Fixes a bug in MIPS simulator runtime call handling.

Fixes a bug in MIPS simulator runtime call handling.

Also:
- Fixes a bug in the MIPS SimulatorDebugger that
  causes a crash if the pc is an illegal address
- Reenables the FindCodeObject test on MIPS when
  checking is off
Committed: https://code.google.com/p/dart/source/detail?r=20924

[zra, 2013-04-04 15:52:26]

This fixes the problem we saw on Mac yesterday. My initial idea that there was a
bug in frame alignment across both MIPS and ARM was wrong. It was just a bug
caused by the MIPS simulator returning to one instruction too far after a
runtime call. The instruction being skipped in CallLeafRuntimeStubCode was for
restoring the stack pointer. Skipping this instruction was only working by
chance on Linux since the frame alignment requirement is smaller than on Mac.

[regis, 2013-04-04 15:58:27]

LGTM

[srdjan, 2013-04-04 16:19:03]

lgtm

https://codereview.chromium.org/13489008/diff/1/runtime/tests/vm/vm.status
File runtime/tests/vm/vm.status (right):

https://codereview.chromium.org/13489008/diff/1/runtime/tests/vm/vm.status#ne...
runtime/tests/vm/vm.status:79: cc/FindCodeObject: Skip
Is it too slow in checked mode?

[zra, 2013-04-04 16:23:41]

Thanks!

https://codereview.chromium.org/13489008/diff/1/runtime/tests/vm/vm.status
File runtime/tests/vm/vm.status (right):

https://codereview.chromium.org/13489008/diff/1/runtime/tests/vm/vm.status#ne...
runtime/tests/vm/vm.status:79: cc/FindCodeObject: Skip
On 2013/04/04 16:19:03, srdjan wrote:
> Is it too slow in checked mode?

The type checks needed for this test aren't yet implemented on MIPS.

[zra, 2013-04-04 16:56:04]

Committed patchset #1 manually as r20924 (presubmit successful).