Adds muls, sdiv, udiv instructions to ARM simulator, assembler, disassembler

Adds muls, sdiv, udiv instructions to ARM simulator, assembler, disassembler
Committed: https://code.google.com/p/dart/source/detail?r=19511

[regis, 2013-03-05 19:11:34]

LGTM with comments

https://codereview.chromium.org/12459005/diff/5003/runtime/vm/assembler_arm_t...
File runtime/vm/assembler_arm_test.cc (right):

https://codereview.chromium.org/12459005/diff/5003/runtime/vm/assembler_arm_t...
runtime/vm/assembler_arm_test.cc:1471: }
I would add tests for division by zero as well as 0x80000000 / 0xFFFFFFFF ==
0x80000000. See spec. None of these should cause the host (simulator) to throw.
See my other comment.

https://codereview.chromium.org/12459005/diff/5003/runtime/vm/simulator_arm.cc
File runtime/vm/simulator_arm.cc (right):

https://codereview.chromium.org/12459005/diff/5003/runtime/vm/simulator_arm.c...
runtime/vm/simulator_arm.cc:2003: Register rm = instr->RmField();
You need to test if rm is zero. In that case, the result should be set to zero,
assuming we expect IntegerZeroDivideTrappingEnabled() to be false. See spec.
We may behave differently if the default behavior on the target is different. I
would add a TODO(zra): Revisit once we test on ARM hardware.
In any case, we do not want the host to throw.

[zra, 2013-03-05 19:52:02]

On 2013/03/05 19:11:34, regis wrote:
>
https://codereview.chromium.org/12459005/diff/5003/runtime/vm/assembler_arm_t...
> runtime/vm/assembler_arm_test.cc:1471: }
> I would add tests for division by zero as well as 0x80000000 / 0xFFFFFFFF ==
> 0x80000000. See spec. None of these should cause the host (simulator) to
throw.
> See my other comment.
>  
>
https://codereview.chromium.org/12459005/diff/5003/runtime/vm/simulator_arm.c...
> runtime/vm/simulator_arm.cc:2003: Register rm = instr->RmField();
> You need to test if rm is zero. In that case, the result should be set to
zero,
> assuming we expect IntegerZeroDivideTrappingEnabled() to be false. See spec.
> We may behave differently if the default behavior on the target is different.
I
> would add a TODO(zra): Revisit once we test on ARM hardware.
> In any case, we do not want the host to throw.

Oh, right. I checked for these cases when I wrote the intrinsic and then tricked
myself into thinking that I didn't need to do it here. In any case, I've handled
these cases now in the simulator and added tests.

[zra, 2013-03-05 19:53:48]

Committed patchset #3 manually as r19511 (presubmit successful).