[runtime] Implement %_ToLength via ToLengthStub.

[runtime] Implement %_ToLength via ToLengthStub.

Use %_ToLength for TO_LENGTH, implemented via a ToLengthStub
that supports a fast path for small integers. Everything else is still
handled in the runtime.

CQ_INCLUDE_TRYBOTS=tryserver.v8:v8_linux_nosnap_rel
BUG=v8:4494
LOG=n

Committed: https://crrev.com/e678a0f9a961165ab43cd1a6080fcb11995b701a
Cr-Commit-Position: refs/heads/master@{#31358}

[Benedikt Meurer, 2015-10-18 16:40:12]

Patchset #2 (id:20001) has been deleted

[Benedikt Meurer, 2015-10-18 16:40:19]

Patchset #2 (id:40001) has been deleted

[Benedikt Meurer, 2015-10-18 17:15:06]

Patchset #3 (id:80001) has been deleted

[Benedikt Meurer, 2015-10-18 17:15:15]

Patchset #1 (id:1) has been deleted

[Benedikt Meurer, 2015-10-18 17:15:26]

Patchset #1 (id:60001) has been deleted

[Benedikt Meurer, 2015-10-18 17:29:48]

Patchset #1 (id:100001) has been deleted

[Benedikt Meurer, 2015-10-18 17:31:52]

The CQ bit was checked by bmeurer@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2015-10-18 17:31:56]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1412963002/120001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1412963002/120001

[Benedikt Meurer, 2015-10-18 17:32:19]

The CQ bit was unchecked by bmeurer@chromium.org

[Benedikt Meurer, 2015-10-18 17:32:24]

The CQ bit was checked by bmeurer@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2015-10-18 17:32:42]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1412963002/120001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1412963002/120001

[commit-bot: I haz the power, 2015-10-18 18:46:46]

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

[commit-bot: I haz the power, 2015-10-18 18:46:46]

Dry run: Try jobs failed on following builders:
  v8_linux_nosnap_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_nosnap_rel/builds/...)

[Benedikt Meurer, 2015-10-19 04:05:50]

Patchset #1 (id:120001) has been deleted

[Benedikt Meurer, 2015-10-19 04:05:59]

The CQ bit was checked by bmeurer@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2015-10-19 04:06:09]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1412963002/140001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1412963002/140001

[Benedikt Meurer, 2015-10-19 04:19:52]

Patchset #1 (id:140001) has been deleted

[Benedikt Meurer, 2015-10-19 04:20:01]

The CQ bit was checked by bmeurer@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2015-10-19 04:20:10]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1412963002/160001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1412963002/160001

[Benedikt Meurer, 2015-10-19 04:20:49]

bmeurer@chromium.org changed reviewers:
+ jarin@chromium.org, littledan@chromium.org

[Benedikt Meurer, 2015-10-19 04:20:49]

Jaro: PTAL
Dan: FYI

[commit-bot: I haz the power, 2015-10-19 04:57:32]

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

[commit-bot: I haz the power, 2015-10-19 04:57:32]

Dry run: This issue passed the CQ dry run.

[Jarin, 2015-10-19 06:38:02]

jarin@chromium.org changed reviewers:
+ mstarzinger@chromium.org

[Jarin, 2015-10-19 06:50:16]

That's an awful lot of platform specific code for what it tries to do. If it is
really necessary, then lgtm. (Adding Michi as a reviewer as he has more informed
opinion on this.)

https://codereview.chromium.org/1412963002/diff/160001/src/full-codegen/arm/f...
File src/full-codegen/arm/full-codegen-arm.cc (right):

https://codereview.chromium.org/1412963002/diff/160001/src/full-codegen/arm/f...
src/full-codegen/arm/full-codegen-arm.cc:3809: context()->Plug(r0);
I see nearly the same code in EmitToString, EmitToNumber and EmitToName. Could
we somehow unify that?

Also, could you just say PlugTOS() here and move the method(s) into
full-codegen.cc?

[Benedikt Meurer, 2015-10-19 07:08:13]

>
https://codereview.chromium.org/1412963002/diff/160001/src/full-codegen/arm/f...
> src/full-codegen/arm/full-codegen-arm.cc:3809: context()->Plug(r0);
> I see nearly the same code in EmitToString, EmitToNumber and EmitToName. Could
> we somehow unify that?
> 
> Also, could you just say PlugTOS() here and move the method(s) into
> full-codegen.cc?

PlugTOS is not really what you want, as it's about TOS not accumulator. Also the
methods are not exactly the same, because MIPS and MIPS64 are at least different
due to a0 != v0. I'm fine with refactoring some stuff if necessary, but I'd like
to limit the amount of code motion/refactoring in fullcodegen to the bare
minimum for now.

[Jarin, 2015-10-19 07:10:26]

On 2015/10/19 07:08:13, Benedikt Meurer wrote:
> >
>
https://codereview.chromium.org/1412963002/diff/160001/src/full-codegen/arm/f...
> > src/full-codegen/arm/full-codegen-arm.cc:3809: context()->Plug(r0);
> > I see nearly the same code in EmitToString, EmitToNumber and EmitToName.
Could
> > we somehow unify that?
> > 
> > Also, could you just say PlugTOS() here and move the method(s) into
> > full-codegen.cc?
> 
> PlugTOS is not really what you want, as it's about TOS not accumulator. Also
the
> methods are not exactly the same, because MIPS and MIPS64 are at least
different
> due to a0 != v0. I'm fine with refactoring some stuff if necessary, but I'd
like
> to limit the amount of code motion/refactoring in fullcodegen to the bare
> minimum for now.

Yeah I meant Plug(result_register()), but that does not work because of mips.
Unification of EmitTo* would be still useful, but perhaps as a separate CL.

[Benedikt Meurer, 2015-10-19 07:15:34]

On 2015/10/19 07:10:26, Jarin wrote:
> On 2015/10/19 07:08:13, Benedikt Meurer wrote:
> > >
> >
>
https://codereview.chromium.org/1412963002/diff/160001/src/full-codegen/arm/f...
> > > src/full-codegen/arm/full-codegen-arm.cc:3809: context()->Plug(r0);
> > > I see nearly the same code in EmitToString, EmitToNumber and EmitToName.
> Could
> > > we somehow unify that?
> > > 
> > > Also, could you just say PlugTOS() here and move the method(s) into
> > > full-codegen.cc?
> > 
> > PlugTOS is not really what you want, as it's about TOS not accumulator. Also
> the
> > methods are not exactly the same, because MIPS and MIPS64 are at least
> different
> > due to a0 != v0. I'm fine with refactoring some stuff if necessary, but I'd
> like
> > to limit the amount of code motion/refactoring in fullcodegen to the bare
> > minimum for now.
> 
> Yeah I meant Plug(result_register()), but that does not work because of mips.
> Unification of EmitTo* would be still useful, but perhaps as a separate CL.

Ok, so that will be done in a follow up CL, and this CL is about the %_ToLength
unblocking for Dan only.

[Dan Ehrenberg, 2015-10-19 07:39:52]

lgtm

At a high level, looks good to me, that this should hopefully solve the ToLength
performance issue.

[Michael Starzinger, 2015-10-19 08:21:08]

LGTM, as much as it hurts to see a new architecture-specific stub being added, I
agree that this one probably makes sense.

https://codereview.chromium.org/1412963002/diff/160001/src/full-codegen/arm/f...
File src/full-codegen/arm/full-codegen-arm.cc (right):

https://codereview.chromium.org/1412963002/diff/160001/src/full-codegen/arm/f...
src/full-codegen/arm/full-codegen-arm.cc:3809: context()->Plug(r0);
On 2015/10/19 06:50:16, Jarin wrote:
> I see nearly the same code in EmitToString, EmitToNumber and EmitToName. Could
> we somehow unify that?
> 
> Also, could you just say PlugTOS() here and move the method(s) into
> full-codegen.cc?

Nah, PlugTOS targets top-of-stack, not the accumulator/result register.

[Benedikt Meurer, 2015-10-19 08:33:32]

The CQ bit was checked by bmeurer@chromium.org

[commit-bot: I haz the power, 2015-10-19 08:33:45]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1412963002/160001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1412963002/160001

[commit-bot: I haz the power, 2015-10-19 08:35:05]

Committed patchset #1 (id:160001)

[commit-bot: I haz the power, 2015-10-19 08:35:27]

Patchset 1 (id:??) landed as
https://crrev.com/e678a0f9a961165ab43cd1a6080fcb11995b701a
Cr-Commit-Position: refs/heads/master@{#31358}