Bug 143843 - VarargsForwardingPhase should use bytecode liveness in addition to other uses to determine the last point that a candidate is used
Summary: VarargsForwardingPhase should use bytecode liveness in addition to other uses...
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: JavaScriptCore (show other bugs)
Version: 528+ (Nightly build)
Hardware: All All
: P2 Normal
Assignee: Filip Pizlo
URL:
Keywords:
Depends on:
Blocks: 143736
  Show dependency tree / graph
 
Reported: 2015-04-16 14:03 PDT by Filip Pizlo
Modified: 2015-04-28 10:56 PDT (History)
1 user (show)

See Also:

Attachments
WRONG PATCH (43.26 KB, patch)
2015-04-16 15:28 PDT, Filip Pizlo 		no flags Details | Formatted Diff | Diff
the patch (9.18 KB, patch)
2015-04-25 01:38 PDT, Filip Pizlo 		ggaren: review+
Details | Formatted Diff | Diff
Show Obsolete (1) 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 Filip Pizlo 2015-04-16 14:03:45 PDT 
...
Comment 1 Filip Pizlo 2015-04-16 15:28:28 PDT 
Created attachment 250955 [details]
WRONG PATCH
Comment 2 Filip Pizlo 2015-04-25 01:38:01 PDT 
Created attachment 251626 [details]
the patch
Comment 3 WebKit Commit Bot 2015-04-25 01:40:06 PDT 
Attachment 251626 [details] did not pass style-queue:


ERROR: Source/JavaScriptCore/dfg/DFGVarargsForwardingPhase.cpp:141:  Place brace on its own line for function definitions.  [whitespace/braces] [4]
ERROR: Source/JavaScriptCore/dfg/DFGPhantomInsertionPhase.cpp:127:  Place brace on its own line for function definitions.  [whitespace/braces] [4]
Total errors found: 2 in 5 files


If any of these errors are false positives, please file a bug against check-webkit-style.
Comment 4 Geoffrey Garen 2015-04-27 11:09:42 PDT 
Comment on attachment 251626 [details]
the patch

r=me
Comment 5 Filip Pizlo 2015-04-27 11:50:49 PDT 
Landed in http://trac.webkit.org/changeset/183406
Comment 6 Filip Pizlo 2015-04-28 10:39:40 PDT 
This appears to have regressed V8Spider/raytrace by 30% and Octane/raytrace by 9%.  I think I broke VarargsForwarding with this change.
Comment 7 Filip Pizlo 2015-04-28 10:56:05 PDT 
(In reply to comment #6)
> This appears to have regressed V8Spider/raytrace by 30% and Octane/raytrace
> by 9%.  I think I broke VarargsForwarding with this change.

The fix is: https://bugs.webkit.org/show_bug.cgi?id=144340