Uses slt and sltu for signed vs. unsigned comparison by the MIPS assembler.

Uses slt and sltu for signed vs. unsigned comparison by the MIPS assembler.

Previously, the sequence subu/b{gt,le,...}z was used, which fails to
distinguish between signed and unsigned comparison.

R=regis@google.com

Committed: https://code.google.com/p/dart/source/detail?r=22159

[zra, 2013-04-29 23:56:09]

Hopefully this will fix the problem with the timeout in the interrupt tests.
Before they were doing subu/bltz for the comparison of the stack pointer, which
doesn't always work. I've re-enabled them in vm.status.

[regis, 2013-04-30 00:14:50]

LGTM with one comment about naming conventions.

https://codereview.chromium.org/14556002/diff/1/runtime/vm/assembler_mips.h
File runtime/vm/assembler_mips.h (right):

https://codereview.chromium.org/14556002/diff/1/runtime/vm/assembler_mips.h#n...
runtime/vm/assembler_mips.h:722: void BranchGreater(Register rd, Register rs,
Label* l) {
I suppose this is an unsigned compare?

The name is misleading if you are used to Intel terminology: 'less' and
'greater' imply signed comparison, whereas 'below' and 'above' imply unsigned
comparison.
The terms 'lower' and 'higher' ('or same') are used for unsigned comparison on
ARM.

I think we could avoid future bugs if we make this explicit
(BranchUnsignedGreater) or if we use the Intel conventions (BranchAbove).

[zra, 2013-04-30 00:30:07]

https://codereview.chromium.org/14556002/diff/1/runtime/vm/assembler_mips.h
File runtime/vm/assembler_mips.h (right):

https://codereview.chromium.org/14556002/diff/1/runtime/vm/assembler_mips.h#n...
runtime/vm/assembler_mips.h:722: void BranchGreater(Register rd, Register rs,
Label* l) {
On 2013/04/30 00:14:50, regis wrote:
> I suppose this is an unsigned compare?
> 
> The name is misleading if you are used to Intel terminology: 'less' and
> 'greater' imply signed comparison, whereas 'below' and 'above' imply unsigned
> comparison.
> The terms 'lower' and 'higher' ('or same') are used for unsigned comparison on
> ARM.
> 
> I think we could avoid future bugs if we make this explicit
> (BranchUnsignedGreater) or if we use the Intel conventions (BranchAbove).
> 
>  

Done.

[zra, 2013-04-30 00:35:14]

Committed patchset #2 manually as r22159 (presubmit successful).