ozone: choose a card useful for display rather than by default

ozone: choose a card useful for display rather than by default

Due to the introduction of VGEM, /dev/dri/card0 is not necessarily capable of
display. This patch adds a procedure that opens each card in order starting
from card0 and returns an open fd of the first card usable for display. Also,
the defaults are taken out of the ozone_platforms and dri_wrapper will use the
new procedure when no default card path is given. Usability for the purposes of
display is determined by success of the DRM_IOCTL_MODE_GETRESOURCES ioctl.

TEST=run chromiumos freon in qemu
BUG=
R=dnicoara@chromium.org

Committed: https://crrev.com/d555b733bcb6b6a1ce551d15f3c642e7c91a6295
Cr-Commit-Position: refs/heads/master@{#316632}

[Zach Reizner, 2015-02-13 00:52:14]

zachr@google.com changed reviewers:
+ marcheu@chromium.org

[dnicoara, 2015-02-13 02:38:26]

Overall I think the approach is fine. Would you mind sync-ing up with ToT.
There's been a lot of changes to add UDL support and some of the code in here
has changed.

https://codereview.chromium.org/913273006/diff/1/ui/ozone/platform/dri/ozone_...
File ui/ozone/platform/dri/ozone_platform_dri.cc (right):

https://codereview.chromium.org/913273006/diff/1/ui/ozone/platform/dri/ozone_...
ui/ozone/platform/dri/ozone_platform_dri.cc:153: LOG(ERROR) << "zachr: Using
Platform Dri";
debug log

[marcheu, 2015-02-13 04:34:42]

https://codereview.chromium.org/913273006/diff/1/ui/ozone/platform/dri/dri_ut...
File ui/ozone/platform/dri/dri_util.cc (right):

https://codereview.chromium.org/913273006/diff/1/ui/ozone/platform/dri/dri_ut...
ui/ozone/platform/dri/dri_util.cc:155: int ret = drmIoctl(fd,
DRM_IOCTL_MODE_GETRESOURCES, &res);
I'm going to nitpick. In some cases, we can have a driver which answers
positively to the getresources ioctl, but which has no resources to use. So, you
probably want to add a check that the number of CRTCs is >0.

[Zach Reizner, 2015-02-13 18:45:14]

zachr@google.com changed reviewers:
+ zachr@google.com

[Zach Reizner, 2015-02-13 18:45:15]

PTAL

https://codereview.chromium.org/913273006/diff/1/ui/ozone/platform/dri/dri_ut...
File ui/ozone/platform/dri/dri_util.cc (right):

https://codereview.chromium.org/913273006/diff/1/ui/ozone/platform/dri/dri_ut...
ui/ozone/platform/dri/dri_util.cc:155: int ret = drmIoctl(fd,
DRM_IOCTL_MODE_GETRESOURCES, &res);
On 2015/02/13 04:34:42, marcheu wrote:
> I'm going to nitpick. In some cases, we can have a driver which answers
> positively to the getresources ioctl, but which has no resources to use. So,
you
> probably want to add a check that the number of CRTCs is >0.

I agree.

https://codereview.chromium.org/913273006/diff/1/ui/ozone/platform/dri/ozone_...
File ui/ozone/platform/dri/ozone_platform_dri.cc (right):

https://codereview.chromium.org/913273006/diff/1/ui/ozone/platform/dri/ozone_...
ui/ozone/platform/dri/ozone_platform_dri.cc:153: LOG(ERROR) << "zachr: Using
Platform Dri";
On 2015/02/13 02:38:26, dnicoara wrote:
> debug log

Whoops. Line removed.

[dnicoara, 2015-02-13 19:53:49]

You've forgot to rebase.

[marcheu, 2015-02-13 21:57:32]

lgtm

[Zach Reizner, 2015-02-17 17:46:02]

On 2015/02/13 21:57:32, marcheu wrote:
> lgtm

I think I still need a reply from dnicoara, right?

[dnicoara, 2015-02-17 19:20:02]

On 2015/02/17 17:46:02, Zach Reizner wrote:
> On 2015/02/13 21:57:32, marcheu wrote:
> > lgtm
> 
> I think I still need a reply from dnicoara, right?

lgtm, You should post a reply after updating, otherwise I don't get a
notification that a new patchset is available.

[Zach Reizner, 2015-02-17 19:21:52]

The CQ bit was checked by zachr@google.com

[commit-bot: I haz the power, 2015-02-17 19:23:18]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/913273006/40001

[commit-bot: I haz the power, 2015-02-17 19:53:09]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-02-17 19:53:32]

Patchset 3 (id:??) landed as
https://crrev.com/d555b733bcb6b6a1ce551d15f3c642e7c91a6295
Cr-Commit-Position: refs/heads/master@{#316632}