Merge branch 'drm-next-4.18' of git://people.freedesktop.org/~agd5f/linux into drm-next

Fixes for 4.18. Highlights:
- Fixes for gfxoff on Raven
- Remove an ATPX quirk now that the root cause is fixed
- Runtime PM fixes
- Vega20 register header update
- Wattman fixes
- Misc bug fixes

Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180614141428.2909-1-alexander.deucher@amd.com

drm/amd/powerplay: Set higher SCLK&MCLK frequency than dpm7 in OD (v2)

Fix the issue that SCLK&MCLK can't be set higher than dpm7 when
OD is enabled in SMU7.

v2: fix warning (Alex)

Signed-off-by: Kenneth Feng <kenneth.feng@amd.com>
Acked-by: Rex Zhu<rezhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: remove uncessary extra gfxoff control call

Gfxoff is already enabled in amdgpu_device_ip_set_powergating_state.
So, no need to enable it again in pp_late_init.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix parsing indirect register list v2

WARN_ON possible buffer overflow and avoid unnecessary dereference.

v2: change BUG_ON to WARN_ON

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/include: Update df 3.6 mask and shift definition

The register field hsas been changed in df 3.6, update to correct setting

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

drm/amd/pp: Fix OD feature enable failed on Vega10 workstation cards

As hw required, soc clock must large than mclk, So we set max soc
clock to OD Max Memory clk.
But on workstation, vbios do not support OD feature, the OD max memory
clock is equal to 0. In this case, driver can support underclocking.
and set od max memory clock to the value in highest memory dpm level.
So the od max memory clock should be less than highest soc clock.
and driver should not change the soc clock.

caused by commit ca57b9b0a156
("drm/amd/pp: Allow underclocking when od table is empty in vbios")

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

drm/amd/display: Fix stale buffer object (bo) use

Fixes stale buffer object (bo) usage for cursor plane

Cursor plane's bo operations are handled in DC code.
Currently, atomic_commit() does not handle bo operations
for cursor plane, as a result the bo assigned for cursor
plane in dm_plane_helper_prepare_fb() is not coherent
with the updates to the same made in dc code.This mismatch
leads to "bo" corruption and hence crashes during S3 entry.

This patch cleans up the code which was added as a hack
for 4.9 version only.

Reviewed-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Signed-off-by: Pratik Vishwakarma <Pratik.Vishwakarma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pp: initialize result to before or'ing in data

The current use of result is or'ing in values and checking for
a non-zero result, however, result is not initialized to zero
so it potentially contains garbage to start with. Fix this by
initializing it to the first return from the call to
vega10_program_didt_config_registers.

Detected by cppcheck:
"(error) Uninitialized variable: result"

Fixes: 9b7b8154cdb8 ("drm/amd/powerplay: added didt support for vega10")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Huang Rui <ray.huang@amd.com>
[Fix the subject as Colin's comment]
Signed-off-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

drm/amd/powerplay: fix wrong clock adjust sequence

The clocks should be adjusted after display configuration changed.
Otherwise, the socclk and memclk may be forced on an unnecessary higher
level.

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

drm/amdgpu: Grab/put runtime PM references in atomic_commit_tail()

So, unfortunately I recently made the discovery that in the upstream
kernel, the only reason that amdgpu is not currently suffering from
issues with runtime PM putting the GPU into suspend while it's driving
displays is due to the fact that on most prime systems, we have sound
devices associated with the GPU that hold their own runtime PM ref for
the GPU.

What this means however, is that in the event that there isn't any kind
of sound device active (which can easily be reproduced by building a
kernel with sound drivers disabled), the GPU will fall asleep even when
there's displays active. This appears to be in part due to the fact that
amdgpu has not actually ever relied on it's rpm_idle() function to be
the only thing keeping it running, and normally grabs it's own power
references whenever there are displays active (as can be seen with the
original pre-DC codepath in amdgpu_display_crtc_set_config() in
amdgpu_display.c). This means it's very likely that this bug was
introduced during the switch over the DC.

So to fix this, we start grabbing runtime PM references every time we
enable a previously disabled CRTC in atomic_commit_tail(). This appears
to be the correct solution, as it matches up with what i915 does in
i915/intel_runtime_pm.c.

The one sideaffect of this is that we ignore the variable that the
pre-DC code used to use for tracking when it needed runtime PM refs,
adev->have_disp_power_ref. This is mainly because there's no way for a
driver to tell whether or not all of it's CRTCs are enabled or disabled
when we've begun committing an atomic state, as there may be CRTC
commits happening in parallel that aren't contained within the atomic
state being committed. So, it's safer to just get/put a reference for
each CRTC being enabled or disabled in the new atomic state.

Signed-off-by: Lyude Paul <lyude@redhat.com>
Acked-by: Christian König <christian.koenig@amd.com>.
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

drm/amd/powerplay: fix missed hwmgr check warning before call gfx_off_control handler

Patch 9667849bbb8d: "drm/amd/powerplay: add control gfxoff enabling in late
init" from Mar 13, 2018, leads to the following static checker warning:

drivers/gpu/drm/amd/amdgpu/../powerplay/amd_powerplay.c:194
pp_late_init()
error: we previously assumed 'hwmgr' could be null (see line 185)

drivers/gpu/drm/amd/amdgpu/../powerplay/amd_powerplay.c

This patch fixes the warning to add hwmgr checking.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix CG enabling hang with gfxoff enabled

After defer the execution of clockgating enabling, at that time, gfx already
enter into "off" state. Howerver, clockgating enabling will use MMIO to access
the gfx registers, then get the gfx hung.

So here we should move the gfx powergating and gfxoff enabling behavior at the
end of initialization behind clockgating.

Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Cc: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix clear_all and replace handling in the VM (v2)

v2: assign bo_va as well

We need to put the lose ends on the invalid list because it is possible
that we need to split up huge pages for them.

Cc: stable@vger.kernel.org
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Junwei Zhang <Jerry.Zhang@amd.com> (v2)
Reviewed-by: David Zhou <david1.zhou@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add checking for sos version

The sos ucode version will be changed to align with the value of
mmMP0_SMN_C2PMSG_58. Then we add a checking for this. Meanwhile, we have to be
compatibility backwards. So it adds serveral recent legacy versions as the white
list for the version checking.

Signed-off-by: Huang Rui <ray.huang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix the missed vcn fw version report

It missed vcn.fw_version setting when init vcn microcode, and it will be used to
report vcn ucode version via amdgpu_firmware_info sysfs interface.

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

Merge tag 'drm-intel-next-fixes-2018-06-08-2' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

First batch of i915 fixes for v4.18:
- gvt fixes that missed v4.17, potentially need to be backported
- eDP resolution regression revert
- remove broken nv12 special casing
- remove stale asserts from find active requests

Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/87y3fp4h15.fsf@intel.com

Merge tag 'drm-misc-next-fixes-2018-05-31' of git://anongit.freedesktop.org/drm/drm-misc into drm-next

drm-misc-next-fixes for v4.18:

Driver changes:
- Plug small memory leak in vc4. (anholt)
- Depend on MMU in v3d. (arnd)

Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/12faab25-e809-e73c-409c-5e9c08aa351c@linux.intel.com

Merge tag 'drm-msm-next-2018-06-04' of git://people.freedesktop.org/~robclark/linux into drm-next

A bunch of msm fixes and some atomic work.

Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/CAF6AEGvXe157nd70=GPjre=HjiAH91tGM50+XtM59b2MEChvXQ@mail.gmail.com

Merge tag 'gvt-fixes-2018-04-19' of https://github.com/intel/gvt-linux into drm-intel-next-fixes

gvt-fixes-2018-04-19

- cmd parser error path mem leak fix (Colin)
- fix dp aux header validation (Changbin)
- sanity check on pfn after vfio pin page (Changbin)
- fix msi eventfd put (Xiong)

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180419073948.4mojv7xaxxvfuyud@zhen-hp.sh.intel.com

drm/i915/icl: Don't update enabled dbuf slices struct until updated in hw

Do not update number of enabled dbuf slices in dev_priv struct until we
actually enable/disable dbuf slice in hw. This is leading to never
updating dbuf slices and resulting in DBuf slice mismatch warning.

Fixes: aa9664ffe863 ("drm/i915/icl: Enable 2nd DBuf slice only when needed")
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Mahesh Kumar <mahesh1.kumar@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180517132626.5885-1-mahesh1.kumar@intel.com
(cherry picked from commit 6ceb7277173597eeed8d635e08db51e35235ab21)
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

drm/i915/icl: fix icl_unmap/map_plls_to_ports

All connectors may not have best_encoder attached, so don't dereference
encoder pointer for each connector.

Fixes: c27e917e2bda ("drm/i915/icl: add basic support for the ICL clocks")
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Mahesh Kumar <mahesh1.kumar@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180525155238.7054-1-lucas.demarchi@intel.com
(cherry picked from commit c46ef57d2008e5195f086a538550488b49644db7)
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

drm/i915: Remove bogus NV12 PLANE_COLOR_CTL setup

We already handle the color encoding mode properly. Remove the broken
NV12 special case.

Cc: Vidya Srinivas <vidya.srinivas@intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Fixes: 8ed30ab6aced ("drm/i915: Enable YUV to RGB for Gen10 in Plane Ctrl Reg")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180521185613.5097-1-ville.syrjala@linux.intel.com
Reviewed-By: Vidya Srinivas <vidya.srinivas@intel.com>
(cherry picked from commit 012d79e6a33f095c293fe2a02b2b3b26d8c6402c)
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

drm/msm: Fix NULL deref on bind/probe deferral

This patch avoids dereferencing msm_host->dev when it is NULL.

If we find ourselves tearing down dsi before calling
(mdp4|mdp5|dpu)_kms_init(), we'll end up in a state where the dev
pointer is NULL and trying to extract priv from it will fail.

This was introduced in a seemingly innocuous commit to ensure the
arguments to msm_gem_put_iova() are correct (even though that
function has been a stub for ~5 years). Correctness FTW! \o/

Fixes: b01884a286b0 drm/msm: use correct aspace pointer in msm_gem_put_iova()
Cc: Daniel Mack <daniel@zonque.org>
Cc: Rob Clark <robdclark@gmail.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm: Switch to atomic_helper_commit()

Now that all of the msm-specific goo is tucked safely away we can switch
over to using the atomic helper commit directly. \o/

Changes in v2:
- None
Changes in v3:
- Rebased on Archit's private_obj set
Changes in v4:
- None

Cc: Abhinav Kumar <abhinavk@codeaurora.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm: Remove msm_commit/worker, use atomic helper commit

Moving further towards switching fully to the the atomic helpers, this
patch removes the hand-rolled worker nonblock commit code and uses the
atomic helpers commit_work model.

Changes in v2:
- Remove commit_destroy()
- Shuffle order of commit_tail calls to further serialize commits
- Use stall in swap_state to avoid abandoned events on disable
Changes in v3:
- Rebased on Archit's private_obj set
Changes in v4:
- None

Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm: Issue queued events when disabling crtc

Ensure that any queued events are issued when disabling the crtc. This
avoids timeouts when we come back and wait for dependencies (like the
previous frame's flip_done).

Changes in v2:
- None
Changes in v3:
- Rebased on Archit's private_obj set
Changes in v4:
- None

Reviewed-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm: Move implicit sync handling to prepare_fb

In preparation for moving to atomic helpers, move the implicit sync
fence handling out of atomic commit and into the plane->prepare_fb()
hook. While we're at it, de-duplicate the mdp*_prepare_fb functions.

Changes in v4:
- Added

Reported-by: Rob Clark <robdclark@gmail.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm: Refactor complete_commit() to look more the helpers

Factor out the commit_tail() portions of complete_commit() into a
separate function to facilitate moving to the atomic helpers in future
patches.

Changes in v2:
- None
Changes in v3:
- Rebased on Archit's private_obj set
Changes in v4:
- None

Cc: Jeykumar Sankaran <jsanka@codeaurora.org>
Reviewed-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm: Don't subclass drm_atomic_state anymore

With the addition of "private_objs" in drm_atomic_state, we no longer
need to subclass drm_atomic_state to store state of share resources
that don't perfectly fit within planes/crtc/connector state information.
We can now save this state within drm_atomic_state itself using
the private objects.

Remove the infrastructure that allowed subclassing of drm_atomic_state
in the driver.

Changes in v3:
- Added to the msm atomic helper patch set
Changes in v4:
- None

Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>

drm/msm/mdp5: Use the new private_obj state

This replaces the usage of the subclassed atomic state (mdp5_state)
with a private_obj state embedded within drm_atomic_state. The latter
method is the preferred approach, since it's simpler to implement
and less prone to errors.

The new API replaces the older and equivalent mdp5_state usage in the
following pattern:
- References to "mdp5_kms->state" (i.e, the old/existing state) is
  replaced with mdp5_get_existing_global_state(). In the atomic_check
  path, this should be called with the glob_state_lock drm_modeset_lock
  alredy taken.
- References to "mdp5_get_state()" are replaced with
  mdp5_get_global_state(). This acquires glob_state_lock and uses
  drm_atomic_get_private_obj_state() to create a new duplicated state.

Changes in v3:
- Acquire glob_state_lock in mdp5_smp.c
- Added to the msm atomic helper patch set
Changes in v4:
- None

Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>

drm/msm/mdp5: Add global state as a private atomic object

Global shared resources (hwpipes, hwmixers and SMP) for MDP5 are
implemented as a part of atomic state by subclassing drm_atomic_state.

The preferred approach is to use the drm_private_obj infrastructure
available in the atomic core.

mdp5_global_state is introduced as a drm atomic private object. The two
funcs mdp5_get_global_state() and mdp5_get_existing_global_state() are
the two variants that will be used to access mdp5_global_state.

This will replace the existing mdp5_state struct (which subclasses
drm_atomic_state) and the funcs around it. These will be removed later
once we mdp5_global_state is put to use everywhere.

Changes in v3:
- Added glob_state_lock instead of pushing it into the core
- Added to the msm atomic helper patch set
Changes in v4:
- None

Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>

drm/msm: use correct aspace pointer in msm_gem_put_iova()

Even though msm_gem_put_iova() is currently a NOP function, the caller
should pass in the address space pointer it used to obtain the object.

Other call sites were changed in 8bdcd949bbe7e ("drm/msm: pass
address-space to _get_iova() and friends"), but this one seems to have
been forgotten.

Signed-off-by: Daniel Mack <daniel@zonque.org>
Cc: Rob Clark <robdclark@gmail.com>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm: remove unbalanced mutex unlock

This regression stems from 0e08270a1f01 ("drm/msm: Separate locking of
buffer resources from struct_mutex").

Signed-off-by: Daniel Mack <daniel@zonque.org>
Cc: Sushmita Susheelendra <ssusheel@codeaurora.org>
Cc: Rob Clark <rclark@redhat.com>
Fixes: 0e08270a1f01 ("drm/msm: Separate locking of buffer resources from struct_mutex")
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm: don't deref error pointer in the msm_fbdev_create error path

Currently the error pointer returned by msm_alloc_stolen_fb gets passed
to drm_framebuffer_remove. The latter handles only NULL pointers, thus
a nasty crash will occur.

Drop the unnecessary fail label and the associated checks - both err and
fb will be set at this stage.

Cc: Rob Clark <robdclark@gmail.com>
Cc: linux-arm-msm@vger.kernel.org
Cc: dri-devel@lists.freedesktop.org
Cc: freedreno@lists.freedesktop.org
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/dsi: use correct enum in dsi_get_cmd_fmt

The function dsi_get_cmd_fmt returns enum dsi_cmd_dst_format,
use the correct enum value also for MIPI_DSI_FMT_RGB666/_PACKED.

This has been discovered using clang:
  drivers/gpu/drm/msm/dsi/dsi_host.c:743:35: warning: implicit conversion
        from enumeration type 'enum dsi_vid_dst_format' to different
        enumeration type 'enum dsi_cmd_dst_format' [-Wenum-conversion]
          case MIPI_DSI_FMT_RGB666:       return VID_DST_FORMAT_RGB666;
                                          ~~~~~~ ^~~~~~~~~~~~~~~~~~~~~

Signed-off-by: Stefan Agner <stefan@agner.ch>
Reviewed-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm: Fix possible null dereference on failure of get_pages()

Commit 62e3a3e342af changed get_pages() to initialise
msm_gem_object::pages before trying to initialise msm_gem_object::sgt,
so that put_pages() would properly clean up pages in the failure
case.

However, this means that put_pages() now needs to check that
msm_gem_object::sgt is not null before trying to clean it up, and
this check was only applied to part of the cleanup code.  Move
it all into the conditional block.  (Strictly speaking we don't
need to make the kfree() conditional, but since we can't avoid
checking for null ourselves we may as well do so.)

Fixes: 62e3a3e342af ("drm/msm: fix leak in failed get_pages")
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Reviewed-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm: Add modifier to mdp_get_format arguments

This change plumbs the new fb modifier through the various mdp/disp
get_format hooks.

Signed-off-by: Jeykumar Sankaran <jsanka@codeaurora.org>
[seanpaul pimped out commit message a bit]
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm: Mark the crtc->state->event consumed

Don't leave the event != NULL once it's consumed, this is used a signal
to the atomic helpers that the event will be handled by the driver.

Changes in v2:
- None
Changes in v3:
- Rebased on Archit's private_obj set
Changes in v4:
- None

Cc: Jeykumar Sankaran <jsanka@codeaurora.org>
Reviewed-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/dsi: implement auto PHY timing calculator for 10nm PHY

Currently the DSI PHY timings are hard-coded for a specific panel
for the 10nm PHY.

Replace this with the auto PHY timing calculator which can calculate
the PHY timings for any panel.

Changes in v4:
- None

Changes in v3:
- None

Changes in v2:
- None

Reviewed-by: Sean Paul <seanpaul@chromium.org>
Reviewed-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Abhinav Kumar <abhinavk@codeaurora.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/dsi: check video mode engine status before waiting

Make sure the video mode engine is on before waiting
for the video done interrupt.

Changes in v4:
- Move setting enabled to false earlier

Changes in v3:
- Move the return value check to another
  patch

Changes in v2:
- Replace pr_err with dev_err
- Changed error message

Signed-off-by: Abhinav Kumar <abhinavk@codeaurora.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/dsi: check return value for video done waits

Check for the return value of wait for video
done waits and print appropriate error message.

Signed-off-by: Abhinav Kumar <abhinavk@codeaurora.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

Revert "drm/amdgpu: Add an ATPX quirk for hybrid laptop"

This reverts commit 13b40935cf64f59b93cf1c716a2033488e5a228c.

This was a workaround for a bug in the HDA driver that prevented
the HDA audio chip from going into runtime pm which prevented
the GPU from going into runtime pm.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=106597
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/df: fix potential array out-of-bounds read

The comparison with the number of elements in array df_v3_7_channel_number
is off-by-one and can produce an array out-of-bounds read if
fb_channel_number is equal to the number of elements of the array. Fix
this by changing the comparison to >= instead of >.

Detected by CoverityScan, CID#1469489 ("Out-of-bounds read")

Fixes: 13b581502d51 ("drm/amdgpu/df: implement df v3_6 callback functions (v2)")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Fix NULL pointer when load kfd driver with PP block is disabled

When PP block is disabled, return a fix value(100M) for mclk and sclk on
bare-metal mode. This will cover the emulation mode as well.

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

drm/gfx9: Update gc goldensetting for vega20.

Update mmCB_DCC_CONFIG register goldensetting.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pp: Allow underclocking when od table is empty in vbios

if max od engine clock limit and memory clock limit are not set
in vbios. driver will allow underclocking instand of disable od feature
completely.

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

drm/i915: Remove stale asserts from i915_gem_find_active_request()

Since we use i915_gem_find_active_request() from inside
intel_engine_dump() and may call that at any time, we do not guarantee
that the engine is paused nor that the signal kthreads and irq handler
are suspended, so we cannot assert that the breadcrumb doesn't advance
and that the irq hasn't happened on another CPU signaling the request we
believe to be idle.

The second assert removed (that request->engine == engine) remains
valid, but is now more rigorously checked during retirement.

Fixes: f636edb214a5 ("drm/i915: Make i915_engine_info pretty printer to standalone")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180529132922.6831-1-chris@chris-wilson.co.uk
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
(cherry picked from commit cc7cc5343584d90e74b7c929ff2c9a2ec8b49cfe)
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

Revert "drm/i915/edp: Allow alternate fixed mode for eDP if available."

This reverts commit dc911f5bd8aacfcf8aabd5c26c88e04c837a938e.

Per the report, no matter what display mode you select with xrandr, the
i915 driver will always select the alternate fixed mode. For the
reporter this means that the display will always run at 40Hz which is
quite annoying. This may be due to the mode comparison.

But there are some other potential issues. The choice of alt_fixed_mode
seems dubious. It's the first non-preferred mode, but there are no
guarantees that the only difference would be refresh rate. Similarly,
there may be more than one preferred mode in the probed modes list, and
the commit changes the preferred mode selection to choose the last one
on the list instead of the first.

(Note that the probed modes list is the raw, unfiltered, unsorted list
of modes from drm_add_edid_modes(), not the pretty result after a
drm_helper_probe_single_connector_modes() call.)

Finally, we already have eerily similar code in place to find the
downclock mode for DRRS that seems like could be reused here.

Back to the drawing board.

Note: This is a hand-crafted revert due to conflicts. If it fails to
backport, please just try reverting the original commit directly.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105469
Reported-by: Rune Petersen <rune@megahurts.dk>
Reported-by: Mark Spencer <n7u4722r35@ynzlx.anonbox.net>
Fixes: dc911f5bd8aa ("drm/i915/edp: Allow alternate fixed mode for eDP if available.")
Cc: Clint Taylor <clinton.a.taylor@intel.com>
Cc: David Weinehall <david.weinehall@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Jim Bride <jim.bride@linux.intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: intel-gfx@lists.freedesktop.org
Cc: <stable@vger.kernel.org> # v4.14+
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180516080110.22770-1-jani.nikula@intel.com
(cherry picked from commit d93fa1b47b8fcd149b5091f18385304f402a8e15)
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

drm/i915: Nul-terminate legacy debug string

Make sure that when we don't have any scheduler attributes for the
request, the string is terminated.

Fixes: 247870ac8ea7 ("drm/i915: Build request info on stack before printk")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180517152824.11619-1-chris@chris-wilson.co.uk
(cherry picked from commit 96d4f03c20d04c80026b1ec3643c090cf4f0eb20)
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

drm/v3d: add CONFIG_MMU dependency

Without CONFIG_MMU, we get a link error:

drivers/gpu/drm/v3d/v3d_bo.o: In function `v3d_gem_fault':
v3d_bo.c:(.text+0x3ca): undefined reference to `vm_insert_mixed'

The other drivers with this problem already depend on CONFIG_MMU,
so let's do the same thing here.

Fixes: 57692c94dcbe ("drm/v3d: Introduce a new DRM driver for Broadcom V3D V3.x+")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Eric Anholt <eric@anholt.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20180525155030.3667352-5-arnd@arndb.de

drm/vc4: Fix leak of the file_priv that stored the perfmon.

Signed-off-by: Eric Anholt <eric@anholt.net>
Fixes: 65101d8c9108 ("drm/vc4: Expose performance counters to userspace")
Link: https://patchwork.freedesktop.org/patch/msgid/20180409205813.7077-1-eric@anholt.net
Reviewed-by: Boris Brezillon <boris.brezillon@bootlin.com>

drm/amdgpu/display: check if ppfuncs exists before using it

Fixes a crash on asics without powerplay yet (e.g., vega20).

Reviewed-by: Rex Zhu<rezhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pp: Connect display_clock_voltage_request to a function pointer

Get rid of an empty dublicate of smu10_display_clock_voltage_request

Add display_clock_voltage_request to smu10 functions struct so it
can be called from outside the class and connect the pointer to
the function.

That way Display driver can finally apply clock voltage requests
when needed.

Signed-off-by: Mikita Lipski <mikita.lipski@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: typo fix for vega20 cg flags

The AMD_CG_SUPPORT_HDP_LS was wrongly written as
AMD_CG_SUPPORT_BIF_LS.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: bug fixs for getsmuclockinfo

The .syspll_id and .dfsdid are not initialzed correctly. And
le32_to_cpu transfer is needed on the output.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: correct SMU11 SYSPLL0 clock id values

The SMU11 SYSPLL0 clock ids were assigned wrong values.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: gds bo must not be per-vm-bo

In per-vm-bo case, there could be no bo list.
But gds bo created from user space  must be passed to bo list.
So adding a check to prevent to creat gds bo as per-vm-bo.

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

drm/amdgpu: To get gds, gws and oa from adev->gds (v2)

As now enabled per vm bo feature, the user mode driver won't supply the
bo_list generally, for this case, the gdb_base, gds_size, gws_base, gws_size and
oa_base, oa_size won't be set.

v2: fix warning (Chunming)

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 tag 'drm-amdkfd-next-2018-05-28' of git://people.freedesktop.org/~gabbayo/linux into drm-next

- Build amdkfd's related files inside amdgpu only if amdkfd is built
- Fix compile warning
- Print info message in case ASIC is not supported by amdkfd

Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180528110435.GA17960@odedg-x270

Merge tag 'exynos-drm-next-for-v4.18-v2' of git://git./linux/kernel/git/daeinki/drm-exynos into drm-next

Add more HW overlays support
- It enables hardware overlay number 4 and 5. For this,
  this patch series adds required clocks.

Several fixups
- Fix default value of zpos according to real hardware overlay number.
- Fix error value of exynos_Drm_crtc_get_by_type function correctly.
- Fix static checker warning of scaler_task_done function.
- Fix signedness bug in fimc_setup_clocks function.

One cleanup
- Disable framedone interrupt of DSI device which is not required.

Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1527229919-25665-1-git-send-email-inki.dae@samsung.com

drm/amd/display: Fix BUG_ON during CRTC atomic check update

For cases where the CRTC is inactive (DPMS off), where a modeset is not
required, yet the CRTC is still in the atomic state, we should not
attempt to update anything on it.

Previously, we were relying on the modereset_required() helper to check
the above condition. However, the function returns false immediately if
a modeset is not required, ignoring the CRTC's enable/active state
flags. The correct way to filter is by looking at these flags instead.

Fixes: e277adc5a06c "drm/amd/display: Hookup color management functions"
Bugzilla: https://bugs.freedesktop.org/106194

Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Tested-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Read DP_SINK_COUNT_ESI range on HPD for DP 1.4

DP 1.4 compliance now requires that registers at DP_SINK_COUNT_ESI range
(0x2002-0x2003, 0x200c-0x200f) are read instead of DP_SINK_COUNT range
(0x200-0x2005.

Signed-off-by: Nikola Cornij <nikola.cornij@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Read DPCD link caps up to and including DP_ADAPTER_CAP

DP 1.4 compliance requires 16 bytes to be read when reading link caps,
i.e. it requires DP_ADAPTER_CAP to be included. Included it for all DP
versions because reading more than required won't fail.

Signed-off-by: Nikola Cornij <nikola.cornij@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Do not program interrupt status on disabled crtc

Prevent interrupt programming of a crtc on which the stream is disabled and
it doesn't have an OTG to reference.

Signed-off-by: Mikita Lipski <mikita.lipski@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix dscl_manual_ratio_init

This change will fix wb and display scaling when ratios of
4 or more are involved

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Destroy connector state on reset

When a DRM mode reset is called on resume, the connector state's
destructor is not called. This leaves a dangling reference on the CRTC
commit object, which was obtained by the connector state during commit
setup.

Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Set TMZ and DCC for secondary surface

Add register programming to support TMZ and DCC on
secondary surfaces.

Signed-off-by: Eric Bernstein <eric.bernstein@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: DP component depth 16 bpc

Add register programming to support 16bpc component
depth for DP.

Signed-off-by: Eric Bernstein <eric.bernstein@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: check if audio clk enable is applicable

Fixing warning on dce10 with HDMI display.

Signed-off-by: Roman Li <Roman.Li@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add register offset != 0 check.

Signed-off-by: Charlene Liu <charlene.liu@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Release fake sink

If connector doesn't have a sink, fake sink is created, but
never released as it assumed that its destroyed with the
stream it is used for. But now sink is released before the
stream maintaing refcount consistency.

This way we also avoid assigning anything to connector keeping
all the operation local.

Signed-off-by: Mikita Lipski <mikita.lipski@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: replace msleep with udelay in fbc path

FBC enabling and disabling path has msleep which leads to
BUG hit when called in atomic context, hence this patch
replaces msleeps with udelays appropriately.

Signed-off-by: Shirish S <shirish.s@amd.com>
Signed-off-by: Roman Li <Roman.Li@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Do not limit color depth to 8bpc

Delete if statement that would force any display's color depth higher
than 8 bpc to 8

Signed-off-by: Mikita Lipski <mikita.lipski@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove unnecessary scheduler entity for VCN

It should be stateless, and no need for scheduler to take care specially.

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

drm/amdgpu: Add helper function to get buffer domain

Move logic of getting supported domain to a helper
function

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

drm/amdgpu: Use GTT for dumb buffer if sg display enabled (v2)

When vram size <= THRESHOLD(256M) lets use GTT for dumb buffer
allocation. As SG will be enabled with vram size <= 256M
scan out will not be an issue.

v2: Use amdgpu_display_supported_domains to get supported domain.

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

drm/amd/pp: Add cases for getting phys and disp clks for SMU10

Add case options to retrieve either physical or display clocks with
voltage from SMU controller that are needed by display driver.

Signed-off-by: Mikita Lipski <mikita.lipski@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix wrong latency assignment for VEGA clock levels

Also drop wrong 10kHz comment

Fixes: drm/amd/display: Implement dm_pp_get_clock_levels_by_type_with_latency
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/scheduler: fix a corner case in dependency optimization

When checking for a dependency fence for belonging to the same entity
compare it with scheduled as well finished fence. Earlier we were only
comparing it with the scheduled fence.

Signed-off-by: Nayan Deshmukh <nayan26deshmukh@gmail.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: vcn_v1_0_is_idle() can be static

Fixes: 9b4c412a654c ("drm/amdgpu: Add static CG control for VCN on RV")
Signed-off-by: kbuild test robot <fengguang.wu@intel.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Implement dm_pp_get_clock_levels_by_type_with_latency

This is required so we use the correct minimum clocks for Vega. Without
this pplib will never be able to enter the lowest clock states.

Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix 32-bit build warning

Casting a pointer to a 64-bit type causes a warning on 32-bit targets:

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c:473:24: error: cast from pointer to integer of different size [-Werror=pointer-to-int-cast]
          lower_32_bits((uint64_t)wptr));
                        ^
drivers/gpu/drm/amd/amdgpu/amdgpu.h:1701:53: note: in definition of macro 'WREG32'
 #define WREG32(reg, v) amdgpu_mm_wreg(adev, (reg), (v), 0)
                                                     ^
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c:473:10: note: in expansion of macro 'lower_32_bits'
          lower_32_bits((uint64_t)wptr));
          ^~~~~~~~~~~~~

The correct method is to cast to 'uintptr_t'.

Fixes: d5a114a6c5f7 ("drm/amdgpu: Add GFXv9 kfd2kgd interface functions")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>

drm/exynos: fimc: signedness bug in fimc_setup_clocks()

"id" needs to be signed for the error handling to work.

Fixes: 7a2d5c77c558 ("drm/exynos: fimc: Convert driver to IPP v2 core API")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>

drm/exynos: scaler: fix static checker warning

drivers/gpu/drm/exynos/exynos_drm_scaler.c:402 scaler_task_done()
warn: signedness bug returning '(-22)'

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>

Merge branch 'drm-next-4.18' of git://people.freedesktop.org/~agd5f/linux into drm-next

Last feature request for 4.18.  Mostly vega20 support.
- Vega20 support
- clock and powergating for VCN
- misc bug fixes

Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180524152427.32713-1-alexander.deucher@amd.com

drm/amdgpu: Use dev_info() to report amdkfd is not supported for this ASIC

This is an important message, so it should be visible to users without
having to enable extra debugging.

Signed-off-by: Tom Stellard <tstellar@redhat.com>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>

drm/amd/display: Remove use of division operator for long longs

In fixed31_32.h, in dc_fixpt_shl,'/' was used for division of one long
long int by another long long int.  As there is no inbuilt long long
int division function in c, gcc inserted its own.  However, gcc does not
link the library that contains this function.  To avoid this, use
bitwise operators instead of /

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Update GFX info structure to match what vega20 used

Update to the latest version from the vbios team.

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

drm/amdgpu/pp: remove duplicate assignment

is_dpm_running callback was assigned to the same value
twice.  Drop the duplicate.

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/sched: add rcu_barrier after entity fini

To free the fence from the amdgpu_fence_slab, need twice call_rcu, to avoid
the amdgpu_fence_slab_fini call kmem_cache_destroy(amdgpu_fence_slab) before
kmem_cache_free(amdgpu_fence_slab, fence), add rcu_barrier after drm_sched_entity_fini.

The kmem_cache_free(amdgpu_fence_slab, fence)'s call trace as below:
1.drm_sched_entity_fini ->
drm_sched_entity_cleanup ->
dma_fence_put(entity->last_scheduled) ->
drm_sched_fence_release_finished ->
drm_sched_fence_release_scheduled ->
call_rcu(&fence->finished.rcu, drm_sched_fence_free)

2.drm_sched_fence_free ->
dma_fence_put(fence->parent) ->
amdgpu_fence_release ->
call_rcu(&f->rcu, amdgpu_fence_free) ->
kmem_cache_free(amdgpu_fence_slab, fence);

v2:put the barrier before the kmem_cache_destroy
v3:put the dma_fence_put(fence->parent) before call_rcu in
drm_sched_fence_release_scheduled

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>

drm/amdgpu: move VM BOs on LRU again

Move all BOs belonging to a VM on the LRU with every submission.

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

drm/amdgpu: consistenly use VM moved flag

Instead of sometimes checking if the vm_status is empty use the moved
flag and also reset it when the BO leaves the state machine.

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

drm/amdgpu: kmap PDs/PTs in amdgpu_vm_update_directories

In theory it is possible that PDs/PTs can move without eviction.

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

drm/amdgpu: further optimize amdgpu_vm_handle_moved

Splice the moved list to a local one to avoid taking the lock over and
over again.

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

drm/amdgpu: cleanup amdgpu_vm_validate_pt_bos v2

Use list_for_each_entry_safe here.

v2: Drop the optimization, it doesn't work as expected.

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

drm/amdgpu: rework VM state machine lock handling v2

Only the moved state needs a separate spin lock protection. All other
states are protected by reserving the VM anyway.

v2: fix some more incorrect cases

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

drm/amdgpu: Add runtime VCN PG support

Enable support for dynamically powering up/down VCN on demand.

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

drm/amdgpu: Enable VCN static PG by default on RV

Enable static VCN powergating by default on Raven.

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

drm/amdgpu: Add VCN static PG support on RV

Implement static powergating suport on VCN.

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

drm/amdgpu: Enable VCN CG by default on RV

Enable VCN clockgating by default on Raven.

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