qiurui
/
glibc
mirror of git://sourceware.org/git/glibc.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
glibc/sysdeps/x86/elide.h

120 lines
3.9 KiB
C
Raw Normal View History

Add adaptive elision to rwlocks This patch relies on the C version of the rwlocks posted earlier. With C rwlocks it is very straight forward to do adaptive elision using TSX. It is based on the infrastructure added earlier for mutexes, but uses its own elision macros. The macros are fairly general purpose and could be used for other elision purposes too. This version is much cleaner than the earlier assembler based version, and in particular implements adaptation which makes it safer. I changed the behavior slightly to not require any changes in the test suite and fully conform to all expected behaviors (generally at the cost of not eliding in various situations). In particular this means the timedlock variants are not elided. Nested trylock aborts.
2014-03-31 15:07:46 +00:00
/* elide.h: Generic lock elision support.
Update copyright dates with scripts/update-copyrights. * All files with FSF copyright notices: Update copyright dates using scripts/update-copyrights. * locale/programs/charmap-kw.h: Regenerated. * locale/programs/locfile-kw.h: Likewise.
2018-01-01 00:32:25 +00:00
Copyright (C) 2014-2018 Free Software Foundation, Inc.
Add adaptive elision to rwlocks This patch relies on the C version of the rwlocks posted earlier. With C rwlocks it is very straight forward to do adaptive elision using TSX. It is based on the infrastructure added earlier for mutexes, but uses its own elision macros. The macros are fairly general purpose and could be used for other elision purposes too. This version is much cleaner than the earlier assembler based version, and in particular implements adaptation which makes it safer. I changed the behavior slightly to not require any changes in the test suite and fully conform to all expected behaviors (generally at the cost of not eliding in various situations). In particular this means the timedlock variants are not elided. Nested trylock aborts.
2014-03-31 15:07:46 +00:00
This file is part of the GNU C Library.
The GNU C Library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
The GNU C Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with the GNU C Library; if not, see
<http://www.gnu.org/licenses/>. */
#ifndef ELIDE_H
#define ELIDE_H 1
#include <hle.h>
#include <elision-conf.h>
Use C11-like atomics instead of plain memory accesses in x86 lock elision. This uses atomic operations to access lock elision metadata that is accessed concurrently (ie, adapt_count fields). The size of the data is less than a word but accessed only with atomic loads and stores; therefore, we add support for shorter-size atomic load and stores too. * include/atomic.h (__atomic_check_size_ls): New. (atomic_load_relaxed, atomic_load_acquire, atomic_store_relaxed, atomic_store_release): Use it. * sysdeps/x86/elide.h (ACCESS_ONCE): Remove. (elision_adapt, ELIDE_LOCK): Use atomics. * sysdeps/unix/sysv/linux/x86/elision-lock.c (__lll_lock_elision): Use atomics and improve code comments. * sysdeps/unix/sysv/linux/x86/elision-trylock.c (__lll_trylock_elision): Likewise.
2016-11-30 16:53:11 +00:00
#include <atomic.h>
Add adaptive elision to rwlocks This patch relies on the C version of the rwlocks posted earlier. With C rwlocks it is very straight forward to do adaptive elision using TSX. It is based on the infrastructure added earlier for mutexes, but uses its own elision macros. The macros are fairly general purpose and could be used for other elision purposes too. This version is much cleaner than the earlier assembler based version, and in particular implements adaptation which makes it safer. I changed the behavior slightly to not require any changes in the test suite and fully conform to all expected behaviors (generally at the cost of not eliding in various situations). In particular this means the timedlock variants are not elided. Nested trylock aborts.
2014-03-31 15:07:46 +00:00
/* Adapt elision with ADAPT_COUNT and STATUS and decide retries. */
static inline bool
Fix build warning in pthread_rwlock_* The first argument of elision_adapt and that of ELISION_*LOCK have different signs since __elision_rwcount is signed char * and the argument of elision_adapt is uint8_t *. Modified elision_adapt to accept signed char * instead of uint8_t *.
2014-06-24 10:42:43 +00:00
elision_adapt(signed char *adapt_count, unsigned int status)
Add adaptive elision to rwlocks This patch relies on the C version of the rwlocks posted earlier. With C rwlocks it is very straight forward to do adaptive elision using TSX. It is based on the infrastructure added earlier for mutexes, but uses its own elision macros. The macros are fairly general purpose and could be used for other elision purposes too. This version is much cleaner than the earlier assembler based version, and in particular implements adaptation which makes it safer. I changed the behavior slightly to not require any changes in the test suite and fully conform to all expected behaviors (generally at the cost of not eliding in various situations). In particular this means the timedlock variants are not elided. Nested trylock aborts.
2014-03-31 15:07:46 +00:00
{
if (status & _XABORT_RETRY)
return false;
if ((status & _XABORT_EXPLICIT)
&& _XABORT_CODE (status) == _ABORT_LOCK_BUSY)
{
/* Right now we skip here. Better would be to wait a bit
and retry. This likely needs some spinning. Be careful
Use C11-like atomics instead of plain memory accesses in x86 lock elision. This uses atomic operations to access lock elision metadata that is accessed concurrently (ie, adapt_count fields). The size of the data is less than a word but accessed only with atomic loads and stores; therefore, we add support for shorter-size atomic load and stores too. * include/atomic.h (__atomic_check_size_ls): New. (atomic_load_relaxed, atomic_load_acquire, atomic_store_relaxed, atomic_store_release): Use it. * sysdeps/x86/elide.h (ACCESS_ONCE): Remove. (elision_adapt, ELIDE_LOCK): Use atomics. * sysdeps/unix/sysv/linux/x86/elision-lock.c (__lll_lock_elision): Use atomics and improve code comments. * sysdeps/unix/sysv/linux/x86/elision-trylock.c (__lll_trylock_elision): Likewise.
2016-11-30 16:53:11 +00:00
to avoid writing the lock.
Using relaxed MO and separate atomic accesses is sufficient because
adapt_count is just a hint. */
if (atomic_load_relaxed (adapt_count) != __elision_aconf.skip_lock_busy)
atomic_store_relaxed (adapt_count, __elision_aconf.skip_lock_busy);
Add adaptive elision to rwlocks This patch relies on the C version of the rwlocks posted earlier. With C rwlocks it is very straight forward to do adaptive elision using TSX. It is based on the infrastructure added earlier for mutexes, but uses its own elision macros. The macros are fairly general purpose and could be used for other elision purposes too. This version is much cleaner than the earlier assembler based version, and in particular implements adaptation which makes it safer. I changed the behavior slightly to not require any changes in the test suite and fully conform to all expected behaviors (generally at the cost of not eliding in various situations). In particular this means the timedlock variants are not elided. Nested trylock aborts.
2014-03-31 15:07:46 +00:00
}
/* Internal abort. There is no chance for retry.
Use the normal locking and next time use lock.
Use C11-like atomics instead of plain memory accesses in x86 lock elision. This uses atomic operations to access lock elision metadata that is accessed concurrently (ie, adapt_count fields). The size of the data is less than a word but accessed only with atomic loads and stores; therefore, we add support for shorter-size atomic load and stores too. * include/atomic.h (__atomic_check_size_ls): New. (atomic_load_relaxed, atomic_load_acquire, atomic_store_relaxed, atomic_store_release): Use it. * sysdeps/x86/elide.h (ACCESS_ONCE): Remove. (elision_adapt, ELIDE_LOCK): Use atomics. * sysdeps/unix/sysv/linux/x86/elision-lock.c (__lll_lock_elision): Use atomics and improve code comments. * sysdeps/unix/sysv/linux/x86/elision-trylock.c (__lll_trylock_elision): Likewise.
2016-11-30 16:53:11 +00:00
Be careful to avoid writing to the lock. See above for MO. */
else if (atomic_load_relaxed (adapt_count)
!= __elision_aconf.skip_lock_internal_abort)
atomic_store_relaxed (adapt_count,
__elision_aconf.skip_lock_internal_abort);
Add adaptive elision to rwlocks This patch relies on the C version of the rwlocks posted earlier. With C rwlocks it is very straight forward to do adaptive elision using TSX. It is based on the infrastructure added earlier for mutexes, but uses its own elision macros. The macros are fairly general purpose and could be used for other elision purposes too. This version is much cleaner than the earlier assembler based version, and in particular implements adaptation which makes it safer. I changed the behavior slightly to not require any changes in the test suite and fully conform to all expected behaviors (generally at the cost of not eliding in various situations). In particular this means the timedlock variants are not elided. Nested trylock aborts.
2014-03-31 15:07:46 +00:00
return true;
}
/* is_lock_free must be executed inside the transaction */
/* Returns true if lock defined by IS_LOCK_FREE was elided.
Use C11-like atomics instead of plain memory accesses in x86 lock elision. This uses atomic operations to access lock elision metadata that is accessed concurrently (ie, adapt_count fields). The size of the data is less than a word but accessed only with atomic loads and stores; therefore, we add support for shorter-size atomic load and stores too. * include/atomic.h (__atomic_check_size_ls): New. (atomic_load_relaxed, atomic_load_acquire, atomic_store_relaxed, atomic_store_release): Use it. * sysdeps/x86/elide.h (ACCESS_ONCE): Remove. (elision_adapt, ELIDE_LOCK): Use atomics. * sysdeps/unix/sysv/linux/x86/elision-lock.c (__lll_lock_elision): Use atomics and improve code comments. * sysdeps/unix/sysv/linux/x86/elision-trylock.c (__lll_trylock_elision): Likewise.
2016-11-30 16:53:11 +00:00
ADAPT_COUNT is a per-lock state variable; it must be accessed atomically
to avoid data races but is just a hint, so using relaxed MO and separate
atomic loads and stores instead of atomic read-modify-write operations is
sufficient. */
Add adaptive elision to rwlocks This patch relies on the C version of the rwlocks posted earlier. With C rwlocks it is very straight forward to do adaptive elision using TSX. It is based on the infrastructure added earlier for mutexes, but uses its own elision macros. The macros are fairly general purpose and could be used for other elision purposes too. This version is much cleaner than the earlier assembler based version, and in particular implements adaptation which makes it safer. I changed the behavior slightly to not require any changes in the test suite and fully conform to all expected behaviors (generally at the cost of not eliding in various situations). In particular this means the timedlock variants are not elided. Nested trylock aborts.
2014-03-31 15:07:46 +00:00
#define ELIDE_LOCK(adapt_count, is_lock_free) \
({ \
int ret = 0; \
\
Use C11-like atomics instead of plain memory accesses in x86 lock elision. This uses atomic operations to access lock elision metadata that is accessed concurrently (ie, adapt_count fields). The size of the data is less than a word but accessed only with atomic loads and stores; therefore, we add support for shorter-size atomic load and stores too. * include/atomic.h (__atomic_check_size_ls): New. (atomic_load_relaxed, atomic_load_acquire, atomic_store_relaxed, atomic_store_release): Use it. * sysdeps/x86/elide.h (ACCESS_ONCE): Remove. (elision_adapt, ELIDE_LOCK): Use atomics. * sysdeps/unix/sysv/linux/x86/elision-lock.c (__lll_lock_elision): Use atomics and improve code comments. * sysdeps/unix/sysv/linux/x86/elision-trylock.c (__lll_trylock_elision): Likewise.
2016-11-30 16:53:11 +00:00
if (atomic_load_relaxed (&(adapt_count)) <= 0) \
Add adaptive elision to rwlocks This patch relies on the C version of the rwlocks posted earlier. With C rwlocks it is very straight forward to do adaptive elision using TSX. It is based on the infrastructure added earlier for mutexes, but uses its own elision macros. The macros are fairly general purpose and could be used for other elision purposes too. This version is much cleaner than the earlier assembler based version, and in particular implements adaptation which makes it safer. I changed the behavior slightly to not require any changes in the test suite and fully conform to all expected behaviors (generally at the cost of not eliding in various situations). In particular this means the timedlock variants are not elided. Nested trylock aborts.
2014-03-31 15:07:46 +00:00
{ \
for (int i = __elision_aconf.retry_try_xbegin; i > 0; i--) \
{ \
unsigned int status; \
if ((status = _xbegin ()) == _XBEGIN_STARTED) \
{ \
if (is_lock_free) \
{ \
ret = 1; \
break; \
} \
_xabort (_ABORT_LOCK_BUSY); \
} \
if (!elision_adapt (&(adapt_count), status)) \
break; \
} \
} \
else \
Use C11-like atomics instead of plain memory accesses in x86 lock elision. This uses atomic operations to access lock elision metadata that is accessed concurrently (ie, adapt_count fields). The size of the data is less than a word but accessed only with atomic loads and stores; therefore, we add support for shorter-size atomic load and stores too. * include/atomic.h (__atomic_check_size_ls): New. (atomic_load_relaxed, atomic_load_acquire, atomic_store_relaxed, atomic_store_release): Use it. * sysdeps/x86/elide.h (ACCESS_ONCE): Remove. (elision_adapt, ELIDE_LOCK): Use atomics. * sysdeps/unix/sysv/linux/x86/elision-lock.c (__lll_lock_elision): Use atomics and improve code comments. * sysdeps/unix/sysv/linux/x86/elision-trylock.c (__lll_trylock_elision): Likewise.
2016-11-30 16:53:11 +00:00
atomic_store_relaxed (&(adapt_count), \
atomic_load_relaxed (&(adapt_count)) - 1); \
Add adaptive elision to rwlocks This patch relies on the C version of the rwlocks posted earlier. With C rwlocks it is very straight forward to do adaptive elision using TSX. It is based on the infrastructure added earlier for mutexes, but uses its own elision macros. The macros are fairly general purpose and could be used for other elision purposes too. This version is much cleaner than the earlier assembler based version, and in particular implements adaptation which makes it safer. I changed the behavior slightly to not require any changes in the test suite and fully conform to all expected behaviors (generally at the cost of not eliding in various situations). In particular this means the timedlock variants are not elided. Nested trylock aborts.
2014-03-31 15:07:46 +00:00
ret; \
})
/* Returns true if lock defined by IS_LOCK_FREE was try-elided.
Use C11-like atomics instead of plain memory accesses in x86 lock elision. This uses atomic operations to access lock elision metadata that is accessed concurrently (ie, adapt_count fields). The size of the data is less than a word but accessed only with atomic loads and stores; therefore, we add support for shorter-size atomic load and stores too. * include/atomic.h (__atomic_check_size_ls): New. (atomic_load_relaxed, atomic_load_acquire, atomic_store_relaxed, atomic_store_release): Use it. * sysdeps/x86/elide.h (ACCESS_ONCE): Remove. (elision_adapt, ELIDE_LOCK): Use atomics. * sysdeps/unix/sysv/linux/x86/elision-lock.c (__lll_lock_elision): Use atomics and improve code comments. * sysdeps/unix/sysv/linux/x86/elision-trylock.c (__lll_trylock_elision): Likewise.
2016-11-30 16:53:11 +00:00
ADAPT_COUNT is a per-lock state variable. */
Add adaptive elision to rwlocks This patch relies on the C version of the rwlocks posted earlier. With C rwlocks it is very straight forward to do adaptive elision using TSX. It is based on the infrastructure added earlier for mutexes, but uses its own elision macros. The macros are fairly general purpose and could be used for other elision purposes too. This version is much cleaner than the earlier assembler based version, and in particular implements adaptation which makes it safer. I changed the behavior slightly to not require any changes in the test suite and fully conform to all expected behaviors (generally at the cost of not eliding in various situations). In particular this means the timedlock variants are not elided. Nested trylock aborts.
2014-03-31 15:07:46 +00:00
#define ELIDE_TRYLOCK(adapt_count, is_lock_free, write) ({ \
int ret = 0; \
if (__elision_aconf.retry_try_xbegin > 0) \
{ \
if (write) \
_xabort (_ABORT_NESTED_TRYLOCK); \
ret = ELIDE_LOCK (adapt_count, is_lock_free); \
} \
ret; \
})
nptl: Document crash due to incorrect use of locks
2015-08-20 06:44:09 +00:00
/* Returns true if lock defined by IS_LOCK_FREE was elided. The call
to _xend crashes if the application incorrectly tries to unlock a
lock which has not been locked. */
Add adaptive elision to rwlocks This patch relies on the C version of the rwlocks posted earlier. With C rwlocks it is very straight forward to do adaptive elision using TSX. It is based on the infrastructure added earlier for mutexes, but uses its own elision macros. The macros are fairly general purpose and could be used for other elision purposes too. This version is much cleaner than the earlier assembler based version, and in particular implements adaptation which makes it safer. I changed the behavior slightly to not require any changes in the test suite and fully conform to all expected behaviors (generally at the cost of not eliding in various situations). In particular this means the timedlock variants are not elided. Nested trylock aborts.
2014-03-31 15:07:46 +00:00
#define ELIDE_UNLOCK(is_lock_free) \
({ \
int ret = 0; \
if (is_lock_free) \
{ \
_xend (); \
ret = 1; \
} \
ret; \
})
#endif