On 2013/09/10 11:29:21, Benedikt Meurer wrote:
https://codereview.chromium.org/23480027/diff/10001/src/arm/macro-assembler-arm.cc
File src/arm/macro-assembler-arm.cc (right):
https://codereview.chromium.org/23480027/diff/10001/src/arm/macro-assembler-arm.cc#newcode3200
src/arm/macro-assembler-arm.cc:3200: sub(ip, length, Operand(4));
Hm, this use of ip doesn't sit well with me. Using ip in the ARM backend
is
like
calling for trouble, because it's not always obvious to anyone whether it
is
preserved across a bunch of operations or not (an inherent design flaw in
the
ARM backend).
That being said, I don't like the new code. Looking at this from a
highlevel
point of view, I think the main optimization opportunity is copying by
word.
But
although we optimize depending on src alignment, we never check to see
whether
dst is actually word aligned.
I also don't like the overlap with the custom MemCopy function in
codegen-arm.cc.
Is it possible to commit only test case code if you think there is no need
to
optimize this function?
https://codereview.chromium.org/23480027/
--
--
v8-dev mailing list
v8-dev@googlegroups.com
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 v8-dev+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.