1784233 - Crash in [@ OOM | small]

Status: RESOLVED INCOMPLETE

(Core :: General, defect)

Description

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

Observed twice in the last 3 days with the latest Nightly version (32-bit( at that time - OS Windows 8.1

Firefox 105.0a1's parent process was using 2.9 GB soon after launch, the UI got unresponsive. Once I terminated it, the other time it eventually crashed on its own. For the second occurrence: Nightly had already been running or a few minutes and the issue started when the chat.mozilla.org tab (pinned) got focus (that's the moment when it sometimes finishes loading parts of the UI) and quickly switched between channels listed there. Maybe a serviceworker issue?

The same installation had experience crashes soon after startup in bug 1764356 but that hadn't been observed for more than a week or so.

Crash report: https://crash-stats.mozilla.org/report/index/491911bf-3187-4201-9863-346ed0220811

MOZ_CRASH Reason: out of memory: 0x0000000000001000 bytes requested

Top 10 frames of crashing thread:

0 mozglue.dll mozalloc_abort memory/mozalloc/mozalloc_abort.cpp:26
1 mozglue.dll moz_xmalloc memory/mozalloc/mozalloc.cpp:58
2 xul.dll xul.dll@0x01712872 
3 xul.dll xul.dll@0x00d50307 
4 xul.dll xul.dll@0x00cd8545 
5 xul.dll xul.dll@0x007ed0f5 
6 xul.dll xul.dll@0x015f6558 
7 xul.dll xul.dll@0x015f4d9b 
8 xul.dll xul.dll@0x008e757e 
9 xul.dll xul.dll@0x0001c99e 

Comment 1

[Gabriele Svelto [:gsvelto]]

This particular dump appears to be running out of virtual memory, there's only ~260 MiB free which is a very low amount. I'd be curious to see a memory report of this particular instance.

Comment 2

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

Hit this again after the latest update. Parent process stack:

0x0000000000000000
VCRUNTIME140.dll!memset+0x3c
xul.dll!JSStructuredCloneWriter::writePrimitive+0x379
xul.dll!JSAutoStructuredCloneBuffer::write+0x9fc
xul.dll!mozilla::dom::StructuredCloneHolderBase::Write+0x7a
xul.dll!mozilla::dom::StructuredCloneHolder::Write+0x16
xul.dll!mozilla::dom::StructuredCloneHolder::Write+0x41
xul.dll!mozilla::dom::StructuredCloneBlob::Constructor+0x14a
xul.dll!mozilla::dom::StructuredCloneHolder_Binding::_constructor+0x18a
xul.dll!InternalConstruct+0x231
xul.dll!js::Construct+0x33
xul.dll!js::jit::InvokeFunction+0x347
0x0000000000000000
xul.dll!js::jit::MaybeEnterJit+0x2f1
xul.dll!js::CallGetter+0x5ef
xul.dll!js::NativeGetProperty+0x731
xul.dll!Str+0x2cc
xul.dll!Str+0x54a
xul.dll!js::Stringify+0x290
xul.dll!JS_Stringify+0x9a
xul.dll!nsContentUtils::StringifyJSON+0x88
xul.dll!nsFrameMessageManager::GetParamsForMessage+0x288
xul.dll!mozilla::dom::JSActor::SendQuery+0x46
xul.dll!mozilla::dom::JSWindowActorParent_Binding::sendQuery_promiseWrapper+0xe0
0x0000000000000000
xul.dll!?QueryInterface@BackstagePass@@WDA@AG?AW4nsresult@@ABUnsID@@PAPAX@Z+0xe7
xul.dll!nsCOMPtr_base::assign_with_AddRef+0x2f

The first child process kept one CPU core and half of the whole GPU busy:

0x0000000000000000
wow64cpu.dll!CpupSyscallStub+0x2
wow64cpu.dll!WaitForMultipleObjects32+0x1d
wow64.dll!RunCpuSimulation+0xa
wow64.dll!Wow64LdrpInitialize+0x172
ntdll.dll!LdrpInitializeProcess+0x1591
ntdll.dll!_LdrpInitialize+0x891fe
ntdll.dll!LdrInitializeThunk+0xe
ntdll.dll!_ZwWaitForMultipleObjects@20+0xc
ntdll.dll!RtlDeactivateActivationContextUnsafeFast+0x9c
xul.dll!mozilla::widget::WinUtils::WaitForMessage+0x77
xul.dll!nsAppShell::ProcessNextNativeEvent+0x3ac
xul.dll!nsThread::ProcessNextEvent+0x809
xul.dll!mozilla::ipc::MessagePump::Run+0x19b
xul.dll!MessageLoop::RunHandler+0x4f
xul.dll!MessageLoop::Run+0x3f
xul.dll!nsBaseAppShell::Run+0x25
xul.dll!nsAppShell::Run+0x2c
xul.dll!XRE_RunAppShell+0x40
xul.dll!mozilla::ipc::MessagePumpForChildProcess::Run+0x6a
xul.dll!MessageLoop::RunHandler+0x4f
xul.dll!MessageLoop::Run+0x3f
xul.dll!XRE_InitChildProcess+0x4b3
xul.dll!mozilla::BootstrapImpl::XRE_InitChildProcess+0x11
firefox.exe!wmain+0x56c
firefox.exe!__scrt_common_main_seh+0xfa
KERNEL32.DLL!@BaseThreadInitThunk@12+0x24
ntdll.dll!__RtlUserThreadStart+0x2f
ntdll.dll!__RtlUserThreadStart@8+0x1b

Comment 3

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

Hit this once when Nightly had been already for a long time after two documents in Google Docs got opened.

Comment 4

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

The Bugbug bot thinks this bug should belong to the 'Core::Memory Allocator' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.

Comment 5

[Marco Castelluccio [:marco]]

Sebastian, can you still reproduce? If so, could you collect a memory report?

Comment 6

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

This still reproduces intermittently but the UI never gets in a stable, responsive state which would enable one to open about:memory and save a memory report.

Comment 7

[Marco Castelluccio [:marco]]

Gabriele, is there something Sebastian could do to get the info?

Comment 8

[Gabriele Svelto [:gsvelto]]

Maybe collect a memory dump from an ordinary run in one of the sites that crashed? If we find something disproportionately large in there it might be the culprit, unless it really happens intermittently and we find nothing.

Comment 9

[Marco Castelluccio [:marco]]

Reducing severity to S3 as this is likely not so frequent.

Comment 10

[Andrew McCreight [:mccr8]]

I don't think we can do anything here without more information, unfortunately.