drm/amd/display: remove min reduction for abm 2.2 level 3

[Why]
Image brightness compensation for solid color full screen images is
expected to be optimal for ABM 2.2 at level 3. The min reduction that is
currently being enforced prevents this from being achieved.

[How]
Remove the min reduction for ABM 2.2 at level 3

Signed-off-by: Josip Pavic <Josip.Pavic@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.25

Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: return correct dc_status for dcn10_validate_global

[Why]
Before it was returning false in the case of failure even though return type should be enum dc_status

[How]
Return DC_FAIL_UNSUPPORTED_1 instead

Signed-off-by: Su Sung Chung <Su.Chung@amd.com>
Reviewed-by: Eric Yang <eric.yang2@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Use plane->color_space for dpp if specified

[Why]
The input color space for the plane was previously ignored even if it
was set.

If a limited range YUV format was given to DC then the
wrong color transformation matrix was being used since DC assumed that
it was full range instead.

[How]
Respect the given color_space format for the plane if it isn't
COLOR_SPACE_UNKNOWN. Otherwise, use the implicit default since DM
didn't specify.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Sun peng Li <Sunpeng.Li@amd.com>
Acked-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Calculate link bandwidth in a common function

[why]
Currently link bandwidth is calculated in two places, using the same
formula. They should be unified into calling one function.

[how]
Replace all implementations of link bandwidth calculation with a call
to a function.

Signed-off-by: Nikola Cornij <nikola.cornij@amd.com>
Reviewed-by: Nikola Cornij <Nikola.Cornij@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix clk_mgr naming

clk_mgr is called dccg in dc_state, this change fixes that

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Eric Bernstein <Eric.Bernstein@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: use proper formula to calculate bandwidth from timing

[why]
The existing calculation uses a wrong formula to
calculate bandwidth from timing.

[how]
Expose the existing proper function that calculates the bandwidth,
so dc_link can use it to calculate timing bandwidth correctly.

Signed-off-by: Wenjing Liu <Wenjing.Liu@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix dp_hdmi_max_pixel_clk units

[Why]
We are incorrectly using dp_hdmi_max_pixel_clk because the units are not clear.

[How]
Rename to dp_hdmi_max_pixel_clk_in_khz, and change mode timing validation to use
the value correctly.

Signed-off-by: SivapiriyanKumarasamy <sivapiriyan.kumarasamy@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add debugfs entry for amdgpu_dm_visual_confirm

[Why]
DC provides a few visual confirmation debug options that can be
dynamically changed at runtime to help debug surface programming issues
but we don't have any way to access it from userspace.

[How]
Add the amdgpu_dm_visual_confirm debugfs entry.
It accepts a string containing the DC visual confirm enum value using
the debugfs attribute helpers.

The debugfs_create_file_unsafe can be used instead of
debugfs_create_file as per the documentation.

v2: Use debugfs helpers for getting and setting the value (Christian)

Cc: Leo Li <sunpeng.li@amd.com>
Cc: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Initialize stream_update with memset

The brace initialization used here generates warnings on some
compilers. For example, on GCC 4.9:

[...] In function ‘dm_determine_update_type_for_commit’:
[...] error: missing braces around initializer [-Werror=missing-braces]
   struct dc_stream_update stream_update = { 0 };
          ^

Use memset to make this more portable.

v2: Specify the compiler / diagnostic in the commit message (Paul)

Cc: Sun peng Li <Sunpeng.Li@amd.com>
Cc: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove semicolon from to_dm_plane_state definition

The extra ; in the macro definition creates an empty statement
preventing any variable declarations from occuring after
any use of to_dm_plane_state(...).

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: support dpm level modification under virtualization v3

Under vega10 virtualuzation, smu ip block will not be added.
Therefore, we need add pp clk query and force dpm level function
at amdgpu_virt_ops to support the feature.

v2: add get_pp_clk existence check and use kzalloc to allocate buf

v3: return -ENOMEM for allocation failure and correct the coding style

Signed-off-by: Yintian Tao <yttao@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Always enable memory sharing within same XGMI hive

XGMI Memory sharing will be disbaled by default for security reason after
boot up, it depends on driver to enable the memory sharing

Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
Acked-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add a check to avoid panic because of unexpected irqs

IP initialize ras in late_init, because of the BUGs of PSP or any
other components, driver receives unexpected irqs. It is ok to add such
check anyway.

Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: sdma use amdgpu_ras_feature_enable_on_boot

handle ras enable on boot.

Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: gmc use amdgpu_ras_feature_enable_on_boot

handle ras enable on boot.

Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: gfx use amdgpu_ras_feature_enable_on_boot

handle ras enable on boot.

Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Introduce another ras enable function

Many parts of the whole SW stack can program the ras enablement state
during the boot. Now we handle that case by adding one function which
check the ras flags and choose different code path.

Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Make default ras error type to none

Unless IP has implemented its own ras, use ERROR_NONE as the default
type.

Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: simplify the code of [get|set]_activity_monitor_coeff

use smu_update_table_with_arg to replace old code logic

Signed-off-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: optimization function of smu_update_table

in fact, the firmware need 2 parameter: 1.table_id, 2.XferArg
so change the function interface to match the firmware code

Signed-off-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: check for invalid profile mode before switching

Need to check for invalid profile mode settings before determining
to switch to that.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Prevent cursor hotspot overflow for RV overlay planes

[Why]
The actual position for the cursor on the screen is essentially:

x_out = x - x_plane - x_hotspot
y_out = y - y_plane - y_hotspot

The register values for cursor position and cursor hotspot need to be
greater than zero when programmed, but we also need to subtract off
the plane position to display the cursor at the correct position.

Since we don't want x or y to be less than zero, we add the plane
position as a positive value to x_hotspot or y_hotspot. However, what
this doesn't take into account is that the hotspot registers are limited
by the maximum cursor size.

On DCN10 the cursor hotspot regitsers are masked to 0xFF, so they have
a maximum value of 0-255. Values greater this will wrap, causing the
cursor to display in the wrong position.

In practice this means that for sufficiently large plane positions, the
cursor will be drawn twice on the screen, and can cause screen flashes
or p-state WARNS depending on what the wrapped value is.

So we need a way to remove the value from x_plane and y_plane without
exceeding the maximum cursor size.

[How]
Subtract as much as x_plane/y_plane as possible from x and y and place
the remainder in the cursor hotspot register.

The value for x_hotspot and y_hotspot can still wrap around but it
won't happen in a case where the cursor is actually enabled.

The cursor plane needs to intersect at least one pixel of the plane's
rectangle to be enabled, so the cursor position + hotspot provided by
userspace must always be strictly less than the maximum cursor size for
the cursor to actually be enabled.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Sun peng Li <Sunpeng.Li@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix "dc has no member named dml" compile error

For DCN disabled builds, dc->dml is stripped out. Therefore, guard usage
in dc_create_state() with CONFIG_DRM_AMD_DC_DCN1_0.

It fixes the following error:

drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc.c: In function 'dc_create_state':
>> drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc.c:1237:34: error: 'struct dc' has no member named 'dml'
     memcpy(&context->bw_ctx.dml, &dc->dml, sizeof(struct display_mode_lib));
                                     ^~

Signed-off-by: Leo Li <sunpeng.li@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix multi-thread writing to 1 state

[Why]
Multiple threads were writing back to one global VBA in DC resulting
in multiple threads overwriting eachother's data

[How]
Add an instance of DML (which contains VBA) to each context and
change all calls that used dc->dml to use context->dml. Created a
seperate copy constructor for linux in a case where there is no
access to DC.

Signed-off-by: Aidan Wood <Aidan.Wood@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Make pageflip event delivery compatible with VRR.

We want vblank counts and timestamps of flip completion as sent
in pageflip completion events to be consistent with the vblank
count and timestamp of the vblank of flip completion, like in non
VRR mode.

In VRR mode, drm_update_vblank_count() - and thereby vblank
count and timestamp updates - must be delayed until after the
end of front-porch of each vblank, as it is only safe to
calculate vblank timestamps outside of the front-porch, when
we actually know when the vblank will end or has ended.

The function drm_update_vblank_count() which updates timestamps
and counts gets called by drm_crtc_accurate_vblank_count() or by
drm_crtc_handle_vblank().

Therefore we must make sure that pageflip events for a completed
flip are only sent out after drm_crtc_accurate_vblank_count() or
drm_crtc_handle_vblank() is executed, after end of front-porch
for the vblank of flip completion.

Two cases:

a) Pageflip irq handler executes inside front-porch:
   In this case we must defer sending pageflip events until
   drm_crtc_handle_vblank() executes after end of front-porch,
   and thereby calculates proper vblank count and timestamp.
   Iow. the pflip irq handler must just arm a pageflip event
   to be sent out by drm_crtc_handle_vblank() later on.

b) Pageflip irq handler executes after end of front-porch, e.g.,
   after flip completion in back-porch or due to a massively
   delayed handler invocation into the active scanout of the new
   frame. In this case we can call drm_crtc_accurate_vblank_count()
   to safely force calculation of a proper vblank count and
   timestamp, and must send the pageflip completion event
   ourselves from the pageflip irq handler.

   This is the same behaviour as needed for standard fixed refresh
   rate mode.

To decide from within pageflip handler if we are in case a) or b),
we check the current scanout position against the boundary of
front-porch. In non-VRR mode we just do what we did in the past.

Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: In VRR mode, do DRM core vblank handling at end of vblank. (v2)

In VRR mode, proper vblank/pageflip timestamps can only be computed
after the display scanout position has left front-porch. Therefore
delay calls to drm_crtc_handle_vblank(), and thereby calls to
drm_update_vblank_count() and pageflip event delivery, to after the
end of front-porch when in VRR mode.

We add a new vupdate irq, which triggers at the end of the vupdate
interval, ie. at the end of vblank, and calls the core vblank handler
function. The new irq handler is not executed in standard non-VRR
mode, so vblank handling for fixed refresh rate mode is identical
to the past implementation.

v2: Implement feedback by Nicholas and Paul Menzel.

Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Rework vrr flip throttling for late vblank irq.

For throttling to work correctly, we always need a baseline vblank
count last_flip_vblank that increments at start of front-porch.

This is the case for drm_crtc_vblank_count() in non-VRR mode, where
the vblank irq fires at start of front-porch and triggers DRM core
vblank handling, but it is no longer the case in VRR mode, where
core vblank handling is done later, after end of front-porch.

Therefore drm_crtc_vblank_count() is no longer useful for this.
We also can't use drm_crtc_accurate_vblank_count(), as that would
screw up vblank timestamps in VRR mode when called in front-porch.

To solve this, use the cooked hardware vblank counter returned by
amdgpu_get_vblank_counter_kms() instead, as that one is cooked to
always increment at start of front-porch, independent of when
vblank related irq's fire.

This patch allows vblank irq handling to happen anywhere within
vblank of even after it, without a negative impact on flip
throttling, so followup patches can shift the vblank core
handling trigger point wherever they need it.

Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Prevent vblank irq disable while VRR is active. (v3)

During VRR mode we can not allow vblank irq dis-/enable
transitions, as an enable after a disable can happen at
an arbitrary time during the video refresh cycle, e.g.,
with a high likelyhood inside vblank front-porch. An
enable during front-porch would cause vblank timestamp
updates/calculations which are completely bogus, given
the code can't know when the vblank will end as long
as we are in front-porch with no page flip completed.

Hold a permanent vblank reference on the crtc while
in active VRR mode to prevent a vblank disable, and
drop the reference again when switching back to fixed
refresh rate non-VRR mode.

v2: Make sure transition is also handled if vrr is
    disabled and stream gets disabled in the same
    atomic commit by moving the call to the transition
    function outside of plane commit.
    Suggested by Nicholas.

v3: Trivial rebase onto previous patch.

Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Update VRR state earlier in atomic_commit_tail.

We need the VRR active/inactive state info earlier in
the commit sequence, so VRR related setup functions like
amdgpu_dm_handle_vrr_transition() know the final VRR state
when they need to do their hw setup work.

Split update_freesync_state_on_stream() into an early part,
that can run at the beginning of commit tail before the
vrr transition handling, and a late part that must run after
vrr transition handling inside the commit planes code for
enabled crtc's.

Suggested by Nicholas Kazlauskas.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Allow switching to CUSTOM profile on Vega20

Vega20 stores a CUSTOM profile on the GPU, but it may not be valid. Add
a bool to vega20_hwmgr to determine whether or not a valid CUSTOM
profile has been set, and use that to check when a user requests
switching to the CUSTOM profile without passing in any arguments. Then
if the CUSTOM profile has been set already, we can switch to it without
providing the parameters again

Signed-off-by: Kent Russell <kent.russell@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Allow switching to CUSTOM profile on Vega10 v2

Don't return an error if the CUSTOM profile is selected, just apply it
with the values saved to the GPU. But ensure that we zero out the
copy stored in adev to ensure that a valid profile has been submitted at
some point first

v2: Fix comment that wasn't updated from previous patch

Signed-off-by: Kent Russell <kent.russell@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Allow switching to CUSTOM profile on smu7 v2

Allow changing to the CUSTOM profile without requiring the
parameters being passed in each time. Store the values in
the smu7_profiling table since it's defined here anyways

v2: Add check that CUSTOM was previously set

Signed-off-by: Kent Russell <kent.russell@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add preferred_domain check when determine XGMI state

Avoid unnecessary XGMI hight pstate trigger when mapping none-vram memory for peer device

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

drm/amdgpu: provide the page fault queue to the VM code

We are going to need that for recoverable page faults.

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

drm/amdgpu: handle leaf PDEs as PTEs on Vega

This way we get retry faults for missing PDs.

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

drm/amdgpu: fix ATC handling for Ryzen

Otherwise we don't correctly use translate further.

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

drm/amdgpu: Adjust TMR address alignment as per HW requirement

According to HW engineer, they prefer the TMR address be "naturally aligned", e.g. the start address
must be an integer divide of TME size.

Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Correct the irq types' num of sdma

Fix the issue about TDR-2 will have "fallback timer expired on ring sdma1".
It is because the wrong number of irq types setting.

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

Merge commit 'refs/for-upstream/mali-dp' of git://linux-arm.org/linux-ld into drm-next

This pull requests adds initial Mali D71 support into the Arm "komeda" DRM
driver. The code has been reviewed at the end of last year, I just been
too slow with pushing it into mainline. Since it started baking in
linux-next we had a kbuild-bot issue raised and one from Joe Perches on
the MAINTAINERS entry, for which I'm including fixes here.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Liviu Dudau <Liviu.Dudau@arm.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190401192833.GW21747@e110455-lin.cambridge.arm.com

Merge branch 'drm-next-5.2' of git://people.freedesktop.org/~agd5f/linux into drm-next

amdgpu:
- Switch to HMM for userptr (reverted until HMM fixes land)
- New experimental SMU 11 replacement for powerplay for vega20 (not enabled by default)
- Initial RAS support for vega20
- BACO support for vega12
- BACO fixes for vega20
- Rework IH handling for page fault and retry interrupts
- Cleanly split CPU and GPU paths for GPUVM updates
- Powerplay fixes
- XGMI fixes
- Rework how DC interacts with atomic for planes
- Clean up and simplify DC/Powerplay interfaces
- Misc cleanups and bug fixes

amdkfd:
- Switch to HMM for userptr (reverted until HMM fixes land)
- Add initial RAS support
- MQD fixes

ttm:
- Unify DRM_FILE_PAGE_OFFSET handling
- Account for kernel allocations in kernel zone only
- Misc cleanups

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexdeucher@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190402170820.22197-1-alexander.deucher@amd.com

drm/amdgpu/smu11: fix warning on 32bit arches

Fixes
warning: cast from pointer to integer of different size
[-Wpointer-to-int-cast]
on 32 bit platforms.

Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

MAINTAINERS: Fix pattern for Documentation path for Arm Mali Komeda

The Device Tree bindings documentation file is in a simplified path
that doesn't match the MAINTAINERS entry.

Reported-by: Joe Perches <joe@perches.com>
Reviewed-by: James Qian Wang (Arm Technology China) <james.qian.wang@arm.com>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>

arm/komeda: Compile komeda_debugfs_init() only if CONFIG_DEBUG_FS is enabled

We don't call this function if CONFIG_DEBUG_FS is not defined, but we
should not be compiling it either, as the declaration of the debugfs
core functions is not included.

Reported by the kbuild test robot.

Reviewed-by: James Qian Wang (Arm Technology China) <james.qian.wang@arm.com>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>

drm/komeda: Add debugfs node "register" for register dump

Add a debugfs node "register" and entry function dump_register to
dev/pipeline/component to register dump, then user can read
"/sys/kernel/debug/komeda/register" to get the register values via these
chip function.

Signed-off-by: James Qian Wang (Arm Technology China) <james.qian.wang@arm.com>
[Added the d71_layer_dump() function that was in a previous commit]
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>

drm/komeda: Add irq handling

1. Added irq_handler/irq_enable/irq_disable to komeda_dev_func, then the
   Komeda-CORE can control the HW irq via these chip function.
2. Install irq and register irq_handler to system by DRM, so once the IRQ
   coming, the handling sequence is:

   komeda_kms_irq_handler(int irq, void *data)
        /* step 1. call into the CHIP to recognize event */
mdev->funcs->irq_handler(mdev, &evts);

/* step 2. notify the crtc to handle the events */
for (i = 0; i < kms->n_crtcs; i++)
komeda_crtc_handle_event(&kms->crtcs[i], &evts);

v2:
- Move get IRQ number into this change.
- Enable irq before drm_dev_register.

Signed-off-by: James Qian Wang (Arm Technology China) <james.qian.wang@arm.com>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>

drm/komeda: Add komeda_assemble_pipelines

komeda_accemble_pipelines is for:

1. Verifing the component->supported_inputs according to the
   pipeline->avail_components.
2. Generating component->supported_outputs.

v2: Lower the debug message of komeda_component_dump to DRM_DEBUG.

Signed-off-by: James Qian Wang (Arm Technology China) <james.qian.wang@arm.com>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>

drm/komeda: Add D71 improc and timing_ctrlr

Add and initialize improc and timing_ctrlr according to D71 capablitites

v2: Rebase.

Signed-off-by: James Qian Wang (Arm Technology China) <james.qian.wang@arm.com>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>

drm/komeda: Add d71 compiz component

Implement d71_compiz_init and add compiz component to komeda-CORE

v2: Rebase.

Signed-off-by: James Qian Wang (Arm Technology China) <james.qian.wang@arm.com>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>

drm/komeda: Add d71 layer

1. Add detailed layer/layer_state definitions
2. Add d71_layer_init to report layer features and capabilities according
   to D71 layer block.
3. Add d71_layer_updat/disable

v2: Rebase.

Signed-off-by: James Qian Wang (Arm Technology China) <james.qian.wang@arm.com>
[removed d71_layer_dump() from this commit]
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>

Merge tag 'drm-misc-next-2019-03-28-1' of git://anongit.freedesktop.org/drm/drm-misc into drm-next

drm-misc-next for 5.2:

UAPI Changes:
- Remove unused DRM_DISPLAY_INFO_LEN (Ville)

Cross-subsystem Changes:
- None

Core Changes:
- Fix compilation when CONFIG_FBDEV not selected (Daniel)
- fbdev: Make skip_vt_switch default (Daniel)
- Merge fb_helper_fill_fix, fb_helper_fill_var into fb_helper_fill_info (Daniel)
- Remove unused fields in connector, display_info, and edid_quirks (Ville)

Driver Changes:
- virtio: package function args in virtio_gpu_object_params (Gerd)
- vkms: Fix potential NULL-dereference bug (Kangjie)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Sean Paul <sean@poorly.run>
Link: https://patchwork.freedesktop.org/patch/msgid/20190328183045.GA44823@art_vandelay

Merge tag 'du-next-20190328' of git://linuxtv.org/pinchartl/media into drm-next

R-Car DU miscellaneous changes for v5.2

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190328042035.GA4846@pendragon.ideasonboard.com

Merge tag 'drm-intel-next-2019-03-28' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

UAPI Changes:
- Make mmap code more asynchronous. Avoid full SET_DOMAIN on GTT mmap pagefault,
  and flushes pages on acquisition instead. Moves some of the work from mmap fault
  time to execbuf time to avoid lock contention during mmap access.

  Has neutral to positive impact on perf as the flushing moves to execbuf time
  in real world workloads on the current known userspaces due to recycling of BOs.

  If there exist an unknown non-recycling userspace, they should explicitly do the
  SET_DOMAIN and not rely on kernel doing implicit SET_DOMAIN because swapout/in
  might have happenedt.

- Restore the accidentally removed behaviour of returning object size on GEM_CREATE
  From 2011: ff72145badb8 ("drm: dumb scanout create/mmap for intel/radeon (v3)")

- Includes a some neutered patches to prepare to complete the earlier Mesa
  recovery feature uAPI. Looking to enable this in the next PR.

Driver Changes:

- Add Elkhartlake (Gen11) support code and PCI IDs
- Add missing Amberlake PCI ID 0x87CA (Ville)
- Fix to Bugzilla #109780: Pick the first mode from EDID as the fixed mode when there is no preferred mode (Ville)
- Fix GCC 4.8 build by using __is_constexpr() (Chris, Randy, Uma)
- Add "Broadcast RGB", "force_audio" and "max_bpc" properties to DP MST (Ville)
- Remove 8bpc limitation from DP MST (Ville)
- Fix changing between limited and full range RGB output in DP fastsets (Ville)
- Reject unsupported HDR formats (Maarten)
- Handle YUV subpixel support better (Maarten)

- Various plane watermarks fixes and cleaning of the code (Ville)
- Icelake port sync master select fix (Manasi)
- Icelake VEBOX disable bitmask fix (Jose)
- Close a race where userspace could see incompletely initialized GEM context (Chris)
- Avoid C3 on i945gm to keep vblank interrupts steady (Ville)
- Avoid recalculating PLL HW readout each time (Lucas)
- A ton of patches to modularize uncore code (Daniel)

- Instead of storing media fuse value, immediately derive engine masks (Daniele)
- Reduce struct_mutex usage (Chris)
- Iterate over child devices to initialize ddi_port_info (Jani)
- Fixes to return correct error values when bailing out of functions (Dan)
- Use bitmap_zalloc() (Andy)
- Reorder and clarify Gen3/4 code (Ville)
- Refactor out common code in display mode handling (Ville)
- GuC code fixes (Sujaritha, Michal)
- Selftest improvements (Chris)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190328151515.GA9606@jlahtine-desk.ger.corp.intel.com

Revert "drm/amdgpu: use HMM callback to replace mmu notifier"

This reverts commit 57731a07795ae80790c6ae7d8e7001cdbd6d14c0.

This depends on an HMM fix which is not upstream yet.

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

Revert "drm/amdkfd: avoid HMM change cause circular lock"

This reverts commit 8dd69e69f42397c9b17764a951c44480b340858e.

This depends on an HMM fix which is not upstream yet.

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

Revert "drm/amdgpu: replace get_user_pages with HMM mirror helpers"

This reverts commit 915d3eecfa23693bac9e54cdacf84fb4efdcc5c4.

This depends on an HMM fix which is not upstream yet.

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

Revert "drm/amdgpu: fix HMM config dependency issue"

This reverts commit 6b8f7e3dee7883084932bbdfce471a2960c6db5d.

This depends on an HMM fix which is not upstream yet.

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

Revert "drm/amdkfd: support concurrent userptr update for HMM"

This reverts commit 386a68e78620c793ed1ba9bc5789b90c18be262a.

This depends on an HMM fix which is not upstream yet.

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

Revert "drm/amdgpu: support userptr cross VMAs case with HMM"

This reverts commit 5aeaccca30023c00d982dfa7bfd25b384523460a.

This depends on an HMM fix which is not upstream yet.

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

Revert "drm/amdgpu: more descriptive message if HMM not enabled"

This reverts commit 194f87ddffe26bbbd124c549a9bf12a96a7f2919.

This depends on an HMM fix which is not upstream yet.

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

drm/i915: Update DRIVER_DATE to 20190328

Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

drm/i915: Update DRIVER_DATE to 20190328

Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

drm/i915: Update DRIVER_DATE to 20190328

Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

drm/virtio: rework resource creation workflow.

This patch moves the virtio_gpu_cmd_create_resource() call (which
notifies the host about the new resource created) into the
virtio_gpu_object_create() function.  That way we can call
virtio_gpu_cmd_create_resource() before ttm_bo_init(), so the host
already knows about the object when ttm initializes the object and calls
our driver callbacks.

Specifically the object is already created when the
virtio_gpu_ttm_tt_bind() callback invokes virtio_gpu_object_attach(),
so the extra virtio_gpu_object_attach() calls done after
virtio_gpu_object_create() are not needed any more.

The fence support for the create ioctl becomes a bit more tricky though.
The code moved into virtio_gpu_object_create() too.  We first submit the
(fenced) virtio_gpu_cmd_create_resource() command, then initialize the
ttm object, and finally attach just created object to the fence for the
command in case it didn't finish yet.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190318113332.10900-6-kraxel@redhat.com

drm/virtio: params struct for virtio_gpu_cmd_create_resource_3d()

Add 3d resource parameters to virtio_gpu_object_params struct.  With
that in place we can use it for virtio_gpu_cmd_resource_create_3d()
calls.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190318113332.10900-5-kraxel@redhat.com

drm/virtio: params struct for virtio_gpu_cmd_create_resource()

Add format, width and height fields to the virtio_gpu_object_params
struct.  With that in place we can use the parameter struct for
virtio_gpu_cmd_create_resource() calls too.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190318113332.10900-4-kraxel@redhat.com

drm/virtio: use struct to pass params to virtio_gpu_object_create()

Create virtio_gpu_object_params, use that to pass object parameters to
virtio_gpu_object_create.  This is just the first step, followup patches
will add more parameters to the struct.  The plan is to use the struct
for all object parameters.

Drop unused "kernel" parameter for virtio_gpu_alloc_object(), it is
unused and always false.

Also drop "pinned" parameter.  virtio-gpu doesn't shuffle around
objects, so effecively they all are pinned anyway.  Hardcode
TTM_PL_FLAG_NO_EVICT so ttm knows.  Doesn't change much for the moment
as virtio-gpu supports TTM_PL_FLAG_TT only so there is no opportunity to
move around objects.  That'll probably change in the future though.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190318113332.10900-3-kraxel@redhat.com

drm/virtio: move virtio_gpu_object_{attach, detach} calls.

Drop the dummy ttm backend implementation, add a real one for
TTM_PL_FLAG_TT objects.  The bin/unbind callbacks will call
virtio_gpu_object_{attach,detach}, to update the object state
on the host side, instead of invoking those calls from the
move_notify() callback.

With that in place the move and move_notify callbacks are not
needed any more, so drop them.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190318113332.10900-2-kraxel@redhat.com

drm/virtio: add virtio-gpu-features debugfs file.

This file prints which features the virtio-gpu device has.

Also add "virtio-gpu-" prefix to the existing fence file,
to make clear this is a driver-specific debugfs file.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190320083612.32067-1-kraxel@redhat.com

drm/virtio: remove set but not used variable 'vgdev'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/gpu/drm/virtio/virtgpu_ttm.c: In function 'virtio_gpu_init_mem_type':
drivers/gpu/drm/virtio/virtgpu_ttm.c:117:28: warning:
 variable 'vgdev' set but not used [-Wunused-but-set-variable]

drivers/gpu/drm/virtio/virtgpu_ttm.c: In function 'virtio_gpu_bo_swap_notify':
drivers/gpu/drm/virtio/virtgpu_ttm.c:300:28: warning:
 variable 'vgdev' set but not used [-Wunused-but-set-variable]

It is never used since introduction in dc5698e80cf7 ("Add virtio gpu driver.")

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Reviewed-by: Mukesh Ojha <mojha@codeaurora.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190325092631.152060-1-yuehaibing@huawei.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm: rcar-du: lvds: Set LVEN and LVRES bits together on D3

On the D3 SoC the LVDS PHY must be enabled in the same register write
that enables the LVDS output. Skip writing the LVEN bit independently
on that platform, it will be set by the write that sets LVRES.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo+renesas@jmondi.org>

drm: rcar-du: lvds: Adjust operating frequency for D3 and E3

The D3 and E3 SoCs have different pixel clock frequency limits for the
LVDS encoder than the other SoCs in the Gen3 family. Adjust the mode
fixup implementation accordingly.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo+renesas@jmondi.org>

drm: rcar-du: Link CRTCs to the DU device

The rcar_du_crtc functions have a heavy reliance on the rcar_du_group
structure, in many cases just to access the DU device context.

To better separate the groups out of the CRTC handling code, give the
rcar_du_crtc its own pointer to the device and remove the indirection
through the group pointers.

Signed-off-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>

drm: rcar-du: lvds: Fix post-DLL divider calculation

The PLL parameters are computed by looping over the range of acceptable
M, N and E values, and selecting the combination that produces the
output frequency closest to the target. The internal frequency
constraints are taken into account by restricting the tested values for
the PLL parameters, reducing the search space. The target frequency,
however, is only taken into account when computing the post-PLL divider,
which can result in a 0 value for the divider when the PLL output
frequency being tested is lower than half of the target frequency.
Subsequent loops will produce a better set of PLL parameters, but for
some of the iterations this can result in a division by 0.

Fix it by clamping the divider value. We could instead restrict the E
values being tested in the inner loop, but that would require additional
calculation that would likely be less efficient as the E parameter can
only take three different values.

Fixes: c25c01361199 ("drm: rcar-du: lvds: D3/E3 support")
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>

drm: Forward-declare struct drm_format_info in drm_framebuffer.h

drm_framebuffer.h makes use of a pointer to struct drm_format_info but
doesn't include drm_fourcc.h (neither directly nor indirectly).
Forward-declare the structure.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Sam Ravnborg <sam@ravnborg.org>

drm: rcar-du: Remove unused prototypes

The CRTC suspend and resume functions have been replaced, but the
prototypes were not removed.

Remove the redundant definitions.

Signed-off-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>

drm: rcar-du: crtc: Make local functions static

The rcar_du_crtc_mode_valid() and rcar_du_crtc_get_crc_sources()
functions are accessed only through a function pointer table.

Convert the function definitions to be static to the module.

Signed-off-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>

drm: Fix subtle spelling error in drm_crtc_state

The drm_crtc_state documentation contains a subtle misspelling of the
word subtle. Correct it.

Signed-off-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>

drm: rcar-du: Support panels connected directly to the DPAD outputs

The R-Car DU driver assumes that a bridge is always connected to the DU
output. This is valid for the LVDS and HDMI outputs, but the DPAD
outputs can be connected directly to a panel, in which case no bridge is
available.

To support this use case, detect whether the entities connected to the
DU DPAD outputs are encoders or panels based on the number of ports of
their DT node, and retrieve the corresponding type of DRM objects. For
panels, additionally create panel bridge instances.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Tested-by: Kevin Key <kevin.key@gentex.com>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo+renesas@jmondi.org>

drm/amd/powerplay: update current profile mode only when it's really applied

No need to update current profile mode if the new profile mode
does not take effect in fact.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Kent Russell <kent.russell@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix odm output gamma programming

Currently only top pipe gets output tf programmed. This change
makes all odm head pipes get output tf programmed.

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Nikola Cornij <Nikola.Cornij@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Populate macro_tile_size field for dml

Create a functions to return swizzle types for dml

Signed-off-by: Joshua Aberback <joshua.aberback@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: use dc_is_virtual instead of ENUM

Signed-off-by: Eric Bernstein <eric.bernstein@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Nevenko Stupar <Nevenko.Stupar@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.24

Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Pass SDP spliting in parameters

pass SDP splitting when setting stream attributes for future use

Signed-off-by: Nikola Cornij <nikola.cornij@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Create clock funcs

Create dccg_init and init_clocks for future use

Signed-off-by: Eryk Brol <eryk.brol@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Clean up old pplib interface functions

[Why]
set_display_requirement, dcn1_pplib_apply_display_requirements
are no longer used and should be removed.

Signed-off-by: Fatemeh Darbehani <fatemeh.darbehani@amd.com>
Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Pass init_data into DCN resource creation

[WHY]
The resource constructor currently needs num_virtual_links from
init_data but will need access to other items provided by DM.

[HOW]
Pass init_data into DCN create_resource_pool functions.

Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Acked-by: Hersen Wu <hersenxs.wu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Handle branch device with DFP count = 0 case.

[Why]
When you have a SST branch device the driver, Even no sink device connected,
it also send HPD with a valid EDID. Driver will config it to DP sink.
Therefore, there're two displays in display setting.

DPCD 0x05, DFP_PRESENT = 1 (branch device),
DFP_TYPE = 00 (Display Port)

[How]
Driver determine DPCD 0x05 DFP_PRESENT = 1(branch) as an active dongle
And check DFP count.

Signed-off-by: Hugo Hu <hugo.hu@amd.com>
Reviewed-by: Hugo Hu <Hugo.Hu@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: program default output gamma

program default output gamma if no user specific gamma
parameters passed.

Signed-off-by: hersen wu <hersenxs.wu@amd.com>
Reviewed-by: Sun peng Li <Sunpeng.Li@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add preferred pipe split logic

[why]
existing logic finds "first free pipe from 5 -> 0" to split
this will cause certain sequences to require DC to move
an MPCC from one tree to another, which is unsupported
this leads to blackscreen

to mitigate this problem, we will always try to acquire the
"preferred" pipe, and each pipe has a unique preferred pipe
this means we avoid most of the scenarios where
pipe splitting leads to moving MPCC from one tree
to another

Signed-off-by: Jun Lei <Jun.Lei@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: don't put the root PD into the relocated list

Instead of skipping the root PD while processing the relocated list just never
put it on the list in the first place.

This avoids walking the list all together when the root PD is the only entry
and so also avoids trying to submit a zero sized IB to the SDMA.

Signed-off-by: Christian König <christian.koenig@amd.com>
Tested-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: drop the ib from the VM update parameters

It is redundant with the job pointer.

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

drm/amdgpu: move VM table mapping into the backend as well

Clean that up further and also fix another case where the BO
wasn't kmapped for CPU based updates.

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

drm/amdgpu: XGMI pstate switch initial support

Driver vote low to high pstate switch whenever there is an outstanding
XGMI mapping request. Driver vote high to low pstate when all the
outstanding XGMI mapping is terminated.

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

drm/amdgpu: use the new VM backend for clears

And remove the existing code when it is unused.

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

drm/amdgpu: defer cmd/fence/fw buffers destroy on hw_init failure

As the cleanup jobs performed in pre_fini may still need these
buffers. NULL pointer dereference will be triggered without them.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add more debug friendly prompts

Large piece of codes share one error prompt. That
is not friendly for debugging.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: error out on mode1 reset failure

The error return value should be correctly reflected.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: trivial typo fix

"error" was not correctly spelled.

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