Bug 209827 - CRASH in MediaPlayerPrivateMediaSourceAVFObjC::addAudioRenderer(), uncaught ObjC exception
Summary: CRASH in MediaPlayerPrivateMediaSourceAVFObjC::addAudioRenderer(), uncaught O...
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: New Bugs (show other bugs)
Version: WebKit Nightly Build
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Jer Noble
URL:
Keywords: InRadar
Depends on:
Blocks:
 
Reported: 2020-03-31 12:41 PDT by Jer Noble
Modified: 2020-04-01 14:15 PDT (History)
10 users (show)

See Also:

Attachments
Patch (5.82 KB, patch)
2020-03-31 12:46 PDT, Jer Noble 		no flags Details | Formatted Diff | Diff
Patch for landing (10.60 KB, patch)
2020-03-31 18:03 PDT, Jer Noble 		no flags Details | Formatted Diff | Diff
Patch for landing (10.56 KB, patch)
2020-04-01 11:57 PDT, Jer Noble 		no flags Details | Formatted Diff | Diff
Show Obsolete (2) View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Jer Noble 2020-03-31 12:41:39 PDT 
CRASH in MediaPlayerPrivateMediaSourceAVFObjC::addAudioRenderer(), uncaught ObjC exception
Comment 1 Jer Noble 2020-03-31 12:42:07 PDT 
<rdar://problem/61113080>
Comment 2 Radar WebKit Bug Importer 2020-03-31 12:42:20 PDT 
<rdar://problem/61120016>
Comment 3 Jer Noble 2020-03-31 12:46:55 PDT 
Created attachment 395081 [details]
Patch
Comment 4 Darin Adler 2020-03-31 12:51:50 PDT 
Comment on attachment 395081 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=395081&action=review

> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:903
> +                ASSERT_NOT_REACHED();

I don’t really like using assertions like this. As a project we intend to use assertions for things that must *never* happen, not for things that *normally* don’t happen.

The other cases in this patch seem to be truly "should never happen" cases, although I might be wrong about that.
Comment 5 Jer Noble 2020-03-31 12:57:45 PDT 
Comment on attachment 395081 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=395081&action=review

>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:903
>> +                ASSERT_NOT_REACHED();
> 
> I don’t really like using assertions like this. As a project we intend to use assertions for things that must *never* happen, not for things that *normally* don’t happen.
> 
> The other cases in this patch seem to be truly "should never happen" cases, although I might be wrong about that.

My goal here is to catch this case in layout tests; this scenario could easily cause layout test flakiness in a totally un-diagnosable way. If we upgrade a bot to a new OS and this ASSERTion starts getting hit and generates crashlogs, we can immediately trace the cause to a particular OS revision.

But maybe we can achieve this some other way, like through a WKTR or DRT only console log that would show up in the test .diff?
Comment 6 Darin Adler 2020-03-31 13:15:51 PDT 
(In reply to Jer Noble from comment #5)
> My goal here is to catch this case in layout tests

Got it!

> But maybe we can achieve this some other way, like through a WKTR or DRT only console log that would show up in the test .diff?

Yes, I think there is quite a bit of logging that shows up in regression test results. Let's check on that.
Comment 7 Jer Noble 2020-03-31 18:03:29 PDT 
Created attachment 395117 [details]
Patch for landing
Comment 8 Jer Noble 2020-04-01 11:57:25 PDT 
Created attachment 395191 [details]
Patch for landing
Comment 9 EWS 2020-04-01 14:15:23 PDT 
Committed r259363: <https://trac.webkit.org/changeset/259363>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 395191 [details].