New issue
Advanced search Search tips

Issue 733323 link

Starred by 2 users

Issue metadata

Status: Untriaged
Owner: ----
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 3
Type: Bug


Sign in to add a comment

Potential memory leaks in gmail/renderer.

Project Member Reported by erikc...@chromium.org, Jun 14 2017

Issue description

Debugging memory usage of gmail. Roughly:
1) take a native heap dump + JS dump
2) force GC
3) take a native heap dump + JS dump
4) navigate to about:blank and force GC.
5) take a native heap dump + JS dump

process number for renderer: 10717
 
Cc: erikc...@chromium.org etienneb@chromium.org
Labels: Performance-Memory
Files:
https://drive.google.com/a/google.com/file/d/0B6ho_7_ut5e1ZEViQmUxTjd1aEE/view?usp=sharing
Attaching potential leaks from the post-about:blank post-GC native heap dump.
process_10717_malloc-leaks.json
164 KB View Download
process_10717_partition_alloc-leaks.json
427 KB View Download
process_10721_blink_gc-objects.json
183 KB View Download
Cc: keishi@chromium.org
+ keishi

Since we're now looking at allocations in a post-GC about:blank renderer, all of these giant allocations are Blink leaks, right?
Attaching blink-gc leaks. c#2 had blink_gc-objects.
process_10717_blink_gc-leaks.json
264 KB View Download

Comment 5 by keishi@chromium.org, Jun 15 2017

Which tool should I use to look at the json files from c#2 and c#4?

Judging from the heapsnapshot files from c#1, post_blank_and_gc seems to be much smaller. But do you think there is a leak persisting post navigation to about:blank?
Any JSON viewer, like http://jsonviewer.stack.hu/.
Each leak consists of, # of objects alive, and trace of the stack that caused the object to be allocated.
Summary: Potential memory leaks in gmail/renderer. (was: gmail debugging)
Looking at partition_alloc leaks...
first potential leak: 48k objects created in blink::CSSSelectorList::AdoptSelectorVector

second potential leak: 34k objects created in blink::CSSSelectorList::AdoptSelectorVector

third potential leak: 20k objects created in blink::CSSSelectorList::AdoptSelectorVector

fourth potential leak: 20k objects created in blink::CSSSelectorList::AdoptSelectorVector

fifth potential leak: 16k strings created in blink::CSSSelectorParser::ConsumeClass.

and so on.
Blockedon: 733714
Blockedon: 733730
Blockedon: 733734
Blockedon: 733735
Labels: Needs-Investigation
Status: Available (was: Untriaged)
By looking to #c9, there is a probably repro for this leak:
  https://clusterfuzz.com/v2/testcase-detail/5515632539074560
Project Member

Comment 17 by sheriffbot@chromium.org, Aug 3

Labels: Hotlist-Recharge-Cold
Status: Untriaged (was: Available)
This issue has been Available for over a year. If it's no longer important or seems unlikely to be fixed, please consider closing it out. If it is important, please re-triage the issue.

Sorry for the inconvenience if the bug really should have been left as Available.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

Sign in to add a comment