VideoCaptureAndroid: Disable YV12 support for XT1097

VideoCaptureAndroid: Disable YV12 support for XT1097

This CL adds XT1097 to COLORSPACE_BUGGY_DEVICE_LIST, which will cause it to use NV21 instead of YV12.

BUG=595623, 577435
Committed: https://crrev.com/6185b7c9462a082c8552d324ceacb24c91fe2e43
Cr-Commit-Position: refs/heads/master@{#381964}

[magjed_chromium, 2016-03-17 12:12:17]

Description was changed from

==========
VideoCaptureAndroid: Disable YV12 support for XT1097

This CL adds XT1097 to COLORSPACE_BUGGY_DEVICE_LIST, which will cause it to use
NV21 instead of YV12.

BUG=595623,577435
==========

to

==========
VideoCaptureAndroid: Disable YV12 support for XT1097

This CL adds XT1097 to COLORSPACE_BUGGY_DEVICE_LIST, which will cause it to use
NV21 instead of YV12.

BUG=595623,577435
==========

[magjed_chromium, 2016-03-17 12:12:17]

magjed@chromium.org changed reviewers:
+ mcasas@chromium.org

[magjed_chromium, 2016-03-17 12:14:21]

Miguel - Please take a look.

[mcasas, 2016-03-17 17:25:41]

LGTM

Shouldn't we add a UMA or similar to track the amount
of special devices?

[magjed_chromium, 2016-03-18 13:24:45]

On 2016/03/17 17:25:41, mcasas wrote:
> LGTM
> 
> Shouldn't we add a UMA or similar to track the amount
> of special devices?

I'm not sure if that statistics is interesting or useful. I'm landing this CL
without adding stats.

[magjed_chromium, 2016-03-18 13:24:49]

The CQ bit was checked by magjed@chromium.org

[commit-bot: I haz the power, 2016-03-18 13:25:00]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1812833002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1812833002/1

[commit-bot: I haz the power, 2016-03-18 14:43:05]

Description was changed from

==========
VideoCaptureAndroid: Disable YV12 support for XT1097

This CL adds XT1097 to COLORSPACE_BUGGY_DEVICE_LIST, which will cause it to use
NV21 instead of YV12.

BUG=595623,577435
==========

to

==========
VideoCaptureAndroid: Disable YV12 support for XT1097

This CL adds XT1097 to COLORSPACE_BUGGY_DEVICE_LIST, which will cause it to use
NV21 instead of YV12.

BUG=595623,577435
==========

[commit-bot: I haz the power, 2016-03-18 14:43:06]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-03-18 14:44:41]

Description was changed from

==========
VideoCaptureAndroid: Disable YV12 support for XT1097

This CL adds XT1097 to COLORSPACE_BUGGY_DEVICE_LIST, which will cause it to use
NV21 instead of YV12.

BUG=595623,577435
==========

to

==========
VideoCaptureAndroid: Disable YV12 support for XT1097

This CL adds XT1097 to COLORSPACE_BUGGY_DEVICE_LIST, which will cause it to use
NV21 instead of YV12.

BUG=595623,577435

Committed: https://crrev.com/6185b7c9462a082c8552d324ceacb24c91fe2e43
Cr-Commit-Position: refs/heads/master@{#381964}
==========

[commit-bot: I haz the power, 2016-03-18 14:44:42]

Patchset 1 (id:??) landed as
https://crrev.com/6185b7c9462a082c8552d324ceacb24c91fe2e43
Cr-Commit-Position: refs/heads/master@{#381964}

[mcasas, 2016-03-18 18:08:05]

On 2016/03/18 13:24:45, magjed_chromium wrote:
> On 2016/03/17 17:25:41, mcasas wrote:
> > LGTM
> > 
> > Shouldn't we add a UMA or similar to track the amount
> > of special devices?
> 
> I'm not sure if that statistics is interesting or useful. 

They are IMHO. Otherwise you'll never know how
impactful your changes are and/or if you ever have
the need or will to revert these changes, you won't
have the information at hand and will need to land
UMA stats there and then.

> I'm landing this CL without adding stats.

Please don't be so dismissive.

For instance, I see that there are 3 XT models already
blacklisted, and now you added another one. How 
can you tell what's the impact in relative terms of 
each pseudo blacklisting?

[magjed_chromium, 2016-03-21 12:07:01]

On 2016/03/18 18:08:05, mcasas wrote:
> On 2016/03/18 13:24:45, magjed_chromium wrote:
> > On 2016/03/17 17:25:41, mcasas wrote:
> > > LGTM
> > > 
> > > Shouldn't we add a UMA or similar to track the amount
> > > of special devices?
> > 
> > I'm not sure if that statistics is interesting or useful. 
> 
> They are IMHO. Otherwise you'll never know how
> impactful your changes are and/or if you ever have
> the need or will to revert these changes, you won't
> have the information at hand and will need to land
> UMA stats there and then.
> 
> > I'm landing this CL without adding stats.
> 
> Please don't be so dismissive.
> 
> For instance, I see that there are 3 XT models already
> blacklisted, and now you added another one. How 
> can you tell what's the impact in relative terms of 
> each pseudo blacklisting?

I thought you were OK with landing this CL as is since you stamped it.

I think adding UMA stats for this is not particularly useful because we would
only track the devices we are already aware of that has this problem. Also, the
impact of changing from YV12 to NV21 is not very significant because the
performance of libyuv::I420Rotate() vs libyuv::NV12ToI420Rotate() is pretty
close. In fact, I'm ok with removing the black list altogether and switch to
always using NV21 instead - this is what we ended up doing in WebRTC.

If you feel strongly about adding UMA stats to track this, we can of course do
it, but I don't think it's worth it.