diff options
| author |   Stefan Liebler <stli@linux.ibm.com> | 2018-10-17 12:23:04 +0200 |
|---|---|---|
| committer | Stefan Liebler <stli@linux.ibm.com> | 2018-10-17 12:23:04 +0200 |
| commit | 403b4feb22dcbc85ace72a361d2a951380372471 (patch) | |
| tree | fe2235a8bce38cad1bbef3b4f28da82dc353e138 /nptl/pthreadP.h | |
| parent | Don't reduce test timeout to less than default (diff) | |
| download | glibc-403b4feb22dcbc85ace72a361d2a951380372471.tar.gz glibc-403b4feb22dcbc85ace72a361d2a951380372471.tar.bz2 glibc-403b4feb22dcbc85ace72a361d2a951380372471.zip | |
Fix race in pthread_mutex_lock while promoting to PTHREAD_MUTEX_ELISION_NP [BZ #23275]
The race leads either to pthread_mutex_destroy returning EBUSY
or triggering an assertion (See description in bugzilla).
This patch is fixing the race by ensuring that the elision path is
used in all cases if elision is enabled by the GLIBC_TUNABLES framework.
The __kind variable in struct __pthread_mutex_s is accessed concurrently.
Therefore we are now using the atomic macros.
The new testcase tst-mutex10 is triggering the race on s390x and intel.
Presumably also on power, but I don't have access to a power machine
with lock-elision. At least the code for power is the same as on the other
two architectures.
ChangeLog:
[BZ #23275]
* nptl/tst-mutex10.c: New File.
* nptl/Makefile (tests): Add tst-mutex10.
(tst-mutex10-ENV): New variable.
* sysdeps/unix/sysv/linux/s390/force-elision.h: (FORCE_ELISION):
Ensure that elision path is used if elision is available.
* sysdeps/unix/sysv/linux/powerpc/force-elision.h (FORCE_ELISION):
Likewise.
* sysdeps/unix/sysv/linux/x86/force-elision.h: (FORCE_ELISION):
Likewise.
* nptl/pthreadP.h (PTHREAD_MUTEX_TYPE, PTHREAD_MUTEX_TYPE_ELISION)
(PTHREAD_MUTEX_PSHARED): Use atomic_load_relaxed.
* nptl/pthread_mutex_consistent.c (pthread_mutex_consistent): Likewise.
* nptl/pthread_mutex_getprioceiling.c (pthread_mutex_getprioceiling):
Likewise.
* nptl/pthread_mutex_lock.c (__pthread_mutex_lock_full)
(__pthread_mutex_cond_lock_adjust): Likewise.
* nptl/pthread_mutex_setprioceiling.c (pthread_mutex_setprioceiling):
Likewise.
* nptl/pthread_mutex_timedlock.c (__pthread_mutex_timedlock): Likewise.
* nptl/pthread_mutex_trylock.c (__pthread_mutex_trylock): Likewise.
* nptl/pthread_mutex_unlock.c (__pthread_mutex_unlock_full): Likewise.
* sysdeps/nptl/bits/thread-shared-types.h (struct __pthread_mutex_s):
Add comments.
* nptl/pthread_mutex_destroy.c (__pthread_mutex_destroy):
Use atomic_load_relaxed and atomic_store_relaxed.
* nptl/pthread_mutex_init.c (__pthread_mutex_init):
Use atomic_store_relaxed.
Diffstat (limited to 'nptl/pthreadP.h')
| -rw-r--r-- | nptl/pthreadP.h | 12 |
1 files changed, 8 insertions, 4 deletions
diff --git a/nptl/pthreadP.h b/nptl/pthreadP.h index 13bdb11133..19efe1e35f 100644 --- a/nptl/pthreadP.h +++ b/nptl/pthreadP.h @@ -110,19 +110,23 @@ enum }; #define PTHREAD_MUTEX_PSHARED_BIT 128 +/* See concurrency notes regarding __kind in struct __pthread_mutex_s + in sysdeps/nptl/bits/thread-shared-types.h. */ #define PTHREAD_MUTEX_TYPE(m) \ - ((m)->__data.__kind & 127) + (atomic_load_relaxed (&((m)->__data.__kind)) & 127) /* Don't include NO_ELISION, as that type is always the same as the underlying lock type. */ #define PTHREAD_MUTEX_TYPE_ELISION(m) \ - ((m)->__data.__kind & (127|PTHREAD_MUTEX_ELISION_NP)) + (atomic_load_relaxed (&((m)->__data.__kind)) \ + & (127 | PTHREAD_MUTEX_ELISION_NP)) #if LLL_PRIVATE == 0 && LLL_SHARED == 128 # define PTHREAD_MUTEX_PSHARED(m) \ - ((m)->__data.__kind & 128) + (atomic_load_relaxed (&((m)->__data.__kind)) & 128) #else # define PTHREAD_MUTEX_PSHARED(m) \ - (((m)->__data.__kind & 128) ? LLL_SHARED : LLL_PRIVATE) + ((atomic_load_relaxed (&((m)->__data.__kind)) & 128) \ + ? LLL_SHARED : LLL_PRIVATE) #endif /* The kernel when waking robust mutexes on exit never uses |