparity++; rc->rc_error = ESTALE; rc->rc_skipped = 1; continue; } #endif if (c >= rm->rm_firstdatacol || rm->rm_missingdata > 0) { rc->rc_error = cvd->v_read(cvd, NULL, rc->rc_data, rc->rc_offset, rc->rc_size); rc->rc_tried = 1; rc->rc_skipped = 0; } } reconstruct: unexpected_errors = 0; parity_errors = 0; parity_untried = 0; data_errors = 0; total_errors = 0; ASSERT(rm->rm_missingparity <= rm->rm_firstdatacol); ASSERT(rm->rm_missingdata <= rm->rm_cols - rm->rm_firstdatacol); for (c = 0; c < rm->rm_cols; c++) { rc = &rm->rm_col[c]; if (rc->rc_error) { ASSERT(rc->rc_error != ECKSUM); /* child has no bp */ if (c < rm->rm_firstdatacol) parity_errors++; else data_errors++; if (!rc->rc_skipped) unexpected_errors++; total_errors++; } else if (c < rm->rm_firstdatacol && !rc->rc_tried) { parity_untried++; } } /* * There are three potential phases for a read: * 1. produce valid data from the columns read * 2. read all disks and try again * 3. perform combinatorial reconstruction * * Each phase is progressively both more expensive and less likely to * occur. If we encounter more errors than we can repair or all phases * fail, we have no choice but to return an error. */ /* * If the number of errors we saw was correctable -- less than or equal * to the number of parity disks read -- attempt to produce data that * has a valid checksum. Naturally, this case applies in the absence of * any errors. */ if (total_errors <= rm->rm_firstdatacol - parity_untried) { if (data_errors == 0) { - if (raidz_checksum_verify(bp, data, bytes) == 0) { + if (raidz_checksum_verify(vd->spa, bp, data, bytes) == 0) { /* * If we read parity information (unnecessarily * as it happens since no reconstruction was * needed) regenerate and verify the parity. * We also regenerate parity when resilvering * so we can write it out to the failed device * later. */ if (parity_errors + parity_untried < rm->rm_firstdatacol) { n = raidz_parity_verify(rm); unexpected_errors += n; ASSERT(parity_errors + n <= rm->rm_firstdatacol); } goto done; } } else { /* * We either attempt to read all the parity columns or * none of them. If we didn't try to read parity, we * wouldn't be here in the correctable case. There must * also have been fewer parity errors than parity * columns or, again, we wouldn't be in this code path. */ ASSERT(parity_untried == 0); ASSERT(parity_errors < rm->rm_firstdatacol); /* * Identify the data columns that reported an error. */ n = 0; for (c = rm->rm_firstdatacol; c < rm->rm_cols; c++) { rc = &rm->rm_col[c]; if (rc->rc_error != 0) { ASSERT(n < VDEV_RAIDZ_MAXPARITY); tgts[n++] = c; } } ASSERT(rm->rm_firstdatacol >= n); code = vdev_raidz_reconstruct(rm, tgts, n); - if (raidz_checksum_verify(bp, data, bytes) == 0) { + if (raidz_checksum_verify(vd->spa, bp, data, bytes) == 0) { /* * If we read more parity disks than were used * for reconstruction, confirm that the other * parity disks produced correct data. This * routine is suboptimal in that it regenerates * the parity that we already used in addition * to the parity that we're attempting to * verify, but this should be a relatively * uncommon case, and can be optimized if it * becomes a problem. Note that we regenerate * parity when resilvering so we can write it * out to failed devices later. */ if (parity_errors < rm->rm_firstdatacol - n) { n = raidz_parity_verify(rm); unexpected_errors += n; ASSERT(parity_errors + n <= rm->rm_firstdatacol); } goto done; } } } /* * This isn't a typical situation -- either we got a read * error or a child silently returned bad data. Read every * block so we can try again with as much data and parity as * we can track down. If we've already been through once * before, all children will be marked as tried so we'll * proceed to combinatorial reconstruction. */ unexpected_errors = 1; rm->rm_missingdata = 0; rm->rm_missingparity = 0; n = 0; for (c = 0; c < rm->rm_cols; c++) { rc = &rm->rm_col[c]; if (rc->rc_tried) continue; cvd = vdev_child(vd, rc->rc_devidx); ASSERT(cvd != NULL); rc->rc_error = cvd->v_read(cvd, NULL, rc->rc_data, rc->rc_offset, rc->rc_size); if (rc->rc_error == 0) n++; rc->rc_tried = 1; rc->rc_skipped = 0; } /* * If we managed to read anything more, retry the * reconstruction. */ if (n > 0) goto reconstruct; /* * At this point we've attempted to reconstruct the data given the * errors we detected, and we've attempted to read all columns. There * must, therefore, be one or more additional problems -- silent errors * resulting in invalid data rather than explicit I/O errors resulting * in absent data. We check if there is enough additional data to * possibly reconstruct the data and then perform combinatorial * reconstruction over all possible combinations. If that fails, * we're cooked. */ if (total_errors > rm->rm_firstdatacol) { error = EIO; } else if (total_errors < rm->rm_firstdatacol && - (code = vdev_raidz_combrec(rm, bp, data, offset, bytes, + (code = vdev_raidz_combrec(vd->spa, rm, bp, data, offset, bytes, total_errors, data_errors)) != 0) { /* * If we didn't use all the available parity for the * combinatorial reconstruction, verify that the remaining * parity is correct. */ if (code != (1 << rm->rm_firstdatacol) - 1) (void) raidz_parity_verify(rm); } else { /* * We're here because either: * * total_errors == rm_first_datacol, or * vdev_raidz_combrec() failed * * In either case, there is enough bad data to prevent * reconstruction. * * Start checksum ereports for all children which haven't * failed, and the IO wasn't speculative. */ error = ECKSUM; } done: vdev_raidz_map_free(rm); return (error); } Index: stable/11 =================================================================== --- stable/11 (revision 328865) +++ stable/11 (revision 328866) Property changes on: stable/11 ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head:r304321,304753-304754,306751,316077,316110