Bug 94611

Summary: Steam client hangs with Radeon DDX + DRI3
Product: xorg Reporter: Maxim Sheviakov <mrader3940>
Component: Driver/RadeonAssignee: xf86-video-ati maintainers <xorg-driver-ati>
Status: RESOLVED DUPLICATE QA Contact: Xorg Project Team <xorg-team>
Severity: normal    
Priority: medium CC: mrader3940
Version: unspecified   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features:

Description Maxim Sheviakov 2016-03-18 12:21:49 UTC
I've managed to find out that it's not Steam's issue: with Radeon DDX with DRI3 enabled there's a *random* hang in Steam client, leaving the only option - "kill -9". You can find the original issue on GitHub with the video attached (showing the process of making the client "dead") here: https://github.com/ValveSoftware/steam-for-linux/issues/4357
Right now I'm on Modesetting driver and everything seems to work fine. Either it's the issue of a recent 1.18.4 X.org update, either it's a regression in DRI3 code. I'm thinking of the former, because everything was fine a while ago.
Comment 1 Maxim Sheviakov 2016-03-18 12:28:31 UTC
UPD: client also seems to work with these options:
>Section "Device"
>	Identifier "R7 370"
>	Driver "radeon"
>	Option "ColorTiling" "on"
>	Option "ColorTiling2D" "on"
>	# Option "DRI" "3"
>	# Option "AccelMethod" "glamor"
>EndSection
Thus with DRI2 enabled, not DRI3.
Comment 2 Laurent carlier 2016-03-18 12:37:21 UTC
Could this be related ? https://bugs.freedesktop.org/show_bug.cgi?id=94596
Comment 3 Maxim Sheviakov 2016-03-18 14:40:07 UTC
(In reply to Laurent carlier from comment #2)
> Could this be related ? https://bugs.freedesktop.org/show_bug.cgi?id=94596

Maybe. However, I haven't noticed any memleaks or something, and heavy games run fine. It's likely a regression somewhere.
Also, NFS Most Wanted is no more launchable via latest Wine and is arguing on GL features missing; it's also beeb happening since the last X.org update. Once I'm home I'll provide some logs.
Comment 4 Maxim Sheviakov 2016-03-18 15:10:16 UTC
This is the log from Wine, as promised:

------------

wine .wine/drive_c/Games/Need\ for\ Speed\ Most\ Wanted/speed.exe 
fixme:winediag:start_process Wine Staging 1.9.5 is a testing version containing experimental patches.
fixme:winediag:start_process Please mention your exact version when filing bug reports on winehq.org.
fixme:module:load_dll Loader redirect from L"wined3d.dll" to L"wined3d-csmt.dll"
fixme:winediag:wined3d_dll_init Experimental wined3d CSMT feature is currently enabled.
err:winediag:init_driver_info Could not find GPU info for 1002:6811.
Mesa: User error: GL_INVALID_ENUM in glRenderbufferStorage(internalFormat=GL_SLUMINANCE8)
Mesa: User error: GL_INVALID_ENUM in glRenderbufferStorage(internalFormat=GL_SLUMINANCE8_ALPHA8)
Mesa: User error: GL_INVALID_OPERATION in glTexImage3D(bad target for depth texture)
Mesa: User error: GL_INVALID_OPERATION in glTexImage3D(bad target for depth texture)
Mesa: User error: GL_INVALID_OPERATION in glTexImage3D(bad target for depth texture)
Mesa: User error: GL_INVALID_OPERATION in glTexImage3D(bad target for depth texture)
Mesa: User error: GL_INVALID_OPERATION in glTexImage3D(bad target for depth texture)
Mesa: User error: GL_INVALID_OPERATION in glTexImage3D(bad target for depth texture)
Mesa: User error: GL_INVALID_OPERATION in glTexImage3D(bad target for depth texture)
Mesa: User error: GL_INVALID_OPERATION in glTexImage3D(bad target for depth texture)
Mesa: User error: GL_INVALID_OPERATION in glTexImage3D(bad target for depth texture)
Mesa: User error: GL_INVALID_OPERATION in glTexImage3D(bad target for depth texture)
fixme:d3d:debug_d3dformat Unrecognized 0x36314644 (as fourcc: DF16) WINED3DFORMAT!
fixme:d3d:wined3d_get_format Can't find format unrecognized (0x36314644) in the format lookup table
fixme:dsound:DSOUND_RecalcFormat Conversion from 4 to 2 channels is not implemented, falling back to stereo
fixme:dsound:DSOUND_RecalcFormat Conversion from 6 to 2 channels is not implemented, falling back to stereo
wine: Unhandled page fault on read access to 0x00e30000 at address 0x65aa27 (thread 0009), starting debugger...
Unhandled exception: page fault on read access to 0x00e30000 in 32-bit code (0x0065aa27).
Comment 5 Ernst Sjöstrand 2016-03-18 15:26:25 UTC
Which version of libxcb do you have?

Actually what distribution do you use?
Comment 6 Maxim Sheviakov 2016-03-18 16:05:59 UTC
(In reply to Ernst Sjöstrand from comment #5)
> Which version of libxcb do you have?
> 
> Actually what distribution do you use?

My distro is Gentoo ~amd64, libxcb is 1.11.1.
Comment 7 Laurent carlier 2016-03-18 16:27:18 UTC
(In reply to Maxim Sheviakov from comment #3)
> (In reply to Laurent carlier from comment #2)
> > Could this be related ? https://bugs.freedesktop.org/show_bug.cgi?id=94596
> 
> Maybe. However, I haven't noticed any memleaks or something, and heavy games
> run fine. It's likely a regression somewhere.
> Also, NFS Most Wanted is no more launchable via latest Wine and is arguing
> on GL features missing; it's also beeb happening since the last X.org
> update. Once I'm home I'll provide some logs.

I don't have memleak with my radeon card, only high cpu usage on one core with Xorg leading to the freeze of the related application.
Comment 8 Maxim Sheviakov 2016-03-24 08:44:01 UTC
(In reply to Maxim Sheviakov from comment #0)
> issue of a recent 1.18.4 X.org update
Sorry, 1.18.2, of course - my bad typo. Anyway, I'll be doing some testing with the workaround patch published here: https://bugs.freedesktop.org/show_bug.cgi?id=94596
If it fixes the issue, I'll let everyone know, and it's likely that the patch will be needed upstream.
Comment 9 Maxim Sheviakov 2016-03-24 08:50:18 UTC
(In reply to Maxim Sheviakov from comment #8)
> (In reply to Maxim Sheviakov from comment #0)
> > issue of a recent 1.18.4 X.org update
> Sorry, 1.18.2, of course - my bad typo. Anyway, I'll be doing some testing
> with the workaround patch published here:
> https://bugs.freedesktop.org/show_bug.cgi?id=94596
> If it fixes the issue, I'll let everyone know, and it's likely that the
> patch will be needed upstream.

Yuppie, it fixes the problem. Devs, please send it upstream.
Comment 10 Maxim Sheviakov 2016-03-24 08:51:54 UTC
See
> https://bugs.freedesktop.org/show_bug.cgi?id=94596
Comment 11 Michel Dänzer 2016-03-24 08:53:25 UTC

*** This bug has been marked as a duplicate of bug 94596 ***

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.