Closed
Bug 731783
Opened 13 years ago
Closed 13 years ago
GC: Eliminate per-compartment barrier marker
Categories
(Core :: JavaScript Engine, defect)
Core
JavaScript Engine
Tracking
()
RESOLVED
FIXED
mozilla13
People
(Reporter: billm, Assigned: billm)
Details
(Whiteboard: [MemShrink])
Attachments
(1 file)
16.14 KB,
patch
|
igor
:
review+
|
Details | Diff | Splinter Review |
We don't need this anymore since the runtime is now single-threaded.
I also fixed GCMarker::stop so that it releases memory that was allocated during the collection.
Attachment #601754 -
Flags: review?(igor)
Updated•13 years ago
|
Whiteboard: [MemShrink]
Comment 1•13 years ago
|
||
Comment on attachment 601754 [details] [diff] [review]
patch
Review of attachment 601754 [details] [diff] [review]:
-----------------------------------------------------------------
Nice!
Attachment #601754 -
Flags: review?(igor) → review+
Assignee | ||
Comment 2•13 years ago
|
||
Just so it's clear, this won't decrease memory usage by much at all. The stuff that we were pushing onto the per-compartment mark stack will now be pushed onto the per-runtime mark stack. But I guess that's a little better from a memshrink perspective because we have a memory reporter for the latter and not for the former.
Comment 3•13 years ago
|
||
> Just so it's clear, this won't decrease memory usage by much at all.
I clarified this with billm on IRC. With heavy usage the MarkStack can grow a lot. E.g. with MemBench it gets up to 2.8MB. Prior to this patch, once grown it would never shrink again, but with this patch it will shrink back to 256KB.
Assignee | ||
Comment 4•13 years ago
|
||
Target Milestone: --- → mozilla13
Comment 5•13 years ago
|
||
Status: NEW → RESOLVED
Closed: 13 years ago
Resolution: --- → FIXED
You need to log in
before you can comment on or make changes to this bug.
Description
•