agp: efficeon: constify pci_device_id.

pci_device_id are not supposed to change at runtime. All functions
working with pci_device_id provided by <linux/pci.h> work with
const pci_device_id. So mark the non-const structs as const.

Signed-off-by: Arvind Yadav <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

agp: ati: constify pci_device_id.

pci_device_id are not supposed to change at runtime. All functions
working with pci_device_id provided by <linux/pci.h> work with
const pci_device_id. So mark the non-const structs as const.

Signed-off-by: Arvind Yadav <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

agp: ali: constify pci_device_id.

pci_device_id are not supposed to change at runtime. All functions
working with pci_device_id provided by <linux/pci.h> work with
const pci_device_id. So mark the non-const structs as const.

Signed-off-by: Arvind Yadav <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

agp: intel: constify pci_device_id.

pci_device_id are not supposed to change at runtime. All functions
working with pci_device_id provided by <linux/pci.h> work with
const pci_device_id. So mark the non-const structs as const.

Signed-off-by: Arvind Yadav <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

agp: amd-k7: constify pci_device_id.

pci_device_id are not supposed to change at runtime. All functions
working with pci_device_id provided by <linux/pci.h> work with
const pci_device_id. So mark the non-const structs as const.

Signed-off-by: Arvind Yadav <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

agp: uninorth: constify pci_device_id.

pci_device_id are not supposed to change at runtime. All functions
working with pci_device_id provided by <linux/pci.h> work with
const pci_device_id. So mark the non-const structs as const.

Signed-off-by: Arvind Yadav <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

Merge branch 'topic-arcpgu-updates' of https://github.com/foss-for-synopsys-dwc-arc-processors/linux into drm-next

arcgpu minor updates.

* 'topic-arcpgu-updates' of https://github.com/foss-for-synopsys-dwc-arc-processors/linux:
  drm: arcpgu: Allow some clock deviation in crtc->mode_valid() callback
  drm: arcpgu: Fix module unload
  drm: arcpgu: Fix mmap() callback
  arcpgu: Simplify driver name
  drm/arcpgu: Opt in debugfs

Merge tag 'drm-next-du-20170803' of git://linuxtv.org/pinchartl/media into drm-next

rcar-du updates, contains vsp1 updates as well.

* tag 'drm-next-du-20170803' of git://linuxtv.org/pinchartl/media: (24 commits)
  drm: rcar-du: Use new iterator macros
  drm: rcar-du: Repair vblank for DRM page flips using the VSP
  drm: rcar-du: Fix race condition when disabling planes at CRTC stop
  drm: rcar-du: Wait for flip completion instead of vblank in commit tail
  drm: rcar-du: Use the VBK interrupt for vblank events
  drm: rcar-du: Add HDMI outputs to R8A7796 device description
  drm: rcar-du: Remove an unneeded NULL check
  drm: rcar-du: Setup planes before enabling CRTC to avoid flicker
  drm: rcar-du: Configure DPAD0 routing through last group on Gen3
  drm: rcar-du: Restrict DPLL duty cycle workaround to H3 ES1.x
  drm: rcar-du: Support multiple sources from the same VSP
  drm: rcar-du: Fix comments to comply with the kernel coding style
  drm: rcar-du: Use of_graph_get_remote_endpoint()
  v4l: vsp1: Add support for header display lists in continuous mode
  v4l: vsp1: Add support for multiple DRM pipelines
  v4l: vsp1: Add support for multiple LIF instances
  v4l: vsp1: Add support for new VSP2-BS, VSP2-DL and VSP2-D instances
  v4l: vsp1: Add support for the BRS entity
  v4l: vsp1: Add pipe index argument to the VSP-DU API
  v4l: vsp1: Don't create links for DRM pipeline
  ...

drm: rcar-du: Use new iterator macros

for_each_obj_in_state is about to be removed, so use the correct new
iterator macros.

Also look at new_plane_state instead of plane->state when looking up
the hw planes in use. They should be the same except when reallocating,
(in which case this code is skipped) and we should really stop looking
at obj->state whenever possible.

Signed-off-by: Maarten Lankhorst <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Signed-off-by: Laurent Pinchart <[email protected]>

drm: rcar-du: Repair vblank for DRM page flips using the VSP

The driver recently switched from handling page flip completion in the
DU vertical blanking handler to the VSP frame end handler to fix a race
condition. This unfortunately resulted in incorrect timestamps in the
vertical blanking events sent to userspace as vertical blanking is now
handled after sending the event.

To fix this we must reverse the order of the two operations. The easiest
way is to handle vertical blanking in the VSP frame end handler before
sending the event. The VSP frame end interrupt occurs approximately 50µs
earlier than the DU frame end interrupt, but this should not cause any
undue harm.

As we need to handle vertical blanking even when page flip completion is
delayed, the VSP driver now needs to call the frame end completion
callback unconditionally, with a new argument to report whether page
flip has completed.

With this new scheme the DU vertical blanking interrupt isn't needed
anymore, so we can stop enabling it.

Fixes: d503a43ac06a ("drm: rcar-du: Register a completion callback with VSP1")
Signed-off-by: Kieran Bingham <[email protected]>
Signed-off-by: Laurent Pinchart <[email protected]>
Acked-by: Mauro Carvalho Chehab <[email protected]>

drm: rcar-du: Fix race condition when disabling planes at CRTC stop

When stopping the CRTC the driver must disable all planes and wait for
the change to take effect at the next vblank. Merely calling
drm_crtc_wait_one_vblank() is not enough, as the function doesn't
include any mechanism to handle the race with vblank interrupts.

Replace the drm_crtc_wait_one_vblank() call with a manual mechanism that
handles the vblank interrupt race.

Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>

drm: rcar-du: Wait for flip completion instead of vblank in commit tail

Page flips can take more than one vertical blanking to complete if
arming the page flips races with the vertical blanking interrupt.
Waiting for one vblank to complete the atomic commit in the commit tail
handler is thus incorrect, and can lead to framebuffers being released
while still being scanned out.

Fix this by waiting for flip completion instead, using the
drm_atomic_helper_wait_for_flip_done() helper.

Fixes: 0d230422d256 ("drm: rcar-du: Register a completion callback with VSP1")
Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>

drm: rcar-du: Use the VBK interrupt for vblank events

When implementing support for interlaced modes, the driver switched from
reporting vblank events on the vertical blanking (VBK) interrupt to the
frame end interrupt (FRM). This incorrectly divided the reported refresh
rate by two. Fix it by moving back to the VBK interrupt.

Fixes: 906eff7fcada ("drm: rcar-du: Implement support for interlaced modes")
Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>

drm: rcar-du: Add HDMI outputs to R8A7796 device description

Update the device description with the HDMI output.

Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>

drm: rcar-du: Remove an unneeded NULL check

"params" can't be NULL here.  The next lines assume that we either
hit the break statement of "params->mpixelclock == ~0UL".  The
inconsistent NULL checking makes static checkers complain.  I've just
removed the test.

Signed-off-by: Dan Carpenter <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Signed-off-by: Laurent Pinchart <[email protected]>

drm: rcar-du: Setup planes before enabling CRTC to avoid flicker

Commit 52055bafa1ff ("drm: rcar-du: Move plane commit code from CRTC
start to CRTC resume") changed the order of the plane commit and CRTC
enable operations to accommodate the runtime PM requirements. However,
this introduced corruption in the first displayed frame, as the CRTC is
now enabled without any plane configured. On Gen2 hardware the first
frame will be black and likely unnoticed, but on Gen3 hardware we end up
starting the display before the VSP compositor, which is more
noticeable.

To fix this, revert the order of the commit operations back, and handle
runtime PM requirements in the CRTC .atomic_begin() and .atomic_enable()
helper operation handlers.

Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>

drm: rcar-du: Configure DPAD0 routing through last group on Gen3

On Gen3 SoCs DPAD0 routing is configured through the last CRTC group,
unlike on Gen2 where it is configured through the first CRTC group. Fix
the driver accordingly.

Fixes: 2427b3037710 ("drm: rcar-du: Add R8A7795 device support")
Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>

drm: rcar-du: Restrict DPLL duty cycle workaround to H3 ES1.x

The H3 ES1.x exhibits dot clock duty cycle stability issues. We can work
around them by configuring the DPLL to twice the desired frequency,
coupled with a /2 post-divider. This isn't needed on other SoCs and
breaks HDMI output on M3-W for a currently unknown reason, so restrict
the workaround to H3 ES1.x.

From an implementation point of view, move work around handling outside
of the rcar_du_dpll_divider() function by requesting a x2 DPLL output
frequency explicitly. The existing post-divider calculation mechanism
will then take care of dividing the clock by two automatically.

While at it, print a more useful debugging message to ease debugging
clock rate issues.

Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>

drm: rcar-du: Support multiple sources from the same VSP

On R-Car H3 ES2.0, DU channels 0 and 3 are served by two separate
pipelines from the same VSP. Support this in the DU driver.

Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>

drm: rcar-du: Fix comments to comply with the kernel coding style

To avoid mixing comment styles when new comments complying with the
kernel coding style are introduced, fix all multiline comments in one
go.

Signed-off-by: Laurent Pinchart <[email protected]>
Acked-by: Kieran Bingham <[email protected]>

drm: rcar-du: Use of_graph_get_remote_endpoint()

Now, we can use of_graph_get_remote_endpoint(). Let's use it.

Signed-off-by: Kuninori Morimoto <[email protected]>
Signed-off-by: Laurent Pinchart <[email protected]>

v4l: vsp1: Add support for header display lists in continuous mode

The VSP supports both header and headerless display lists. The latter is
easier to use when the VSP feeds data directly to the DU in continuous
mode, and the driver thus uses headerless display lists for DU operation
and header display lists otherwise.

Headerless display lists are only available on WPF.0. This has never
been an issue so far, as only WPF.0 is connected to the DU. However, on
H3 ES2.0, the VSP-DL instance has both WPF.0 and WPF.1 connected to the
DU. We thus can't use headerless display lists unconditionally for DU
operation.

Implement support for continuous mode with header display lists, and use
it for DU operation on WPF outputs that don't support headerless mode.

Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>
Acked-by: Mauro Carvalho Chehab <[email protected]>

v4l: vsp1: Add support for multiple DRM pipelines

The R-Car H3 ES2.0 VSP-DL instance has two LIF entities and can drive
two display pipelines at the same time. Refactor the VSP DRM code to
support that by introducing a vsp_drm_pipeline object that models one
display pipeline.

Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>
Acked-by: Mauro Carvalho Chehab <[email protected]>

v4l: vsp1: Add support for multiple LIF instances

The VSP2-DL instance (present in the H3 ES2.0 and M3-N SoCs) has two LIF
instances. Adapt the driver infrastructure to support multiple LIFs.
Support for multiple display pipelines will be added separately.

The change to the entity routing table removes the ability to connect
the LIF output to the HGO or HGT histogram generators. This feature is
only available on Gen2 hardware, isn't supported by the rest of the
driver, and has no known use case, so this isn't an issue.

Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>
Acked-by: Mauro Carvalho Chehab <[email protected]>

v4l: vsp1: Add support for new VSP2-BS, VSP2-DL and VSP2-D instances

New Gen3 SoCs come with two new VSP2 variants names VSP2-BS and VSP2-DL,
as well as a new VSP2-D variant on V3M and V3H SoCs. Add new entries for
them in the VSP device info table.

Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>
Acked-by: Mauro Carvalho Chehab <[email protected]>

drm: arcpgu: Allow some clock deviation in crtc->mode_valid() callback

Currently we expect that clock driver produces the exact same value
as we are requiring. There can, and will, be some deviation
however so we need to take that into account instead of just
rejecting the mode.

According to the HDMI spec we have a max of +-0.5% for the pixel clock
frequency deviation. Lets take that into an advantage and use it to
calculate how much deviation we can support.

Signed-off-by: Jose Abreu <[email protected]>
Acked-by: Alexey Brodkin <[email protected]>
Cc: Carlos Palminha <[email protected]>
Cc: Alexey Brodkin <[email protected]>
Cc: Daniel Vetter <[email protected]>
Cc: Dave Airlie <[email protected]>
Signed-off-by: Alexey Brodkin <[email protected]>

drm: arcpgu: Fix module unload

At module unload we are expecting a struct drm_device but at
probing we are not setting it right. Fix this and correct the
arcpgu module unload.

Signed-off-by: Jose Abreu <[email protected]>
Fixes: 0c4250e7b15e ("drm: Add support of ARC PGU display controller")
Cc: Carlos Palminha <[email protected]>
Cc: Alexey Brodkin <[email protected]>
Cc: Daniel Vetter <[email protected]>
Cc: Dave Airlie <[email protected]>
Signed-off-by: Alexey Brodkin <[email protected]>

drm: arcpgu: Fix mmap() callback

Now that ARC properly supports DMA mmap() we can use the standard
CMA helper to map dumb buffers. This makes ARC PGU works with
standard DRM consumer applications like, for example, mpv/mplayer
via DRM. While at it, use the DEFINE_DRM_GEM_CMA_FOPS() helper.

This fixes the use of dumb buffers.

Signed-off-by: Jose Abreu <[email protected]>
Fixes: 0c4250e7b15e ("drm: Add support of ARC PGU display controller")
Cc: Carlos Palminha <[email protected]>
Cc: Alexey Brodkin <[email protected]>
Cc: Daniel Vetter <[email protected]>
Cc: Dave Airlie <[email protected]>
Signed-off-by: Alexey Brodkin <[email protected]>

arcpgu: Simplify driver name

This very minor change is still useful because it aligns
ARC PGU driver name with other DRM drivers and makes usage of
that driver name a bit easier.

For example in libdrm's test app we'll use "arcpgu" instead of
a bit more ugly "drm-arcpgu".

Signed-off-by: Alexey Brodkin <[email protected]>
Cc: Daniel Vetter <[email protected]>

drm/arcpgu: Opt in debugfs

This change adopts debugfs usage for outputting useful data.
As of today we print:
 * Mode and real HW clock values
 * Standard FB info

Code is heavily borrowed from ARM's HDLCD thus adding Liviu in Cc.

Signed-off-by: Alexey Brodkin <[email protected]>
Reviewed-by: Liviu Dudau <[email protected]>
Cc: Daniel Vetter <[email protected]>
Cc: David Airlie <[email protected]>
Cc: Jose Abreu <[email protected]>

Merge branch 'drm-next-4.14' of git://people.freedesktop.org/~agd5f/linux into drm-next

- Stop reprogramming the MC, the vbios already does this in asic_init
- Reduce internal gart to 256M (this does not affect the ttm GTT pool size)
- Initial support for huge pages
- Rework bo migration logic
- Lots of improvements for vega10
- Powerplay fixes
- Additional Raven enablement
- SR-IOV improvements
- Bug fixes
- Code cleanup

* 'drm-next-4.14' of git://people.freedesktop.org/~agd5f/linux: (138 commits)
  drm/amdgpu: fix header on gfx9 clear state
  drm/amdgpu: reduce the time of reading VBIOS
  drm/amdgpu/virtual_dce: Remove the rmmod error message
  drm/amdgpu/gmc9: disable legacy vga features in gmc init
  drm/amdgpu/gmc8: disable legacy vga features in gmc init
  drm/amdgpu/gmc7: disable legacy vga features in gmc init
  drm/amdgpu/gmc6: disable legacy vga features in gmc init (v2)
  drm/radeon: Set depth on low mem to 16 bpp instead of 8 bpp
  drm/amdgpu: fix the incorrect scratch reg number on gfx v6
  drm/amdgpu: fix the incorrect scratch reg number on gfx v7
  drm/amdgpu: fix the incorrect scratch reg number on gfx v8
  drm/amdgpu: fix the incorrect scratch reg number on gfx v9
  drm/amd/powerplay: add support for 3DP 4K@120Hz on vega10.
  drm/amdgpu: enable huge page handling in the VM v5
  drm/amdgpu: increase fragmentation size for Vega10 v2
  drm/amdgpu: ttm_bind only when user needs gpu_addr in bo pin
  drm/amdgpu: correct clock info for SRIOV
  drm/amdgpu/gmc8: SRIOV need to program fb location
  drm/amdgpu: disable firmware loading for psp v10
  drm/amdgpu:fix gfx fence allocate size
  ...

drm/ast: Actually load DP501 firmware when required

The ast driver has a code to load the DP501 firmware, but it's never
used.  This patch implements its actual usage by requesting the
firmware on demand, and release the firmware at exit as well.

Also the path contains a few cleanups and makes relevant functions
static.

Signed-off-by: Egbert Eich <[email protected]>
Signed-off-by: Takashi Iwai <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

drm/ast: Add an crtc_disable callback to the crtc helper funcs

Implement the proper CRTC disablement, just like done in mgag200
driver.

Signed-off-by: Egbert Eich <[email protected]>
Signed-off-by: Takashi Iwai <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

drm/ast: Fix memleak in error path in ast_bo_create()

The allocated struct ast_bo was not freed in all error paths.
This patch consolidates error handling and fixes this.

Signed-off-by: Egbert Eich <[email protected]>
Signed-off-by: Takashi Iwai <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

drm/ast: Free container instead of member in ast_user_framebuffer_destroy()

Technically freeing ast_fb->base is the same as freeing ast_fb as 'base'
the first member of the data structure.
Still this makes it cleaner.

Signed-off-by: Egbert Eich <[email protected]>
Signed-off-by: Takashi Iwai <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

drm/ast: Simplify function ast_bo_unpin()

Just a code refactoring, no functional change.

Signed-off-by: Egbert Eich <[email protected]>
Signed-off-by: Takashi Iwai <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

v4l: vsp1: Add support for the BRS entity

The Blend/ROP Sub Unit (BRS) is a stripped-down version of the BRU found
in several VSP2 instances. Compared to a regular BRU, it supports two
inputs only, and thus has no ROP unit.

Add support for the BRS by modelling it as a new entity type, but reuse
the vsp1_bru object underneath. Chaining the BRU and BRS entities seems
to be supported by the hardware but isn't implemented yet as it isn't
the primary use case for the BRS.

Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>
Acked-by: Mauro Carvalho Chehab <[email protected]>

v4l: vsp1: Add pipe index argument to the VSP-DU API

In the H3 ES2.0 SoC the VSP2-DL instance has two connections to DU
channels that need to be configured independently. Extend the VSP-DU API
with a pipeline index to identify which pipeline the caller wants to
operate on.

Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>
Acked-by: Mauro Carvalho Chehab <[email protected]>

v4l: vsp1: Don't create links for DRM pipeline

When the VSP1 is used in a DRM pipeline the driver doesn't register the
media device. Links between entities are not exposed to userspace, but
are still used internally for the sole purpose of setting up internal
source to sink pointers through the link setup handler.

Instead of going through this complex procedure, remove link creation
and set the sink pointers directly.

Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>
Acked-by: Mauro Carvalho Chehab <[email protected]>

v4l: vsp1: Store source and sink pointers as vsp1_entity

The internal VSP entity source and sink pointers are stored as
media_entity pointers, which are then cast to a vsp1_entity. As all
sources and sinks are vsp1_entity instances, we can store the
vsp1_entity pointers directly.

Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>
Acked-by: Mauro Carvalho Chehab <[email protected]>

v4l: vsp1: Don't set WPF sink pointer

The sink pointer is used to configure routing inside the VSP, and as
such must point to the next VSP entity in the pipeline. The WPF being a
pipeline terminal sink, its output route can't be configured. The
routing configuration code already handles this correctly without
referring to the sink pointer, which thus doesn't need to be set.

Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>
Acked-by: Mauro Carvalho Chehab <[email protected]>

v4l: vsp1: Don't recycle active list at display start

When the display start interrupt occurs, we know that the hardware has
finished loading the active display list. The driver then proceeds to
recycle the list, assuming it won't be needed anymore.

This assumption holds true for headerless display lists, as the VSP
doesn't reload the list for the next frame if it hasn't changed.
However, this isn't true anymore for header display lists, as they are
loaded at every frame start regardless of whether they have been
updated.

To prepare for header display lists usage in display pipelines, we need
to postpone recycling the list until it gets replaced by a new one
through a page flip. The driver already does so in the frame end
interrupt handler, so all we need is to skip list recycling in the
display start interrupt handler.

While the active list can be recycled at display start for headerless
display lists, there's no real harm in postponing that to the frame end
interrupt handler in all cases. This simplifies interrupt handling as we
don't need to process the display start interrupt anymore.

Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>
Acked-by: Mauro Carvalho Chehab <[email protected]>

v4l: vsp1: Fill display list headers without holding dlm spinlock

The display list headers are filled using information from the display
list only. Lower the display list manager spinlock contention by filling
the headers without holding the lock.

Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Kieran Bingham <[email protected]>
Acked-by: Mauro Carvalho Chehab <[email protected]>

drm/amdgpu: fix header on gfx9 clear state

This got missed when we open sourced this.

Reviewed-by: Christian König <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
Cc: [email protected]

drm: linux-next: build failure after merge of the drm-misc tree

Hi all,

After merging the drm-misc tree, today's linux-next build (x86_64
allmodconfig) failed like this:

drivers/staging/vboxvideo/vbox_drv.c:235:2: error: unknown field 'set_busid' specified in initializer
  .set_busid = drm_pci_set_busid,
  ^
drivers/staging/vboxvideo/vbox_drv.c:235:15: error: 'drm_pci_set_busid' undeclared here (not in a function)
  .set_busid = drm_pci_set_busid,
               ^
drivers/staging/vboxvideo/vbox_drv.c: In function 'vbox_init':
drivers/staging/vboxvideo/vbox_drv.c:273:9: error: implicit declaration of function 'drm_pci_init' [-Werror=implicit-function-declaration]
  return drm_pci_init(&driver, &vbox_pci_driver);
         ^
drivers/staging/vboxvideo/vbox_drv.c: In function 'vbox_exit':
drivers/staging/vboxvideo/vbox_drv.c:278:2: error: implicit declaration of function 'drm_pci_exit' [-Werror=implicit-function-declaration]
  drm_pci_exit(&driver, &vbox_pci_driver);
  ^

Caused by commits

  5c484cee7ef9 ("drm: Remove drm_driver->set_busid hook")
  10631d724def ("drm/pci: Deprecate drm_pci_init/exit completely")

interacting with commit

  dd55d44f4084 ("staging: vboxvideo: Add vboxvideo to drivers/staging")

from the staging.current tree.

I have applied the following merge fix patch - please check that it
is correct.

From: Stephen Rothwell <[email protected]>
Date: Wed, 19 Jul 2017 11:41:01 +1000
Subject: [PATCH] drm: fixes for staging due to API changes in the drm core

Signed-off-by: Stephen Rothwell <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

Backmerge tag 'v4.13-rc2' into drm-next

Linux 4.13-rc2

This is required for drm-misc fixing.

Merge tag 'drm-misc-next-2017-07-26' of git://anongit.freedesktop.org/git/drm-misc into drm-next

drm-misc-next-2017-07-18:
Core Changes:
- A couple fixes to only opening crc when needed (Maarten)
- Change atomic helper swap_state to be interruptible (Maarten)
- fb_helper: Support waiting for an output before setting up (Daniel)
- Allow drivers supporting runtime_pm to use helper_commit_tail (Maxime)

Driver Changes:
- misc: Use %pOF to print device node names (Rob)
- Miscellaneous fixes

drm-misc-next-2017-07-18:
UAPI Changes:
- Fail commits which request an event without including a crtc (Andrey)

Core Changes:
- Add YCBCR 4:2:0 support (Shashank)
- s/drm_atomic_replace_property_blob/drm_property_replace_blob/ (Peter)
- Add proper base class for private objs instead of using void* (Ville)
- Remove pending_read/write_domains from drm_gem_object (Chris)
- Add async plane update support (ie: cursor) to atomic helpers (Gustavo)
- Add old state to .enable and rename to .atomic_enable (Laurent)
- Add drm_atomic_helper_wait_for_flip_done() (Boris)
- Remove drm_driver->set_busid hook (Daniel)
- Migrate vblank documentation into the source files (Daniel)
- Add fb_helper->lock instead of abusing modeset lock (Thierry/Daniel)

Driver Changes:
- stm: Add STM32 DSI controller driver (Phillipe)
- amdgpu: Numerous small/misc fixes
- bridge: Add Synopsys Designware MIPI DSI host bridge driver (Phillipe)
- tinydrm: Add support for Pervasive Displays RePaper displays (Noralf)
- misc: Replace for_each_[obj]_in_state to prep for removal (Maarten)
- misc: Use .atomic_disable for atomic drivers (Laurent)
- vgem: Pin pages when mapped/exported (Chris)
- dw_hdmi: Add support for Rockchip RK3399 (Mark)
- atmel-hlcdc: Add 8-bit color look-up table format (Peter)
- vc4: Send vblank event when disabling a crtc (Boris)
- vc4: Use atomic helpers for fence waits (Eric)
- misc: drop drm_vblank_cleanup cargo-cult (Daniel)

Cc: Daniel Vetter <[email protected]>
Cc: Boris Brezillon <[email protected]>
Cc: Eric Anholt <[email protected]>
Cc: Peter Rosin <[email protected]>
Cc: Mark Yao <[email protected]>
Cc: Chris Wilson <[email protected]>
Cc: Andrey Grodzovsky <[email protected]>
Cc: Laurent Pinchart <[email protected]>
Cc: Gustavo Padovan <[email protected]>
Cc: Thierry Reding <[email protected]>
Cc: Ville Syrjälä <[email protected]>
Cc: Peter Rosin <[email protected]>
Cc: Shashank Sharma <[email protected]>
Cc: Philippe CORNU <[email protected]>
Cc: Maarten Lankhorst <[email protected]>
Cc: Rob Herring <[email protected]>
Cc: Maxime Ripard <[email protected]>
* tag 'drm-misc-next-2017-07-26' of git://anongit.freedesktop.org/git/drm-misc: (171 commits)
  drm/hisilicon: fix build error without fbdev emulation
  drm/atomic: implement drm_atomic_helper_commit_tail for runtime_pm users
  drm: Improve kerneldoc for drm_modeset_lock
  drm/hisilicon: Remove custom FB helper deferred setup
  drm/exynos: Remove custom FB helper deferred setup
  drm/fb-helper: Support deferred setup
  dma-fence: Don't BUG_ON when not absolutely needed
  drm: Convert to using %pOF instead of full_name
  drm/syncobj: Fix kerneldoc
  drm/atomic: Allow drm_atomic_helper_swap_state to fail
  drm/atomic: Add __must_check to drm_atomic_helper_swap_state.
  drm/vc4: Handle drm_atomic_helper_swap_state failure
  drm/tilcdc: Handle drm_atomic_helper_swap_state failure
  drm/tegra: Handle drm_atomic_helper_swap_state failure
  drm/msm: Handle drm_atomic_helper_swap_state failure
  drm/mediatek: Handle drm_atomic_helper_swap_state failure
  drm/i915: Handle drm_atomic_helper_swap_state failure
  drm/atmel-hlcdc: Handle drm_atomic_helper_swap_state failure
  drm/nouveau: Handle drm_atomic_helper_swap_state failure
  drm/atomic: Change drm_atomic_helper_swap_state to return an error.
  ...

drm/hisilicon: fix build error without fbdev emulation

We cannot reference priv->fbdev outside of the #ifdef:

drivers/net/virtio_net.c:1881:12: error: 'virtnet_restore_up' defined but not used [-Werror=unused-function]
 static int virtnet_restore_up(struct virtio_device *vdev)
drivers/net/virtio_net.c:1859:13: error: 'virtnet_freeze_down' defined but not used [-Werror=unused-function]
 static void virtnet_freeze_down(struct virtio_device *vdev)

As the #ifdef is a bit annoying here, this removes it entirely
and uses an IS_ENABLED() check in it place where needed.

Fixes: b4dd9f1ffaba ("drm/hisilicon: Remove custom FB helper deferred setup")
Signed-off-by: Arnd Bergmann <[email protected]>
[danvet: One step further, also remove the IS_ENABLED checks, core
no-ops out the fb helper functions that the cma helpers use. Discussed
with Arnd on dri-devel.]
Signed-off-by: Daniel Vetter <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
Signed-off-by: Daniel Vetter <[email protected]>

drm/atomic: implement drm_atomic_helper_commit_tail for runtime_pm users

The current drm_atomic_helper_commit_tail helper works only if the CRTC is
accessible, and documents an alternative implementation that is supposed to
be used if that happens.

That implementation is then duplicated by some drivers. Instead of
documenting it, let's implement an helper that all the relevant users can
use directly.

Signed-off-by: Maxime Ripard <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/a8f92dc70048bab746e94dadd1c23200626aff60.1500555652.git-series.maxime.ripard@free-electrons.com
Signed-off-by: Daniel Vetter <[email protected]>

drm: Improve kerneldoc for drm_modeset_lock

Explain better when the drm_modeset_acquire_ctx parameter can
be skipped for drm_modeset_lock() call.

Signed-off-by: Liviu Dudau <[email protected]>
Cc: Daniel Vetter <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
Signed-off-by: Daniel Vetter <[email protected]>

drm/hisilicon: Remove custom FB helper deferred setup

The FB helper core now supports deferred setup, so the driver's custom
implementation can be removed.

v2: Dont' resurrect drm_vblank_cleanup.

Cc: Xinliang Liu <[email protected]>
Cc: Rongrong Zou <[email protected]>
Cc: Xinwei Kong <[email protected]>
Cc: Chen Feng <[email protected]>
Signed-off-by: Thierry Reding <[email protected]> (v1)
Reviewed-by: Sean Paul <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
Signed-off-by: Daniel Vetter <[email protected]>

drm/exynos: Remove custom FB helper deferred setup

The FB helper core now supports deferred setup, so the driver's custom
implementation can be removed.

v2: Drop NULL check, drm_fb_helper_hotplug_event handles that already.

Cc: Inki Dae <[email protected]>
Cc: Joonyoung Shim <[email protected]>
Cc: Seung-Woo Kim <[email protected]>
Cc: Kyungmin Park <[email protected]>
Signed-off-by: Thierry Reding <[email protected]> (v1)
Reviewed-by: Inki Dae <[email protected]>
Tested-by: Inki Dae <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
Signed-off-by: Daniel Vetter <[email protected]>

drm/fb-helper: Support deferred setup

FB helper code falls back to a 1024x768 mode if no outputs are connected
or don't report back any modes upon initialization. This can be annoying
because outputs that are added to FB helper later on can't be used with
FB helper if they don't support a matching mode.

The fallback is in place because VGA connectors can happen to report an
unknown connection status even when they are in fact connected.

Some drivers have custom solutions in place to defer FB helper setup
until at least one output is connected. But the logic behind these
solutions is always the same and there is nothing driver-specific about
it, so a better alterative is to fix the FB helper core and add support
for all drivers automatically.

This patch adds support for deferred FB helper setup. It checks all the
connectors for their connection status, and if all of them report to be
disconnected marks the FB helper as needing deferred setup. Whet setup
is deferred, the FB helper core will automatically retry setup after a
hotplug event, and it will keep trying until it succeeds.

v2: Rebase onto my entirely reworked fbdev helper locking. One big
difference is that this version again drops&reacquires the fbdev lock
(which is now fb_helper->lock, but before this patch series it was
mode_config->mutex), because register_framebuffer must be able to
recurse back into fbdev helper code for the initial screen setup.

v3: __drm_fb_helper_initial_config must hold fb_helper->lock upon
return, I've fumbled that in the deferred setup case (Liviu).

v4: I was blind, redo this all. __drm_fb_helper_initial_config
shouldn't need to reacquire fb_helper->lock, that just confuses
callers. I myself got confused by kernel_fb_helper_lock and somehow
thought it's the same as fb_helper->lock. Tsk.

Also simplify the logic a bit (we don't need two functions to probe
connectors), we can stick much closer to the existing code. And update
some comments I've spotted that are outdated.

v5: Don't pass -EAGAIN to drivers, it's just an internal error code
(Liviu).

v6: Add _and_unlock suffix to clarify locking (Maarten)

Cc: Liviu Dudau <[email protected]>
Cc: John Stultz <[email protected]>
Cc: Thierry Reding <[email protected]>
Cc: Maarten Lankhorst <[email protected]>
Tested-by: John Stultz <[email protected]>
Signed-off-by: Thierry Reding <[email protected]> (v1)
Reviewed-by: Maarten Lankhorst <[email protected]>
Reviewed-by: Liviu Dudau <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
Signed-off-by: Daniel Vetter <[email protected]>

dma-fence: Don't BUG_ON when not absolutely needed

It makes debugging a massive pain.

Signed-off-by: Daniel Vetter <[email protected]>
Cc: Sumit Semwal <[email protected]>
Cc: Gustavo Padovan <[email protected]>
Cc: [email protected]
Cc: [email protected]
Reviewed-by: Lucas Stach <[email protected]>
Acked-by: Sumit Semwal <[email protected]>
Acked-by: Maarten Lankhorst <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
Signed-off-by: Daniel Vetter <[email protected]>

drm: Convert to using %pOF instead of full_name

Now that we have a custom printf format specifier, convert users of
full_name to use %pOF instead. This is preparation to remove storing
of the full path string for each node.

Signed-off-by: Rob Herring <[email protected]>
Cc: Russell King <[email protected]>
Cc: David Airlie <[email protected]>
Cc: Daniel Vetter <[email protected]>
Cc: Jani Nikula <[email protected]>
Cc: Sean Paul <[email protected]>
Cc: Inki Dae <[email protected]>
Cc: Joonyoung Shim <[email protected]>
Cc: Seung-Woo Kim <[email protected]>
Cc: Kyungmin Park <[email protected]>
Cc: Kukjin Kim <[email protected]>
Cc: Krzysztof Kozlowski <[email protected]>
Cc: Javier Martinez Canillas <[email protected]>
Cc: Xinliang Liu <[email protected]>
Cc: Rongrong Zou <[email protected]>
Cc: Xinwei Kong <[email protected]>
Cc: Chen Feng <[email protected]>
Cc: CK Hu <[email protected]>
Cc: Philipp Zabel <[email protected]>
Cc: Matthias Brugger <[email protected]>
Cc: Neil Armstrong <[email protected]>
Cc: Carlo Caione <[email protected]>
Cc: Kevin Hilman <[email protected]>
Cc: Thierry Reding <[email protected]>
Cc: Laurent Pinchart <[email protected]>
Cc: Mark Yao <[email protected]>
Cc: Heiko Stuebner <[email protected]>
Cc: Maxime Ripard <[email protected]>
Cc: Chen-Yu Tsai <[email protected]>
Cc: Jyri Sarha <[email protected]>
Cc: Tomi Valkeinen <[email protected]>
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Partially-Reviewed-by: Laurent Pinchart <[email protected]>
Reviewed-by: Philipp Zabel <[email protected]>
Acked-by: Maxime Ripard <[email protected]>
[seanpaul changed subject prefix and fixed conflict in stm/ltdc.c]
Signed-off-by: Sean Paul <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>

Merge airlied/drm-next into drm-misc-next

I need this to be able to apply the deferred fbdev setup patches, I
need the relevant prep work that landed through the drm-intel tree.

Also squash in conflict fixup from Laurent Pinchart.

Signed-off-by: Daniel Vetter <[email protected]>

drm/syncobj: Fix kerneldoc

make htmldocs helps with catching these.

Cc: Dave Airlie <[email protected]>
Acked-by: Dave Airlie <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
Signed-off-by: Daniel Vetter <[email protected]>

drm/atomic: Allow drm_atomic_helper_swap_state to fail

Signed-off-by: Maarten Lankhorst <[email protected]>
Link: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Reviewed-by: Daniel Vetter <[email protected]>
Reviewed-by: Sean Paul <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>

drm/atomic: Add __must_check to drm_atomic_helper_swap_state.

Now that all drivers check the return value, convert swap_state to
__must_check. This is done separately to force build warnings if we
missed a driver.

Signed-off-by: Maarten Lankhorst <[email protected]>
Link: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Reviewed-by: Daniel Vetter <[email protected]>
Reviewed-by: Sean Paul <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>

drm/vc4: Handle drm_atomic_helper_swap_state failure

drm_atomic_helper_swap_state() will be changed to interruptible waiting
in the next few commits, so all drivers have to be changed to handling
failure.

VC4 has its own nonblocking modeset tracking through the vc4->async_modeset
semaphore, so it doesn't need to stall in swap_state. Pass stall = false
and BUG_ON when it returns an error. This should never happen for !stall.

Signed-off-by: Maarten Lankhorst <[email protected]>
Cc: Eric Anholt <[email protected]>
Link: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Reviewed-by: Daniel Vetter <[email protected]>
Reviewed-by: Sean Paul <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>

drm/tilcdc: Handle drm_atomic_helper_swap_state failure

drm_atomic_helper_swap_state() will be changed to interruptible waiting
in the next few commits, so all drivers have to be changed to handling
failure.

Signed-off-by: Maarten Lankhorst <[email protected]>
Cc: Jyri Sarha <[email protected]>
Cc: Tomi Valkeinen <[email protected]>
Link: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Reviewed-by: Sean Paul <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>

drm/tegra: Handle drm_atomic_helper_swap_state failure

drm_atomic_helper_swap_state() will be changed to interruptible waiting
in the next few commits, so all drivers have to be changed to handling
failure.

Signed-off-by: Maarten Lankhorst <[email protected]>
Cc: Thierry Reding <[email protected]>
Cc: Jonathan Hunter <[email protected]>
Cc: [email protected]
Link: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Reviewed-by: Daniel Vetter <[email protected]>
Reviewed-by: Sean Paul <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>

drm/msm: Handle drm_atomic_helper_swap_state failure

drm_atomic_helper_swap_state() will be changed to interruptible waiting
in the next few commits, so all drivers have to be changed to handling
failure.

MSM has its own busy tracking, which means the swap_state call can be
done with stall = false, in which case it should never return an error.
Handle failure with BUG_ON for this reason.

Signed-off-by: Maarten Lankhorst <[email protected]>
Cc: Rob Clark <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Reviewed-by: Daniel Vetter <[email protected]>
Reviewed-by: Sean Paul <[email protected]>
Tested-by: Archit Taneja <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>

drm/mediatek: Handle drm_atomic_helper_swap_state failure

drm_atomic_helper_swap_state() will be changed to interruptible waiting
in the next few commits, so all drivers have to be changed to handling
failure.

Signed-off-by: Maarten Lankhorst <[email protected]>
Cc: CK Hu <[email protected]>
Cc: Philipp Zabel <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Reviewed-by: Philipp Zabel <[email protected]>
Reviewed-by: Sean Paul <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>

drm/i915: Handle drm_atomic_helper_swap_state failure

drm_atomic_helper_swap_state() will be changed to interruptible waiting
in the next few commits, so all drivers have to be changed to handling
failure.

Signed-off-by: Maarten Lankhorst <[email protected]>
Link: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Reviewed-by: Sean Paul <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>

drm/atmel-hlcdc: Handle drm_atomic_helper_swap_state failure

drm_atomic_helper_swap_state() will be changed to interruptible waiting
in the next few commits, so all drivers have to be changed to handling
failure.

Atmel tracks pending commits through dc->commit.pending, so it can
ignore the changes by setting stall = false. We never return failure in
this case, so make failure a BUG_ON.

Signed-off-by: Maarten Lankhorst <[email protected]>
Cc: Boris Brezillon <[email protected]>
Link: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Reviewed-by: Sean Paul <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>

drm/nouveau: Handle drm_atomic_helper_swap_state failure

drm_atomic_helper_swap_state() will be changed to interruptible waiting
in the next few commits, so all drivers have to be changed to handling
failure.

Signed-off-by: Maarten Lankhorst <[email protected]>
Cc: Ben Skeggs <[email protected]>
Cc: [email protected]
Link: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Reviewed-by: Sean Paul <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>

drm/atomic: Change drm_atomic_helper_swap_state to return an error.

We want to change swap_state to wait indefinitely, but to do this
swap_state should wait interruptibly. This requires propagating
the error to each driver.

Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Signed-off-by: Maarten Lankhorst <[email protected]>
Link: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Reviewed-by: Daniel Vetter <[email protected]>
[mlankhorst: Fix typos in swap_state documentation (seanpaul)]
Reviewed-by: Sean Paul <[email protected]>]
Signed-off-by: Daniel Vetter <[email protected]>

drm/nouveau: Fix error handling in nv50_disp_atomic_commit

Make it more clear that post commit return ret is really return 0,

and add a missing drm_atomic_helper_cleanup_planes when
drm_atomic_helper_wait_for_fences fails.

Fixes: 839ca903f12e ("drm/nouveau/kms/nv50: transition to atomic interfaces internally")
Cc: Ben Skeggs <[email protected]>
Cc: [email protected]
Cc: [email protected]
Cc: <[email protected]> # v4.10+
Signed-off-by: Maarten Lankhorst <[email protected]>
Link: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Reviewed-by: Sean Paul <[email protected]>
[mlankhorst: Use if (ret) to remove the goto in success case.]
Reviewed-by: Daniel Vetter <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>

drm/<drivers>: Drop fbdev info flags

- FBINFO_CAN_FORCE_OUTPUT has been a lie ever since we nerfed&removed
  the entire panic handling code in our fbdev emulation. We might
  restore kms panic output, but not through the bazillion of legacy
  code layers called fbdev/fbcon, there's just no way to make that
  work safely.

- With the module check change FBINFO_DEFAULT is always 0, so can be
  removed too.

That removes another change to cargo-cult stuff in kms drivers, yay!

Reviewed-by: Sean Paul <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
Signed-off-by: Daniel Vetter <[email protected]>

drm/qxl: Drop fbdev hwaccel flags

It's not accelarated, just system memory. Note we don't even need to
set the default flag since that's now always 0.

Note that originally qxl had acceleration support, but that was all
ripped out in

commit c0fe07aa50befe2e6e6525181e2080377a1c1494
Author: Gerd Hoffmann <[email protected]>
Date:   Tue May 5 13:52:49 2015 +0200

    drm/qxl: rewrite framebuffer support

v2: Amend commit message a bit after irc chat with Dave.

Cc: Dave Airlie <[email protected]>
Cc: Gerd Hoffmann <[email protected]>
Cc: [email protected]
Acked-by: Dave Airlie <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
Signed-off-by: Daniel Vetter <[email protected]>

drm: Update docs around gem_free_object

Not all places correctly stated that gem_free_object_unlocked is the
one to use.

Reported-by: Eric Anholt <[email protected]
Cc: Eric Anholt <[email protected]
Reviewed-by: Eric Anholt <[email protected]>
Signed-off-by: Daniel Vetter <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
Signed-off-by: Daniel Vetter <[email protected]>

drm/amdgpu: reduce the time of reading VBIOS

VRAM is usually marked write combined, so change ioremap mode from
noncache to write combine for reading vbios from VRAM.

This will reduce cost time of reading vbios from 188ms to 8ms.

Signed-off-by: Xiangliang Yu <[email protected]>
Reviewed-by: Christian König <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu/virtual_dce: Remove the rmmod error message

Remove the error message "[drm:amdgpu_irq_disable_all
[amdgpu]] *ERROR* error disabling interrupt (-22)".

For virtual dce, it only use AMDGPU_CRTC_IRQ_VBLANK1 -
AMDGPU_CRTC_IRQ_VBLANK6, and don't use AMDGPU_CRTC_IRQ_VLINE1
- AMDGPU_CRTC_IRQ_VLINE6. And when rmmod amdgpu, it will disable
all interrupts, it will return error when the type of crtc irq
interrupt is AMDGPU_CRTC_IRQ_VLINE1 - AMDGPU_CRTC_IRQ_VLINE6.

BUG: SWDEV-121607

Signed-off-by: Emily Deng <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu/gmc9: disable legacy vga features in gmc init

Needs to be done when the MC is set up.

Acked-by: Christian König <[email protected]>
Acked-by: Harry Wentland <[email protected]>
Reviewed-by: Junwei Zhang <[email protected]>
Reviewed-by: Michel Dänzer <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu/gmc8: disable legacy vga features in gmc init

Needs to be done when the MC is set up.

Acked-by: Christian König <[email protected]>
Acked-by: Harry Wentland <[email protected]>
Reviewed-by: Junwei Zhang <[email protected]>
Reviewed-by: Michel Dänzer <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu/gmc7: disable legacy vga features in gmc init

Needs to be done when the MC is set up.

Acked-by: Christian König <[email protected]>
Acked-by: Harry Wentland <[email protected]>
Reviewed-by: Junwei Zhang <[email protected]>
Reviewed-by: Michel Dänzer <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu/gmc6: disable legacy vga features in gmc init (v2)

Needs to be done when the MC is set up.

v2: make consistent with other asics

Acked-by: Christian König <[email protected]>
Acked-by: Harry Wentland <[email protected]>
Reviewed-by: Junwei Zhang <[email protected]>
Reviewed-by: Michel Dänzer <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/radeon: Set depth on low mem to 16 bpp instead of 8 bpp

The radeon driver reduces the framebuffer resolution to 8bpp if a
device with less than 32MB VRAM is found.  This causes the framebuffer
to run in 8 bit paletted mode.  For a text console this is not an
issue as 256 different colors is more than one gets on a VGA text
console.  However this leads to a poor 8bit pseudo-color visual when
running X on fbdev, too, which is quite ugly.

In this patch, we try to give some moderate compromise: limit the
framebuffer bpp to 8 only when VRAM is 8MB or less, and use 16 bpp
otherwise for 32MB or less VRAM.

Reviewed-by: Michel Dänzer <[email protected]>
Signed-off-by: Egbert Eich <[email protected]>
Signed-off-by: Takashi Iwai <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu: fix the incorrect scratch reg number on gfx v6

Signed-off-by: Huang Rui <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Reviewed-by: Junwei Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu: fix the incorrect scratch reg number on gfx v7

Signed-off-by: Huang Rui <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Reviewed-by: Junwei Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu: fix the incorrect scratch reg number on gfx v8

Signed-off-by: Huang Rui <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Reviewed-by: Junwei Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu: fix the incorrect scratch reg number on gfx v9

Signed-off-by: Huang Rui <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Reviewed-by: Junwei Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amd/powerplay: add support for 3DP 4K@120Hz on vega10.

Signed-off-by: Rex Zhu <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu: enable huge page handling in the VM v5

The hardware can use huge pages to map 2MB of address space with only one PDE.

v2: few cleanups and rebased
v3: skip PT updates if we are using the PDE
v4: rebased, added support for CPU based updates
v5: fix CPU based updates once more
v6: fix ndw estimation

Signed-off-by: Christian König <[email protected]>
Reviewed-and-tested-by: Felix Kuehling <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu: increase fragmentation size for Vega10 v2

The fragment bits work differently for Vega10 compared to previous generations.

Increase the fragment size to 2MB for now to better handle that.

v2: handle the hardware setup as well

Signed-off-by: Christian König <[email protected]>
Reviewed-and-tested-by: Felix Kuehling <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu: ttm_bind only when user needs gpu_addr in bo pin

Signed-off-by: Chunming Zhou <[email protected]>
Reviewed-by: Christian König <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu: correct clock info for SRIOV

Currently, get clock info from default clk of pm if dpm is disable.
Buf SRIOV doesn't support dpm and pm, can't get anything from pm.
Only get clock info only from default clk of amdgpu for SRIOV.

And driver get pm default clk also from amdgpu default clk and never
be changed by others. So use amdgpu default clk value for SRIOV
and non-dpm cases.

Signed-off-by: Xiangliang Yu <[email protected]>
Acked-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu/gmc8: SRIOV need to program fb location

SRIOV won't do vbios post in guest OS, and the mmMC_VM_FB_LOCATION
is pf and vf copy, so still need to program fb location for SRIOV.

v2: No need to stop mc, and update gmc_v8_0_vram_gtt_location as well.

v3: New line after the stack variables

BUG: SWDEV-126629

Signed-off-by: Emily Deng <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu: disable firmware loading for psp v10

Now asd firmware is not ready for psp v10, will enable it when it's available

Signed-off-by: Junwei Zhang <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu:fix gfx fence allocate size

1, for sriov, we need 8dw for the gfx fence due to CP
behaviour
2, cleanup wrong logic in wptr/rptr wb alloc and free

Change-Id: Ifbfed17a4621dae57244942ffac7de1743de0294
Signed-off-by: Monk Liu <[email protected]>
Signed-off-by: Xiangliang Yu <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Reviewed-by: Christian König <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu: Implement ttm_bo_driver.access_memory callback v2

Allows gdb to access contents of user mode mapped VRAM BOs.

v2: return error for non-VRAM pools

Signed-off-by: Felix Kuehling <[email protected]>
Reviewed-by: Michel Dänzer <[email protected]>
Reviewed-by: Christian König <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/ttm: Implement vm_operations_struct.access v2

Allows gdb to access contents of user mode mapped BOs. System memory
is handled by TTM using kmap. Other memory pools require a new driver
callback in ttm_bo_driver.

v2:
* kmap only one page at a time
* swap in BO if needed
* make driver callback more generic to handle private memory pools
* document callback return value
* WARN_ON -> WARN_ON_ONCE

Signed-off-by: Felix Kuehling <[email protected]>
Reviewed-by: Christian König <[email protected]>
Reviewed-by: Michel Dänzer <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amd/powerplay: fix AVFS voltage offset for Vega10

Signed-off-by: Eric Huang <[email protected]>
Acked-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu: read reg in each iterator of psp_wait_for loop

v2: fix the SOS loading failure for PSP v3.1

Signed-off-by: Junwei Zhang <[email protected]>
Cc: [email protected]
Acked-by: Alex Deucher <[email protected]> (v1)
Acked-by: Huang Rui <[email protected]> (v1)
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu/gfx9: simplify and fix GRBM index selection

Copy the approach taken by gfx8, which simplifies the code, and set the
instance index properly. The latter is required for debugging, e.g. for
reading wave status by UMR.

Signed-off-by: Nicolai Hähnle <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu: add ring_destroy for psp v10

Signed-off-by: Junwei Zhang <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu: add ring_create function for psp v10

Signed-off-by: Junwei Zhang <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu: add init microcode function for psp v10

Signed-off-by: Junwei Zhang <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

drm/amdgpu: remove unncessary code in psp v10 ring init func

Signed-off-by: Junwei Zhang <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>