[arm] Fix test failures on old architectures.

[arm] Fix test failures on old architectures.

This mostly affects ARMv6, but also fixes some failures on ARMv7 when
hardware integer division is not available.

- Fix a case where a CodePatcher accumulates literal pool entries, but
  does not actually emit them.
- Don't treat division as safe if we can't use the hardware instruction.
  Our fallback implementation returns the wrong result if the divisor is
  zero.
- Support deoptimization tables bigger than 8 bits. (kMaxNumberOfEntries
  requires a 16-bit table index.)
- Correct a TurboFan instruction encoding to encode the Operand2 mode.

BUG=

Committed: https://crrev.com/886b259e229572257e6271c3f6d1532c510e15ef
Cr-Commit-Position: refs/heads/master@{#36636}

[jbramley, 2016-05-31 15:29:32]

jacob.bramley@arm.com changed reviewers:
+ bmeurer@chromium.org, epertoso@chromium.org, jarin@chromium.org

[epertoso, 2016-05-31 15:35:29]

lgtm

[jbramley, 2016-06-01 08:05:47]

The CQ bit was checked by jacob.bramley@arm.com

[commit-bot: I haz the power, 2016-06-01 08:05:49]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/2021343002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/2021343002/1

[commit-bot: I haz the power, 2016-06-01 08:11:22]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-06-01 08:11:23]

Try jobs failed on following builders:
  v8_presubmit on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_presubmit/builds/16268)

[jbramley, 2016-06-01 08:27:10]

On 2016/06/01 08:11:23, commit-bot: I haz the power wrote:
> Try jobs failed on following builders:
>   v8_presubmit on tryserver.v8 (JOB_FAILED,
> http://build.chromium.org/p/tryserver.v8/builders/v8_presubmit/builds/16268)

** Presubmit ERRORS **
Missing LGTM from an OWNER for these files:
    src/compiler/arm/instruction-selector-arm.cc

Can someone help with that please?

Thanks,
Jacob

[jbramley, 2016-06-01 09:01:57]

The CQ bit was checked by jacob.bramley@arm.com

[commit-bot: I haz the power, 2016-06-01 09:02:07]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/2021343002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/2021343002/1

[epertoso, 2016-06-01 09:02:36]

On 2016/06/01 at 08:27:10, jacob.bramley wrote:
> On 2016/06/01 08:11:23, commit-bot: I haz the power wrote:
> > Try jobs failed on following builders:
> >   v8_presubmit on tryserver.v8 (JOB_FAILED,
> > http://build.chromium.org/p/tryserver.v8/builders/v8_presubmit/builds/16268)
> 
> ** Presubmit ERRORS **
> Missing LGTM from an OWNER for these files:
>     src/compiler/arm/instruction-selector-arm.cc
> 
> Can someone help with that please?
> 
> Thanks,
> Jacob

https://codereview.chromium.org/2025273002 should have fixed the issue.

[commit-bot: I haz the power, 2016-06-01 09:03:49]

Committed patchset #1 (id:1)

[jbramley, 2016-06-01 09:03:53]

> https://codereview.chromium.org/2025273002 should have fixed the issue.

It looks promising. Thanks!

[commit-bot: I haz the power, 2016-06-01 09:04:40]

Description was changed from

==========
[arm] Fix test failures on old architectures.

This mostly affects ARMv6, but also fixes some failures on ARMv7 when
hardware integer division is not available.

- Fix a case where a CodePatcher accumulates literal pool entries, but
  does not actually emit them.
- Don't treat division as safe if we can't use the hardware instruction.
  Our fallback implementation returns the wrong result if the divisor is
  zero.
- Support deoptimization tables bigger than 8 bits. (kMaxNumberOfEntries
  requires a 16-bit table index.)
- Correct a TurboFan instruction encoding to encode the Operand2 mode.

BUG=
==========

to

==========
[arm] Fix test failures on old architectures.

This mostly affects ARMv6, but also fixes some failures on ARMv7 when
hardware integer division is not available.

- Fix a case where a CodePatcher accumulates literal pool entries, but
  does not actually emit them.
- Don't treat division as safe if we can't use the hardware instruction.
  Our fallback implementation returns the wrong result if the divisor is
  zero.
- Support deoptimization tables bigger than 8 bits. (kMaxNumberOfEntries
  requires a 16-bit table index.)
- Correct a TurboFan instruction encoding to encode the Operand2 mode.

BUG=

Committed: https://crrev.com/886b259e229572257e6271c3f6d1532c510e15ef
Cr-Commit-Position: refs/heads/master@{#36636}
==========

[commit-bot: I haz the power, 2016-06-01 09:04:41]

Patchset 1 (id:??) landed as
https://crrev.com/886b259e229572257e6271c3f6d1532c510e15ef
Cr-Commit-Position: refs/heads/master@{#36636}