Bug 49736

Summary: [NV43] piglit quick.driver tests kill X server
Product: Mesa Reporter: Andrew Randrianasulu <randrik>
Component: Drivers/DRI/nouveauAssignee: Nouveau Project <nouveau>
Status: RESOLVED WONTFIX QA Contact:
Severity: normal    
Priority: medium    
Version: git   
Hardware: x86 (IA32)   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Attachments: piglit stdout and stderr log (bz2 compressed)
X log
piglit html results summary, 7z compressed
nv43 dmesg
X log

Description Andrew Randrianasulu 2012-05-10 07:35:47 UTC 
Using this  configuration

software:

ddx:
commit 581564466c47ab0435ecca3cf4d82c03e1b30259
Author: Francisco Jerez <currojerez@riseup.net>
Date:   Tue May 8 14:02:44 2012 +0200

    dri2: Don't try to page-flip pixmaps (fdo bug 49351).

libdrm:
commit 5288729823ee1c243023758c35fbe8e3a70ccf9c
Author: Marcin Slusarz <marcin.slusarz@gmail.com>
Date:   Tue May 1 23:24:45 2012 +0200

    nouveau: fix channel closing

kernel:
nouveau tree
commit 6de446e248e4e0bbb040abc1e76c9876750aac78
Author: Ben Skeggs <bskeggs@redhat.com>
Date:   Thu May 10 19:03:11 2012 +1000

    drm/nv04-nv40: fix regression from a previous commit (SQUASH)

mesa:
commit 788fd04dacb9eb1e32010050c57cd2f49779311b
Author: Tom Stellard <thomas.stellard@amd.com>
Date:   Wed May 9 11:50:02 2012 -0400

    radeon/llvm: Remove AMDGPUConstants.pm

_plus_ patches  from http://lists.freedesktop.org/archives/nouveau/2012-May/010322.html

nvfx_Fix_generic_passing_to_fragment_program_in_NV34.diff
nvfx_SSG_only_requires_one_source_register.diff
nvfx_Silence_pipe_cap_warnings.diff


piglit - 
commit bbed4bd9e7b614cf8b3e016d1f7116c323b23332
Author: Kenneth Graunke <kenneth@whitecape.org>
Date:   Fri May 4 17:07:03 2012 -0700

    Add a clipping test that draws an extremely large triangle.

X server  1.10.6

hw:
01:00.0 VGA compatible controller: NVIDIA Corporation NV43 [GeForce 6600 GT] (rev a2) (prog-if 00 [VGA controller])
        Subsystem: XFX Pine Group Inc. GeForce 6600 GT AGP
        Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B- DisINTx-
        Status: Cap+ 66MHz+ UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
        Latency: 32 (1250ns min, 250ns max)
        Interrupt: pin A routed to IRQ 16
        Region 0: Memory at fd000000 (32-bit, non-prefetchable) [size=16M]
        Region 1: Memory at e0000000 (32-bit, prefetchable) [size=256M]
        Region 2: Memory at fc000000 (32-bit, non-prefetchable) [size=16M]
        Expansion ROM at fe8e0000 [disabled] [size=128K]
        Capabilities: [60] Power Management version 2
                Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
                Status: D0 NoSoftRst- PME-Enable- DSel=0 DScale=0 PME-
        Capabilities: [44] AGP version 3.0
                Status: RQ=32 Iso- ArqSz=0 Cal=0 SBA+ ITACoh- GART64- HTrans- 64bit- FW+ AGP3- Rate=x1,x2,x4
                Command: RQ=32 ArqSz=0 Cal=0 SBA+ AGP+ GART64- 64bit- FW+ Rate=x4
        Kernel driver in use: nouveau
        Kernel modules: nouveau


I  run into X crash during quck.driver testing.
Comment 1 Andrew Randrianasulu 2012-05-10 07:37:36 UTC 
Created attachment 61372 [details]
piglit stdout and  stderr log (bz2  compressed)
Comment 2 Andrew Randrianasulu 2012-05-10 07:39:13 UTC 
Created attachment 61373 [details]
X log
Comment 3 Andrew Randrianasulu 2012-05-10 07:40:56 UTC 
Created attachment 61374 [details]
piglit html results summary, 7z  compressed
Comment 4 Emil Velikov 2012-11-01 17:38:29 UTC 
The bug is in the obsolete, unsupported and now missing gallium/nvfx. Consider updating Mesa 9.0 or greater that includes gallium/nv30

Marking as Resolved/Wontfix
Comment 5 Emil Velikov 2012-11-01 17:47:03 UTC 
You are using nv30, by bad
Comment 6 Ilia Mirkin 2013-09-06 22:47:42 UTC 
It'd be really interesting to see what the Xorg backtrace looks like with debug libraries (so that we can see the symbol names). BTW, does this still happen with fresh software (xf86-video-nouveau 1.0.9, mesa 9.2, libdrm 2.4.46)?
Comment 7 Andrew Randrianasulu 2013-09-19 10:58:05 UTC 
(In reply to comment #6)
> It'd be really interesting to see what the Xorg backtrace looks like with
> debug libraries (so that we can see the symbol names). BTW, does this still
> happen with fresh software (xf86-video-nouveau 1.0.9, mesa 9.2, libdrm
> 2.4.46)?

using kernel 3.12.0-rc1-00020-ga1d2bdc and mesa git master (OpenGL version string: 2.1 Mesa 9.3.0-devel (git-e150c0d)) it still crash, but now I  think it  just  hang GPU and  X  crashes  due  to  this ....
Comment 8 Andrew Randrianasulu 2013-09-19 11:01:44 UTC 
Created attachment 86134 [details]
nv43 dmesg
Comment 9 Andrew Randrianasulu 2013-09-19 11:03:00 UTC 
Created attachment 86135 [details]
X log
Comment 10 Ilia Mirkin 2013-09-19 14:34:19 UTC 
There's still a segfault, but the timing corresponds nicely to the GPU lockup. The segfault happens in libexa, without direct nouveau involvement though. You might try to also upgrade Xorg and see if that helps, although a bit unlikely.

BTW, you should grab my patch at http://lists.freedesktop.org/archives/mesa-dev/2013-September/044399.html (and the 2/2). Although it won't apply without a bit of massaging (nv30 dir got moved, should be able to just hand-edit the patch). This disables ARB_framebuffer_object on nv30. There's another patch I need to write that will disallow incompatible color/zeta settings too.
Comment 11 Ilia Mirkin 2015-10-22 03:05:46 UTC 
I believe that the nv4x piglit situation has improved considerably in the intervening years. If the GPU locks up, then an Xorg crash is expected (albeit unfortunate).

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.