1375873 - all tabs crash in nightly (thread '<unnamed>' panicked at 'SinkInfo contains invalid flags')

Status: RESOLVED FIXED

(Core :: Audio/Video: cubeb, defect, P1)

Description

[Peter Gerber (:pege)]

Attachment: full_terminal_output

User Agent: Mozilla/5.0 (Windows NT 6.1; rv:52.0) Gecko/20100101 Firefox/52.0
Build ID: 20100101

Steps to reproduce:

open the browser or a new tab


Actual results:

tab crashes:

thread '<unnamed>' panicked at 'SinkInfo contains invalid flags', /checkout/src/libcore/option.rs:794
stack backtrace:
   0: std::sys::imp::backtrace::tracing::imp::unwind_backtrace
   1: std::panicking::default_hook::{{closure}}
   2: std::panicking::rust_panic_with_hook
   3: std::panicking::begin_panic
   4: std::panicking::begin_panic_fmt
   5: core::panicking::panic_fmt
   6: core::option::expect_failed
   7: pulse::context::Context::get_sink_info_by_name::wrapped
   8: <unknown>
   9: <unknown>
  10: pa_pdispatch_run
  11: <unknown>
  12: <unknown>
  13: <unknown>
  14: <unknown>
  15: <unknown>
  16: pa_mainloop_dispatch
  17: pa_mainloop_iterate
  18: pa_mainloop_run
  19: <unknown>
  20: <unknown>
  21: start_thread
  22: __clone
Redirecting call to abort() to mozalloc_abort



Expected results:

tab shouldn't crash

Comment 1

[Peter Gerber (:pege)]

Forgot to mention the platform:

peter@ivy:~/src/mozilla-central$ lsb_release -a
No LSB modules are available.
Distributor ID: Debian
Description:    Debian GNU/Linux 9.0 (stretch)
Release:        9.0
Codename:       stretch
peter@ivy:~/src/mozilla-central$ uname -a
Linux ivy 4.11.6 #117 SMP PREEMPT Sat Jun 17 17:44:45 CEST 2017 x86_64 GNU/Linux
peter@pg:~/src/mozilla-central$ dpkg -l '*pulse*' |grep ii
ii  libpulse-mainloop-glib0:amd64 10.0-1       amd64        PulseAudio client libraries (glib support)
ii  libpulse0:amd64               10.0-1       amd64        PulseAudio client libraries
ii  libpulsedsp:amd64             10.0-1       amd64        PulseAudio OSS pre-load library
ii  pulseaudio                    10.0-1       amd64        PulseAudio sound server
ii  pulseaudio-module-gconf       10.0-1       amd64        GConf module for PulseAudio sound server
ii  pulseaudio-utils              10.0-1       amd64        Command line tools for the PulseAudio sound server


Tried to reproduce the issue on my home machine which is powered by the same OS, no luck so far.

Comment 2

[Ethan Glasser-Camp (:glasserc)]

I also just encountered this behavior on my work laptop.

> lsb_release -a
LSB Version:	:core-4.1-amd64:core-4.1-noarch:cxx-4.1-amd64:cxx-4.1-noarch:desktop-4.1-amd64:desktop-4.1-noarch:languages-4.1-amd64:languages-4.1-noarch:printing-4.1-amd64:printing-4.1-noarch
Distributor ID:	Fedora
Description:	Fedora release 25 (Twenty Five)
Release:	25
Codename:	TwentyFive

> dnf list installed '*pulse*'
Last metadata expiration check: 9 days, 0:12:22 ago on Wed Jun 14 16:01:16 2017.
Installed Packages
alsa-plugins-pulseaudio.x86_64            1.1.1-1.fc25         @fedora          
mpg123-plugins-pulseaudio.x86_64          1.23.8-4.fc25        @updates         
pulseaudio.x86_64                         10.0-2.fc25          @updates         
pulseaudio-debuginfo.x86_64               9.0-1.fc25           @fedora-debuginfo
pulseaudio-gdm-hooks.x86_64               10.0-2.fc25          @updates         
pulseaudio-libs.i686                      10.0-2.fc25          @updates         
pulseaudio-libs.x86_64                    10.0-2.fc25          @updates         
pulseaudio-libs-devel.x86_64              10.0-2.fc25          @updates         
pulseaudio-libs-glib2.x86_64              10.0-2.fc25          @updates         
pulseaudio-module-bluetooth.x86_64        10.0-2.fc25          @updates         
pulseaudio-module-x11.x86_64              10.0-2.fc25          @updates         
pulseaudio-utils.x86_64                   10.0-2.fc25          @updates

Comment 3

[Peter Gerber (:pege)]

I took a short look at the code, the thread appears to panic in Context::new() [1]. I'm not an expert but just based on the names of the flags in SinkFlags::try_from(), I'd say the wrong flags are used. I believe the pa_sink_flags [3] is what is needed (just based on the name SinkFlags) and not pa_source_flags. The two sets of flags are mostly overlapping with exception of PA_SINK_SET_FORMATS with value 0x0100U [5] where no there is no equivalent for PA_SOURCE_* [6]. I guess this flag is set on my work machine causing the error. The code was introduced by 5ff54dc9d0ad [7] / bug 1374494.

[1]: https://hg.mozilla.org/mozilla-central/file/tip/media/libcubeb/cubeb-pulse-rs/src/backend/context.rs#l123
[2]: https://hg.mozilla.org/mozilla-central/file/tip/media/libcubeb/cubeb-pulse-rs/pulse-rs/src/lib.rs#l414
[3]: https://www.freedesktop.org/software/pulseaudio/doxygen/def_8h.html#a20e0a15bebf78a29893f3b73ef7dadc0
[4]: https://www.freedesktop.org/software/pulseaudio/doxygen/def_8h.html#a6102010fffc6e49f6d378263e7b474b0
[5]: https://www.freedesktop.org/software/pulseaudio/doxygen/def_8h.html#a20e0a15bebf78a29893f3b73ef7dadc0ad3f643f079f57e16e3b272c948956ba4
[6]: https://www.freedesktop.org/software/pulseaudio/doxygen/def_8h.html#a6102010fffc6e49f6d378263e7b474b0
[7]: https://hg.mozilla.org/mozilla-central/rev/5ff54dc9d0ad

Comment 4

[Peter Gerber (:pege)]

Attachment: Bug 1375873 - fix panic in cubeb caused by PA_SINK_SET_FORMATS flag

Review commit: https://reviewboard.mozilla.org/r/152352/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/152352/

Comment 5

[skierpage]

This is happening to me with today's 64-bit Nightly too, on Fedora 25 KDE spin. Will Mozilla patch its local libcubeb or must the fix wait for a patch to https://github.com/djg/cubeb-pulse-rs/blob/dev/pulse-rs/src/lib.rs ?  Should this bug and/or review be assigned to Dan Glastonbury:kamidphish aka @djg on GitHub? I took the liberty of cc'ing him on this bug.

(Cool to see more Rust in Firefox O_O )

Comment 6

[u480271]

Comment on attachment 8880989 [details]
Bug 1375873 - fix panic in cubeb caused by PA_SINK_SET_FORMATS flag

https://reviewboard.mozilla.org/r/152352/#review157414

Thanks.  This is the correct fix for SinkFlags::try_from().  I had the change locally ready to upstream, but we can take this.

Comment 7

[u480271]

Comment on attachment 8880989 [details]
Bug 1375873 - fix panic in cubeb caused by PA_SINK_SET_FORMATS flag

https://reviewboard.mozilla.org/r/152352/#review157416

Comment 8

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/b52b0b9cd722
fix panic in cubeb caused by PA_SINK_SET_FORMATS flag r=kamidphish

Comment 10

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/b52b0b9cd722