Bug 151321

Summary:

B3 should be be clever about choosing which child to reuse for result in two-operand commutative operations

Product:

WebKit

Reporter:

Filip Pizlo <fpizlo>

Component:

JavaScriptCore

Assignee:

Filip Pizlo <fpizlo>

Status:

RESOLVED FIXED

Severity:

Normal

CC:

barraclough, benjamin, buildbot, commit-queue, ggaren, mark.lam, mhahnenb, msaboff, nrotem, oliver, rniwa, saam, sam

Priority:

Version:

WebKit Nightly Build

Hardware:

All

OS:

All

Bug Depends on:

Bug Blocks:

150279

Attachments:

Description	Flags
I wrote some code	none
almost happy with it	none
the patch	buildbot: commit-queue-
Archive of layout-test-results from ews100 for mac-yosemite	none
Archive of layout-test-results from ews105 for mac-yosemite-wk2	none
Archive of layout-test-results from ews117 for mac-yosemite	none
the patch	ggaren: review+

Filip Pizlo

Reported 2015-11-16 13:19:53 PST

If we have: Int32 @3 = Add(@1, @2) and we know that @1 will die, then we should lower this to: Move %tmp1, %tmp3 Add32 %tmp2, %tmp3 But if we know that @2 will die, then we should lower this to: Move %tmp2, %tmp3 Add32 %tmp1, %tmp3

Attachments
I wrote some code (8.65 KB, patch) 2015-11-28 11:50 PST, Filip Pizlo	no flags	Details Formatted Diff Diff
almost happy with it (10.18 KB, patch) 2015-11-29 16:06 PST, Filip Pizlo	no flags	Details Formatted Diff Diff
the patch (29.38 KB, patch) 2015-11-29 17:29 PST, Filip Pizlo	buildbot: commit-queue-	Details Formatted Diff Diff
Archive of layout-test-results from ews100 for mac-yosemite (1.08 MB, application/zip) 2015-11-29 18:14 PST, Build Bot	no flags	Details
Archive of layout-test-results from ews105 for mac-yosemite-wk2 (1.02 MB, application/zip) 2015-11-29 18:15 PST, Build Bot	no flags	Details
Archive of layout-test-results from ews117 for mac-yosemite (1.12 MB, application/zip) 2015-11-29 18:16 PST, Build Bot	no flags	Details
the patch (28.79 KB, patch) 2015-11-30 10:36 PST, Filip Pizlo	ggaren: review+	Details Formatted Diff Diff
Show Obsolete (6) View All Add attachment proposed patch, testcase, etc.

Filip Pizlo

Comment 1 2015-11-28 11:50:56 PST

Created attachment 266214 [details] I wrote some code

Filip Pizlo

Comment 2 2015-11-29 16:06:13 PST

Created attachment 266227 [details] almost happy with it This is a regression on imaging-gaussian-blur because it commutes something that really should not have been commuted. I'm thinking about what to do.

Filip Pizlo

Comment 3 2015-11-29 17:29:31 PST

Created attachment 266229 [details] the patch

Filip Pizlo

Comment 4 2015-11-29 17:30:32 PST

Comment on attachment 266229 [details] the patch View in context: https://bugs.webkit.org/attachment.cgi?id=266229&action=review > Source/JavaScriptCore/dfg/DFGCommon.h:41 > -#define FTL_USES_B3 0 > +#define FTL_USES_B3 1 I will revert, I promise!

WebKit Commit Bot

Comment 5 2015-11-29 17:32:01 PST

Attachment 266229 [details] did not pass style-queue: ERROR: Source/JavaScriptCore/b3/B3PhiChildren.h:154: Place brace on its own line for function definitions. [whitespace/braces] [4] Total errors found: 1 in 9 files If any of these errors are false positives, please file a bug against check-webkit-style.

Build Bot

Comment 6 2015-11-29 18:14:09 PST

Comment on attachment 266229 [details] the patch Attachment 266229 [details] did not pass mac-ews (mac): Output: http://webkit-queues.webkit.org/results/494819 Number of test failures exceeded the failure limit.

Build Bot

Comment 7 2015-11-29 18:14:12 PST

Created attachment 266231 [details] Archive of layout-test-results from ews100 for mac-yosemite The attached test failures were seen while running run-webkit-tests on the mac-ews. Bot: ews100 Port: mac-yosemite Platform: Mac OS X 10.10.5

Build Bot

Comment 8 2015-11-29 18:15:33 PST

Comment on attachment 266229 [details] the patch Attachment 266229 [details] did not pass mac-wk2-ews (mac-wk2): Output: http://webkit-queues.webkit.org/results/494816 Number of test failures exceeded the failure limit.

Build Bot

Comment 9 2015-11-29 18:15:35 PST

Created attachment 266232 [details] Archive of layout-test-results from ews105 for mac-yosemite-wk2 The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews. Bot: ews105 Port: mac-yosemite-wk2 Platform: Mac OS X 10.10.5

Build Bot

Comment 10 2015-11-29 18:16:05 PST

Comment on attachment 266229 [details] the patch Attachment 266229 [details] did not pass mac-debug-ews (mac): Output: http://webkit-queues.webkit.org/results/494812 Number of test failures exceeded the failure limit.

Build Bot

Comment 11 2015-11-29 18:16:07 PST

Created attachment 266233 [details] Archive of layout-test-results from ews117 for mac-yosemite The attached test failures were seen while running run-webkit-tests on the mac-debug-ews. Bot: ews117 Port: mac-yosemite Platform: Mac OS X 10.10.5

Filip Pizlo

Comment 12 2015-11-30 10:36:16 PST

Created attachment 266251 [details] the patch

WebKit Commit Bot

Comment 13 2015-11-30 10:37:23 PST

Attachment 266251 [details] did not pass style-queue: ERROR: Source/JavaScriptCore/b3/B3PhiChildren.h:154: Place brace on its own line for function definitions. [whitespace/braces] [4] Total errors found: 1 in 8 files If any of these errors are false positives, please file a bug against check-webkit-style.

Geoffrey Garen

Comment 14 2015-11-30 12:14:21 PST

Comment on attachment 266251 [details] the patch View in context: https://bugs.webkit.org/attachment.cgi?id=266251&action=review r=me > Source/JavaScriptCore/ChangeLog:8 > + When lowering a commutative operation two a two-operand instruction, you have a choice of which to > Source/JavaScriptCore/b3/B3LowerToAir.cpp:542 > + // then we can flip things around to held coalescing, which then kills the move instruction. help > Source/JavaScriptCore/b3/B3PhiChildren.h:109 > + UpsilonCollection() > + { > + } Do we use this null initializer? I couldn't find its use. If we do use it, we should probably null-initialize our data members.

Filip Pizlo

Comment 15 2015-11-30 12:18:55 PST

(In reply to comment #14) > Comment on attachment 266251 [details] > the patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=266251&action=review > > r=me > > > Source/JavaScriptCore/ChangeLog:8 > > + When lowering a commutative operation two a two-operand instruction, you have a choice of which > > to > > > Source/JavaScriptCore/b3/B3LowerToAir.cpp:542 > > + // then we can flip things around to held coalescing, which then kills the move instruction. > > help > > > Source/JavaScriptCore/b3/B3PhiChildren.h:109 > > + UpsilonCollection() > > + { > > + } > > Do we use this null initializer? I couldn't find its use. > > If we do use it, we should probably null-initialize our data members. I added the new-style { } initializers. I like it when these classes have a null initializer because then you can use them more freely, like: PhiChildren::UpsilonCollection x; if (things) x = phis[index]; else x = otherPhis[index];

Filip Pizlo

Comment 16 2015-11-30 15:20:23 PST

Landed in http://trac.webkit.org/changeset/192816

Note You need to log in before you can comment on or make changes to this bug.