On MIPS, uses more branch delay slots.

On MIPS, uses more branch delay slots.

Also, tries to reorganize loops to look like what gcc produces.

R=regis@google.com

Committed: https://code.google.com/p/dart/source/detail?r=22262

[regis, 2013-05-01 01:21:46]

LGTM with a suggestion.

https://codereview.chromium.org/14733002/diff/1/runtime/vm/assembler_mips.h
File runtime/vm/assembler_mips.h (right):

https://codereview.chromium.org/14733002/diff/1/runtime/vm/assembler_mips.h#n...
runtime/vm/assembler_mips.h:176: void LeaveStubFrameSPRestore(bool uses_pp =
false);
I think it would be clearer to leave LeaveStubFrame untouched (it would restore
sp as before) and add instead a LeaveStubFrameAndRet that would make use of the
delay slot.

https://codereview.chromium.org/14733002/diff/1/runtime/vm/stub_code_arm.cc
File runtime/vm/stub_code_arm.cc (right):

https://codereview.chromium.org/14733002/diff/1/runtime/vm/stub_code_arm.cc#n...
runtime/vm/stub_code_arm.cc:520: __ b(&loop, CC);  // Loop until R1 == R7.
Thanks!

https://codereview.chromium.org/14733002/diff/1/runtime/vm/stub_code_mips.cc
File runtime/vm/stub_code_mips.cc (right):

https://codereview.chromium.org/14733002/diff/1/runtime/vm/stub_code_mips.cc#...
runtime/vm/stub_code_mips.cc:987: __ LeaveStubFrameSPRestore();
Why didn't you use the delay slot here?

[zra, 2013-05-01 16:53:17]

https://codereview.chromium.org/14733002/diff/1/runtime/vm/assembler_mips.h
File runtime/vm/assembler_mips.h (right):

https://codereview.chromium.org/14733002/diff/1/runtime/vm/assembler_mips.h#n...
runtime/vm/assembler_mips.h:176: void LeaveStubFrameSPRestore(bool uses_pp =
false);
On 2013/05/01 01:21:46, regis wrote:
> I think it would be clearer to leave LeaveStubFrame untouched (it would
restore
> sp as before) and add instead a LeaveStubFrameAndRet that would make use of
the
> delay slot.

Done.

https://codereview.chromium.org/14733002/diff/1/runtime/vm/stub_code_mips.cc
File runtime/vm/stub_code_mips.cc (right):

https://codereview.chromium.org/14733002/diff/1/runtime/vm/stub_code_mips.cc#...
runtime/vm/stub_code_mips.cc:987: __ LeaveStubFrameSPRestore();
On 2013/05/01 01:21:46, regis wrote:
> Why didn't you use the delay slot here?

Missed it. Thanks!

[zra, 2013-05-01 16:56:36]

Committed patchset #2 manually as r22262 (presubmit successful).