From:  Andrew McCreight <>
Date:  13 Nov 2015 03:44:42 Hong Kong Time

Re: MemShrink Triage (12th Nov 2015)


On Thu, Nov 12, 2015 at 11:30 AM, Eric Rahm <> wrote:
MemShrink triage: 2015-11-12

Triage URL

(Add name and topic to discuss here)

Bug List
Vote for:
  - P1: High importance, will follow up periodically
  - P2: Important issue, possibly already being worked on
  - P3: Real issue, someone will get to it if they have time
  - moreinfo: We need logs, or clarifying information
  - invalid: Misclassified, remove the MemShrink tag

5 bugs to triage

We should also triage this one that I just added the tag to:
Bug 1217571 - The imagelib cache is broken in e10s.

In IRC, Seth said this could result in having multiple copies of an image. I think this should be P1.
  1200514 - Core :: XPCOM - protect against leaks from CycleCollectedJSRuntime::RunInStableState() late in shutdown

It's not particularly clear to me what the issue is. Sounds like a potential shutdown leak that has been worked around. Perhaps P2?

P2 sounds good. This is likely just some kind of teardown issue, but it makes it harder to know when there are real media leaks.

  khuey, what do you think?
  1221503 - Firefox :: General - Firefox is slow (Windows swapping all the time)

moreinfo, looks like a lot of memory being used by style-sets in embedded facebook-esque widgets. Most likely an ad-block-like add-on.
  1222092 - Core :: Widget - leaking nsWindow with e10s + active content (cursor)


This sounds like more of a P2 to me. It sounds like some global isn't being cleared soon enough, so we can't have an unbounded leak. Maybe I'm wrong.
  1223445 - Core :: DOM: Animation - KeyframeEffectReadOnly objects end up keeping lots of other objects alive too long

P1, animation related leak causing excessive GC/CC times.

  mccr8, what do you think?

Definitely. I don't know how widely this is used, but sites that use it get horrible problems.
  1223691 - Core :: Web Audio - MediaStreamGraph leak

WebAudio shutdown leak, karlt thinks it's related to bug 1200514, so maybe P2 as well.

Yeah, same basic thing.

dev-memory mailing list