Bug 21249

Summary: [GM965 GEM KMS] Spurious GPU wedge while running compiz
Product: xorg Reporter: Ben Gamari <bgamari>
Component: Driver/intelAssignee: Eric Anholt <eric>
Status: RESOLVED DUPLICATE QA Contact: Xorg Project Team <xorg-team>
Severity: critical    
Priority: medium CC: mmar
Version: unspecified   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
intel_gpu_dump output none

Description Ben Gamari 2009-04-17 08:27:26 UTC
Created attachment 24899 [details]
intel_gpu_dump output

While I was switching windows while running compiz (not sure if compiz is a cause) the GPU wedged. Luckily, I had a remote console and got a backtrace and gpu dump. The backtrace is as follows,

0x00000033b40d84d7 in ioctl () from /lib64/libc.so.6
(gdb) bt
#0  0x00000033b40d84d7 in ioctl () from /lib64/libc.so.6
#1  0x0000003a82a05a15 in drm_intel_gem_bo_map_gtt (bo=0x3f2d860)
    at intel_bufmgr_gem.c:721
#2  0x00007f9fe34efb85 in i830_uxa_prepare_access (pixmap=0x4a91d50, 
    access=UXA_ACCESS_RW) at i830_exa.c:865
#3  0x00007f9fe350e1d4 in uxa_check_poly_fill_rect (pDrawable=0x4a91d50, 
    pGC=0x37bb0f0, nrect=1, prect=0x7fffec204980) at uxa-unaccel.c:255
#4  0x00007f9fe350b03f in uxa_create_alpha_picture (pScreen=0xa41d60, 
    pDst=<value optimized out>, pPictFormat=0xa42b48, width=10, height=6)
    at uxa-render.c:841
#5  0x00007f9fe350b690 in uxa_trapezoids (op=<value optimized out>, 
    pSrc=0x7d78c10, pDst=0x71f7620, maskFormat=<value optimized out>, 
    xSrc=<value optimized out>, ySrc=<value optimized out>, ntrap=8, 
    traps=0x7b204a8) at uxa-render.c:909
#6  0x00000000004cb1db in ProcRenderTrapezoids (client=0x39f6e80)
    at render.c:781
#7  0x000000000042a224 in Dispatch () at dispatch.c:428
#8  0x00000000004243a5 in main (argc=8, argv=0x7ce0c8, 
    envp=<value optimized out>) at main.c:396
Comment 1 Marcel Martin 2009-04-23 13:21:50 UTC
I'm getting similar freezes every few hours on Intel 855GM, KDE 4.2, X server 1.6, libdrm 2.4.6, intel driver 2.7.0, Linux 2.6.29. I'm not running compiz.

Since comment 12 in bug 19911 mentions a similar backtrace, could someone tell me whether that bug is related to this one?

Would be great to see this fixed as currently I need to reboot frequently to continue working.

#0  0xffffe424 in __kernel_vsyscall ()
#1  0xb7ad4459 in ioctl () from /lib/libc.so.6
#2  0xb7986208 in drm_intel_gem_bo_map_gtt (bo=0xa452800) at intel_bufmgr_gem.c:721
#3  0xb7920c3e in i830_uxa_prepare_access (pixmap=0xa644b60, access=UXA_ACCESS_RW) at i830_exa.c:865
#4  0xb79399f4 in uxa_prepare_access (pDrawable=0xa644b60, access=UXA_ACCESS_RW) at uxa.c:158
#5  0xb7940233 in uxa_check_composite (op=12 '\f', pSrc=0x995fad0, pMask=0x0, pDst=0xa8136f8, xSrc=16, ySrc=48, xMask=0,
    yMask=0, xDst=0, yDst=0, width=8, height=12) at uxa-unaccel.c:376
#6  0xb793ea8d in uxa_composite_rects (op=12 '\f', pSrc=0x995fad0, pDst=0xa8136f8, nrect=1, rects=0xbfb283a8)
    at uxa-render.c:418
#7  0xb793d48c in uxa_glyphs (op=3 '\003', pSrc=0xa4a3f68, pDst=0x98f89e8, maskFormat=0x98cec60, xSrc=0, ySrc=0, nlist=0,
    list=0xbfb29528, glyphs=0xbfb2912c) at uxa-glyphs.c:553
#8  0x0817df70 in damageGlyphs (op=0 '\0', pSrc=0xa4a3f68, pDst=0x98f89e8, maskFormat=0x0, xSrc=0, ySrc=0, nlist=1,
    list=0xbfb29528, glyphs=0xbfb29128) at damage.c:721
#9  0x0816c2c4 in CompositeGlyphs (op=0 '\0', pSrc=0xa4a3f68, pDst=0x98f89e8, maskFormat=0x0, xSrc=0, ySrc=0, nlist=1,
    lists=0xbfb29528, glyphs=0xbfb29128) at glyph.c:632
#10 0x0817961e in ProcRenderCompositeGlyphs (client=0xa468390) at render.c:1459
#11 0x081728c5 in ProcRenderDispatch (client=0x400c645f) at render.c:2086
#12 0x0808b53f in Dispatch () at dispatch.c:437
#13 0x0806ff8d in main (argc=10, argv=0xbfb29984, envp=Cannot access memory at address 0x400c6467
) at main.c:397
Comment 2 Jesse Barnes 2009-05-11 11:21:50 UTC
Adjusting severity: crashes & hangs should be marked critical.
Comment 3 Eric Anholt 2009-05-12 15:58:17 UTC
Same looking dump as the other one that I've put out a proposed patch for.  Let's consolidate things there.


*** This bug has been marked as a duplicate of bug 21488 ***

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.