LGTM, with a nit (and a question).
https://codereview.chromium.org/952233002/diff/1/src/mips64/lithium-codegen-mips64.cc
File src/mips64/lithium-codegen-mips64.cc (right):
https://codereview.chromium.org/952233002/diff/1/src/mips64/lithium-codegen-mips64.cc#newcode2367
src/mips64/lithium-codegen-mips64.cc:2367:
Operand(static_cast<int32_t>(kHoleNanUpper32)));
this looks like nice solution to remove some instructions! Curious if
this changes the number of instructions emitted by the li() in the
branch, though. Is our li() clever enough to sign-extend without needing
to explicitly load all 4 16-bit chunks?
https://codereview.chromium.org/952233002/diff/1/src/mips64/macro-assembler-mips64.cc
File src/mips64/macro-assembler-mips64.cc (right):
https://codereview.chromium.org/952233002/diff/1/src/mips64/macro-assembler-mips64.cc#newcode3891
src/mips64/macro-assembler-mips64.cc:3891: // scratch1 is now effective
address of the double element.
nit: It looks like this comment is now stale - scratch1 is only used in
CheckMap() above, and should have map pointer. Can you correct this as
necessary.
https://codereview.chromium.org/952233002/
--
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed to the Google Groups "v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
