Remove runtime NEON detection

Remove runtime NEON detection

Chrome does not detect NEON instruction set at runtime in WebRTC code starting
with M50, which is now in Stable. Remove support for runtime detection for
simplicity.

The only remaining piece of Chrome that will continue to depend on runtime
detection is /net, where devices with _broken_ neon support are also detected,
and it is not configurable via GYP/GN.

BUG=522035
NOPRESUBMIT=true

Committed: https://crrev.com/e305d956c0717a28ca88cd8547e5b310dfa74594
Cr-Commit-Position: refs/heads/master@{#12778}

[pasko, 2016-05-10 08:20:17]

Description was changed from

==========
Subject: [PATCH] Remove runtime NEON detection

Chrome does not detect NEON instruction set at runtime in WebRTC code starting
with M50, which is now in Stable. Remove support for runtime detection for
simplicity.

The only remaining piece of Chrome that will continue to depend on runtime
detection is /net, where devices with _broken_ neon support are also detected,
and it is not configurable via GYP/GN.

BUG=522035
==========

to

==========
Remove runtime NEON detection

Chrome does not detect NEON instruction set at runtime in WebRTC code starting
with M50, which is now in Stable. Remove support for runtime detection for
simplicity.

The only remaining piece of Chrome that will continue to depend on runtime
detection is /net, where devices with _broken_ neon support are also detected,
and it is not configurable via GYP/GN.

BUG=522035
==========

[pasko, 2016-05-10 09:56:22]

pasko@chromium.org changed reviewers:
+ henrika@chromium.org

[pasko, 2016-05-10 10:56:15]

henrika: I am not sure who the best reviewer is for this, please route
appropriately.

[henrika_webrtc, 2016-05-10 11:11:55]

henrika@webrtc.org changed reviewers:
+ tina.legrand@webrtc.org
- henrika@chromium.org

[henrika_webrtc, 2016-05-10 11:13:46]

Reassigning to tlegrand@

[pasko, 2016-05-12 12:36:59]

On 2016/05/10 11:13:46, henrika_webrtc wrote:
> Reassigning to tlegrand@

friendly ping ;-)

[minyue-webrtc, 2016-05-12 13:01:38]

On 2016/05/12 12:36:59, pasko wrote:
> On 2016/05/10 11:13:46, henrika_webrtc wrote:
> > Reassigning to tlegrand@
> 
> friendly ping ;-)

how are y

[minyue-webrtc, 2016-05-12 13:02:59]

On 2016/05/12 13:01:38, minyue-webrtc wrote:
> On 2016/05/12 12:36:59, pasko wrote:
> > On 2016/05/10 11:13:46, henrika_webrtc wrote:
> > > Reassigning to tlegrand@
> > 
> > friendly ping ;-)
> 
> how are y

sorry, pressed something wrong before finishing it.

How are we sure that we use neon code when we have neon?

[pasko, 2016-05-12 13:17:37]

On 2016/05/12 13:02:59, minyue-webrtc wrote:
> On 2016/05/12 13:01:38, minyue-webrtc wrote:
> > On 2016/05/12 12:36:59, pasko wrote:
> > > On 2016/05/10 11:13:46, henrika_webrtc wrote:
> > > > Reassigning to tlegrand@
> > > 
> > > friendly ping ;-)
> > 
> > how are y
> 
> sorry, pressed something wrong before finishing it.
> 
> How are we sure that we use neon code when we have neon?

My understanding is that in webrtc code we have neon iff the macro
WEBRTC_HAS_NEON is defined.

Is your question about testing whether we put WEBRTC_HAS_NEON correctly?

I do not know the answer to this question, I assume it is somehow already done
correctly and in this change I am simplifying the code under the assumption that
defined(WEBRTC_DETECT_NEON) is always false. That's all.

See this recent chromium-dev@ discussion for more context:
https://groups.google.com/a/chromium.org/d/msg/chromium-dev/FeUOhsADqvM/Cmp4b...

[tlegrand-webrtc, 2016-05-17 12:32:48]

On 2016/05/12 13:17:37, pasko wrote:
> On 2016/05/12 13:02:59, minyue-webrtc wrote:
> > On 2016/05/12 13:01:38, minyue-webrtc wrote:
> > > On 2016/05/12 12:36:59, pasko wrote:
> > > > On 2016/05/10 11:13:46, henrika_webrtc wrote:
> > > > > Reassigning to tlegrand@
> > > > 
> > > > friendly ping ;-)
> > > 
> > > how are y
> > 
> > sorry, pressed something wrong before finishing it.
> > 
> > How are we sure that we use neon code when we have neon?
> 
> My understanding is that in webrtc code we have neon iff the macro
> WEBRTC_HAS_NEON is defined.
> 
> Is your question about testing whether we put WEBRTC_HAS_NEON correctly?
> 
> I do not know the answer to this question, I assume it is somehow already done
> correctly and in this change I am simplifying the code under the assumption
that
> defined(WEBRTC_DETECT_NEON) is always false. That's all.
> 
> See this recent chromium-dev@ discussion for more context:
>
https://groups.google.com/a/chromium.org/d/msg/chromium-dev/FeUOhsADqvM/Cmp4b...

Thanks for cleaning this up. LGTM

[pasko, 2016-05-17 13:16:10]

The CQ bit was checked by pasko@chromium.org

[commit-bot: I haz the power, 2016-05-17 13:16:20]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1955413003/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1955413003/20001

[commit-bot: I haz the power, 2016-05-17 13:21:30]

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

[commit-bot: I haz the power, 2016-05-17 13:21:31]

Try jobs failed on following builders:
  presubmit on tryserver.webrtc (JOB_FAILED,
http://build.chromium.org/p/tryserver.webrtc/builders/presubmit/builds/5602)

[pasko, 2016-05-17 14:27:51]

Presumbit complains:

> ** Presubmit Messages **
>
> You seem to be changing native API header files. Please make sure that you:
>   1. Make compatible changes that don't break existing clients.
>   2. Mark the old stuff as deprecated.
>   3. Create a timeline and plan for when the deprecated stuff will be
>      removed. (The amount of time we give users to change their code
>      should be informed by how much work it is for them. If they just
>      need to replace one name with another or something equally
>      simple, 1-2 weeks might be good; if they need to do serious work,
>      up to 3 months may be called for.)
>   4. Update/inform existing downstream code owners to stop using the
>      deprecated stuff. (Send announcements to
>      discuss-webrtc@googlegroups.com and webrtc-users@google.com.)
>   5. Remove the deprecated stuff, once the agreed-upon amount of time
>      has passed.
> Related files:
>
>   webrtc/typedefs.h

This is done, thanks.

> As you're changing GYP files: please make sure corresponding BUILD.gn files
are also updated.
> Changed GYP files:
>   webrtc/build/common.gypi

Is that an error or a warning? This change updates webrtc/build/webrtc.gni, so I
assume it should be good to go.

>
/b/build/slave/linux64/build/src/tools/autoroller/unittests/roll_chromium_revision_test.py
(0.20s)
>
> If this change requires manual test instructions to QA team, add
TEST=[instructions].
>
> mb_validate (0.10s)
>
> ** Presubmit Warnings **
> License must match:
> .*? Copyright( \(c\))?
> (2016|2015|2014|2013|2012|2011|2010|2009|2008|2007|2006|2005|2004|2003) The
> WebRTC [Pp]roject [Aa]uthors\. All [Rr]ights [Rr]eserved\.\n.*?\n.*? Use of
> this source code is governed by a BSD-style license\n.*? that can be found in
> the LICENSE file in the root of the source\n.*? tree\. An additional
> intellectual property rights grant can be found\n.*? in the file PATENTS\.
> All contributing project authors may\n.*? be found in the AUTHORS file in the
> root of the source tree\.\n
> Found a bad license header in these files:
>   webrtc/modules/audio_processing/aec/aec_rdft.cc

I think we should not change the copyright in this file on mechanical changes
like mine.

Should I NOPRESUBMIT=true it?

[kjellander_webrtc, 2016-05-17 14:29:05]

On 2016/05/17 14:27:51, pasko wrote:
> Presumbit complains:
> 
> > ** Presubmit Messages **
> >
> > You seem to be changing native API header files. Please make sure that you:
> >   1. Make compatible changes that don't break existing clients.
> >   2. Mark the old stuff as deprecated.
> >   3. Create a timeline and plan for when the deprecated stuff will be
> >      removed. (The amount of time we give users to change their code
> >      should be informed by how much work it is for them. If they just
> >      need to replace one name with another or something equally
> >      simple, 1-2 weeks might be good; if they need to do serious work,
> >      up to 3 months may be called for.)
> >   4. Update/inform existing downstream code owners to stop using the
> >      deprecated stuff. (Send announcements to
> >      mailto:discuss-webrtc@googlegroups.com and
mailto:webrtc-users@google.com.)
> >   5. Remove the deprecated stuff, once the agreed-upon amount of time
> >      has passed.
> > Related files:
> >
> >   webrtc/typedefs.h
> 
> This is done, thanks.
> 
> > As you're changing GYP files: please make sure corresponding BUILD.gn files
> are also updated.
> > Changed GYP files:
> >   webrtc/build/common.gypi
> 
> Is that an error or a warning? This change updates webrtc/build/webrtc.gni, so
I
> assume it should be good to go.
> 
> >
>
/b/build/slave/linux64/build/src/tools/autoroller/unittests/roll_chromium_revision_test.py
> (0.20s)
> >
> > If this change requires manual test instructions to QA team, add
> TEST=[instructions].
> >
> > mb_validate (0.10s)
> >
> > ** Presubmit Warnings **
> > License must match:
> > .*? Copyright( \(c\))?
> > (2016|2015|2014|2013|2012|2011|2010|2009|2008|2007|2006|2005|2004|2003) The
> > WebRTC [Pp]roject [Aa]uthors\. All [Rr]ights [Rr]eserved\.\n.*?\n.*? Use of
> > this source code is governed by a BSD-style license\n.*? that can be found
in
> > the LICENSE file in the root of the source\n.*? tree\. An additional
> > intellectual property rights grant can be found\n.*? in the file PATENTS\.
> > All contributing project authors may\n.*? be found in the AUTHORS file in
the
> > root of the source tree\.\n
> > Found a bad license header in these files:
> >   webrtc/modules/audio_processing/aec/aec_rdft.cc
> 
> I think we should not change the copyright in this file on mechanical changes
> like mine.
> 
> Should I NOPRESUBMIT=true it?

sgtm; https://bugs.chromium.org/p/webrtc/issues/detail?id=5023 addresses this
issue.

[pasko, 2016-05-17 15:55:34]

Description was changed from

==========
Remove runtime NEON detection

Chrome does not detect NEON instruction set at runtime in WebRTC code starting
with M50, which is now in Stable. Remove support for runtime detection for
simplicity.

The only remaining piece of Chrome that will continue to depend on runtime
detection is /net, where devices with _broken_ neon support are also detected,
and it is not configurable via GYP/GN.

BUG=522035
==========

to

==========
Remove runtime NEON detection

Chrome does not detect NEON instruction set at runtime in WebRTC code starting
with M50, which is now in Stable. Remove support for runtime detection for
simplicity.

The only remaining piece of Chrome that will continue to depend on runtime
detection is /net, where devices with _broken_ neon support are also detected,
and it is not configurable via GYP/GN.

BUG=522035
NOPRESUBMIT=true
==========

[pasko, 2016-05-17 15:55:45]

The CQ bit was checked by pasko@chromium.org

[commit-bot: I haz the power, 2016-05-17 15:55:56]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1955413003/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1955413003/20001

[commit-bot: I haz the power, 2016-05-17 17:56:44]

Description was changed from

==========
Remove runtime NEON detection

Chrome does not detect NEON instruction set at runtime in WebRTC code starting
with M50, which is now in Stable. Remove support for runtime detection for
simplicity.

The only remaining piece of Chrome that will continue to depend on runtime
detection is /net, where devices with _broken_ neon support are also detected,
and it is not configurable via GYP/GN.

BUG=522035
NOPRESUBMIT=true
==========

to

==========
Remove runtime NEON detection

Chrome does not detect NEON instruction set at runtime in WebRTC code starting
with M50, which is now in Stable. Remove support for runtime detection for
simplicity.

The only remaining piece of Chrome that will continue to depend on runtime
detection is /net, where devices with _broken_ neon support are also detected,
and it is not configurable via GYP/GN.

BUG=522035
NOPRESUBMIT=true
==========

[commit-bot: I haz the power, 2016-05-17 17:56:46]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-05-17 17:56:57]

Description was changed from

==========
Remove runtime NEON detection

Chrome does not detect NEON instruction set at runtime in WebRTC code starting
with M50, which is now in Stable. Remove support for runtime detection for
simplicity.

The only remaining piece of Chrome that will continue to depend on runtime
detection is /net, where devices with _broken_ neon support are also detected,
and it is not configurable via GYP/GN.

BUG=522035
NOPRESUBMIT=true
==========

to

==========
Remove runtime NEON detection

Chrome does not detect NEON instruction set at runtime in WebRTC code starting
with M50, which is now in Stable. Remove support for runtime detection for
simplicity.

The only remaining piece of Chrome that will continue to depend on runtime
detection is /net, where devices with _broken_ neon support are also detected,
and it is not configurable via GYP/GN.

BUG=522035
NOPRESUBMIT=true

Committed: https://crrev.com/e305d956c0717a28ca88cd8547e5b310dfa74594
Cr-Commit-Position: refs/heads/master@{#12778}
==========

[commit-bot: I haz the power, 2016-05-17 17:56:58]

Patchset 2 (id:??) landed as
https://crrev.com/e305d956c0717a28ca88cd8547e5b310dfa74594
Cr-Commit-Position: refs/heads/master@{#12778}