[gentoo-user] X11 crashes anyone?

[gentoo-user] X11 crashes anyone?

[Nikos Chantziaras]

Has anyone here noticed that x.org likes to crash sometimes as of late? 
Never happened before, going years and years back. The last month or so, 
I've got three x.org crashes:

systemd-coredump[204553]: [🡕] Process 453 (X) of user 0 dumped core.

This is x11-base/xorg-server-21.1.2-r2.

qlop -v x11-base/xorg-server says:

   2021-08-04T17:58:15 >>> x11-base/xorg-server-1.20.13-r1: 52s
   2021-11-30T16:15:54 >>> x11-base/xorg-server-21.1.1: 28s
   2021-12-03T18:49:22 >>> x11-base/xorg-server-21.1.1-r2: 27s
   2021-12-16T17:08:40 >>> x11-base/xorg-server-21.1.2: 24s
   2021-12-19T21:57:08 >>> x11-base/xorg-server-21.1.2-r2: 25s

So it seems this coincides with the upgrade from 1.20.13 to 21.1.1 about 
a month ago.

Anyone else seeing this?

Re: [gentoo-user] X11 crashes anyone?

[Michael]

[-- Attachment #1: Type: text/plain, Size: 2040 bytes --]

On Monday, 20 December 2021 07:10:59 GMT Nikos Chantziaras wrote:
> Has anyone here noticed that x.org likes to crash sometimes as of late?
> Never happened before, going years and years back. The last month or so,
> I've got three x.org crashes:
> 
> systemd-coredump[204553]: [🡕] Process 453 (X) of user 0 dumped core.
> 
> This is x11-base/xorg-server-21.1.2-r2.
> 
> qlop -v x11-base/xorg-server says:
> 
>    2021-08-04T17:58:15 >>> x11-base/xorg-server-1.20.13-r1: 52s
>    2021-11-30T16:15:54 >>> x11-base/xorg-server-21.1.1: 28s
>    2021-12-03T18:49:22 >>> x11-base/xorg-server-21.1.1-r2: 27s
>    2021-12-16T17:08:40 >>> x11-base/xorg-server-21.1.2: 24s
>    2021-12-19T21:57:08 >>> x11-base/xorg-server-21.1.2-r2: 25s
> 
> So it seems this coincides with the upgrade from 1.20.13 to 21.1.1 about
> a month ago.
> 
> Anyone else seeing this?

I have been suffering similar symptoms[1] on a AMD Kaveri APU powered box, 
running plasma with two monitors, which worsened[2] in the last couple of 
weeks.  All other boxen work fine, so I assumed some Radeon driver issue 
specific to this machine.

[1]  For some months now the RH monitor would not acquire the correct 
resolution.  The workaround I came up with, until I have some time to 
troubleshoot it, was to go into SystemSettings/Display, disable the RH monitor 
and then re-enable it.  This process is more painful than it sounds, because 
clicking on Plasma desktop GUI menus does not update the screen.  I have to 
drop into a console and back to Alt+F7 before I can see the output of whatever 
I had clicked on.  Interestingly, Plasma-Wayland worked fine, both monitors 
came up with the correct resolution.

[2] Last week's update broke both xorg and wayland.  Plasma crashes when 
launched and partially recovers.  I glanced at the errors and it seems to be 
related to some Qt library.  Again, I haven't troubleshooted it further.

Not sure if this helps, but would be interested to try in whatever fix you may 
come up with.

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[gentoo-user] Re: X11 crashes anyone?

[Nikos Chantziaras]

On 20/12/2021 13:21, Michael wrote:
> On Monday, 20 December 2021 07:10:59 GMT Nikos Chantziaras wrote:
>> Has anyone here noticed that x.org likes to crash sometimes as of late?
>> [...]
> 
> I have been suffering similar symptoms[1] on a AMD Kaveri APU powered box,
> running plasma with two monitors, which worsened[2] in the last couple of
> weeks.  All other boxen work fine, so I assumed some Radeon driver issue
> specific to this machine.
> 
> [1]  For some months now the RH monitor would not acquire the correct
> resolution. [...]
> 
> [2] Last week's update broke both xorg and wayland.  Plasma crashes when
> launched and partially recovers. [...]

Seems to be a different issue then. I'm on an nvidia card using the 
binary driver, and there's no problems like the ones you're having. It's 
a straight disappearance of the desktop here when it happens due to the 
X11 process segfault.

Re: [gentoo-user] Re: X11 crashes anyone?

[Ionen Wolkens]

[-- Attachment #1: Type: text/plain, Size: 1044 bytes --]

On Tue, Dec 21, 2021 at 08:31:55AM +0200, Nikos Chantziaras wrote:
> Seems to be a different issue then. I'm on an nvidia card using the 
> binary driver, and there's no problems like the ones you're having. It's 
> a straight disappearance of the desktop here when it happens due to the 
> X11 process segfault.

If using xorg-21, I assume using nvidia-drivers-495.xx as well?

I was never able to reproduce these, but did run into users running
into segfault-at-start issues with 495.

With 495.44-r2 and 495.46-r10, Xorg log may have a backtrace
mentioning libnvidia-glcore.so, if so try 495.46-r0.

Alternatively, 495.46* may mention failed in libpthread (or libc if
glibc-2.34), in that case try 495.44-r2 instead.

The more failsafe approach however is to just use stable drivers though.

echo "x11-drivers/nvidia-drivers -~amd64" >> /etc/portage/package.accept_keywords

Or 470.94 is fine too (due to be next stable soon).

If still happening, then I don't know what might be wrong just from
this.

-- 
ionen

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

[gentoo-user] Re: X11 crashes anyone?

[Nikos Chantziaras]

On 21/12/2021 08:50, Ionen Wolkens wrote:
> On Tue, Dec 21, 2021 at 08:31:55AM +0200, Nikos Chantziaras wrote:
>> Seems to be a different issue then. I'm on an nvidia card using the
>> binary driver, and there's no problems like the ones you're having. It's
>> a straight disappearance of the desktop here when it happens due to the
>> X11 process segfault.
> [...]
> 
> With 495.44-r2 and 495.46-r10, Xorg log may have a backtrace
> mentioning libnvidia-glcore.so, if so try 495.46-r0.

Just got another one. I was smart enough this time to save 
Xorg.0.log.old before it gets overwritten. No mention of nvidia anywhere:

[ 47526.314] (EE) Backtrace:
[ 47526.321] (EE) 0: /usr/bin/X (xorg_backtrace+0x5b) [0x55c605c611cb]
[ 47526.321] (EE) 1: /usr/bin/X (0x55c605b16000+0x14f0f5) [0x55c605c650f5]
[ 47526.321] (EE) 2: /lib64/libpthread.so.0 (0x7fd0097df000+0x12660) 
[0x7fd0097f1660]
[ 47526.321] (EE) 3: /lib64/libc.so.6 (0x7fd00961c000+0x15b0b3) 
[0x7fd0097770b3]
[ 47526.321] (EE) 4: /usr/bin/X (0x55c605b16000+0x83397) [0x55c605b99397]
[ 47526.321] (EE) 5: /usr/bin/X (0x55c605b16000+0x84e11) [0x55c605b9ae11]
[ 47526.321] (EE) 6: /usr/bin/X (0x55c605b16000+0x85335) [0x55c605b9b335]
[ 47526.321] (EE) 7: /usr/bin/X (0x55c605b16000+0x863c6) [0x55c605b9c3c6]
[ 47526.321] (EE) 8: /usr/bin/X (0x55c605b16000+0x1058e5) [0x55c605c1b8e5]
[ 47526.321] (EE) 9: /usr/bin/X (XkbHandleActions+0x2b0) [0x55c605c522f0]
[ 47526.321] (EE) 10: /usr/bin/X (0x55c605b16000+0x135401) [0x55c605c4b401]
[ 47526.321] (EE) 11: /usr/bin/X (0x55c605b16000+0x1355fa) [0x55c605c4b5fa]
[ 47526.321] (EE) 12: /usr/bin/X (0x55c605b16000+0x148686) [0x55c605c5e686]
[ 47526.321] (EE) 13: /usr/bin/X (WaitForSomething+0x268) [0x55c605c5e918]
[ 47526.321] (EE) 14: /usr/bin/X (0x55c605b16000+0x77f25) [0x55c605b8df25]
[ 47526.321] (EE) 15: /usr/bin/X (0x55c605b16000+0x7c4c3) [0x55c605b924c3]
[ 47526.321] (EE) 16: /lib64/libc.so.6 (__libc_start_main+0xcd) 
[0x7fd00963f7fd]
[ 47526.321] (EE) 17: /usr/bin/X (_start+0x2a) [0x55c605b5365a]
[ 47526.321] (EE)
[ 47526.321] (EE) Segmentation fault at address 0x7fff46403180


> Alternatively, 495.46* may mention failed in libpthread (or libc if
> glibc-2.34), in that case try 495.44-r2 instead.

It does mention libc, but I'm on glibc-2.33-r7.

I guess I'll just downgrade the nvidia drivers to 470.94 anyway and see 
if it happens again.

[gentoo-user] Re: X11 crashes anyone?

[Nikos Chantziaras]

On 26/12/2021 09:34, Nikos Chantziaras wrote:
> On 21/12/2021 08:50, Ionen Wolkens wrote:
>> On Tue, Dec 21, 2021 at 08:31:55AM +0200, Nikos Chantziaras wrote:
>>> Seems to be a different issue then. I'm on an nvidia card using the
>>> binary driver, and there's no problems like the ones you're having. It's
>>> a straight disappearance of the desktop here when it happens due to the
>>> X11 process segfault.
>> [...]
>>
>> With 495.44-r2 and 495.46-r10, Xorg log may have a backtrace
>> mentioning libnvidia-glcore.so, if so try 495.46-r0.
> 
> Just got another one. I was smart enough this time to save 
> Xorg.0.log.old before it gets overwritten. No mention of nvidia anywhere:
> 
> [ 47526.314] (EE) Backtrace:
> [ 47526.321] (EE) 0: /usr/bin/X (xorg_backtrace+0x5b) [0x55c605c611cb]
> [ 47526.321] (EE) 1: /usr/bin/X (0x55c605b16000+0x14f0f5) [0x55c605c650f5]
> [ 47526.321] (EE) 2: /lib64/libpthread.so.0 (0x7fd0097df000+0x12660) 
> [0x7fd0097f1660]
> [ 47526.321] (EE) 3: /lib64/libc.so.6 (0x7fd00961c000+0x15b0b3) 
> [...]
> [ 47526.321] (EE) Segmentation fault at address 0x7fff46403180
> 
> [...]
> 
> I guess I'll just downgrade the nvidia drivers to 470.94 anyway and see 
> if it happens again.

It just happened again with 470.94. :-/

Re: [gentoo-user] Re: X11 crashes anyone?

[Michael]

[-- Attachment #1: Type: text/plain, Size: 1495 bytes --]

On Thursday, 6 January 2022 16:51:49 GMT Nikos Chantziaras wrote:
> On 26/12/2021 09:34, Nikos Chantziaras wrote:
> > On 21/12/2021 08:50, Ionen Wolkens wrote:
> >> On Tue, Dec 21, 2021 at 08:31:55AM +0200, Nikos Chantziaras wrote:
> >>> Seems to be a different issue then. I'm on an nvidia card using the
> >>> binary driver, and there's no problems like the ones you're having. It's
> >>> a straight disappearance of the desktop here when it happens due to the
> >>> X11 process segfault.
> >> 
> >> [...]
> >> 
> >> With 495.44-r2 and 495.46-r10, Xorg log may have a backtrace
> >> mentioning libnvidia-glcore.so, if so try 495.46-r0.
> > 
> > Just got another one. I was smart enough this time to save
> > Xorg.0.log.old before it gets overwritten. No mention of nvidia anywhere:
> > 
> > [ 47526.314] (EE) Backtrace:
> > [ 47526.321] (EE) 0: /usr/bin/X (xorg_backtrace+0x5b) [0x55c605c611cb]
> > [ 47526.321] (EE) 1: /usr/bin/X (0x55c605b16000+0x14f0f5) [0x55c605c650f5]
> > [ 47526.321] (EE) 2: /lib64/libpthread.so.0 (0x7fd0097df000+0x12660)
> > [0x7fd0097f1660]
> > [ 47526.321] (EE) 3: /lib64/libc.so.6 (0x7fd00961c000+0x15b0b3)
> > [...]
> > [ 47526.321] (EE) Segmentation fault at address 0x7fff46403180
> > 
> > [...]
> > 
> > I guess I'll just downgrade the nvidia drivers to 470.94 anyway and see
> > if it happens again.
> 
> It just happened again with 470.94. :-/

Just an idea, have you tried rebuilding sys-libs/glibc and xorg, since this is 
what it barfs at as it segfaults?

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[gentoo-user] Re: X11 crashes anyone?

[Nikos Chantziaras]

On 20/12/2021 09:10, Nikos Chantziaras wrote:
> Has anyone here noticed that x.org likes to crash sometimes as of late? 
> Never happened before, going years and years back. The last month or so, 
> I've got three x.org crashes:
> 
> systemd-coredump[204553]: [🡕] Process 453 (X) of user 0 dumped core.

So half a year later and I was still getting these crashes. I found the 
possible issue. More and more people started having the same crashes as 
different distros were updating their X11 stuff. It seems the issue is this:

   https://gitlab.freedesktop.org/xorg/xserver/-/issues/1261

A patch was merged in xorg upstream about 4 months ago:

   https://gitlab.freedesktop.org/xorg/xserver/-/merge_requests/865

Downloadable diff:

   https://gitlab.freedesktop.org/xorg/xserver/-/merge_requests/865.diff

This patch is not carried in portage. I copied it to 
/etc/portage/patches/x11-base/xorg-server/ and rebuilt xorg-server three 
days ago, and I haven't had any more crashes.

Fingers crossed...