http://bugs.winehq.org/show_bug.cgi?id=20181
Summary: Warcraft 3 crashes on start with GLXBadDrawable Product: Wine Version: 1.1.25 Platform: PC OS/Version: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: -unknown AssignedTo: wine-bugs@winehq.org ReportedBy: infobox.oleg@wo.cz
Created an attachment (id=23786) --> (http://bugs.winehq.org/attachment.cgi?id=23786) wine output
I have Warcraft 3 Frozen Throne, vesion 1.22a. It crashes under wine 1.25 on. Tested with all wine versions up to 1.1.29 (1.1.30 is not compilable under gentoo currently, waiting for fix from gentoo). Under wine 1.1.24 it runs ok.
Reproduction: Command line: wine euroloader.exe -opengl wine war3.exe -opengl wine war3.exe
All above leads to this error: X Error of failed request: GLXBadDrawable Major opcode of failed request: 152 (GLX) Minor opcode of failed request: 5 (X_GLXMakeCurrent) Serial number of failed request: 794 Current serial number in output stream: 794
Switch -opengl has only one impact - resolution changes before application crash and stays changed after the crash.
Distribution: Gentoo Linux 2.6.31-gentoo #5 SMP PREEMPT Intel(R) Core(TM)2 Duo CPU T6670 @ 2.20GHz GenuineIntel GNU/Linux
Versions of software: x11-drivers/xf86-video-intel-2.8.1 x11-base/xorg-server-1.6.3.901-r2 media-libs/mesa-7.5.1 x11-libs/libdrm-2.4.14
Graphics is Intel G4500.
Oleg
http://bugs.winehq.org/show_bug.cgi?id=20181
infobox.oleg@wo.cz changed:
What |Removed |Added ---------------------------------------------------------------------------- Keywords| |regression
http://bugs.winehq.org/show_bug.cgi?id=20181
Vitaliy Margolen vitaliy@kievinfo.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Attachment #23786|application/octet-stream |text/plain mime type| |
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #1 from infobox.oleg@wo.cz 2009-09-28 04:42:51 --- I've reproduced the bug on wine 1.1.30 under Ubuntu. I'm planning to do git bisect.
Oleg
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #2 from LanoxxthShaddow sbastig@gmx.net 2009-10-14 18:47:58 --- Created an attachment (id=24141) --> (http://bugs.winehq.org/attachment.cgi?id=24141) Warcraft TFT Log on Karmic with 1.1.31
http://bugs.winehq.org/show_bug.cgi?id=20181
LanoxxthShaddow sbastig@gmx.net changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |sbastig@gmx.net
--- Comment #3 from LanoxxthShaddow sbastig@gmx.net 2009-10-14 18:50:50 --- I can confirm this bug.
Im using wine1.2 aka wine-1.1.31 from scott richies PPA on ubuntu karmic 9.10 I purged my whole wine directory just resently and have only warcraft 3 installed at the moment. I have an intel X3100 onboard graphics card with the latest drivers from Ubuntu Karmic. Warcraft is patched to the latest version.
xorg-video-intel: Version: 2:2.9.0-1ubuntu2 Linux mars 2.6.31-14-generic #46-Ubuntu SMP Tue Oct 13 16:47:28 UTC 2009 x86_64 GNU/Linux
http://bugs.winehq.org/show_bug.cgi?id=20181
Alistair ajftonner@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |ajftonner@gmail.com
--- Comment #4 from Alistair ajftonner@gmail.com 2009-10-18 21:11:14 --- Lenovo T400 00:02.0 VGA compatible controller: Intel Corporation Mobile 4 Series Chipset Integrated Graphics Controller (rev 07) (prog-if 00 [VGA controller]) Subsystem: Lenovo Device 20e4 Flags: bus master, fast devsel, latency 0, IRQ 34 [ebuild R ] x11-drivers/xf86-video-intel-2.9.0 USE="dri -debug" 0 kB
Memory at f4400000 (64-bit, non-prefetchable) [size=4M] Memory at d0000000 (64-bit, prefetchable) [size=256M] I/O ports at 1800 [size=8] Capabilities: [90] Message Signalled Interrupts: Mask- 64bit- Count=1/1 Enable+ Capabilities: [d0] Power Management version 3 Kernel driver in use: i915 Kernel modules: i915
Linux lrssbra0240 2.6.31-gentoo-r2 #3 SMP Wed Oct 14 22:08:52 EDT 2009 x86_64 Intel(R) Core(TM)2 Duo CPU P8600 @ 2.40GHz GenuineIntel GNU/Linux
[ebuild R ] x11-drivers/xf86-video-intel-2.9.0 USE="dri -debug" 0 kB [ebuild R ] media-libs/mesa-7.5.2 USE="motif nptl -debug -pic -xcb" VIDEO_CARDS="intel -mach64 -mga -none -r128 -radeon -radeonhd -s3virge -savage -sis (-sunffb) -tdfx -trident -via" 0 kB [ebuild R ] app-emulation/wine-1.1.30 USE="X alsa cups fontconfig gecko lcms ncurses opengl png samba ssl threads xcomposite xml -capi (-dbus) -esd (-gnutls) (-gphoto2) (-gsm) (-hal) -jack (-jpeg) (-ldap) (-mp3) (-nas) (-openal) -oss (-scanner) -test -win64 -xinerama" 0 kB
finally have almost everything on this laptop working, except wow, I have the following on launching
wine /opt/windows/Games/World\ of\ Warcraft/Wow.exe -opengl
fixme:win:EnumDisplayDevicesW ((null),0,0x3aed4c,0x00000000), stub! X Error of failed request: GLXBadDrawable Major opcode of failed request: 152 (GLX) Minor opcode of failed request: 5 (X_GLXMakeCurrent) Serial number of failed request: 514 Current serial number in output stream: 514
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #5 from LanoxxthShaddow sbastig@gmx.net 2009-10-25 12:42:26 --- Created an attachment (id=24348) --> (http://bugs.winehq.org/attachment.cgi?id=24348) Output of starting TrackMania Nations Forever
Interestingly TmNations starts at first but then crashes because of some other error. Then when I restart it I get the same GLX error than when I start warcraft.
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #6 from infobox.oleg@wo.cz 2009-11-07 12:55:51 --- I've installed Icewind Dale and the problem is the same - application won't start with current (1.1.32) wine, GLXBadDrawable error.
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #7 from infobox.oleg@wo.cz 2009-11-16 13:23:00 --- Results of git bisect:
390d3929f2b2a9641ecc8114cf94ae09fa851a15 is the first bad commit commit 390d3929f2b2a9641ecc8114cf94ae09fa851a15 Author: Roderick Colenbrander thunderbird2k@gmail.com Date: Fri Jul 31 22:50:36 2009 +0200
wgl: Fix usage of non-GLX visual in glXCreateContext.
:040000 040000 411c38a1206c84d87e52240249d32344f3fba0a4 bf24c5844a1cc215cc7e62fbb9e596a16c0ebfd2 M dlls bisect run success
Application used for test - Icewind Dale 1 + data disk.
Versions of software: x11-drivers/xf86-video-intel-2.9.1 media-libs/mesa-7.5.2 x11-base/xorg-server-1.7.1 x11-libs/libdrm-2.4.15
Problem persists with wine-1.1.33.
Git bisect log: git bisect start # good: [578aeb87b59b04a85db6e8e545a6065cfab8a79b] Release 1.1.26. git bisect good 578aeb87b59b04a85db6e8e545a6065cfab8a79b # bad: [b79639699766e7c8f3e236f3fef54e5daac0e5f9] Release 1.1.27. git bisect bad b79639699766e7c8f3e236f3fef54e5daac0e5f9 # good: [dddbfcdeda238946ccd79876e862a6e3d1a3a2a3] hhctrl.ocx: Convert the Russian resources to UTF-8. git bisect good dddbfcdeda238946ccd79876e862a6e3d1a3a2a3 # bad: [ae45fa6864962bf186ecc80d69b22d1543458bf9] setupapi: Add stub implementations of CM_Get_Device_ID_ListW and CM_Enumerate_Classes. git bisect bad ae45fa6864962bf186ecc80d69b22d1543458bf9 # bad: [43bfb240add6c722993801bbbb82b4244b3535a4] wined3d: Use the context's current render target the get the swapchain drawable size. git bisect bad 43bfb240add6c722993801bbbb82b4244b3535a4 # good: [f26431dc47449ee6f433218416aa8afe7cb0c51e] shdoclc: Convert the Russian resources to UTF-8. git bisect good f26431dc47449ee6f433218416aa8afe7cb0c51e # good: [6f8026821b7b1276b6901daed14ed0e470dd15a9] cmdlgtst: Add the UTF-8 pragma for the Italian resources. git bisect good 6f8026821b7b1276b6901daed14ed0e470dd15a9 # good: [77ff2df83a387eebb95f1322a4487735221cc46b] mshtml: Added IHTMLStyle3::wordWrap property implementation. git bisect good 77ff2df83a387eebb95f1322a4487735221cc46b # bad: [390d3929f2b2a9641ecc8114cf94ae09fa851a15] wgl: Fix usage of non-GLX visual in glXCreateContext. git bisect bad 390d3929f2b2a9641ecc8114cf94ae09fa851a15 # good: [09436d69f181b8bae4afea28be73edb612a6f05b] urlmon: Use one wininet internet handle for all connections. git bisect good 09436d69f181b8bae4afea28be73edb612a6f05b
http://bugs.winehq.org/show_bug.cgi?id=20181
infobox.oleg@wo.cz changed:
What |Removed |Added ---------------------------------------------------------------------------- Component|-unknown |opengl Version|1.1.25 |1.1.26
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #8 from infobox.oleg@wo.cz 2009-11-16 13:52:38 --- Also one note: actual error has changed, now it states:
X Error of failed request: BadMatch (invalid parameter attributes) Major opcode of failed request: 151 (GLX) Minor opcode of failed request: 5 (X_GLXMakeCurrent) Serial number of failed request: 532 Current serial number in output stream: 532
Versions of software has changed from comment 0 to comment 7.
Please, give me further directions, if you need to check something to help fixing the problem.
Oleg
http://bugs.winehq.org/show_bug.cgi?id=20181
Roderick Colenbrander thunderbird2k@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |thunderbird2k@gmail.com
--- Comment #9 from Roderick Colenbrander thunderbird2k@gmail.com 2009-11-16 14:03:40 --- Are you sure that you completed the regression test? The GLX patch in question was added in Wine 1.1.27, so if the issue appeared in 1.1.25 then this can't be the bug. I fear that the issue is outside Wine though and in the Intel drivers.
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #10 from infobox.oleg@wo.cz 2009-11-16 14:15:58 --- I've found out, that issue is not in 1.1.25 nor 1.1.26, but in 1.1.27. Well, at least not this "X Error of failed request: GLXBadDrawable" from comment 0. I can't reproduce that one.
It IS possible, that issue is in some library. But offending change has very few lines: git diff 09436d69f181b8bae4afea28be73edb612a6f05b 390d3929f2b2a9641ecc8114cf94ae09fa851a15
I'm trying to revert offending change in wine-1.1.33 and see, whether the problem will persist. But it is just a random change to the code, I don't really understand :-)
Any other ideas?
Oleg
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #11 from Roderick Colenbrander thunderbird2k@gmail.com 2009-11-16 14:28:43 --- So it occurs on Wine 1.1.27 or later? Could you also check if the issue is around on the demo of Warcraft III? If it occurs there then I could easily report to an Intel developer if the bug is in their drivers which I fear it is.
Roderick
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #12 from Roderick Colenbrander thunderbird2k@gmail.com 2009-11-16 14:34:47 --- Further it would be useful to have a WINEDEBUG=+wgl log when the issue occurs. Also try to create the same log using WINEDEBUG=+synchronous,+wgl (in this case the game might even work).
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #13 from infobox.oleg@wo.cz 2009-11-16 15:24:39 --- Warcraft 3 Demo 1.01 does not work (GLXBadDrawable). Problem is in 1.1.24... It is not in 1.1.16. I don't have binary packages in between, so I'll do bisect again, for another error (from comment 0).
http://bugs.winehq.org/show_bug.cgi?id=20181
Roderick Colenbrander thunderbird2k@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- URL| |http://www.gamershell.com/d | |ownload_668.shtml
--- Comment #14 from Roderick Colenbrander thunderbird2k@gmail.com 2009-11-16 15:30:25 --- Add url for obtaining the demo.
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #15 from Alistair ajftonner@gmail.com 2009-11-16 17:05:15 --- (In reply to comment #12)
Further it would be useful to have a WINEDEBUG=+wgl log when the issue occurs. Also try to create the same log using WINEDEBUG=+synchronous,+wgl (in this case the game might even work).
Roderic will attempt to dump these for you .. one point I will make is that I've only just gotten this laptop running, so cannot provide historical *it worked then* data. Am inclined personally to agree that its likely intel driver -- would (along with the logs above) such like glxinfo provide some sort of further detail?
I will note that every once in a blue moon I can get WoW to start on the laptop, but the image is somewhat like looking through a kaleidescope with cracked mirrors.
would shooting for debug in the video drivers be of any assistance? {at wrk atm, have both my work vms up otherwise would consider dropping this all now, but the management types would be upset}
http://bugs.winehq.org/show_bug.cgi?id=20181
Dmitry Timoshkov dmitry@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Component|opengl |-unknown Version|1.1.26 |1.1.25
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #16 from infobox.oleg@wo.cz 2009-11-17 04:26:34 --- Created an attachment (id=24801) --> (http://bugs.winehq.org/attachment.cgi?id=24801) GlxBadDrawable error with WINEDEBUG=+wgl
Hi, the issue is somehow stateful. E.g. now I have wine-1.1.26. Sometimes it crashes on GlxBadDrawable, sometimes it doesn't. I can't find any pattern. The worst thing - if I compare log from successful run with the log from error, it is the same till error.
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #17 from infobox.oleg@wo.cz 2009-11-17 04:27:24 --- Created an attachment (id=24802) --> (http://bugs.winehq.org/attachment.cgi?id=24802) Successful run with WINEDEBUG=+wgl
compare with previous, it's the same until error occurs.
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #18 from Roderick Colenbrander thunderbird2k@gmail.com 2009-11-17 04:58:58 --- What I find suspicious is the following: trace:wgl:X11DRV_WineGL_InitOpenglInfo Direct rendering enabled: False
Do you have proper 32-bit intel libs around? You might be seeing libGL using GLX to perform GL rendering which is slower and offers less 3D features. It can also be that you are using Compiz or some other composition manager. If you are using Compiz or something similar could you disable desktop effects? Then direct rendering should be enabled. The issue might not even occur then.
http://bugs.winehq.org/show_bug.cgi?id=20181
Roderick Colenbrander thunderbird2k@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Component|-unknown |opengl
--- Comment #19 from Roderick Colenbrander thunderbird2k@gmail.com 2009-11-17 05:24:53 --- Lets keep it at opengl as it is either a bug in our code (unlikely though) or intel.
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #20 from infobox.oleg@wo.cz 2009-11-17 05:59:47 --- (In reply to comment #18) Well, I have 64bit system. glxinfo says Direct rendering: enabled. But no 32bit intel drivers. I'm not sure, whether it is possible to install 32bit drivers into 64bit X server. Should I try? Does it make sense?
I don't use compiz nor any other effects.
I think I have two issues here: - GlxBadDrawable - comment 0 - appears somewhere between 1.1.16 and 1.1.26 - bad parameters - comment 7 and comment 8 - reliably appears in 1.1.27 onward. - I've reverted the patch from comment 7 in source of wine 1.1.33. Game started ok (Icewind Dale), but crashed a few menus later.
Oleg
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #21 from Roderick Colenbrander thunderbird2k@gmail.com 2009-11-17 06:19:03 --- I don't have much experience with the plain dri drivers, so what I'm saying might not be fully correct. When you want to run 32-bit 3d apps you need a 32-bit libGL.so. In case of dri drivers Mesa can access the hardware directly using a 'dri' module on a 64-bit ubuntu these are in /usr/lib32/dri. The directory contains a dozen of these modules like 'i810_dri.so, i965_dri.so, radeon_dri.so, nouveau_dri.so' and so on. When such a library isn't around I think Mesa falls back to GLX. This means that it talks to the Xserver and then the 64-bit Xserver will pass the 3D commands to the driver likely using the 64-bit libGL.so.
I would check if you have a 32-bit i965_dri.so and if not try to install it.
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #22 from Alistair ajftonner@gmail.com 2009-11-17 08:57:17 --- (In reply to comment #21)
I don't have much experience with the plain dri drivers, so what I'm saying might not be fully correct.
<snippage>
I would check if you have a 32-bit i965_dri.so and if not try to install it.
On my system - Gentoo x86_64 :
wine --version wine-1.1.30 atonner@lrssbra0240 /usr/lib32/dri $ X -version
X.Org X Server 1.6.4 Release Date: 2009-9-27 X Protocol Version 11, Revision 0 Build Operating System: Linux 2.6.30-gentoo-r5 x86_64 Current Operating System: Linux lrssbra0240 2.6.31-gentoo-r2 #3 SMP Wed Oct 14 22:08:52 EDT 2009 x86_64 Build Date: 13 October 2009 10:53:26PM
Before reporting problems, check http://wiki.x.org to make sure that you have the latest version. atonner@lrssbra0240 /usr/lib32/dri $ glxinfo |grep -e render -e version direct rendering: Yes server glx version string: 1.2 client glx version string: 1.4 GLX version: 1.2 OpenGL renderer string: Mesa DRI Mobile Intel® GM45 Express Chipset GEM 20090712 2009Q2 RC3 OpenGL version string: 2.1 Mesa 7.5.2 OpenGL shading language version string: 1.20
I find the following rather interesting:
tonner@lrssbra0240 ~ $ ls -l /usr/lib/dri/i915* -rwxr-xr-x 1 root root 2636840 Nov 7 09:52 /usr/lib/dri/i915_dri.so atonner@lrssbra0240 ~ $ ls -l /usr/lib32/dri/i915* -rwxr-xr-x 1 root root 2255868 Nov 26 2007 /usr/lib32/dri/i915_dri.so -rwxr-xr-x 1 root root 2256188 Nov 26 2007 /usr/lib32/dri/i915tex_dri.so
The linked libraries are ***NOT*** consistent between 32bit and 64bit libraries (barring bitness, I'd presume that we need the same code sets in the ldd tree)
I find this odd since as far as I know these two libraries should have been built at the same time. I'll try rebuilding both xf86-video-intel and wine to see if I can clean up the inconsistencies before I dump the debugs.
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #23 from Roderick Colenbrander thunderbird2k@gmail.com 2009-11-17 09:10:56 --- For your gpu you should need the i965_dri driver because your GM45 is using that GPU.
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #24 from infobox.oleg@wo.cz 2009-11-17 10:15:47 --- (In reply to comment #22) I have Gentoo x86_64 as well.
tonner@lrssbra0240 ~ $ ls -l /usr/lib/dri/i915*
Here you have 64bit libs.
32bit libs are in /usr/lib32/dri/.
What is you output of equery b /usr/lib32/dri/i915_dri.so? I mean which package owns that file? I have only 64bit version of the driver.
Oleg
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #25 from infobox.oleg@wo.cz 2009-11-17 11:53:17 --- Ok, I've ran emerge -a1v "=app-emulation/emul-linux-x86-xlibs-10.1" and now I have 32bit i965_dri.so. But in Xorg.log I have (II) AIGLX: Loaded and initialized /usr/lib64/dri/i965_dri.so and I still have trace:wgl:X11DRV_WineGL_InitOpenglInfo Direct rendering enabled: False
How do I ensure, 32bit library is used for wine?
Oleg
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #26 from Roderick Colenbrander thunderbird2k@gmail.com 2009-11-17 12:43:51 --- You should install a 32-bit glxinfo and once that says direct rendering is enabled try Wine again. I have no experience with gentoo, so you have to ask on their forums or in their irc channel.
http://bugs.winehq.org/show_bug.cgi?id=20181
infobox.oleg@wo.cz changed:
What |Removed |Added ---------------------------------------------------------------------------- URL|http://www.gamershell.com/d |http://www.gamershell.com/d |ownload_668.shtml |ownload_668.shtml, | |http://bugs.gentoo.org/show | |_bug.cgi?id=293529
--- Comment #27 from infobox.oleg@wo.cz 2009-11-17 12:47:22 --- Ok, I've got there:
~/.wine/drive_c/Warcraft III Demo $ LD_PRELOAD=/usr/lib32/dri/i965_dri.so wine War3Demo.exe -opengl wine: symbol lookup error: /usr/lib32/dri/i965_dri.so: undefined symbol: _glapi_tls_Context
Now I'll crosspost a bug to Gentoo bugzilla: http://bugs.gentoo.org/show_bug.cgi?id=293529. I hope someone there will help to narrow down 32bit libs issues.
Oleg
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #28 from Roderick Colenbrander thunderbird2k@gmail.com 2009-11-17 13:03:13 --- The _dri.so files are sort of plugins for Mesa. You can't just preload them. I would just try to get a 32-bit glxinfo from somewhere and try that.
http://bugs.winehq.org/show_bug.cgi?id=20181
Austin English austinenglish@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Keywords| |download
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #29 from infobox.oleg@wo.cz 2009-11-18 00:13:59 --- I've installed 32bit chroot inside my adm64 OS. With wine 1.1.33 inside this chroot I have: - glxinfo DRI: true - DRI in wine: true - started War 3 Demo - it started, but it is extremely slow. I will examine it further
Oleg
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #30 from infobox.oleg@wo.cz 2009-11-18 00:29:45 --- After adding user to proper groups (video, audio, games) War III Demo works nicely. Where do we go from here? Under 32bit OS wine works OK, under 64bit OS not so much - with the same application.
Oleg
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #31 from Roderick Colenbrander thunderbird2k@gmail.com 2009-11-18 03:09:43 --- I would try to run outside a chroot. Just run the 32-bit glxinfo and also try to copy the i965_dri.so library to the 32-bit dri directory. If it works in a chroot it is quite similar to a mixed 32-bit / 64-bit system because it still uses the same 64-bit kernel module or Xserver.
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #32 from Alistair ajftonner@gmail.com 2009-11-18 08:44:38 --- Created an attachment (id=24821) --> (http://bugs.winehq.org/attachment.cgi?id=24821) run with WINEDEBUG=+wgl to GLXBADDRAWABLE exit
rebuilt x video driver, libdrm, mesa and emul-linux-x86-xlibs for consistency with current X and kernel.
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #33 from Alistair ajftonner@gmail.com 2009-11-18 08:54:11 --- (In reply to comment #23)
For your gpu you should need the i965_dri driver because your GM45 is using that GPU.
um assumption being the mother of all ....
You are of course correct sir, lsof of X shows it loading the i965_drv.
since the kernel device driver was i915 I made a silly assumption. Now however I seem to be missing any 32 bit intel drivers - I ended up with app-emulation/emul-linux-x86-xlibs-20081109 instead of 10.1 -- Hmm I attached the log anyway -- 24821.
I'll see what dependencies I need to satisfy 10.1 emulation libs.
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #34 from infobox.oleg@wo.cz 2009-11-18 09:19:54 --- (In reply to comment #33)
since the kernel device driver was i915 I made a silly assumption. Now however I seem to be missing any 32 bit intel drivers - I ended up with app-emulation/emul-linux-x86-xlibs-20081109 instead of 10.1 -- Hmm I attached the log anyway -- 24821.
I'll see what dependencies I need to satisfy 10.1 emulation libs.
Yep, gentoo removed those drivers in later versions of emul libs. I don't understand why. I still hope, that someone from gentoo will respond and advice us. I'll try an approach without chroot. If it succeeds, than we will need help from gentoo emul libs team IMO (the list of amd64 developers is here http://www.gentoo.org/proj/en/base/amd64/).
P;eg
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #35 from Alistair ajftonner@gmail.com 2009-11-18 19:55:28 --- (In reply to comment #34)
(In reply to comment #33)
since the kernel device driver was i915 I made a silly assumption. Now
<snippage>
us. I'll try an approach without chroot. If it succeeds, than we will need help from gentoo emul libs team IMO (the list of amd64 developers is here http://www.gentoo.org/proj/en/base/amd64/).
Oleg: if you've built the i965_drv.so in a 32 bit chroot, and you move it to your 64bit env, what happens?
I have the emul 10. installed, and I now more *rarely* get the GLXBadDrawable error (still getting it one in six tries) but still have the "Broken Kaleidescope" effect.
Loading the game and getting the kaleidescope effect, I can find nowhere loaded /usr/lib32/dri/i965_dri.so, rather it seems to load
/usr/lib32/opengl/xorg-x11/lib/libGL.so.1.2
and in my case:
equery b /usr/lib32/opengl/xorg-x11/lib/libGL.so.1.2 [ Searching for file(s) /usr/lib32/opengl/xorg-x11/lib/libGL.so.1.2 in *... ] app-emulation/emul-linux-x86-xlibs-10.1 (/usr/lib32/opengl/xorg-x11/lib/libGL.so.1.2)
I somehow believe that we're chasing the wrong library here -- we need to look at libGL.so --- I believe that it talks api to api to the driver rather than loading the 32bit driver and talking to it.
http://bugs.winehq.org/show_bug.cgi?id=20181
Jason D. Clinton me@jasonclinton.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |me@jasonclinton.com
--- Comment #36 from Jason D. Clinton me@jasonclinton.com 2009-11-18 22:21:41 --- I was having this same issue on Debian sid/experimental since at least the 2.27 time frame. I just installed Fedora 12 and the issue has vanished. F12 is using the very latest Xorg+Mesa+Intel driver, so it appears it was fixed upstream.
Now, Bnet doesn't work but that's another issue...
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #37 from infobox.oleg@wo.cz 2009-11-19 01:51:14 --- (In reply to comment #36)
I was having this same issue on Debian sid/experimental since at least the 2.27 time frame. I just installed Fedora 12 and the issue has vanished. F12 is using the very latest Xorg+Mesa+Intel driver, so it appears it was fixed upstream.
Jason,
Do you have 64bit system? If not, never mind :-)
Do you have 32bit libs we were discussing? Do you have 32bit glxinfo? Could you provide output of export LIBGL_DEBUG=verobse glxinfo <-- must be 32bit
or export LIBGL_DEBUG=verobse WINEDEBUG=+wgl wine 32bitgame
LIBGL_DEBUG displays libraries loaded by mesa. WINEDEBUG will should, if DRI is enabled for wine.
Thanks. Oleg
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #38 from infobox.oleg@wo.cz 2009-11-19 02:05:13 --- (In reply to comment #35)
if you've built the i965_drv.so in a 32 bit chroot, and you move it to your64bit env, what happens?
I did mount -o bind /chroot32/usr/lib /usr/lib32. I saw i965 started loading, but loading crashed, because dynamic library loaded found 64bit library for 32bit i965 -> something like "bad ELF" error.
I have the emul 10. installed, and I now more *rarely* get theGLXBadDrawable error (still getting it one in six tries) but still have the "Broken Kaleidescope" effect.
Loading the game and getting the kaleidescope effect, I can find nowhere loaded /usr/lib32/dri/i965_dri.so, rather it seems to load
/usr/lib32/opengl/xorg-x11/lib/libGL.so.1.2
See previous post please, what does LIBGL_DEBUG=verbose show? Try it with 64bit glxinfo, so you see what to expect.
and in my case:
equery b /usr/lib32/opengl/xorg-x11/lib/libGL.so.1.2 [ Searching for file(s) /usr/lib32/opengl/xorg-x11/lib/libGL.so.1.2 in *... ] app-emulation/emul-linux-x86-xlibs-10.1 (/usr/lib32/opengl/xorg-x11/lib/libGL.so.1.2)
I somehow believe that we're chasing the wrong library here -- we need to look at libGL.so --- I believe that it talks api to api to the driver rather than loading the 32bit driver and talking to it.
I think we are chasing the right library. My output for LIBGL_DEBUG in 32chroot shows i965 loading. But when I start 32bit glxinfo in 64bit environment, I don't see any library loading. You may still be right, because libGL.so may be used to load i965. But if libGL doesn't find i965, libGL.so is still loaded.
Oleg
http://bugs.winehq.org/show_bug.cgi?id=20181
Dmitry Timoshkov dmitry@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Keywords|regression | Status|UNCONFIRMED |RESOLVED Resolution| |INVALID
--- Comment #39 from Dmitry Timoshkov dmitry@codeweavers.com 2009-11-19 06:00:51 --- This appears to be not a Wine bug, and not a regression. Please consult your distro support for further help.
http://bugs.winehq.org/show_bug.cgi?id=20181
Dmitry Timoshkov dmitry@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #40 from Dmitry Timoshkov dmitry@codeweavers.com 2009-11-19 06:01:06 --- Closing invalid.
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #41 from infobox.oleg@wo.cz 2009-11-19 06:45:00 --- (In reply to comment #40)
Closing invalid.
Ok, it is probably issue of the distribution. Still, Wine could check for absense of DRI support and display some warning about it (trace:wgl:X11DRV_WineGL_InitOpenglInfo Direct rendering enabled: False is not descriptive enough).
Alistair, everyone else concerned,
I propose we move this discussion to Gentoo bugzilla: http://bugs.gentoo.org/show_bug.cgi?id=293529.
Jason, please,
could you still respond to comment 37. It could help us.
Roderick,
I'll do my best to report back if/when we'll solve this in 64bit OS outside chroot.
Thanks everyone. Oleg
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #42 from Roderick Colenbrander thunderbird2k@gmail.com 2009-11-19 07:34:40 --- (In reply to comment #41)
(In reply to comment #40)
Closing invalid.
Ok, it is probably issue of the distribution. Still, Wine could check for absense of DRI support and display some warning about it (trace:wgl:X11DRV_WineGL_InitOpenglInfo Direct rendering enabled: False is not descriptive enough).
I have thought about turning it into a warning in the past but there are a lot of valid cases where you use indirect rendering. Indirect rendering doesn't have to mean 'slow'. For instance when you use a composition manager like compiz and don't run apps fullscreen you are using indirect rendering. You also use indirect rendering when you use remote X.
Roderick
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #43 from infobox.oleg@wo.cz 2009-11-19 07:42:50 --- (In reply to comment #42)
(In reply to comment #41)
(In reply to comment #40)
Closing invalid.
Ok, it is probably issue of the distribution. Still, Wine could check for absense of DRI support and display some warning about it (trace:wgl:X11DRV_WineGL_InitOpenglInfo Direct rendering enabled: False is not descriptive enough).
I have thought about turning it into a warning in the past but there are a lot of valid cases where you use indirect rendering. Indirect rendering doesn't have to mean 'slow'. For instance when you use a composition manager like compiz and don't run apps fullscreen you are using indirect rendering. You also use indirect rendering when you use remote X.
Roderick
You are right. Actually when I run Warcraft with -opengl option on my 64bit without DRI, it runs OK. The problem is, when application actually crashes due to absence of DRI and we don't know what it's the issue.
From this bug I see, that for wine 1.1.27+ absence of DRI becomes issue on
64bit. It would be interesting to purposefully break DRI on 32bit platform and see, if wine crashes then.
Oleg
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #44 from Jason D. Clinton me@jasonclinton.com 2009-11-19 10:41:03 --- Created an attachment (id=24833) --> (http://bugs.winehq.org/attachment.cgi?id=24833) Everything working fine on Fedora 12
(In reply to comment #37)
Do you have 64bit system? If not, never mind :-)
Yes.
export LIBGL_DEBUG=verobse WINEDEBUG=+wgl wine 32bitgame
Attached.
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #45 from infobox.oleg@wo.cz 2009-11-19 10:43:45 --- (In reply to comment #44)
Created an attachment (id=24833)
--> (http://bugs.winehq.org/attachment.cgi?id=24833) [details]
Everything working fine on Fedora 12
(In reply to comment #37)
Do you have 64bit system? If not, never mind :-)
Yes.
export LIBGL_DEBUG=verobse WINEDEBUG=+wgl wine 32bitgame
Attached.
Interesting. Could you please run "file /usr/lib/dri/i965_dri.so"? I'm just checking, that this file is a 32bit driver.
Oleg
http://bugs.winehq.org/show_bug.cgi?id=20181
--- Comment #46 from Jason D. Clinton me@jasonclinton.com 2009-11-19 10:45:08 --- (In reply to comment #45)
Interesting. Could you please run "file /usr/lib/dri/i965_dri.so"? I'm just checking, that this file is a 32bit driver.
Oleg
It is. 64-bit libs are (stupidly) installed to /usr/lib64 on Fedora/Red Hat.
-
wine-bugs＠winehq.org