Archive-Team
/
zfs
2
0
Fork 0
Code Issues Pull Requests Packages Projects Releases 103 Wiki Activity

Simplify threads, mutexs, cvs and rwlocks 

* Simplify threads, mutexs, cvs and rwlocks

* Update the zk_thread_create() function to use the same trick
  as Illumos.  Specifically, cast the new pthread_t to a void
  pointer and return that as the kthread_t *.  This avoids the
  issues associated with managing a wrapper structure and is
  safe as long as the callers never attempt to dereference it.

* Update all function prototypes passed to pthread_create() to
  match the expected prototype.  We were getting away this with
  before since the function were explicitly cast.

* Replaced direct zk_thread_create() calls with thread_create()
  for code consistency.  All consumers of libzpool now use the
  proper wrappers.

* The mutex_held() calls were converted to MUTEX_HELD().

* Removed all mutex_owner() calls and retired the interface.
  Instead use MUTEX_HELD() which provides the same information
  and allows the implementation details to be hidden.  In this
  case the use of the pthread_equals() function.

* The kthread_t, kmutex_t, krwlock_t, and krwlock_t types had
  any non essential fields removed.  In the case of kthread_t
  and kcondvar_t they could be directly typedef'd to pthread_t
  and pthread_cond_t respectively.

* Removed all extra ASSERTS from the thread, mutex, rwlock, and
  cv wrapper functions.  In practice, pthreads already provides
  the vast majority of checks as long as we check the return
  code.  Removing this code from our wrappers help readability.

* Added TS_JOINABLE state flag to pass to request a joinable rather
  than detached thread.  This isn't a standard thread_create() state
  but it's the least invasive way to pass this information and is
  only used by ztest.

TEST_ZTEST_TIMEOUT=3600

Chunwei Chen <tuxoko@gmail.com>
Reviewed-by: Tom Caputi <tcaputi@datto.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4547 
Closes #5503 
Closes #5523 
Closes #6377 
Closes #6495
This commit is contained in:
Brian Behlendorf 2017-08-11 08:51:44 -07:00 committed by GitHub
parent 21df134f4c
commit c25b8f99f8
10 changed files with 143 additions and 356 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
cmd/raidz_test/raidz_test.c
View File

@ -702,10 +702,8 @@ run_sweep(void)
opts->rto_dsize = size_v[s]; opts->rto_dsize = size_v[s];
opts->rto_v = 0; /* be quiet */ opts->rto_v = 0; /* be quiet */
VERIFY3P(zk_thread_create(NULL, 0, VERIFY3P(thread_create(NULL, 0, sweep_thread, (void *) opts,
(thread_func_t)sweep_thread, 0, NULL, TS_RUN, defclsyspri), !=, NULL);
(void *) opts, 0, NULL, TS_RUN, 0,
PTHREAD_CREATE_JOINABLE), !=, NULL);
} }
exit: exit:

46
cmd/ztest/ztest.c
View File

@ -2218,7 +2218,7 @@ ztest_lookup(ztest_ds_t *zd, ztest_od_t *od, int count)
int error; int error;
int i; int i;
ASSERT(mutex_held(&zd->zd_dirobj_lock)); ASSERT(MUTEX_HELD(&zd->zd_dirobj_lock));
for (i = 0; i < count; i++, od++) { for (i = 0; i < count; i++, od++) {
od->od_object = 0; od->od_object = 0;
@ -2259,7 +2259,7 @@ ztest_create(ztest_ds_t *zd, ztest_od_t *od, int count)
int missing = 0; int missing = 0;
int i; int i;
ASSERT(mutex_held(&zd->zd_dirobj_lock)); ASSERT(MUTEX_HELD(&zd->zd_dirobj_lock));
for (i = 0; i < count; i++, od++) { for (i = 0; i < count; i++, od++) {
if (missing) { if (missing) {
@ -2305,7 +2305,7 @@ ztest_remove(ztest_ds_t *zd, ztest_od_t *od, int count)
int error; int error;
int i; int i;
ASSERT(mutex_held(&zd->zd_dirobj_lock)); ASSERT(MUTEX_HELD(&zd->zd_dirobj_lock));
od += count - 1; od += count - 1;
@ -6081,7 +6081,7 @@ ztest_resume(spa_t *spa)
(void) zio_resume(spa); (void) zio_resume(spa);
} }
static void * static void
ztest_resume_thread(void *arg) ztest_resume_thread(void *arg)
{ {
spa_t *spa = arg; spa_t *spa = arg;
@ -6105,8 +6105,6 @@ ztest_resume_thread(void *arg)
} }
thread_exit(); thread_exit();
return (NULL);
} }
#define GRACE 300 #define GRACE 300
@ -6140,7 +6138,7 @@ ztest_execute(int test, ztest_info_t *zi, uint64_t id)
(double)functime / NANOSEC, zi->zi_funcname); (double)functime / NANOSEC, zi->zi_funcname);
} }
static void * static void
ztest_thread(void *arg) ztest_thread(void *arg)
{ {
int rand; int rand;
@ -6180,8 +6178,6 @@ ztest_thread(void *arg)
} }
thread_exit(); thread_exit();
return (NULL);
} }
static void static void
@ -6309,10 +6305,10 @@ ztest_dataset_close(int d)
static void static void
ztest_run(ztest_shared_t *zs) ztest_run(ztest_shared_t *zs)
{ {
kt_did_t *tid;
spa_t *spa; spa_t *spa;
objset_t *os; objset_t *os;
kthread_t *resume_thread; kthread_t *resume_thread;
kthread_t **run_threads;
uint64_t object; uint64_t object;
int error; int error;
int t, d; int t, d;
@ -6373,9 +6369,8 @@ ztest_run(ztest_shared_t *zs)
/* /*
* Create a thread to periodically resume suspended I/O. * Create a thread to periodically resume suspended I/O.
*/ */
VERIFY3P((resume_thread = zk_thread_create(NULL, 0, resume_thread = thread_create(NULL, 0, ztest_resume_thread,
(thread_func_t)ztest_resume_thread, spa, 0, NULL, TS_RUN, 0, spa, 0, NULL, TS_RUN | TS_JOINABLE, defclsyspri);
PTHREAD_CREATE_JOINABLE)), !=, NULL);
#if 0 #if 0
/* /*
@ -6409,7 +6404,7 @@ ztest_run(ztest_shared_t *zs)
} }
zs->zs_enospc_count = 0; zs->zs_enospc_count = 0;
tid = umem_zalloc(ztest_opts.zo_threads * sizeof (kt_did_t), run_threads = umem_zalloc(ztest_opts.zo_threads * sizeof (kthread_t *),
UMEM_NOFAIL); UMEM_NOFAIL);
if (ztest_opts.zo_verbose >= 4) if (ztest_opts.zo_verbose >= 4)
@ -6419,20 +6414,15 @@ ztest_run(ztest_shared_t *zs)
* Kick off all the tests that run in parallel. * Kick off all the tests that run in parallel.
*/ */
for (t = 0; t < ztest_opts.zo_threads; t++) { for (t = 0; t < ztest_opts.zo_threads; t++) {
kthread_t *thread; if (t < ztest_opts.zo_datasets && ztest_dataset_open(t) != 0) {
umem_free(run_threads, ztest_opts.zo_threads *
if (t < ztest_opts.zo_datasets && sizeof (kthread_t *));
ztest_dataset_open(t) != 0) {
umem_free(tid,
ztest_opts.zo_threads * sizeof (kt_did_t));
return; return;
} }
VERIFY3P(thread = zk_thread_create(NULL, 0, run_threads[t] = thread_create(NULL, 0, ztest_thread,
(thread_func_t)ztest_thread, (void *)(uintptr_t)t, 0, NULL, TS_RUN | TS_JOINABLE,
(void *)(uintptr_t)t, 0, NULL, TS_RUN, 0, defclsyspri);
PTHREAD_CREATE_JOINABLE), !=, NULL);
tid[t] = thread->t_tid;
} }
/* /*
@ -6440,7 +6430,7 @@ ztest_run(ztest_shared_t *zs)
* so we don't close datasets while threads are still using them. * so we don't close datasets while threads are still using them.
*/ */
for (t = ztest_opts.zo_threads - 1; t >= 0; t--) { for (t = ztest_opts.zo_threads - 1; t >= 0; t--) {
thread_join(tid[t]); VERIFY0(thread_join(run_threads[t]));
if (t < ztest_opts.zo_datasets) if (t < ztest_opts.zo_datasets)
ztest_dataset_close(t); ztest_dataset_close(t);
} }
@ -6450,11 +6440,11 @@ ztest_run(ztest_shared_t *zs)
zs->zs_alloc = metaslab_class_get_alloc(spa_normal_class(spa)); zs->zs_alloc = metaslab_class_get_alloc(spa_normal_class(spa));
zs->zs_space = metaslab_class_get_space(spa_normal_class(spa)); zs->zs_space = metaslab_class_get_space(spa_normal_class(spa));
umem_free(tid, ztest_opts.zo_threads * sizeof (kt_did_t)); umem_free(run_threads, ztest_opts.zo_threads * sizeof (kthread_t *));
/* Kill the resume thread */ /* Kill the resume thread */
ztest_exiting = B_TRUE; ztest_exiting = B_TRUE;
thread_join(resume_thread->t_tid); VERIFY0(thread_join(resume_thread));
ztest_resume(spa); ztest_resume(spa);
/* /*

104
include/sys/zfs_context.h
View File

@ -207,15 +207,23 @@ extern int aok;
(unsigned long)i) (unsigned long)i)
/* /*
* Threads. TS_STACK_MIN is dictated by the minimum allowed pthread stack * Threads.
* size. While TS_STACK_MAX is somewhat arbitrary, it was selected to be
* large enough for the expected stack depth while small enough to avoid
* exhausting address space with high thread counts.
*/ */
#define TS_MAGIC 0x72f158ab4261e538ull typedef pthread_t kthread_t;
#define TS_RUN 0x00000002 #define TS_RUN 0x00000002
#define TS_STACK_MIN MAX(PTHREAD_STACK_MIN, 32768) #define TS_JOINABLE 0x00000004
#define TS_STACK_MAX (256 * 1024)
#define curthread ((void *)(uintptr_t)pthread_self())
#define kpreempt(x) yield()
#define getcomm() "unknown"
#define thread_create(stk, stksize, func, arg, len, pp, state, pri) \
zk_thread_create(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 */ /* in libzpool, p0 exists only to have its address taken */
typedef struct proc { typedef struct proc {
@ -225,81 +233,42 @@ typedef struct proc {
extern struct proc p0; extern struct proc p0;
#define curproc (&p0) #define curproc (&p0)
typedef void (*thread_func_t)(void *);
typedef void (*thread_func_arg_t)(void *);
typedef pthread_t kt_did_t;
#define kpreempt(x) ((void)0)
typedef struct kthread {
kt_did_t t_tid;
thread_func_t t_func;
void * t_arg;
pri_t t_pri;
} kthread_t;
#define curthread zk_thread_current()
#define getcomm() "unknown"
#define thread_exit zk_thread_exit
#define thread_create(stk, stksize, func, arg, len, pp, state, pri) \
zk_thread_create(stk, stksize, (thread_func_t)func, arg, \
len, NULL, state, pri, PTHREAD_CREATE_DETACHED)
#define thread_join(t) zk_thread_join(t)
#define newproc(f, a, cid, pri, ctp, pid) (ENOSYS)
extern kthread_t *zk_thread_current(void);
extern void zk_thread_exit(void);
extern kthread_t *zk_thread_create(caddr_t stk, size_t stksize,
thread_func_t func, void *arg, uint64_t len,
proc_t *pp, int state, pri_t pri, int detachstate);
extern void zk_thread_join(kt_did_t tid);
#define kpreempt_disable() ((void)0)
#define kpreempt_enable() ((void)0)
#define PS_NONE -1 #define PS_NONE -1
extern kthread_t *zk_thread_create(void (*func)(void *), void *arg,
size_t stksize, int state);
#define issig(why) (FALSE) #define issig(why) (FALSE)
#define ISSIG(thr, why) (FALSE) #define ISSIG(thr, why) (FALSE)
#define kpreempt_disable() ((void)0)
#define kpreempt_enable() ((void)0)
/* /*
* Mutexes * Mutexes
*/ */
#define MTX_MAGIC 0x9522f51362a6e326ull
#define MTX_INIT ((void *)NULL)
#define MTX_DEST ((void *)-1UL)
typedef struct kmutex { typedef struct kmutex {
void *m_owner;
uint64_t m_magic;
pthread_mutex_t m_lock; pthread_mutex_t m_lock;
pthread_t m_owner;
} kmutex_t; } kmutex_t;
#define MUTEX_DEFAULT 0 #define MUTEX_DEFAULT 0
#define MUTEX_NOLOCKDEP MUTEX_DEFAULT #define MUTEX_NOLOCKDEP MUTEX_DEFAULT
#define MUTEX_HELD(m) ((m)->m_owner == curthread) #define MUTEX_HELD(mp) pthread_equal((mp)->m_owner, pthread_self())
#define MUTEX_NOT_HELD(m) (!MUTEX_HELD(m)) #define MUTEX_NOT_HELD(mp) !MUTEX_HELD(mp)
extern void mutex_init(kmutex_t *mp, char *name, int type, void *cookie); extern void mutex_init(kmutex_t *mp, char *name, int type, void *cookie);
extern void mutex_destroy(kmutex_t *mp); extern void mutex_destroy(kmutex_t *mp);
extern void mutex_enter(kmutex_t *mp); extern void mutex_enter(kmutex_t *mp);
extern void mutex_exit(kmutex_t *mp); extern void mutex_exit(kmutex_t *mp);
extern int mutex_tryenter(kmutex_t *mp); extern int mutex_tryenter(kmutex_t *mp);
extern void *mutex_owner(kmutex_t *mp);
extern int mutex_held(kmutex_t *mp);
/* /*
* RW locks * RW locks
*/ */
#define RW_MAGIC 0x4d31fb123648e78aull
#define RW_INIT ((void *)NULL)
#define RW_DEST ((void *)-1UL)
typedef struct krwlock { typedef struct krwlock {
void *rw_owner;
void *rw_wr_owner;
uint64_t rw_magic;
pthread_rwlock_t rw_lock; pthread_rwlock_t rw_lock;
pthread_t rw_owner;
uint_t rw_readers; uint_t rw_readers;
} krwlock_t; } krwlock_t;
@ -310,15 +279,9 @@ typedef int krw_t;
#define RW_DEFAULT RW_READER #define RW_DEFAULT RW_READER
#define RW_NOLOCKDEP RW_READER #define RW_NOLOCKDEP RW_READER
#define RW_READ_HELD(x) ((x)->rw_readers > 0) #define RW_READ_HELD(rw) ((rw)->rw_readers > 0)
#define RW_WRITE_HELD(x) ((x)->rw_wr_owner == curthread) #define RW_WRITE_HELD(rw) pthread_equal((rw)->rw_owner, pthread_self())
#define RW_LOCK_HELD(x) (RW_READ_HELD(x) || RW_WRITE_HELD(x)) #define RW_LOCK_HELD(rw) (RW_READ_HELD(rw) || RW_WRITE_HELD(rw))
#undef RW_LOCK_HELD
#define RW_LOCK_HELD(x) (RW_READ_HELD(x) || RW_WRITE_HELD(x))
#undef RW_LOCK_HELD
#define RW_LOCK_HELD(x) (RW_READ_HELD(x) || RW_WRITE_HELD(x))
extern void rw_init(krwlock_t *rwlp, char *name, int type, void *arg); extern void rw_init(krwlock_t *rwlp, char *name, int type, void *arg);
extern void rw_destroy(krwlock_t *rwlp); extern void rw_destroy(krwlock_t *rwlp);
@ -328,6 +291,9 @@ extern int rw_tryupgrade(krwlock_t *rwlp);
extern void rw_exit(krwlock_t *rwlp); extern void rw_exit(krwlock_t *rwlp);
#define rw_downgrade(rwlp) do { } while (0) #define rw_downgrade(rwlp) do { } while (0)
/*
* Credentials
*/
extern uid_t crgetuid(cred_t *cr); extern uid_t crgetuid(cred_t *cr);
extern uid_t crgetruid(cred_t *cr); extern uid_t crgetruid(cred_t *cr);
extern gid_t crgetgid(cred_t *cr); extern gid_t crgetgid(cred_t *cr);
@ -337,12 +303,7 @@ extern gid_t *crgetgroups(cred_t *cr);
/* /*
* Condition variables * Condition variables
*/ */
#define CV_MAGIC 0xd31ea9a83b1b30c4ull typedef pthread_cond_t kcondvar_t;
typedef struct kcondvar {
uint64_t cv_magic;
pthread_cond_t cv;
} kcondvar_t;
#define CV_DEFAULT 0 #define CV_DEFAULT 0
#define CALLOUT_FLAG_ABSOLUTE 0x2 #define CALLOUT_FLAG_ABSOLUTE 0x2
@ -355,6 +316,7 @@ extern clock_t cv_timedwait_hires(kcondvar_t *cvp, kmutex_t *mp, hrtime_t tim,
hrtime_t res, int flag); hrtime_t res, int flag);
extern void cv_signal(kcondvar_t *cv); extern void cv_signal(kcondvar_t *cv);
extern void cv_broadcast(kcondvar_t *cv); extern void cv_broadcast(kcondvar_t *cv);
#define cv_timedwait_sig(cv, mp, at) cv_timedwait(cv, mp, at) #define cv_timedwait_sig(cv, mp, at) cv_timedwait(cv, mp, at)
#define cv_wait_sig(cv, mp) cv_wait(cv, mp) #define cv_wait_sig(cv, mp) cv_wait(cv, mp)
#define cv_wait_io(cv, mp) cv_wait(cv, mp) #define cv_wait_io(cv, mp) cv_wait(cv, mp)

292
lib/libzpool/kernel.c
View File

@ -64,104 +64,29 @@ struct proc p0;
* ========================================================================= * =========================================================================
* threads * 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)
pthread_cond_t kthread_cond = PTHREAD_COND_INITIALIZER; /*ARGSUSED*/
pthread_mutex_t kthread_lock = PTHREAD_MUTEX_INITIALIZER;
pthread_key_t kthread_key;
int kthread_nr = 0;
static void
thread_init(void)
{
kthread_t *kt;
VERIFY3S(pthread_key_create(&kthread_key, NULL), ==, 0);
/* Create entry for primary kthread */
kt = umem_zalloc(sizeof (kthread_t), UMEM_NOFAIL);
kt->t_tid = pthread_self();
kt->t_func = NULL;
VERIFY3S(pthread_setspecific(kthread_key, kt), ==, 0);
/* Only the main thread should be running at the moment */
ASSERT3S(kthread_nr, ==, 0);
kthread_nr = 1;
}
static void
thread_fini(void)
{
kthread_t *kt = curthread;
ASSERT(pthread_equal(kt->t_tid, pthread_self()));
ASSERT3P(kt->t_func, ==, NULL);
umem_free(kt, sizeof (kthread_t));
/* Wait for all threads to exit via thread_exit() */
VERIFY3S(pthread_mutex_lock(&kthread_lock), ==, 0);
kthread_nr--; /* Main thread is exiting */
while (kthread_nr > 0)
VERIFY0(pthread_cond_wait(&kthread_cond, &kthread_lock));
ASSERT3S(kthread_nr, ==, 0);
VERIFY3S(pthread_mutex_unlock(&kthread_lock), ==, 0);
VERIFY3S(pthread_key_delete(kthread_key), ==, 0);
}
kthread_t * kthread_t *
zk_thread_current(void) zk_thread_create(void (*func)(void *), void *arg, size_t stksize, int state)
{ {
kthread_t *kt = pthread_getspecific(kthread_key);
ASSERT3P(kt, !=, NULL);
return (kt);
}
void *
zk_thread_helper(void *arg)
{
kthread_t *kt = (kthread_t *)arg;
VERIFY3S(pthread_setspecific(kthread_key, kt), ==, 0);
VERIFY3S(pthread_mutex_lock(&kthread_lock), ==, 0);
kthread_nr++;
VERIFY3S(pthread_mutex_unlock(&kthread_lock), ==, 0);
(void) setpriority(PRIO_PROCESS, 0, kt->t_pri);
kt->t_tid = pthread_self();
((thread_func_arg_t)kt->t_func)(kt->t_arg);
/* Unreachable, thread must exit with thread_exit() */
abort();
return (NULL);
}
kthread_t *
zk_thread_create(caddr_t stk, size_t stksize, thread_func_t func, void *arg,
uint64_t len, proc_t *pp, int state, pri_t pri, int detachstate)
{
kthread_t *kt;
pthread_attr_t attr; pthread_attr_t attr;
pthread_t tid;
char *stkstr; char *stkstr;
int detachstate = PTHREAD_CREATE_DETACHED;
ASSERT0(state & ~TS_RUN);
ASSERT0(len);
kt = umem_zalloc(sizeof (kthread_t), UMEM_NOFAIL);
kt->t_func = func;
kt->t_arg = arg;
kt->t_pri = pri;
VERIFY0(pthread_attr_init(&attr)); VERIFY0(pthread_attr_init(&attr));
if (state & TS_JOINABLE)
detachstate = PTHREAD_CREATE_JOINABLE;
VERIFY0(pthread_attr_setdetachstate(&attr, detachstate)); VERIFY0(pthread_attr_setdetachstate(&attr, detachstate));
/* /*
@ -183,6 +108,7 @@ zk_thread_create(caddr_t stk, size_t stksize, thread_func_t func, void *arg,
VERIFY3S(stksize, >, 0); VERIFY3S(stksize, >, 0);
stksize = P2ROUNDUP(MAX(stksize, TS_STACK_MIN), PAGESIZE); stksize = P2ROUNDUP(MAX(stksize, TS_STACK_MIN), PAGESIZE);
/* /*
* If this ever fails, it may be because the stack size is not a * If this ever fails, it may be because the stack size is not a
* multiple of system page size. * multiple of system page size.
@ -190,36 +116,10 @@ zk_thread_create(caddr_t stk, size_t stksize, thread_func_t func, void *arg,
VERIFY0(pthread_attr_setstacksize(&attr, stksize)); VERIFY0(pthread_attr_setstacksize(&attr, stksize));
VERIFY0(pthread_attr_setguardsize(&attr, PAGESIZE)); VERIFY0(pthread_attr_setguardsize(&attr, PAGESIZE));
VERIFY0(pthread_create(&kt->t_tid, &attr, &zk_thread_helper, kt)); VERIFY0(pthread_create(&tid, &attr, (void *(*)(void *))func, arg));
VERIFY0(pthread_attr_destroy(&attr)); VERIFY0(pthread_attr_destroy(&attr));
return (kt); return ((void *)(uintptr_t)tid);
}
void
zk_thread_exit(void)
{
kthread_t *kt = curthread;
ASSERT(pthread_equal(kt->t_tid, pthread_self()));
umem_free(kt, sizeof (kthread_t));
VERIFY0(pthread_mutex_lock(&kthread_lock));
kthread_nr--;
VERIFY0(pthread_mutex_unlock(&kthread_lock));
VERIFY0(pthread_cond_broadcast(&kthread_cond));
pthread_exit((void *)TS_MAGIC);
}
void
zk_thread_join(kt_did_t tid)
{
void *ret;
pthread_join((pthread_t)tid, &ret);
VERIFY3P(ret, ==, (void *)TS_MAGIC);
} }
/* /*
@ -291,46 +191,34 @@ kstat_set_raw_ops(kstat_t *ksp,
void void
mutex_init(kmutex_t *mp, char *name, int type, void *cookie) mutex_init(kmutex_t *mp, char *name, int type, void *cookie)
{ {
ASSERT3S(type, ==, MUTEX_DEFAULT); VERIFY0(pthread_mutex_init(&mp->m_lock, NULL));
ASSERT3P(cookie, ==, NULL); memset(&mp->m_owner, 0, sizeof (pthread_t));
mp->m_owner = MTX_INIT;
mp->m_magic = MTX_MAGIC;
VERIFY3S(pthread_mutex_init(&mp->m_lock, NULL), ==, 0);
} }
void void
mutex_destroy(kmutex_t *mp) mutex_destroy(kmutex_t *mp)
{ {
ASSERT3U(mp->m_magic, ==, MTX_MAGIC); VERIFY0(pthread_mutex_destroy(&mp->m_lock));
ASSERT3P(mp->m_owner, ==, MTX_INIT);
ASSERT0(pthread_mutex_destroy(&(mp)->m_lock));
mp->m_owner = MTX_DEST;
mp->m_magic = 0;
} }
void void
mutex_enter(kmutex_t *mp) mutex_enter(kmutex_t *mp)
{ {
ASSERT3U(mp->m_magic, ==, MTX_MAGIC); VERIFY0(pthread_mutex_lock(&mp->m_lock));
ASSERT3P(mp->m_owner, !=, MTX_DEST); mp->m_owner = pthread_self();
ASSERT3P(mp->m_owner, !=, curthread);
VERIFY3S(pthread_mutex_lock(&mp->m_lock), ==, 0);
ASSERT3P(mp->m_owner, ==, MTX_INIT);
mp->m_owner = curthread;
} }
int int
mutex_tryenter(kmutex_t *mp) mutex_tryenter(kmutex_t *mp)
{ {
int err; int error;
ASSERT3U(mp->m_magic, ==, MTX_MAGIC);
ASSERT3P(mp->m_owner, !=, MTX_DEST); error = pthread_mutex_trylock(&mp->m_lock);
if (0 == (err = pthread_mutex_trylock(&mp->m_lock))) { if (error == 0) {
ASSERT3P(mp->m_owner, ==, MTX_INIT); mp->m_owner = pthread_self();
mp->m_owner = curthread;
return (1); return (1);
} else { } else {
VERIFY3S(err, ==, EBUSY); VERIFY3S(error, ==, EBUSY);
return (0); return (0);
} }
} }
@ -338,23 +226,8 @@ mutex_tryenter(kmutex_t *mp)
void void
mutex_exit(kmutex_t *mp) mutex_exit(kmutex_t *mp)
{ {
ASSERT3U(mp->m_magic, ==, MTX_MAGIC); memset(&mp->m_owner, 0, sizeof (pthread_t));
ASSERT3P(mutex_owner(mp), ==, curthread); VERIFY0(pthread_mutex_unlock(&mp->m_lock));
mp->m_owner = MTX_INIT;
VERIFY3S(pthread_mutex_unlock(&mp->m_lock), ==, 0);
}
void *
mutex_owner(kmutex_t *mp)
{
ASSERT3U(mp->m_magic, ==, MTX_MAGIC);
return (mp->m_owner);
}
int
mutex_held(kmutex_t *mp)
{
return (mp->m_owner == curthread);
} }
/* /*
@ -366,89 +239,60 @@ mutex_held(kmutex_t *mp)
void void
rw_init(krwlock_t *rwlp, char *name, int type, void *arg) rw_init(krwlock_t *rwlp, char *name, int type, void *arg)
{ {
ASSERT3S(type, ==, RW_DEFAULT); VERIFY0(pthread_rwlock_init(&rwlp->rw_lock, NULL));
ASSERT3P(arg, ==, NULL);
VERIFY3S(pthread_rwlock_init(&rwlp->rw_lock, NULL), ==, 0);
rwlp->rw_owner = RW_INIT;
rwlp->rw_wr_owner = RW_INIT;
rwlp->rw_readers = 0; rwlp->rw_readers = 0;
rwlp->rw_magic = RW_MAGIC; rwlp->rw_owner = 0;
} }
void void
rw_destroy(krwlock_t *rwlp) rw_destroy(krwlock_t *rwlp)
{ {
ASSERT3U(rwlp->rw_magic, ==, RW_MAGIC); VERIFY0(pthread_rwlock_destroy(&rwlp->rw_lock));
ASSERT(rwlp->rw_readers == 0 && rwlp->rw_wr_owner == RW_INIT);
VERIFY3S(pthread_rwlock_destroy(&rwlp->rw_lock), ==, 0);
rwlp->rw_magic = 0;
} }
void void
rw_enter(krwlock_t *rwlp, krw_t rw) rw_enter(krwlock_t *rwlp, krw_t rw)
{ {
ASSERT3U(rwlp->rw_magic, ==, RW_MAGIC);
ASSERT3P(rwlp->rw_owner, !=, curthread);
ASSERT3P(rwlp->rw_wr_owner, !=, curthread);
if (rw == RW_READER) { if (rw == RW_READER) {
VERIFY3S(pthread_rwlock_rdlock(&rwlp->rw_lock), ==, 0); VERIFY0(pthread_rwlock_rdlock(&rwlp->rw_lock));
ASSERT3P(rwlp->rw_wr_owner, ==, RW_INIT);
atomic_inc_uint(&rwlp->rw_readers); atomic_inc_uint(&rwlp->rw_readers);
} else { } else {
VERIFY3S(pthread_rwlock_wrlock(&rwlp->rw_lock), ==, 0); VERIFY0(pthread_rwlock_wrlock(&rwlp->rw_lock));
ASSERT3P(rwlp->rw_wr_owner, ==, RW_INIT); rwlp->rw_owner = pthread_self();
ASSERT3U(rwlp->rw_readers, ==, 0);
rwlp->rw_wr_owner = curthread;
} }
rwlp->rw_owner = curthread;
} }
void void
rw_exit(krwlock_t *rwlp) rw_exit(krwlock_t *rwlp)
{ {
ASSERT3U(rwlp->rw_magic, ==, RW_MAGIC);
ASSERT(RW_LOCK_HELD(rwlp));
if (RW_READ_HELD(rwlp)) if (RW_READ_HELD(rwlp))
atomic_dec_uint(&rwlp->rw_readers); atomic_dec_uint(&rwlp->rw_readers);
else else
rwlp->rw_wr_owner = RW_INIT; rwlp->rw_owner = 0;
rwlp->rw_owner = RW_INIT; VERIFY0(pthread_rwlock_unlock(&rwlp->rw_lock));
VERIFY3S(pthread_rwlock_unlock(&rwlp->rw_lock), ==, 0);
} }
int int
rw_tryenter(krwlock_t *rwlp, krw_t rw) rw_tryenter(krwlock_t *rwlp, krw_t rw)
{ {
int rv; int error;
ASSERT3U(rwlp->rw_magic, ==, RW_MAGIC);
if (rw == RW_READER) if (rw == RW_READER)
rv = pthread_rwlock_tryrdlock(&rwlp->rw_lock); error = pthread_rwlock_tryrdlock(&rwlp->rw_lock);
else else
rv = pthread_rwlock_trywrlock(&rwlp->rw_lock); error = pthread_rwlock_trywrlock(&rwlp->rw_lock);
if (rv == 0) {
ASSERT3P(rwlp->rw_wr_owner, ==, RW_INIT);
if (error == 0) {
if (rw == RW_READER) if (rw == RW_READER)
atomic_inc_uint(&rwlp->rw_readers); atomic_inc_uint(&rwlp->rw_readers);
else { else
ASSERT3U(rwlp->rw_readers, ==, 0); rwlp->rw_owner = pthread_self();
rwlp->rw_wr_owner = curthread;
}
rwlp->rw_owner = curthread;
return (1); return (1);
} }
VERIFY3S(rv, ==, EBUSY); VERIFY3S(error, ==, EBUSY);
return (0); return (0);
} }
@ -456,8 +300,6 @@ rw_tryenter(krwlock_t *rwlp, krw_t rw)
int int
rw_tryupgrade(krwlock_t *rwlp) rw_tryupgrade(krwlock_t *rwlp)
{ {
ASSERT3U(rwlp->rw_magic, ==, RW_MAGIC);
return (0); return (0);
} }
@ -470,27 +312,21 @@ rw_tryupgrade(krwlock_t *rwlp)
void void
cv_init(kcondvar_t *cv, char *name, int type, void *arg) cv_init(kcondvar_t *cv, char *name, int type, void *arg)
{ {
ASSERT3S(type, ==, CV_DEFAULT); VERIFY0(pthread_cond_init(cv, NULL));
cv->cv_magic = CV_MAGIC;
VERIFY0(pthread_cond_init(&cv->cv, NULL));
} }
void void
cv_destroy(kcondvar_t *cv) cv_destroy(kcondvar_t *cv)
{ {
ASSERT3U(cv->cv_magic, ==, CV_MAGIC); VERIFY0(pthread_cond_destroy(cv));
VERIFY0(pthread_cond_destroy(&cv->cv));
cv->cv_magic = 0;
} }
void void
cv_wait(kcondvar_t *cv, kmutex_t *mp) cv_wait(kcondvar_t *cv, kmutex_t *mp)
{ {
ASSERT3U(cv->cv_magic, ==, CV_MAGIC); memset(&mp->m_owner, 0, sizeof (pthread_t));
ASSERT3P(mutex_owner(mp), ==, curthread); VERIFY0(pthread_cond_wait(cv, &mp->m_lock));
mp->m_owner = MTX_INIT; mp->m_owner = pthread_self();
VERIFY0(pthread_cond_wait(&cv->cv, &mp->m_lock));
mp->m_owner = curthread;
} }
clock_t clock_t
@ -501,8 +337,6 @@ cv_timedwait(kcondvar_t *cv, kmutex_t *mp, clock_t abstime)
timestruc_t ts; timestruc_t ts;
clock_t delta; clock_t delta;
ASSERT3U(cv->cv_magic, ==, CV_MAGIC);
delta = abstime - ddi_get_lbolt(); delta = abstime - ddi_get_lbolt();
if (delta <= 0) if (delta <= 0)
return (-1); return (-1);
@ -516,10 +350,9 @@ cv_timedwait(kcondvar_t *cv, kmutex_t *mp, clock_t abstime)
ts.tv_nsec -= NANOSEC; ts.tv_nsec -= NANOSEC;
} }
ASSERT3P(mutex_owner(mp), ==, curthread); memset(&mp->m_owner, 0, sizeof (pthread_t));
mp->m_owner = MTX_INIT; error = pthread_cond_timedwait(cv, &mp->m_lock, &ts);
error = pthread_cond_timedwait(&cv->cv, &mp->m_lock, &ts); mp->m_owner = pthread_self();
mp->m_owner = curthread;
if (error == ETIMEDOUT) if (error == ETIMEDOUT)
return (-1); return (-1);
@ -548,7 +381,7 @@ cv_timedwait_hires(kcondvar_t *cv, kmutex_t *mp, hrtime_t tim, hrtime_t res,
if (delta <= 0) if (delta <= 0)
return (-1); return (-1);
VERIFY(gettimeofday(&tv, NULL) == 0); VERIFY0(gettimeofday(&tv, NULL));
ts.tv_sec = tv.tv_sec + delta / NANOSEC; ts.tv_sec = tv.tv_sec + delta / NANOSEC;
ts.tv_nsec = tv.tv_usec * NSEC_PER_USEC + (delta % NANOSEC); ts.tv_nsec = tv.tv_usec * NSEC_PER_USEC + (delta % NANOSEC);
@ -557,10 +390,9 @@ cv_timedwait_hires(kcondvar_t *cv, kmutex_t *mp, hrtime_t tim, hrtime_t res,
ts.tv_nsec -= NANOSEC; ts.tv_nsec -= NANOSEC;
} }
ASSERT(mutex_owner(mp) == curthread); memset(&mp->m_owner, 0, sizeof (pthread_t));
mp->m_owner = MTX_INIT; error = pthread_cond_timedwait(cv, &mp->m_lock, &ts);
error = pthread_cond_timedwait(&cv->cv, &mp->m_lock, &ts); mp->m_owner = pthread_self();
mp->m_owner = curthread;
if (error == ETIMEDOUT) if (error == ETIMEDOUT)
return (-1); return (-1);
@ -573,15 +405,13 @@ cv_timedwait_hires(kcondvar_t *cv, kmutex_t *mp, hrtime_t tim, hrtime_t res,
void void
cv_signal(kcondvar_t *cv) cv_signal(kcondvar_t *cv)
{ {
ASSERT3U(cv->cv_magic, ==, CV_MAGIC); VERIFY0(pthread_cond_signal(cv));
VERIFY0(pthread_cond_signal(&cv->cv));
} }
void void
cv_broadcast(kcondvar_t *cv) cv_broadcast(kcondvar_t *cv)
{ {
ASSERT3U(cv->cv_magic, ==, CV_MAGIC); VERIFY0(pthread_cond_broadcast(cv));
VERIFY0(pthread_cond_broadcast(&cv->cv));
} }
/* /*
@ -1188,7 +1018,6 @@ kernel_init(int mode)
VERIFY0(uname(&hw_utsname)); VERIFY0(uname(&hw_utsname));
thread_init();
system_taskq_init(); system_taskq_init();
icp_init(); icp_init();
@ -1207,7 +1036,6 @@ kernel_fini(void)
icp_fini(); icp_fini();
system_taskq_fini(); system_taskq_fini();
thread_fini();
random_fini(); random_fini();
} }

4
module/zfs/arc.c
View File

@ -4214,7 +4214,7 @@ arc_kmem_reap_now(void)
* using mutex_tryenter() from arc_reclaim_thread(). * using mutex_tryenter() from arc_reclaim_thread().
*/ */
static void static void
arc_reclaim_thread(void) arc_reclaim_thread(void *unused)
{ {
fstrans_cookie_t cookie = spl_fstrans_mark(); fstrans_cookie_t cookie = spl_fstrans_mark();
hrtime_t growtime = 0; hrtime_t growtime = 0;
@ -7515,7 +7515,7 @@ l2arc_write_buffers(spa_t *spa, l2arc_dev_t *dev, uint64_t target_sz)
* heart of the L2ARC. * heart of the L2ARC.
*/ */
static void static void
l2arc_feed_thread(void) l2arc_feed_thread(void *unused)
{ {
callb_cpr_t cpr; callb_cpr_t cpr;
l2arc_dev_t *dev; l2arc_dev_t *dev;

2
module/zfs/dbuf.c
View File

@ -531,7 +531,7 @@ dbuf_evict_one(void)
* out of the cache it is destroyed and becomes eligible for arc eviction. * out of the cache it is destroyed and becomes eligible for arc eviction.
*/ */
static void static void
dbuf_evict_thread(void) dbuf_evict_thread(void *unused)
{ {
callb_cpr_t cpr; callb_cpr_t cpr;

5
module/zfs/mmp.c
View File

@ -123,7 +123,7 @@ uint_t zfs_multihost_import_intervals = MMP_DEFAULT_IMPORT_INTERVALS;
*/ */
uint_t zfs_multihost_fail_intervals = MMP_DEFAULT_FAIL_INTERVALS; uint_t zfs_multihost_fail_intervals = MMP_DEFAULT_FAIL_INTERVALS;
static void mmp_thread(spa_t *spa); static void mmp_thread(void *arg);
void void
mmp_init(spa_t *spa) mmp_init(spa_t *spa)
@ -364,8 +364,9 @@ mmp_write_uberblock(spa_t *spa)
} }
static void static void
mmp_thread(spa_t *spa) mmp_thread(void *arg)
{ {
spa_t *spa = (spa_t *)arg;
mmp_thread_t *mmp = &spa->spa_mmp; mmp_thread_t *mmp = &spa->spa_mmp;
boolean_t last_spa_suspended = spa_suspended(spa); boolean_t last_spa_suspended = spa_suspended(spa);
boolean_t last_spa_multihost = spa_multihost(spa); boolean_t last_spa_multihost = spa_multihost(spa);

10
module/zfs/spa.c
View File

@ -1028,6 +1028,11 @@ spa_create_zio_taskqs(spa_t *spa)
} }
} }
/*
* Disabled until spa_thread() can be adapted for Linux.
*/
#undef HAVE_SPA_THREAD
#if defined(_KERNEL) && defined(HAVE_SPA_THREAD) #if defined(_KERNEL) && defined(HAVE_SPA_THREAD)
static void static void
spa_thread(void *arg) spa_thread(void *arg)
@ -3415,7 +3420,7 @@ spa_open_common(const char *pool, spa_t **spapp, void *tag, nvlist_t *nvpolicy,
* up calling spa_open() again. The real fix is to figure out how to * up calling spa_open() again. The real fix is to figure out how to
* avoid dsl_dir_open() calling this in the first place. * avoid dsl_dir_open() calling this in the first place.
*/ */
if (mutex_owner(&spa_namespace_lock) != curthread) { if (MUTEX_NOT_HELD(&spa_namespace_lock)) {
mutex_enter(&spa_namespace_lock); mutex_enter(&spa_namespace_lock);
locked = B_TRUE; locked = B_TRUE;
} }
@ -6068,8 +6073,9 @@ spa_async_autoexpand(spa_t *spa, vdev_t *vd)
} }
static void static void
spa_async_thread(spa_t *spa) spa_async_thread(void *arg)
{ {
spa_t *spa = (spa_t *)arg;
int tasks, i; int tasks, i;
ASSERT(spa->spa_sync_on); ASSERT(spa->spa_sync_on);

10
module/zfs/txg.c
View File

@ -108,8 +108,8 @@
* now transition to the syncing state. * now transition to the syncing state.
*/ */
static void txg_sync_thread(dsl_pool_t *dp); static void txg_sync_thread(void *dp);
static void txg_quiesce_thread(dsl_pool_t *dp); static void txg_quiesce_thread(void *dp);
int zfs_txg_timeout = 5; /* max seconds worth of delta per txg */ int zfs_txg_timeout = 5; /* max seconds worth of delta per txg */
@ -477,8 +477,9 @@ txg_wait_callbacks(dsl_pool_t *dp)
} }
static void static void
txg_sync_thread(dsl_pool_t *dp) txg_sync_thread(void *arg)
{ {
dsl_pool_t *dp = (dsl_pool_t *)arg;
spa_t *spa = dp->dp_spa; spa_t *spa = dp->dp_spa;
tx_state_t *tx = &dp->dp_tx; tx_state_t *tx = &dp->dp_tx;
callb_cpr_t cpr; callb_cpr_t cpr;
@ -561,8 +562,9 @@ txg_sync_thread(dsl_pool_t *dp)
} }
static void static void
txg_quiesce_thread(dsl_pool_t *dp) txg_quiesce_thread(void *arg)
{ {
dsl_pool_t *dp = (dsl_pool_t *)arg;
tx_state_t *tx = &dp->dp_tx; tx_state_t *tx = &dp->dp_tx;
callb_cpr_t cpr; callb_cpr_t cpr;

2
module/zfs/zfs_sa.c
View File

@ -300,7 +300,7 @@ zfs_sa_upgrade(sa_handle_t *hdl, dmu_tx_t *tx)
* Otherwise, we know we are doing the * Otherwise, we know we are doing the
* sa_update() that caused us to enter this function. * sa_update() that caused us to enter this function.
*/ */
if (mutex_owner(&zp->z_lock) != curthread) { if (MUTEX_NOT_HELD(&zp->z_lock)) {
if (mutex_tryenter(&zp->z_lock) == 0) if (mutex_tryenter(&zp->z_lock) == 0)
return; return;
else else