MIPS: Change inlined cache of intanceof stub to use indirection through cell.

MIPS: Change inlined cache of intanceof stub to use indirection through cell.

Port r10380 (46f646).

Original commit message:
The stub was directly patching caller's code without issuing write barrier which violated incremental marking invariants.

BUG=
TEST=

Committed: https://code.google.com/p/v8/source/detail?r=10457

[kalmard, 2012-01-19 12:45:46]

Should be applied to the 3.7 branch as well (merged in r10387 for ARM).

Successful testing of v8 on MIPS requires the following patches to be landed as
well:
http://codereview.chromium.org/9265007/
http://codereview.chromium.org/9192011/
http://codereview.chromium.org/9150025/
http://codereview.chromium.org/9254025/
http://codereview.chromium.org/9150026/

[Vyacheslav Egorov (Chromium), 2012-01-20 11:50:41]

http://codereview.chromium.org/9159008/diff/1/src/mips/macro-assembler-mips.cc
File src/mips/macro-assembler-mips.cc (right):

http://codereview.chromium.org/9159008/diff/1/src/mips/macro-assembler-mips.c...
src/mips/macro-assembler-mips.cc:4740: Check(eq, "The instruction to patch
should be a lui.",
check message does not match function actions (I know that it does not match on
ARM as well).

http://codereview.chromium.org/9159008/diff/1/src/mips/macro-assembler-mips.c...
src/mips/macro-assembler-mips.cc:4751: Check(eq, "The instruction to patch
should be an ori.",
check message does not match function actions (I know that it does not match on
ARM).

http://codereview.chromium.org/9159008/diff/1/src/mips/macro-assembler-mips.c...
src/mips/macro-assembler-mips.cc:4755: // "scratch" now holds an ori
instruction. Extract the immediscratche.
immediscratche ?

[kalmard, 2012-01-20 12:13:32]

Thanks for the review, the issues are fixed.

http://codereview.chromium.org/9159008/diff/1/src/mips/macro-assembler-mips.cc
File src/mips/macro-assembler-mips.cc (right):

http://codereview.chromium.org/9159008/diff/1/src/mips/macro-assembler-mips.c...
src/mips/macro-assembler-mips.cc:4740: Check(eq, "The instruction to patch
should be a lui.",
On 2012/01/20 11:50:41, Vyacheslav Egorov wrote:
> check message does not match function actions (I know that it does not match
on
> ARM as well).

Done.

http://codereview.chromium.org/9159008/diff/1/src/mips/macro-assembler-mips.c...
src/mips/macro-assembler-mips.cc:4751: Check(eq, "The instruction to patch
should be an ori.",
On 2012/01/20 11:50:41, Vyacheslav Egorov wrote:
> check message does not match function actions (I know that it does not match
on
> ARM).

Done.

http://codereview.chromium.org/9159008/diff/1/src/mips/macro-assembler-mips.c...
src/mips/macro-assembler-mips.cc:4755: // "scratch" now holds an ori
instruction. Extract the immediscratche.
On 2012/01/20 11:50:41, Vyacheslav Egorov wrote:
> immediscratche ?

A search and replace attempt gone horribly wrong.
Done.

[Vyacheslav Egorov (Chromium), 2012-01-20 12:19:06]

LGTM.

Erik will land it for you.