Closed Bug 1435960 Opened 7 years ago Closed 6 years ago

nsNSSComponent::ShutdownNSS() blocks the shutdown of other components

Categories

(Core :: Security: PSM, defect)

58 Branch
defect
Not set
normal

Tracking

()

RESOLVED DUPLICATE of bug 1435343

People

(Reporter: baku, Unassigned)

References

(Blocks 2 open bugs)

Details

Crash Data

https://crash-stats.mozilla.com/report/index/e0dea993-bf09-4bc4-bfe1-119860180206#allthreads
https://crash-stats.mozilla.com/report/index/fae67c93-de78-4229-a591-9a28a0180206#allthreads

here 2 crash reports where nsNSSComponent::ShutdownNSS() blocks the main-thread doing something when xpcom-shutdown notification is received. Doing this, other components do not receive the same notifications (for instance workerinternals::RuntimeService) and Fx ends up crashing for timeout.

I don't know in which component I should file this bug.
Looks like we try to destroy the session cache while still holding the SSL3 RWLock somewhere.
Component: Security → Security: PSM
Blocks: 1437575
Blocks: 1405290
Crash Signature: [@ mozilla::dom::workerinternals::RuntimeService::CrashIfHanging ]
Franziskus - where else could we be holding the SSL3 RWLock in those crash reports? None of the other stacks are in nss. Could this just be a bug in NSS itself?
Flags: needinfo?(franziskuskiefer)
Hm, recent logs for that signature look very different from the two linked in comment 0. I didn't find any other logs that would indicate that this is actually a PSM or NSS issue.
The only references to NSS are actually NSPR locks. Looks more like a worker locking issue now (bug 1435343).
Flags: needinfo?(franziskuskiefer)
Yeah I'm not seeing any PSM or NSS hang-crashes with this signature. I'll duplicate this to bug 1435343 for now - we can reopen if this comes back or if fixing that bug doesn't fix this.
Status: NEW → RESOLVED
Closed: 6 years ago
Resolution: --- → DUPLICATE
Blocks: 1633342
No longer blocks: 1633342
You need to log in before you can comment on or make changes to this bug.