#0 0xdcc71147 in ?? () from /home/desktop/.steam/debian-installation/ubuntu12_32/libaudio.so
I realise this path looks strange for an Arch system, but this is a test machine with my /home shared between multiple OSs, and Steam was originally installed via Debian.
As another data point, Steam seems to be working fine in an installation of Debian 12 on the same hardware (pipewire-pulse 0.3.65-2, libpulse0:i386 16.1+dfsg1-2+b1, kernel 6.1.8-1).
This sounds like https://github.com/ValveSoftware/steam-for-linux/issues/9060, which I haven't been able to reproduce or get useful backtraces from. I thought it was Arch specific so far though, but maybe not.
This sounds like https://github.com/ValveSoftware/steam-for-linux/issues/9060
Yes, it could well be the same crash.
I thought it was Arch specific
Yes, this is Arch too. My results, on a machine with multiple OSs sharing hardware and /home, are:
I don't think I can get Steam to upload minidumps from Arch, because it consistently crashes. But if I run it on Arch, let it crash, reboot to Debian and re-launch Steam from the same $HOME, would that upload a crash dump that would be useful to you? If yes, I'll try that tomorrow.
It looks like Steam beta 2023-02-20 (1676935611) and stable version 2023-02-14 (1676336721) are no longer crashing on my Arch system.
Since 1676336721 is one of the versions that was crashing before, I assume this means the crash I've been experiencing must have been something to do with an OS library and was fixed in Arch. Pipewire was upgraded to 1:0.3.66-2 today, so maybe it's that.
Still reproducible in July 2026 with a new trigger device. Same backtrace as OP — segfault in libaudio.so called from pa_pdispatch_run in the card-info reply path.
System:
1782866176 (updater built Jun 24 2026), steam 1.0.0.85ubuntu12_32/steam segfaults seconds after spawning steamwebhelper, 100% reproducible. Crash thread:#0 libaudio.so + 0x4d942
#1 libpulse.so.0 + 0x1a1bf
#2 libpulsecommon-1.1.so + 0x3225d
#3 pa_pdispatch_run (libpulsecommon-1.1.so + 0x32660)
#4 libpulse.so.0 + 0xd4dd
Uploaded minidumps: bp-a6516fea-8856-47a4-be09-2fd112260704, bp-a85b2be3-0ee7-4c63-ab3f-b00902260704
Bisection: crash occurs iff the Scarlett's pro-audio profile card is visible to the client. pactl set-card-profile alsa_card.usb-Focusrite_Scarlett_2i2_4th_Gen_S27YCMT52ABC24-00 off → steam starts fine; re-enable → crash. All other cards (AMD GPU HDMI, onboard, webcam) irrelevant.
Workaround (keeps pro-audio for all other clients): pipewire-pulse per-client quirk —
# /etc/pipewire/pipewire-pulse.conf.d/99-steam-libaudio.conf
pulse.rules = [
{ matches = [ { application.process.binary = "steam" } ]
actions = { quirks = [ force-s16-info ] } }
]
Suggests libaudio's card/format introspection can't handle non-S16 sample formats in pro-audio profile info. Given the bundled libpulse is 1.1-era, an updated bundle or bounds-checking in the card-info callback would fix the class of crash.
Still reproducible in July 2026 with a new trigger device
The issue that I was experiencing was fixed, so please open a separate issue with your own information: anything that I said here is not necessarily relevant any more.
The fact that you experienced the same symptom as me (a segfault during Steam startup with libpulse calling into libaudio) does not mean that you were experiencing it for the same reason.
Your system information
Steps for reproducing this issue:
Run
steamfrom a terminal prompt (I used Konsole).Please describe your issue in as much detail as possible:
Steam crashes shortly after startup, before the main UI has come up, with a segmentation fault in
libaudio.sowhich was called from the PulseAudio client thread. Using Arch'slib32-libpulseversion16.1-1, I get this backtrace:If I remove
lib32-libpulseand use the Steam Runtime's (very old) libpulse, I get a similar crash, but with less detail shown in gdb.It was running successfully from Debian 11 on the same hardware earlier today.