[turbofan] Don't treat the hole NaN as constant inside the compiler.

[turbofan] Don't treat the hole NaN as constant inside the compiler.

We use a signaling NaN to represent the hole in
FAST_HOLEY_DOUBLE_ELEMENTS backing stores, but on Intel processors, the
C++ compiler may decide to (or be forced to due to calling conventions)
use X87 registers for double values. However transfering to X87
registers automatically quietens the NaNs and there's no way to disable
this. Therefore we should just always load the hole NaN from the canonical
place identified by the address_of_hole_nan external reference instead,
which might even be more efficient in some cases.

R=jarin@chromium.org, jkummerow@chromium.org
BUG=v8:5332
Committed: https://crrev.com/64a7bd3877dec131a7544a695f6a760c46d4ba04
Cr-Commit-Position: refs/heads/master@{#39062}

[Benedikt Meurer, 2016-09-01 05:07:55]

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

[commit-bot: I haz the power, 2016-09-01 05:08:03]

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

[Benedikt Meurer, 2016-09-01 05:08:17]

Jaro: PTAL
Jakob: FYI

[commit-bot: I haz the power, 2016-09-01 05:28:13]

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

[commit-bot: I haz the power, 2016-09-01 05:28:14]

Dry run: This issue passed the CQ dry run.

[Jarin, 2016-09-01 05:42:42]

lgtm

[Jarin, 2016-09-01 05:42:43]

lgtm

lgtm

[Benedikt Meurer, 2016-09-01 05:57:01]

The CQ bit was checked by bmeurer@chromium.org

[commit-bot: I haz the power, 2016-09-01 05:57:03]

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, 2016-09-01 06:01:38]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-09-01 06:02:30]

Description was changed from

==========
[turbofan] Don't treat the hole NaN as constant inside the compiler.

We use a signaling NaN to represent the hole in
FAST_HOLEY_DOUBLE_ELEMENTS backing stores, but on Intel processors, the
C++ compiler may decide to (or be forced to due to calling conventions)
use X87 registers for double values. However transfering to X87
registers automatically quietens the NaNs and there's no way to disable
this. Therefore we should just always load the hole NaN from the canonical
place identified by the address_of_hole_nan external reference instead,
which might even be more efficient in some cases.

R=jarin@chromium.org, jkummerow@chromium.org
BUG=v8:5332
==========

to

==========
[turbofan] Don't treat the hole NaN as constant inside the compiler.

We use a signaling NaN to represent the hole in
FAST_HOLEY_DOUBLE_ELEMENTS backing stores, but on Intel processors, the
C++ compiler may decide to (or be forced to due to calling conventions)
use X87 registers for double values. However transfering to X87
registers automatically quietens the NaNs and there's no way to disable
this. Therefore we should just always load the hole NaN from the canonical
place identified by the address_of_hole_nan external reference instead,
which might even be more efficient in some cases.

R=jarin@chromium.org, jkummerow@chromium.org
BUG=v8:5332

Committed: https://crrev.com/64a7bd3877dec131a7544a695f6a760c46d4ba04
Cr-Commit-Position: refs/heads/master@{#39062}
==========

[commit-bot: I haz the power, 2016-09-01 06:02:31]

Patchset 1 (id:??) landed as
https://crrev.com/64a7bd3877dec131a7544a695f6a760c46d4ba04
Cr-Commit-Position: refs/heads/master@{#39062}

[Jakob Kummerow, 2016-09-01 08:33:18]

Thanks for the quick fix!