FS#67684 - [pulseaudio] 13.0-3 dumping core during boot on kernel 5.8.3.arch1-1

Attached to Project: Arch Linux
Opened by kortewegdevries (kortewegdevries) - Sunday, 23 August 2020, 04:03 GMT
Last edited by Toolybird (Toolybird) - Friday, 22 September 2023, 08:13 GMT
Task Type Bug Report
Category Packages: Extra
Status Closed
Assigned To Jan Alexander Steffens (heftig)
Architecture All
Severity Low
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No

Details

Atleast 10-12 in a given journalctl -b -0:

Aug 23 03:28:13 systemd-coredump[2985]: Process 2983 (pulseaudio) of user 1000 dumped core.

Stack trace of thread 196:
#0 0x00007f536b17e55b personality (libc.so.6 + 0x10055b)
#1 0x00007f536af80f4c pa_reset_personality (libpulsecommon-13.0.so + 0x42f4c)
#2 0x000055eebe05c158 main (pulseaudio + 0x5158)
#3 0x00007f536b0a6152 __libc_start_main (libc.so.6 + 0x28152)
#4 0x000055eebe0609ee n/a (pulseaudio + 0x99ee)


gdb backtrace from coredumpctl:

Reading symbols from /usr/bin/pulseaudio...
(No debugging symbols found in /usr/bin/pulseaudio)
[New LWP 196]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/usr/lib/libthread_db.so.1".
Core was generated by `/usr/bin/pulseaudio --start --log-target=syslog'.
Program terminated with signal SIGSYS, Bad system call.
#0 0x00007f536b17e55b in personality () from /usr/lib/libc.so.6
#0 0x00007f536b17e55b in personality () from /usr/lib/libc.so.6
#1 0x00007f536af80f4c in pa_reset_personality () from /usr/lib/pulseaudio/libpulsecommon-13.0.so
#2 0x000055eebe05c158 in main ()

Versions:

libpulse 13.0-3
pulseaudio 13.0-3
pulseaudio-alsa 1:1.2.2-2
linux 5.8.3.arch1-1
linux-firmware 20200817.7a30af1-1
systemd 246.2-2

Steps to reproduce:

Not sure how to reproduce but according to coredumpctl I have logs of the same crash since Sun 2020-08-09 10:34:04 UTC, the day I started using this Arch system and around 1220 coredumps of pulseaudio.
This task depends upon

Closed by  Toolybird (Toolybird)
Friday, 22 September 2023, 08:13 GMT
Reason for closing:  Works for me
Additional comments about closing:  Old and stale. If still an issue with latest pkgs, please post a backtrace with debug info. Arch has support for debuginfod these days.
Comment by freswa (frederik) - Sunday, 23 August 2020, 15:51 GMT
Could you try pulseaudio from testing please?
Comment by kortewegdevries (kortewegdevries) - Monday, 24 August 2020, 08:34 GMT
Hi, @freswa I don't think it changed.

# journalctl -b -0

Aug 24 08:24:30 korte systemd-coredump[1081]: Process 1079 (pulseaudio) of user 1000 dumped core.

Stack trace of thread 42:
#0 0x00007fd41aa5b55b personality (libc.so.6 + 0x10055b)
#1 0x00007fd41a843c0c pa_reset_personality (libpulsecommon-13.99.so + 0x28c0c)
#2 0x000055b19ed42150 main (pulseaudio + 0x5150)
#3 0x00007fd41a983152 __libc_start_main (libc.so.6 + 0x28152)
#4 0x000055b19ed468fe n/a (pulseaudio + 0x98fe)

# coredumpctl info 1079

PID: 1079 (pulseaudio)
UID: 1000 (korte)
GID: 1000 (korte)
Signal: 31 (SYS)
Timestamp: Mon 2020-08-24 08:24:29 UTC (3min 1s ago)
Command Line: /usr/bin/pulseaudio --start --log-target=syslog
Executable: /usr/bin/pulseaudio
Control Group: /user.slice/user-1000.slice/session-2.scope
Unit: session-2.scope
Slice: user-1000.slice
Session: 2
Owner UID: 1000 (korte)
Boot ID: <>
Machine ID: <>
Hostname: korte
Storage: /var/lib/systemd/coredump/core.pulseaudio.1000.37ec76843c6b459aa9802deef02915d0.1079.1598257469000000000000.zst
Message: Process 1079 (pulseaudio) of user 1000 dumped core.

Stack trace of thread 42:
#0 0x00007fd41aa5b55b personality (libc.so.6 + 0x10055b)
#1 0x00007fd41a843c0c pa_reset_personality (libpulsecommon-13.99.so + 0x28c0c)
#2 0x000055b19ed42150 main (pulseaudio + 0x5150)
#3 0x00007fd41a983152 __libc_start_main (libc.so.6 + 0x28152)
#4 0x000055b19ed468fe n/a (pulseaudio + 0x98fe)

Backtrace:

Reading symbols from /usr/bin/pulseaudio...
(No debugging symbols found in /usr/bin/pulseaudio)
[New LWP 42]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/usr/lib/libthread_db.so.1".
Core was generated by `/usr/bin/pulseaudio --start --log-target=syslog'.
Program terminated with signal SIGSYS, Bad system call.
#0 0x00007fd41aa5b55b in personality () from /usr/lib/libc.so.6
(gdb) bt
#0 0x00007fd41aa5b55b in personality () from /usr/lib/libc.so.6
#1 0x00007fd41a843c0c in pa_reset_personality () from /usr/lib/pulseaudio/libpulsecommon-13.99.so
#2 0x000055b19ed42150 in main ()

Versions:

libpulse 13.99.1+67+g414644dff-1
pulseaudio 13.99.1+67+g414644dff-1
pulseaudio-alsa 1:1.2.2-2
pulseaudio-jack 13.99.1+67+g414644dff-1

Notably libseccomp was also updated...
Comment by Jan Alexander Steffens (heftig) - Monday, 24 August 2020, 23:46 GMT
What's your /proc/cmdline?
Comment by kortewegdevries (kortewegdevries) - Tuesday, 25 August 2020, 01:07 GMT
# cat /proc/cmdline

BOOT_IMAGE=/boot/vmlinuz-linux root=UUID=e81e5b9e-04f2-473c-92d1-db1ad91f2a00 rw audit=1 apparmor=1 lsm=apparmor,lockdown,yama nmi_watchdog=0 acpi_backlight=amdgpu_bl0 vt.global_cursor_default=0 loglevel=3 rd.systemd.show_status=auto rd.udev.log_priority=3 loglevel=3 quiet splash cryptdevice=UUID=b2b38cc5-0da1-43c5-a16e-2e39fb0f68be:luks-b2b38cc5-0da1-43c5-a16e-2e39fb0f68be root=/dev/mapper/luks-b2b38cc5-0da1-43c5-a16e-2e39fb0f68be nowatchdog
Comment by Jan Alexander Steffens (heftig) - Tuesday, 25 August 2020, 01:14 GMT
I suspect one of your enabled LSMs (probably AppArmor) is disallowing the syscall.
Comment by kortewegdevries (kortewegdevries) - Tuesday, 25 August 2020, 09:32 GMT
I'll remove the lsms and try again.Is it supposed to? I use apparmor elsewhere and I don't make any custom profiles other than setting up notify and audit.log.
Comment by Jan Alexander Steffens (heftig) - Tuesday, 25 August 2020, 09:38 GMT
I don't know anything about how to use AppArmor on Arch Linux, sorry.
Comment by kortewegdevries (kortewegdevries) - Tuesday, 25 August 2020, 09:42 GMT
Still dumping core, removed lsms, update-grub:

coredumpctl info 868
PID: 868 (pulseaudio)
UID: 1000 (korte)
GID: 1000 (korte)
Signal: 31 (SYS)
Timestamp: Tue 2020-08-25 09:37:05 UTC (3min 7s ago)
Command Line: /usr/bin/pulseaudio --start --log-target=syslog
Executable: /usr/bin/pulseaudio
Control Group: /user.slice/user-1000.slice/session-2.scope
Unit: session-2.scope
Slice: user-1000.slice
Session: 2
Owner UID: 1000 (korte)
Boot ID: <>
Machine ID: <>
Hostname: korte
Storage: /var/lib/systemd/coredump/core.pulseaudio.1000.94d5208fc16f42caa6466786256e6756.868.1598348225000000000000.zst
Message: Process 868 (pulseaudio) of user 1000 dumped core.

Stack trace of thread 42:
#0 0x00007fb76508655b personality (libc.so.6 + 0x10055b)
#1 0x00007fb764e88f4c pa_reset_personality (libpulsecommon-13.0.so + 0x42f4c)
#2 0x000055d57af5b158 main (pulseaudio + 0x5158)
#3 0x00007fb764fae152 __libc_start_main (libc.so.6 + 0x28152)
#4 0x000055d57af5f9ee n/a (pulseaudio + 0x99ee)
Comment by Jan Alexander Steffens (heftig) - Tuesday, 25 August 2020, 11:03 GMT
Is there something else you might have changed that affects how seccomp operates? pulseaudio.service sets LockPersonality but according to systemd code that means the call should return EPERM instead of triggering SIGSYS.
Comment by kortewegdevries (kortewegdevries) - Tuesday, 25 August 2020, 14:14 GMT
No I don't remember something that could've changed seccomp behavior. I'll try to start a VM and retrace my steps, to see if it happens also there...
Comment by kortewegdevries (kortewegdevries) - Saturday, 29 August 2020, 14:25 GMT
I've looked into it, turning on apparmor and audit registers this syscall (personality):

type=ANOM_ABEND msg=audit(1598707694.859:207): auid=1000 uid=1000 gid=1000 ses=2 subj==unconfined pid=3297 comm="pulseaudio" exe="/usr/bin/pulseaudio" sig=31 res=1

type=SECCOMP msg=audit(1598707696.575:214): auid=1000 uid=1000 gid=1000 ses=2 subj==unconfined pid=3305 comm="pulseaudio" exe="/usr/bin/pulseaudio" sig=31 arch=c000003e syscall=135 compat=0 ip=0x7f0bff7d355b code=0x0

Maybe this has something to do?
Comment by Buggy McBugFace (bugbot) - Tuesday, 08 August 2023, 19:11 GMT
This is an automated comment as this bug is open for more then 2 years. Please reply if you still experience this bug otherwise this issue will be closed after 1 month.

Loading...