Bug 98567 - [V8] toV8(Node*, ...) does more work than needed (6% faster on dom-traverse)
Summary: [V8] toV8(Node*, ...) does more work than needed (6% faster on dom-traverse)
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: New Bugs (show other bugs)
Version: 528+ (Nightly build)
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Adam Barth
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2012-10-05 15:42 PDT by Adam Barth
Modified: 2012-10-07 23:52 PDT (History)
3 users (show)

See Also:

Attachments
work in progress (3.29 KB, patch)
2012-10-05 15:42 PDT, Adam Barth 		no flags Details | Formatted Diff | Diff
dom-traverse results (6% better; first 3 runs are with patch, last 3 are without) (21.62 KB, text/html)
2012-10-05 15:52 PDT, Adam Barth 		no flags Details
Patch (4.10 KB, patch)
2012-10-05 16:08 PDT, Adam Barth 		no flags Details | Formatted Diff | Diff
Patch (4.13 KB, patch)
2012-10-05 16:09 PDT, Adam Barth 		no flags Details | Formatted Diff | Diff
Patch for landing (4.25 KB, patch)
2012-10-05 18:34 PDT, Adam Barth 		no flags Details | Formatted Diff | Diff
Show Obsolete (3) View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Adam Barth 2012-10-05 15:42:03 PDT 
[V8] toV8(Node*, ...) does more work than needed
Comment 1 Adam Barth 2012-10-05 15:42:21 PDT 
Created attachment 167399 [details]
work in progress
Comment 2 Adam Barth 2012-10-05 15:52:55 PDT 
Created attachment 167404 [details]
dom-traverse results (6% better; first 3 runs are with patch, last 3 are without)
Comment 3 Adam Barth 2012-10-05 16:08:38 PDT 
Created attachment 167405 [details]
Patch
Comment 4 Adam Barth 2012-10-05 16:09:45 PDT 
Created attachment 167406 [details]
Patch
Comment 5 Kentaro Hara 2012-10-05 17:25:37 PDT 
Comment on attachment 167406 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=167406&action=review

Awesome!

> Source/WebCore/bindings/scripts/CodeGeneratorV8.pm:534
> +    if (holder->wrapper() && *holder->wrapper() == holderWrapper && node->wrapper())

This looks a bit hacky. Shall we add a comment on this?

> Source/WebCore/bindings/scripts/CodeGeneratorV8.pm:535
> +        return *node->wrapper();

Maybe you can now remove the logic to check node->wrapper() from V8DOMWrapper::getCachedWrapper()? (You can do it in a follow-up patch.)
Comment 6 Adam Barth 2012-10-05 18:13:21 PDT 
> > Source/WebCore/bindings/scripts/CodeGeneratorV8.pm:534
> > +    if (holder->wrapper() && *holder->wrapper() == holderWrapper && node->wrapper())
> 
> This looks a bit hacky. Shall we add a comment on this?

Add a comment is a good idea.  It's a subtle trick.

> > Source/WebCore/bindings/scripts/CodeGeneratorV8.pm:535
> > +        return *node->wrapper();
> 
> Maybe you can now remove the logic to check node->wrapper() from V8DOMWrapper::getCachedWrapper()? (You can do it in a follow-up patch.)

We can once we move all callers over to toV8Fast.
Comment 7 Adam Barth 2012-10-05 18:34:44 PDT 
Created attachment 167433 [details]
Patch for landing
Comment 8 WebKit Review Bot 2012-10-05 19:08:42 PDT 
Comment on attachment 167433 [details]
Patch for landing

Clearing flags on attachment: 167433

Committed r130574: <http://trac.webkit.org/changeset/130574>
Comment 9 WebKit Review Bot 2012-10-05 19:08:45 PDT 
All reviewed patches have been landed.  Closing bug.