Status: New
Owner: ----
New issue 2387 by [email protected]: Parser/tiny-innerHTML is spending
17.5% of time in sem_post under v8?
http://code.google.com/p/v8/issues/detail?id=2387
Parser/tiny-innerHTML is spending 17.5% of time in sem_post under v8?
Tools/Scripts/run-perf-tests Parser/tiny-innerHTML.html
or directly:
http://trac.webkit.org/browser/trunk/PerformanceTests/Parser/tiny-innerHTML.html
first 10 lines from pprof --text output:
Running Parser/tiny-innerHTML.html (10 of 12)
363 17.5% 17.5% 363 17.5% sem_post
45 2.2% 19.7% 186 9.0% WebCore::HTMLTokenizer::nextToken
43 2.1% 21.7% 43 2.1% MaskPtr (inline)
34 1.6% 23.4% 34 1.6% append (inline)
32 1.5% 24.9% 66 3.2%
WebCore::SegmentedString::advanceAndUpdateLineNumber
30 1.4% 26.3% 695 33.5%
WebCore::HTMLDocumentParser::pumpTokenizer
29 1.4% 27.7% 34 1.6%
WebCore::SegmentedString::advanceAndUpdateLineNumber (inline)
27 1.3% 29.0% 65 3.1% tcmalloc::FL_Next
25 1.2% 30.3% 36 1.7%
WebCore::HTMLConstructionSite::reconstructTheActiveFormattingElements
23 1.1% 31.4% 45 2.2% SegmentedString (inline)
(The sample was generated by using the run-perf-tests --profile patch,
currently attached to https://bugs.webkit.org/show_bug.cgi?id=99517.)
I suspect we may be tickling some pathological case in v8, but I don't yet
know how to get v8 symbols to see.
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
