RESOLVED FIXED 170459
Don't need to Air::reportUsedRegisters for wasm at -O1
https://bugs.webkit.org/show_bug.cgi?id=170459
Summary Don't need to Air::reportUsedRegisters for wasm at -O1
Filip Pizlo
Reported 2017-04-04 11:08:17 PDT
Patch forthcoming.
Attachments
the patch (19.12 KB, patch)
2017-04-04 11:19 PDT, Filip Pizlo
saam: review+
Filip Pizlo
Comment 1 2017-04-04 11:19:43 PDT
Created attachment 306181 [details] the patch
Saam Barati
Comment 2 2017-04-04 11:39:33 PDT
Comment on attachment 306181 [details] the patch View in context: https://bugs.webkit.org/attachment.cgi?id=306181&action=review r=me > Source/JavaScriptCore/b3/B3StackmapGenerationParams.h:63 > + // NOTE: This will report bogus information if you did proc.setNeedsUsedRegisters(false). Can we make it crash, at least in debug builds? > Source/JavaScriptCore/b3/air/AirLiveness.h:42 > + SuperSamplerScope samplingScope(false); > + WTF::Liveness<Adapter>::compute(); Is the reason for this change just so you can time it?
Filip Pizlo
Comment 3 2017-04-04 11:49:05 PDT
(In reply to Saam Barati from comment #2) > Comment on attachment 306181 [details] > the patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=306181&action=review > > r=me > > > Source/JavaScriptCore/b3/B3StackmapGenerationParams.h:63 > > + // NOTE: This will report bogus information if you did proc.setNeedsUsedRegisters(false). > > Can we make it crash, at least in debug builds? I'll add an assertion. > > > Source/JavaScriptCore/b3/air/AirLiveness.h:42 > > + SuperSamplerScope samplingScope(false); > > + WTF::Liveness<Adapter>::compute(); > > Is the reason for this change just so you can time it? Yeah, it was just for timing. There was a prior patch where I had done this same hack and then lost it. It's super useful.
Filip Pizlo
Comment 4 2017-04-04 12:09:29 PDT
Note You need to log in before you can comment on or make changes to this bug.