1377914 - Don't trigger a full GC in nsGlobalWindow::DetachFromDocShell()

Status: RESOLVED WONTFIX

(Core :: DOM: Core & HTML, defect, P3)

Description

[Andrew McCreight [:mccr8]]

It looks like we may not actually need to do this.

Comment 1

[Olli Pettay [:smaug][[e-mail]]]

Attachment: no_full_gc_at_docshell_detach.diff

Comment 2

[Andrew McCreight [:mccr8]]

Comment on attachment 8883058 [details] [diff] [review]
no_full_gc_at_docshell_detach.diff

Review of attachment 8883058 [details] [diff] [review]:
-----------------------------------------------------------------

Hopefully this will be okay!

Comment 3

[Olli Pettay [:smaug][[e-mail]]]

Windows 7 VM opt R isn't too happy with this
https://treeherder.mozilla.org/#/jobs?repo=try&revision=6361d4aba26ba62748249ef3e5fcab94cbe7a828&selectedJob=111539321&filter-searchStr=Windows%207%20VM%20opt

Comment 4

[Andrew McCreight [:mccr8]]

I wonder if in some cases we have to GC the same zone twice in a row, but the PokeGC approach only does it once. That might be hard to fix, because we have to worry about zones going away.

Comment 5

[Andrew Overholt [:overholt]]

Where did we end up with this work?

Comment 6

[Andrew McCreight [:mccr8]]

I'm not sure, but based on comment 3, I assume it made Win32 reftests OOM like they always do when we try to GC/CC less aggressively.

Comment 7

[Olli Pettay [:smaug][[e-mail]]]

Hmm, we have changed GC/CC scheduling a bit, worth to push to try again.

remote: 
remote: View the pushlog for these changes here:
remote:   https://hg.mozilla.org/try/pushloghtml?changeset=42e80c4e478642c06406f58767da8fb19a517191
remote: 
remote: Follow the progress of your build on Treeherder:
remote:   https://treeherder.mozilla.org/#/jobs?repo=try&revision=42e80c4e478642c06406f58767da8fb19a517191
remote: recorded changegroup in replication log in 0.019s

Comment 8

[Olli Pettay [:smaug][[e-mail]]]

Looks quite reasonable. Possibly worth to try to land for FF58.

Comment 9

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/a50a428b23fc
Don't trigger a full GC in nsGlobalWindow::DetachFromDocShell(), r=mccr8

Comment 10

[Olli Pettay [:smaug][[e-mail]]]

Crossing fingers.

Comment 11

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

Backed out for frequent leaks during a11y run on Linux x64 asan:

https://hg.mozilla.org/integration/mozilla-inbound/rev/27ba09aa4be87e7dc6e036bbbe1045e2a2a19f01

Push with failures: https://treeherder.mozilla.org/#/jobs?repo=mozilla-inbound&revision=a50a428b23fcfeaa84cb7f781d9a9f592216eaa8&filter-resultStatus=testfailed&filter-resultStatus=busted&filter-resultStatus=exception&filter-resultStatus=retry&filter-resultStatus=usercancel&filter-resultStatus=runnable
Failure log: https://treeherder.mozilla.org/logviewer.html#?job_id=133147779&repo=mozilla-inbound

[task 2017-09-25T16:03:42.167Z] 16:03:42     INFO - TEST-INFO | LeakSanitizer | To show the addresses of leaked objects add report_objects=1 to LSAN_OPTIONS
[task 2017-09-25T16:03:42.167Z] 16:03:42     INFO - TEST-INFO | LeakSanitizer | This can be done in testing/mozbase/mozrunner/mozrunner/utils.py
[task 2017-09-25T16:03:42.168Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_malloc, maybe_pod_malloc, js::Shape::hashify, maybeCreateTableForLookup
[task 2017-09-25T16:03:42.169Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_calloc, maybe_pod_calloc, AllocScriptData, js::detail::CopyScript
[task 2017-09-25T16:03:42.170Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_calloc, maybe_pod_calloc, NewEmptyScopeData, js::Scope::XDRSizedBindingNames
[task 2017-09-25T16:03:42.171Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_malloc, maybe_pod_malloc, mozilla::UniquePtr, js::GlobalScope::create
[task 2017-09-25T16:03:42.171Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_malloc, maybe_pod_malloc, js::NewStringCopyNDontDeflate, AtomizeAndCopyChars
[task 2017-09-25T16:03:42.172Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_malloc, maybe_pod_malloc, js::NewStringCopyNDontDeflate, js::JSONParser
[task 2017-09-25T16:03:42.173Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_malloc, maybe_pod_malloc, mozilla::UniquePtr, js::LexicalScope::create
[task 2017-09-25T16:03:42.174Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at start_thread, clone, js_pod_malloc, maybe_pod_malloc
[task 2017-09-25T16:03:42.175Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_malloc, maybe_pod_malloc, mozilla::UniquePtr, js::FunctionScope::create
[task 2017-09-25T16:03:42.176Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_calloc, maybe_pod_calloc, NewEmptyScopeData, js::GlobalScope::create
[task 2017-09-25T16:03:42.176Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_malloc, maybe_pod_malloc, NewStringDeflated, AtomizeAndCopyChars
[task 2017-09-25T16:03:42.177Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_malloc, maybe_pod_malloc, js::Shape::hashify, js::NativeObject::toDictionaryMode
[task 2017-09-25T16:03:42.178Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at nsStringBuffer::Alloc, Atom::Atom, CreateDynamic, NS_Atomize
[task 2017-09-25T16:03:42.178Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at Alloc, nsTSubstring, nsTSubstring, nsTSubstring
[task 2017-09-25T16:03:42.180Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at AllocateProtoAndIfaceCache, xpc::CreateGlobalObject, XPCWrappedNative::WrapNewGlobal, xpc::InitClassesWithNewWrappedGlobal
[task 2017-09-25T16:03:42.180Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_malloc, maybe_pod_malloc, NewStringDeflated, JS_NewUCStringCopyN
[task 2017-09-25T16:03:42.181Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_malloc, maybe_pod_malloc, js::NewStringCopyNDontDeflate, NewStringCopyN
[task 2017-09-25T16:03:42.182Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_malloc, maybe_pod_malloc, mozilla::UniquePtr, js::FunctionScope::clone
[task 2017-09-25T16:03:42.183Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_malloc, maybe_pod_malloc, js::TypeNewScript::make, js::ObjectGroup::defaultNewGroup
[task 2017-09-25T16:03:42.183Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at nsStringBuffer::Alloc, nsAttrValue::GetStringBuffer, nsAttrValue::SetTo, mozilla::dom::Element::SetAttr
[task 2017-09-25T16:03:42.184Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_calloc, maybe_pod_calloc, AllocScriptData, partiallyInit
[task 2017-09-25T16:03:42.185Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_malloc, maybe_pod_malloc, mozilla::UniquePtr, js::VarScope::create
[task 2017-09-25T16:03:42.186Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_calloc, maybe_pod_calloc, AllocScriptData, JSScript::partiallyInit
[task 2017-09-25T16:03:42.186Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_malloc, maybe_pod_malloc, mozilla::UniquePtr, js::Scope::clone
[task 2017-09-25T16:03:42.187Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_calloc, maybe_pod_calloc, NewEmptyScopeData, js::FunctionScope::create
[task 2017-09-25T16:03:42.188Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_new, HashChildren, js::PropertyTree::insertChild, inlinedGetChild
[task 2017-09-25T16:03:42.189Z] 16:03:42    ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at js_pod_malloc, maybe_pod_malloc, js::LazyScript::CreateRaw, js::LazyScript::Create

Comment 12

[Olli Pettay [:smaug][[e-mail]]]

Given that this caused issues, closing.
(doing some end-of-the-year clearance on really old needinfos)