| Message ID | 4adc080f-9975-4ee0-003e-675f3c9f1203@linux.vnet.ibm.com |
|---|---|
| State | New |
| Headers | show |
On 12/13/2016 09:54 AM, Stefan Liebler wrote: > Hi, > > the macros lll_trylock, lll_cond_trylock are extended by a > __glibc_unlikely hint. Now the trylock macros are based on the > same assumption about a free/busy lock as lll_lock. > With the hint gcc emits code in e.g. pthread_mutex_trylock which does > not use jumps if the lock is free. Without the hint it had to jump away > if the lock is free. > > Tested on s390x, ppc. > > Okay to commit? > > Bye > Stefan > > ChangeLog: > > * sysdeps/nptl/lowlevellock.h (lll_trylock, lll_cond_trylock): > Add __glibc_unlikely hint. ping
On Tuesday 13 December 2016 02:24 PM, Stefan Liebler wrote: > Hi, > > the macros lll_trylock, lll_cond_trylock are extended by a > __glibc_unlikely hint. Now the trylock macros are based on the > same assumption about a free/busy lock as lll_lock. > With the hint gcc emits code in e.g. pthread_mutex_trylock which does > not use jumps if the lock is free. Without the hint it had to jump away > if the lock is free. > > Tested on s390x, ppc. > > Okay to commit? Yes, looks good. Siddhesh
On 02/06/2017 09:33 AM, Siddhesh Poyarekar wrote: > On Tuesday 13 December 2016 02:24 PM, Stefan Liebler wrote: >> Hi, >> >> the macros lll_trylock, lll_cond_trylock are extended by a >> __glibc_unlikely hint. Now the trylock macros are based on the >> same assumption about a free/busy lock as lll_lock. >> With the hint gcc emits code in e.g. pthread_mutex_trylock which does >> not use jumps if the lock is free. Without the hint it had to jump away >> if the lock is free. >> >> Tested on s390x, ppc. >> >> Okay to commit? > > Yes, looks good. > > Siddhesh > Thanks. Committed.
commit 2e11862aaf6d563518780a470b97b59a6e0b2717
Author: Stefan Liebler <stli@linux.vnet.ibm.com>
Date: Tue Dec 13 09:14:09 2016 +0100
Add __glibc_unlikely hint in lll_trylock, lll_cond_trylock.
The macros lll_trylock, lll_cond_trylock are extended by an __glibc_unlikely
hint. Now the trylock macros are based on the same assumption about a
free/busy lock as lll_lock.
With the hint gcc emits code in e.g. pthread_mutex_trylock which does
not use jumps if the lock is free. Without the hint it had to jump away
if the lock is free.
Tested on s390x, ppc.
ChangeLog:
* sysdeps/nptl/lowlevellock.h (lll_trylock, lll_cond_trylock):
Add __glibc_unlikely hint.
diff --git a/sysdeps/nptl/lowlevellock.h b/sysdeps/nptl/lowlevellock.h
index 3890145..b77043b 100644
--- a/sysdeps/nptl/lowlevellock.h
+++ b/sysdeps/nptl/lowlevellock.h
@@ -64,13 +64,13 @@
0. Otherwise leave lock unchanged and return non-zero to indicate that the
lock was not acquired. */
#define lll_trylock(lock) \
- atomic_compare_and_exchange_bool_acq (&(lock), 1, 0)
+ __glibc_unlikely (atomic_compare_and_exchange_bool_acq (&(lock), 1, 0))
/* If LOCK is 0 (not acquired), set to 2 (acquired, possibly with waiters) and
return 0. Otherwise leave lock unchanged and return non-zero to indicate
that the lock was not acquired. */
#define lll_cond_trylock(lock) \
- atomic_compare_and_exchange_bool_acq (&(lock), 2, 0)
+ __glibc_unlikely (atomic_compare_and_exchange_bool_acq (&(lock), 2, 0))
extern void __lll_lock_wait_private (int *futex) attribute_hidden;
extern void __lll_lock_wait (int *futex, int private) attribute_hidden;