887288 - Assertion failure: _mOwningThread.GetThread() == PR_GetCurrentThread() (nsDOMMemoryFileDataOwnerMemoryReporter not thread-safe)

Status: RESOLVED FIXED

(Core :: DOM: Workers, defect)

Description

[David Teller [:Yoric] - still alive but not very active]

Attachment: Stack trace

Reproductible assertion failure when attempting to migrate OS.File to the new module loader.

Comment 1

[David Teller [:Yoric] - still alive but not very active]

Marking as blocker, as I won't be able to land OS.File v2 until this is fixed.

Comment 2

[David Teller [:Yoric] - still alive but not very active]

Attachment: Marking nsDOMMemoryFileDataOwnerMemoryReporter thread-safe

Justin, would that be sufficient?

Comment 3

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

Comment on attachment 768362 [details] [diff] [review]
Marking nsDOMMemoryFileDataOwnerMemoryReporter thread-safe

Review of attachment 768362 [details] [diff] [review]:
-----------------------------------------------------------------

No, we need to make sure that calling nsDOMMemoryFile::DataOwner::EnsureMemoryReporterRegistered on multiple threads can't race too.

Comment 4

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

Comment on attachment 768362 [details] [diff] [review]
Marking nsDOMMemoryFileDataOwnerMemoryReporter thread-safe

Actually I take that back, we're already holding the lock in the only callsite.

r=me if you add an assertion to EnsureMemoryReporterRegistered that makes sure sDataOwnerMutex is held.

Comment 5

[David Teller [:Yoric] - still alive but not very active]

I don't really see how to do that. |StaticMutex| only seems to provide Lock/Unlock, which doesn't really help me check whether it is held.

Comment 6

[Justin Lebar (not reading bugmail)]

(In reply to David Rajchenbach Teller [:Yoric] <on workweek, will not follow bugzilla until July 1st> from comment #5)
> I don't really see how to do that. |StaticMutex| only seems to provide
> Lock/Unlock, which doesn't really help me check whether it is held.

I have a patch in my queue to add the necessary assertion; I'll post it in a bug.

Comment 7

[Justin Lebar (not reading bugmail)]

Comment on attachment 768362 [details] [diff] [review]
Marking nsDOMMemoryFileDataOwnerMemoryReporter thread-safe

The requisite blocking bug has landed; you can now add the assertion Kyle asked for.

Comment 8

[David Teller [:Yoric] - still alive but not very active]

Attachment: Marking nsDOMMemoryFileDataOwnerMemoryReporter thread-safe, v2

Comment 9

[David Teller [:Yoric] - still alive but not very active]

Try: https://tbpl.mozilla.org/?tree=Try&rev=e0179324a4a3

Comment 10

[David Teller [:Yoric] - still alive but not very active]

More try: https://tbpl.mozilla.org/?tree=Try&rev=04667e06324f

Comment 11

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/e21563063d60

Comment 12

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/e21563063d60