4 * The contents of this file are subject to the terms of the
5 * Common Development and Distribution License (the "License").
6 * You may not use this file except in compliance with the License.
8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 * or http://www.opensolaris.org/os/licensing.
10 * See the License for the specific language governing permissions
11 * and limitations under the License.
13 * When distributing Covered Code, include this CDDL HEADER in each
14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 * If applicable, add the following below this CDDL HEADER, with the
16 * fields enclosed by brackets "[]" replaced with your own identifying
17 * information: Portions Copyright [yyyy] [name of copyright owner]
22 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
32 #include <sys/signal.h>
35 #include <sys/processor.h>
36 #include <sys/zfs_context.h>
37 #include <sys/rrwlock.h>
38 #include <sys/utsname.h>
40 #include <sys/systeminfo.h>
43 * Emulation of kernel services in userland.
48 vnode_t
*rootdir
= (vnode_t
*)0xabcd1234;
49 char hw_serial
[HW_HOSTID_LEN
];
51 struct utsname utsname
= {
52 "userland", "libzpool", "1", "1", "na"
55 /* this only exists to have its address taken */
59 * =========================================================================
61 * =========================================================================
64 pthread_cond_t kthread_cond
= PTHREAD_COND_INITIALIZER
;
65 pthread_mutex_t kthread_lock
= PTHREAD_MUTEX_INITIALIZER
;
66 pthread_key_t kthread_key
;
74 VERIFY3S(pthread_key_create(&kthread_key
, NULL
), ==, 0);
76 /* Create entry for primary kthread */
77 kt
= umem_zalloc(sizeof (kthread_t
), UMEM_NOFAIL
);
78 kt
->t_tid
= pthread_self();
81 VERIFY3S(pthread_setspecific(kthread_key
, kt
), ==, 0);
83 /* Only the main thread should be running at the moment */
84 ASSERT3S(kthread_nr
, ==, 0);
91 kthread_t
*kt
= curthread
;
93 ASSERT(pthread_equal(kt
->t_tid
, pthread_self()));
94 ASSERT3P(kt
->t_func
, ==, NULL
);
96 umem_free(kt
, sizeof (kthread_t
));
98 /* Wait for all threads to exit via thread_exit() */
99 VERIFY3S(pthread_mutex_lock(&kthread_lock
), ==, 0);
101 kthread_nr
--; /* Main thread is exiting */
103 while (kthread_nr
> 0)
104 VERIFY3S(pthread_cond_wait(&kthread_cond
, &kthread_lock
), ==,
107 ASSERT3S(kthread_nr
, ==, 0);
108 VERIFY3S(pthread_mutex_unlock(&kthread_lock
), ==, 0);
110 VERIFY3S(pthread_key_delete(kthread_key
), ==, 0);
114 zk_thread_current(void)
116 kthread_t
*kt
= pthread_getspecific(kthread_key
);
118 ASSERT3P(kt
, !=, NULL
);
124 zk_thread_helper(void *arg
)
126 kthread_t
*kt
= (kthread_t
*) arg
;
128 VERIFY3S(pthread_setspecific(kthread_key
, kt
), ==, 0);
130 VERIFY3S(pthread_mutex_lock(&kthread_lock
), ==, 0);
132 VERIFY3S(pthread_mutex_unlock(&kthread_lock
), ==, 0);
134 kt
->t_tid
= pthread_self();
135 ((thread_func_arg_t
) kt
->t_func
)(kt
->t_arg
);
137 /* Unreachable, thread must exit with thread_exit() */
144 zk_thread_create(caddr_t stk
, size_t stksize
, thread_func_t func
, void *arg
,
145 size_t len
, proc_t
*pp
, int state
, pri_t pri
, int detachstate
)
151 ASSERT3S(state
& ~TS_RUN
, ==, 0);
153 kt
= umem_zalloc(sizeof (kthread_t
), UMEM_NOFAIL
);
158 * The Solaris kernel stack size is 24k for x86/x86_64.
159 * The Linux kernel stack size is 8k for x86/x86_64.
161 * We reduce the default stack size in userspace, to ensure
162 * we observe stack overruns in user space as well as in
163 * kernel space. In practice we can't set the userspace stack
164 * size to 8k because differences in stack usage between kernel
165 * space and userspace could lead to spurious stack overflows
166 * (especially when debugging is enabled). Nevertheless, we try
167 * to set it to the lowest value that works (currently 8k*4).
168 * PTHREAD_STACK_MIN is the minimum stack required for a NULL
169 * procedure in user space and is added in to the stack
172 * Some buggy NPTL threading implementations include the
173 * guard area within the stack size allocations. In
174 * this case we allocate an extra page to account for the
175 * guard area since we only have two pages of usable stack
179 stack
= PTHREAD_STACK_MIN
+ MAX(stksize
, STACK_SIZE
) * 4;
181 VERIFY3S(pthread_attr_init(&attr
), ==, 0);
182 VERIFY3S(pthread_attr_setstacksize(&attr
, stack
), ==, 0);
183 VERIFY3S(pthread_attr_setguardsize(&attr
, PAGESIZE
), ==, 0);
184 VERIFY3S(pthread_attr_setdetachstate(&attr
, detachstate
), ==, 0);
186 VERIFY3S(pthread_create(&kt
->t_tid
, &attr
, &zk_thread_helper
, kt
),
189 VERIFY3S(pthread_attr_destroy(&attr
), ==, 0);
197 kthread_t
*kt
= curthread
;
199 ASSERT(pthread_equal(kt
->t_tid
, pthread_self()));
201 umem_free(kt
, sizeof (kthread_t
));
203 pthread_mutex_lock(&kthread_lock
);
205 pthread_mutex_unlock(&kthread_lock
);
207 pthread_cond_broadcast(&kthread_cond
);
208 pthread_exit((void *)TS_MAGIC
);
212 zk_thread_join(kt_did_t tid
)
216 pthread_join((pthread_t
)tid
, &ret
);
217 VERIFY3P(ret
, ==, (void *)TS_MAGIC
);
221 * =========================================================================
223 * =========================================================================
227 kstat_create(const char *module
, int instance
, const char *name
,
228 const char *class, uchar_t type
, ulong_t ndata
, uchar_t ks_flag
)
235 kstat_install(kstat_t
*ksp
)
240 kstat_delete(kstat_t
*ksp
)
245 kstat_waitq_enter(kstat_io_t
*kiop
)
250 kstat_waitq_exit(kstat_io_t
*kiop
)
255 kstat_runq_enter(kstat_io_t
*kiop
)
260 kstat_runq_exit(kstat_io_t
*kiop
)
265 kstat_waitq_to_runq(kstat_io_t
*kiop
)
270 kstat_runq_back_to_waitq(kstat_io_t
*kiop
)
274 kstat_set_raw_ops(kstat_t
*ksp
,
275 int (*headers
)(char *buf
, size_t size
),
276 int (*data
)(char *buf
, size_t size
, void *data
),
277 void *(*addr
)(kstat_t
*ksp
, loff_t index
))
281 * =========================================================================
283 * =========================================================================
287 mutex_init(kmutex_t
*mp
, char *name
, int type
, void *cookie
)
289 ASSERT3S(type
, ==, MUTEX_DEFAULT
);
290 ASSERT3P(cookie
, ==, NULL
);
291 mp
->m_owner
= MTX_INIT
;
292 mp
->m_magic
= MTX_MAGIC
;
293 VERIFY3S(pthread_mutex_init(&mp
->m_lock
, NULL
), ==, 0);
297 mutex_destroy(kmutex_t
*mp
)
299 ASSERT3U(mp
->m_magic
, ==, MTX_MAGIC
);
300 ASSERT3P(mp
->m_owner
, ==, MTX_INIT
);
301 VERIFY3S(pthread_mutex_destroy(&(mp
)->m_lock
), ==, 0);
302 mp
->m_owner
= MTX_DEST
;
307 mutex_enter(kmutex_t
*mp
)
309 ASSERT3U(mp
->m_magic
, ==, MTX_MAGIC
);
310 ASSERT3P(mp
->m_owner
, !=, MTX_DEST
);
311 ASSERT3P(mp
->m_owner
, !=, curthread
);
312 VERIFY3S(pthread_mutex_lock(&mp
->m_lock
), ==, 0);
313 ASSERT3P(mp
->m_owner
, ==, MTX_INIT
);
314 mp
->m_owner
= curthread
;
318 mutex_tryenter(kmutex_t
*mp
)
320 ASSERT3U(mp
->m_magic
, ==, MTX_MAGIC
);
321 ASSERT3P(mp
->m_owner
, !=, MTX_DEST
);
322 if (0 == pthread_mutex_trylock(&mp
->m_lock
)) {
323 ASSERT3P(mp
->m_owner
, ==, MTX_INIT
);
324 mp
->m_owner
= curthread
;
332 mutex_exit(kmutex_t
*mp
)
334 ASSERT3U(mp
->m_magic
, ==, MTX_MAGIC
);
335 ASSERT3P(mutex_owner(mp
), ==, curthread
);
336 mp
->m_owner
= MTX_INIT
;
337 VERIFY3S(pthread_mutex_unlock(&mp
->m_lock
), ==, 0);
341 mutex_owner(kmutex_t
*mp
)
343 ASSERT3U(mp
->m_magic
, ==, MTX_MAGIC
);
344 return (mp
->m_owner
);
348 mutex_held(kmutex_t
*mp
)
350 return (mp
->m_owner
== curthread
);
354 * =========================================================================
356 * =========================================================================
360 rw_init(krwlock_t
*rwlp
, char *name
, int type
, void *arg
)
362 ASSERT3S(type
, ==, RW_DEFAULT
);
363 ASSERT3P(arg
, ==, NULL
);
364 VERIFY3S(pthread_rwlock_init(&rwlp
->rw_lock
, NULL
), ==, 0);
365 rwlp
->rw_owner
= RW_INIT
;
366 rwlp
->rw_wr_owner
= RW_INIT
;
367 rwlp
->rw_readers
= 0;
368 rwlp
->rw_magic
= RW_MAGIC
;
372 rw_destroy(krwlock_t
*rwlp
)
374 ASSERT3U(rwlp
->rw_magic
, ==, RW_MAGIC
);
376 VERIFY3S(pthread_rwlock_destroy(&rwlp
->rw_lock
), ==, 0);
381 rw_enter(krwlock_t
*rwlp
, krw_t rw
)
383 ASSERT3U(rwlp
->rw_magic
, ==, RW_MAGIC
);
384 ASSERT3P(rwlp
->rw_owner
, !=, curthread
);
385 ASSERT3P(rwlp
->rw_wr_owner
, !=, curthread
);
387 if (rw
== RW_READER
) {
388 VERIFY3S(pthread_rwlock_rdlock(&rwlp
->rw_lock
), ==, 0);
389 ASSERT3P(rwlp
->rw_wr_owner
, ==, RW_INIT
);
391 atomic_inc_uint(&rwlp
->rw_readers
);
393 VERIFY3S(pthread_rwlock_wrlock(&rwlp
->rw_lock
), ==, 0);
394 ASSERT3P(rwlp
->rw_wr_owner
, ==, RW_INIT
);
395 ASSERT3U(rwlp
->rw_readers
, ==, 0);
397 rwlp
->rw_wr_owner
= curthread
;
400 rwlp
->rw_owner
= curthread
;
404 rw_exit(krwlock_t
*rwlp
)
406 ASSERT3U(rwlp
->rw_magic
, ==, RW_MAGIC
);
407 ASSERT(RW_LOCK_HELD(rwlp
));
409 if (RW_READ_HELD(rwlp
))
410 atomic_dec_uint(&rwlp
->rw_readers
);
412 rwlp
->rw_wr_owner
= RW_INIT
;
414 rwlp
->rw_owner
= RW_INIT
;
415 VERIFY3S(pthread_rwlock_unlock(&rwlp
->rw_lock
), ==, 0);
419 rw_tryenter(krwlock_t
*rwlp
, krw_t rw
)
423 ASSERT3U(rwlp
->rw_magic
, ==, RW_MAGIC
);
426 rv
= pthread_rwlock_tryrdlock(&rwlp
->rw_lock
);
428 rv
= pthread_rwlock_trywrlock(&rwlp
->rw_lock
);
431 ASSERT3P(rwlp
->rw_wr_owner
, ==, RW_INIT
);
434 atomic_inc_uint(&rwlp
->rw_readers
);
436 ASSERT3U(rwlp
->rw_readers
, ==, 0);
437 rwlp
->rw_wr_owner
= curthread
;
440 rwlp
->rw_owner
= curthread
;
444 VERIFY3S(rv
, ==, EBUSY
);
450 rw_tryupgrade(krwlock_t
*rwlp
)
452 ASSERT3U(rwlp
->rw_magic
, ==, RW_MAGIC
);
458 * =========================================================================
459 * condition variables
460 * =========================================================================
464 cv_init(kcondvar_t
*cv
, char *name
, int type
, void *arg
)
466 ASSERT3S(type
, ==, CV_DEFAULT
);
467 cv
->cv_magic
= CV_MAGIC
;
468 VERIFY3S(pthread_cond_init(&cv
->cv
, NULL
), ==, 0);
472 cv_destroy(kcondvar_t
*cv
)
474 ASSERT3U(cv
->cv_magic
, ==, CV_MAGIC
);
475 VERIFY3S(pthread_cond_destroy(&cv
->cv
), ==, 0);
480 cv_wait(kcondvar_t
*cv
, kmutex_t
*mp
)
482 ASSERT3U(cv
->cv_magic
, ==, CV_MAGIC
);
483 ASSERT3P(mutex_owner(mp
), ==, curthread
);
484 mp
->m_owner
= MTX_INIT
;
485 int ret
= pthread_cond_wait(&cv
->cv
, &mp
->m_lock
);
487 VERIFY3S(ret
, ==, EINTR
);
488 mp
->m_owner
= curthread
;
492 cv_timedwait(kcondvar_t
*cv
, kmutex_t
*mp
, clock_t abstime
)
499 ASSERT3U(cv
->cv_magic
, ==, CV_MAGIC
);
502 delta
= abstime
- ddi_get_lbolt();
506 VERIFY(gettimeofday(&tv
, NULL
) == 0);
508 ts
.tv_sec
= tv
.tv_sec
+ delta
/ hz
;
509 ts
.tv_nsec
= tv
.tv_usec
* 1000 + (delta
% hz
) * (NANOSEC
/ hz
);
510 if (ts
.tv_nsec
>= NANOSEC
) {
512 ts
.tv_nsec
-= NANOSEC
;
515 ASSERT3P(mutex_owner(mp
), ==, curthread
);
516 mp
->m_owner
= MTX_INIT
;
517 error
= pthread_cond_timedwait(&cv
->cv
, &mp
->m_lock
, &ts
);
518 mp
->m_owner
= curthread
;
520 if (error
== ETIMEDOUT
)
526 VERIFY3S(error
, ==, 0);
533 cv_timedwait_hires(kcondvar_t
*cv
, kmutex_t
*mp
, hrtime_t tim
, hrtime_t res
,
543 delta
= tim
- gethrtime();
547 ts
.tv_sec
= delta
/ NANOSEC
;
548 ts
.tv_nsec
= delta
% NANOSEC
;
550 ASSERT(mutex_owner(mp
) == curthread
);
552 error
= pthread_cond_timedwait(&cv
->cv
, &mp
->m_lock
, &ts
);
553 mp
->m_owner
= curthread
;
567 cv_signal(kcondvar_t
*cv
)
569 ASSERT3U(cv
->cv_magic
, ==, CV_MAGIC
);
570 VERIFY3S(pthread_cond_signal(&cv
->cv
), ==, 0);
574 cv_broadcast(kcondvar_t
*cv
)
576 ASSERT3U(cv
->cv_magic
, ==, CV_MAGIC
);
577 VERIFY3S(pthread_cond_broadcast(&cv
->cv
), ==, 0);
581 * =========================================================================
583 * =========================================================================
586 * Note: for the xxxat() versions of these functions, we assume that the
587 * starting vp is always rootdir (which is true for spa_directory.c, the only
588 * ZFS consumer of these interfaces). We assert this is true, and then emulate
589 * them by adding '/' in front of the path.
594 vn_open(char *path
, int x1
, int flags
, int mode
, vnode_t
**vpp
, int x2
, int x3
)
603 realpath
= umem_alloc(MAXPATHLEN
, UMEM_NOFAIL
);
606 * If we're accessing a real disk from userland, we need to use
607 * the character interface to avoid caching. This is particularly
608 * important if we're trying to look at a real in-kernel storage
609 * pool from userland, e.g. via zdb, because otherwise we won't
610 * see the changes occurring under the segmap cache.
611 * On the other hand, the stupid character device returns zero
612 * for its size. So -- gag -- we open the block device to get
613 * its size, and remember it for subsequent VOP_GETATTR().
615 #if defined(__sun__) || defined(__sun)
616 if (strncmp(path
, "/dev/", 5) == 0) {
621 fd
= open64(path
, O_RDONLY
);
627 if (fstat64(fd
, &st
) == -1) {
634 (void) sprintf(realpath
, "%s", path
);
635 dsk
= strstr(path
, "/dsk/");
637 (void) sprintf(realpath
+ (dsk
- path
) + 1, "r%s",
640 (void) sprintf(realpath
, "%s", path
);
641 if (!(flags
& FCREAT
) && stat64(realpath
, &st
) == -1) {
648 if (!(flags
& FCREAT
) && S_ISBLK(st
.st_mode
)) {
652 /* We shouldn't be writing to block devices in userspace */
653 VERIFY(!(flags
& FWRITE
));
657 old_umask
= umask(0);
660 * The construct 'flags - FREAD' conveniently maps combinations of
661 * FREAD and FWRITE to the corresponding O_RDONLY, O_WRONLY, and O_RDWR.
663 fd
= open64(realpath
, flags
- FREAD
, mode
);
667 (void) umask(old_umask
);
672 if (fstat64_blk(fd
, &st
) == -1) {
678 (void) fcntl(fd
, F_SETFD
, FD_CLOEXEC
);
680 *vpp
= vp
= umem_zalloc(sizeof (vnode_t
), UMEM_NOFAIL
);
683 vp
->v_size
= st
.st_size
;
684 vp
->v_path
= spa_strdup(path
);
691 vn_openat(char *path
, int x1
, int flags
, int mode
, vnode_t
**vpp
, int x2
,
692 int x3
, vnode_t
*startvp
, int fd
)
694 char *realpath
= umem_alloc(strlen(path
) + 2, UMEM_NOFAIL
);
697 ASSERT(startvp
== rootdir
);
698 (void) sprintf(realpath
, "/%s", path
);
700 /* fd ignored for now, need if want to simulate nbmand support */
701 ret
= vn_open(realpath
, x1
, flags
, mode
, vpp
, x2
, x3
);
703 umem_free(realpath
, strlen(path
) + 2);
710 vn_rdwr(int uio
, vnode_t
*vp
, void *addr
, ssize_t len
, offset_t offset
,
711 int x1
, int x2
, rlim64_t x3
, void *x4
, ssize_t
*residp
)
713 ssize_t rc
, done
= 0, split
;
715 if (uio
== UIO_READ
) {
716 rc
= pread64(vp
->v_fd
, addr
, len
, offset
);
719 * To simulate partial disk writes, we split writes into two
720 * system calls so that the process can be killed in between.
722 int sectors
= len
>> SPA_MINBLOCKSHIFT
;
723 split
= (sectors
> 0 ? rand() % sectors
: 0) <<
725 rc
= pwrite64(vp
->v_fd
, addr
, split
, offset
);
728 rc
= pwrite64(vp
->v_fd
, (char *)addr
+ split
,
729 len
- split
, offset
+ split
);
734 if (rc
== -1 && errno
== EINVAL
) {
736 * Under Linux, this most likely means an alignment issue
737 * (memory or disk) due to O_DIRECT, so we abort() in order to
738 * catch the offender.
749 *residp
= len
- done
;
750 else if (done
!= len
)
756 vn_close(vnode_t
*vp
)
759 spa_strfree(vp
->v_path
);
760 umem_free(vp
, sizeof (vnode_t
));
764 * At a minimum we need to update the size since vdev_reopen()
765 * will no longer call vn_openat().
768 fop_getattr(vnode_t
*vp
, vattr_t
*vap
)
773 if (fstat64_blk(vp
->v_fd
, &st
) == -1) {
779 vap
->va_size
= st
.st_size
;
784 * =========================================================================
785 * Figure out which debugging statements to print
786 * =========================================================================
789 static char *dprintf_string
;
790 static int dprintf_print_all
;
793 dprintf_find_string(const char *string
)
795 char *tmp_str
= dprintf_string
;
796 int len
= strlen(string
);
799 * Find out if this is a string we want to print.
800 * String format: file1.c,function_name1,file2.c,file3.c
803 while (tmp_str
!= NULL
) {
804 if (strncmp(tmp_str
, string
, len
) == 0 &&
805 (tmp_str
[len
] == ',' || tmp_str
[len
] == '\0'))
807 tmp_str
= strchr(tmp_str
, ',');
809 tmp_str
++; /* Get rid of , */
815 dprintf_setup(int *argc
, char **argv
)
820 * Debugging can be specified two ways: by setting the
821 * environment variable ZFS_DEBUG, or by including a
822 * "debug=..." argument on the command line. The command
823 * line setting overrides the environment variable.
826 for (i
= 1; i
< *argc
; i
++) {
827 int len
= strlen("debug=");
828 /* First look for a command line argument */
829 if (strncmp("debug=", argv
[i
], len
) == 0) {
830 dprintf_string
= argv
[i
] + len
;
831 /* Remove from args */
832 for (j
= i
; j
< *argc
; j
++)
839 if (dprintf_string
== NULL
) {
840 /* Look for ZFS_DEBUG environment variable */
841 dprintf_string
= getenv("ZFS_DEBUG");
845 * Are we just turning on all debugging?
847 if (dprintf_find_string("on"))
848 dprintf_print_all
= 1;
852 * =========================================================================
854 * =========================================================================
857 __dprintf(const char *file
, const char *func
, int line
, const char *fmt
, ...)
863 * Get rid of annoying "../common/" prefix to filename.
865 newfile
= strrchr(file
, '/');
866 if (newfile
!= NULL
) {
867 newfile
= newfile
+ 1; /* Get rid of leading / */
872 if (dprintf_print_all
||
873 dprintf_find_string(newfile
) ||
874 dprintf_find_string(func
)) {
875 /* Print out just the function name if requested */
877 if (dprintf_find_string("pid"))
878 (void) printf("%d ", getpid());
879 if (dprintf_find_string("tid"))
880 (void) printf("%u ", (uint_t
) pthread_self());
881 if (dprintf_find_string("cpu"))
882 (void) printf("%u ", getcpuid());
883 if (dprintf_find_string("time"))
884 (void) printf("%llu ", gethrtime());
885 if (dprintf_find_string("long"))
886 (void) printf("%s, line %d: ", newfile
, line
);
887 (void) printf("%s: ", func
);
889 (void) vprintf(fmt
, adx
);
896 * =========================================================================
897 * cmn_err() and panic()
898 * =========================================================================
900 static char ce_prefix
[CE_IGNORE
][10] = { "", "NOTICE: ", "WARNING: ", "" };
901 static char ce_suffix
[CE_IGNORE
][2] = { "", "\n", "\n", "" };
904 vpanic(const char *fmt
, va_list adx
)
906 (void) fprintf(stderr
, "error: ");
907 (void) vfprintf(stderr
, fmt
, adx
);
908 (void) fprintf(stderr
, "\n");
910 abort(); /* think of it as a "user-level crash dump" */
914 panic(const char *fmt
, ...)
924 vcmn_err(int ce
, const char *fmt
, va_list adx
)
928 if (ce
!= CE_NOTE
) { /* suppress noise in userland stress testing */
929 (void) fprintf(stderr
, "%s", ce_prefix
[ce
]);
930 (void) vfprintf(stderr
, fmt
, adx
);
931 (void) fprintf(stderr
, "%s", ce_suffix
[ce
]);
937 cmn_err(int ce
, const char *fmt
, ...)
942 vcmn_err(ce
, fmt
, adx
);
947 * =========================================================================
949 * =========================================================================
952 kobj_open_file(char *name
)
957 /* set vp as the _fd field of the file */
958 if (vn_openat(name
, UIO_SYSSPACE
, FREAD
, 0, &vp
, 0, 0, rootdir
,
960 return ((void *)-1UL);
962 file
= umem_zalloc(sizeof (struct _buf
), UMEM_NOFAIL
);
963 file
->_fd
= (intptr_t)vp
;
968 kobj_read_file(struct _buf
*file
, char *buf
, unsigned size
, unsigned off
)
972 vn_rdwr(UIO_READ
, (vnode_t
*)file
->_fd
, buf
, size
, (offset_t
)off
,
973 UIO_SYSSPACE
, 0, 0, 0, &resid
);
975 return (size
- resid
);
979 kobj_close_file(struct _buf
*file
)
981 vn_close((vnode_t
*)file
->_fd
);
982 umem_free(file
, sizeof (struct _buf
));
986 kobj_get_filesize(struct _buf
*file
, uint64_t *size
)
989 vnode_t
*vp
= (vnode_t
*)file
->_fd
;
991 if (fstat64(vp
->v_fd
, &st
) == -1) {
1000 * =========================================================================
1002 * =========================================================================
1006 delay(clock_t ticks
)
1008 poll(0, 0, ticks
* (1000 / hz
));
1012 * Find highest one bit set.
1013 * Returns bit number + 1 of highest bit that is set, otherwise returns 0.
1014 * High order bit is 31 (or 63 in _LP64 kernel).
1024 if (i
& 0xffffffff00000000ul
) {
1028 if (i
& 0xffff0000) {
1046 static int random_fd
= -1, urandom_fd
= -1;
1049 random_get_bytes_common(uint8_t *ptr
, size_t len
, int fd
)
1056 while (resid
!= 0) {
1057 bytes
= read(fd
, ptr
, resid
);
1058 ASSERT3S(bytes
, >=, 0);
1067 random_get_bytes(uint8_t *ptr
, size_t len
)
1069 return (random_get_bytes_common(ptr
, len
, random_fd
));
1073 random_get_pseudo_bytes(uint8_t *ptr
, size_t len
)
1075 return (random_get_bytes_common(ptr
, len
, urandom_fd
));
1079 ddi_strtoul(const char *hw_serial
, char **nptr
, int base
, unsigned long *result
)
1083 *result
= strtoul(hw_serial
, &end
, base
);
1090 ddi_strtoull(const char *str
, char **nptr
, int base
, u_longlong_t
*result
)
1094 *result
= strtoull(str
, &end
, base
);
1101 * =========================================================================
1102 * kernel emulation setup & teardown
1103 * =========================================================================
1106 umem_out_of_memory(void)
1108 char errmsg
[] = "out of memory -- generating core dump\n";
1110 (void) fprintf(stderr
, "%s", errmsg
);
1116 kernel_init(int mode
)
1118 extern uint_t rrw_tsd_key
;
1120 umem_nofail_callback(umem_out_of_memory
);
1122 physmem
= sysconf(_SC_PHYS_PAGES
);
1124 dprintf("physmem = %llu pages (%.2f GB)\n", physmem
,
1125 (double)physmem
* sysconf(_SC_PAGE_SIZE
) / (1ULL << 30));
1127 (void) snprintf(hw_serial
, sizeof (hw_serial
), "%ld",
1128 (mode
& FWRITE
) ? gethostid() : 0);
1130 VERIFY((random_fd
= open("/dev/random", O_RDONLY
)) != -1);
1131 VERIFY((urandom_fd
= open("/dev/urandom", O_RDONLY
)) != -1);
1134 system_taskq_init();
1138 tsd_create(&rrw_tsd_key
, rrw_tsd_destroy
);
1146 system_taskq_fini();
1157 crgetuid(cred_t
*cr
)
1163 crgetruid(cred_t
*cr
)
1169 crgetgid(cred_t
*cr
)
1175 crgetngroups(cred_t
*cr
)
1181 crgetgroups(cred_t
*cr
)
1187 zfs_secpolicy_snapshot_perms(const char *name
, cred_t
*cr
)
1193 zfs_secpolicy_rename_perms(const char *from
, const char *to
, cred_t
*cr
)
1199 zfs_secpolicy_destroy_perms(const char *name
, cred_t
*cr
)
1205 ksid_lookupdomain(const char *dom
)
1209 kd
= umem_zalloc(sizeof (ksiddomain_t
), UMEM_NOFAIL
);
1210 kd
->kd_name
= spa_strdup(dom
);
1215 ksiddomain_rele(ksiddomain_t
*ksid
)
1217 spa_strfree(ksid
->kd_name
);
1218 umem_free(ksid
, sizeof (ksiddomain_t
));
1222 kmem_vasprintf(const char *fmt
, va_list adx
)
1227 va_copy(adx_copy
, adx
);
1228 VERIFY(vasprintf(&buf
, fmt
, adx_copy
) != -1);
1235 kmem_asprintf(const char *fmt
, ...)
1241 VERIFY(vasprintf(&buf
, fmt
, adx
) != -1);
1249 zfs_onexit_fd_hold(int fd
, minor_t
*minorp
)
1257 zfs_onexit_fd_rele(int fd
)
1263 zfs_onexit_add_cb(minor_t minor
, void (*func
)(void *), void *data
,
1264 uint64_t *action_handle
)
1271 zfs_onexit_del_cb(minor_t minor
, uint64_t action_handle
, boolean_t fire
)
1278 zfs_onexit_cb_data(minor_t minor
, uint64_t action_handle
, void **data
)