Hi,
while trying to speed up Xing DVD Player I came across the following rather
uncool problem:
Critical Sections take way too much time. I have added the appended patch
to EnterCriticalSection():
It prints an S every 20 seconds on my K6-200, meaning it did at least
3*1000 (wine->wineserver->wine) process context switches (not counting
the switches caused by the owning thread).
I can't say for sure, but this might be sucking up way too expensive CPU
power ;)
A solution would be too change handling of local critical sections to use
either UNIX IPC semaphores direct or 'busy wait' using a local wait
(like usleep(), select() and/or sched_yield()).
Both would avoid 2 of 3 process context switches.
But before I start hacking like madly for myself... Any opinions?
Ciao, Marcus
Index: critsection.c
===================================================================
RCS file: /home/wine/wine/scheduler/critsection.c,v
retrieving revision 1.20
diff -u -r1.20 critsection.c
--- critsection.c 2000/06/07 02:15:40 1.20
+++ critsection.c 2000/07/28 14:09:48
@@ -69,6 +69,8 @@
*/
void WINAPI EnterCriticalSection( CRITICAL_SECTION *crit )
{
+ static int slowpathhit = 0;
+
if (InterlockedIncrement( &crit->LockCount ))
{
if (crit->OwningThread == GetCurrentThreadId())
@@ -76,6 +78,13 @@
crit->RecursionCount++;
return;
}
+
+ if (!HIWORD(crit->LockSemaphore)) {
+ slowpathhit++;
+ if (slowpathhit>1000) {
+ fprintf(stderr,"S");slowpathhit = 0;
+ }
+ }
/* Now wait for it */
for (;;)
