1962022 - Crash in [@ mozilla::MozPromise<T>::ThenValueBase::ResolveOrRejectRunnable::~ResolveOrRejectRunnable]

Status: REOPENED

(Toolkit :: Startup and Profile System, defect)

Description

[Mayank Bansal]

Crash report: https://crash-stats.mozilla.org/report/index/c6c7fcac-02b3-4f8c-a278-127250250423

MOZ_CRASH Reason:

MozPromise::ThenValue created from 'AsyncFlush' destroyed without being either disconnected, resolved, or rejected (dispatchRv: not dispatched)

Top 10 frames:

0  xul.dll  MOZ_Crash(char const*, int, char const*)  mfbt/Assertions.h:381
0  xul.dll  mozilla::MozPromise<ProcessInfo, nsresult, 0>::ThenValueBase::AssertIsDead()  xpcom/threads/MozPromise.h:527
1  xul.dll  mozilla::MozPromise<bool, nsresult, 0>::AssertIsDead()  xpcom/threads/MozPromise.h:1253
2  xul.dll  mozilla::MozPromise<nsTAutoStringN<char16_t, 64>, nsresult, 0>::ThenValueBase...  xpcom/threads/MozPromise.h:482
3  xul.dll  mozilla::MozPromise<nsTAutoStringN<char16_t, 64>, nsresult, 0>::ThenValueBase...  xpcom/threads/MozPromise.h:480
4  xul.dll  mozilla::Runnable::Release()  xpcom/threads/nsThreadUtils.cpp:66
4  xul.dll  IdleRunnableWrapper::Release()  xpcom/threads/nsThreadUtils.cpp:151
5  xul.dll  mozilla::RefPtrTraits<nsIRunnable>::Release(nsIRunnable*)  mfbt/RefPtr.h:49
5  xul.dll  nsCOMPtr<nsIRunnable>::~nsCOMPtr()  xpcom/base/nsCOMPtr.h:344
5  xul.dll  mozilla::TaskQueue::Dispatch(already_AddRefed<nsIRunnable>, unsigned int)  xpcom/threads/TaskQueue.h:80

Been getting this crash since the last few days... Also seeing pages randomly stop loading on Firefox while Chrome loads them perfectly. Also seeing many instances where firefox keeps on running in the background when you close it.

Comment 1

[Mayank Bansal]

See also https://bugzilla.mozilla.org/show_bug.cgi?id=1824294#c17, where padenot suggested the crash is around the ToolkitProfileService.

Comment 2

[BugBot [:suhaib / :marco/ :calixte]]

The bug is linked to a topcrash signature, which matches the following criterion:

• Top 10 AArch64 and ARM crashes on nightly

For more information, please visit BugBot documentation.

Comment 3

[Dave Townsend [:mossop]]

The crash reason certainly indicates nsToolkitProfileService::AsyncFlush however the stack ends at MOZ_CRASH being called from MozPromise<ProcessInfo, nsresult, 0>::ThenValueBase::AssertIsDead() which would suggest that it is this promise and I don't see anything cancelling that promise if the global goes out of scope.

Am I misunderstanding this Paul?

Comment 4

[Mayank Bansal]

Attachment: about:support

Comment 5

[Paul Adenot (:padenot)]

I think we need to add something that handles the process going away indeed.

Displaying all threads in the crash report, we can see that we're in shutdown: NS_ShutdownXPCOM is present on the stack of the main thread.

I don't immediately see anything that listen for shutdown events here, but it's one way of solving this. A common pattern is to stash the promises on the instance (or globally if static), and reject all of them in a loop when entering shutdown.

Comment 6

[Dave Townsend [:mossop]]

Attachment: Bug 1962022: Disconnect system info promises when the calling global is disconnected. r=chutten!

The crash reason from bug 1962022 points to promises in the profile service, but the stacks end
in promises in nsSystemInfo. I'm not sure why this is the case but none of the promises in
nsSystemInfo are set to disconnect when their calling global goes away so this seems like a useful
improvement nonetheless. Once this lands we can watch the crash frequency to see if it ends up
resolving the bug or not.

Comment 7

[Mayank Bansal]

FWIW, since i filed this bug, i have not seen this crash.

Comment 8

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/87d4d529bf2c
Disconnect system info promises when the calling global is disconnected. r=chutten

Comment 9

[Serban Stanca [:SerbanS]]

https://hg.mozilla.org/mozilla-central/rev/87d4d529bf2c

Comment 10

[Dave Townsend [:mossop]]

(In reply to Mayank Bansal from comment #7)

FWIW, since i filed this bug, i have not seen this crash.

We're still getting new reports as recently as yesterday so it's not just you. I'm going to hold this bug open and check back in a bit to see whether the crashes stop.

Comment 11

[Dave Townsend [:mossop]]

Ok, definitely still not fixed

Comment 12

[BugBot [:suhaib / :marco/ :calixte]]

The bug is linked to a topcrash signature, which matches the following criterion:

• Top 20 desktop browser crashes on beta (startup)

For more information, please visit BugBot documentation.

Comment 13

[BugBot [:suhaib / :marco/ :calixte]]

The severity field is not set for this bug.
:mossop, could you have a look please?

For more information, please visit BugBot documentation.