MIPS64: Fix 'Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.''

MIPS64: Fix 'Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.''

This CL fixes the bugs caused by the following CL:
50a394d -- MIPS64: Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.'
  Port 1f5b84e467763510b08aac64577292205dfc19c
  MIPS: use DAHI/DATH for li macro on mips64r6.

LUI instruction would sign extend into higher 32bits, in that case we might need to use DAHI, DATI to overwrite the extension.

The bug will occur when we are loading some addresses such as 0x00007fffffffxxxx.

BUG=
TEST=test-run-native-calls/Run_Int32_Select_*, test-run-native-calls/Run_Int32_WeightedSum_*, test-run-native-calls/Run_Int32_WeightedSum_*, test-run-native-calls/Run_Int32_Select_*

Committed: https://crrev.com/4c57e05d5f50ead2f0c9b9351ca4ac672dac31ff
Cr-Commit-Position: refs/heads/master@{#34467}

[Alan Li, 2016-03-03 17:30:41]

Description was changed from

==========
Fix DAHI/DATI loading wrong constant failures.

LUI instruction would sign extend into higher 32bits, in that case we might need
to use DAHI, DATI to overwrite the extension.

This is quite a bug when loading some addresses such as 0x00007fffffffxxxx.

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

to

==========
Fix DAHI/DATI loading wrong constant failures.

LUI instruction would sign extend into higher 32bits, in that case we might need
to use DAHI, DATI to overwrite the extension.

This is quite a bug when loading some addresses such as 0x00007fffffffxxxx.

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

[Alan Li, 2016-03-03 17:30:41]

alan.li@imgtec.com changed reviewers:
+ akos.palfi@imgtec.com, balazs.kilvady@imgtec.com, gergely.kis@imgtec.com,
ilija.pavlovic.imgtec@gmail.com, ivica.bogosavljevic@imgtec.com

[balazs.kilvady, 2016-03-03 18:09:13]

Please fix the description. You remember? 
MIP64: Fix ''

Port ....

and so on :)

[Alan Li, 2016-03-03 18:16:59]

Description was changed from

==========
Fix DAHI/DATI loading wrong constant failures.

LUI instruction would sign extend into higher 32bits, in that case we might need
to use DAHI, DATI to overwrite the extension.

This is quite a bug when loading some addresses such as 0x00007fffffffxxxx.

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

to

==========
Fix DAHI/DATI loading wrong constant failures. 

The following 2 CLs are buggy:
50a394d -- MIPS64: Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.'
1f5b84e -- MIPS: use DAHI/DATH for li macro on mips64r6.

LUI instruction would sign extend into higher 32bits, in that case we might need
to use DAHI, DATI to overwrite the extension.

This is quite a bug when loading some addresses such as 0x00007fffffffxxxx.


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

[Alan Li, 2016-03-03 18:17:54]

On 2016/03/03 18:09:13, balazs.kilvady wrote:
> Please fix the description. You remember? 
> MIP64: Fix ''
> 
> Port ....
> 
> and so on :)

Right... I forgot again ... Sorry. Fixed.

[balazs.kilvady, 2016-03-03 18:23:50]

lgtm

[balazs.kilvady, 2016-03-03 18:25:51]

NOT LGTM. You fixed the message, not the description, right?

[balazs.kilvady, 2016-03-03 18:28:09]

On 2016/03/03 18:25:51, balazs.kilvady wrote:
> NOT LGTM. You fixed the message, not the description, right?

The description will be added with git commit so that should be fine. I don't
change the message to make it easy to follow the topic in email clients.

[Alan Li, 2016-03-03 18:32:47]

Description was changed from

==========
Fix DAHI/DATI loading wrong constant failures. 

The following 2 CLs are buggy:
50a394d -- MIPS64: Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.'
1f5b84e -- MIPS: use DAHI/DATH for li macro on mips64r6.

LUI instruction would sign extend into higher 32bits, in that case we might need
to use DAHI, DATI to overwrite the extension.

This is quite a bug when loading some addresses such as 0x00007fffffffxxxx.


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

to

==========
MIPS64: Fix 'Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.''

This CL fixes the bugs caused by the following 2 CLs:
50a394d -- MIPS64: Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.'
1f5b84e -- MIPS: use DAHI/DATH for li macro on mips64r6.

Description:
LUI instruction would sign extend into higher 32bits, in that case we might need
to use DAHI, DATI to overwrite the extension.

The bug will occur when we are loading some addresses such as
0x00007fffffffxxxx.


BUG=
TEST=test-run-native-calls/Run_Int32_Select_*,
test-run-native-calls/Run_Int32_WeightedSum_*,
test-run-native-calls/Run_Int32_WeightedSum_*,
test-run-native-calls/Run_Int32_Select_*
==========

[Alan Li, 2016-03-03 18:33:29]

On 2016/03/03 18:28:09, balazs.kilvady wrote:
> On 2016/03/03 18:25:51, balazs.kilvady wrote:
> > NOT LGTM. You fixed the message, not the description, right?
> 
> The description will be added with git commit so that should be fine. I don't
> change the message to make it easy to follow the topic in email clients.

I also added the missing failure test lists.

[balazs.kilvady, 2016-03-03 18:35:02]

lgtm

[Alan Li, 2016-03-03 18:36:32]

Description was changed from

==========
MIPS64: Fix 'Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.''

This CL fixes the bugs caused by the following 2 CLs:
50a394d -- MIPS64: Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.'
1f5b84e -- MIPS: use DAHI/DATH for li macro on mips64r6.

Description:
LUI instruction would sign extend into higher 32bits, in that case we might need
to use DAHI, DATI to overwrite the extension.

The bug will occur when we are loading some addresses such as
0x00007fffffffxxxx.


BUG=
TEST=test-run-native-calls/Run_Int32_Select_*,
test-run-native-calls/Run_Int32_WeightedSum_*,
test-run-native-calls/Run_Int32_WeightedSum_*,
test-run-native-calls/Run_Int32_Select_*
==========

to

==========
MIPS64: Fix 'Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.''

This CL fixes the bugs caused by the following CL:
50a394d -- MIPS64: Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.'  
  Port 1f5b84e467763510b08aac64577292205dfc19c
  MIPS: use DAHI/DATH for li macro on mips64r6.

LUI instruction would sign extend into higher 32bits, in that case we might need
to use DAHI, DATI to overwrite the extension.

The bug will occur when we are loading some addresses such as
0x00007fffffffxxxx.


BUG=
TEST=test-run-native-calls/Run_Int32_Select_*,
test-run-native-calls/Run_Int32_WeightedSum_*,
test-run-native-calls/Run_Int32_WeightedSum_*,
test-run-native-calls/Run_Int32_Select_*
==========

[Alan Li, 2016-03-03 18:37:22]

The CQ bit was checked by alan.li@imgtec.com

[commit-bot: I haz the power, 2016-03-03 18:37:29]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1763733002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1763733002/1

[commit-bot: I haz the power, 2016-03-03 18:53:50]

Description was changed from

==========
MIPS64: Fix 'Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.''

This CL fixes the bugs caused by the following CL:
50a394d -- MIPS64: Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.'  
  Port 1f5b84e467763510b08aac64577292205dfc19c
  MIPS: use DAHI/DATH for li macro on mips64r6.

LUI instruction would sign extend into higher 32bits, in that case we might need
to use DAHI, DATI to overwrite the extension.

The bug will occur when we are loading some addresses such as
0x00007fffffffxxxx.


BUG=
TEST=test-run-native-calls/Run_Int32_Select_*,
test-run-native-calls/Run_Int32_WeightedSum_*,
test-run-native-calls/Run_Int32_WeightedSum_*,
test-run-native-calls/Run_Int32_Select_*
==========

to

==========
MIPS64: Fix 'Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.''

This CL fixes the bugs caused by the following CL:
50a394d -- MIPS64: Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.'  
  Port 1f5b84e467763510b08aac64577292205dfc19c
  MIPS: use DAHI/DATH for li macro on mips64r6.

LUI instruction would sign extend into higher 32bits, in that case we might need
to use DAHI, DATI to overwrite the extension.

The bug will occur when we are loading some addresses such as
0x00007fffffffxxxx.


BUG=
TEST=test-run-native-calls/Run_Int32_Select_*,
test-run-native-calls/Run_Int32_WeightedSum_*,
test-run-native-calls/Run_Int32_WeightedSum_*,
test-run-native-calls/Run_Int32_Select_*
==========

[commit-bot: I haz the power, 2016-03-03 18:53:52]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-03-03 18:55:12]

Description was changed from

==========
MIPS64: Fix 'Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.''

This CL fixes the bugs caused by the following CL:
50a394d -- MIPS64: Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.'  
  Port 1f5b84e467763510b08aac64577292205dfc19c
  MIPS: use DAHI/DATH for li macro on mips64r6.

LUI instruction would sign extend into higher 32bits, in that case we might need
to use DAHI, DATI to overwrite the extension.

The bug will occur when we are loading some addresses such as
0x00007fffffffxxxx.


BUG=
TEST=test-run-native-calls/Run_Int32_Select_*,
test-run-native-calls/Run_Int32_WeightedSum_*,
test-run-native-calls/Run_Int32_WeightedSum_*,
test-run-native-calls/Run_Int32_Select_*
==========

to

==========
MIPS64: Fix 'Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.''

This CL fixes the bugs caused by the following CL:
50a394d -- MIPS64: Fix 'MIPS: use DAHI/DATH for li macro on mips64r6.'
  Port 1f5b84e467763510b08aac64577292205dfc19c
  MIPS: use DAHI/DATH for li macro on mips64r6.

LUI instruction would sign extend into higher 32bits, in that case we might need
to use DAHI, DATI to overwrite the extension.

The bug will occur when we are loading some addresses such as
0x00007fffffffxxxx.

BUG=
TEST=test-run-native-calls/Run_Int32_Select_*,
test-run-native-calls/Run_Int32_WeightedSum_*,
test-run-native-calls/Run_Int32_WeightedSum_*,
test-run-native-calls/Run_Int32_Select_*

Committed: https://crrev.com/4c57e05d5f50ead2f0c9b9351ca4ac672dac31ff
Cr-Commit-Position: refs/heads/master@{#34467}
==========

[commit-bot: I haz the power, 2016-03-03 18:55:14]

Patchset 1 (id:??) landed as
https://crrev.com/4c57e05d5f50ead2f0c9b9351ca4ac672dac31ff
Cr-Commit-Position: refs/heads/master@{#34467}