mbox series

[RFC,0/4,v2] drm_hwcomposer: Changes to support HiKey/HiKey960

Message ID 1516749399-29504-1-git-send-email-john.stultz@linaro.org
Headers show
Series drm_hwcomposer: Changes to support HiKey/HiKey960 | expand

Message

John Stultz Jan. 23, 2018, 11:16 p.m. UTC 
Here is a second revision of the early RFC patch set I wanted to
send out to get some review and feedback on.

I've been working to enable the drm_hwcomposer for HiKey and
HiKey960 boards in AOSP, and this patchset contains the required
changes to the drm_hwcomposer code to get things working.

I'm really quite naive when it comes to graphics, and I'm sure I
don't fully understand the drm_hwcomposer code, so forgive me if
I've done anything terribly stupid here.

The first two patches are, I think, mostly straightforward, the
first being a bug fix, and the second adding an importer for
the ARM gralloc implementations.

The next two patches, while largely reworked from v1,  are still
a bit iffy, and I'd appreciate guidance on how to improve them.
The third forcing client composition if there is only one layer
and the fourth providing a glFinish() fallback which fixes
on-screen tearing when the gl compositior fails to initalize.

I also have two additional patches, which I've not included
here - as they still need work, which get the glcompositor to
properly function on HiKey960. But the curious can find them
here:
  https://github.com/johnstultz-work/drm_hwcomposer/commits/drm_hwc

If anyone wants to try the code out on either a HiKey or
HiKey960, you will need just the 4 patches here on top of the
freedesktop/master branch of drm_hwcomposer. You'll also need
the freedesktop/master branch of libdrm. You'll need the
following patches to the device/linaro/hikey project:
 https://github.com/johnstultz-work/android_device_linaro_hikey/commits/drm_hwc

And you'll need to replace the kernel Image-dtb with one built
from the following tree:
 https://git.linaro.org/people/john.stultz/android-dev.git/log/?h=dev/hikey-hwcomposer-deps

Any thoughts or feedback will be very much appreciated!

Many thanks to Rob Herring and Matt Szczesiak for help to get
this working so far! And Sean Paul for his recent help working
out the crtc activiation bug.

Thanks so much for your time!
-john

Cc: Marissa Wall <marissaw@google.com>
Cc: Sean Paul <seanpaul@google.com>
Cc: Dmitry Shmidt <dimitrysh@google.com>
Cc: Robert Foss <robert.foss@collabora.com>
Cc: Matt Szczesiak <matt.szczesiak@arm.com>
Cc: Liviu Dudau <Liviu.Dudau@arm.com>
Cc: David Hanna <david.hanna11@gmail.com>
Cc: Rob Herring <rob.herring@linaro.org>


John Stultz (4):
  drm_hwcomposer: Make sure we set the active state when doing modesets
  drm_hwcomposer: Add platformhisi buffer importer for hikey and
    hikey960
  drm_hwcomposer: Use client compositing if there is only one plane
  drm_hwcomposer: Try to fallback if GLCompisition fails

 Android.mk               |  15 +++-
 drmdisplaycompositor.cpp |  11 +++
 drmhwctwo.cpp            |   7 ++
 platformhisi.cpp         | 200 +++++++++++++++++++++++++++++++++++++++++++++++
 platformhisi.h           |  50 ++++++++++++
 5 files changed, 282 insertions(+), 1 deletion(-)
 create mode 100644 platformhisi.cpp
 create mode 100644 platformhisi.h

-- 
2.7.4