| Summary: | xorg-server-1.9, mesa-7.9, xf86-video-vesa: segfault | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Andrey Grozin <grozin> |
| Component: | [OLD] Unspecified | Assignee: | Gentoo X packagers <x11> |
| Status: | RESOLVED WORKSFORME | ||
| Severity: | normal | CC: | Hugo.Mildenberger, jan.narovec, rb6 |
| Priority: | High | ||
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: |
Xorg.0.log
emerge --info |
||
|
Description
Andrey Grozin
2010-12-04 14:11:19 UTC
Created attachment 256317 [details]
Xorg.0.log
Tbh gallium useflag HAS no effect on performance unless you select the engine by hand. Does switching between classic and gallium sw driver (with eselect mesa) make a difference? (In reply to comment #3) > Does switching between classic and gallium sw driver (with eselect mesa) make a > difference? OK, I again emerged mesa-7.9, xorg-server-1.9.2.902, and re-emerged xf86-video-vesa-2.3.0, xf86-input-evdev-2.5.0 (these things take quite considerable time on eeepc!). If I eselect mesa set sw gallium, kde starts; I get a message from kwin that desktop effects are too slow and have been disabled; video (800x600 resilution) is impossibly slow. Is I eselect mesa set sw classic, X crashes: Backtrace: [ 77077.199] 0: /usr/bin/X (xorg_backtrace+0x3c) [0x80e8ecc] [ 77077.199] 1: /usr/bin/X (0x8048000+0x60376) [0x80a8376] [ 77077.199] 2: (vdso) (__kernel_rt_sigreturn+0x0) [0xb77c840c] [ 77077.200] 3: /usr/lib/xorg/modules/libfb.so (fbBltStip+0x74) [0xb71d3384] [ 77077.200] 4: /usr/lib/xorg/modules/libfb.so (fbGetImage+0x1a1) [0xb71d8651] [ 77077.200] 5: /usr/lib/xorg/modules/libshadow.so (0xb71e3000+0x124d) [0xb71e 424d] [ 77077.200] 6: /usr/bin/X (0x8048000+0x161d2f) [0x81a9d2f] [ 77077.200] 7: /usr/lib/xorg/modules/extensions/libglx.so (0xb7392000+0x3954c ) [0xb73cb54c] [ 77077.200] 8: /usr/lib/dri/swrast_dri.so (0xb654e000+0xf46e) [0xb655d46e] [ 77077.200] 9: /usr/lib/dri/swrast_dri.so (0xb654e000+0xf4b8) [0xb655d4b8] [ 77077.200] 10: /usr/lib/xorg/modules/extensions/libglx.so (0xb7392000+0x394a e) [0xb73cb4ae] [ 77077.200] 11: /usr/lib/xorg/modules/extensions/libglx.so (0xb7392000+0x35b1 a) [0xb73c7b1a] [ 77077.200] 12: /usr/lib/xorg/modules/extensions/libglx.so (0xb7392000+0x3699 b) [0xb73c899b] [ 77077.200] 13: /usr/lib/xorg/modules/extensions/libglx.so (0xb7392000+0x38a6 0) [0xb73caa60] [ 77077.200] 14: /usr/bin/X (0x8048000+0x2a767) [0x8072767] [ 77077.200] 15: /usr/bin/X (0x8048000+0x1d665) [0x8065665] [ 77077.200] 16: /lib/libc.so.6 (__libc_start_main+0xe6) [0xb74dbcc6] [ 77077.200] 17: /usr/bin/X (0x8048000+0x1d221) [0x8065221] [ 77077.201] Segmentation fault at address (nil) [ 77077.201] Fatal server error: [ 77077.201] Caught signal 11 (Segmentation fault). Server aborting Going back to mesa-7.8.2 and xorg-server-1.7.7-r1, the latest version which works... Similar behaviour here, but on completely different hardware (amd64/radeon): Backtrace: [ 3042.661] 0: /usr/bin/X (xorg_backtrace+0x28) [0x49feb8] [ 3042.661] 1: /usr/bin/X (0x400000+0x62d69) [0x462d69] [ 3042.661] 2: /lib/libpthread.so.0 (0x7fdb72a6f000+0xee90) [0x7fdb72a7de90] [ 3042.661] 3: /usr/lib64/xorg/modules/extensions/libglx.so (0x7fdb70afd000+0x37c21) [0x7fdb70b34c21] [ 3042.661] 4: /usr/lib64/xorg/modules/extensions/libglx.so (0x7fdb70afd000+0x3f66a) [0x7fdb70b3c66a] [ 3042.661] 5: /usr/lib64/xorg/modules/extensions/libglx.so (0x7fdb70afd000+0x36aff) [0x7fdb70b33aff] [ 3042.661] 6: /usr/bin/X (FreeResource+0x143) [0x44d5e3] [ 3042.661] 7: /usr/lib64/xorg/modules/extensions/libglx.so (0x7fdb70afd000+0x33bb9) [0x7fdb70b30bb9] [ 3042.661] 8: /usr/lib64/xorg/modules/extensions/libglx.so (0x7fdb70afd000+0x36922) [0x7fdb70b33922] [ 3042.661] 9: /usr/bin/X (0x400000+0x2ef49) [0x42ef49] [ 3042.662] 10: /usr/bin/X (0x400000+0x248fa) [0x4248fa] [ 3042.662] 11: /lib/libc.so.6 (__libc_start_main+0xfd) [0x7fdb71c67bad] [ 3042.662] 12: /usr/bin/X (0x400000+0x24499) [0x424499] [ 3042.662] Segmentation fault at address 0x290 [ 3042.662] Fatal server error: [ 3042.662] Caught signal 11 (Segmentation fault). Server aborting media-libs/mesa-7.9 x11-base/xorg-server-1.9.2 Previous version of xorg was rock stable at my machine. Now, after reverting back to x11-base/xorg-server-1.7.7-r1 and x11-apps/xinit-1.2.0-r4, is my X session running for more than three hours without any sign of instability. Created attachment 258668 [details]
emerge --info
I have a very similar problem with xf86-videe-intel. Had to go back to mesa 7.8.2. Any update on this? (In reply to comment #8) If you get a crash with a stacktrace containing libshadow on intel, then something is very wrong. Regarding xf86-video-vesa, I doubt that anybody at freedesktop.org still cares enough to have it fixed. Radeon crashes have mostly been due to vesafb/uvesafb/radeonfb (bug 350326) or strange things happening with gallium (bug 350280). My backtrace from kdm.log is as follows: Backtrace: 0: /usr/bin/X (xorg_backtrace+0x28) [0x4a1c58] 1: /usr/bin/X (0x400000+0x643f9) [0x4643f9] 2: /lib/libpthread.so.0 (0x7fcb73c72000+0xf410) [0x7fcb73c81410] 3: /lib/libc.so.6 (memcpy+0x15b) [0x7fcb72ebdddb] 4: /usr/lib64/xorg/modules/libfb.so (fbBlt+0x100) [0x7fcb70a3cbc0] 5: /usr/lib64/xorg/modules/libfb.so (fbBltStip+0x40) [0x7fcb70a3d8f0] 6: /usr/lib64/xorg/modules/libfb.so (fbGetImage+0x20a) [0x7fcb70a4251a] 7: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7fcb71062000+0x28fc5) [0x7fcb7108afc5] 8: /usr/bin/X (0x400000+0x1573c0) [0x5573c0] 9: /usr/lib64/xorg/modules/extensions/libglx.so (0x7fcb71ad5000+0x36c5a) [0x7fcb71b0bc5a] 10: /usr/lib64/dri/swrast_dri.so (0x7fcb704be000+0x2590a) [0x7fcb704e390a] 11: /usr/lib64/xorg/modules/extensions/libglx.so (0x7fcb71ad5000+0x36bf8) [0x7fcb71b0bbf8] 12: /usr/lib64/xorg/modules/extensions/libglx.so (0x7fcb71ad5000+0x32d7d) [0x7fcb71b07d7d] 13: /usr/lib64/xorg/modules/extensions/libglx.so (0x7fcb71ad5000+0x320e2) [0x7fcb71b070e2] 14: /usr/lib64/xorg/modules/extensions/libglx.so (0x7fcb71ad5000+0x367e2) [0x7fcb71b0b7e2] 15: /usr/bin/X (0x400000+0x30119) [0x430119] 16: /usr/bin/X (0x400000+0x257da) [0x4257da] 17: /lib/libc.so.6 (__libc_start_main+0xfd) [0x7fcb72e5bd2d] 18: /usr/bin/X (0x400000+0x25379) [0x425379] Segmentation fault at address (nil) Different problem? Then your issue is not the one described in the bug. (In reply to comment #5) > Similar behaviour here, but on completely different hardware (amd64/radeon): libglx has still a double-free and a use-after-free problem. You may try the patch attached to https://bugs.freedesktop.org/show_bug.cgi?id=32822 Recently, someone described here a quick method to check patches: 1.) create a directory /etc/portage/patches/x11-base/xorg-server ( or, e.g., /etc/portage/patches/x11-base/xorg-server-1.9.2.902 ) 2.) put the patch into this directory 3.) emerge x11-base/xorg-server 4.) Watch for the message that your patch had been applied (In reply to comment #12) Thank you. I've "solved" the issue by turning on KMS. Now, with vanilla kernel 2.6.37, KMS on, xorg-server-1.9.2, mesa-7.9 and rendering set to "classic", everything works as expected. I've read that the kernel 2.6.37 fixes regression - black screen on some Intel graphics hardware. Regression indeed - the screen became black still in the text mode, when the kernel (with KMS) decided to switch to the native 1024x600 resolution during boot (and it remained black if I blindly log in and say startx). So, I installed 2.6.37 with KMS and intel kernel driver, and, indeed, the screen is no longer black. X (I use kde-4.5) started in the 1024x600 mode, but at first it was very unstable - after some usual actions (like closing a window) the mouse pointer moves, but clicks have no effect; keyboard input also does not work (perhaps thare is no focus window), except ctrl-alt-f1, and I had to kill X by ctrl-c. But after the recent upgrade of xf86-video-intel, xf86-input-evdev, xorg-server, mesa these effects disappeared. And I have a normally working kde on my eeepc with the intel driver and KMS. So, closing as WORKSFORME. However, on another computer, also with Intel graphics (but a different one), the X server always segfaults when I exit kde. This does not interfere with the normal work, but still shows that xorg-server-1.9 (or maybe xf86-video-intel, or mesa) is fragile and unstable. (In reply to comment #14) > > However, on another computer, also with Intel graphics (but a different one), > the X server always segfaults when I exit kde. This does not interfere with the > normal work, but still shows that xorg-server-1.9 (or maybe xf86-video-intel, > or mesa) is fragile and unstable. > I have the same experience, here is the end of Xorg.log (Radeon HD3300, KMS, vanilla 2.6.37, classic driver - r600, xorg-server-1.9.2): [ 13053.262] (II) RADEON(0): Modeline "1280x960"x0.0 108.00 1280 1376 1488 1800 960 961 964 1000 +hsync +vsync (60.0 kHz) [ 13053.262] (II) RADEON(0): Modeline "1680x1050"x0.0 119.00 1680 1728 1760 1840 1050 1053 1059 1080 +hsync -vsync (64.7 kHz) [ 13263.847] Backtrace: [ 13263.894] 0: /usr/bin/X (xorg_backtrace+0x28) [0x49feb8] [ 13263.894] 1: /usr/bin/X (0x400000+0x62d69) [0x462d69] [ 13263.894] 2: /lib/libpthread.so.0 (0x7f192046f000+0xee90) [0x7f192047de90] [ 13263.894] 3: /usr/lib64/xorg/modules/extensions/libglx.so (0x7f191e4fd000+0x36a68) [0x7f191e533a68] [ 13263.894] 4: /usr/bin/X (FreeClientResources+0xe6) [0x44d116] [ 13263.894] 5: /usr/bin/X (FreeAllResources+0x3e) [0x44d1ce] [ 13263.894] 6: /usr/bin/X (0x400000+0x2492e) [0x42492e] [ 13263.895] 7: /lib/libc.so.6 (__libc_start_main+0xfd) [0x7f191f667bad] [ 13263.895] 8: /usr/bin/X (0x400000+0x24499) [0x424499] [ 13263.895] Segmentation fault at address 0x4 [ 13263.895] Fatal server error: [ 13263.895] Caught signal 11 (Segmentation fault). Server aborting [ 13263.895] [ 13263.907] Please consult the The X.Org Foundation support at http://wiki.x.org for help. [ 13263.908] Please also check the log file at "/var/log/Xorg.0.log" for additional information. [ 13263.908] [ 13263.932] (II) Power Button: Close [ 13263.954] (II) UnloadModule: "evdev" [ 13263.996] (II) Power Button: Close [ 13263.996] (II) UnloadModule: "evdev" [ 13264.028] (II) HID 04cf:0022: Close [ 13264.028] (II) UnloadModule: "evdev" [ 13264.060] (II) HID 04cf:0022: Close [ 13264.060] (II) UnloadModule: "evdev" [ 13264.124] (II) Logitech USB-PS/2 Optical Mouse: Close [ 13264.124] (II) UnloadModule: "evdev" [ 13264.124] (II) AIGLX: Suspending AIGLX clients for VT switch (In reply to comment #15) > (In reply to comment #14) did you try the patch I proposed to test in comment #12? (In reply to comment #16) Not yet. After applying those patches (four in total), the problem with segmentation faults when exiting X-server seems to be gone. Thank you. (In reply to comment #17) > (In reply to comment #16) > > Not yet. After applying those patches (four in total), the problem with > segmentation faults when exiting X-server seems to be gone. Thank you. > Well, I was too fast. Yesterday, when turning off the computer, the segfault was there again, so the patches didn't help. After a hour of debugging, I found the line, where the X server crashes: Program received signal SIGSEGV, Segmentation fault. 0x00007fd774e98a68 in DrawableGone (glxPriv=0x398e2a0, xid=0) at glxext.c:133 Unfortunately, I didn't manage to get a full backtrace from gdb, but better than nothing. Another observation: the segfault appears always when quitting KDE session during which the compositing effects were disabled and enabled again. (In reply to comment #19) > Program received signal SIGSEGV, Segmentation fault. > 0x00007fd774e98a68 in DrawableGone (glxPriv=0x398e2a0, xid=0) at glxext.c:133 Same symptoms, same line, same xid. See: https://bugs.freedesktop.org/show_bug.cgi?id=32822 >Unfortunately, I didn't manage to get a full backtrace from gdb, but >better than nothing. You probably don't get a full backtrace, because eselect opengl (or may be also eselect mesa) shuffles around libglx.so and friends (libdri,libdri2) while not caring about to sym-link libglx.so.debug too. Hence gdb usually does not find them. |