http://bugs.winehq.org/show_bug.cgi?id=2618
Summary: dsound buffer underrun lockup stutter .. again... Product: Wine Version: CVS Platform: Other OS/Version: Linux Status: UNCONFIRMED Severity: major Priority: P2 Component: wine-directx AssignedTo: wine-bugs@winehq.org ReportedBy: spamalltheway@bigfoot.com
Wine (CVS 2004-12-21)
Game tried: Heroes of Might and Magic 3 / III Complete (Exe version v4.0)
Files: - Log of full dsound,wave output - Log of only the default error output - lspci dump of the used hardware - ver_linux output
No full log since it was hitting 2.4 meg lines after only 6 seconds runtime ;-)
Tried sound output methods + "results": - ALSA: In the intro the sound output consists of something sounding like a very _fast_ reset being done on the soundcard every .1 to .2 seconds (accompanied by tons of messages along the lines of:
fixme:dsound:DSOUND_MixOne problem with underrun detection (mixlen=8832 < primary_done=8900) err:dsound:DSOUND_MixOne underrun on sound buffer 0x403f9d30
).
The sound starts playing in the Mainmenu afterwards. When ever you put only a little additional stress on the CPU like eg. moving the mouse cursor (graphical cursor in HoMM 3) or doing a "ps ax" in a shell window repeatedly) the sound stutters _badly_, accompanied by messages along the lines of:
fixme:dsound:DSOUND_MixOne problem with underrun detection (mixlen=2302 < primary_done=4782) err:dsound:DSOUND_MixOne underrun on sound buffer 0x403f9d30
At this point they are mostly of the first type though.
Creating additional stress for 1.5 seconds or above often makes the sound gui crash (no sound, but the rest of Wine and the game are still running). Sometimes whole Wine decends into a dead lock without _any_ error messages being printed (=> pkill -9 wine*).
In case of deadlock the usual message occurs ;-) : err:ntdll:RtlpWaitForCriticalSection section 0x403f86f4 "?" wait timed out in thread 000c, blocked by 0009, retrying (60 sec)
- OSS: No sound at all with hardwareacceleration != Emulation, only a black screen instead of the loading game with hardacceleration == Emulation
- Jack: A little sound, although stuttering more badly (mostly due to jackd creating a little stress to the cpu all the time, just like moving your mouse in the ALSA 'chapter').
- ArtS: Same as jack, but stuttering without needing any additional stress on the CPU (maybe this is again CPU load since artsd is surely creating a little when having clients).
Changing the settings in dsound to incredibly high/low values didn't help at all (except for creating a constant stutter when making the buffer size too small).
Software: Wine is the CVS version as of 2004-12-21. Linux 2.6.8.1 Debian Sarge / Unstable (if related at all)
Hardware: CPU: Pentium 4 Celeron 2.4Ghz For the rest see lspci output.