RESOLVED FIXED 104497
Profiler should say things about OSR exits 
https://bugs.webkit.org/show_bug.cgi?id=104497
Summary Profiler should say things about OSR exits
Filip Pizlo
Reported 2012-12-09 15:38:24 PST
Where they happened. How many times they happened. Patch forthcoming.
Attachments
sample profiling session (122.53 KB, text/plain)
2012-12-09 15:46 PST, Filip Pizlo 		no flags
Details
the patch (56.92 KB, patch)
2012-12-09 15:49 PST, Filip Pizlo 		no flags
DetailsFormatted DiffDiff
the patch (60.21 KB, patch)
2012-12-09 15:57 PST, Filip Pizlo 		no flags
DetailsFormatted DiffDiff
the patch (58.80 KB, patch)
2012-12-09 16:08 PST, Filip Pizlo 		webkit-ews: commit-queue-
DetailsFormatted DiffDiff
sample profiling session (130.10 KB, text/plain)
2012-12-09 16:13 PST, Filip Pizlo 		no flags
Details
the patch (60.05 KB, patch)
2012-12-09 16:24 PST, Filip Pizlo 		no flags
DetailsFormatted DiffDiff
the patch (60.04 KB, patch)
2012-12-09 16:30 PST, Filip Pizlo 		no flags
DetailsFormatted DiffDiff
the patch (60.03 KB, patch)
2012-12-09 16:43 PST, Filip Pizlo 		oliver: review+
DetailsFormatted DiffDiff
Show Obsolete (6) View All Add attachment proposed patch, testcase, etc.
Filip Pizlo
Comment 1 2012-12-09 15:46:58 PST
Created attachment 178441 [details] sample profiling session Shows a sample profiling session where I investigate a code block that has a lot of exits and recompilations. Note how I can see all of the exits that happened just by looking at the bytecode. This is so full of win, I'm really proud of myself. Then I look at the most recent DFG compilation and see that we're exiting because an "undefined" is flowing into something that we statically assume to be a cell.
Filip Pizlo
Comment 2 2012-12-09 15:49:44 PST
Created attachment 178442 [details] the patch This will cause some of the build systems to puke. I'll fix that shortly.
Filip Pizlo
Comment 3 2012-12-09 15:57:53 PST
Created attachment 178443 [details] the patch Fix things maybe
Filip Pizlo
Comment 4 2012-12-09 16:08:07 PST
Created attachment 178445 [details] the patch Rebased.
Filip Pizlo
Comment 5 2012-12-09 16:13:08 PST
Created attachment 178446 [details] sample profiling session
Early Warning System Bot
Comment 6 2012-12-09 16:16:52 PST
Comment on attachment 178445 [details] the patch Attachment 178445 [details] did not pass qt-ews (qt): Output: http://queues.webkit.org/results/15238252
Early Warning System Bot
Comment 7 2012-12-09 16:20:37 PST
Comment on attachment 178445 [details] the patch Attachment 178445 [details] did not pass qt-wk2-ews (qt): Output: http://queues.webkit.org/results/15242047
Filip Pizlo
Comment 8 2012-12-09 16:24:36 PST
Created attachment 178448 [details] the patch More build system fixes.
WebKit Review Bot
Comment 9 2012-12-09 16:26:56 PST
Attachment 178448 [details] did not pass style-queue: Failed to run "['Tools/Scripts/check-webkit-style', '--diff-files', u'Source/JavaScriptCore/CMakeLists.txt', u'S..." exit_code: 1 Source/JavaScriptCore/bytecode/ExitKind.h:46: The parameter name "kind" adds no information, so it should be removed. [readability/parameter_name] [5] Source/JavaScriptCore/bytecode/ExitKind.h:47: The parameter name "kind" adds no information, so it should be removed. [readability/parameter_name] [5] Total errors found: 2 in 29 files If any of these errors are false positives, please file a bug against check-webkit-style.
Filip Pizlo
Comment 10 2012-12-09 16:30:14 PST
Created attachment 178449 [details] the patch Fix style
Filip Pizlo
Comment 11 2012-12-09 16:43:48 PST
Created attachment 178451 [details] the patch ungimp the gtk
Filip Pizlo
Comment 12 2012-12-10 10:26:52 PST
Landed in http://trac.webkit.org/changeset/137175
Note You need to log in before you can comment on or make changes to this bug.