S390X: Prevent upper 32bit corruption in 32bit ops

S390X: Prevent upper 32bit corruption in 32bit ops

Fix A couple places which could cause upper 32bit corruption

R=joransiu@ca.ibm.com, mbrandy@us.ibm.com, michael_dawson@ca.ibm.com
BUG=

Committed: https://crrev.com/59b9ff3f7a7cbe848e77ffc86ead998ffd9508bf
Cr-Commit-Position: refs/heads/master@{#35952}

[john.yan, 2016-05-02 17:55:09]

PTAL

[JoranSiu, 2016-05-02 18:22:22]

https://codereview.chromium.org/1939763003/diff/1/src/compiler/s390/code-gene...
File src/compiler/s390/code-generator-s390.cc (right):

https://codereview.chromium.org/1939763003/diff/1/src/compiler/s390/code-gene...
src/compiler/s390/code-generator-s390.cc:811: if
(i.OutputRegister().is(i.InputRegister(1)) &&
ShiftLeft masm shoudl really handle all of these cases.  It at least has a check
for DISTINCT_OPS.  Can we fix it there instead?

https://codereview.chromium.org/1939763003/diff/1/src/compiler/s390/code-gene...
src/compiler/s390/code-generator-s390.cc:1102: #if V8_TARGET_ARCH_S390X
We should introduce a Load_ExtendSrc() in masm so we don't have so many #if
S390X's sprinkled through our code.

https://codereview.chromium.org/1939763003/diff/1/src/compiler/s390/code-gene...
src/compiler/s390/code-generator-s390.cc:1148: #if V8_TARGET_ARCH_S390X
I believe this is the same as : LoadAndTestP_ExtendSrc

[john.yan, 2016-05-02 21:03:57]

On 2016/05/02 18:22:22, JoranSiu wrote:
>
https://codereview.chromium.org/1939763003/diff/1/src/compiler/s390/code-gene...
> File src/compiler/s390/code-generator-s390.cc (right):
> 
>
https://codereview.chromium.org/1939763003/diff/1/src/compiler/s390/code-gene...
> src/compiler/s390/code-generator-s390.cc:811: if
> (i.OutputRegister().is(i.InputRegister(1)) &&
> ShiftLeft masm shoudl really handle all of these cases.  It at least has a
check
> for DISTINCT_OPS.  Can we fix it there instead?
> 
>
https://codereview.chromium.org/1939763003/diff/1/src/compiler/s390/code-gene...
> src/compiler/s390/code-generator-s390.cc:1102: #if V8_TARGET_ARCH_S390X
> We should introduce a Load_ExtendSrc() in masm so we don't have so many #if
> S390X's sprinkled through our code.
> 

overload LoadW and LoadlW instead.

>
https://codereview.chromium.org/1939763003/diff/1/src/compiler/s390/code-gene...
> src/compiler/s390/code-generator-s390.cc:1148: #if V8_TARGET_ARCH_S390X
> I believe this is the same as : LoadAndTestP_ExtendSrc

updated.

[MTBrandyberry, 2016-05-02 21:30:04]

lgtm

[john.yan, 2016-05-02 21:33:56]

The CQ bit was checked by jyan@ca.ibm.com

[commit-bot: I haz the power, 2016-05-02 21:34:06]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1939763003/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1939763003/20001

[commit-bot: I haz the power, 2016-05-02 21:56:39]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-05-02 21:57:09]

Description was changed from

==========
S390X: Prevent upper 32bit corruption in 32bit ops

Fix A couple places which could cause upper 32bit corruption

R=joransiu@ca.ibm.com, mbrandy@us.ibm.com, michael_dawson@ca.ibm.com
BUG=
==========

to

==========
S390X: Prevent upper 32bit corruption in 32bit ops

Fix A couple places which could cause upper 32bit corruption

R=joransiu@ca.ibm.com, mbrandy@us.ibm.com, michael_dawson@ca.ibm.com
BUG=

Committed: https://crrev.com/59b9ff3f7a7cbe848e77ffc86ead998ffd9508bf
Cr-Commit-Position: refs/heads/master@{#35952}
==========

[commit-bot: I haz the power, 2016-05-02 21:57:10]

Patchset 2 (id:??) landed as
https://crrev.com/59b9ff3f7a7cbe848e77ffc86ead998ffd9508bf
Cr-Commit-Position: refs/heads/master@{#35952}