|
Description
martin
2009-02-24 03:43:35 UTC
I just installed the 29-rc6 kernel on the G45 machine that the bug repros using that config as well. I just found out that this bug repros using EXA as well. I didn't notice first because pretty often the xserver doesn't SEGV on the first time the test case is run but if you run it 2-3 times you are almost certain so see the crash using EXA+G45 as well (this goes both for the .29-rc6 and the ubuntu stock kernels). Gordon, because this bug also repros on G45+EXA (perhaps you can try to verify this), I don't think it should be lumped up with the other glean failures. What if the glean failures get postponed again? Ubuntu really needs stable EXA for the April jaunty release. When I repro the bug in EXA, the stack is different of course:
Program received signal SIGSEGV, Segmentation fault.
0x00007ff647512d51 in ?? () from /usr/lib/dri/i965_dri.so
(gdb) bt
#0 0x00007ff647512d51 in ?? () from /usr/lib/dri/i965_dri.so
#1 0x00007ff658ebc936 in __glXForceCurrent (cl=0x3155710, tag=<value optimized out>, error=0x7fff63eab92c) at ../../glx/glxext.c:394
#2 0x00007ff658ebaaae in DoMakeCurrent (cl=0x3155710, drawId=<value optimized out>, readId=73400327, contextId=<value optimized out>, tag=1)
at ../../glx/glxcmds.c:605
#3 0x00007ff658ebcea2 in __glXDispatch (client=0x37b2540) at ../../glx/glxext.c:523
#4 0x000000000044e1d4 in Dispatch () at ../../dix/dispatch.c:437
#5 0x0000000000433c5d in main (argc=10, argv=0x7fff63eabb48, envp=<value optimized out>) at ../../dix/main.c:397
(gdb) bt full
#0 0x00007ff647512d51 in ?? () from /usr/lib/dri/i965_dri.so
No symbol table info available.
#1 0x00007ff658ebc936 in __glXForceCurrent (cl=0x3155710, tag=<value optimized out>, error=0x7fff63eab92c) at ../../glx/glxext.c:394
cx = (__GLXcontext *) 0x708b780
#2 0x00007ff658ebaaae in DoMakeCurrent (cl=0x3155710, drawId=<value optimized out>, readId=73400327, contextId=<value optimized out>, tag=1)
at ../../glx/glxcmds.c:605
client = (ClientPtr) 0x37b2540
reply = {type = 96 '`', unused = 219 '�', sequenceNumber = 125, length = 0, contextTag = 0, pad2 = 0, pad3 = 8248216, pad4 = 0,
pad5 = 1487362742, pad6 = 32758}
glxc = (__GLXcontext *) 0x64a86d0
prevglxc = (__GLXcontext *) 0x708b780
drawPriv = (__GLXdrawable *) 0x21cc350
readPriv = (__GLXdrawable *) 0x21cc350
error = <value optimized out>
mask = <value optimized out>
__PRETTY_FUNCTION__ = "DoMakeCurrent"
#3 0x00007ff658ebcea2 in __glXDispatch (client=0x37b2540) at ../../glx/glxext.c:523
stuff = (xGLXSingleReq *) 0x4dc63e8
opcode = <value optimized out>
cl = (__GLXclientState *) 0x3155710
retval = 1
#4 0x000000000044e1d4 in Dispatch () at ../../dix/dispatch.c:437
result = <value optimized out>
client = (ClientPtr) 0x37b2540
nready = 0
start_tick = 7200
#5 0x0000000000433c5d in main (argc=10, argv=0x7fff63eabb48, envp=<value optimized out>) at ../../dix/main.c:397
i = 1
alwaysCheckForInput = {0, 1}
(gdb)
Here is the EXA stack with mesa dbg symbols installed properly:
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fd7b03e16f0 (LWP 7707)]
0x00007fd79ba9f000 in ?? () from /usr/lib/dri/i965_dri.so
(gdb) bt
#0 0x00007fd79ba9f000 in ?? () from /usr/lib/dri/i965_dri.so
#1 0x00007fd79bb3695d in _mesa_delete_texture_image () from /usr/lib/dri/i965_dri.so
#2 0x00007fd79bb3d1d1 in _mesa_delete_texture_object () from /usr/lib/dri/i965_dri.so
#3 0x00007fd79bad7dfc in ?? () from /usr/lib/dri/i965_dri.so
#4 0x00007fd79bad8bbd in _mesa_free_context_data () from /usr/lib/dri/i965_dri.so
#5 0x00007fd79ba78a40 in ?? () from /usr/lib/dri/i965_dri.so
#6 0x00007fd7ad42c8b9 in __glXDRIcontextDestroy (baseContext=0x4652360) at ../../glx/glxdri.c:290
#7 0x00007fd7ad422bac in __glXFreeContext (cx=0x4652360) at ../../glx/glxext.c:160
#8 0x00007fd7ad422bf3 in ContextGone (cx=0x61bb090, id=4294967294) at ../../glx/glxext.c:109
#9 0x0000000000435fcb in FreeClientResources (client=0x4650ea0) at ../../dix/resource.c:807
#10 0x0000000000448840 in CloseDownClient (client=0x4650ea0) at ../../dix/dispatch.c:3433
#11 0x000000000044e068 in Dispatch () at ../../dix/dispatch.c:448
#12 0x0000000000433c5d in main (argc=10, argv=0x7fffb840f0a8, envp=<value optimized out>) at ../../dix/main.c:397
(gdb) bt full
#0 0x00007fd79ba9f000 in ?? () from /usr/lib/dri/i965_dri.so
No symbol table info available.
#1 0x00007fd79bb3695d in _mesa_delete_texture_image () from /usr/lib/dri/i965_dri.so
No symbol table info available.
#2 0x00007fd79bb3d1d1 in _mesa_delete_texture_object () from /usr/lib/dri/i965_dri.so
No symbol table info available.
#3 0x00007fd79bad7dfc in ?? () from /usr/lib/dri/i965_dri.so
No symbol table info available.
#4 0x00007fd79bad8bbd in _mesa_free_context_data () from /usr/lib/dri/i965_dri.so
No symbol table info available.
#5 0x00007fd79ba78a40 in ?? () from /usr/lib/dri/i965_dri.so
No symbol table info available.
#6 0x00007fd7ad42c8b9 in __glXDRIcontextDestroy (baseContext=0x4652360) at ../../glx/glxdri.c:290
No locals.
#7 0x00007fd7ad422bac in __glXFreeContext (cx=0x4652360) at ../../glx/glxext.c:160
No locals.
#8 0x00007fd7ad422bf3 in ContextGone (cx=0x61bb090, id=4294967294) at ../../glx/glxext.c:109
No locals.
#9 0x0000000000435fcb in FreeClientResources (client=0x4650ea0) at ../../dix/resource.c:807
rtype = 56
resources = <value optimized out>
this = (ResourcePtr) 0x4ab8160
j = 13
#10 0x0000000000448840 in CloseDownClient (client=0x4650ea0) at ../../dix/dispatch.c:3433
really_close_down = 1
#11 0x000000000044e068 in Dispatch () at ../../dix/dispatch.c:448
result = 8203840
client = (ClientPtr) 0x4650ea0
nready = 0
start_tick = 1640
#12 0x0000000000433c5d in main (argc=10, argv=0x7fffb840f0a8, envp=<value optimized out>) at ../../dix/main.c:397
i = 1
alwaysCheckForInput = {0, 1}
Jian, can you reproduce this on G45 with EXA and UXA? I think we'd just directly use the case from glean repository instead of from piglit. We can reproduce it on G45-64bit but only with UXA. I can run it more than 10 times consecutively with EXA. martin, Could you try the glean case from the original glean CVS instead of from piglit? Yes, we use piglit. But for glean, we prefer to use the original repository. We can reproduce it on G45-64bit but only with UXA. I can run it more than 10 times consecutively with EXA. Using this version: cvs -z3 -d:pserver:anonymous@glean.cvs.sourceforge.net:/cvsroot/glean co -P glean I put these lines into a .sh script: GLEAN_ROOT=`pwd` ; export GLEAN_ROOT cd src make install And then I just ran GLEAN_ROOT/bin/glean -r makeCurrent This crashes for me using EXA. However, with piglit the crash comes really fast but with pure glean like above it opens and closes a window lots of times, prints some status messages along the way and then it crashes. For this test I used an upstream 2.6.29-rc8 kernel compiled with ubuntu's kernel config, and then userspace bits from the xorg-edgers PPA: xserver-xorg-core 2:1.6.0-0ubuntu3 xserver-xorg-core-dbg 2:1.6.0-0ubuntu3 libdrm-intel1 2.4.5+git20090314.82eac806-0ubuntu0tormod2 libdrm-intel1-dbg 2.4.5+git20090314.82eac806-0ubuntu0tormod2 xserver-xorg-video-intel 2:2.6.99.1+git20090319.bedc894a-0ubuntu0tormod xserver-xorg-video-intel-dbg 2:2.6.99.1+git20090319.bedc894a-0ubuntu0tormod libgl1-mesa-dev 7.3+git20090312+mesa-7-4-branch.a6f7e909-0ubuntu0tormod libgl1-mesa-dri 7.3+git20090312+mesa-7-4-branch.a6f7e909-0ubuntu0tormod libgl1-mesa-dri-dbg 7.3+git20090312+mesa-7-4-branch.a6f7e909-0ubuntu0tormod libgl1-mesa-glx 7.3+git20090312+mesa-7-4-branch.a6f7e909-0ubuntu0tormod libglu1-mesa 7.3+git20090312+mesa-7-4-branch.a6f7e909-0ubuntu0tormod libglu1-mesa-dev 7.3+git20090312+mesa-7-4-branch.a6f7e909-0ubuntu0tormod mesa-common-dev 7.3+git20090312+mesa-7-4-branch.a6f7e909-0ubuntu0tormod mesa-utils 7.3+git20090312+mesa-7-4-branch.a6f7e909-0ubuntu0tormod xlibmesa-gl-dev 1:7.4~5ubuntu16 libdrm-dev 2.4.5+git20090314.82eac806-0ubuntu0tormod2 libdrm-intel1 2.4.5+git20090314.82eac806-0ubuntu0tormod2 libdrm-intel1-dbg 2.4.5+git20090314.82eac806-0ubuntu0tormod2 libdrm-nouveau1 2.4.5+git20090314.82eac806-0ubuntu0tormod2 libdrm2 2.4.5+git20090314.82eac806-0ubuntu0tormod2 libdrm2-dbg 2.4.5+git20090314.82eac806-0ubuntu0tormod2 compiz 1:0.8.2-0ubuntu5 compiz-core 1:0.8.2-0ubuntu5 compiz-dev 1:0.8.2-0ubuntu5 compiz-fusion-plugins-extra 0.8.2-0ubuntu1 compiz-fusion-plugins-main 0.8.2-0ubuntu1 compiz-gnome 1:0.8.2-0ubuntu5 compiz-kde 1:0.8.2-0ubuntu5 compiz-plugins 1:0.8.2-0ubuntu5 compiz-wrapper 1:0.8.2-0ubuntu5 compizconfig-backend-gconf 0.8.2-0ubuntu1 compizconfig-backend-kconfig 0.8.2-0ubuntu1 compizconfig-settings-manager 0.8.2-0ubuntu1 libcompizconfig0 0.8.2-0ubuntu1 The last thing CVS glean prints before crashing is:
basicPerf: PASS rgba8, db, z24, s8, accrgba16, win+pmap, slow, id 168
AVERAGE = 0.1000036 Range = [0.100033, 0.100038]
When I repro this using EXA with gdb attached, I see the following stack:
(gdb) c
Continuing.
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fe043d75700 (LWP 4250)]
driBindContext (pcp=0x7b9c160, pdp=0x502fc0, prp=0x502fc0) at ../common/dri_util.c:178
178 ../common/dri_util.c: No such file or directory.
in ../common/dri_util.c
(gdb) bt
#0 driBindContext (pcp=0x7b9c160, pdp=0x502fc0, prp=0x502fc0) at ../common/dri_util.c:178
#1 0x00007fe0405f0876 in __glXForceCurrent (cl=0x696fe20, tag=<value optimized out>, error=0x7fff4bda6abc) at ../../glx/glxext.c:394
#2 0x00007fe0405ee9ee in DoMakeCurrent (cl=0x696fe20, drawId=<value optimized out>, readId=62914831, contextId=<value optimized out>, tag=1)
at ../../glx/glxcmds.c:605
#3 0x00007fe0405f0de2 in __glXDispatch (client=0x5dcaf00) at ../../glx/glxext.c:523
#4 0x000000000044e354 in Dispatch () at ../../dix/dispatch.c:437
#5 0x0000000000433ddd in main (argc=10, argv=0x7fff4bda6cd8, envp=<value optimized out>) at ../../dix/main.c:397
(gdb) bt full
#0 driBindContext (pcp=0x7b9c160, pdp=0x502fc0, prp=0x502fc0) at ../common/dri_util.c:178
psp = (__DRIscreenPrivate *) 0x3eb46f0
#1 0x00007fe0405f0876 in __glXForceCurrent (cl=0x696fe20, tag=<value optimized out>, error=0x7fff4bda6abc) at ../../glx/glxext.c:394
cx = (__GLXcontext *) 0xb52dd50
#2 0x00007fe0405ee9ee in DoMakeCurrent (cl=0x696fe20, drawId=<value optimized out>, readId=62914831, contextId=<value optimized out>, tag=1)
at ../../glx/glxcmds.c:605
client = (ClientPtr) 0x5dcaf00
reply = {type = 128 '\200', unused = 226 '�', sequenceNumber = 125, length = 0, contextTag = 0, pad2 = 0, pad3 = 8250040, pad4 = 0,
pad5 = 1075485366, pad6 = 32736}
glxc = (__GLXcontext *) 0x7c02590
prevglxc = (__GLXcontext *) 0xb52dd50
drawPriv = (__GLXdrawable *) 0x79ab750
readPriv = (__GLXdrawable *) 0x79ab750
error = <value optimized out>
mask = <value optimized out>
__PRETTY_FUNCTION__ = "DoMakeCurrent"
#3 0x00007fe0405f0de2 in __glXDispatch (client=0x5dcaf00) at ../../glx/glxext.c:523
stuff = (xGLXSingleReq *) 0x6554008
opcode = <value optimized out>
cl = (__GLXclientState *) 0x696fe20
retval = 1
#4 0x000000000044e354 in Dispatch () at ../../dix/dispatch.c:437
result = <value optimized out>
client = (ClientPtr) 0x5dcaf00
nready = 0
start_tick = 720
#5 0x0000000000433ddd in main (argc=10, argv=0x7fff4bda6cd8, envp=<value optimized out>) at ../../dix/main.c:397
i = 1
alwaysCheckForInput = {0, 1}
(gdb)
Created attachment 24086 [details]
dmesg (EXA makeCurrent SEGV, logs captured while GDB had just stopped at SIGSEGV)
Created attachment 24087 [details]
gpu registers (EXA makeCurrent SEGV, logs captured while GDB had just stopped at SIGSEGV)
Created attachment 24088 [details]
proc_gem_objects (EXA makeCurrent SEGV, logs captured while GDB had just stopped at SIGSEGV)
Created attachment 24089 [details]
gem_interrupts (EXA makeCurrent SEGV, logs captured while GDB had just stopped at SIGSEGV)
Created attachment 24090 [details]
package versions and exact kernel version used (EXA makeCurrent SEGV, logs captured while GDB had just stopped at SIGSEGV)
Created attachment 24091 [details]
xorg conf (EXA makeCurrent SEGV, logs captured while GDB had just stopped at SIGSEGV)
Created attachment 24092 [details]
xorg.log.old (EXA makeCurrent SEGV, logs captured while GDB had just stopped at SIGSEGV)
Created attachment 24093 [details]
xorg.log (EXA makeCurrent SEGV, logs captured while GDB had just stopped at SIGSEGV)
Created attachment 24094 [details]
xsession-errors (EXA makeCurrent SEGV, logs captured while GDB had just stopped at SIGSEGV)
The logs above was captured while gdb had first stopped inside GDB with the SIGSEG signal. _After_ I let it continue I get back to gdm of course and then I login. At this point, if I open the xorg.log.old file, the following backtrace has also been printed (but this is probably just a poor version of the gdb backtrace I pasted before I attached all the logs). Backtrace: 0: /usr/X11R6/bin/X(xorg_backtrace+0x26) [0x4f1976] 1: /usr/X11R6/bin/X(xf86SigHandler+0x41) [0x485a91] 2: /lib/libc.so.6 [0x7fe0418d6040] 3: /usr/lib/dri/i965_dri.so [0x7fe02eb51d51] 4: /usr/lib/xorg/modules/extensions//libglx.so(__glXForceCurrent+0x96) [0x7fe0405f0876] 5: /usr/lib/xorg/modules/extensions//libglx.so [0x7fe0405ee9ee] 6: /usr/lib/xorg/modules/extensions//libglx.so [0x7fe0405f0de2] 7: /usr/X11R6/bin/X(Dispatch+0x364) [0x44e354] 8: /usr/X11R6/bin/X(main+0x3bd) [0x433ddd] 9: /lib/libc.so.6(__libc_start_main+0xe6) [0x7fe0418c15a6] 10: /usr/X11R6/bin/X [0x433269] Saw signal 11. Server aborting. [ 545.546298] (II) Razer Reclusa Keyboard: Close [ 545.546340] (II) UnloadModule: "evdev" [ 545.562149] (II) Razer Reclusa Keyboard: Close [ 545.562183] (II) UnloadModule: "evdev" [ 545.594244] (II) Macintosh mouse button emulation: Close [ 545.594268] (II) UnloadModule: "evdev" [ 545.617308] (II) Microsoft Microsoft 5-Button Mouse with IntelliEye(TM): Close [ 545.617346] (II) UnloadModule: "evdev" [ 545.617370] (II) AIGLX: Suspending AIGLX clients for VT switch [ 545.832855] (II) intel(0): xf86UnbindGARTMemory: unbind key 0 ddxSigGiveUp: Closing log I have to reassign this to Eric since I can't reproduce this with EXA, and only see the problem with UXA (reported in bug#19177). *** Bug 21193 has been marked as a duplicate of this bug. *** Information from duplicate bug 21193: user liuhaien <haien.liu@intel.com> was able to repro this bug (in UXA) using Fedora 9 plus the following versions from GIT. System Environment: -------------------------- Host: x-945gm Arch: i386 Platform: 945GM OSD: Fedora release 9 (Sulphur) Kernel_version: 2.6.29.1 Libdrm: (master)07646002c6835537c6ae44ef9b3f8480762279b8 Mesa: (mesa_7_4_branch)de197cf991416f0cd65ad2e2d2ca9aa599b52075 Xserver: (server-1.6-branch)60c161545af80eb78eb790a05bde79409dfdf16e Xf86_video_intel: (2.7)121bd7ff7cfd9a43fbb61fa56f06ba2d2b55035e Kernel: (drm-intel-2.6.29)0aa7e8a61658193f90198d579c5ef8297b45be9b Bug detailed description: ------------------------- startx with uxa, then run glean case makeCurrent,X will crash soon.the backtrace from gdb is as below: #0 0xb7eb3d1b in DRI2GetBuffers (pDraw=0x0, width=0x9754f04, height=0x9754f08, attachments=0xbff0c478, count=4, out_count=0xbff0c4a4) at dri2.c:139 #1 0xb7e51295 in dri2GetBuffers (driDrawable=0x9754ee0, width=0x9754f04, height=0x9754f08, attachments=0xbff0c478, count=4, out_count=0xbff0c4a4, loaderPrivate=0x9754e38) at glxdri2.c:363 #2 0xa7a4be10 in intel_update_renderbuffers (context=0x93ce2d0, drawable=0x9754ee0) at intel_context.c:235 #3 0xa7a4c261 in intelMakeCurrent (driContextPriv=0x93ce2d0, driDrawPriv=0x9754ee0, driReadPriv=0x9754ee0) at intel_context.c:837 #4 0xa7a272be in driBindContext (pcp=0x93ce2d0, pdp=0x9754ee0, prp=0x9754ee0) at ../common/dri_util.c:203 #5 0xb7e5113c in __glXDRIcontextForceCurrent (baseContext=0x93d3350) at glxdri2.c:202 #6 0xb7e46356 in __glXForceCurrent (cl=0x93cf2d0, tag=1, error=0xbff0c5a8) at glxext.c:394 #7 0xb7e43fe7 in DoMakeCurrent (cl=0x93cf2d0, drawId=<value optimized out>, readId=0, contextId=0, tag=1) at glxcmds.c:605 #8 0xb7e46876 in __glXDispatch (client=0x93ebec8) at glxext.c:523 #9 0x0808672f in Dispatch () at dispatch.c:437 #10 0x0806c65d in main (argc=2, argv=0xbff0c724, envp=0x9754e38) at main.c:397 Reproduce Steps: ------------------------ 1.xinit& (with uxa) 2. run glean case makeCurrent Please validate if below commit fixes: Author: Brian Paul <brianp@vmware.com> Date: Wed Jun 17 07:58:35 2009 -0600 GLX: attempt to fix glean makeCurrent test cases. Two parts to this: One we don't keep pointers to possibly freed memory anymore once we unbind the drawables from the context. Brian I need to figure out what the comment you made there, can we get a glean/piglit test so we can fix it properly? If the new gc is the same as the oldGC, we call the unbind even though we just bound it in that function. doh. (cherry picked from master, commit 77506dac8e81e9548a7e9680ce367175fe5747af) zhao jian, since you were able to repro can you check with UXA and the suggested patch? My G45 machine is packed in a moving box in another country, I will be only able to check the repro with the suggested patch after mid-August at the earliest. I found the similar failure when running glean makeCurrent on Moblin2, Mesa 7.4 and xserver 1.6.2. I think it could be the same issue talking about here. The root cause I believe is that Mesa tries to send a request to X to unbind a destroyed context. I have to check oldGC->xid. If None, then means it is destroyed and don't send request to X. I attach the patch. It works for my case. Hopefully it also resolve your issue and any comment is welcome. Created attachment 27179 [details]
patch made for mesa/src/glx/x11/glxcurrent.c
The patch is based on mesa/gallium-mesa-7.4 branch
Add mesa-dev mailing list to review my previous patch for glean makeCurrent issue. Appears to work now with current Mesa and X Server. Yes. I tried with the newest code on G45, it works well. Libdrm: (master)de1ed01214874dcdd6116ff2587c8710d6ed4d2d Mesa: (master)5e6b593d35156a0068dc0eb3e55dec086f1cadd3 Xserver: (master)3525d140567e0ad5f0184e4b37893c47239e1628 Xf86_video_intel: (master)1e4784bf26e3c154f5673f7b5add3ef7af3b1474 |
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.