http://bugs.winehq.org/show_bug.cgi?id=14395
Summary: World of Warcraft regular crashes using NVidia Geforce 9600 GT Product: Wine Version: 1.1.0 Platform: PC-x86-64 OS/Version: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: -unknown AssignedTo: wine-bugs@winehq.org ReportedBy: vallidor@gangrenegang.com
Created an attachment (id=14709) --> (http://bugs.winehq.org/attachment.cgi?id=14709) Terse output of wine.
Frequent in-game crashes with no criteria required to reproduce other than logging into the game and "standing around".
The only required factor is patience. The first crash typically occurs 30-minutes into a session, with the second 15-20 minutes after restarting, with gradually increasing frequency.
Eventually, crashes will occur within seconds or minutes of each other, and by this threshold frequently cause the X server to crash.
I have tested various NVidia driver versions, and two AMD64 systems with the same video card on two different distributions (Ubuntu Gutsy/Hardy x86_64 hybrid, and Sabayon 3.5 x86_64 fresh) as well as older versions of wine (0.9.6 and up) up to and including 1.1.0.
Although I suspect this is an NVidia driver issue related to this card, I have read similar reports of World of Warcraft itself crashing on the Win32 platform with newer generation hardware as well, and as such am reporting this bug to Nvidia, Wine, and Blizzard.
Attached is the wine output in text format.
When crashing, a memory violation error window is displayed as well as the Blizzard bug report feedback window.
http://bugs.winehq.org/show_bug.cgi?id=14395
Adys adys.wh+winehqdotorg@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |adys.wh+winehqdotorg@gmail.c | |om
--- Comment #1 from Adys adys.wh+winehqdotorg@gmail.com 2008-11-06 06:13:33 --- err:wgl:X11DRV_wglCreatePbufferARB ((nil)): unexpected iPixelFormat(0) <= 0, returns NULL
I haven't seen this error before. Does this still happen in 1.1.7? If so, could you attach a relevant error log (found in World of Warcraft/Errors folder, sorted by date).
http://bugs.winehq.org/show_bug.cgi?id=14395
Jason Rosenberg smitty_the_smith@yahoo.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |smitty_the_smith@yahoo.com
--- Comment #2 from Jason Rosenberg smitty_the_smith@yahoo.com 2008-11-18 16:08:13 ---
Although I suspect this is an NVidia driver issue related to this card,
I have a GeForce 8400 GS and I've noticed Ubuntu (maybe Debian or linux as a whole I don't know) have installed Nvidia drivers poorly, so I use the .run files from Nvidia.com and they work much better. I hope that helps. good luck.
http://bugs.winehq.org/show_bug.cgi?id=14395
Karsten Elfenbein kelfe@gmx.de changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |kelfe@gmx.de
--- Comment #3 from Karsten Elfenbein kelfe@gmx.de 2008-11-28 08:49:00 --- do you have more than 4GB memory installed on that machine? if so what does top say in the VIRT colum of the wow process when you get close to a crash?
http://bugs.winehq.org/show_bug.cgi?id=14395
CHRIS@BAD2BEEF chris@bad2beef.net changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |chris@bad2beef.net
--- Comment #4 from CHRIS@BAD2BEEF chris@bad2beef.net 2008-12-03 23:42:52 --- (In reply to comment #3)
do you have more than 4GB memory installed on that machine? if so what does top say in the VIRT colum of the wow process when you get close to a crash?
I am having a similar problem, and for me the answer to your first question is yes.
I was monitoring memory utilization and the virtual image size slowly crept up to 4092MB where it stayed for some time. I never did see it cross the 4096 mark. WoW crashed while still at 4092 but may have been trying to allocate more.
I tend to notice this problem when switching areas, and new textures for those areas need to be loaded. If I'm very lucky, the actual crash is proceeded by various missing textures, mostly in the UI and map. On very rare occasions I will see a shaded but untextured polygon.
I've tried this with multiple versions of Wine and the NVIDIA driver blobs. I will attempt to through some notes together when I have time.
http://bugs.winehq.org/show_bug.cgi?id=14395
Dan Kegel dank@kegel.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |dank@kegel.com
--- Comment #5 from Dan Kegel dank@kegel.com 2008-12-04 04:43:24 --- Sweet. I wonder if a short run of the app under valgrind would reveal the leak.
http://bugs.winehq.org/show_bug.cgi?id=14395
--- Comment #6 from Karsten Elfenbein kelfe@gmx.de 2008-12-04 12:46:51 --- ok what I know about the leaking of memory under wine with more than 4gb ram
first I used 8gb memory wow and other apps start at around 3.8gb virt size and increase at different speed to 4gb mark where the apps starts to act up/crash apps showing this are wow while changing zones for just over time, company of heroes at the loading screen (has about 3-4k fps then drops to 150fps when it hits the 4gb mark after a few seconds) team fortress 2 is affected as well (probably all steam apps)
happens with wine and crossover a way to reduce the effects are basicly to not use the 8gb (I used the "mem" kernel parameter to limit the memory amount to 4gb) swap size is about 12gb on my system
http://bugs.winehq.org/show_bug.cgi?id=14395
Denis Zhdanov altwazar@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |altwazar@gmail.com
--- Comment #7 from Denis Zhdanov altwazar@gmail.com 2008-12-06 09:43:38 --- Confirm 8800 gtx ultra. WoW crushed, when WoW.exe near 4066 of virtual memory. 5953 sad 20 0 4063m 1.0g 28m S 4 34.5 33:54.25 WoW.exe
http://bugs.winehq.org/show_bug.cgi?id=14395
--- Comment #8 from CHRIS@BAD2BEEF chris@bad2beef.net 2008-12-06 17:02:12 --- So assuming this is the problem, what do we do about it? Limiting my entire 4 core 8GB system for 4GB of memory just for one app seems silly. Why does something that provide a 32 bit API allow an app to allocate more memory than it can address?
Being not at all familiar with the Wine code, what I can I do to help get this resolved? What information or testing is needed to get a better grip on this? Are there any workarounds in the mean time?
If I attempt ulimit at 2 or 3GB, I have issues loading libraries and cannot run the app. If I set it just under 4GB, it will eventually still blow up.
http://bugs.winehq.org/show_bug.cgi?id=14395
--- Comment #9 from CHRIS@BAD2BEEF chris@bad2beef.net 2008-12-06 17:20:21 --- Created an attachment (id=17696) --> (http://bugs.winehq.org/attachment.cgi?id=17696) Last few lines of stderr during a crash
This happened during a ulimit test. Numbers are:
10911 chris 20 0 3911m 510m 204m S 3 6.4 8:39.78 WoW.exe
http://bugs.winehq.org/show_bug.cgi?id=14395
--- Comment #10 from CHRIS@BAD2BEEF chris@bad2beef.net 2008-12-06 17:38:21 --- Created an attachment (id=17697) --> (http://bugs.winehq.org/attachment.cgi?id=17697) Last few lines of stderr during crash without ulimit
Much more useful.
http://bugs.winehq.org/show_bug.cgi?id=14395
--- Comment #11 from Adys adys.wh+winehqdotorg@gmail.com 2008-12-06 22:31:58 --- (In reply to comment #10)
Created an attachment (id=17697)
--> (http://bugs.winehq.org/attachment.cgi?id=17697) [details]
Last few lines of stderr during crash without ulimit
Much more useful.
Could you have a read through bug 16131? This sounds like the same issue.
http://bugs.winehq.org/show_bug.cgi?id=14395
--- Comment #12 from Denis Zhdanov altwazar@gmail.com 2008-12-07 00:03:05 --- sad@sad-desktop ~ $ tail -n 30 WOWERROR fixme:win:EnumDisplayDevicesW ((null),0,0x3af59c,0x00000000), stub! fixme:win:EnumDisplayDevicesW ((null),0,0x3af530,0x00000000), stub! fixme:win:EnumDisplayDevicesW ((null),0,0x3af520,0x00000000), stub! fixme:win:EnumDisplayDevicesW ((null),0,0x3af018,0x00000000), stub! fixme:win:EnumDisplayDevicesW ((null),0,0x3af150,0x00000000), stub! fixme:dsalsa:IDsDriverBufferImpl_SetVolumePan (0x139cc8,0x139bc8): stub fixme:win:EnumDisplayDevicesW ((null),0,0x3adf1c,0x00000000), stub! fixme:win:EnumDisplayDevicesW ((null),0,0x3adf44,0x00000000), stub! fixme:wininet:InternetSetOptionW Option INTERNET_OPTION_CONNECT_TIMEOUT (5000): STUB fixme:wininet:InternetSetOptionW INTERNET_OPTION_SEND/RECEIVE_TIMEOUT 5000 fixme:wininet:InternetSetOptionW Option INTERNET_OPTION_CONNECT_TIMEOUT (5000): STUB fixme:wininet:InternetSetOptionW INTERNET_OPTION_SEND/RECEIVE_TIMEOUT 5000 fixme:reg:GetNativeSystemInfo (0x374025c4) using GetSystemInfo() fixme:wininet:InternetSetOptionW Option INTERNET_OPTION_CONTEXT_VALUE; STUB fixme:wininet:InternetSetOptionW Option INTERNET_OPTION_CONTEXT_VALUE; STUB fixme:imm:ImmReleaseContext (0x30050, 0x138bd0): stub fixme:winsock:WSAIoctl unsupported WS_IOCTL cmd (9800000c) fixme:win:EnumDisplayDevicesW ((null),0,0x3adae4,0x00000000), stub! fixme:win:EnumDisplayDevicesW ((null),0,0x3ad30c,0x00000000), stub! fixme:win:EnumDisplayDevicesW ((null),0,0x3ad30c,0x00000000), stub! fixme:win:EnumDisplayDevicesW ((null),0,0x3ad30c,0x00000000), stub! fixme:win:EnumDisplayDevicesW ((null),0,0x3ad30c,0x00000000), stub! fixme:win:EnumDisplayDevicesW ((null),0,0x3aded4,0x00000000), stub! fixme:winsock:WSAIoctl unsupported WS_IOCTL cmd (9800000c) fixme:win:EnumDisplayDevicesW ((null),0,0x3adae4,0x00000000), stub! err:ntdll:RtlpWaitForCriticalSection section 0x624d33c "?" wait timed out in thread 0035, blocked by 0027, retrying (60 sec) fixme:win:EnumDisplayDevicesW ((null),0,0x3ad30c,0x00000000), stub! fixme:win:EnumDisplayDevicesW ((null),0,0x3ad30c,0x00000000), stub! fixme:win:EnumDisplayDevicesW ((null),0,0x3ad30c,0x00000000), stub! fixme:win:EnumDisplayDevicesW ((null),0,0x3ad30c,0x00000000), stub!
Last 30 lines from stderr.
http://bugs.winehq.org/show_bug.cgi?id=14395
--- Comment #13 from Denis Zhdanov altwazar@gmail.com 2008-12-07 00:05:14 --- sad@sad-desktop ~ $ ulimit -a | grep virtual virtual memory (kbytes, -v) unlimited
I don't try to limit virtual memory.
http://bugs.winehq.org/show_bug.cgi?id=14395
Alexander Trauzzi atrauzzi@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |atrauzzi@gmail.com
--- Comment #14 from Alexander Trauzzi atrauzzi@gmail.com 2008-12-07 09:31:02 --- I have to note that I have been experiencing intermittent freezes as well.
The most I have observed is that they are totally random, do not occur immediately and cause the game to hang (but not my X server). My system is not running hot at all, and this never used to happen in previous combinations of Ubuntu and WINE.
Is there any way I can generate crash information without having the console interfere with gameplay? Let me know!
http://bugs.winehq.org/show_bug.cgi?id=14395
Mathieu Rini lawcorn@lawcorn.net changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|UNCONFIRMED |NEW Ever Confirmed|0 |1
--- Comment #15 from Mathieu Rini lawcorn@lawcorn.net 2008-12-09 01:51:13 --- *** This bug has been confirmed by popular vote. ***
http://bugs.winehq.org/show_bug.cgi?id=14395
Mathieu Rini lawcorn@lawcorn.net changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |lawcorn@lawcorn.net
http://bugs.winehq.org/show_bug.cgi?id=14395
Kai Blaschke webmaster@thw-theorie.de changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |webmaster@thw-theorie.de
--- Comment #16 from Kai Blaschke webmaster@thw-theorie.de 2008-12-09 03:01:00 --- I'd like to confirm this bug, and add my own experience.
I'm using a GeForce 7800GT, my system is running with 2 GB of RAM. During gameplay, the virtual memory usage of WoW is around 3050 MB, but when I enter Dalaran, or a zone from where it can be seen, memory usage jumps up to 4000 MB. As mentioned in comment #4, WoW is likely to crash if the value reaches 4096 MB.
To reproduce it, I just need to set my graphic settings to high, and walk around in dalaran on prime time. After a few seconds, random, untextured polygons start to appear for a short while, and finally WoW crashes. Crashes in other areas are less frequent, but *always* happen when the virtual memory usage exceeds 4GB.
http://bugs.winehq.org/show_bug.cgi?id=14395
Mike Krüger mike@icsharpcode.net changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |mike@icsharpcode.net
--- Comment #17 from Mike Krüger mike@icsharpcode.net 2008-12-09 09:16:44 --- confirmed with Geforce 9800GTX and 8GB RAM. I now using a cheap ATI card (their bugs are less serve, maybe I'll upgrade it to a better one if the ATI drivers become more decent).
However the best would be fixing this bug :)
http://bugs.winehq.org/show_bug.cgi?id=14395
Bas Koopmans winebugs@badako.nl changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |winebugs@badako.nl
--- Comment #18 from Bas Koopmans winebugs@badako.nl 2008-12-22 02:30:37 --- Confirmed with a NVidia 9800GTX+ running on Gentoo x86_64 Running it on my previous card (NV 7800GTX) yields the same crashes and a slightly reduced framerate.
I have no non-NV card to test with, unfortunately.
top shows over 4090MB of virtual memory usage, but if I do not venture into Dalaran, or stay away from certain sections of Dragonblight and Crystalsong Forest, the crash _will not_ occur and I can play pretty much indefinetly. (This was tested... Thouroughly).
http://bugs.winehq.org/show_bug.cgi?id=14395
--- Comment #19 from Kai Blaschke webmaster@thw-theorie.de 2008-12-22 03:20:07 --- The workaround posted for bug #10778 resolved both the crashes and the fps drop for me. Now I can finally look down from Dalaran without getting a crash. Memory usage has dropped to about 2500 MB vmem, even on busy days with lots of players in visible range.
BUT - even if this helps, I won't consider this patch a solution, as it just moves the limit a bit further. As mentioned in the comments, the actual problem lies deep inside the Wine virtual memory handler, is is not an easy fix even for the responsible developers. There is a lengthy discussion at bug #13335, start here:
http://bugs.winehq.org/show_bug.cgi?id=13335#c96
http://bugs.winehq.org/show_bug.cgi?id=14395
Jason Rosenberg smitty_the_smith@yahoo.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC|smitty_the_smith@yahoo.com |
http://bugs.winehq.org/show_bug.cgi?id=14395
--- Comment #20 from Adys adys.wh+winehqdotorg@gmail.com 2009-01-21 18:13:21 --- This bug is a duplicate of bug 16131, which itself is a duplicate of bug 10778. Probably best to group those up.
http://bugs.winehq.org/show_bug.cgi?id=14395
Jeff Zaroyko jeffz@jeffz.name changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE
--- Comment #21 from Jeff Zaroyko jeffz@jeffz.name 2009-05-23 09:30:08 --- marking duplicate as per comment 19 and 20.
*** This bug has been marked as a duplicate of bug 10778 ***
http://bugs.winehq.org/show_bug.cgi?id=14395
Jeff Zaroyko jeffz@jeffz.name changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #22 from Jeff Zaroyko jeffz@jeffz.name 2009-05-23 09:32:24 --- closing
-
wine-bugs＠winehq.org