Issue 9323007: Tweak compaction candidate selection to avoid keeping page with low occupancy around.

Keyboard Shortcuts

	File
u :	up to issue
j / k :	jump to file after / before current file
J / K :	jump to next file with a comment after / before current file
	Side-by-side diff
i :	toggle intra-line diffs
e :	expand all comments
c :	collapse all comments
s :	toggle showing all comments
n / p :	next / previous diff chunk or comment
N / P :	next / previous comment
<Up> / <Down> :	next / previous line

	Issue
u :	up to list of issues
j / k :	jump to patch after / before current patch
o / <Enter> :	open current patch in side-by-side view
i :	open current patch in unified diff view

	Issue List
j / k :	jump to issue after / before current issue
o / <Enter> :	open current issue

Issue 9323007: Tweak compaction candidate selection to avoid keeping page with low occupancy around. (Closed)

Created:
8 years, 10 months ago by Vyacheslav Egorov (Chromium)

Modified:
8 years, 10 months ago

Reviewers:
Erik Corry

CC:
v8-dev

Base URL:
https://v8.googlecode.com/svn/branches/bleeding_edge

Visibility:
Public.

Description

Tweak compaction candidate selection to avoid keeping page with low occupancy around. Increase slots buffer chain length to 15 to make compaction more aggressive and usefull. Pass gc and collector selection reasons to GCTracer to allow more meaningull --gc-trace. Print fragmentation of spaces that we do not compact. R=erik.corry@gmail.com Committed: https://code.google.com/p/v8/source/detail?r=10601

Patch Set 1 #

Total comments: 1

Created: 8 years, 10 months ago

	Unified diffs	Side-by-side diffs	Stats (+294 lines, -112 lines)
M	src/api.cc	View	1 chunk	+1 line, -1 line	0 comments
M	src/debug.cc	View	3 chunks	+7 lines, -4 lines	0 comments
M	src/execution.cc	View	1 chunk	+1 line, -1 line	0 comments
M	src/extensions/gc-extension.cc	View	1 chunk	+1 line, -1 line	0 comments
M	src/heap.h	View	5 chunks	+28 lines, -9 lines	0 comments
M	src/heap.cc	View	19 chunks	+71 lines, -31 lines	0 comments
M	src/heap-inl.h	View	4 chunks	+8 lines, -5 lines	0 comments
M	src/log.cc	View	3 chunks	+6 lines, -3 lines	0 comments
M	src/mark-compact.h	View	2 chunks	+3 lines, -1 line	0 comments
M	src/mark-compact.cc	View	7 chunks	+147 lines, -3 lines	1 comment
M	src/mark-compact-inl.h	View	1 chunk	+1 line, -0 lines	0 comments
M	src/mksnapshot.cc	View	1 chunk	+1 line, -1 line	0 comments
	src/profile-generator.cc	View	2 chunks	+8 lines, -3 lines	0 comments
M	src/runtime.cc	View	4 chunks	+9 lines, -5 lines	0 comments
M	src/spaces.h	View	1 chunk	+2 lines, -44 lines	0 comments

Messages

Total messages: 2 (0 generated)

Expand Messages | Collapse Messages

Erik Corry

8 years, 10 months ago (2012-02-03 10:12:39 UTC) #2

LGTM

https://chromiumcodereview.appspot.com/9323007/diff/1/src/mark-compact.cc
File src/mark-compact.cc (right):

https://chromiumcodereview.appspot.com/9323007/diff/1/src/mark-compact.cc#new...
src/mark-compact.cc:578: space->CountFreeListItems(p, &sizes);
This call worries me.  It can potentially take a long time, and really all we
need to know is whether there is a small number of live objects on the page.  If
the page consists of 100k objects and 100k gaps then we are not interested in it
for the purposes of REDUCE_MEMORY_FOOTPRINT.  So I think CountFreeList should
take a maximum number of free list nodes to traverse.

I think the other 2 places this is used would also benefit from a max argument.

Expand Messages | Collapse Messages