...
Created attachment 351727 [details] WIP - Patch
Created attachment 351730 [details] WIP - Patch Patch doesn't compile.
Created attachment 351934 [details] WIP - Patch
Created attachment 352878 [details] Patch
Created attachment 352880 [details] Patch
Created attachment 352885 [details] Benchmarks Changes are perf neutral.
Created attachment 353569 [details] Patch
Created attachment 353570 [details] Patch
Comment on attachment 353570 [details] Patch Investigating performance regression
Created attachment 353609 [details] Patch
Created attachment 353611 [details] Benchmarks This patch is perf neutral.
Ping Review
Created attachment 354235 [details] Patch
Comment on attachment 354235 [details] Patch Attachment 354235 [details] did not pass jsc-ews (mac): Output: https://webkit-queues.webkit.org/results/9908802 New failing tests: stress/ftl-put-by-id-setter-exception.js.dfg-eager-no-cjit-validate apiTests
Created attachment 354559 [details] Patch
Created attachment 354681 [details] Patch
Created attachment 354813 [details] Patch
Created attachment 355243 [details] Patch
Ping review
Created attachment 355629 [details] Patch
Created attachment 355630 [details] Patch
Created attachment 355863 [details] Patch
Comment on attachment 355863 [details] Patch r=me
Comment on attachment 355863 [details] Patch Clearing flags on attachment: 355863 Committed r238732: <https://trac.webkit.org/changeset/238732>
All reviewed patches have been landed. Closing bug.
<rdar://problem/46371247>
Comment on attachment 355863 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=355863&action=review > Source/JavaScriptCore/dfg/DFGAbstractInterpreterInlines.h:398 > + case ValueBitXor: > case ValueBitAnd: > + case ValueBitOr: Can we add constant folding for this in the future if the inputs are constant? > Source/JavaScriptCore/dfg/DFGFixupPhase.cpp:167 > + case ArithBitXor: > + node->setOpAndDefaultFlags(ValueBitXor); > + break; Did we end up switching this back to be the opposite direction? First emit ValueBitXor, then optimize to ArithBitXor.