StringCache is an Isolate-local data structure. Currently, V8 bindings need to look up an Isolate to find the StringCache. By using AccessorInfo.GetIsolate() or Arguments.GetIsolate(), we can remove the Isolate look-up.
Created attachment 137440 [details] Performance tests
Created attachment 137441 [details] Patch
Comment on attachment 137441 [details] Patch Attachment 137441 [details] did not pass chromium-ews (chromium-xvfb): Output: http://queues.webkit.org/results/12412665
Created attachment 137499 [details] Patch
Rolled chromium DEPS to r132474 and re-uploaded the patch. The patch requires V8 r11306.
Created attachment 137502 [details] Patch
Comment on attachment 137502 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=137502&action=review > Source/WebCore/ChangeLog:34 > + Performance test: https://bugs.webkit.org/attachment.cgi?id=137440 > + div.id: 30.70ns -> 26.70ns (+15%) > + div.className: 31.10ns -> 26.40ns (+18%) > + div.nodeName: 37.70ns -> 33.00ns (+14%) > + text.nodeValue: 31.40ns -> 25.90ns (+21%) > + text.textContent: 51.50ns -> 45.90ns (+12%) Should we add it as a performance test?
(In reply to comment #7) > > Source/WebCore/ChangeLog:34 > > + Performance test: https://bugs.webkit.org/attachment.cgi?id=137440 > > + div.id: 30.70ns -> 26.70ns (+15%) > > + div.className: 31.10ns -> 26.40ns (+18%) > > + div.nodeName: 37.70ns -> 33.00ns (+14%) > > + text.nodeValue: 31.40ns -> 25.90ns (+21%) > > + text.textContent: 51.50ns -> 45.90ns (+12%) > > Should we add it as a performance test? Bindings/dom-attributes.html should cover it. I'll update it later. (Currently, the div.id test in dom-attributes.html returns null, which is a different call-path from the performance test attached in this bug.)
Comment on attachment 137502 [details] Patch Clearing flags on attachment: 137502 Committed r114421: <http://trac.webkit.org/changeset/114421>
All reviewed patches have been landed. Closing bug.
Reverted r114421 for reason: Chromium crash Committed r114645: <http://trac.webkit.org/changeset/114645>
Created attachment 138368 [details] Patch for relanding. It is revealed that the patch was innocent.
Committed r114912: <http://trac.webkit.org/changeset/114912>