[PATCH] Add support for high priority scheduling in amdgpu v7

[Andres Rodriguez]

This is the v6 series rebased on the 4.12-wip branch.

It now also includes support for gfx9. I've tried to minimize the changes
to gfx9 since I don't have pre-release hardware. The small changes made can
be found in patches:
patches/0007-drm-amdgpu-take-ownership-of-per-pipe-configuration.patch
patches/0009-drm-amdgpu-allow-split-of-queues-with-kfd-at-queue-g.patch
patches/0014-drm-amdgpu-remove-hardcoded-queue_mask-in-PACKET3_SE.patch
patches/0020-drm-amdgpu-convert-srbm-lock-to-a-spinlock-v2.patch

This series is also available at:
https://github.com/lostgoat/linux/commits/wip-high-priority-4.12

The list of patches is getting a little large to maintain and keep rebasing.
If we could start merging at least the first part of the series that has
a lot of the prep work and cleanup patches that would be helpful for me.

On that note, the following patches are still missing review/ack:

[PATCH 10/25] drm/amdgpu: teach amdgpu how to enable interrupts for
[PATCH 14/25] drm/amdgpu: remove hardcoded queue_mask i
[PATCH 17/25] drm/amdgpu: implement lru amdgpu_queue_mgr policy for
[PATCH 19/25] drm/amdgpu: add framework for HW specific priority
[PATCH 20/25] drm/amdgpu: convert srbm lock to a spinlock v2
[PATCH 22/25] drm/amdgpu: condense mqd programming sequence
[PATCH 23/25] drm/amdgpu: workaround tonga HW bug in HQD programming

Regards,
Andres