1500109 - Crash in audiounit_create_unit

Status: RESOLVED FIXED

(Core :: Audio/Video: cubeb, defect, P2)

Description

[Marcia Knous [:marcia]]

This bug was filed from the Socorro interface and is
report bp-264771fe-ade3-41f6-baf8-92fff0181005.
=============================================================

Seen while looking at Mac nightly crash stats: https://bit.ly/2q1316I. Crashes seem to go back to at least 20181005102516. Further down in the stack there appears to be some cubeb related signatures. There are also a handful of crashes in 63b13 and b14.

Top 10 frames of crashing thread:

0  @0x7fff57daa20a 
1 XUL google_breakpad::CrashGenerationClient::RequestDumpForException toolkit/crashreporter/google-breakpad/src/common/mac/MachIPC.mm:249
2 XUL google_breakpad::ExceptionHandler::WriteMinidumpWithException toolkit/crashreporter/breakpad-client/mac/handler/exception_handler.cc:382
3 XUL google_breakpad::ExceptionHandler::SignalHandler toolkit/crashreporter/breakpad-client/mac/handler/exception_handler.cc:628
4  @0x7fff57f71f59 
5  @0x114c0d28f 
6  @0x7fff57d0f1ad 
7  @0x7fff57cd71ab 
8 XUL audiounit_create_unit media/libcubeb/src/cubeb_audiounit.cpp:2032
9 XUL audiounit_setup_stream media/libcubeb/src/cubeb_audiounit.cpp:2553

=============================================================

Comment 1

[Alex Chronopoulos [:achronop]]

Probably the audiounit_setup_stream() failed at the output AU. The input AU which is initialized first has not cleared out the stm->input_unit is not null at the point of reconfigure. I would suggest to call the audiounit_close_stream(stm) between the 1st setup failure [1] and the reconfigure at [2].

[1] https://searchfox.org/mozilla-central/rev/eef79962ba73f7759fd74da658f6e5ceae0fc730/media/libcubeb/src/cubeb_audiounit.cpp#835
[2] https://searchfox.org/mozilla-central/rev/eef79962ba73f7759fd74da658f6e5ceae0fc730/media/libcubeb/src/cubeb_audiounit.cpp#840

Comment 2

[Nils Ohlmeier [:drno]]

Looks like something in 63 made this happening a lot more often then before. It seems the first crash was in 60.

Comment 3

[Jean-Yves Avenard [:jya]]

yes, it's an existing bug, but with BT headset we reset the device much more often now

Comment 4

[Alex Chronopoulos [:achronop]]

Fix up for review in cubeb repo:
https://github.com/kinetiknz/cubeb/pull/468

Comment 5

[Alex Chronopoulos [:achronop]]

(In reply to Nils Ohlmeier [:drno] from comment #2)
> Looks like something in 63 made this happening a lot more often then before.
> It seems the first crash was in 60.

Crashes in 60 are different call stack. This one is a regression from Bug 1489052 and we need to uplift the fix accordingly.

Comment 6

[Alex Chronopoulos [:achronop]]

Fixed by Bug 1501605.

Comment 7

[Liz Henry (:lizzard) (relman/hg->git project)]

Heading over to bug 1501605 to figure out if we may want uplift there.