diff --git a/include/sys/zfs_context.h b/include/sys/zfs_context.h index 9ec2f73b366c..8f264b50e995 100644 --- a/include/sys/zfs_context.h +++ b/include/sys/zfs_context.h @@ -1,792 +1,792 @@ /* * 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 https://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 */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2012, 2018 by Delphix. All rights reserved. * Copyright (c) 2012, Joyent, Inc. All rights reserved. */ #ifndef _SYS_ZFS_CONTEXT_H #define _SYS_ZFS_CONTEXT_H #ifdef __cplusplus extern "C" { #endif /* * This code compiles in three different contexts. When __KERNEL__ is defined, * the code uses "unix-like" kernel interfaces. When _STANDALONE is defined, the * code is running in a reduced capacity environment of the boot loader which is * generally a subset of both POSIX and kernel interfaces (with a few unique * interfaces too). When neither are defined, it's in a userland POSIX or * similar environment. */ #if defined(__KERNEL__) || defined(_STANDALONE) #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #else /* _KERNEL || _STANDALONE */ #define _SYS_MUTEX_H #define _SYS_RWLOCK_H #define _SYS_CONDVAR_H #define _SYS_VNODE_H #define _SYS_VFS_H #define _SYS_SUNDDI_H #define _SYS_CALLB_H #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * Stack */ #define noinline __attribute__((noinline)) #define likely(x) __builtin_expect((x), 1) #define unlikely(x) __builtin_expect((x), 0) /* * Debugging */ /* * Note that we are not using the debugging levels. */ #define CE_CONT 0 /* continuation */ #define CE_NOTE 1 /* notice */ #define CE_WARN 2 /* warning */ #define CE_PANIC 3 /* panic */ #define CE_IGNORE 4 /* print nothing */ /* * ZFS debugging */ extern void dprintf_setup(int *argc, char **argv); extern void cmn_err(int, const char *, ...) __attribute__((format(printf, 2, 3))); extern void vcmn_err(int, const char *, va_list) __attribute__((format(printf, 2, 0))); extern void panic(const char *, ...) __attribute__((format(printf, 1, 2), noreturn)); extern void vpanic(const char *, va_list) __attribute__((format(printf, 1, 0), noreturn)); #define fm_panic panic /* * DTrace SDT probes have different signatures in userland than they do in * the kernel. If they're being used in kernel code, re-define them out of * existence for their counterparts in libzpool. * * Here's an example of how to use the set-error probes in userland: * zfs$target:::set-error /arg0 == EBUSY/ {stack();} * * Here's an example of how to use DTRACE_PROBE probes in userland: * If there is a probe declared as follows: * DTRACE_PROBE2(zfs__probe_name, uint64_t, blkid, dnode_t *, dn); * Then you can use it as follows: * zfs$target:::probe2 /copyinstr(arg0) == "zfs__probe_name"/ * {printf("%u %p\n", arg1, arg2);} */ #ifdef DTRACE_PROBE #undef DTRACE_PROBE #endif /* DTRACE_PROBE */ #define DTRACE_PROBE(a) #ifdef DTRACE_PROBE1 #undef DTRACE_PROBE1 #endif /* DTRACE_PROBE1 */ #define DTRACE_PROBE1(a, b, c) #ifdef DTRACE_PROBE2 #undef DTRACE_PROBE2 #endif /* DTRACE_PROBE2 */ #define DTRACE_PROBE2(a, b, c, d, e) #ifdef DTRACE_PROBE3 #undef DTRACE_PROBE3 #endif /* DTRACE_PROBE3 */ #define DTRACE_PROBE3(a, b, c, d, e, f, g) #ifdef DTRACE_PROBE4 #undef DTRACE_PROBE4 #endif /* DTRACE_PROBE4 */ #define DTRACE_PROBE4(a, b, c, d, e, f, g, h, i) /* * Tunables. */ typedef struct zfs_kernel_param { const char *name; /* unused stub */ } zfs_kernel_param_t; #define ZFS_MODULE_PARAM(scope_prefix, name_prefix, name, type, perm, desc) #define ZFS_MODULE_PARAM_ARGS void #define ZFS_MODULE_PARAM_CALL(scope_prefix, name_prefix, name, setfunc, \ getfunc, perm, desc) /* * Threads. */ typedef pthread_t kthread_t; #define TS_RUN 0x00000002 #define TS_JOINABLE 0x00000004 #define curthread ((void *)(uintptr_t)pthread_self()) #define getcomm() "unknown" #define thread_create_named(name, stk, stksize, func, arg, len, \ pp, state, pri) \ - zk_thread_create(func, arg, stksize, state) + zk_thread_create(name, func, arg, stksize, state) #define thread_create(stk, stksize, func, arg, len, pp, state, pri) \ - zk_thread_create(func, arg, stksize, state) + zk_thread_create(#func, func, arg, stksize, state) #define thread_exit() pthread_exit(NULL) #define thread_join(t) pthread_join((pthread_t)(t), NULL) #define newproc(f, a, cid, pri, ctp, pid) (ENOSYS) /* in libzpool, p0 exists only to have its address taken */ typedef struct proc { uintptr_t this_is_never_used_dont_dereference_it; } proc_t; extern struct proc p0; #define curproc (&p0) #define PS_NONE -1 -extern kthread_t *zk_thread_create(void (*func)(void *), void *arg, - size_t stksize, int state); +extern kthread_t *zk_thread_create(const char *name, void (*func)(void *), + void *arg, size_t stksize, int state); #define issig(why) (FALSE) #define ISSIG(thr, why) (FALSE) #define KPREEMPT_SYNC (-1) #define kpreempt(x) sched_yield() #define kpreempt_disable() ((void)0) #define kpreempt_enable() ((void)0) /* * Mutexes */ typedef struct kmutex { pthread_mutex_t m_lock; pthread_t m_owner; } kmutex_t; #define MUTEX_DEFAULT 0 #define MUTEX_NOLOCKDEP MUTEX_DEFAULT #define MUTEX_HELD(mp) pthread_equal((mp)->m_owner, pthread_self()) #define MUTEX_NOT_HELD(mp) !MUTEX_HELD(mp) extern void mutex_init(kmutex_t *mp, char *name, int type, void *cookie); extern void mutex_destroy(kmutex_t *mp); extern void mutex_enter(kmutex_t *mp); extern int mutex_enter_check_return(kmutex_t *mp); extern void mutex_exit(kmutex_t *mp); extern int mutex_tryenter(kmutex_t *mp); #define NESTED_SINGLE 1 #define mutex_enter_nested(mp, class) mutex_enter(mp) #define mutex_enter_interruptible(mp) mutex_enter_check_return(mp) /* * RW locks */ typedef struct krwlock { pthread_rwlock_t rw_lock; pthread_t rw_owner; uint_t rw_readers; } krwlock_t; typedef int krw_t; #define RW_READER 0 #define RW_WRITER 1 #define RW_DEFAULT RW_READER #define RW_NOLOCKDEP RW_READER #define RW_READ_HELD(rw) ((rw)->rw_readers > 0) #define RW_WRITE_HELD(rw) pthread_equal((rw)->rw_owner, pthread_self()) #define RW_LOCK_HELD(rw) (RW_READ_HELD(rw) || RW_WRITE_HELD(rw)) extern void rw_init(krwlock_t *rwlp, char *name, int type, void *arg); extern void rw_destroy(krwlock_t *rwlp); extern void rw_enter(krwlock_t *rwlp, krw_t rw); extern int rw_tryenter(krwlock_t *rwlp, krw_t rw); extern int rw_tryupgrade(krwlock_t *rwlp); extern void rw_exit(krwlock_t *rwlp); #define rw_downgrade(rwlp) do { } while (0) /* * Credentials */ extern uid_t crgetuid(cred_t *cr); extern uid_t crgetruid(cred_t *cr); extern gid_t crgetgid(cred_t *cr); extern int crgetngroups(cred_t *cr); extern gid_t *crgetgroups(cred_t *cr); /* * Condition variables */ typedef pthread_cond_t kcondvar_t; #define CV_DEFAULT 0 #define CALLOUT_FLAG_ABSOLUTE 0x2 extern void cv_init(kcondvar_t *cv, char *name, int type, void *arg); extern void cv_destroy(kcondvar_t *cv); extern void cv_wait(kcondvar_t *cv, kmutex_t *mp); extern int cv_wait_sig(kcondvar_t *cv, kmutex_t *mp); extern int cv_timedwait(kcondvar_t *cv, kmutex_t *mp, clock_t abstime); extern int cv_timedwait_hires(kcondvar_t *cvp, kmutex_t *mp, hrtime_t tim, hrtime_t res, int flag); extern void cv_signal(kcondvar_t *cv); extern void cv_broadcast(kcondvar_t *cv); #define cv_timedwait_io(cv, mp, at) cv_timedwait(cv, mp, at) #define cv_timedwait_idle(cv, mp, at) cv_timedwait(cv, mp, at) #define cv_timedwait_sig(cv, mp, at) cv_timedwait(cv, mp, at) #define cv_wait_io(cv, mp) cv_wait(cv, mp) #define cv_wait_idle(cv, mp) cv_wait(cv, mp) #define cv_wait_io_sig(cv, mp) cv_wait_sig(cv, mp) #define cv_timedwait_sig_hires(cv, mp, t, r, f) \ cv_timedwait_hires(cv, mp, t, r, f) #define cv_timedwait_idle_hires(cv, mp, t, r, f) \ cv_timedwait_hires(cv, mp, t, r, f) /* * Thread-specific data */ #define tsd_get(k) pthread_getspecific(k) #define tsd_set(k, v) pthread_setspecific(k, v) #define tsd_create(kp, d) pthread_key_create((pthread_key_t *)kp, d) #define tsd_destroy(kp) /* nothing */ #ifdef __FreeBSD__ typedef off_t loff_t; #endif /* * kstat creation, installation and deletion */ extern kstat_t *kstat_create(const char *, int, const char *, const char *, uchar_t, ulong_t, uchar_t); extern void kstat_install(kstat_t *); extern void kstat_delete(kstat_t *); extern void kstat_set_raw_ops(kstat_t *ksp, int (*headers)(char *buf, size_t size), int (*data)(char *buf, size_t size, void *data), void *(*addr)(kstat_t *ksp, loff_t index)); /* * procfs list manipulation */ typedef struct procfs_list { void *pl_private; kmutex_t pl_lock; list_t pl_list; uint64_t pl_next_id; size_t pl_node_offset; } procfs_list_t; #ifndef __cplusplus struct seq_file { }; void seq_printf(struct seq_file *m, const char *fmt, ...); typedef struct procfs_list_node { list_node_t pln_link; uint64_t pln_id; } procfs_list_node_t; void procfs_list_install(const char *module, const char *submodule, const char *name, mode_t mode, procfs_list_t *procfs_list, int (*show)(struct seq_file *f, void *p), int (*show_header)(struct seq_file *f), int (*clear)(procfs_list_t *procfs_list), size_t procfs_list_node_off); void procfs_list_uninstall(procfs_list_t *procfs_list); void procfs_list_destroy(procfs_list_t *procfs_list); void procfs_list_add(procfs_list_t *procfs_list, void *p); #endif /* * Kernel memory */ #define KM_SLEEP UMEM_NOFAIL #define KM_PUSHPAGE KM_SLEEP #define KM_NOSLEEP UMEM_DEFAULT #define KM_NORMALPRI 0 /* not needed with UMEM_DEFAULT */ #define KMC_NODEBUG UMC_NODEBUG #define KMC_KVMEM 0x0 #define kmem_alloc(_s, _f) umem_alloc(_s, _f) #define kmem_zalloc(_s, _f) umem_zalloc(_s, _f) #define kmem_free(_b, _s) umem_free(_b, _s) #define vmem_alloc(_s, _f) kmem_alloc(_s, _f) #define vmem_zalloc(_s, _f) kmem_zalloc(_s, _f) #define vmem_free(_b, _s) kmem_free(_b, _s) #define kmem_cache_create(_a, _b, _c, _d, _e, _f, _g, _h, _i) \ umem_cache_create(_a, _b, _c, _d, _e, _f, _g, _h, _i) #define kmem_cache_destroy(_c) umem_cache_destroy(_c) #define kmem_cache_alloc(_c, _f) umem_cache_alloc(_c, _f) #define kmem_cache_free(_c, _b) umem_cache_free(_c, _b) #define kmem_debugging() 0 #define kmem_cache_reap_now(_c) umem_cache_reap_now(_c); #define kmem_cache_set_move(_c, _cb) /* nothing */ #define POINTER_INVALIDATE(_pp) /* nothing */ #define POINTER_IS_VALID(_p) 0 typedef umem_cache_t kmem_cache_t; typedef enum kmem_cbrc { KMEM_CBRC_YES, KMEM_CBRC_NO, KMEM_CBRC_LATER, KMEM_CBRC_DONT_NEED, KMEM_CBRC_DONT_KNOW } kmem_cbrc_t; /* * Task queues */ #define TASKQ_NAMELEN 31 typedef uintptr_t taskqid_t; typedef void (task_func_t)(void *); typedef struct taskq_ent { struct taskq_ent *tqent_next; struct taskq_ent *tqent_prev; task_func_t *tqent_func; void *tqent_arg; uintptr_t tqent_flags; } taskq_ent_t; typedef struct taskq { char tq_name[TASKQ_NAMELEN + 1]; kmutex_t tq_lock; krwlock_t tq_threadlock; kcondvar_t tq_dispatch_cv; kcondvar_t tq_wait_cv; kthread_t **tq_threadlist; int tq_flags; int tq_active; int tq_nthreads; int tq_nalloc; int tq_minalloc; int tq_maxalloc; kcondvar_t tq_maxalloc_cv; int tq_maxalloc_wait; taskq_ent_t *tq_freelist; taskq_ent_t tq_task; } taskq_t; #define TQENT_FLAG_PREALLOC 0x1 /* taskq_dispatch_ent used */ #define TASKQ_PREPOPULATE 0x0001 #define TASKQ_CPR_SAFE 0x0002 /* Use CPR safe protocol */ #define TASKQ_DYNAMIC 0x0004 /* Use dynamic thread scheduling */ #define TASKQ_THREADS_CPU_PCT 0x0008 /* Scale # threads by # cpus */ #define TASKQ_DC_BATCH 0x0010 /* Mark threads as batch */ #define TQ_SLEEP KM_SLEEP /* Can block for memory */ #define TQ_NOSLEEP KM_NOSLEEP /* cannot block for memory; may fail */ #define TQ_NOQUEUE 0x02 /* Do not enqueue if can't dispatch */ #define TQ_FRONT 0x08 /* Queue in front */ #define TASKQID_INVALID ((taskqid_t)0) extern taskq_t *system_taskq; extern taskq_t *system_delay_taskq; extern taskq_t *taskq_create(const char *, int, pri_t, int, int, uint_t); extern taskq_t *taskq_create_synced(const char *, int, pri_t, int, int, uint_t, kthread_t ***); #define taskq_create_proc(a, b, c, d, e, p, f) \ (taskq_create(a, b, c, d, e, f)) #define taskq_create_sysdc(a, b, d, e, p, dc, f) \ ((void) sizeof (dc), taskq_create(a, b, maxclsyspri, d, e, f)) extern taskqid_t taskq_dispatch(taskq_t *, task_func_t, void *, uint_t); extern taskqid_t taskq_dispatch_delay(taskq_t *, task_func_t, void *, uint_t, clock_t); extern void taskq_dispatch_ent(taskq_t *, task_func_t, void *, uint_t, taskq_ent_t *); extern int taskq_empty_ent(taskq_ent_t *); extern void taskq_init_ent(taskq_ent_t *); extern void taskq_destroy(taskq_t *); extern void taskq_wait(taskq_t *); extern void taskq_wait_id(taskq_t *, taskqid_t); extern void taskq_wait_outstanding(taskq_t *, taskqid_t); extern int taskq_member(taskq_t *, kthread_t *); extern taskq_t *taskq_of_curthread(void); extern int taskq_cancel_id(taskq_t *, taskqid_t); extern void system_taskq_init(void); extern void system_taskq_fini(void); #define XVA_MAPSIZE 3 #define XVA_MAGIC 0x78766174 extern char *vn_dumpdir; #define AV_SCANSTAMP_SZ 32 /* length of anti-virus scanstamp */ typedef struct xoptattr { inode_timespec_t xoa_createtime; /* Create time of file */ uint8_t xoa_archive; uint8_t xoa_system; uint8_t xoa_readonly; uint8_t xoa_hidden; uint8_t xoa_nounlink; uint8_t xoa_immutable; uint8_t xoa_appendonly; uint8_t xoa_nodump; uint8_t xoa_settable; uint8_t xoa_opaque; uint8_t xoa_av_quarantined; uint8_t xoa_av_modified; uint8_t xoa_av_scanstamp[AV_SCANSTAMP_SZ]; uint8_t xoa_reparse; uint8_t xoa_offline; uint8_t xoa_sparse; } xoptattr_t; typedef struct vattr { uint_t va_mask; /* bit-mask of attributes */ u_offset_t va_size; /* file size in bytes */ } vattr_t; typedef struct xvattr { vattr_t xva_vattr; /* Embedded vattr structure */ uint32_t xva_magic; /* Magic Number */ uint32_t xva_mapsize; /* Size of attr bitmap (32-bit words) */ uint32_t *xva_rtnattrmapp; /* Ptr to xva_rtnattrmap[] */ uint32_t xva_reqattrmap[XVA_MAPSIZE]; /* Requested attrs */ uint32_t xva_rtnattrmap[XVA_MAPSIZE]; /* Returned attrs */ xoptattr_t xva_xoptattrs; /* Optional attributes */ } xvattr_t; typedef struct vsecattr { uint_t vsa_mask; /* See below */ int vsa_aclcnt; /* ACL entry count */ void *vsa_aclentp; /* pointer to ACL entries */ int vsa_dfaclcnt; /* default ACL entry count */ void *vsa_dfaclentp; /* pointer to default ACL entries */ size_t vsa_aclentsz; /* ACE size in bytes of vsa_aclentp */ } vsecattr_t; #define AT_MODE 0x00002 #define AT_UID 0x00004 #define AT_GID 0x00008 #define AT_FSID 0x00010 #define AT_NODEID 0x00020 #define AT_NLINK 0x00040 #define AT_SIZE 0x00080 #define AT_ATIME 0x00100 #define AT_MTIME 0x00200 #define AT_CTIME 0x00400 #define AT_RDEV 0x00800 #define AT_BLKSIZE 0x01000 #define AT_NBLOCKS 0x02000 #define AT_SEQ 0x08000 #define AT_XVATTR 0x10000 #define CRCREAT 0 #define F_FREESP 11 #define FIGNORECASE 0x80000 /* request case-insensitive lookups */ /* * Random stuff */ #define ddi_get_lbolt() (gethrtime() >> 23) #define ddi_get_lbolt64() (gethrtime() >> 23) #define hz 119 /* frequency when using gethrtime() >> 23 for lbolt */ #define ddi_time_before(a, b) (a < b) #define ddi_time_after(a, b) ddi_time_before(b, a) #define ddi_time_before_eq(a, b) (!ddi_time_after(a, b)) #define ddi_time_after_eq(a, b) ddi_time_before_eq(b, a) #define ddi_time_before64(a, b) (a < b) #define ddi_time_after64(a, b) ddi_time_before64(b, a) #define ddi_time_before_eq64(a, b) (!ddi_time_after64(a, b)) #define ddi_time_after_eq64(a, b) ddi_time_before_eq64(b, a) extern void delay(clock_t ticks); #define SEC_TO_TICK(sec) ((sec) * hz) #define MSEC_TO_TICK(msec) (howmany((hrtime_t)(msec) * hz, MILLISEC)) #define USEC_TO_TICK(usec) (howmany((hrtime_t)(usec) * hz, MICROSEC)) #define NSEC_TO_TICK(nsec) (howmany((hrtime_t)(nsec) * hz, NANOSEC)) #define max_ncpus 64 #define boot_ncpus (sysconf(_SC_NPROCESSORS_ONLN)) /* * Process priorities as defined by setpriority(2) and getpriority(2). */ #define minclsyspri 19 #define maxclsyspri -20 #define defclsyspri 0 #define CPU_SEQID ((uintptr_t)pthread_self() & (max_ncpus - 1)) #define CPU_SEQID_UNSTABLE CPU_SEQID #define kcred NULL #define CRED() NULL #define ptob(x) ((x) * PAGESIZE) #define NN_DIVISOR_1000 (1U << 0) #define NN_NUMBUF_SZ (6) extern uint64_t physmem; extern const char *random_path; extern const char *urandom_path; extern int highbit64(uint64_t i); extern int lowbit64(uint64_t i); extern int random_get_bytes(uint8_t *ptr, size_t len); extern int random_get_pseudo_bytes(uint8_t *ptr, size_t len); static __inline__ uint32_t random_in_range(uint32_t range) { uint32_t r; ASSERT(range != 0); if (range == 1) return (0); (void) random_get_pseudo_bytes((uint8_t *)&r, sizeof (r)); return (r % range); } extern void kernel_init(int mode); extern void kernel_fini(void); extern void random_init(void); extern void random_fini(void); struct spa; extern void show_pool_stats(struct spa *); extern int set_global_var(char const *arg); typedef struct callb_cpr { kmutex_t *cc_lockp; } callb_cpr_t; #define CALLB_CPR_INIT(cp, lockp, func, name) { \ (cp)->cc_lockp = lockp; \ } #define CALLB_CPR_SAFE_BEGIN(cp) { \ ASSERT(MUTEX_HELD((cp)->cc_lockp)); \ } #define CALLB_CPR_SAFE_END(cp, lockp) { \ ASSERT(MUTEX_HELD((cp)->cc_lockp)); \ } #define CALLB_CPR_EXIT(cp) { \ ASSERT(MUTEX_HELD((cp)->cc_lockp)); \ mutex_exit((cp)->cc_lockp); \ } #define zone_dataset_visible(x, y) (1) #define INGLOBALZONE(z) (1) extern uint32_t zone_get_hostid(void *zonep); extern char *kmem_vasprintf(const char *fmt, va_list adx); extern char *kmem_asprintf(const char *fmt, ...); #define kmem_strfree(str) kmem_free((str), strlen(str) + 1) #define kmem_strdup(s) strdup(s) #ifndef __cplusplus extern int kmem_scnprintf(char *restrict str, size_t size, const char *restrict fmt, ...); #endif /* * Hostname information */ extern int ddi_strtoull(const char *str, char **nptr, int base, u_longlong_t *result); typedef struct utsname utsname_t; extern utsname_t *utsname(void); /* ZFS Boot Related stuff. */ struct _buf { intptr_t _fd; }; struct bootstat { uint64_t st_size; }; typedef struct ace_object { uid_t a_who; uint32_t a_access_mask; uint16_t a_flags; uint16_t a_type; uint8_t a_obj_type[16]; uint8_t a_inherit_obj_type[16]; } ace_object_t; #define ACE_ACCESS_ALLOWED_OBJECT_ACE_TYPE 0x05 #define ACE_ACCESS_DENIED_OBJECT_ACE_TYPE 0x06 #define ACE_SYSTEM_AUDIT_OBJECT_ACE_TYPE 0x07 #define ACE_SYSTEM_ALARM_OBJECT_ACE_TYPE 0x08 extern int zfs_secpolicy_snapshot_perms(const char *name, cred_t *cr); extern int zfs_secpolicy_rename_perms(const char *from, const char *to, cred_t *cr); extern int zfs_secpolicy_destroy_perms(const char *name, cred_t *cr); extern int secpolicy_zfs(const cred_t *cr); extern int secpolicy_zfs_proc(const cred_t *cr, proc_t *proc); extern zoneid_t getzoneid(void); /* SID stuff */ typedef struct ksiddomain { uint_t kd_ref; uint_t kd_len; char *kd_name; } ksiddomain_t; ksiddomain_t *ksid_lookupdomain(const char *); void ksiddomain_rele(ksiddomain_t *); #define DDI_SLEEP KM_SLEEP #define ddi_log_sysevent(_a, _b, _c, _d, _e, _f, _g) \ sysevent_post_event(_c, _d, _b, "libzpool", _e, _f) #define zfs_sleep_until(wakeup) \ do { \ hrtime_t delta = wakeup - gethrtime(); \ struct timespec ts; \ ts.tv_sec = delta / NANOSEC; \ ts.tv_nsec = delta % NANOSEC; \ (void) nanosleep(&ts, NULL); \ } while (0) typedef int fstrans_cookie_t; extern fstrans_cookie_t spl_fstrans_mark(void); extern void spl_fstrans_unmark(fstrans_cookie_t); extern int __spl_pf_fstrans_check(void); extern int kmem_cache_reap_active(void); /* * Kernel modules */ #define __init #define __exit #endif /* _KERNEL || _STANDALONE */ #ifdef __cplusplus }; #endif #endif /* _SYS_ZFS_CONTEXT_H */ diff --git a/lib/libzpool/kernel.c b/lib/libzpool/kernel.c index ffad7fc02bc9..a3930ee07f73 100644 --- a/lib/libzpool/kernel.c +++ b/lib/libzpool/kernel.c @@ -1,1472 +1,1475 @@ /* * 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 https://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 */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2012, 2018 by Delphix. All rights reserved. * Copyright (c) 2016 Actifio, Inc. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * Emulation of kernel services in userland. */ uint64_t physmem; uint32_t hostid; struct utsname hw_utsname; /* If set, all blocks read will be copied to the specified directory. */ char *vn_dumpdir = NULL; /* this only exists to have its address taken */ struct proc p0; /* * ========================================================================= * threads * ========================================================================= * * TS_STACK_MIN is dictated by the minimum allowed pthread stack size. While * TS_STACK_MAX is somewhat arbitrary, it was selected to be large enough for * the expected stack depth while small enough to avoid exhausting address * space with high thread counts. */ #define TS_STACK_MIN MAX(PTHREAD_STACK_MIN, 32768) #define TS_STACK_MAX (256 * 1024) struct zk_thread_wrapper { void (*func)(void *); void *arg; }; static void * zk_thread_wrapper(void *arg) { struct zk_thread_wrapper ztw; memcpy(&ztw, arg, sizeof (ztw)); free(arg); ztw.func(ztw.arg); return (NULL); } kthread_t * -zk_thread_create(void (*func)(void *), void *arg, size_t stksize, int state) +zk_thread_create(const char *name, void (*func)(void *), void *arg, + size_t stksize, int state) { pthread_attr_t attr; pthread_t tid; char *stkstr; struct zk_thread_wrapper *ztw; int detachstate = PTHREAD_CREATE_DETACHED; VERIFY0(pthread_attr_init(&attr)); if (state & TS_JOINABLE) detachstate = PTHREAD_CREATE_JOINABLE; VERIFY0(pthread_attr_setdetachstate(&attr, detachstate)); /* * We allow the default stack size in user space to be specified by * setting the ZFS_STACK_SIZE environment variable. This allows us * the convenience of observing and debugging stack overruns in * user space. Explicitly specified stack sizes will be honored. * The usage of ZFS_STACK_SIZE is discussed further in the * ENVIRONMENT VARIABLES sections of the ztest(1) man page. */ if (stksize == 0) { stkstr = getenv("ZFS_STACK_SIZE"); if (stkstr == NULL) stksize = TS_STACK_MAX; else stksize = MAX(atoi(stkstr), TS_STACK_MIN); } VERIFY3S(stksize, >, 0); stksize = P2ROUNDUP(MAX(stksize, TS_STACK_MIN), PAGESIZE); /* * If this ever fails, it may be because the stack size is not a * multiple of system page size. */ VERIFY0(pthread_attr_setstacksize(&attr, stksize)); VERIFY0(pthread_attr_setguardsize(&attr, PAGESIZE)); VERIFY(ztw = malloc(sizeof (*ztw))); ztw->func = func; ztw->arg = arg; VERIFY0(pthread_create(&tid, &attr, zk_thread_wrapper, ztw)); VERIFY0(pthread_attr_destroy(&attr)); + pthread_setname_np(tid, name); + return ((void *)(uintptr_t)tid); } /* * ========================================================================= * kstats * ========================================================================= */ kstat_t * kstat_create(const char *module, int instance, const char *name, const char *class, uchar_t type, ulong_t ndata, uchar_t ks_flag) { (void) module, (void) instance, (void) name, (void) class, (void) type, (void) ndata, (void) ks_flag; return (NULL); } void kstat_install(kstat_t *ksp) { (void) ksp; } void kstat_delete(kstat_t *ksp) { (void) ksp; } void kstat_set_raw_ops(kstat_t *ksp, int (*headers)(char *buf, size_t size), int (*data)(char *buf, size_t size, void *data), void *(*addr)(kstat_t *ksp, loff_t index)) { (void) ksp, (void) headers, (void) data, (void) addr; } /* * ========================================================================= * mutexes * ========================================================================= */ void mutex_init(kmutex_t *mp, char *name, int type, void *cookie) { (void) name, (void) type, (void) cookie; VERIFY0(pthread_mutex_init(&mp->m_lock, NULL)); memset(&mp->m_owner, 0, sizeof (pthread_t)); } void mutex_destroy(kmutex_t *mp) { VERIFY0(pthread_mutex_destroy(&mp->m_lock)); } void mutex_enter(kmutex_t *mp) { VERIFY0(pthread_mutex_lock(&mp->m_lock)); mp->m_owner = pthread_self(); } int mutex_enter_check_return(kmutex_t *mp) { int error = pthread_mutex_lock(&mp->m_lock); if (error == 0) mp->m_owner = pthread_self(); return (error); } int mutex_tryenter(kmutex_t *mp) { int error = pthread_mutex_trylock(&mp->m_lock); if (error == 0) { mp->m_owner = pthread_self(); return (1); } else { VERIFY3S(error, ==, EBUSY); return (0); } } void mutex_exit(kmutex_t *mp) { memset(&mp->m_owner, 0, sizeof (pthread_t)); VERIFY0(pthread_mutex_unlock(&mp->m_lock)); } /* * ========================================================================= * rwlocks * ========================================================================= */ void rw_init(krwlock_t *rwlp, char *name, int type, void *arg) { (void) name, (void) type, (void) arg; VERIFY0(pthread_rwlock_init(&rwlp->rw_lock, NULL)); rwlp->rw_readers = 0; rwlp->rw_owner = 0; } void rw_destroy(krwlock_t *rwlp) { VERIFY0(pthread_rwlock_destroy(&rwlp->rw_lock)); } void rw_enter(krwlock_t *rwlp, krw_t rw) { if (rw == RW_READER) { VERIFY0(pthread_rwlock_rdlock(&rwlp->rw_lock)); atomic_inc_uint(&rwlp->rw_readers); } else { VERIFY0(pthread_rwlock_wrlock(&rwlp->rw_lock)); rwlp->rw_owner = pthread_self(); } } void rw_exit(krwlock_t *rwlp) { if (RW_READ_HELD(rwlp)) atomic_dec_uint(&rwlp->rw_readers); else rwlp->rw_owner = 0; VERIFY0(pthread_rwlock_unlock(&rwlp->rw_lock)); } int rw_tryenter(krwlock_t *rwlp, krw_t rw) { int error; if (rw == RW_READER) error = pthread_rwlock_tryrdlock(&rwlp->rw_lock); else error = pthread_rwlock_trywrlock(&rwlp->rw_lock); if (error == 0) { if (rw == RW_READER) atomic_inc_uint(&rwlp->rw_readers); else rwlp->rw_owner = pthread_self(); return (1); } VERIFY3S(error, ==, EBUSY); return (0); } uint32_t zone_get_hostid(void *zonep) { /* * We're emulating the system's hostid in userland. */ (void) zonep; return (hostid); } int rw_tryupgrade(krwlock_t *rwlp) { (void) rwlp; return (0); } /* * ========================================================================= * condition variables * ========================================================================= */ void cv_init(kcondvar_t *cv, char *name, int type, void *arg) { (void) name, (void) type, (void) arg; VERIFY0(pthread_cond_init(cv, NULL)); } void cv_destroy(kcondvar_t *cv) { VERIFY0(pthread_cond_destroy(cv)); } void cv_wait(kcondvar_t *cv, kmutex_t *mp) { memset(&mp->m_owner, 0, sizeof (pthread_t)); VERIFY0(pthread_cond_wait(cv, &mp->m_lock)); mp->m_owner = pthread_self(); } int cv_wait_sig(kcondvar_t *cv, kmutex_t *mp) { cv_wait(cv, mp); return (1); } int cv_timedwait(kcondvar_t *cv, kmutex_t *mp, clock_t abstime) { int error; struct timeval tv; struct timespec ts; clock_t delta; delta = abstime - ddi_get_lbolt(); if (delta <= 0) return (-1); VERIFY(gettimeofday(&tv, NULL) == 0); ts.tv_sec = tv.tv_sec + delta / hz; ts.tv_nsec = tv.tv_usec * NSEC_PER_USEC + (delta % hz) * (NANOSEC / hz); if (ts.tv_nsec >= NANOSEC) { ts.tv_sec++; ts.tv_nsec -= NANOSEC; } memset(&mp->m_owner, 0, sizeof (pthread_t)); error = pthread_cond_timedwait(cv, &mp->m_lock, &ts); mp->m_owner = pthread_self(); if (error == ETIMEDOUT) return (-1); VERIFY0(error); return (1); } int cv_timedwait_hires(kcondvar_t *cv, kmutex_t *mp, hrtime_t tim, hrtime_t res, int flag) { (void) res; int error; struct timeval tv; struct timespec ts; hrtime_t delta; ASSERT(flag == 0 || flag == CALLOUT_FLAG_ABSOLUTE); delta = tim; if (flag & CALLOUT_FLAG_ABSOLUTE) delta -= gethrtime(); if (delta <= 0) return (-1); VERIFY0(gettimeofday(&tv, NULL)); ts.tv_sec = tv.tv_sec + delta / NANOSEC; ts.tv_nsec = tv.tv_usec * NSEC_PER_USEC + (delta % NANOSEC); if (ts.tv_nsec >= NANOSEC) { ts.tv_sec++; ts.tv_nsec -= NANOSEC; } memset(&mp->m_owner, 0, sizeof (pthread_t)); error = pthread_cond_timedwait(cv, &mp->m_lock, &ts); mp->m_owner = pthread_self(); if (error == ETIMEDOUT) return (-1); VERIFY0(error); return (1); } void cv_signal(kcondvar_t *cv) { VERIFY0(pthread_cond_signal(cv)); } void cv_broadcast(kcondvar_t *cv) { VERIFY0(pthread_cond_broadcast(cv)); } /* * ========================================================================= * procfs list * ========================================================================= */ void seq_printf(struct seq_file *m, const char *fmt, ...) { (void) m, (void) fmt; } void procfs_list_install(const char *module, const char *submodule, const char *name, mode_t mode, procfs_list_t *procfs_list, int (*show)(struct seq_file *f, void *p), int (*show_header)(struct seq_file *f), int (*clear)(procfs_list_t *procfs_list), size_t procfs_list_node_off) { (void) module, (void) submodule, (void) name, (void) mode, (void) show, (void) show_header, (void) clear; mutex_init(&procfs_list->pl_lock, NULL, MUTEX_DEFAULT, NULL); list_create(&procfs_list->pl_list, procfs_list_node_off + sizeof (procfs_list_node_t), procfs_list_node_off + offsetof(procfs_list_node_t, pln_link)); procfs_list->pl_next_id = 1; procfs_list->pl_node_offset = procfs_list_node_off; } void procfs_list_uninstall(procfs_list_t *procfs_list) { (void) procfs_list; } void procfs_list_destroy(procfs_list_t *procfs_list) { ASSERT(list_is_empty(&procfs_list->pl_list)); list_destroy(&procfs_list->pl_list); mutex_destroy(&procfs_list->pl_lock); } #define NODE_ID(procfs_list, obj) \ (((procfs_list_node_t *)(((char *)obj) + \ (procfs_list)->pl_node_offset))->pln_id) void procfs_list_add(procfs_list_t *procfs_list, void *p) { ASSERT(MUTEX_HELD(&procfs_list->pl_lock)); NODE_ID(procfs_list, p) = procfs_list->pl_next_id++; list_insert_tail(&procfs_list->pl_list, p); } /* * ========================================================================= * vnode operations * ========================================================================= */ /* * ========================================================================= * Figure out which debugging statements to print * ========================================================================= */ static char *dprintf_string; static int dprintf_print_all; int dprintf_find_string(const char *string) { char *tmp_str = dprintf_string; int len = strlen(string); /* * Find out if this is a string we want to print. * String format: file1.c,function_name1,file2.c,file3.c */ while (tmp_str != NULL) { if (strncmp(tmp_str, string, len) == 0 && (tmp_str[len] == ',' || tmp_str[len] == '\0')) return (1); tmp_str = strchr(tmp_str, ','); if (tmp_str != NULL) tmp_str++; /* Get rid of , */ } return (0); } void dprintf_setup(int *argc, char **argv) { int i, j; /* * Debugging can be specified two ways: by setting the * environment variable ZFS_DEBUG, or by including a * "debug=..." argument on the command line. The command * line setting overrides the environment variable. */ for (i = 1; i < *argc; i++) { int len = strlen("debug="); /* First look for a command line argument */ if (strncmp("debug=", argv[i], len) == 0) { dprintf_string = argv[i] + len; /* Remove from args */ for (j = i; j < *argc; j++) argv[j] = argv[j+1]; argv[j] = NULL; (*argc)--; } } if (dprintf_string == NULL) { /* Look for ZFS_DEBUG environment variable */ dprintf_string = getenv("ZFS_DEBUG"); } /* * Are we just turning on all debugging? */ if (dprintf_find_string("on")) dprintf_print_all = 1; if (dprintf_string != NULL) zfs_flags |= ZFS_DEBUG_DPRINTF; } /* * ========================================================================= * debug printfs * ========================================================================= */ void __dprintf(boolean_t dprint, const char *file, const char *func, int line, const char *fmt, ...) { /* Get rid of annoying "../common/" prefix to filename. */ const char *newfile = zfs_basename(file); va_list adx; if (dprint) { /* dprintf messages are printed immediately */ if (!dprintf_print_all && !dprintf_find_string(newfile) && !dprintf_find_string(func)) return; /* Print out just the function name if requested */ flockfile(stdout); if (dprintf_find_string("pid")) (void) printf("%d ", getpid()); if (dprintf_find_string("tid")) (void) printf("%ju ", (uintmax_t)(uintptr_t)pthread_self()); if (dprintf_find_string("cpu")) (void) printf("%u ", getcpuid()); if (dprintf_find_string("time")) (void) printf("%llu ", gethrtime()); if (dprintf_find_string("long")) (void) printf("%s, line %d: ", newfile, line); (void) printf("dprintf: %s: ", func); va_start(adx, fmt); (void) vprintf(fmt, adx); va_end(adx); funlockfile(stdout); } else { /* zfs_dbgmsg is logged for dumping later */ size_t size; char *buf; int i; size = 1024; buf = umem_alloc(size, UMEM_NOFAIL); i = snprintf(buf, size, "%s:%d:%s(): ", newfile, line, func); if (i < size) { va_start(adx, fmt); (void) vsnprintf(buf + i, size - i, fmt, adx); va_end(adx); } __zfs_dbgmsg(buf); umem_free(buf, size); } } /* * ========================================================================= * cmn_err() and panic() * ========================================================================= */ static char ce_prefix[CE_IGNORE][10] = { "", "NOTICE: ", "WARNING: ", "" }; static char ce_suffix[CE_IGNORE][2] = { "", "\n", "\n", "" }; __attribute__((noreturn)) void vpanic(const char *fmt, va_list adx) { (void) fprintf(stderr, "error: "); (void) vfprintf(stderr, fmt, adx); (void) fprintf(stderr, "\n"); abort(); /* think of it as a "user-level crash dump" */ } __attribute__((noreturn)) void panic(const char *fmt, ...) { va_list adx; va_start(adx, fmt); vpanic(fmt, adx); va_end(adx); } void vcmn_err(int ce, const char *fmt, va_list adx) { if (ce == CE_PANIC) vpanic(fmt, adx); if (ce != CE_NOTE) { /* suppress noise in userland stress testing */ (void) fprintf(stderr, "%s", ce_prefix[ce]); (void) vfprintf(stderr, fmt, adx); (void) fprintf(stderr, "%s", ce_suffix[ce]); } } void cmn_err(int ce, const char *fmt, ...) { va_list adx; va_start(adx, fmt); vcmn_err(ce, fmt, adx); va_end(adx); } /* * ========================================================================= * misc routines * ========================================================================= */ void delay(clock_t ticks) { (void) poll(0, 0, ticks * (1000 / hz)); } /* * Find highest one bit set. * Returns bit number + 1 of highest bit that is set, otherwise returns 0. * The __builtin_clzll() function is supported by both GCC and Clang. */ int highbit64(uint64_t i) { if (i == 0) return (0); return (NBBY * sizeof (uint64_t) - __builtin_clzll(i)); } /* * Find lowest one bit set. * Returns bit number + 1 of lowest bit that is set, otherwise returns 0. * The __builtin_ffsll() function is supported by both GCC and Clang. */ int lowbit64(uint64_t i) { if (i == 0) return (0); return (__builtin_ffsll(i)); } const char *random_path = "/dev/random"; const char *urandom_path = "/dev/urandom"; static int random_fd = -1, urandom_fd = -1; void random_init(void) { VERIFY((random_fd = open(random_path, O_RDONLY | O_CLOEXEC)) != -1); VERIFY((urandom_fd = open(urandom_path, O_RDONLY | O_CLOEXEC)) != -1); } void random_fini(void) { close(random_fd); close(urandom_fd); random_fd = -1; urandom_fd = -1; } static int random_get_bytes_common(uint8_t *ptr, size_t len, int fd) { size_t resid = len; ssize_t bytes; ASSERT(fd != -1); while (resid != 0) { bytes = read(fd, ptr, resid); ASSERT3S(bytes, >=, 0); ptr += bytes; resid -= bytes; } return (0); } int random_get_bytes(uint8_t *ptr, size_t len) { return (random_get_bytes_common(ptr, len, random_fd)); } int random_get_pseudo_bytes(uint8_t *ptr, size_t len) { return (random_get_bytes_common(ptr, len, urandom_fd)); } int ddi_strtoull(const char *str, char **nptr, int base, u_longlong_t *result) { errno = 0; *result = strtoull(str, nptr, base); if (*result == 0) return (errno); return (0); } utsname_t * utsname(void) { return (&hw_utsname); } /* * ========================================================================= * kernel emulation setup & teardown * ========================================================================= */ static int umem_out_of_memory(void) { char errmsg[] = "out of memory -- generating core dump\n"; (void) fprintf(stderr, "%s", errmsg); abort(); return (0); } void kernel_init(int mode) { extern uint_t rrw_tsd_key; umem_nofail_callback(umem_out_of_memory); physmem = sysconf(_SC_PHYS_PAGES); dprintf("physmem = %llu pages (%.2f GB)\n", (u_longlong_t)physmem, (double)physmem * sysconf(_SC_PAGE_SIZE) / (1ULL << 30)); hostid = (mode & SPA_MODE_WRITE) ? get_system_hostid() : 0; random_init(); VERIFY0(uname(&hw_utsname)); system_taskq_init(); icp_init(); zstd_init(); spa_init((spa_mode_t)mode); fletcher_4_init(); tsd_create(&rrw_tsd_key, rrw_tsd_destroy); } void kernel_fini(void) { fletcher_4_fini(); spa_fini(); zstd_fini(); icp_fini(); system_taskq_fini(); random_fini(); } uid_t crgetuid(cred_t *cr) { (void) cr; return (0); } uid_t crgetruid(cred_t *cr) { (void) cr; return (0); } gid_t crgetgid(cred_t *cr) { (void) cr; return (0); } int crgetngroups(cred_t *cr) { (void) cr; return (0); } gid_t * crgetgroups(cred_t *cr) { (void) cr; return (NULL); } int zfs_secpolicy_snapshot_perms(const char *name, cred_t *cr) { (void) name, (void) cr; return (0); } int zfs_secpolicy_rename_perms(const char *from, const char *to, cred_t *cr) { (void) from, (void) to, (void) cr; return (0); } int zfs_secpolicy_destroy_perms(const char *name, cred_t *cr) { (void) name, (void) cr; return (0); } int secpolicy_zfs(const cred_t *cr) { (void) cr; return (0); } int secpolicy_zfs_proc(const cred_t *cr, proc_t *proc) { (void) cr, (void) proc; return (0); } ksiddomain_t * ksid_lookupdomain(const char *dom) { ksiddomain_t *kd; kd = umem_zalloc(sizeof (ksiddomain_t), UMEM_NOFAIL); kd->kd_name = spa_strdup(dom); return (kd); } void ksiddomain_rele(ksiddomain_t *ksid) { spa_strfree(ksid->kd_name); umem_free(ksid, sizeof (ksiddomain_t)); } char * kmem_vasprintf(const char *fmt, va_list adx) { char *buf = NULL; va_list adx_copy; va_copy(adx_copy, adx); VERIFY(vasprintf(&buf, fmt, adx_copy) != -1); va_end(adx_copy); return (buf); } char * kmem_asprintf(const char *fmt, ...) { char *buf = NULL; va_list adx; va_start(adx, fmt); VERIFY(vasprintf(&buf, fmt, adx) != -1); va_end(adx); return (buf); } /* * kmem_scnprintf() will return the number of characters that it would have * printed whenever it is limited by value of the size variable, rather than * the number of characters that it did print. This can cause misbehavior on * subsequent uses of the return value, so we define a safe version that will * return the number of characters actually printed, minus the NULL format * character. Subsequent use of this by the safe string functions is safe * whether it is snprintf(), strlcat() or strlcpy(). */ int kmem_scnprintf(char *restrict str, size_t size, const char *restrict fmt, ...) { int n; va_list ap; /* Make the 0 case a no-op so that we do not return -1 */ if (size == 0) return (0); va_start(ap, fmt); n = vsnprintf(str, size, fmt, ap); va_end(ap); if (n >= size) n = size - 1; return (n); } zfs_file_t * zfs_onexit_fd_hold(int fd, minor_t *minorp) { (void) fd; *minorp = 0; return (NULL); } void zfs_onexit_fd_rele(zfs_file_t *fp) { (void) fp; } int zfs_onexit_add_cb(minor_t minor, void (*func)(void *), void *data, uintptr_t *action_handle) { (void) minor, (void) func, (void) data, (void) action_handle; return (0); } fstrans_cookie_t spl_fstrans_mark(void) { return ((fstrans_cookie_t)0); } void spl_fstrans_unmark(fstrans_cookie_t cookie) { (void) cookie; } int __spl_pf_fstrans_check(void) { return (0); } int kmem_cache_reap_active(void) { return (0); } void zvol_create_minor(const char *name) { (void) name; } void zvol_create_minors_recursive(const char *name) { (void) name; } void zvol_remove_minors(spa_t *spa, const char *name, boolean_t async) { (void) spa, (void) name, (void) async; } void zvol_rename_minors(spa_t *spa, const char *oldname, const char *newname, boolean_t async) { (void) spa, (void) oldname, (void) newname, (void) async; } /* * Open file * * path - fully qualified path to file * flags - file attributes O_READ / O_WRITE / O_EXCL * fpp - pointer to return file pointer * * Returns 0 on success underlying error on failure. */ int zfs_file_open(const char *path, int flags, int mode, zfs_file_t **fpp) { int fd = -1; int dump_fd = -1; int err; int old_umask = 0; zfs_file_t *fp; struct stat64 st; if (!(flags & O_CREAT) && stat64(path, &st) == -1) return (errno); if (!(flags & O_CREAT) && S_ISBLK(st.st_mode)) flags |= O_DIRECT; if (flags & O_CREAT) old_umask = umask(0); fd = open64(path, flags, mode); if (fd == -1) return (errno); if (flags & O_CREAT) (void) umask(old_umask); if (vn_dumpdir != NULL) { char *dumppath = umem_zalloc(MAXPATHLEN, UMEM_NOFAIL); const char *inpath = zfs_basename(path); (void) snprintf(dumppath, MAXPATHLEN, "%s/%s", vn_dumpdir, inpath); dump_fd = open64(dumppath, O_CREAT | O_WRONLY, 0666); umem_free(dumppath, MAXPATHLEN); if (dump_fd == -1) { err = errno; close(fd); return (err); } } else { dump_fd = -1; } (void) fcntl(fd, F_SETFD, FD_CLOEXEC); fp = umem_zalloc(sizeof (zfs_file_t), UMEM_NOFAIL); fp->f_fd = fd; fp->f_dump_fd = dump_fd; *fpp = fp; return (0); } void zfs_file_close(zfs_file_t *fp) { close(fp->f_fd); if (fp->f_dump_fd != -1) close(fp->f_dump_fd); umem_free(fp, sizeof (zfs_file_t)); } /* * Stateful write - use os internal file pointer to determine where to * write and update on successful completion. * * fp - pointer to file (pipe, socket, etc) to write to * buf - buffer to write * count - # of bytes to write * resid - pointer to count of unwritten bytes (if short write) * * Returns 0 on success errno on failure. */ int zfs_file_write(zfs_file_t *fp, const void *buf, size_t count, ssize_t *resid) { ssize_t rc; rc = write(fp->f_fd, buf, count); if (rc < 0) return (errno); if (resid) { *resid = count - rc; } else if (rc != count) { return (EIO); } return (0); } /* * Stateless write - os internal file pointer is not updated. * * fp - pointer to file (pipe, socket, etc) to write to * buf - buffer to write * count - # of bytes to write * off - file offset to write to (only valid for seekable types) * resid - pointer to count of unwritten bytes * * Returns 0 on success errno on failure. */ int zfs_file_pwrite(zfs_file_t *fp, const void *buf, size_t count, loff_t pos, ssize_t *resid) { ssize_t rc, split, done; int sectors; /* * To simulate partial disk writes, we split writes into two * system calls so that the process can be killed in between. * This is used by ztest to simulate realistic failure modes. */ sectors = count >> SPA_MINBLOCKSHIFT; split = (sectors > 0 ? rand() % sectors : 0) << SPA_MINBLOCKSHIFT; rc = pwrite64(fp->f_fd, buf, split, pos); if (rc != -1) { done = rc; rc = pwrite64(fp->f_fd, (char *)buf + split, count - split, pos + split); } #ifdef __linux__ if (rc == -1 && errno == EINVAL) { /* * Under Linux, this most likely means an alignment issue * (memory or disk) due to O_DIRECT, so we abort() in order * to catch the offender. */ abort(); } #endif if (rc < 0) return (errno); done += rc; if (resid) { *resid = count - done; } else if (done != count) { return (EIO); } return (0); } /* * Stateful read - use os internal file pointer to determine where to * read and update on successful completion. * * fp - pointer to file (pipe, socket, etc) to read from * buf - buffer to write * count - # of bytes to read * resid - pointer to count of unread bytes (if short read) * * Returns 0 on success errno on failure. */ int zfs_file_read(zfs_file_t *fp, void *buf, size_t count, ssize_t *resid) { int rc; rc = read(fp->f_fd, buf, count); if (rc < 0) return (errno); if (resid) { *resid = count - rc; } else if (rc != count) { return (EIO); } return (0); } /* * Stateless read - os internal file pointer is not updated. * * fp - pointer to file (pipe, socket, etc) to read from * buf - buffer to write * count - # of bytes to write * off - file offset to read from (only valid for seekable types) * resid - pointer to count of unwritten bytes (if short write) * * Returns 0 on success errno on failure. */ int zfs_file_pread(zfs_file_t *fp, void *buf, size_t count, loff_t off, ssize_t *resid) { ssize_t rc; rc = pread64(fp->f_fd, buf, count, off); if (rc < 0) { #ifdef __linux__ /* * Under Linux, this most likely means an alignment issue * (memory or disk) due to O_DIRECT, so we abort() in order to * catch the offender. */ if (errno == EINVAL) abort(); #endif return (errno); } if (fp->f_dump_fd != -1) { int status; status = pwrite64(fp->f_dump_fd, buf, rc, off); ASSERT(status != -1); } if (resid) { *resid = count - rc; } else if (rc != count) { return (EIO); } return (0); } /* * lseek - set / get file pointer * * fp - pointer to file (pipe, socket, etc) to read from * offp - value to seek to, returns current value plus passed offset * whence - see man pages for standard lseek whence values * * Returns 0 on success errno on failure (ESPIPE for non seekable types) */ int zfs_file_seek(zfs_file_t *fp, loff_t *offp, int whence) { loff_t rc; rc = lseek(fp->f_fd, *offp, whence); if (rc < 0) return (errno); *offp = rc; return (0); } /* * Get file attributes * * filp - file pointer * zfattr - pointer to file attr structure * * Currently only used for fetching size and file mode * * Returns 0 on success or error code of underlying getattr call on failure. */ int zfs_file_getattr(zfs_file_t *fp, zfs_file_attr_t *zfattr) { struct stat64 st; if (fstat64_blk(fp->f_fd, &st) == -1) return (errno); zfattr->zfa_size = st.st_size; zfattr->zfa_mode = st.st_mode; return (0); } /* * Sync file to disk * * filp - file pointer * flags - O_SYNC and or O_DSYNC * * Returns 0 on success or error code of underlying sync call on failure. */ int zfs_file_fsync(zfs_file_t *fp, int flags) { (void) flags; if (fsync(fp->f_fd) < 0) return (errno); return (0); } /* * fallocate - allocate or free space on disk * * fp - file pointer * mode (non-standard options for hole punching etc) * offset - offset to start allocating or freeing from * len - length to free / allocate * * OPTIONAL */ int zfs_file_fallocate(zfs_file_t *fp, int mode, loff_t offset, loff_t len) { #ifdef __linux__ return (fallocate(fp->f_fd, mode, offset, len)); #else (void) fp, (void) mode, (void) offset, (void) len; return (EOPNOTSUPP); #endif } /* * Request current file pointer offset * * fp - pointer to file * * Returns current file offset. */ loff_t zfs_file_off(zfs_file_t *fp) { return (lseek(fp->f_fd, SEEK_CUR, 0)); } /* * unlink file * * path - fully qualified file path * * Returns 0 on success. * * OPTIONAL */ int zfs_file_unlink(const char *path) { return (remove(path)); } /* * Get reference to file pointer * * fd - input file descriptor * * Returns pointer to file struct or NULL. * Unsupported in user space. */ zfs_file_t * zfs_file_get(int fd) { (void) fd; abort(); return (NULL); } /* * Drop reference to file pointer * * fp - pointer to file struct * * Unsupported in user space. */ void zfs_file_put(zfs_file_t *fp) { abort(); (void) fp; } void zfsvfs_update_fromname(const char *oldname, const char *newname) { (void) oldname, (void) newname; } void spa_import_os(spa_t *spa) { (void) spa; } void spa_export_os(spa_t *spa) { (void) spa; } void spa_activate_os(spa_t *spa) { (void) spa; } void spa_deactivate_os(spa_t *spa) { (void) spa; } diff --git a/lib/libzpool/taskq.c b/lib/libzpool/taskq.c index 99a181ec3c93..5fb2283cf0b1 100644 --- a/lib/libzpool/taskq.c +++ b/lib/libzpool/taskq.c @@ -1,416 +1,416 @@ /* * 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 https://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 */ /* * Copyright 2010 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ /* * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright 2012 Garrett D'Amore . All rights reserved. * Copyright (c) 2014 by Delphix. All rights reserved. */ #include int taskq_now; taskq_t *system_taskq; taskq_t *system_delay_taskq; static pthread_key_t taskq_tsd; #define TASKQ_ACTIVE 0x00010000 static taskq_ent_t * task_alloc(taskq_t *tq, int tqflags) { taskq_ent_t *t; int rv; again: if ((t = tq->tq_freelist) != NULL && tq->tq_nalloc >= tq->tq_minalloc) { ASSERT(!(t->tqent_flags & TQENT_FLAG_PREALLOC)); tq->tq_freelist = t->tqent_next; } else { if (tq->tq_nalloc >= tq->tq_maxalloc) { if (!(tqflags & KM_SLEEP)) return (NULL); /* * We don't want to exceed tq_maxalloc, but we can't * wait for other tasks to complete (and thus free up * task structures) without risking deadlock with * the caller. So, we just delay for one second * to throttle the allocation rate. If we have tasks * complete before one second timeout expires then * taskq_ent_free will signal us and we will * immediately retry the allocation. */ tq->tq_maxalloc_wait++; rv = cv_timedwait(&tq->tq_maxalloc_cv, &tq->tq_lock, ddi_get_lbolt() + hz); tq->tq_maxalloc_wait--; if (rv > 0) goto again; /* signaled */ } mutex_exit(&tq->tq_lock); t = kmem_alloc(sizeof (taskq_ent_t), tqflags); mutex_enter(&tq->tq_lock); if (t != NULL) { /* Make sure we start without any flags */ t->tqent_flags = 0; tq->tq_nalloc++; } } return (t); } static void task_free(taskq_t *tq, taskq_ent_t *t) { if (tq->tq_nalloc <= tq->tq_minalloc) { t->tqent_next = tq->tq_freelist; tq->tq_freelist = t; } else { tq->tq_nalloc--; mutex_exit(&tq->tq_lock); kmem_free(t, sizeof (taskq_ent_t)); mutex_enter(&tq->tq_lock); } if (tq->tq_maxalloc_wait) cv_signal(&tq->tq_maxalloc_cv); } taskqid_t taskq_dispatch(taskq_t *tq, task_func_t func, void *arg, uint_t tqflags) { taskq_ent_t *t; if (taskq_now) { func(arg); return (1); } mutex_enter(&tq->tq_lock); ASSERT(tq->tq_flags & TASKQ_ACTIVE); if ((t = task_alloc(tq, tqflags)) == NULL) { mutex_exit(&tq->tq_lock); return (0); } if (tqflags & TQ_FRONT) { t->tqent_next = tq->tq_task.tqent_next; t->tqent_prev = &tq->tq_task; } else { t->tqent_next = &tq->tq_task; t->tqent_prev = tq->tq_task.tqent_prev; } t->tqent_next->tqent_prev = t; t->tqent_prev->tqent_next = t; t->tqent_func = func; t->tqent_arg = arg; t->tqent_flags = 0; cv_signal(&tq->tq_dispatch_cv); mutex_exit(&tq->tq_lock); return (1); } taskqid_t taskq_dispatch_delay(taskq_t *tq, task_func_t func, void *arg, uint_t tqflags, clock_t expire_time) { (void) tq, (void) func, (void) arg, (void) tqflags, (void) expire_time; return (0); } int taskq_empty_ent(taskq_ent_t *t) { return (t->tqent_next == NULL); } void taskq_init_ent(taskq_ent_t *t) { t->tqent_next = NULL; t->tqent_prev = NULL; t->tqent_func = NULL; t->tqent_arg = NULL; t->tqent_flags = 0; } void taskq_dispatch_ent(taskq_t *tq, task_func_t func, void *arg, uint_t flags, taskq_ent_t *t) { ASSERT(func != NULL); /* * Mark it as a prealloc'd task. This is important * to ensure that we don't free it later. */ t->tqent_flags |= TQENT_FLAG_PREALLOC; /* * Enqueue the task to the underlying queue. */ mutex_enter(&tq->tq_lock); if (flags & TQ_FRONT) { t->tqent_next = tq->tq_task.tqent_next; t->tqent_prev = &tq->tq_task; } else { t->tqent_next = &tq->tq_task; t->tqent_prev = tq->tq_task.tqent_prev; } t->tqent_next->tqent_prev = t; t->tqent_prev->tqent_next = t; t->tqent_func = func; t->tqent_arg = arg; cv_signal(&tq->tq_dispatch_cv); mutex_exit(&tq->tq_lock); } void taskq_wait(taskq_t *tq) { mutex_enter(&tq->tq_lock); while (tq->tq_task.tqent_next != &tq->tq_task || tq->tq_active != 0) cv_wait(&tq->tq_wait_cv, &tq->tq_lock); mutex_exit(&tq->tq_lock); } void taskq_wait_id(taskq_t *tq, taskqid_t id) { (void) id; taskq_wait(tq); } void taskq_wait_outstanding(taskq_t *tq, taskqid_t id) { (void) id; taskq_wait(tq); } static __attribute__((noreturn)) void taskq_thread(void *arg) { taskq_t *tq = arg; taskq_ent_t *t; boolean_t prealloc; VERIFY0(pthread_setspecific(taskq_tsd, tq)); mutex_enter(&tq->tq_lock); while (tq->tq_flags & TASKQ_ACTIVE) { if ((t = tq->tq_task.tqent_next) == &tq->tq_task) { if (--tq->tq_active == 0) cv_broadcast(&tq->tq_wait_cv); cv_wait(&tq->tq_dispatch_cv, &tq->tq_lock); tq->tq_active++; continue; } t->tqent_prev->tqent_next = t->tqent_next; t->tqent_next->tqent_prev = t->tqent_prev; t->tqent_next = NULL; t->tqent_prev = NULL; prealloc = t->tqent_flags & TQENT_FLAG_PREALLOC; mutex_exit(&tq->tq_lock); rw_enter(&tq->tq_threadlock, RW_READER); t->tqent_func(t->tqent_arg); rw_exit(&tq->tq_threadlock); mutex_enter(&tq->tq_lock); if (!prealloc) task_free(tq, t); } tq->tq_nthreads--; cv_broadcast(&tq->tq_wait_cv); mutex_exit(&tq->tq_lock); thread_exit(); } taskq_t * taskq_create(const char *name, int nthreads, pri_t pri, int minalloc, int maxalloc, uint_t flags) { (void) pri; taskq_t *tq = kmem_zalloc(sizeof (taskq_t), KM_SLEEP); int t; if (flags & TASKQ_THREADS_CPU_PCT) { int pct; ASSERT3S(nthreads, >=, 0); ASSERT3S(nthreads, <=, 100); pct = MIN(nthreads, 100); pct = MAX(pct, 0); nthreads = (sysconf(_SC_NPROCESSORS_ONLN) * pct) / 100; nthreads = MAX(nthreads, 1); /* need at least 1 thread */ } else { ASSERT3S(nthreads, >=, 1); } rw_init(&tq->tq_threadlock, NULL, RW_DEFAULT, NULL); mutex_init(&tq->tq_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&tq->tq_dispatch_cv, NULL, CV_DEFAULT, NULL); cv_init(&tq->tq_wait_cv, NULL, CV_DEFAULT, NULL); cv_init(&tq->tq_maxalloc_cv, NULL, CV_DEFAULT, NULL); (void) strlcpy(tq->tq_name, name, sizeof (tq->tq_name)); tq->tq_flags = flags | TASKQ_ACTIVE; tq->tq_active = nthreads; tq->tq_nthreads = nthreads; tq->tq_minalloc = minalloc; tq->tq_maxalloc = maxalloc; tq->tq_task.tqent_next = &tq->tq_task; tq->tq_task.tqent_prev = &tq->tq_task; tq->tq_threadlist = kmem_alloc(nthreads * sizeof (kthread_t *), KM_SLEEP); if (flags & TASKQ_PREPOPULATE) { mutex_enter(&tq->tq_lock); while (minalloc-- > 0) task_free(tq, task_alloc(tq, KM_SLEEP)); mutex_exit(&tq->tq_lock); } for (t = 0; t < nthreads; t++) - VERIFY((tq->tq_threadlist[t] = thread_create(NULL, 0, - taskq_thread, tq, 0, &p0, TS_RUN, pri)) != NULL); + VERIFY((tq->tq_threadlist[t] = thread_create_named(tq->tq_name, + NULL, 0, taskq_thread, tq, 0, &p0, TS_RUN, pri)) != NULL); return (tq); } void taskq_destroy(taskq_t *tq) { int nthreads = tq->tq_nthreads; taskq_wait(tq); mutex_enter(&tq->tq_lock); tq->tq_flags &= ~TASKQ_ACTIVE; cv_broadcast(&tq->tq_dispatch_cv); while (tq->tq_nthreads != 0) cv_wait(&tq->tq_wait_cv, &tq->tq_lock); tq->tq_minalloc = 0; while (tq->tq_nalloc != 0) { ASSERT(tq->tq_freelist != NULL); taskq_ent_t *tqent_nexttq = tq->tq_freelist->tqent_next; task_free(tq, tq->tq_freelist); tq->tq_freelist = tqent_nexttq; } mutex_exit(&tq->tq_lock); kmem_free(tq->tq_threadlist, nthreads * sizeof (kthread_t *)); rw_destroy(&tq->tq_threadlock); mutex_destroy(&tq->tq_lock); cv_destroy(&tq->tq_dispatch_cv); cv_destroy(&tq->tq_wait_cv); cv_destroy(&tq->tq_maxalloc_cv); kmem_free(tq, sizeof (taskq_t)); } /* * Create a taskq with a specified number of pool threads. Allocate * and return an array of nthreads kthread_t pointers, one for each * thread in the pool. The array is not ordered and must be freed * by the caller. */ taskq_t * taskq_create_synced(const char *name, int nthreads, pri_t pri, int minalloc, int maxalloc, uint_t flags, kthread_t ***ktpp) { taskq_t *tq; kthread_t **kthreads = kmem_zalloc(sizeof (*kthreads) * nthreads, KM_SLEEP); (void) pri; (void) minalloc; (void) maxalloc; flags &= ~(TASKQ_DYNAMIC | TASKQ_THREADS_CPU_PCT | TASKQ_DC_BATCH); tq = taskq_create(name, nthreads, minclsyspri, nthreads, INT_MAX, flags | TASKQ_PREPOPULATE); VERIFY(tq != NULL); VERIFY(tq->tq_nthreads == nthreads); for (int i = 0; i < nthreads; i++) { kthreads[i] = tq->tq_threadlist[i]; } *ktpp = kthreads; return (tq); } int taskq_member(taskq_t *tq, kthread_t *t) { int i; if (taskq_now) return (1); for (i = 0; i < tq->tq_nthreads; i++) if (tq->tq_threadlist[i] == t) return (1); return (0); } taskq_t * taskq_of_curthread(void) { return (pthread_getspecific(taskq_tsd)); } int taskq_cancel_id(taskq_t *tq, taskqid_t id) { (void) tq, (void) id; return (ENOENT); } void system_taskq_init(void) { VERIFY0(pthread_key_create(&taskq_tsd, NULL)); system_taskq = taskq_create("system_taskq", 64, maxclsyspri, 4, 512, TASKQ_DYNAMIC | TASKQ_PREPOPULATE); system_delay_taskq = taskq_create("delay_taskq", 4, maxclsyspri, 4, 512, TASKQ_DYNAMIC | TASKQ_PREPOPULATE); } void system_taskq_fini(void) { taskq_destroy(system_taskq); system_taskq = NULL; /* defensive */ taskq_destroy(system_delay_taskq); system_delay_taskq = NULL; VERIFY0(pthread_key_delete(taskq_tsd)); }