[turbofan] Implemented the optional Float32RoundDown operator.

[turbofan] Implemented the optional Float32RoundDown operator.

I implemented the optional Float32RoundDown operator on x64, ia32, arm,
and arm64.

For arm I also had to adjust the simulator.

R=titzer@chromium.org

Committed: https://crrev.com/74434403f6d3e8a367992ce672bf112dd3955554
Cr-Commit-Position: refs/heads/master@{#32261}

[ahaas, 2015-11-24 21:26:15]

ahaas@chromium.org changed reviewers:
+ v8-arm-ports@googlegroups.com

[titzer, 2015-11-25 10:33:25]

lgtm

https://codereview.chromium.org/1471913006/diff/20001/src/ia32/assembler-ia32.cc
File src/ia32/assembler-ia32.cc (right):

https://codereview.chromium.org/1471913006/diff/20001/src/ia32/assembler-ia32...
src/ia32/assembler-ia32.cc:2205: // Mask precision exeption.
s/exeption/exception/

[ahaas, 2015-11-25 10:37:05]

The CQ bit was checked by ahaas@chromium.org

[commit-bot: I haz the power, 2015-11-25 10:37:24]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1471913006/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1471913006/20001

[jbramley, 2015-11-25 10:46:15]

jacob.bramley@arm.com changed reviewers:
+ jacob.bramley@arm.com

[jbramley, 2015-11-25 10:46:16]

https://codereview.chromium.org/1471913006/diff/20001/src/arm/simulator-arm.cc
File src/arm/simulator-arm.cc (right):

https://codereview.chromium.org/1471913006/diff/20001/src/arm/simulator-arm.c...
src/arm/simulator-arm.cc:3925: sd_value = ceilf(sm_value);
Why not std::ceil? (Ditto for the other Float32 operations.)

https://codereview.chromium.org/1471913006/diff/20001/test/cctest/compiler/te...
File test/cctest/compiler/test-run-machops.cc (right):

https://codereview.chromium.org/1471913006/diff/20001/test/cctest/compiler/te...
test/cctest/compiler/test-run-machops.cc:5153: FOR_FLOAT32_INPUTS(i) {
CheckFloatEq(floorf(*i), m.Call(*i)); }
Again, why not std::floor?

[jbramley, 2015-11-25 10:52:17]

https://codereview.chromium.org/1471913006/diff/1/src/arm/simulator-arm.cc
File src/arm/simulator-arm.cc (right):

https://codereview.chromium.org/1471913006/diff/1/src/arm/simulator-arm.cc#ne...
src/arm/simulator-arm.cc:3921: sd_value = nearbyintf(sm_value);
If this has behaviour equivalent to the explicit code above, can't we use this
there too?

[ahaas, 2015-11-25 11:01:55]

On 2015/11/25 at 10:46:16, jacob.bramley wrote:
> https://codereview.chromium.org/1471913006/diff/20001/src/arm/simulator-arm.cc
> File src/arm/simulator-arm.cc (right):
> 
>
https://codereview.chromium.org/1471913006/diff/20001/src/arm/simulator-arm.c...
> src/arm/simulator-arm.cc:3925: sd_value = ceilf(sm_value);
> Why not std::ceil? (Ditto for the other Float32 operations.)
> 
>
https://codereview.chromium.org/1471913006/diff/20001/test/cctest/compiler/te...
> File test/cctest/compiler/test-run-machops.cc (right):
> 
>
https://codereview.chromium.org/1471913006/diff/20001/test/cctest/compiler/te...
> test/cctest/compiler/test-run-machops.cc:5153: FOR_FLOAT32_INPUTS(i) {
CheckFloatEq(floorf(*i), m.Call(*i)); }
> Again, why not std::floor?

I don't have any particular reason for using floorf instead of std::floor. Is
there a reason why I should use std::floor?

[ahaas, 2015-11-25 11:04:58]

On 2015/11/25 at 10:52:17, jacob.bramley wrote:
> https://codereview.chromium.org/1471913006/diff/1/src/arm/simulator-arm.cc
> File src/arm/simulator-arm.cc (right):
> 
>
https://codereview.chromium.org/1471913006/diff/1/src/arm/simulator-arm.cc#ne...
> src/arm/simulator-arm.cc:3921: sd_value = nearbyintf(sm_value);
> If this has behaviour equivalent to the explicit code above, can't we use this
there too?

We could use this function for float64 rounding too. I would prefer to do it in
a separate CL though.

[commit-bot: I haz the power, 2015-11-25 11:05:28]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-11-25 11:06:23]

Patchset 2 (id:??) landed as
https://crrev.com/74434403f6d3e8a367992ce672bf112dd3955554
Cr-Commit-Position: refs/heads/master@{#32261}

[jbramley, 2015-11-25 11:12:01]

On 2015/11/25 11:01:55, ahaas wrote:
> On 2015/11/25 at 10:46:16, jacob.bramley wrote:
> >
https://codereview.chromium.org/1471913006/diff/20001/src/arm/simulator-arm.cc
> > File src/arm/simulator-arm.cc (right):
> > 
> >
>
https://codereview.chromium.org/1471913006/diff/20001/src/arm/simulator-arm.c...
> > src/arm/simulator-arm.cc:3925: sd_value = ceilf(sm_value);
> > Why not std::ceil? (Ditto for the other Float32 operations.)
> > 
> >
>
https://codereview.chromium.org/1471913006/diff/20001/test/cctest/compiler/te...
> > File test/cctest/compiler/test-run-machops.cc (right):
> > 
> >
>
https://codereview.chromium.org/1471913006/diff/20001/test/cctest/compiler/te...
> > test/cctest/compiler/test-run-machops.cc:5153: FOR_FLOAT32_INPUTS(i) {
> CheckFloatEq(floorf(*i), m.Call(*i)); }
> > Again, why not std::floor?
> 
> I don't have any particular reason for using floorf instead of std::floor. Is
> there a reason why I should use std::floor?

Consistency, I suppose. There might be other reasons. For example, this CL
changed several calls to use the 'std::' versions in the ARM64 simulator:
https://codereview.chromium.org/207543003

[jbramley, 2015-11-25 11:12:19]

On 2015/11/25 11:04:58, ahaas wrote:
> On 2015/11/25 at 10:52:17, jacob.bramley wrote:
> > https://codereview.chromium.org/1471913006/diff/1/src/arm/simulator-arm.cc
> > File src/arm/simulator-arm.cc (right):
> > 
> >
>
https://codereview.chromium.org/1471913006/diff/1/src/arm/simulator-arm.cc#ne...
> > src/arm/simulator-arm.cc:3921: sd_value = nearbyintf(sm_value);
> > If this has behaviour equivalent to the explicit code above, can't we use
this
> there too?
> 
> We could use this function for float64 rounding too. I would prefer to do it
in
> a separate CL though.

Fine by me.

[jbramley, 2015-11-25 11:13:54]

Also, vrintm float tests should be added to test/cctest/test-assembler-arm.cc,
similar to the vrintz tests that I mentioned in
https://codereview.chromium.org/1468303005/.

[ahaas, 2015-11-25 12:32:54]

On 2015/11/25 at 11:13:54, jacob.bramley wrote:
> Also, vrintm float tests should be added to test/cctest/test-assembler-arm.cc,
similar to the vrintz tests that I mentioned in
https://codereview.chromium.org/1468303005/.

I will add the test in https://codereview.chromium.org/1468303005/, together
with the test for vrintz.