Bug#1035938: Additional info

2023-05-12 Thread Kamil Wilczek

My colleague found this post from the past:
https://lowendtalk.com/discussion/174556/segfaults-when-booting-debian-11-installer

It describes a very similar situation to ours. The problem seems to
be affecting only certain combinations of AMD CPUs and specific kernels, 
and in our case the affected KVM guests are on the only hypervisor in

our server room that has an AMD CPU. Before them I upgraded several
other systems -- hosted on servers with Intel CPUs -- and I encountered
no problems at all.

Kind regards
--
Kamil Wilczek [https://keys.openpgp.org/]
[6C4BE20A90A1DBFB3CBE2947A832BF5A491F9F2A]



Bug#1035938: Workaround to the problem

2023-05-11 Thread Kamil Wilczek

Installing the 6.1.X kernel from the bullseye-backports
apparently solves the problem, so this might be a temporary
solution to affected systems (if applicable).

Kind regards,
--
Kamil Wilczek [https://keys.openpgp.org/]
[6C4BE20A90A1DBFB3CBE2947A832BF5A491F9F2A]


OpenPGP_signature
Description: OpenPGP digital signature


Bug#1035938: KVM guests did not boot properly after upgrade from 11.6 to 11.7

2023-05-11 Thread Kamil Wilczek
temd-sysuser[331]: segfault at 7f4a6768e090 ip 
7f4a683291ef sp 7fff92a6dc58 error 27 in 
ld-2.31.so[7f4a6830a000+2]
[2.33] setfont[332]: segfault at 7ffe6ed25fe8 ip 
7f3807b173d6 sp 7ffe6ed25fe8 error 25 in 
libc-2.31.so[7f3807a4e000+159000]

[2.222686] Code: Unable to access opcode bytes at RIP 0x7f4a683291c5.
[2.223633] Code: Unable to access opcode bytes at RIP 0x7f3807b173ac.
[2.225057] iTCO_vendor_support: vendor-support=0
[2.225520] cryptd: max_cpu_qlen set to 1000
[2.226497] fuse: init (API version 7.32)
[2.228231] iTCO_wdt: Intel TCO WatchDog Timer Driver v1.11
[2.228728] iTCO_wdt: Found a ICH9 TCO device (Version=2, TCOBASE=0x0660)
[2.229398] iTCO_wdt: initialized. heartbeat=30 sec (nowayout=0)
[2.232657] AVX2 version of gcm_enc/dec engaged.
[2.233042] AES CTR mode by8 optimization enabled
[2.287356] EXT4-fs (vdc1): mounted filesystem with ordered data 
mode. Opts: (null)
[2.288381] EXT4-fs (vdb1): mounted filesystem with ordered data 
mode. Opts: (null)
[2.293006] systemd-journald[255]: Received client request to flush 
runtime journal.
[2.301747] FAT-fs (vda1): Volume was not properly unmounted. Some 
data may be corrupt. Please run fsck.
[2.304947] systemd-journal[255]: segfault at 7f1513dd9e00 ip 
7f15157e8683 sp 7fff6feda320 error 25 in 
libsystemd-shared-247.so[7f151567a000+18d000]

[2.306036] Code: Unable to access opcode bytes at RIP 0x7f15157e8659.
[2.307205] systemd[1]: systemd-journal-flush.service: Main process 
exited, code=exited, status=1/FAILURE
[2.308002] systemd[1]: systemd-journal-flush.service: Failed with 
result 'exit-code'.



Kind regards,
--
Kamil Wilczek [https://keys.openpgp.org/]
[6C4BE20A90A1DBFB3CBE2947A832BF5A491F9F2A]



Bug#977514: mutter-common: gnome-shell crashed on Wayland and lags on Xorg with nVidia drivers (nouveau unusable)

2020-12-15 Thread Kamil Wilczek
Package: mutter-common
Version: 3.38.2-1
Severity: important
Tags: upstream
X-Debbugs-Cc: kamil.wilc...@protonmail.com


I have an nVidia GTX 1650 GPU and installed weekly
testing system with GNOME desktop.


* nouveau is basically unusable, it displays the desktop,
  but lags heavily and simple tasks, like editing a file
  in Vim is impossible, keyboard strokes are missed, or
  unwanted symbols appear (multiple letters are added
  or several letters are removed after single backspace press).


So I installed nvidia-driver (currently 450.80.02) from non-free.
I had to do it from a TTY, otherwise it was impossible.
After reboot Wayland was still enabled (as default).
Performance was great, but after a moment gnome-shell crashed,
rebooted itself, then, for example gnome-terminal was lagging,
some windows too (not often). This was in the logs just before
the crash:


gru 15 21:33:35 wilk gnome-shell[1438]: Can't update stage views actor
overviewGroup is on because it needs an allocation.
gru 15 21:33:35 wilk gnome-shell[1438]: Can't update stage views actor overview
is on because it needs an allocation.
gru 15 21:33:35 wilk gnome-shell[1438]: Can't update stage views actor
Gjs_ui_overviewControls_ControlsManager is on because it needs an allocation.
gru 15 21:33:35 wilk gnome-shell[1438]: Can't update stage views actor
overview-group is on because it needs an allocation.
gru 15 21:33:35 wilk gnome-shell[1438]: Can't update stage views actor
Gjs_ui_overviewControls_ThumbnailsSlider is on because it needs an allocation.
gru 15 21:33:35 wilk gnome-shell[1438]: Can't update stage views actor
Gjs_ui_workspaceThumbnail_ThumbnailsBox is on because it needs an allocation.
gru 15 21:33:35 wilk gnome-shell[1438]: Can't update stage views actor
Gjs_ui_workspaceThumbnail_WorkspaceThumbnail is on because it needs an
allocation.
gru 15 21:33:35 wilk gnome-shell[1438]: Can't update stage views actor
ClutterActor is on because it needs an allocation.
gru 15 21:33:35 wilk gnome-shell[1438]: Can't update stage views actor
Gjs_ui_workspaceThumbnail_WindowClone is on because it needs an allocation.
gru 15 21:33:47 wilk gnome-shell[1438]: Window manager warning: MetaSyncRing:
Sync object is not ready -- were events handled properly?
gru 15 21:33:58 wilk /usr/libexec/gdm-x-session[1322]: (II) event5  -
SteelSeries SteelSeries Sensei 310 eSports Mouse: SYN_DROPPED event - some
input events have been lost.
gru 15 21:33:58 wilk /usr/libexec/gdm-x-session[1322]: (EE) client bug: timer
event5 debounce: scheduled expiry is in the past (-877ms), your system is too
slow
gru 15 21:33:58 wilk /usr/libexec/gdm-x-session[1322]: (EE) client bug: timer
event5 debounce short: scheduled expiry is in the past (-890ms), your system is
too slow
gru 15 21:34:03 wilk /usr/libexec/gdm-x-session[1322]: (WW) NVIDIA: Wait for
channel idle timed out.
gru 15 21:34:04 wilk gnome-shell[1438]: Window manager warning: Failed to start
restart helper: Wykonanie procesu potomnego (child process failed)
„/usr/libexec/mutter-restart-helper” (Nie ma takiego pliku ani katalogu) się
nie powio> (no such file or directory)
gru 15 21:34:04 wilk gnome-shell[1438]: Window manager warning:
META_CURRENT_TIME used to choose focus window; focus window may not be correct.


This is not a keyboard/mouse issue. I used both on two other computers without
problems (gnome-shell 3.36) or on Intel integrated GPU. I also connected
a different keyboard to this machine with the same effect.


I switched to Xorg (WaylandEnable=false in /etc/gdm3/daemon.conf).
Performance is good, but there are issues unfortunately:

* gnome-terminal is lagging, I have to wait for a cursor to move
  sometimes, it can hang even for a second or two.
* I have to wait for a response from simple, usually immediate
  programs like htop, sometimes quitting is lagging to,
  quittin vim, etc.
* switching between terminal tabs also hangs sometimes for a moment.
* Not all programs are affected, browsers are fine mostly (I had
  one issue, but couldn't reproduce).


I checked KDE and as far as I know it is unaffected at all.
I only experience this on GNOME.


I tried xterm instead of gnome-terminal, it seems unaffected.
I wrote this report using reportbug-gtk and I had some problems
with lags (waiting seconds for Enter to kick in, etc.), same in gnome-terminal.
Scrolling sometimes hangs too.
libinput messages about compositor being slow are appearing in logs.


Similar problems were reported on GNOME's Gitlab, it looks
like only some programs are affected.
https://gitlab.gnome.org/GNOME/mutter/-/issues/1561
https://gitlab.gnome.org/GNOME/mutter/-/issues/1516


Someone in those issues pinpointed this behaviour to mutter,
that is why I'm reporting this as a mutter issue (I also
consulted #debian IRC before submitting here).


This issue basically makes terminal annoying to use, and
sometimes unusable. Stuttering and hanging is a no go
for a programmer or a system administrator (like me),
who depends on terminal (I use