Access double fields in C++ as uint64_t fields to preserve signaling bit of a NaN.

Access double fields in C++ as uint64_t fields to preserve signaling bit of a NaN.

Manipulating the signaling NaN used for the hole and uninitialized double
field sentinel in C++, e.g. with bit_cast or HeapNumber::value()/set_value(),
will change its value on ia32 (the x87 stack is used to return values and
stores to the stack silently clear the signalling bit).

BUG=v8:5495
Review-Url: https://codereview.chromium.org/2652553003
Cr-Commit-Position: refs/heads/master@{#42609}
Committed: https://chromium.googlesource.com/v8/v8/+/f9f5fc31fdea4ded8287d6bc8ee36e147b6412d4

[Igor Sheludko, 2017-01-23 14:12:13]

Patchset #1 (id:1) has been deleted

[Igor Sheludko, 2017-01-23 14:17:38]

Description was changed from

==========
Don't use 'double' for the cases where we need to check for kHoleNaN value

formatted

fox

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

to

==========
Always access double fields in C++ as uint64_t fields to preserve signaling bit
of a NaN.

Manipulating the signaling NaN used for the hole and uninitialized
double field sentinel in C++, e.g. with bit_cast or
HeapNumber::value()/set_value(),
will change its value on ia32 (the x87 stack is used to return values
and stores to the stack silently clear the signalling bit).

BUG=v8:5495
==========

[Igor Sheludko, 2017-01-23 14:18:00]

Description was changed from

==========
Always access double fields in C++ as uint64_t fields to preserve signaling bit
of a NaN.

Manipulating the signaling NaN used for the hole and uninitialized
double field sentinel in C++, e.g. with bit_cast or
HeapNumber::value()/set_value(),
will change its value on ia32 (the x87 stack is used to return values
and stores to the stack silently clear the signalling bit).

BUG=v8:5495
==========

to

==========
Access double fields in C++ as uint64_t fields to preserve signaling bit of a
NaN.

Manipulating the signaling NaN used for the hole and uninitialized
double field sentinel in C++, e.g. with bit_cast or
HeapNumber::value()/set_value(),
will change its value on ia32 (the x87 stack is used to return values
and stores to the stack silently clear the signalling bit).

BUG=v8:5495
==========

[Igor Sheludko, 2017-01-23 14:18:23]

Description was changed from

==========
Access double fields in C++ as uint64_t fields to preserve signaling bit of a
NaN.

Manipulating the signaling NaN used for the hole and uninitialized
double field sentinel in C++, e.g. with bit_cast or
HeapNumber::value()/set_value(),
will change its value on ia32 (the x87 stack is used to return values
and stores to the stack silently clear the signalling bit).

BUG=v8:5495
==========

to

==========
Access double fields in C++ as uint64_t fields to preserve signaling bit state
of a NaN.

Manipulating the signaling NaN used for the hole and uninitialized
double field sentinel in C++, e.g. with bit_cast or
HeapNumber::value()/set_value(),
will change its value on ia32 (the x87 stack is used to return values
and stores to the stack silently clear the signalling bit).

BUG=v8:5495
==========

[Igor Sheludko, 2017-01-23 14:18:56]

Description was changed from

==========
Access double fields in C++ as uint64_t fields to preserve signaling bit state
of a NaN.

Manipulating the signaling NaN used for the hole and uninitialized
double field sentinel in C++, e.g. with bit_cast or
HeapNumber::value()/set_value(),
will change its value on ia32 (the x87 stack is used to return values
and stores to the stack silently clear the signalling bit).

BUG=v8:5495
==========

to

==========
Access double fields in C++ as uint64_t fields to preserve signaling bit state
of a NaN.

Manipulating the signaling NaN used for the hole and uninitialized double
field sentinel in C++, e.g. with bit_cast or HeapNumber::value()/set_value(),
will change its value on ia32 (the x87 stack is used to return values and
stores to the stack silently clear the signalling bit).

BUG=v8:5495
==========

[Igor Sheludko, 2017-01-23 14:24:18]

Patchset #1 (id:20001) has been deleted

[Igor Sheludko, 2017-01-23 14:29:05]

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

[commit-bot: I haz the power, 2017-01-23 14:29:08]

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

[Igor Sheludko, 2017-01-23 14:29:16]

ishell@chromium.org changed reviewers:
+ jkummerow@chromium.org

[Igor Sheludko, 2017-01-23 14:29:17]

PTAL

[commit-bot: I haz the power, 2017-01-23 14:29:38]

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

[commit-bot: I haz the power, 2017-01-23 14:29:39]

Dry run: Try jobs failed on following builders:
  v8_win64_rel_ng on master.tryserver.v8 (JOB_FAILED, no build URL)
  v8_win_compile_dbg on master.tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_win_compile_dbg/builds/3...)
  v8_win_nosnap_shared_rel_ng on master.tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_win_nosnap_shared_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/21427)

[Igor Sheludko, 2017-01-23 14:31:43]

Description was changed from

==========
Access double fields in C++ as uint64_t fields to preserve signaling bit state
of a NaN.

Manipulating the signaling NaN used for the hole and uninitialized double
field sentinel in C++, e.g. with bit_cast or HeapNumber::value()/set_value(),
will change its value on ia32 (the x87 stack is used to return values and
stores to the stack silently clear the signalling bit).

BUG=v8:5495
==========

to

==========
Access double fields in C++ as uint64_t fields to preserve signaling bit of a
NaN.

Manipulating the signaling NaN used for the hole and uninitialized double
field sentinel in C++, e.g. with bit_cast or HeapNumber::value()/set_value(),
will change its value on ia32 (the x87 stack is used to return values and
stores to the stack silently clear the signalling bit).

BUG=v8:5495
==========

[Michael Lippautz, 2017-01-23 14:32:54]

On 2017/01/23 14:29:17, Igor Sheludko wrote:
> PTAL

heap/ lgtm

[Jakob Kummerow, 2017-01-23 15:07:10]

lgtm

https://codereview.chromium.org/2652553003/diff/40001/src/objects-inl.h
File src/objects-inl.h (right):

https://codereview.chromium.org/2652553003/diff/40001/src/objects-inl.h#newco...
src/objects-inl.h:2371: // we use normal double setters here.
How about:

DCHECK(!value->IsHeapNumber() || 
       HeapNumber::cast(value)->value_as_bits() !=
static_cast<uint64_t>(kHoleNanInt64));

[Igor Sheludko, 2017-01-23 16:22:49]

Patchset #2 (id:60001) has been deleted

[Igor Sheludko, 2017-01-23 16:29:49]

https://codereview.chromium.org/2652553003/diff/40001/src/objects-inl.h
File src/objects-inl.h (right):

https://codereview.chromium.org/2652553003/diff/40001/src/objects-inl.h#newco...
src/objects-inl.h:2371: // we use normal double setters here.
On 2017/01/23 15:07:10, Jakob Kummerow wrote:
> How about:
> 
> DCHECK(!value->IsHeapNumber() || 
>        HeapNumber::cast(value)->value_as_bits() !=
> static_cast<uint64_t>(kHoleNanInt64));

The kHoleNanInt64 value can be constructed using typed arrays so such an assert
can fail. But it will not harm constant field tracking.

To make the C++ code be in sync with the compiled code (which preserves
signalling bit when storing to double fields) I modified this code to store
double values as bits.

[Igor Sheludko, 2017-01-23 16:35:08]

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

[commit-bot: I haz the power, 2017-01-23 16:35:16]

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, 2017-01-23 16:42:05]

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

[commit-bot: I haz the power, 2017-01-23 16:42:05]

Dry run: Try jobs failed on following builders:
  v8_win_compile_dbg on master.tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_win_compile_dbg/builds/3...)

[Igor Sheludko, 2017-01-23 16:47:13]

The CQ bit was checked by ishell@chromium.org

[Igor Sheludko, 2017-01-23 16:47:13]

The patchset sent to the CQ was uploaded after l-g-t-m from
mlippautz@chromium.org, jkummerow@chromium.org
Link to the patchset: https://codereview.chromium.org/2652553003/#ps100001
(title: "Fix windows build")

[commit-bot: I haz the power, 2017-01-23 16:47:15]

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

[Igor Sheludko, 2017-01-23 16:58:28]

The CQ bit was unchecked by ishell@chromium.org

[Igor Sheludko, 2017-01-23 17:26:53]

The CQ bit was checked by ishell@chromium.org

[Igor Sheludko, 2017-01-23 17:26:53]

The patchset sent to the CQ was uploaded after l-g-t-m from
mlippautz@chromium.org, jkummerow@chromium.org
Link to the patchset: https://codereview.chromium.org/2652553003/#ps120001
(title: "More fixes")

[commit-bot: I haz the power, 2017-01-23 17:26:59]

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, 2017-01-23 17:30:07]

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

[commit-bot: I haz the power, 2017-01-23 17:30:08]

Try jobs failed on following builders:
  v8_win64_rel_ng on master.tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_win64_rel_ng/builds/21359)

[Jakob Kummerow, 2017-01-23 17:31:35]

lgtm

[Igor Sheludko, 2017-01-23 17:32:43]

Patchset #4 (id:120001) has been deleted

[Igor Sheludko, 2017-01-23 17:35:57]

Patchset #4 (id:140001) has been deleted

[Igor Sheludko, 2017-01-23 17:37:25]

Patchset #4 (id:20002) has been deleted

[Igor Sheludko, 2017-01-23 17:41:24]

Patchset #4 (id:160001) has been deleted

[Igor Sheludko, 2017-01-23 17:43:30]

The CQ bit was checked by ishell@chromium.org

[Igor Sheludko, 2017-01-23 17:43:30]

The patchset sent to the CQ was uploaded after l-g-t-m from
mlippautz@chromium.org, jkummerow@chromium.org
Link to the patchset: https://codereview.chromium.org/2652553003/#ps180001
(title: "More fixes")

[commit-bot: I haz the power, 2017-01-23 17:43:32]

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, 2017-01-23 18:18:50]

CQ is committing da patch.

Bot data: {"patchset_id": 180001, "attempt_start_ts": 1485193410415040,
"parent_rev": "d80781442f490e39641fcef14c65082565a4d3c5", "commit_rev":
"f9f5fc31fdea4ded8287d6bc8ee36e147b6412d4"}

[commit-bot: I haz the power, 2017-01-23 18:18:56]

Description was changed from

==========
Access double fields in C++ as uint64_t fields to preserve signaling bit of a
NaN.

Manipulating the signaling NaN used for the hole and uninitialized double
field sentinel in C++, e.g. with bit_cast or HeapNumber::value()/set_value(),
will change its value on ia32 (the x87 stack is used to return values and
stores to the stack silently clear the signalling bit).

BUG=v8:5495
==========

to

==========
Access double fields in C++ as uint64_t fields to preserve signaling bit of a
NaN.

Manipulating the signaling NaN used for the hole and uninitialized double
field sentinel in C++, e.g. with bit_cast or HeapNumber::value()/set_value(),
will change its value on ia32 (the x87 stack is used to return values and
stores to the stack silently clear the signalling bit).

BUG=v8:5495

Review-Url: https://codereview.chromium.org/2652553003
Cr-Commit-Position: refs/heads/master@{#42609}
Committed:
https://chromium.googlesource.com/v8/v8/+/f9f5fc31fdea4ded8287d6bc8ee36e147b6...
==========

[commit-bot: I haz the power, 2017-01-23 18:18:57]

Committed patchset #4 (id:180001) as
https://chromium.googlesource.com/v8/v8/+/f9f5fc31fdea4ded8287d6bc8ee36e147b6...