Improve integer division on ARM in favor of power of 2 constant divisor

Improve integer division on ARM in favor of power of 2 constant divisor
BUG=none
TEST=none

Committed: https://code.google.com/p/v8/source/detail?r=13819

[rkrithiv, 2013-01-23 01:03:38]

Hi Danno, I would appreciate it if you could review my change. Thanks!

[danno, 2013-01-23 13:02:20]

I'm handing this off of Ulan.

However, I have two pieces of feedback:
- How does this patch affect the standard benchmarks. Is the improvement
measurable?
- You should add specific test cases that trigger all of the cases that your CL
improves, including the edge cases of deopting on odd values, handling of
negative devisors and handling of -0 (or at least ensure that existing test
cases cover them)

[rkrithiv, 2013-02-25 19:06:22]

On 2013/01/23 13:02:20, danno wrote:
> I'm handing this off of Ulan.
> 
> However, I have two pieces of feedback:
> - How does this patch affect the standard benchmarks. Is the improvement
> measurable?
This change improves SunSpider (around 1%).

> - You should add specific test cases that trigger all of the cases that your
CL
> improves, including the edge cases of deopting on odd values, handling of
> negative devisors and handling of -0 (or at least ensure that existing test
> cases cover them)

The test cases in math-floor-of-div.js cover most of the cases, like
test_div_deopt_minus_zero, test_div_deopt_overflow, test_div_deopt_minus_zero_v,
test_div2. Those tests are enabled for ARM now.

[ulan, 2013-02-27 14:12:32]

One style nit, otherwise looks good.

https://codereview.chromium.org/12052032/diff/1/src/arm/lithium-codegen-arm.cc
File src/arm/lithium-codegen-arm.cc (right):

https://codereview.chromium.org/12052032/diff/1/src/arm/lithium-codegen-arm.c...
src/arm/lithium-codegen-arm.cc:1371: class DeferredDivI: public LDeferredCode {
Please move this class up to the beginning of the function.

[rkrithiv, 2013-02-27 20:33:36]

I have moved the class up in patch set 3.Thanks!

[ulan, 2013-03-01 09:10:42]

LGTM, I will land it for you.

[ulan, 2013-03-01 09:58:20]

I can't find your email in the list of people who signed "Committers License
Agreement"

Could you please sign it following the instructions in the link?
http://code.google.com/p/v8/wiki/Contributing

[rkrithiv, 2013-03-04 19:10:32]

On 2013/03/01 09:58:20, ulan wrote:
> I can't find your email in the list of people who signed "Committers License
> Agreement"
> 
> Could you please sign it following the instructions in the link?
> http://code.google.com/p/v8/wiki/Contributing

I am a member of the Code Aurora Forum and am mentioned in that CLA. Also I have
contributed patches before and am in the AUTHORS list.

[ulan, 2013-03-05 08:48:06]

Committed patchset #3 manually as r13819 (presubmit successful).

[ulan, 2013-03-05 08:49:07]

Thanks, landed.