drm/amdgpu: use top down allocation for non-CPU accessible vram

Should help avoid fragmentation of vram due to CPU access
requirements.

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: be explicit about cpu vram access for driver BOs (v2)

For kernel driver BOs, be explicit about whether we need
vram access up front.  This avoids unecessary migrations and
avoids using visible vram for buffers were it's not needed.

v2: line wrap fixes

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: set MEC doorbell range for Fiji

Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: implement burst NOP for SDMA

Customize the insert_nop func for SDMA rings, and use burst NOP for
ring/IB submissions in other places as well

Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add insert_nop ring func and default implementation

The insert_nop function is added to amdgpu_ring_funcs structure as
well as the default implementation

Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add amdgpu_get_sdma_instance helper function

This function is added to map the ring to sdma instance

Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add AMDGPU_MAX_SDMA_INSTANCES

Remove the hardcoded usage

Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add burst_nop flag for sdma

The burst NOP is supported for SDMA when feature_version is >= 20.

Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add count field for the SDMA NOP packet v2

This is added to support the burst NOP

v2: squash the typo fix

Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: use PT for VM sync on unmap

Instead of the array which is used for ID management.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Jammy Zhou <Jammy.Zhou@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>

drm/amdgpu: make wait_event uninterruptible in push_job

with interruptible, the push_job maybe return -ERESTARTSYS,
then result in push_job error.

E.g. bug trace:
[  181.618860] *****amdgpu_copy_buffer:fence->seq:0x0000000048d8758b, contxt:1207959552, ref:683967304, r:-512
[  181.618929] BUG: unable to handle kernel paging request at ffffffff811aa266
[  181.625887] IP: [<ffffffff81548ffc>] reservation_object_add_excl_fence+0x3c/0x120
...
[  181.859767]  [<ffffffff811aa266>] ? unmap_mapping_range+0x66/0x110
[  181.865928]  [<ffffffffc0608ac1>] ttm_bo_move_accel_cleanup+0x41/0x3c0 [ttm]
[  181.872971]  [<ffffffffc062d382>] amdgpu_move_blit.isra.18+0x122/0x150 [amdgpu]
[  181.880254]  [<ffffffff811aa266>] ? unmap_mapping_range+0x66/0x110
[  181.886420]  [<ffffffffc062d709>] amdgpu_bo_move+0xa9/0x200 [amdgpu]
[  181.892753]  [<ffffffffc0606e8d>] ttm_bo_handle_move_mem+0x26d/0x5c0 [ttm]

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

drm/amdgpu: fix amdgpu_bo_unreserve order in GEM_OP IOCTL v2

No copy_(to|from)_user while BO is reserved.

v2: handle default path as well

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>

drm/amdgpu: partially revert "modify amdgpu_fence_wait_any() to amdgpu_fence_wait_multiple()" v2

That isn't used any more.

v2: rebase

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Add radeon suspend/resume quirk for HP Compaq dc5750.

With the radeon driver loaded the HP Compaq dc5750
Small Form Factor machine fails to resume from suspend.
Adding a quirk similar to other devices avoids
the problem and the system resumes properly.

Signed-off-by: Jeffery Miller <jmiller@neverware.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

drm/amdgpu: re-work sync_resv

sync_resv is to handle both amdgpu_fence and sched_fence.

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Jammy Zhou <Jammy.Zhou@amd.com>
Reviewed-by: Christian K?nig <christian.koenig@amd.com>

drm/amdgpu/atom: Send out the full AUX address

AUX addresses are 20 bits long. Send out the entire address instead of
just the low 16 bits.

Port of:
drm/radeon/atom: Send out the full AUX address
to amdgpu

Reviewed-by: Jammy Zhou <Jammy.Zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

drm/radeon/native: Send out the full AUX address

AUX addresses are 20 bits long. Send out the entire address instead of
just the low 16 bits.

Port of:
drm/radeon/atom: Send out the full AUX address
to radeon non-atom aux path

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

drm/radeon/atom: Send out the full AUX address

AUX addresses are 20 bits long. Send out the entire address instead of
just the low 16 bits.

Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: "Christian König" <christian.koenig@amd.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

drm/amdgpu: use IB for fill_buffer instead of direct command

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Christian K?nig <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: stop trying to suspend UVD sessions v2

Saving the current UVD state on suspend and restoring it on resume
just doesn't work reliable. Just close cleanup all sessions on suspend.

Ported from radeon commit "12e49feadff6d7b7ebbe852b36943a71524d8d34".

v2: rebased

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v1)

drm/amdgpu: add scheduler dependency callback v2

This way the scheduler doesn't wait in it's work thread any more.

v2: fix race conditions

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Jammy Zhou <Jammy.Zhou@amd.com>

drm/amdgpu: let the scheduler work more with jobs v2

v2: fix another race condition

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Jammy Zhou <Jammy.Zhou@amd.com>

drm/radeon: fix HDMI quantization_range for pre-DCE5 asics

Support for output_csc is only available on DCE5 and newer so
don't mess with the HDMI quantization_range on pre-DCE5 asics.

bug:
https://bugs.freedesktop.org/show_bug.cgi?id=83226

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

drm/nouveau/dispnv04: fix build on powerpc

align with mask code in overlay.c, Ben can clean the naming
up later if necessary.

Signed-off-by: Dave Airlie <airlied@redhat.com>

nouveau: fix powerpc build

Reported-by: Ilia Mirkin on irc.
Signed-off-by: Dave Airlie <airlied@redhat.com>

Merge branch 'linux-4.3' of git://anongit.freedesktop.org/git/nouveau/linux-2.6 into drm-next

Rather large pull request this time around, due to the long-pending
cleanup of the kernel driver being here.  There's a stupidly large
number of commits for that, as I wanted to have the series be
bisectable at a fairly fine-grained level.  That said, a very large
portion of the churn in the rework was automated, and a very large
number of boards from right across the whole range we support have
been tested.  I'm fairly confident there shouldn't be (too many)
issues from this.

Beyond correcting some not-so-great design decisions and making the
code a lot easier to work with, there's not much exciting (lower
memory usage, GPU VM should be a lot faster, etc) to be gained by the
end-user as a result of the cleanup, it mostly lays the groundwork for
future improvements.

A big thanks goes to Alexandre Courbot for testing/debugging the GK20A
codepaths for me :)

Highlights:
- A heap of perfmon work, providing a more useful userspace interface
and specifying counters for a bunch of boards
- Support for GT200 reclocking + other misc pm improvements
- Initial patches towards supporting GM20B (Tegra X1)
- Maxwell DisplayPort fixes
- Cleanup of the kernel driver
- The usual collection of random fixes

* 'linux-4.3' of git://anongit.freedesktop.org/git/nouveau/linux-2.6: (312 commits)
  drm/nouveau: bump driver version for release
  drm/nouveau/tegra: merge platform setup from nouveau drm
  drm/nouveau/pci: merge agp handling from nouveau drm
  drm/nouveau/device: remove pci/platform_device from common struct
  drm/nouveau/device: import pciid list and integrate quirks with it
  drm/nouveau/device: cleaner abstraction for device resource functions
  drm/nouveau/mc: move device irq handling to platform-specific code
  drm/nouveau/mc/gf100-: handle second interrupt tree
  drm/nouveau/mc: abstract interface to master intr registers
  drm/nouveau/pci: new subdev
  drm/nouveau/object: merge with handle
  drm/nouveau/core: remove the remainder of the previous style
  drm/nouveau/mpeg: convert to new-style nvkm_engine
  drm/nouveau/sw: convert to new-style nvkm_engine
  drm/nouveau/pm: convert to new-style nvkm_engine
  drm/nouveau/gr: convert to new-style nvkm_engine
  drm/nouveau/fifo: convert to new-style nvkm_engine
  drm/nouveau/disp: convert to new-style nvkm_engine
  drm/nouveau/dma: convert to new-style nvkm_engine
  drm/nouveau/cipher: convert to new-style nvkm_engine
  ...

drm/nouveau: bump driver version for release

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/tegra: merge platform setup from nouveau drm

The copyright header in nvkm/engine/device/platform.c has been replaced
with the NVIDIA one from drm/nouveau_platform.c, as most of the actual
code is now theirs.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/pci: merge agp handling from nouveau drm

This commit reinstates the pre-DEVINIT AGP fiddling that was broken in
an earlier commit.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/device: remove pci/platform_device from common struct

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/device: import pciid list and integrate quirks with it

PCI IDs taken from the NVIDIA binary driver, with permission.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/device: cleaner abstraction for device resource functions

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mc: move device irq handling to platform-specific code

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mc/gf100-: handle second interrupt tree

Doesn't fix any known issue, but best be safe in case control is handed
to us from firmware with these left enabled.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mc: abstract interface to master intr registers

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/pci: new subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/object: merge with handle

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/core: remove the remainder of the previous style

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mpeg: convert to new-style nvkm_engine

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/sw: convert to new-style nvkm_engine

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/pm: convert to new-style nvkm_engine

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr: convert to new-style nvkm_engine

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo: convert to new-style nvkm_engine

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/disp: convert to new-style nvkm_engine

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/dma: convert to new-style nvkm_engine

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/cipher: convert to new-style nvkm_engine

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/ce: convert to new-style nvkm_engine

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/xtensa: convert to new-style nvkm_engine

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/falcon: convert to new-style nvkm_engine

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/volt: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/tmr: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/therm: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/pmu: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mxm: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mmu: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mc: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/ltc: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/imem: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/devinit: run devinit scripts right after preinit

This ensures we have a valid mask of disabled engines before we start
trying to execute fini()/init() on the subdevs, potentially touching
devices that don't exist.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/ibus: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/i2c: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gpio: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fuse: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fb: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/devinit: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/clk: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/bus: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/bios: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/bar: convert to new-style nvkm_subdev

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/core: kill some (now) dead code

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/sw: convert user classes to new-style nvkm_object

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/pm: convert user classes to new-style nvkm_object

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mpeg: convert user classes to new-style nvkm_object

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr: convert user classes to new-style nvkm_object

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/cipher: convert user classes to new-style nvkm_object

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/ce: convert user classes to new-style nvkm_object

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/xtensa: convert user classes to new-style nvkm_object

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/falcon: convert user classes to new-style nvkm_object

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo: convert user classes to new-style nvkm_object

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo: split user classes out from engine implementations

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/disp: convert user classes to new-style nvkm_object

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/disp: split user classes out from engine implementations

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/dma: convert user classes to new-style nvkm_object

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/dma: split user classes out from engine implementations

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/device: convert ctrl class to new-style nvkm_object

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/device: convert user class to new-style nvkm_object

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/client: convert to new-style nvkm_object

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/sw: remove dependence on namedb/engctx lookup

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mpeg: remove dependence on namedb/engctx lookup

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr: remove dependence on namedb/engctx lookup

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo: remove dependence on namedb/engctx lookup

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/cipher: remove dependence on namedb/engctx lookup

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/falcon: remove dependence on namedb/engctx lookup

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fb: remove dependence on namedb/engctx lookup

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/sw: turn flip completion into an event

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/sw/nv04: replace direct context access with GetRef method

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/ramht: remove dependence on namedb

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gpuobj: separate allocation from nvkm_object

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr: directly use instmem where currently possible

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo: directly use instmem for runlists and polling areas

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>