Bytecode checkpoint fixes for arm64(e)
Created attachment 386998 [details] Patch
Comment on attachment 386998 [details] Patch r=me
rdar://problem/58310178
Comment on attachment 386998 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=386998&action=review > Source/JavaScriptCore/dfg/DFGOSRExitCompilerCommon.cpp:150 > + return LLInt::getCodePtr<NoPtrTag>(checkpoint_osr_exit_from_inlined_call_trampoline).executableAddress(); why not make this function return a typed value instead of void*, and then caller's can unbox as needed?
r=me too
Comment on attachment 386998 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=386998&action=review >> Source/JavaScriptCore/dfg/DFGOSRExitCompilerCommon.cpp:150 >> + return LLInt::getCodePtr<NoPtrTag>(checkpoint_osr_exit_from_inlined_call_trampoline).executableAddress(); > > why not make this function return a typed value instead of void*, and then caller's can unbox as needed? Done, I went with MacroAssemblerCodePtr<JSEntryPtrTag> as the type.
Committed r254142: <https://trac.webkit.org/changeset/254142>
Non-arm64e build fix in: https://trac.webkit.org/changeset/254143.