[x64] Avoid sign extension in TruncateFloat64ToInt32(TruncationMode::kJavaScript)

[x64] Avoid the sign extension in the DoubleToI conversion.

We use the cvttsd2si (float64-to-int32) instruction to avoid that negative numbers get sign extended in the destination register.

@rodolph: Is sign extension also a thing on arm64? I noticed that also on arm64 a float64-to-int64 instruction is used.

Committed: https://crrev.com/6f8630ed7bf82a957408023447934c0961c44ad9
Cr-Commit-Position: refs/heads/master@{#35466}

[ahaas, 2016-04-13 17:18:17]

Description was changed from

==========
[x64] Avoid the sign extension in the DoubleToI conversion.
==========

to

==========
[x64] Avoid the sign extension in the DoubleToI conversion.

We use the cvttsd2si (float64-to-int32) instruction to avoid that negative
numbers get sign extended in the destination register.

@rodolph: Is sign extension also a thing on arm64? I noticed that also on arm64
a float64-to-int64 instruction is used.
==========

[ahaas, 2016-04-13 17:18:18]

ahaas@chromium.org changed reviewers:
+ bmeurer@chromium.org

[Rodolph Perfetta (ARM), 2016-04-13 17:59:13]

On 2016/04/13 17:21:22, ahaas wrote:

So yes on arm64 a float64-to-int64 will sign extend to 64-bits. On arm64 the
only difference between a float64-to-int64 and float64-to-int32 is the size of
the destination register: 32-bit W or 64-bit X, the instruction menmonic is the
same.

After a quick look at the code it looks like it is going to be fine though since
TF's TruncateFloat64ToInt32 is expected to produce a 32-bit value: which ever
instructions consumes the output of the truncate, which has been written to an X
register, should read the associated 32-bit W register not the 64-bit X.

[Benedikt Meurer, 2016-04-13 18:13:00]

Nice, thanks. LGTM.

[ahaas, 2016-04-14 07:22:19]

The CQ bit was checked by ahaas@chromium.org

[commit-bot: I haz the power, 2016-04-14 07:22:29]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1877133004/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1877133004/20001

[commit-bot: I haz the power, 2016-04-14 07:24:29]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-04-14 07:25:50]

Description was changed from

==========
[x64] Avoid the sign extension in the DoubleToI conversion.

We use the cvttsd2si (float64-to-int32) instruction to avoid that negative
numbers get sign extended in the destination register.

@rodolph: Is sign extension also a thing on arm64? I noticed that also on arm64
a float64-to-int64 instruction is used.
==========

to

==========
[x64] Avoid the sign extension in the DoubleToI conversion.

We use the cvttsd2si (float64-to-int32) instruction to avoid that negative
numbers get sign extended in the destination register.

@rodolph: Is sign extension also a thing on arm64? I noticed that also on arm64
a float64-to-int64 instruction is used.

Committed: https://crrev.com/6f8630ed7bf82a957408023447934c0961c44ad9
Cr-Commit-Position: refs/heads/master@{#35466}
==========

[commit-bot: I haz the power, 2016-04-14 07:25:51]

Patchset 2 (id:??) landed as
https://crrev.com/6f8630ed7bf82a957408023447934c0961c44ad9
Cr-Commit-Position: refs/heads/master@{#35466}