jyapayne/musl
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

fix excessive/insufficient wakes in __vm_unlock

Browse source 
there is no need to send a wake when the lock count does not hit zero,
but when it does, all waiters must be woken (since all with the same
sign are eligible to obtain the lock).
This commit is contained in:
Rich Felker 2011-09-28 19:45:37 -04:00
commit de543b05c8
1 changed files with 3 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

6
src/thread/pthread_barrier_wait.c
View file

@ -13,9 +13,9 @@ void __vm_lock(int inc)
void __vm_unlock(void)
{
if (vmlock[0]>0) a_dec(vmlock);
else a_inc(vmlock);
if (vmlock[1]) __wake(vmlock, 1, 1);
int inc = vmlock[0]>0 ? -1 : 1;
if (a_fetch_add(vmlock, inc)==-inc && vmlock[1])
__wake(vmlock, -1, 1);
}
static int pshared_barrier_wait(pthread_barrier_t *b)