MIPS: [wasm] Detect unrepresentability in the float32-to-int32 conversion correctly on arm

MIPS: [wasm] Detect unrepresentability in the float32-to-int32 conversion correctly on arm

Port de369129d272c9cbd2a6282f4097a5d0ce264e85

Original commit message:

In the current implementation of wasm an unrepresentable input of the
float32-to-int32 conversion is detected by first truncating the input, then
converting the truncated input to int32 and back to float32, and then checking
whether the result is the same as the truncated input.

This input check does not work on arm and arm64 for an input of (INT32_MAX + 1)
because on these platforms the float32-to-int32 conversion results in INT32_MAX
if the input is greater than INT32_MAX.  When INT32_MAX is converted back to
float32, then the result is (INT32_MAX + 1) again because INT32_MAX cannot be
represented precisely as float32, and rounding-to-nearest results in (INT32_MAX
+ 1). Since (INT32_MAX + 1) equals the truncated input value, the input appears
to be representable.

With the changes in this CL, the result of the float32-to-int32 conversion is
incremented by 1 if the original result was INT32_MAX. Thereby the detection of
unrepresenable inputs in wasm works. Note that since INT32_MAX cannot be
represented precisely in float32, it can also never be a valid result of the
float32-to-int32 conversion.

BUG=cctest/test-run-wasm/RunWasmCompiled_I32SConvertF32,cctest/test-run-wasm/RunWasmCompiled_I32UConvertF32

Committed: https://crrev.com/db6d8e2a5afc96209ec4dd99e6933b9c7e864f69
Cr-Commit-Position: refs/heads/master@{#37586}

[ivica.bogosavljevic, 2016-07-07 11:01:39]

ivica.bogosavljevic@imgtec.com changed reviewers:
+ akos.palfi@imgtec.com, balazs.kilvady@imgtec.com

[ivica.bogosavljevic, 2016-07-07 11:01:39]

PTAL

Although it may seems strange, this is exactly what is done at ARM. The tests
confirm this is the correct behavior.

[ivica.bogosavljevic, 2016-07-07 12:00:22]

Description was changed from

==========
MIPS: Fix implementation of TruncateFloat32To[U]int32 turbofan operators.

In case of overflow, TruncateFloat32To[U]int32 operators returned MAX_[U]INT.
This behavior
prevents overflow detection and causes two tests to fail. The fix introduces a
new behavior:
in case of overflow, the value returned is MIN_[U]INT. Now
TruncateFloat32To[U]int32
behaves same on MIPS and other architectures.

BUG=cctest/test-run-wasm/RunWasmCompiled_I32SConvertF32,cctest/test-run-wasm/RunWasmCompiled_I32UConvertF32
==========

to

==========
MIPS: [wasm] Detect unrepresentability in the float32-to-int32 conversion
correctly on arm

Port de369129d272c9cbd2a6282f4097a5d0ce264e85

Original commit message:

In the current implementation of wasm an unrepresentable input of the
float32-to-int32 conversion is detected by first truncating the input, then
converting the truncated input to int32 and back to float32, and then checking
whether the result is the same as the truncated input.

This input check does not work on arm and arm64 for an input of (INT32_MAX + 1)
because on these platforms the float32-to-int32 conversion results in INT32_MAX
if the input is greater than INT32_MAX.  When INT32_MAX is converted back to
float32, then the result is (INT32_MAX + 1) again because INT32_MAX cannot be
represented precisely as float32, and rounding-to-nearest results in (INT32_MAX
+ 1). Since (INT32_MAX + 1) equals the truncated input value, the input appears
to be representable.

With the changes in this CL, the result of the float32-to-int32 conversion is
incremented by 1 if the original result was INT32_MAX. Thereby the detection of
unrepresenable inputs in wasm works. Note that since INT32_MAX cannot be
represented precisely in float32, it can also never be a valid result of the
float32-to-int32 conversion.

BUG=cctest/test-run-wasm/RunWasmCompiled_I32SConvertF32,cctest/test-run-wasm/RunWasmCompiled_I32UConvertF32
==========

[balazs.kilvady, 2016-07-07 12:01:24]

lgtm

[ivica.bogosavljevic, 2016-07-07 12:05:42]

The CQ bit was checked by ivica.bogosavljevic@imgtec.com to run a CQ dry run

[commit-bot: I haz the power, 2016-07-07 12:05:54]

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, 2016-07-07 12:33:31]

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

[commit-bot: I haz the power, 2016-07-07 12:33:31]

Dry run: This issue passed the CQ dry run.

[ivica.bogosavljevic, 2016-07-07 12:37:18]

The CQ bit was checked by ivica.bogosavljevic@imgtec.com

[commit-bot: I haz the power, 2016-07-07 12:37:30]

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, 2016-07-07 12:39:25]

Description was changed from

==========
MIPS: [wasm] Detect unrepresentability in the float32-to-int32 conversion
correctly on arm

Port de369129d272c9cbd2a6282f4097a5d0ce264e85

Original commit message:

In the current implementation of wasm an unrepresentable input of the
float32-to-int32 conversion is detected by first truncating the input, then
converting the truncated input to int32 and back to float32, and then checking
whether the result is the same as the truncated input.

This input check does not work on arm and arm64 for an input of (INT32_MAX + 1)
because on these platforms the float32-to-int32 conversion results in INT32_MAX
if the input is greater than INT32_MAX.  When INT32_MAX is converted back to
float32, then the result is (INT32_MAX + 1) again because INT32_MAX cannot be
represented precisely as float32, and rounding-to-nearest results in (INT32_MAX
+ 1). Since (INT32_MAX + 1) equals the truncated input value, the input appears
to be representable.

With the changes in this CL, the result of the float32-to-int32 conversion is
incremented by 1 if the original result was INT32_MAX. Thereby the detection of
unrepresenable inputs in wasm works. Note that since INT32_MAX cannot be
represented precisely in float32, it can also never be a valid result of the
float32-to-int32 conversion.

BUG=cctest/test-run-wasm/RunWasmCompiled_I32SConvertF32,cctest/test-run-wasm/RunWasmCompiled_I32UConvertF32
==========

to

==========
MIPS: [wasm] Detect unrepresentability in the float32-to-int32 conversion
correctly on arm

Port de369129d272c9cbd2a6282f4097a5d0ce264e85

Original commit message:

In the current implementation of wasm an unrepresentable input of the
float32-to-int32 conversion is detected by first truncating the input, then
converting the truncated input to int32 and back to float32, and then checking
whether the result is the same as the truncated input.

This input check does not work on arm and arm64 for an input of (INT32_MAX + 1)
because on these platforms the float32-to-int32 conversion results in INT32_MAX
if the input is greater than INT32_MAX.  When INT32_MAX is converted back to
float32, then the result is (INT32_MAX + 1) again because INT32_MAX cannot be
represented precisely as float32, and rounding-to-nearest results in (INT32_MAX
+ 1). Since (INT32_MAX + 1) equals the truncated input value, the input appears
to be representable.

With the changes in this CL, the result of the float32-to-int32 conversion is
incremented by 1 if the original result was INT32_MAX. Thereby the detection of
unrepresenable inputs in wasm works. Note that since INT32_MAX cannot be
represented precisely in float32, it can also never be a valid result of the
float32-to-int32 conversion.

BUG=cctest/test-run-wasm/RunWasmCompiled_I32SConvertF32,cctest/test-run-wasm/RunWasmCompiled_I32UConvertF32
==========

[commit-bot: I haz the power, 2016-07-07 12:39:26]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-07-07 12:41:37]

Description was changed from

==========
MIPS: [wasm] Detect unrepresentability in the float32-to-int32 conversion
correctly on arm

Port de369129d272c9cbd2a6282f4097a5d0ce264e85

Original commit message:

In the current implementation of wasm an unrepresentable input of the
float32-to-int32 conversion is detected by first truncating the input, then
converting the truncated input to int32 and back to float32, and then checking
whether the result is the same as the truncated input.

This input check does not work on arm and arm64 for an input of (INT32_MAX + 1)
because on these platforms the float32-to-int32 conversion results in INT32_MAX
if the input is greater than INT32_MAX.  When INT32_MAX is converted back to
float32, then the result is (INT32_MAX + 1) again because INT32_MAX cannot be
represented precisely as float32, and rounding-to-nearest results in (INT32_MAX
+ 1). Since (INT32_MAX + 1) equals the truncated input value, the input appears
to be representable.

With the changes in this CL, the result of the float32-to-int32 conversion is
incremented by 1 if the original result was INT32_MAX. Thereby the detection of
unrepresenable inputs in wasm works. Note that since INT32_MAX cannot be
represented precisely in float32, it can also never be a valid result of the
float32-to-int32 conversion.

BUG=cctest/test-run-wasm/RunWasmCompiled_I32SConvertF32,cctest/test-run-wasm/RunWasmCompiled_I32UConvertF32
==========

to

==========
MIPS: [wasm] Detect unrepresentability in the float32-to-int32 conversion
correctly on arm

Port de369129d272c9cbd2a6282f4097a5d0ce264e85

Original commit message:

In the current implementation of wasm an unrepresentable input of the
float32-to-int32 conversion is detected by first truncating the input, then
converting the truncated input to int32 and back to float32, and then checking
whether the result is the same as the truncated input.

This input check does not work on arm and arm64 for an input of (INT32_MAX + 1)
because on these platforms the float32-to-int32 conversion results in INT32_MAX
if the input is greater than INT32_MAX.  When INT32_MAX is converted back to
float32, then the result is (INT32_MAX + 1) again because INT32_MAX cannot be
represented precisely as float32, and rounding-to-nearest results in (INT32_MAX
+ 1). Since (INT32_MAX + 1) equals the truncated input value, the input appears
to be representable.

With the changes in this CL, the result of the float32-to-int32 conversion is
incremented by 1 if the original result was INT32_MAX. Thereby the detection of
unrepresenable inputs in wasm works. Note that since INT32_MAX cannot be
represented precisely in float32, it can also never be a valid result of the
float32-to-int32 conversion.

BUG=cctest/test-run-wasm/RunWasmCompiled_I32SConvertF32,cctest/test-run-wasm/RunWasmCompiled_I32UConvertF32

Committed: https://crrev.com/db6d8e2a5afc96209ec4dd99e6933b9c7e864f69
Cr-Commit-Position: refs/heads/master@{#37586}
==========

[commit-bot: I haz the power, 2016-07-07 12:41:38]

Patchset 1 (id:??) landed as
https://crrev.com/db6d8e2a5afc96209ec4dd99e6933b9c7e864f69
Cr-Commit-Position: refs/heads/master@{#37586}