[wasm] Do float constant folding if the origin is not wasm

[wasm] Do float constant folding if the origin is not wasm

I removed some constant folding optimizations for float instruction in
https://codereview.chromium.org/2647353007 because they were incorrect
if the input was a signalling NaN. Removing these optimizations, however
had an unexpectedly big impact on asm.js performance. With this CL I
restore the optimizations again when the source origin is not wasm. In
JavaScript signalling NaNs are not observable and therefore the
optimizations are correct.

R=titzer@chromium.org
BUG=chromium:686654
Review-Url: https://codereview.chromium.org/2666903002
Cr-Commit-Position: refs/heads/master@{#42850}
Committed: https://chromium.googlesource.com/v8/v8/+/3b2301650eba47706f50104ecd13517dbb20bc9f

[ahaas, 2017-01-31 10:46:58]

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

[commit-bot: I haz the power, 2017-01-31 10:46:59]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-01-31 10:50:29]

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

[commit-bot: I haz the power, 2017-01-31 10:50:30]

Dry run: Try jobs failed on following builders:
  v8_linux64_avx2_rel_ng on master.tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux64_avx2_rel_ng/buil...)
  v8_linux64_verify_csa_rel_ng on master.tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux64_verify_csa_rel_n...)
  v8_linux_verify_csa_rel_ng on master.tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_verify_csa_rel_ng/...)
  v8_win_rel_ng on master.tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_win_rel_ng/builds/21887)

[ahaas, 2017-01-31 11:25:57]

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

[commit-bot: I haz the power, 2017-01-31 11:26:05]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-01-31 11:51:05]

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

[commit-bot: I haz the power, 2017-01-31 11:51:05]

Dry run: This issue passed the CQ dry run.

[titzer, 2017-01-31 17:25:49]

https://codereview.chromium.org/2666903002/diff/20001/src/compiler/machine-op...
File src/compiler/machine-operator-reducer.h (right):

https://codereview.chromium.org/2666903002/diff/20001/src/compiler/machine-op...
src/compiler/machine-operator-reducer.h:27: explicit
MachineOperatorReducer(JSGraph* jsgraph, bool wasm_origin = false);
Can we call this flag something like "preserve_nan_bits"?

[titzer, 2017-01-31 19:18:05]

On 2017/01/31 17:25:49, titzer wrote:
>
https://codereview.chromium.org/2666903002/diff/20001/src/compiler/machine-op...
> File src/compiler/machine-operator-reducer.h (right):
> 
>
https://codereview.chromium.org/2666903002/diff/20001/src/compiler/machine-op...
> src/compiler/machine-operator-reducer.h:27: explicit
> MachineOperatorReducer(JSGraph* jsgraph, bool wasm_origin = false);
> Can we call this flag something like "preserve_nan_bits"?

Or "preserve signaling NaNs"--whichever is closer to the truth.

LGTM otherwise.

[ahaas, 2017-01-31 19:25:55]

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

[commit-bot: I haz the power, 2017-01-31 19:26:06]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[ahaas, 2017-01-31 19:26:48]

On 2017/01/31 at 19:18:05, titzer wrote:
> On 2017/01/31 17:25:49, titzer wrote:
> >
https://codereview.chromium.org/2666903002/diff/20001/src/compiler/machine-op...
> > File src/compiler/machine-operator-reducer.h (right):
> > 
> >
https://codereview.chromium.org/2666903002/diff/20001/src/compiler/machine-op...
> > src/compiler/machine-operator-reducer.h:27: explicit
> > MachineOperatorReducer(JSGraph* jsgraph, bool wasm_origin = false);
> > Can we call this flag something like "preserve_nan_bits"?
> 
> Or "preserve signaling NaNs"--whichever is closer to the truth.
> 
> LGTM otherwise.

Done. I call the flag allow_signalling_nan now.

[commit-bot: I haz the power, 2017-01-31 19:51:40]

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

[commit-bot: I haz the power, 2017-01-31 19:51:41]

Dry run: This issue passed the CQ dry run.

[ahaas, 2017-02-01 07:56:30]

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

[commit-bot: I haz the power, 2017-02-01 07:56:41]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-02-01 08:22:25]

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

[commit-bot: I haz the power, 2017-02-01 08:22:26]

Dry run: This issue passed the CQ dry run.

[ahaas, 2017-02-01 11:49:08]

The CQ bit was checked by ahaas@chromium.org

[ahaas, 2017-02-01 11:49:08]

The patchset sent to the CQ was uploaded after l-g-t-m from titzer@chromium.org
Link to the patchset: https://codereview.chromium.org/2666903002/#ps60001
(title: "Allow signalling NaNs by default")

[commit-bot: I haz the power, 2017-02-01 11:49:17]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-02-01 11:52:00]

CQ is committing da patch.

Bot data: {"patchset_id": 60001, "attempt_start_ts": 1485949748212880,
"parent_rev": "42011d29975898c3648241749cf1e96eac074b85", "commit_rev":
"3b2301650eba47706f50104ecd13517dbb20bc9f"}

[commit-bot: I haz the power, 2017-02-01 11:52:03]

Description was changed from

==========
[wasm] Do float constant folding if the origin is not wasm

I removed some constant folding optimizations for float instruction in
https://codereview.chromium.org/2647353007 because they were incorrect
if the input was a signalling NaN. Removing these optimizations, however
had an unexpectedly big impact on asm.js performance. With this CL I
restore the optimizations again when the source origin is not wasm. In
JavaScript signalling NaNs are not observable and therefore the
optimizations are correct.

R=titzer@chromium.org
BUG=chromium:686654
==========

to

==========
[wasm] Do float constant folding if the origin is not wasm

I removed some constant folding optimizations for float instruction in
https://codereview.chromium.org/2647353007 because they were incorrect
if the input was a signalling NaN. Removing these optimizations, however
had an unexpectedly big impact on asm.js performance. With this CL I
restore the optimizations again when the source origin is not wasm. In
JavaScript signalling NaNs are not observable and therefore the
optimizations are correct.

R=titzer@chromium.org
BUG=chromium:686654

Review-Url: https://codereview.chromium.org/2666903002
Cr-Commit-Position: refs/heads/master@{#42850}
Committed:
https://chromium.googlesource.com/v8/v8/+/3b2301650eba47706f50104ecd13517dbb2...
==========

[commit-bot: I haz the power, 2017-02-01 11:52:05]

Committed patchset #4 (id:60001) as
https://chromium.googlesource.com/v8/v8/+/3b2301650eba47706f50104ecd13517dbb2...