Issue 651019: Improve load of global variables.

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 651019: Improve load of global variables. (Closed)

Created:
10 years, 10 months ago by William Hesse

Modified:
9 years, 7 months ago

Reviewers:
Kevin Millikin (Chromium), Mads Ager (chromium)

CC:
v8-dev

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

Visibility:
Public.

Description

Improve load of global variables. Committed: http://code.google.com/p/v8/source/detail?r=3911

Patch Set 1 #

Patch Set 2 : '' #

Total comments: 1

Created: 10 years, 10 months ago

Download [raw] [tar.bz2]

		Unified diffs	Side-by-side diffs	Delta from patch set	Stats (+3 lines, -0 lines)			Patch
	M	src/ia32/codegen-ia32.cc	View	1	1 chunk	+3 lines, -0 lines	1 comment	Download

Messages

Total messages: 3 (0 generated)

Expand Messages | Collapse Messages

Mads Ager (chromium)

LGTM I think the comment could be extended a bit to make it clear why ...

10 years, 10 months ago (2010-02-19 10:01:33 UTC) #2

Kevin Millikin (Chromium)

10 years, 10 months ago (2010-02-19 10:23:04 UTC) #3

Questions:

1. If this is a good idea, why not do it everywhere
(LoadFromGlobalSlotCheckExtensions, EmitNamedPropertyAssignment for compound
assignments)?

2. Would it be easy to port this change to x64?

http://codereview.chromium.org/651019/diff/1002/6
File src/ia32/codegen-ia32.cc (right):

http://codereview.chromium.org/651019/diff/1002/6#newcode693
src/ia32/codegen-ia32.cc:693: // Spilling eax makes it free, so LoadGlobal loads
directly into eax.
The comment is still not quite right.  LoadIC wants receiver in eax and the
register allocator will prefer eax if free.

I think it's obvious that Spill(eax) makes it free, but not that the register
allocator prefers it.

Named stores do use reference in the case of e.x++ and ++e.x, so don't put that
comment at all since it's not true.

Expand Messages | Collapse Messages