Support FrameLookup vm test on ARM, requiring among other things:

Support FrameLookup vm test on ARM, requiring among other things:
Maintain a separate top_exit_frame_info for simulated frames and use it to
verify that longjumps are safe.
Make sure unwinding of scopes works in the presence of a simulator.
Support inline allocation of objects.
Support checking of inline cache and instance calls.
Support subtype test cache.
Support (some) equality checks.
Support for (some) conditional branches.
Support for pool pointer setup in stubs.
Committed: https://code.google.com/p/dart/source/detail?r=20890

[zra, 2013-04-03 17:39:47]

https://codereview.chromium.org/13502002/diff/1/runtime/vm/intermediate_langu...
File runtime/vm/intermediate_language_arm.cc (right):

https://codereview.chromium.org/13502002/diff/1/runtime/vm/intermediate_langu...
runtime/vm/intermediate_language_arm.cc:373: __ Bind(&false_label);
Nothing branches to this label.

https://codereview.chromium.org/13502002/diff/1/runtime/vm/intermediate_langu...
runtime/vm/intermediate_language_arm.cc:1226: // TODO(vegorov): should be
eliminated earlier by constant propagation.
Unless constant propagation/folding is really bad, maybe there should be an
assert here instead?

https://codereview.chromium.org/13502002/diff/1/runtime/vm/intermediate_langu...
runtime/vm/intermediate_language_arm.cc:1265: // TODO(vegorov): should be
eliminated earlier by constant propagation.
Here too?

https://codereview.chromium.org/13502002/diff/1/runtime/vm/stub_code_arm.cc
File runtime/vm/stub_code_arm.cc (right):

https://codereview.chromium.org/13502002/diff/1/runtime/vm/stub_code_arm.cc#n...
runtime/vm/stub_code_arm.cc:377: const int kInlineInstanceSize = 12;
Why 12? Should this be a flag?

https://codereview.chromium.org/13502002/diff/1/runtime/vm/stub_code_arm.cc#n...
runtime/vm/stub_code_arm.cc:938: // R6: ICData.
R5?

https://codereview.chromium.org/13502002/diff/1/runtime/vm/stub_code_arm.cc#n...
runtime/vm/stub_code_arm.cc:945: #if 0
Maybe rm from the CL?

[regis, 2013-04-03 20:06:07]

Thanks!

https://codereview.chromium.org/13502002/diff/1/runtime/vm/intermediate_langu...
File runtime/vm/intermediate_language_arm.cc (right):

https://codereview.chromium.org/13502002/diff/1/runtime/vm/intermediate_langu...
runtime/vm/intermediate_language_arm.cc:373: __ Bind(&false_label);
On 2013/04/03 17:39:47, zra wrote:
> Nothing branches to this label.

Removed here and on other platforms.

https://codereview.chromium.org/13502002/diff/1/runtime/vm/intermediate_langu...
runtime/vm/intermediate_language_arm.cc:1226: // TODO(vegorov): should be
eliminated earlier by constant propagation.
On 2013/04/03 17:39:47, zra wrote:
> Unless constant propagation/folding is really bad, maybe there should be an
> assert here instead?

I cannot answer for Slava :-)
Since there is already a TODO, I will leave it as is.

https://codereview.chromium.org/13502002/diff/1/runtime/vm/intermediate_langu...
runtime/vm/intermediate_language_arm.cc:1265: // TODO(vegorov): should be
eliminated earlier by constant propagation.
On 2013/04/03 17:39:47, zra wrote:
> Here too?

ditto

https://codereview.chromium.org/13502002/diff/1/runtime/vm/stub_code_arm.cc
File runtime/vm/stub_code_arm.cc (right):

https://codereview.chromium.org/13502002/diff/1/runtime/vm/stub_code_arm.cc#n...
runtime/vm/stub_code_arm.cc:377: const int kInlineInstanceSize = 12;
On 2013/04/03 17:39:47, zra wrote:
> Why 12? Should this be a flag?

The flag would only be useful until the value is tuned. We try to keep the
number of flags low. 12 was copied from Intel.

https://codereview.chromium.org/13502002/diff/1/runtime/vm/stub_code_arm.cc#n...
runtime/vm/stub_code_arm.cc:938: // R6: ICData.
On 2013/04/03 17:39:47, zra wrote:
> R5?

Since the stub is unimplemented, the comment is meaningless. Removed.

https://codereview.chromium.org/13502002/diff/1/runtime/vm/stub_code_arm.cc#n...
runtime/vm/stub_code_arm.cc:945: #if 0
On 2013/04/03 17:39:47, zra wrote:
> Maybe rm from the CL?

Oops. Removed.

[zra, 2013-04-03 20:11:57]

lgtm

[regis, 2013-04-03 20:17:24]

Committed patchset #2 manually as r20890 (presubmit successful).