Implement bigint shift intrinsics on IA32.

Implement bigint shift intrinsics on IA32.
Make shifter register ECX explicit in shld and shrd instructions on IA32.
Fix bad encoding of shld and shrd on X64 and update assembler tests.

Committed: https://code.google.com/p/dart/source/detail?r=45182

[regis, 2015-04-14 19:04:05]

regis@google.com changed reviewers:
+ johnmccutchan@google.com

[regis, 2015-04-14 22:00:32]

On 2015/04/14 19:04:06, regis wrote:

Since using EBP as a general purpose register is causing trouble to the
profiler, I will rewrite this code to use one less register. I also found and
fixed a bad encoding of shld and shrd instructions on X64.
So you can hold off reviewing this cl until I update it.

Thanks!

[regis, 2015-04-14 23:27:44]

On 2015/04/14 22:00:32, regis wrote:
> On 2015/04/14 19:04:06, regis wrote:
> 
> Since using EBP as a general purpose register is causing trouble to the
> profiler, I will rewrite this code to use one less register. I also found and
> fixed a bad encoding of shld and shrd instructions on X64.
> So you can hold off reviewing this cl until I update it.
> 
> Thanks!

PTAL

[regis, 2015-04-14 23:48:16]

On 2015/04/14 23:27:44, regis wrote:
> On 2015/04/14 22:00:32, regis wrote:
> > On 2015/04/14 19:04:06, regis wrote:
> > 
> > Since using EBP as a general purpose register is causing trouble to the
> > profiler, I will rewrite this code to use one less register. I also found
and
> > fixed a bad encoding of shld and shrd instructions on X64.
> > So you can hold off reviewing this cl until I update it.
> > 
> > Thanks!
> 
> PTAL

Since the inner loop of the rewritten Bigint_rsh on IA32 is one instruction
shorter, I ported it back to X64.

[srdjan, 2015-04-15 05:05:37]

srdjan@google.com changed reviewers:
+ srdjan@google.com

[srdjan, 2015-04-15 05:05:37]

https://codereview.chromium.org/1089003002/diff/60001/runtime/vm/intrinsifier...
File runtime/vm/intrinsifier_ia32.cc (right):

https://codereview.chromium.org/1089003002/diff/60001/runtime/vm/intrinsifier...
runtime/vm/intrinsifier_ia32.cc:9: // Dart method was intrinsified.
Please also add comment that EBP may not be touched.

[Cutch, 2015-04-15 17:56:36]

LGTMwC

https://codereview.chromium.org/1089003002/diff/60001/runtime/vm/intrinsifier...
File runtime/vm/intrinsifier_ia32.cc (right):

https://codereview.chromium.org/1089003002/diff/60001/runtime/vm/intrinsifier...
runtime/vm/intrinsifier_ia32.cc:9: // Dart method was intrinsified.
On 2015/04/15 05:05:37, srdjan wrote:
> Please also add comment that EBP may not be touched.

Also on intrinsifer_x64 RBP may not be touched.

https://codereview.chromium.org/1089003002/diff/60001/runtime/vm/intrinsifier...
File runtime/vm/intrinsifier_x64.cc (right):

https://codereview.chromium.org/1089003002/diff/60001/runtime/vm/intrinsifier...
runtime/vm/intrinsifier_x64.cc:5: #include "vm/globals.h"  // Needed here to get
TARGET_ARCH_X64.
comment here about which registers can be modified (RBP cannot be)

[regis, 2015-04-15 20:46:00]

Thanks!

https://codereview.chromium.org/1089003002/diff/60001/runtime/vm/intrinsifier...
File runtime/vm/intrinsifier_ia32.cc (right):

https://codereview.chromium.org/1089003002/diff/60001/runtime/vm/intrinsifier...
runtime/vm/intrinsifier_ia32.cc:9: // Dart method was intrinsified.
On 2015/04/15 05:05:37, srdjan wrote:
> Please also add comment that EBP may not be touched.

Done.

https://codereview.chromium.org/1089003002/diff/60001/runtime/vm/intrinsifier...
runtime/vm/intrinsifier_ia32.cc:9: // Dart method was intrinsified.
On 2015/04/15 17:56:36, Cutch wrote:
> On 2015/04/15 05:05:37, srdjan wrote:
> > Please also add comment that EBP may not be touched.
> 
> Also on intrinsifer_x64 RBP may not be touched.

Done.

https://codereview.chromium.org/1089003002/diff/60001/runtime/vm/intrinsifier...
File runtime/vm/intrinsifier_x64.cc (right):

https://codereview.chromium.org/1089003002/diff/60001/runtime/vm/intrinsifier...
runtime/vm/intrinsifier_x64.cc:5: #include "vm/globals.h"  // Needed here to get
TARGET_ARCH_X64.
On 2015/04/15 17:56:36, Cutch wrote:
> comment here about which registers can be modified (RBP cannot be)

I have edited the existing comment on line 22 and added similar comments to all
intrinsifier_*.cc files.

[regis, 2015-04-15 20:46:35]

Committed patchset #5 (id:80001) manually as r45182 (presubmit successful).