Failure observed on several machines, first build it has been reported for is Nightly 84.0a1 20201021213007

Pushlog between previous build and that one: https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=d8861d51b01e9489672f998648d67662a60a8b3a&tochange=7d6d66062e843a75b7aafb4ec0ae2dff355755e7

Crash report: https://crash-stats.mozilla.org/report/index/63e6e42c-4666-4c82-b306-678d10201022

Reason: EXCEPTION_ACCESS_VIOLATION_WRITE

Top 10 frames of crashing thread:

0 xul.dll js::GlobalHelperThreadState::finishParseTaskCommon js/src/vm/HelperThreads.cpp:1918
1 xul.dll js::GlobalHelperThreadState::finishSingleParseTask js/src/vm/HelperThreads.cpp:2039
2 xul.dll nsJSUtils::ExecutionContext::JoinDecode dom/base/nsJSUtils.cpp:299
3 xul.dll mozilla::dom::ScriptLoader::EvaluateScript dom/script/ScriptLoader.cpp:2937
4 xul.dll mozilla::dom::ScriptLoader::ProcessRequest dom/script/ScriptLoader.cpp:2535
5 xul.dll mozilla::dom::`anonymous namespace'::NotifyOffThreadScriptLoadCompletedRunnable::Run dom/script/ScriptLoader.cpp:2235
6 xul.dll mozilla::SchedulerGroup::Runnable::Run xpcom/threads/SchedulerGroup.cpp:146
7 xul.dll mozilla::TaskController::DoExecuteNextTaskOnlyMainThreadInternal xpcom/threads/TaskController.cpp:515
8 xul.dll mozilla::detail::RunnableFunction<`lambda at /builds/worker/checkouts/gecko/xpcom/threads/TaskController.cpp:85:7'>::Run xpcom/threads/nsThreadUtils.h:577
9 xul.dll nsThread::ProcessNextEvent xpcom/threads/nsThread.cpp:1197

The signature itself seems to go back quite a ways -- I see ESR 68 and 78 crashes in there -- but the reported crash addresses do seem to have changed in character. Other than one 0xe5e5.... crash in 82-beta these didn't really start showing up until 84. But there was clearly something wrong here before, too.

Ted and Jon, any ideas? Bug 1672172 and bug 1657025 are in the regression range in comment 0, and look like the only parser-related patches I can see in there. Thanks.

The problem here was cancelling parse tasks without the browser's knowledge (I
didn't realise that the cancel method did anything beyond waiting).

https://hg.mozilla.org/integration/autoland/rev/68e6b05b17b7e32496b5519f0b0fd9e99e10b3a0
https://hg.mozilla.org/mozilla-central/rev/68e6b05b17b7

As part of a security bug pattern analysis, we are requesting your help with a high level analysis of this bug. It is our hope to develop static analysis (or potentially runtime/dynamic analysis) in the future to identify classes of bugs.

Please visit this google form to reply.

(In reply to Release mgmt bot [:sylvestre / :calixte / :marco for bugbug] from comment #5)
Done.