| Summary: | x11-base/xorg-server-21.1.2-r4 segfaults with x11-drivers/nvidia-drivers-495.46 when using xpresent extensions | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | acab |
| Component: | Current packages | Assignee: | Gentoo X packagers <x11> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | ionen, jstein, sam |
| Priority: | Normal | ||
| Version: | unspecified | ||
| Hardware: | AMD64 | ||
| OS: | Linux | ||
| URL: | https://gitlab.freedesktop.org/xorg/xserver/-/issues/1275 | ||
| See Also: | https://bugs.gentoo.org/show_bug.cgi?id=850679 | ||
| Whiteboard: | Fixed in master and in 21.1.4 | ||
| Package list: | Runtime testing required: | --- | |
| Bug Depends on: | |||
| Bug Blocks: | 837317 | ||
| Attachments: |
xorg log file
emerge --info |
||
|
Description
acab
2022-01-03 02:20:56 UTC
Created attachment 761133 [details]
xorg log file
Thank you for the report. We need to have all information at hand before ticket assignment. Please * paste the emerge info as described on https://wiki.gentoo.org/wiki/Attach_the_logs_to_the_bug_ticket Does it still happen with nvidia-drivers-495.44-r2? (stable 470.94 should be fine too). If it does, I suspect it's https://gitlab.freedesktop.org/xorg/xserver/-/issues/1275 Still unclear if it's a Xorg bug that happened to work by accident on a nvidia-drivers regression. (In reply to Ionen Wolkens from comment #3) > Does it still happen with nvidia-drivers-495.44-r2? (stable 470.94 should be > fine too). > > If it does, I suspect it's err, meant to say if it doesn't Created attachment 761176 [details]
emerge --info
No reason to doubt ionen's diagnosis, as usual, but could possibly be helpful to enable debugging symbols for X (I know you did libunwind though) and glibc, just to see if it makes it any more clearer. (In reply to Ionen Wolkens from comment #3) > Does it still happen with nvidia-drivers-495.44-r2? (stable 470.94 should be > fine too). > > If it does, I suspect it's > https://gitlab.freedesktop.org/xorg/xserver/-/issues/1275 > > Still unclear if it's a Xorg bug that happened to work by accident on a > nvidia-drivers regression. No, with nvidia-drivers-495.44-r2 the bug goes away. Nice catch thanks! (In reply to Sam James from comment #6) > No reason to doubt ionen's diagnosis, as usual, but could possibly be > helpful to enable debugging symbols for X (I know you did libunwind though) > and glibc, just to see if it makes it any more clearer. Here you go Thread 1 "X" received signal SIGSEGV, Segmentation fault. 0x0000000000000000 in ?? () (gdb) bt #0 0x0000000000000000 in ?? () #1 0x000055813fbf9ddd in present_get_crtc (window=window@entry=0x5581419e4320) at ../xorg-server-21.1.2/present/present.c:52 #2 0x000055813fbfd070 in present_scmd_pixmap (window=0x5581419e4320, pixmap=0x558141a44300, serial=0, valid=0x0, update=0x558141a916a0, x_off=<optimized out>, y_off=0, target_crtc=<optimized out>, wait_fence=0x0, idle_fence=0x0, options=0, target_window_msc=0, divisor=1, remainder=0, notifies=0x0, num_notifies=0) at ../xorg-server-21.1.2/present/present_scmd.c:699 #3 0x000055813fbfbb72 in proc_present_pixmap (client=0x5581419b2a20) at ../xorg-server-21.1.2/present/present_request.c:146 #4 0x000055813fb989e4 in Dispatch () at ../xorg-server-21.1.2/dix/dispatch.c:550 #5 0x000055813fb9cb63 in dix_main (argc=8, argv=0x7ffcdec2e688, envp=<optimized out>) at ../xorg-server-21.1.2/dix/main.c:272 #6 0x00007f0ccd7eb2fa in __libc_start_call_main () from /lib64/libc.so.6 #7 0x00007f0ccd7eb3a8 in __libc_start_main () from /lib64/libc.so.6 #8 0x000055813fb5f161 in _start () Thanks, that looks identical to the issue. Technically could be fixed by patching Xorg but unsure if it's something that should be fixed on Xorg or nvidia's side (not that can do much about the latter beside wait). Hopefully something comes out of $URL. (In reply to Ionen Wolkens from comment #9) > Thanks, that looks identical to the issue. > > Technically could be fixed by patching Xorg but unsure if it's something > that should be fixed on Xorg or nvidia's side (not that can do much about > the latter beside wait). > > Hopefully something comes out of $URL. Yeah i think you should just wait for upstream to decide whether those nulls are legit or what. Meanwhile thanks a lot, you've been amazing. Is this still happening with nvidia-drivers-510.39.01? (In reply to Ionen Wolkens from comment #11) > Is this still happening with nvidia-drivers-510.39.01? Unfortunately, got a report that it is Looking at recent developments, looks like the way to go will be to fix this in Xorg. I'll keep 495.44-r2 in the tree until: https://gitlab.freedesktop.org/xorg/xserver/-/merge_requests/841 I'm affected by this issue as well. I'm on x11-drivers/nvidia-drivers-470.94, the latest for my GK. I'm back on x11-base/xorg-server-1.20.14 and have working desktop again. I'll test the patch from above later. So patch did not helped. It must be something else. Back on 1.20.14 again. [ 10.535] (EE) Backtrace: [ 10.535] (EE) 0: /usr/bin/X (xorg_backtrace+0x5b) [0x563aab5c251b] [ 10.535] (EE) 1: /usr/bin/X (0x563aab489000+0x13d185) [0x563aab5c6185] [ 10.535] (EE) 2: /lib64/libpthread.so.0 (0x7f010028c000+0x120e0) [0x7f010029e0e0] [ 10.535] (EE) 3: /usr/bin/X (InitOutput+0x491) [0x563aab5e2c81] [ 10.535] (EE) 4: /usr/bin/X (0x563aab489000+0x76ac3) [0x563aab4ffac3] [ 10.535] (EE) 5: /lib64/libc.so.6 (__libc_start_main+0xcd) [0x7f01000f27fd] [ 10.535] (EE) 6: /usr/bin/X (_start+0x2a) [0x563aab4c508a] [ 10.535] (EE) [ 10.535] (EE) Segmentation fault at address 0x10 Hi, I'd like to chime in here... Same sympthoms on xorg/lightdm/mate combination here. any nvidia-drivers >495.44-r2 would segfault and returned me to login screen. so after masking a few driver versions i found this bug. and the proposed solution to wait for a patch on the xorg side. I looked yesterday and there seemed to be some progress so went the "don't try this at home!" way. emerged xorg-server-9999, tried the latest nvidia-drivers. braced for impact and breakage. no segfault, all seems well at first glance. Guess what i'm trying to say here is... seems hopeful, just hold on a bit longer. my compliments to the teams who are looking into this. I am happy! The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=48948713f238a734ddc80071ddb60fa9e6d5a966 commit 48948713f238a734ddc80071ddb60fa9e6d5a966 Author: Ionen Wolkens <ionen@gentoo.org> AuthorDate: 2022-02-07 20:52:36 +0000 Commit: Ionen Wolkens <ionen@gentoo.org> CommitDate: 2022-02-07 21:11:40 +0000 x11-drivers/nvidia-drivers: drop vulnerable 470.94, 495.* Users affected by bug #830482 (with unpatched Xorg) should use stable 470.103.01 rather than vulnerable 495.44-r2. Bug: https://bugs.gentoo.org/830482 Bug: https://bugs.gentoo.org/832867 Signed-off-by: Ionen Wolkens <ionen@gentoo.org> x11-drivers/nvidia-drivers/Manifest | 21 - .../nvidia-drivers/nvidia-drivers-470.94.ebuild | 445 ------------------- .../nvidia-drivers/nvidia-drivers-495.44-r2.ebuild | 493 --------------------- .../nvidia-drivers-495.46-r10.ebuild | 462 ------------------- 4 files changed, 1421 deletions(-) Would it be possible to backport and stable (or optionally straight-to): https://gitlab.freedesktop.org/xorg/xserver/-/commit/22d58188 Just trivial null checks, should have consequence beside prevent segfaults. Originally it sounded like they were going to do a new release very soon (recall a comment talking of 2 weeks) but it's been taking a long time and some users need >=nvidia-drivers-510 to be stable for new cards. The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=e90981351332cd70992e6458fc7aec721b883dce commit e90981351332cd70992e6458fc7aec721b883dce Author: Sam James <sam@gentoo.org> AuthorDate: 2022-06-03 08:38:02 +0000 Commit: Sam James <sam@gentoo.org> CommitDate: 2022-06-03 08:38:13 +0000 x11-base/xorg-server: backport present crash fix Closes: https://bugs.gentoo.org/830482 Signed-off-by: Sam James <sam@gentoo.org> .../xorg-server-21.1.3-present-null-check.patch | 36 ++++ x11-base/xorg-server/xorg-server-21.1.3-r3.ebuild | 190 +++++++++++++++++++++ 2 files changed, 226 insertions(+) Thanks. |
