Adds mrc instruction to arm simulator, assembler, disassembler.

Adds mrc instruction to arm simulator, assembler, disassembler.
This instruction will be used to query an arm processor to
determine whether it supports the sdiv and udiv instructions.

Also adds a flag --sim_has_int_div to the arm simulator,
which changes the result of the mrc instruction so we can
test both ways.
Committed: https://code.google.com/p/dart/source/detail?r=19439

[regis, 2013-03-04 21:54:52]

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_arm.cc
File runtime/vm/assembler_arm.cc (right):

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_arm.c...
runtime/vm/assembler_arm.cc:1120: int32_t CRn, int32_t CRm, int32_t opc2,
Condition cond) {
crn, crm

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_arm.h
File runtime/vm/assembler_arm.h (right):

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_arm.h...
runtime/vm/assembler_arm.h:491: // MRC
Better to spell out MRC: Move to ARM core register from Coprocessor.

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_arm.h...
runtime/vm/assembler_arm.h:493: int32_t CRn, int32_t CRm, int32_t opc2,
Condition cond = AL);
No upper case for locals.

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_arm_t...
File runtime/vm/assembler_arm_test.cc (right):

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_arm_t...
runtime/vm/assembler_arm_test.cc:1367: // If we are in the simulator, check that
HasDiv returns the expected result
You can run this test even if we are running on real hardware.

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_arm_t...
runtime/vm/assembler_arm_test.cc:1368: #ifdef USING_SIMULATOR
We prefer the form #if defined(USING_SIMULATOR), which makes it easier to add
terms.
I may have missed some in my previous submits. I'll check.

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_arm_t...
runtime/vm/assembler_arm_test.cc:1378: if (FLAG_sim_has_int_div) {
A flag is problematic to maintain when running tests.
Instead of a flag, add a new static integer_division_supported() in class
CPUFeatures. See assembler_arm.h.
This call will be used to decide what instructions to emit on real hardware.
Now, if we want to test mrc under the simulator, the tests should set the
feature directly in CPUFeatures, thereby telling the simulator what to do. Let's
discuss if this explanation is not clear.

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_arm_t...
runtime/vm/assembler_arm_test.cc:1384: #endif
We like to repeat the #if as a comment, unless they are very close to each
other:
#endif  // defined(USING_SIMULATOR)

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_macro...
File runtime/vm/assembler_macros_arm.cc (right):

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_macro...
runtime/vm/assembler_macros_arm.cc:80: 
Move code to CPUFeatures::InitOnce().

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_macro...
File runtime/vm/assembler_macros_arm.h (right):

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_macro...
runtime/vm/assembler_macros_arm.h:49: static void HasDiv(Assembler* assembler,
Register rd);
This method is not necessary. The code testing for this feature should be
executed once in CPUFeatures::InitOnce().

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/simulator_arm.cc
File runtime/vm/simulator_arm.cc (right):

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/simulator_arm.c...
runtime/vm/simulator_arm.cc:27: DEFINE_FLAG(bool, sim_has_int_div, false,
"Simulator implements sdiv and udiv");
not necessary.

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/simulator_arm.c...
runtime/vm/simulator_arm.cc:2592: if (FLAG_sim_has_int_div) {
Test CPUFeatures instead.

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/simulator_arm.h
File runtime/vm/simulator_arm.h (right):

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/simulator_arm.h...
runtime/vm/simulator_arm.h:26: 
Not necessary.

[srdjan, 2013-03-04 22:02:17]

DBC

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_arm.h
File runtime/vm/assembler_arm.h (right):

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_arm.h...
runtime/vm/assembler_arm.h:491: // MRC
Also terminate all comments with a period.

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_arm_t...
File runtime/vm/assembler_arm_test.cc (right):

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/assembler_arm_t...
runtime/vm/assembler_arm_test.cc:1367: // If we are in the simulator, check that
HasDiv returns the expected result
ditto

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/constants_arm.h
File runtime/vm/constants_arm.h (right):

https://codereview.chromium.org/12378080/diff/7004/runtime/vm/constants_arm.h...
runtime/vm/constants_arm.h:278: // MRC instruction offset field encoding
ditto

[zra, 2013-03-05 00:42:30]

Okay. Please have a second look.

Thanks!

[regis, 2013-03-05 00:59:22]

LGTM!

[zra, 2013-03-05 01:03:27]

Committed patchset #7 manually as r19439 (presubmit successful).