Bug#1051418: obs-studio: clicking on an xcomposite window source makes obs segfault
Rhonda D'Vine
Version: 29.1.3+dfsg-2
Severity: important
Hi,
after upgrading from 29.0.2+dfsg-1+b1 to 29.1.3+dfsg-2 (bookworm to trixie) I
couldn't work with the xcomposite window source anymore. In fact, a
pre-existing scene that was set up to use that source makes obs segfault
immediately when I click the entry in the sources tree, not even able to remove
it and finally switch to a pipewire window source (which has its other quirks).
I'm on gnome with wayland, I guess that is needed information to reproduce it.
When starting obs in the console, the only line that is printed after clicking
on the entry is segfault.
Thanks,
Rhonda
-- System Information:
Debian Release: trixie/sid
APT prefers stable-updates
APT policy: (500, 'stable-updates'), (500, 'stable-security'), (500, 'testing'), (500, 'stable')
Architecture: amd64 (x86_64)
Foreign Architectures: i386
Kernel: Linux 6.4.0-4-amd64 (SMP w/8 CPU threads; PREEMPT)
Kernel taint flags: TAINT_PROPRIETARY_MODULE, TAINT_OOT_MODULE, TAINT_UNSIGNED_MODULE
Locale: LANG=de_AT.UTF-8, LC_CTYPE=de_AT.UTF-8 (charmap=UTF-8), LANGUAGE=de_AT:de
Shell: /bin/sh linked to /usr/bin/dash
Init: systemd (via /run/systemd/system)
LSM: AppArmor: enabled
Versions of packages obs-studio depends on:
ii libavcodec60 7:6.0-6
ii libavdevice60 7:6.0-6
ii libavformat60 7:6.0-6
ii libavutil58 7:6.0-6
ii libc6 2.37-7
ii libcurl3-gnutls 8.2.1-1
ii libfontconfig1 2.14.2-4
ii libfreetype6 2.13.2+dfsg-1
ii libgcc-s1 13.2.0-2
ii libjansson4 2.14-2
ii libluajit-5.1-2 2.1.0~beta3+git20220320+dfsg-4.1
ii libmbedcrypto7 2.28.3-1
ii libmbedtls14 2.28.3-1
ii libmbedx509-1 2.28.3-1
ii libobs0 29.1.3+dfsg-2
ii libpci3 1:3.10.0-2
ii libpulse0 16.1+dfsg1-2+b1
ii libpython3.11 3.11.5-3
ii libqt5core5a 5.15.10+dfsg-3
ii libqt5gui5 5.15.10+dfsg-3
ii libqt5network5 5.15.10+dfsg-3
ii libqt5svg5 5.15.10-2
ii libqt5widgets5 5.15.10+dfsg-3
ii libqt5xml5 5.15.10+dfsg-3
ii librist4 0.2.7+dfsg-1
ii libspeexdsp1 1.2.1-1
ii libsrt1.5-openssl 1.5.2-1
ii libstdc++6 13.2.0-2
ii libswscale7 7:6.0-6
ii libudev1 254.1-2
ii libv4l-0 1.24.1-3
ii libva-drm2 2.19.0-1
ii libva2 2.19.0-1
ii libx11-6 2:1.8.6-1
ii libx264-164 2:0.164.3095+gitbaee400-3+b1
ii libxcb-composite0 1.15-1
ii libxcb-randr0 1.15-1
ii libxcb-shm0 1.15-1
ii libxcb-xfixes0 1.15-1
ii libxcb-xinerama0 1.15-1
ii libxcb1 1.15-1
ii python3 3.11.4-5+b1
ii python3.11 3.11.5-3
Versions of packages obs-studio recommends:
ii obs-plugins 29.1.3+dfsg-2
Versions of packages obs-studio suggests:
ii pkexec 123-1
ii policykit-1 123-1
ii v4l2loopback-dkms 0.12.7-2
-- no debconf information
--
Fühlst du dich mutlos, fass endlich Mut, los |
Fühlst du dich hilflos, geh raus und hilf, los | Wir sind Helden
Fühlst du dich machtlos, geh raus und mach, los | 23.55: Alles auf Anfang
Fühlst du dich haltlos, such Halt und lass los |
tha...@t-online.de
Same story on xorg.
-- System Information:
Debian Release: trixie/sid
APT prefers oldoldstable-updates
APT policy: (500, 'oldoldstable-updates'), (500, 'oldoldstable'), (500, 'unstable'), (1, 'experimental')
Architecture: amd64 (x86_64)
Kernel: Linux 6.5.0-2-amd64 (SMP w/12 CPU threads; PREEMPT)
Kernel taint flags: TAINT_FIRMWARE_WORKAROUND
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8), LANGUAGE=en_US:en
Shell: /bin/sh linked to /usr/bin/dash
Init: systemd (via /run/systemd/system)
LSM: AppArmor: enabled
Versions of packages obs-studio depends on:
ii libavcodec60 7:6.0-7+b1
ii libavdevice60 7:6.0-7+b1
ii libavformat60 7:6.0-7+b1
ii libavutil58 7:6.0-7+b1
ii libc6 2.37-12
ii libcurl3-gnutls 8.3.0-3
ii libfontconfig1 2.14.2-6
ii libfreetype6 2.13.2+dfsg-1
ii libgcc-s1 13.2.0-5
ii libjansson4 2.14-2
ii libluajit-5.1-2 2.1.0~beta3+git20220320+dfsg-4.1
ii libmbedcrypto7 2.28.5-1
ii libmbedtls14 2.28.5-1
ii libmbedx509-1 2.28.5-1
ii libobs0 29.1.3+dfsg-2
ii libpci3 1:3.10.0-2
ii libpulse0 16.1+dfsg1-2+b1
ii libpython3.11 3.11.6-3
ii libqt5core5a 5.15.10+dfsg-3
ii libqt5gui5 5.15.10+dfsg-3
ii libqt5network5 5.15.10+dfsg-3
ii libqt5svg5 5.15.10-2
ii libqt5widgets5 5.15.10+dfsg-3
ii libqt5xml5 5.15.10+dfsg-3
ii librist4 0.2.8+dfsg+really0.2.7+dfsg-1
ii libspeexdsp1 1.2.1-1
ii libsrt1.5-openssl 1.5.3-1
ii libstdc++6 13.2.0-5
ii libswscale7 7:6.0-7+b1
ii libudev1 254.5-1
ii libv4l-0 1.24.1-4
ii libva-drm2 2.20.0-2
ii libva2 2.20.0-2
ii libx11-6 2:1.8.7-1
ii libx264-164 2:0.164.3095+gitbaee400-3+b1
ii libxcb-composite0 1.15-1
ii libxcb-randr0 1.15-1
ii libxcb-shm0 1.15-1
ii libxcb-xfixes0 1.15-1
ii libxcb-xinerama0 1.15-1
ii libxcb1 1.15-1
ii python3 3.11.4-5+b1
ii python3.11 3.11.6-3
Versions of packages obs-studio recommends:
ii obs-plugins 29.1.3+dfsg-2
Versions of packages obs-studio suggests:
ii pkexec 123-1
ii policykit-1 123-1
pn v4l2loopback-dkms <none>
-- no debconf information
...
info: - source: 'Window Capture (Xcomposite)' (xcomposite_input)
info: ------------------------------------------------
[Thread 0x7ffeebfff6c0 (LWP 1793192) exited]
[New Thread 0x7ffeebfff6c0 (LWP 1793194)]
[New Thread 0x7fff0cff96c0 (LWP 1793195)]
[New Thread 0x7fff0ffff6c0 (LWP 1793196)]
[Thread 0x7fff15dff6c0 (LWP 1793191) exited]
info: adding 21 milliseconds of audio buffering, total audio buffering is now 21 milliseconds (source: Desktop Audio)
Thread 1 "obs" received signal SIGSEGV, Segmentation fault.
__strcmp_avx2 () at ../sysdeps/x86_64/multiarch/strcmp-avx2.S:287
287 ../sysdeps/x86_64/multiarch/strcmp-avx2.S: No such file or directory
Brian Bostwick
> Sending in plain text so it is readable:
>
>
> I'm seeing the same bug in the latest Debian testing updated this
> morning. Specifically, if I have the following github page open in my
> browser clicking the xcomposite window source cases a seg fault as
> reported.
>
>
> Â Â https://github.com/pjreddie/darknet/issues/553
>
>
> This results in a window name of
>
>
>   "Error on make: 'cuda_runtime.h: No such file or directory' · Issue #553 · pjreddie/darknet - Brave"
>
>
> In xcompcap_props(), the name_lower variable has zero values in the
> struct which is then pushed to window_strings.array. I can't currently
> step into dstr_to_lower().
>
>
> The following shows the array entry in window_strings.array:
>
>
> Â Â $18 = {
> Â Â name_lower = {array = 0x0, len = 0, capacity = 0},
> Â Â name = {
>     array = 0x5555570b88e0 "Error on make: 'cuda_runtime.h: No such file or directory' · Issue #553 · pjreddie/darknet - Brave",
> Â Â Â Â len = 100,
> Â Â Â Â capacity = 101
> Â Â },
> Â Â desc = {
>     array = 0x555555c24560 "54525967\r\nError on make: 'cuda_runtime.h: No such file or directory' · Issue #553 · pjreddie/darknet - Brave\r\nbrave-browser",
> Â Â Â Â len = 125,
> Â Â Â Â capacity = 126
> Â Â }
> Â Â }
>
>
> The following is the gdb backtrace with frame 11 being the frame where
> the above was observed.
>
>
> (gdb) bt
> #0Â __strcmp_avx2 () at ../sysdeps/x86_64/multiarch/strcmp-avx2.S:283
> #1Â 0x00007ffff3e5de65 in msort_with_tmp (p=p@entry=0x7fffffffc510, b=b@entry=0x7fffffffc448, n=n@entry=3) at ./stdlib/msort.c:123
> #2Â 0x00007ffff3e5ddb1 in msort_with_tmp (n=3, b=0x7fffffffc448, p=0x7fffffffc510) at ./stdlib/msort.c:44
> #3Â msort_with_tmp (p=p@entry=0x7fffffffc510, b=b@entry=0x7fffffffc430, n=n@entry=6) at ./stdlib/msort.c:53
> #4Â 0x00007ffff3e5dd94 in msort_with_tmp (n=6, b=0x7fffffffc430, p=0x7fffffffc510) at ./stdlib/msort.c:44
> #5Â msort_with_tmp (p=p@entry=0x7fffffffc510, b=b@entry=0x7fffffffc430, n=n@entry=12) at ./stdlib/msort.c:52
> #6Â 0x00007ffff3e5ddb1 in msort_with_tmp (n=12, b=0x7fffffffc430, p=0x7fffffffc510) at ./stdlib/msort.c:44
> #7Â msort_with_tmp (p=p@entry=0x7fffffffc510, b=0x7fffffffc3d8, n=n@entry=23) at ./stdlib/msort.c:53
> #8Â 0x00007ffff3e5e23b in msort_with_tmp (n=23, b=<optimized out>, p=0x7fffffffc510) at ./stdlib/msort.c:44
> #9Â __GI___qsort_r (b=b@entry=0x5555570e2320, n=n@entry=23, s=s@entry=72, cmp=cmp@entry=0x7fffe1fdaac0 <cmp_wi>, arg=arg@entry=0x0) at ./stdlib/msort.c:253
> #10 0x00007ffff3e5e3c8 in __GI_qsort (b=b@entry=0x5555570e2320, n=n@entry=23, s=s@entry=72, cmp=cmp@entry=0x7fffe1fdaac0 <cmp_wi>) at ./stdlib/msort.c:307
> #11 0x00007fffe1fdbcb5 in xcompcap_props (unused=<optimized out>) at ./plugins/linux-capture/xcomposite-input.c:750
> #12 0x00007ffff65629f6 in obs_source_properties () at /lib/x86_64-linux-gnu/libobs.so.0
> #13 0x000055555564c3d5 in SourceToolbar::SourceToolbar(QWidget*, OBSSafeRef<obs_source*, &obs_source_get_ref, &obs_source_release>) (source=..., parent=0x555555ea89a0, this=0x555556074fd0) at ./libobs/obs.hpp:103
IOhannes m zmoelnig
Version: 30.0.2+dfsg~
Followup-For: Bug #1051418
Unfortunately I can confirm the bug at question with the (not yet uploaded)
obs-studio 30.0.2.
the OBS shipped by flatpak does not have this problem, so it appears to be
Debian packaging problem.
Erich Schubert
It makes OBS almost unusable for me, because any click on a window
source crashes obs. You cannot even remove these sources!
Recording an entire screen still works, but not recording single windows.
The earlier message by Michael Neilly suggests that maybe this Debian
patch is causing the problem:
https://sources.debian.org/src/obs-studio/30.0.2%2Bdfsg-2/debian/patches/0013-Fix-comparision-if-char-is-unsigned.patch/
Note that it diverges from the linked upstream patch
https://github.com/obsproject/obs-studio/pull/9184/files
Can you refresh this patch?
Regards,
Erich Schubert
Erich Schubert
thanks
Removing the bad patch
debian/patches/0013-Fix-comparision-if-char-is-unsigned.patch
and rebuilding the package fixes this bug.