Acknowledgement sent
to Timo Weingärtner <[email protected]>:
New Bug report received and forwarded. Copy sent to Thorsten Glaser <[email protected]>.
(Sat, 14 Sep 2024 21:09:02 GMT) (full text, mbox, link).
Subject: E: /usr/bin/ssh-askpass[39]: too many open files (3 -> 212): Success
Date: Sat, 14 Sep 2024 22:52:27 +0200
Package: kwalletcli
Version: 3.03-1+b1
Severity: normal
File: /usr/bin/kwalletaskpass
$ ssh-keygen -t ed25519 -f /tmp/test
$ ssh-add /tmp/test < /dev/null
E: /usr/bin/ssh-askpass[39]: too many open files (3 -> 212): Success
$ echo $?
1
key is not added to the agent
-- System Information:
Debian Release: trixie/sid
APT prefers testing
APT policy: (900, 'testing'), (900, 'stable'), (800, 'unstable')
Architecture: amd64 (x86_64)
Kernel: Linux 6.10.6-amd64 (SMP w/4 CPU threads; PREEMPT)
Locale: LANG=de_DE.UTF-8, LC_CTYPE=de_DE.UTF-8 (charmap=UTF-8), LANGUAGE=en_US
Shell: /bin/sh linked to /usr/bin/dash
Init: systemd (via /run/systemd/system)
LSM: AppArmor: enabled
Versions of packages kwalletcli depends on:
ii libc6 2.40-2
ii libgcc-s1 14.2.0-3
ii libkf5coreaddons5 5.115.0-2
ii libkf5i18n5 5.115.1-2+b1
ii libkf5wallet-bin 5.115.0-4
ii libkf5wallet5 5.115.0-4
ii libqt5core5t64 5.15.13+dfsg-4
ii libqt5widgets5t64 5.15.13+dfsg-4
ii libstdc++6 14.2.0-3
ii mksh 59c-39
Versions of packages kwalletcli recommends:
ii gpg-agent [gnupg-agent] 2.2.43-8+b1
ii kwalletmanager 4:22.12.3-1+b1
ii openssh-client 1:9.8p1-8
ii pinentry-curses [pinentry] 1.2.1-4+b1
ii pinentry-qt [pinentry] 1.2.1-4+b1
kwalletcli suggests no packages.
-- no debconf information
Subject: Re: Bug#1081794: E: /usr/bin/ssh-askpass[39]: too many open files
(3 -> 212): Success
Date: Sat, 14 Sep 2024 21:45:58 +0000 (UTC)
Hi Timo,
>$ ssh-keygen -t ed25519 -f /tmp/test
>$ ssh-add /tmp/test < /dev/null
>E: /usr/bin/ssh-askpass[39]: too many open files (3 -> 212): Success
>$ echo $?
>1
>
>key is not added to the agent
this is quite puzzling.
First guess: is $DISPLAY set and points to a usable X11 server?
Second guess: /usr/bin/pinentry points to a real one, not to
pinentry-kwallet, right? (Otherwise, there can be recursion.)
Third… please put “set -xv” in the top of the script and retry,
for debugging. Out of ideas, otherwise.
Thanks,
//mirabilos
--
15:39⎜«mika:#grml» mira|AO: "mit XFree86® wär’ das nicht passiert" - muhaha
15:48⎜<thkoehler:#grml> also warum machen die xorg Jungs eigentlich alles
kaputt? :) 15:49⎜<novoid:#grml> thkoehler: weil sie als Kinder nie den
gebauten Turm selber umschmeissen durften? -- ~/.Xmodmap wonders…
Acknowledgement sent
to Timo Weingärtner <[email protected]>:
Extra info received and forwarded to list. Copy sent to Thorsten Glaser <[email protected]>.
(Sun, 15 Sep 2024 09:39:02 GMT) (full text, mbox, link).
reassign 1081794 src:mesa
tags 1081794 + upstream
retitle 1081794 file descriptor leak mesa_shader_cache_db
severity 1081794 important
Hallo,
14.09.24 23:45 Thorsten Glaser:
> First guess: is $DISPLAY set and points to a usable X11 server?
Yes, this is a pure X11 session, no wayland.
> Second guess: […]
> Third… […]
Did all that, including strace'ing, was puzzled by a jump in FD numbers there.
> Out of ideas, otherwise.
Further investigation shows FDs 10-109 and 110-209 filled with:
.cache/mesa_shader_cache_db/part0/mesa_cache.db
.cache/mesa_shader_cache_db/part0/mesa_cache.idx
through
.cache/mesa_shader_cache_db/part49/mesa_cache.db
.cache/mesa_shader_cache_db/part49/mesa_cache.idx
When running in xterm (started from krunner) instead of konsole, there is only
the first such run of cache files and mksh doesn't complain.
These files should definitely be opened CLOEXEC.
https://bbs.archlinux.org/viewtopic.php?id=298957
So while there might be a problem in mksh with many FDs, I'm reassigning to
src:mesa. Because of https://bbs.archlinux.org/viewtopic.php?id=298957 I
assume an upstream bug.
Since this adds 100 FDs (at least in my case) per exec() of a mesa-using
binary in the call chain I'm bumbping severity.
Grüße
Timo
Changed Bug title to 'file descriptor leak mesa_shader_cache_db' from 'E: /usr/bin/ssh-askpass[39]: too many open files (3 -> 212): Success'.
Request was from Timo Weingärtner <[email protected]>
to [email protected].
(Sun, 15 Sep 2024 10:39:02 GMT) (full text, mbox, link).
Acknowledgement sent
to Guillem Jover <[email protected]>:
Extra info received and forwarded to list. Copy sent to Debian X Strike Force <[email protected]>.
(Sat, 09 Nov 2024 03:03:01 GMT) (full text, mbox, link).
Subject: Re: Bug#1081794: E: /usr/bin/ssh-askpass[39]: too many open files (3
-> 212): Success
Date: Sat, 9 Nov 2024 04:00:34 +0100
Control: forwarded -1 https://gitlab.freedesktop.org/mesa/mesa/-/issues/11810
Hi!
On Sun, 2024-09-15 at 11:14:20 +0200, Timo Weingärtner wrote:
> reassign 1081794 src:mesa
> tags 1081794 + upstream
> retitle 1081794 file descriptor leak mesa_shader_cache_db
> severity 1081794 important
> These files should definitely be opened CLOEXEC.
>
> https://bbs.archlinux.org/viewtopic.php?id=298957
>
> So while there might be a problem in mksh with many FDs, I'm reassigning to
> src:mesa. Because of https://bbs.archlinux.org/viewtopic.php?id=298957 I
> assume an upstream bug.
>
> Since this adds 100 FDs (at least in my case) per exec() of a mesa-using
> binary in the call chain I'm bumbping severity.
This has been reported upstream and now fixed in git. There's the
other related upstream report with the excessive amount of file
descriptors used:
https://gitlab.freedesktop.org/mesa/mesa/-/issues/11776
It would be nice to get the fix for that backported into the Debian
package before the next mesa release happens upstream.
Thanks,
Guillem
Debbugs is free software and licensed under the terms of the GNU General
Public License version 2. The current version can be obtained
from https://bugs.debian.org/debbugs-source/.