LGTM, with a couple of comments. Søren, could you have a look as well since you implemented the ia32 version?
http://codereview.chromium.org/661469/diff/12/1005 File src/x64/codegen-x64.cc (right): http://codereview.chromium.org/661469/diff/12/1005#newcode9315 src/x64/codegen-x64.cc:9315: __ Move(undefined, Factory::undefined_value()); __ LoadRoot(undefined, Heap::kUndefinedValueRootIndex); http://codereview.chromium.org/661469/diff/12/1005#newcode9358 src/x64/codegen-x64.cc:9358: // As we are out of registers save the mask on the stack and use that Do we have one more register on x64 at the call sites? http://codereview.chromium.org/661469/diff/12/1006 File src/x64/codegen-x64.h (right): http://codereview.chromium.org/661469/diff/12/1006#newcode781 src/x64/codegen-x64.h:781: // string is found the code falls through with the string in register eax. eax -> rax But really, I think this comment should be changed on ia32 as well. I guess the string will be left in one of the argument registers (c1, c2?) which does not have to be eax/rax. http://codereview.chromium.org/661469 -- v8-dev mailing list [email protected] http://groups.google.com/group/v8-dev
