platform/kernel/linux-starfive.git
7 years agoMerge tag 'drm/tegra/for-4.15-rc1' of git://anongit.freedesktop.org/tegra/linux into...
Dave Airlie [Tue, 24 Oct 2017 06:50:45 +0000 (16:50 +1000)]
Merge tag 'drm/tegra/for-4.15-rc1' of git://anongit.freedesktop.org/tegra/linux into drm-next

drm/tegra: Changes for v4.15-rc1

This contains a bit of cleanup and some minor fixes for the host1x and
Tegra DRM drivers. There's also some more preparatory work for Tegra186
support which I'm not quite ready to send upstream because the GPIO
driver needed for HDMI support has been stuck for months, and we can't
do much without it. Hopefully that driver will land in v4.15, which
would mean we could go ahead with Tegra186 display support in v4.16.

* tag 'drm/tegra/for-4.15-rc1' of git://anongit.freedesktop.org/tegra/linux: (21 commits)
  drm/tegra: hdmi: Add cec-notifier support
  drm/tegra: dc: Perform a complete reset sequence
  drm/tegra: dc: Make sure to set the module clock rate
  drm/tegra: dc: Simplify atomic plane helper functions
  drm/tegra: dc: Move some declarations to dc.h
  drm/tegra: vic: Use of_device_get_match_data()
  drm/tegra: sor: Use of_device_get_match_data()
  drm/tegra: hdmi: Use of_device_get_match_data()
  drm/tegra: dc: Use of_device_get_match_data()
  drm/tegra: Use u64_to_user_ptr helper
  gpu: host1x: Fix incorrect comment for channel_request
  gpu: host1x: Disassemble more instructions
  gpu: host1x: Improve debug disassembly formatting
  gpu: host1x: Enable gather filter
  gpu: host1x: Enable Tegra186 syncpoint protection
  gpu: host1x: Call of_dma_configure() after setting bus
  drm/tegra: Add Tegra186 support for VIC
  gpu: host1x: Add Tegra186 support
  dt-bindings: host1x: Add Tegra186 information
  gpu: host1x: syncpt: Request syncpoints per client
  ...

7 years agodrm: Add CRTC_GET_SEQUENCE and CRTC_QUEUE_SEQUENCE ioctls [v3]
Keith Packard [Fri, 30 Jun 2017 05:49:31 +0000 (22:49 -0700)]
drm: Add CRTC_GET_SEQUENCE and CRTC_QUEUE_SEQUENCE ioctls [v3]

These provide crtc-id based functions instead of pipe-number, while
also offering higher resolution time (ns) and wider frame count (64)
as required by the Vulkan API.

v2:

 * Check for DRIVER_MODESET in new crtc-based vblank ioctls

Failing to check this will oops the driver.

 * Ensure vblank interupt is running in crtc_get_sequence ioctl

The sequence and timing values are not correct while the
interrupt is off, so make sure it's running before asking for
them.

 * Short-circuit get_sequence if the counter is enabled and accurate

Steal the idea from the code in wait_vblank to avoid the
expense of drm_vblank_get/put

 * Return active state of crtc in crtc_get_sequence ioctl

Might be useful for applications that aren't in charge of
modesetting?

 * Use drm_crtc_vblank_get/put in new crtc-based vblank sequence ioctls

Daniel Vetter prefers these over the old drm_vblank_put/get
APIs.

 * Return s64 ns instead of u64 in new sequence event

Suggested-by: Daniel Vetter <daniel@ffwll.ch>
Suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
v3:

 * Removed FIRST_PIXEL_OUT_FLAG
 * Document that the timestamp in the query and event are
   that of the first pixel leaving the display engine for
   the display (using the same wording as the Vulkan spec).

Suggested-by: Michel Dänzer <michel@daenzer.net>
Acked-by: Dave Airlie <airlied@redhat.com>
[airlied: left->leaves (Michel)]

Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
7 years agodrm: Reorganize drm_pending_event to support future event types [v2]
Keith Packard [Wed, 5 Jul 2017 21:34:23 +0000 (14:34 -0700)]
drm: Reorganize drm_pending_event to support future event types [v2]

Place drm_event_vblank in a new union that includes that and a bare
drm_event structure. This will allow new members of that union to be
added in the future without changing code related to the existing vbl
event type.

Assignments to the crtc_id field are now done when the event is
allocated, rather than when delievered. This way, delivery doesn't
need to have the crtc ID available.

v2:
 * Remove 'dev' argument from create_vblank_event

It wasn't being used anyways, and if we need it in the future,
we can always get it from crtc->dev.

 * Check for MODESETTING before looking for crtc in queue_vblank_event

UMS drivers will oops if we try to get a crtc, so make sure
we're modesetting before we try to find a crtc_id to fill into
the event.

Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
(cherry picked from commit dc695b85fde88eca3ef3b03fcd82f15b6bc6e462)

7 years agodrm: Widen vblank count to 64-bits [v3]
Keith Packard [Thu, 12 Oct 2017 18:57:49 +0000 (11:57 -0700)]
drm: Widen vblank count to 64-bits [v3]

This modifies the datatypes used by the vblank code to provide 64 bits
of vblank count.

The driver interfaces have been left using 32 bits of vblank count;
all of the code necessary to widen that value for the user API was
already included to handle devices returning fewer than 32-bits.

This will provide the necessary datatypes for the Vulkan API.

v2:

 * Re-write wait_vblank ioctl to ABSOLUTE sequence

    When an application uses the WAIT_VBLANK ioctl with RELATIVE
    or NEXTONMISS bits set, the target vblank interval is updated
    within the kernel. We need to write that target back to the
    ioctl buffer and update the flags bits so that if the wait is
    interrupted by a signal, when it is re-started, it will target
    precisely the same vblank count as before.

 * Leave driver API with 32-bit vblank count

v3:

 * Rebase on top of Arnd Bergmann's patch which had
   the switch to ktime_t parts.

[airlied: fix conflict with Ville vblank change].

Suggested-by: Michel Dänzer <michel@daenzer.net>
Suggested-by: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
(cherry picked from commit 2affbc16983e4fc90960bc7f70e7615f4228199b)

7 years agodrm/tegra: hdmi: Add cec-notifier support
Hans Verkuil [Mon, 11 Sep 2017 12:29:52 +0000 (14:29 +0200)]
drm/tegra: hdmi: Add cec-notifier support

In order to support CEC the HDMI driver has to inform the CEC driver
whenever the physical address changes. So when the EDID is read the
CEC driver has to be informed and whenever the hotplug detect goes
away.

This is done through the cec-notifier framework.

The link between the HDMI driver and the CEC driver is done through
the hdmi-phandle property in the tegra-cec node in the device tree.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agodrm/tegra: dc: Perform a complete reset sequence
Thierry Reding [Wed, 30 Aug 2017 15:41:00 +0000 (17:41 +0200)]
drm/tegra: dc: Perform a complete reset sequence

In order for the reset to be applied properly, the module clock must be
enabled during the assertion.

Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agodrm/tegra: dc: Make sure to set the module clock rate
Thierry Reding [Wed, 30 Aug 2017 15:38:39 +0000 (17:38 +0200)]
drm/tegra: dc: Make sure to set the module clock rate

When applying the PLL changes from the computed state object, make sure
to set the rate of the display controller module clock. Failing to do so
can yield to a situation where the parent will be set to the proper
pixel clock, but the module clock will be divided down to the rate that
is happened to be set to before the parent rate change.

Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agodrm/tegra: dc: Simplify atomic plane helper functions
Thierry Reding [Wed, 30 Aug 2017 15:34:10 +0000 (17:34 +0200)]
drm/tegra: dc: Simplify atomic plane helper functions

Remove the tegra_dc_disable_window() function whose only purpose was to
allow tegra_plane_atomic_update() to also call it. Fix that by shuffling
tegra_plano_atomic_disable() to before tegra_plane_atomic_update().

While at it, also remove the overlay plane helper functions because they
are exactly the same as the primary plane helper functions.

Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agodrm/tegra: dc: Move some declarations to dc.h
Thierry Reding [Wed, 30 Aug 2017 15:25:34 +0000 (17:25 +0200)]
drm/tegra: dc: Move some declarations to dc.h

Move the DC related declarations in drm.h to dc.h where they belong.

Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agodrm/tegra: vic: Use of_device_get_match_data()
Thierry Reding [Mon, 21 Aug 2017 16:03:27 +0000 (18:03 +0200)]
drm/tegra: vic: Use of_device_get_match_data()

Avoid some boilerplate by calling of_device_get_match_data() instead of
open-coding the equivalent in the driver.

Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agodrm/tegra: sor: Use of_device_get_match_data()
Thierry Reding [Mon, 21 Aug 2017 15:33:14 +0000 (17:33 +0200)]
drm/tegra: sor: Use of_device_get_match_data()

Avoid some boilerplate by calling of_device_get_match_data() instead of
open-coding the equivalent in the driver.

Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agodrm/tegra: hdmi: Use of_device_get_match_data()
Thierry Reding [Mon, 21 Aug 2017 16:05:10 +0000 (18:05 +0200)]
drm/tegra: hdmi: Use of_device_get_match_data()

Avoid some boilerplate by calling of_device_get_match_data() instead of
open-coding the equivalent in the driver.

Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agodrm/tegra: dc: Use of_device_get_match_data()
Thierry Reding [Mon, 21 Aug 2017 14:35:17 +0000 (16:35 +0200)]
drm/tegra: dc: Use of_device_get_match_data()

Avoid some boilerplate by calling of_device_get_match_data() instead of
open-coding the equivalent in the driver.

Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agodrm/tegra: Use u64_to_user_ptr helper
Mikko Perttunen [Thu, 28 Sep 2017 12:50:44 +0000 (15:50 +0300)]
drm/tegra: Use u64_to_user_ptr helper

Use the u64_to_user_ptr helper macro to cast IOCTL argument u64 values
to user pointers instead of writing out the cast manually. Also do
some other cleanup with user pointers to make them stand out more
and look cleaner.

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Tested-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agogpu: host1x: Fix incorrect comment for channel_request
Mikko Perttunen [Thu, 28 Sep 2017 12:50:43 +0000 (15:50 +0300)]
gpu: host1x: Fix incorrect comment for channel_request

This function actually doesn't sleep in the version that was merged.

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agogpu: host1x: Disassemble more instructions
Mikko Perttunen [Thu, 28 Sep 2017 12:50:42 +0000 (15:50 +0300)]
gpu: host1x: Disassemble more instructions

The disassembler for debug dumps was missing some newer host1x opcodes.
Add disassembly support for these.

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Tested-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agogpu: host1x: Improve debug disassembly formatting
Mikko Perttunen [Thu, 28 Sep 2017 12:50:41 +0000 (15:50 +0300)]
gpu: host1x: Improve debug disassembly formatting

The host1x driver prints out "disassembly" dumps of the command FIFO
and gather contents on submission timeouts. However, the output has
been quite difficult to read with unnecessary newlines and occasional
missing parentheses.

Fix these problems by using pr_cont to remove unnecessary newlines
and by fixing other small issues.

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Tested-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agogpu: host1x: Enable gather filter
Mikko Perttunen [Thu, 28 Sep 2017 12:50:40 +0000 (15:50 +0300)]
gpu: host1x: Enable gather filter

The gather filter is a feature present on Tegra124 and newer where the
hardware prevents GATHERed command buffers from executing commands
normally reserved for the CDMA pushbuffer which is maintained by the
kernel driver.

This commit enables the gather filter on all supporting hardware.

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agogpu: host1x: Enable Tegra186 syncpoint protection
Mikko Perttunen [Thu, 28 Sep 2017 12:50:39 +0000 (15:50 +0300)]
gpu: host1x: Enable Tegra186 syncpoint protection

Since Tegra186 the Host1x hardware allows syncpoints to be assigned to
specific channels, preventing any other channels from incrementing
them.

Enable this feature where available and assign syncpoints to channels
when submitting a job. Syncpoints are currently never unassigned from
channels since that would require extra work and is unnecessary with
the current channel allocation model.

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agogpu: host1x: Call of_dma_configure() after setting bus
Mikko Perttunen [Sun, 24 Sep 2017 09:04:53 +0000 (12:04 +0300)]
gpu: host1x: Call of_dma_configure() after setting bus

of_dma_configure() now checks the device's bus before configuring it, so
we need to set the device's bus before calling.

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agodrm/tegra: Add Tegra186 support for VIC
Mikko Perttunen [Tue, 5 Sep 2017 08:43:06 +0000 (11:43 +0300)]
drm/tegra: Add Tegra186 support for VIC

Add Tegra186 support for VIC - no changes are required except for new
firmware and compatibility string.

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agogpu: host1x: Add Tegra186 support
Mikko Perttunen [Tue, 5 Sep 2017 08:43:05 +0000 (11:43 +0300)]
gpu: host1x: Add Tegra186 support

Add support for the implementation of Host1x present on the Tegra186.
The register space has been shuffled around a little bit, requiring
addition of some chip-specific code sections. Tegra186 also adds
several new features, most importantly the hypervisor, but those are
not yet supported with this commit.

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Tested-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agodt-bindings: host1x: Add Tegra186 information
Mikko Perttunen [Tue, 5 Sep 2017 08:43:04 +0000 (11:43 +0300)]
dt-bindings: host1x: Add Tegra186 information

Add the Tegra186-specific hypervisor-related register range
properties.

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agogpu: host1x: syncpt: Request syncpoints per client
Thierry Reding [Wed, 30 Aug 2017 10:48:31 +0000 (12:48 +0200)]
gpu: host1x: syncpt: Request syncpoints per client

Rather than request syncpoints for a struct device *, request them for a
struct host1x_client *. This is important because subsequent patches are
going to break the assumption that host1x will always be the parent for
devices requesting a syncpoint. It's also a more natural choice because
host1x clients are really the only ones that will know how to deal with
syncpoints.

Note that host1x clients are always guaranteed to be children of host1x,
regardless of their location in the device tree.

Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agogpu: host1x: Use of_device_get_match_data()
Thierry Reding [Mon, 21 Aug 2017 16:08:42 +0000 (18:08 +0200)]
gpu: host1x: Use of_device_get_match_data()

Avoid some boilerplate by calling of_device_get_match_data() instead of
open-coding the equivalent in the driver.

While at it, shuffle around some code to avoid unnecessary local
variables.

Signed-off-by: Thierry Reding <treding@nvidia.com>
7 years agoMerge tag 'drm-amdkfd-next-2017-10-18' of git://people.freedesktop.org/~gabbayo/linux...
Dave Airlie [Fri, 20 Oct 2017 05:54:44 +0000 (15:54 +1000)]
Merge tag 'drm-amdkfd-next-2017-10-18' of git://people.freedesktop.org/~gabbayo/linux into drm-next

This is the amdkfd pull request for 4.15 merge window.
The patches here are relevant only for Kaveri and Carrizo. Still no dGPU patches.

The main goal is to continue alignment with the internal AMD development tree.

The following is a summary of the changes:
- Improvements and fixes to suspend/resume code
- Improvements and fixes to process termination code
- Cleanups in the queue unmapping functionality
- Reuse code from amdgpu

* tag 'drm-amdkfd-next-2017-10-18' of git://people.freedesktop.org/~gabbayo/linux:
  drm/amdkfd: Improve multiple SDMA queues support per process
  drm/amdkfd: Limit queue number per process and device to 127
  drm/amdkfd: Clean up process queue management
  drm/amdkfd: Compress unnecessary function parameters
  drm/amdkfd: Improve process termination handling
  drm/amdkfd: Avoid submitting an unnecessary packet to HWS
  drm/amdkfd: Fix MQD updates
  drm/amdkfd: Pass filter params to unmap_queues_cpsch
  drm/amdkfd: move locking outside of unmap_queues_cpsch
  drm/amdkfd: Avoid name confusion involved in queue unmapping
  drm/amdkfd: Drop _nocpsch suffix from shared functions
  drm/amdkfd: Reuse CHIP_* from amdgpu v2
  drm/amdkfd: Use VMID bitmap from KGD v2
  drm/amdkfd: Adjust dequeue latencies and timeouts
  drm/amdkfd: Rectify the jiffies calculation error with milliseconds v2
  drm/amdkfd: Fix suspend/resume issue on Carrizo v2
  drm/amdkfd: Reorganize kfd resume code

7 years agoMerge tag 'drm-intel-next-2017-10-12' of git://anongit.freedesktop.org/drm/drm-intel...
Dave Airlie [Fri, 20 Oct 2017 00:56:10 +0000 (10:56 +1000)]
Merge tag 'drm-intel-next-2017-10-12' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

Last batch of drm/i915 features for v4.15:

- transparent huge pages support (Matthew)
- uapi: I915_PARAM_HAS_SCHEDULER into a capability bitmask (Chris)
- execlists: preemption (Chris)
- scheduler: user defined priorities (Chris)
- execlists optimization (Michał)
- plenty of display fixes (Imre)
- has_ipc fix (Rodrigo)
- platform features definition refactoring (Rodrigo)
- legacy cursor update fix (Maarten)
- fix vblank waits for cursor updates (Maarten)
- reprogram dmc firmware on resume, dmc state fix (Imre)
- remove use_mmio_flip module parameter (Maarten)
- wa fixes (Oscar)
- huc/guc firmware refacoring (Sagar, Michal)
- push encoder specific code to encoder hooks (Jani)
- DP MST fixes (Dhinakaran)
- eDP power sequencing fixes (Manasi)
- selftest updates (Chris, Matthew)
- mmu notifier cpu hotplug deadlock fix (Daniel)
- more VBT parser refactoring (Jani)
- max pipe refactoring (Mika Kahola)
- rc6/rps refactoring and separation (Sagar)
- userptr lockdep fix (Chris)
- tracepoint fixes and defunct tracepoint removal (Chris)
- use rcu instead of abusing stop_machine (Daniel)
- plenty of other fixes all around (Everyone)

* tag 'drm-intel-next-2017-10-12' of git://anongit.freedesktop.org/drm/drm-intel: (145 commits)
  drm/i915: Update DRIVER_DATE to 20171012
  drm/i915: Simplify intel_sanitize_enable_ppgtt
  drm/i915/userptr: Drop struct_mutex before cleanup
  drm/i915/dp: limit sink rates based on rate
  drm/i915/dp: centralize max source rate conditions more
  drm/i915: Allow PCH platforms fall back to BIOS LVDS mode
  drm/i915: Reuse normal state readout for LVDS/DVO fixed mode
  drm/i915: Use rcu instead of stop_machine in set_wedged
  drm/i915: Introduce separate status variable for RC6 and LLC ring frequency setup
  drm/i915: Create generic functions to control RC6, RPS
  drm/i915: Create generic function to setup LLC ring frequency table
  drm/i915: Rename intel_enable_rc6 to intel_rc6_enabled
  drm/i915: Name structure in dev_priv that contains RPS/RC6 state as "gt_pm"
  drm/i915: Move rps.hw_lock to dev_priv and s/hw_lock/pcu_lock
  drm/i915: Name i915_runtime_pm structure in dev_priv as "runtime_pm"
  drm/i915: Separate RPS and RC6 handling for CHV
  drm/i915: Separate RPS and RC6 handling for VLV
  drm/i915: Separate RPS and RC6 handling for BDW
  drm/i915: Remove superfluous IS_BDW checks and non-BDW changes from gen8_enable_rps
  drm/i915: Separate RPS and RC6 handling for gen6+
  ...

7 years agoMerge branch 'drm-next-4.15' of git://people.freedesktop.org/~agd5f/linux into drm...
Dave Airlie [Fri, 20 Oct 2017 00:47:19 +0000 (10:47 +1000)]
Merge branch 'drm-next-4.15' of git://people.freedesktop.org/~agd5f/linux into drm-next

Last set of features for 4.15.  Highlights:
- Add a bo flag to allow buffers to opt out of implicit sync
- Add ctx priority setting interface
- Lots more powerplay cleanups
- Start to plumb through vram lost infrastructure for gpu reset
- ttm support for huge pages
- misc cleanups and bug fixes

* 'drm-next-4.15' of git://people.freedesktop.org/~agd5f/linux: (73 commits)
  drm/amd/powerplay: Place the constant on the right side of the test
  drm/amd/powerplay: Remove useless variable
  drm/amd/powerplay: Don't cast kzalloc() return value
  drm/amdgpu: allow GTT overcommit during bind
  drm/amdgpu: linear validate first then bind to GART
  drm/amd/pp: Fix overflow when setup decf/pix/disp dpm table.
  drm/amd/pp: thermal control not enabled on vega10.
  drm/amdgpu: busywait KIQ register accessing (v4)
  drm/amdgpu: report more amdgpu_fence_info
  drm/amdgpu:don't check soft_reset for sriov
  drm/amdgpu:fix duplicated setting job's vram_lost
  drm/amdgpu:reduce wb to 512 slot
  drm/amdgpu: fix regresstion on SR-IOV gpu reset failed
  drm/amd/powerplay: Tidy up cz_dpm_powerup_vce()
  drm/amd/powerplay: Tidy up cz_dpm_powerdown_vce()
  drm/amd/powerplay: Tidy up cz_dpm_update_vce_dpm()
  drm/amd/powerplay: Tidy up cz_dpm_update_uvd_dpm()
  drm/amd/powerplay: Tidy up cz_dpm_powerup_uvd()
  drm/amd/powerplay: Tidy up cz_dpm_powerdown_uvd()
  drm/amd/powerplay: Tidy up cz_start_dpm()
  ...

7 years agodrm/amd/powerplay: Place the constant on the right side of the test
Georgiana Chelu [Tue, 17 Oct 2017 20:22:09 +0000 (23:22 +0300)]
drm/amd/powerplay: Place the constant on the right side of the test

Move the constant on the right side of the comparison in order to
make the code easier to read.

Issue found by checkpatch script:
* WARNING: Comparisons should place the constant on the right side of
the test

Signed-off-by: Georgiana Chelu <georgiana.chelu93@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amd/powerplay: Remove useless variable
Georgiana Chelu [Tue, 17 Oct 2017 20:22:08 +0000 (23:22 +0300)]
drm/amd/powerplay: Remove useless variable

The result variable is initialized at the beginning of the function, but
its value does not change during the function execution. Thus, remove the
variable and return the SUCCESS value, which is 0.

Issue found by coccinelle script:
* Unneeded variable: "result". Return "0"

Path to the cocci script: scripts/coccinelle/misc/returnvar.cocci

Signed-off-by: Georgiana Chelu <georgiana.chelu93@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amd/powerplay: Don't cast kzalloc() return value
Georgiana Chelu [Tue, 17 Oct 2017 20:22:07 +0000 (23:22 +0300)]
drm/amd/powerplay: Don't cast kzalloc() return value

The kzalloc function returns a void pointer and the assignment
operator converts it to the type of pointer it is assigned to.
Therefore, there is no need to cast.

Issue found by alloc_cast.cocci:
* WARNING: casting value returned by memory allocation function
to <struct type> is useless.

Path to the cocci script: scripts/coccinelle/api/alloc/alloc_cast.cocci

Signed-off-by: Georgiana Chelu <georgiana.chelu93@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: allow GTT overcommit during bind
Christian König [Mon, 16 Oct 2017 09:18:54 +0000 (11:18 +0200)]
drm/amdgpu: allow GTT overcommit during bind

While binding BOs to GART we need to allow a bit overcommit in the GTT
domain. Otherwise we can never use the full GART space when GART size=GTT size.

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>
7 years agodrm/amdgpu: linear validate first then bind to GART
Christian König [Mon, 16 Oct 2017 08:32:04 +0000 (10:32 +0200)]
drm/amdgpu: linear validate first then bind to GART

For VM emulation for old UVD/VCE we need to validate the BO with linear
VRAM flag set first and then eventually bind it to GART.

Validating with linear VRAM flag set can move the BO to GART making
UVD/VCE read/write from an unbound GART BO.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
CC: stable@vger.kernel.org
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amd/pp: Fix overflow when setup decf/pix/disp dpm table.
Rex Zhu [Wed, 18 Oct 2017 09:43:43 +0000 (17:43 +0800)]
drm/amd/pp: Fix overflow when setup decf/pix/disp dpm table.

Clear the count in the single table setup function to
avoid missing any tables.

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>
7 years agodrm/amd/pp: thermal control not enabled on vega10.
Rex Zhu [Wed, 18 Oct 2017 06:31:40 +0000 (14:31 +0800)]
drm/amd/pp: thermal control not enabled on vega10.

regression issue.
caused by "f12f9f5e5d455edebc01"
forget to set start_thermal_controller function point.

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>
7 years agodrm/amdgpu: busywait KIQ register accessing (v4)
pding [Fri, 13 Oct 2017 07:38:35 +0000 (15:38 +0800)]
drm/amdgpu: busywait KIQ register accessing (v4)

Register accessing is performed when IRQ is disabled. Never sleep in
this function.

Known issue: dead sleep in many use cases of index/data registers.

v2:
 - wrap polling fence functions.
 - don't trigger IRQ for polling in case of wrongly fence signal.

v3:
 - handle wrap round gracefully.
 - add comments for polling function

v4:
 - don't return negative timeout confused with error code

Signed-off-by: pding <Pixel.Ding@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: report more amdgpu_fence_info
pding [Thu, 12 Oct 2017 05:53:20 +0000 (13:53 +0800)]
drm/amdgpu: report more amdgpu_fence_info

Only for GFX ring. This can help checking MCBP feature.

The fence at the end of the frame will indicate the completion status.
If the frame completed normally, the fence is written to the address
given in the EVENT_WRITE_EOP packet. If preemption occurred in the
previous IB the address is adjusted by 2 DWs. If work submitted in the
frame was reset before completion, the fence address is adjusted by
four DWs. In the case that preemption occurred, and before preemption
completed a reset was initiated, the address will be adjusted with six
DWs

Signed-off-by: pding <Pixel.Ding@amd.com>
Reviewed-by: Monk Liu <monk.liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu:don't check soft_reset for sriov
Monk Liu [Mon, 16 Oct 2017 11:46:01 +0000 (19:46 +0800)]
drm/amdgpu:don't check soft_reset for sriov

Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Ack-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu:fix duplicated setting job's vram_lost
Monk Liu [Mon, 16 Oct 2017 12:02:08 +0000 (20:02 +0800)]
drm/amdgpu:fix duplicated setting job's vram_lost

Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu:reduce wb to 512 slot
Monk Liu [Tue, 17 Oct 2017 11:23:42 +0000 (19:23 +0800)]
drm/amdgpu:reduce wb to 512 slot

with current WB usage we only use 57 slots, so 512
is extreamly sufficient, and reduce to 512 can
make WB fit into one page.

Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: fix regresstion on SR-IOV gpu reset failed
Rex Zhu [Wed, 18 Oct 2017 09:19:42 +0000 (17:19 +0800)]
drm/amdgpu: fix regresstion on SR-IOV gpu reset failed

fw ucode is corrupted after vf flr by PSP so ucode_init() is
a must in psp_hw_init othewise KIQ/KCQ enabling will fail

Revert "drm/amdgpu: refine code delete duplicated error handling"
This reverts commit e57b87ff828f95efe992468e6d18c2c059b27aa9.
Revert "drm/amdgpu: move amdgpu_ucode_init_bo to amdgpu_device.c"
This reverts commit 815b8f8595148d06a64d2ce4282e8e80dfcb02f1.

Reviewed-by: Monk Liu <monk.liu@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amd/powerplay: Tidy up cz_dpm_powerup_vce()
Tom St Denis [Mon, 16 Oct 2017 17:33:25 +0000 (13:33 -0400)]
drm/amd/powerplay: Tidy up cz_dpm_powerup_vce()

Use PP_CAP macro.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amd/powerplay: Tidy up cz_dpm_powerdown_vce()
Tom St Denis [Mon, 16 Oct 2017 17:32:38 +0000 (13:32 -0400)]
drm/amd/powerplay: Tidy up cz_dpm_powerdown_vce()

Use PP_CAP macro.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amd/powerplay: Tidy up cz_dpm_update_vce_dpm()
Tom St Denis [Mon, 16 Oct 2017 17:31:49 +0000 (13:31 -0400)]
drm/amd/powerplay: Tidy up cz_dpm_update_vce_dpm()

Use PP_CAP and tidy up indentation.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amd/powerplay: Tidy up cz_dpm_update_uvd_dpm()
Tom St Denis [Mon, 16 Oct 2017 17:30:28 +0000 (13:30 -0400)]
drm/amd/powerplay: Tidy up cz_dpm_update_uvd_dpm()

Use PP_CAP and tidy up indentation.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amd/powerplay: Tidy up cz_dpm_powerup_uvd()
Tom St Denis [Mon, 16 Oct 2017 17:28:24 +0000 (13:28 -0400)]
drm/amd/powerplay: Tidy up cz_dpm_powerup_uvd()

Use PP_CAP and simplify enable/disable logic.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amd/powerplay: Tidy up cz_dpm_powerdown_uvd()
Tom St Denis [Mon, 16 Oct 2017 17:25:42 +0000 (13:25 -0400)]
drm/amd/powerplay: Tidy up cz_dpm_powerdown_uvd()

Use PP_CAP and tidy up indentation.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amd/powerplay: Tidy up cz_start_dpm()
Tom St Denis [Mon, 16 Oct 2017 17:23:55 +0000 (13:23 -0400)]
drm/amd/powerplay: Tidy up cz_start_dpm()

Remove unused variables.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: use ARRAY_SIZE
Jérémy Lefaure [Mon, 16 Oct 2017 02:29:23 +0000 (22:29 -0400)]
drm/amdgpu: use ARRAY_SIZE

Using the ARRAY_SIZE macro improves the readability of the code.

Found with Coccinelle with the following semantic patch:
@r depends on (org || report)@
type T;
T[] E;
position p;
@@
(
 (sizeof(E)@p /sizeof(*E))
|
 (sizeof(E)@p /sizeof(E[...]))
|
 (sizeof(E)@p /sizeof(T))
)

Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Jérémy Lefaure <jeremy.lefaure@lse.epita.fr>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: reserve root PD while releasing it
Christian König [Fri, 13 Oct 2017 15:24:31 +0000 (17:24 +0200)]
drm/amdgpu: reserve root PD while releasing it

Otherwise somebody could try to evict it at the same time and try to use
half torn down structures.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-and-Tested-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/ttm: fix the fix for huge compound pages
Christian König [Thu, 12 Oct 2017 17:28:42 +0000 (19:28 +0200)]
drm/ttm: fix the fix for huge compound pages

We don't use compound pages at the moment. Take this into account when
freeing them.

Signed-off-by: Christian König <christian.koenig@amd.comd>
Reviewed-and-Tested-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: bump version for vram lost counter query (v2)
Alex Deucher [Thu, 12 Oct 2017 20:26:34 +0000 (16:26 -0400)]
drm/amdgpu: bump version for vram lost counter query (v2)

v2: vram -> VRAM in comment

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: minor CS optimization
Christian König [Thu, 12 Oct 2017 10:16:33 +0000 (12:16 +0200)]
drm/amdgpu: minor CS optimization

We only need to loop over all IBs for old UVD/VCE command stream patching.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amd/sched: fix job tear down order v2
Christian König [Fri, 13 Oct 2017 08:58:15 +0000 (10:58 +0200)]
drm/amd/sched: fix job tear down order v2

Move the trace before we signal the scheduler fence and drop the
scheduler fence reference directly before we free the job.

v2: keep extra s_fence reference

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Liu, Monk <Monk.Liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: Add amdgpu_find_mm_node()
Harish Kasiviswanathan [Fri, 6 Oct 2017 21:36:35 +0000 (17:36 -0400)]
drm/amdgpu: Add amdgpu_find_mm_node()

Replace some commonly repeated code with a function.

v2: Use amdgpu_find_mm_node() in amdgpu_ttm_io_mem_pfn()

Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: Refactor amdgpu_move_blit
Harish Kasiviswanathan [Tue, 3 Oct 2017 19:41:56 +0000 (15:41 -0400)]
drm/amdgpu: Refactor amdgpu_move_blit

Add more generic function amdgpu_copy_ttm_mem_to_mem() that supports
arbitrary copy size, offsets and two BOs (source & dest.).

This is useful for KFD Cross Memory Attach feature where data needs to
be copied from BOs from different processes

v2: Add struct amdgpu_copy_mem and changed amdgpu_copy_ttm_mem_to_mem()
function parameters to use the struct

v3: Minor function name change

Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: Fix extra call to amdgpu_ctx_put.
Andrey Grodzovsky [Wed, 11 Oct 2017 21:02:02 +0000 (17:02 -0400)]
drm/amdgpu: Fix extra call to amdgpu_ctx_put.

In amdgpu_cs_parser_init() in case of error handling
amdgpu_ctx_put() is called without setting p->ctx to NULL after that,
later amdgpu_cs_parser_fini() also calls amdgpu_ctx_put() again and
mess up the reference count.

Signed-off-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/ttm: Fix unused variables with huge page support
Tom St Denis [Thu, 12 Oct 2017 11:25:08 +0000 (07:25 -0400)]
drm/ttm: Fix unused variables with huge page support

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: add VRAM lost query
Christian König [Mon, 9 Oct 2017 15:53:06 +0000 (17:53 +0200)]
drm/amdgpu: add VRAM lost query

Allows userspace to figure out if VRAM was lost.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: set -ECANCELED when dropping jobs
Christian König [Mon, 9 Oct 2017 13:51:10 +0000 (15:51 +0200)]
drm/amdgpu: set -ECANCELED when dropping jobs

And return from the wait functions the fence error code.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: move the VRAM lost counter per context
Christian König [Mon, 9 Oct 2017 13:18:43 +0000 (15:18 +0200)]
drm/amdgpu: move the VRAM lost counter per context

Instead of per device track the VRAM lost per context and return ECANCELED
instead of ENODEV.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: keep copy of VRAM lost counter in job
Christian König [Mon, 9 Oct 2017 13:04:41 +0000 (15:04 +0200)]
drm/amdgpu: keep copy of VRAM lost counter in job

Instead of reading the current counter from fpriv.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: partial revert VRAM lost handling v2
Christian König [Mon, 9 Oct 2017 12:45:09 +0000 (14:45 +0200)]
drm/amdgpu: partial revert VRAM lost handling v2

Keep blocking the CS, but revert everything else. Mapping BOs and info IOCTL
are harmless and can still happen even when VRAM content ist lost.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/ttm: add transparent huge page support for wc or uc allocations v2
Christian König [Thu, 5 Oct 2017 12:27:34 +0000 (14:27 +0200)]
drm/ttm: add transparent huge page support for wc or uc allocations v2

Add a new huge page pool and try to allocate from it when it makes sense.

v2: avoid compound pages for now

Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/ttm: move more logic into ttm_page_pool_get_pages
Christian König [Thu, 21 Sep 2017 09:28:25 +0000 (11:28 +0200)]
drm/ttm: move more logic into ttm_page_pool_get_pages

Make it easier to add huge page pool.

Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/ttm: add transparent huge page support for cached allocations v2
Christian König [Wed, 20 Sep 2017 13:06:12 +0000 (15:06 +0200)]
drm/ttm: add transparent huge page support for cached allocations v2

Try to allocate huge pages when it makes sense.

v2: avoid compound pages for now

Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Felix Kuehling <Felix.Kuehling@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/ttm: don't use compound pages for now
Christian König [Mon, 9 Oct 2017 12:34:13 +0000 (14:34 +0200)]
drm/ttm: don't use compound pages for now

We need to figure out first how to correctly map them into the CPU page tables.

bug: https://bugs.freedesktop.org/show_bug.cgi?id=103138
Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: correct reference clock value on vega10
Ken Wang [Fri, 29 Sep 2017 07:41:43 +0000 (15:41 +0800)]
drm/amdgpu: correct reference clock value on vega10

Old value from bringup was wrong.

Cc: stable@vger.kernel.org
Signed-off-by: Ken Wang <Ken.Wang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: SR-IOV data exchange between PF&VF
Horace Chen [Mon, 9 Oct 2017 08:17:16 +0000 (16:17 +0800)]
drm/amdgpu: SR-IOV data exchange between PF&VF

SR-IOV need to exchange some data between PF&VF through shared VRAM

PF will copy some necessary firmware and information to the shared
VRAM. It also requires some information from VF. PF will send a
key through mailbox2 to help guest calculate checksum so that it can
verify whether the data is correct.

So check the data on the specified offset of the shared VRAM, if the
checksum is right, read values from it and write some VF information
next to the data from PF.

Signed-off-by: Horace Chen <horace.chen@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: Move old fence waiting before reservation lock is aquired v2
Andrey Grodzovsky [Tue, 10 Oct 2017 20:50:17 +0000 (16:50 -0400)]
drm/amdgpu: Move old fence waiting before reservation lock is aquired v2

Helps avoiding deadlock during GPU reset.
Added mutex to amdgpu_ctx to preserve order of fences on a ring.

v2:
Put waiting logic in a function in a seperate function in amdgpu_ctx.c

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: Refactor amdgpu_cs_ib_vm_chunk and amdgpu_cs_ib_fill.
Andrey Grodzovsky [Tue, 10 Oct 2017 20:50:16 +0000 (16:50 -0400)]
drm/amdgpu: Refactor amdgpu_cs_ib_vm_chunk and amdgpu_cs_ib_fill.

This enables old fence waiting before reservation lock is aquired
which in turn is part of a bigger solution to deadlock happening
when gpu reset with VRAM recovery accures during intensive rendering.

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: make function uvd_v6_0_enc_get_destroy_msg static
Colin Ian King [Wed, 11 Oct 2017 09:21:11 +0000 (10:21 +0100)]
drm/amdgpu: make function uvd_v6_0_enc_get_destroy_msg static

The function uvd_v6_0_enc_get_destroy_msg is local to the source and
does not need to be in global scope, so make it static.

Cleans up sparse warning:
symbol 'uvd_v6_0_enc_get_destroy_msg' was not declared. Should it be
static?

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amd/pp: remove polaris10_smc/smumgr split.
Rex Zhu [Mon, 9 Oct 2017 08:33:06 +0000 (16:33 +0800)]
drm/amd/pp: remove polaris10_smc/smumgr split.

move functions in polaris_smc.c to
polaris10_smumgr.c and make all functions in
polaris10_smumgr.c static.

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>
7 years agodrm/amd/pp: remove tonga_smc/smumgr split.
Rex Zhu [Mon, 9 Oct 2017 08:20:49 +0000 (16:20 +0800)]
drm/amd/pp: remove tonga_smc/smumgr split.

move functions in tonga_smc.c to tonga_smumgr.c
and make all functions in tonga_smumgr.c static.

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>
7 years agodrm/amd/pp: remove iceland_smc/smumgr split.
Rex Zhu [Mon, 9 Oct 2017 08:04:39 +0000 (16:04 +0800)]
drm/amd/pp: remove iceland_smc/smumgr split.

move functions in iceland_smc.c to iceland_smumgr.c
and make all functions in iceland_smumgr.c static.

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>
7 years agodrm/amd/pp: remove fiji_smc/smumgr split.
Rex Zhu [Mon, 9 Oct 2017 07:37:32 +0000 (15:37 +0800)]
drm/amd/pp: remove fiji_smc/smumgr split.

make all functions in fiji_smumgr.c static and
exported by pp_smumgr_func table.

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>
7 years agodrm/amd/pp: rename ci_smc.c to ci_smumgr.c
Rex Zhu [Mon, 9 Oct 2017 07:05:21 +0000 (15:05 +0800)]
drm/amd/pp: rename ci_smc.c to ci_smumgr.c

for consistency.

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>
7 years agodrm/amdgpu: refine code delete duplicated error handling
Rex Zhu [Mon, 9 Oct 2017 05:50:31 +0000 (13:50 +0800)]
drm/amdgpu: refine code delete duplicated error handling

in function amdgpu_ucode_init_bo, when failed, it will
set load_type to AMDGPU_FW_LOAD_DIRECT.

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>
7 years agodrm/amd/pp: move common function to smu7_smumgr.c
Rex Zhu [Mon, 9 Oct 2017 05:17:26 +0000 (13:17 +0800)]
drm/amd/pp: move common function to smu7_smumgr.c

fiji and polaris can share same setup_pwr_virus
function.

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>
7 years agodrm/amd/pp: implement function notify_cac_buffer_info on VI
Rex Zhu [Fri, 15 Sep 2017 11:39:52 +0000 (19:39 +0800)]
drm/amd/pp: implement function notify_cac_buffer_info on VI

Used for smu power logging.

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>
7 years agodrm/amd/pp: implement function notify_cac_buffer_info on Vega
Rex Zhu [Fri, 15 Sep 2017 11:39:09 +0000 (19:39 +0800)]
drm/amd/pp: implement function notify_cac_buffer_info on Vega

Used for smu power logging.

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>
7 years agodrm/amd/pp: add new function point in hwmgr.
Rex Zhu [Mon, 9 Oct 2017 04:21:30 +0000 (12:21 +0800)]
drm/amd/pp: add new function point in hwmgr.

used for notify SMU the allocated buffer address.

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>
7 years agodrm/amd/pp: export new smu messages for vega10
Rex Zhu [Fri, 15 Sep 2017 09:14:09 +0000 (17:14 +0800)]
drm/amd/pp: export new smu messages for vega10

New messages for smu power logging.

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>
7 years agodrm/amd/powerplay: get the temperature on CZ
Satyajit Sahu [Fri, 6 Oct 2017 06:28:16 +0000 (06:28 +0000)]
drm/amd/powerplay: get the temperature on CZ

Setting the function pointer to the get the temperature on CZ.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Satyajit Sahu <satyajit.sahu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: bump version for new AMDGPU_SCHED ioctl
Alex Deucher [Mon, 9 Oct 2017 20:28:16 +0000 (16:28 -0400)]
drm/amdgpu: bump version for new AMDGPU_SCHED ioctl

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: rename context priority levels
Andres Rodriguez [Fri, 13 Oct 2017 18:58:14 +0000 (14:58 -0400)]
drm/amdgpu: rename context priority levels

Don't leak implementation details about how each priority behaves to
usermode. This allows greater flexibility in the future.

Squash into c2636dc53abd8269a0930bccd564f2f195dba729

Signed-off-by: Andres Rodriguez <andresx7@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/plane: drop num_overlay_planes (v3)
Dave Airlie [Thu, 12 Oct 2017 23:52:01 +0000 (00:52 +0100)]
drm/plane: drop num_overlay_planes (v3)

In order to implement plane leasing we need to count things,
just make the code consistent with the counting code currently
used for counting crtcs/encoders/connectors and drop the need
for num_overlay_planes.

v2: don't forget to assign plane_ptr. (keithp)
v3: use correct bounds check, found by igt.

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
7 years agoMerge tag 'tilcdc-4.15' of https://github.com/jsarha/linux into drm-next
Dave Airlie [Tue, 17 Oct 2017 00:13:47 +0000 (10:13 +1000)]
Merge tag 'tilcdc-4.15' of https://github.com/jsarha/linux into drm-next

tilcdc changes for v4.15

* tag 'tilcdc-4.15' of https://github.com/jsarha/linux:
  drm/tilcdc: Remove redundant OF_DETACHED flag setting
  drm/tilcdc: Precalculate total frametime in tilcdc_crtc_set_mode()
  drm/tilcdc: Use tilcdc_crtc_shutdown() in tilcdc_crtc_destroy()
  drm/tilcdc: Remove WARN_ON(!drm_modeset_is_locked(&crtc->mutex)) checks
  drm/tilcdc: Turn raster off in crtc reset, if it was on in the HW
  drm/tilcdc: switch to drm_*{get,put} helpers
  drm/tilcdc: tilcdc_tfp410: make of_device_ids const.
  drm/tilcdc: tilcdc_panel: make of_device_ids const.

7 years agoMerge tag 'drm-misc-next-2017-10-16' of git://anongit.freedesktop.org/drm/drm-misc...
Dave Airlie [Tue, 17 Oct 2017 00:10:17 +0000 (10:10 +1000)]
Merge tag 'drm-misc-next-2017-10-16' of git://anongit.freedesktop.org/drm/drm-misc into drm-next

Quick 4.15 misc pull for the build fix:

Cross-subsystem Changes:
- piles an piles of misc/trivial patches all over, some more from
  outreachy applicants

Core Changes:
- build fix for the bridge/of cleanup (Maarten)
- fix vblank count in arm_vblank_event (Ville)
- some kerneldoc typo fixes from Thierry

Driver Changes:
- vc4: Fix T-format tiling scanout, cleanup clock divider w/a (Anholt)
- sun4i: small cleanups and improved code comments all over (Chen-Yu
  Tsai)

* tag 'drm-misc-next-2017-10-16' of git://anongit.freedesktop.org/drm/drm-misc: (21 commits)
  drm/via: use ARRAY_SIZE
  drm/gma500: use ARRAY_SIZE
  drm/sun4i: hdmi: Move PAD_CTRL1 setting to mode_set function
  drm/sun4i: hdmi: Document PAD_CTRL1 output invert bits
  drm/sun4i: backend: Add comment explaining why registers are cleared
  drm/sun4i: backend: Use drm_fb_cma_get_gem_addr() to get display memory
  drm/sun4i: backend: Create regmap after access is possible
  drm/sun4i: don't add components that are already in the queue
  drm/vc4: Fix pitch setup for T-format scanout.
  drm/vc4: Move the DSI clock divider workaround closer to the clock call.
  drm: Replace kzalloc with kcalloc
  drm/tinydrm: Remove explicit .best_encoder assignment
  drm/tinydrm: Replace dev_error with DRM_DEV_ERROR
  drm/drm_of: Move drm_of_panel_bridge_remove_function into header.
  drm/atomic-helper: Fix reference to drm_crtc_send_vblank_event()
  drm/atomic-helper: Fix typo
  drm: Add missing __user annotation to drm_syncobj_array_find()
  drm/rockchip: add PINCTRL dependency for LVDS
  drm/kirin: Checking for IS_ERR() instead of NULL
  driver:gpu: return -ENOMEM on allocation failure.
  ...

7 years agodrm/via: use ARRAY_SIZE
Jérémy Lefaure [Mon, 16 Oct 2017 02:33:56 +0000 (22:33 -0400)]
drm/via: use ARRAY_SIZE

Using the ARRAY_SIZE macro improves the readability of the code.

Found with Coccinelle with the following semantic patch:
@r depends on (org || report)@
type T;
T[] E;
position p;
@@
(
 (sizeof(E)@p /sizeof(*E))
|
 (sizeof(E)@p /sizeof(E[...]))
|
 (sizeof(E)@p /sizeof(T))
)

Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Jérémy Lefaure <jeremy.lefaure@lse.epita.fr>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20171016023357.20174-1-jeremy.lefaure@lse.epita.fr
7 years agodrm/gma500: use ARRAY_SIZE
Jérémy Lefaure [Mon, 16 Oct 2017 02:30:46 +0000 (22:30 -0400)]
drm/gma500: use ARRAY_SIZE

Using the ARRAY_SIZE macro improves the readability of the code. Also,
it is useless to re-invent it.

Found with Coccinelle with the following semantic patch:
@r depends on (org || report)@
type T;
T[] E;
position p;
@@
(
 (sizeof(E)@p /sizeof(*E))
|
 (sizeof(E)@p /sizeof(E[...]))
|
 (sizeof(E)@p /sizeof(T))
)

Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Jérémy Lefaure <jeremy.lefaure@lse.epita.fr>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20171016023047.19145-1-jeremy.lefaure@lse.epita.fr
7 years agodrm/sun4i: hdmi: Move PAD_CTRL1 setting to mode_set function
Chen-Yu Tsai [Sat, 14 Oct 2017 04:02:52 +0000 (12:02 +0800)]
drm/sun4i: hdmi: Move PAD_CTRL1 setting to mode_set function

Initially we configured the PAD_CTRL1 register at probe/bind time.
However it seems the HDMI controller will modify some of the bits
in this register by itself. On the A10 it is particularly annoying
as it toggles the output invert bits, which inverts the colors on
the display output.

The U-boot driver this driver is based on sets this register twice,
though it seems it's only needed for actual display output. Hence
we move it to the mode_set function.

Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171014040252.9621-8-wens@csie.org
7 years agodrm/sun4i: hdmi: Document PAD_CTRL1 output invert bits
Chen-Yu Tsai [Sat, 14 Oct 2017 04:02:51 +0000 (12:02 +0800)]
drm/sun4i: hdmi: Document PAD_CTRL1 output invert bits

While debugging inverted color from the HDMI output on the A10, I
found that the lowest 3 bits were set. These were cleared on A20
boards that had normal display output. By manually toggling these
bits the mapping of the color components to these bits was found.

While these are not used anywhere, it would be nice to document
them somewhere.

Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171014040252.9621-7-wens@csie.org
7 years agodrm/sun4i: backend: Add comment explaining why registers are cleared
Chen-Yu Tsai [Sat, 14 Oct 2017 04:02:49 +0000 (12:02 +0800)]
drm/sun4i: backend: Add comment explaining why registers are cleared

Many of the backend's layer configuration registers have undefined
default values. This poses a risk as we use regmap_update_bits in
some places, and don't overwrite the whole register.

At probe/bind time we explicitly clear all the control registers
by writing 0 to them. This patch adds a more detailed explanation
on why we're doing this.

Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171014040252.9621-5-wens@csie.org
7 years agodrm/sun4i: backend: Use drm_fb_cma_get_gem_addr() to get display memory
Chen-Yu Tsai [Sat, 14 Oct 2017 04:02:48 +0000 (12:02 +0800)]
drm/sun4i: backend: Use drm_fb_cma_get_gem_addr() to get display memory

Commit 4636ce93d5b2 ("drm/fb-cma-helper: Add drm_fb_cma_get_gem_addr()")
adds a new helper, which covers fetching a drm_framebuffer's GEM object
and calculating the buffer address for a given plane.

This patch uses this helper to replace our own open coded version of the
same function.

Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171014040252.9621-4-wens@csie.org
7 years agodrm/sun4i: backend: Create regmap after access is possible
Chen-Yu Tsai [Sat, 14 Oct 2017 04:02:47 +0000 (12:02 +0800)]
drm/sun4i: backend: Create regmap after access is possible

The backend has various clocks and reset controls that need to be
enabled and deasserted before register access is possible.

Move the creation of the regmap to after the clocks and reset controls
have been configured where it makes more sense.

Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171014040252.9621-3-wens@csie.org
7 years agodrm/sun4i: don't add components that are already in the queue
Chen-Yu Tsai [Sat, 14 Oct 2017 04:02:46 +0000 (12:02 +0800)]
drm/sun4i: don't add components that are already in the queue

Even though the components framework can handle duplicate entries,
the extra entries cause a lot more debug messages to be generated,
which would be confusing to developers not familiar with our driver
and the framework in general.

Instead, we can scan the relatively small queue and check if the
component to be added is already queued up. Since the display
pipelines are symmetrical (not considering the third display
pipeline on the A80), and we add components level by level, when
we get to the second instance at the same level, any shared downstream
components would already be in the queue.

Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171014040252.9621-2-wens@csie.org
7 years agodrm/vc4: Fix pitch setup for T-format scanout.
Eric Anholt [Wed, 27 Sep 2017 19:32:09 +0000 (12:32 -0700)]
drm/vc4: Fix pitch setup for T-format scanout.

The documentation said to use src_w here, and I didn't consider that
we actually needed to be using pitch somewhere in our setup.  Fixes
scanout on my DSI panel when X11 does initial setup with 1920x1080
HDMI and 800x480 DSI both at 0,0 of the same framebuffer.

v2: Add some comments requested by Boris

Signed-off-by: Eric Anholt <eric@anholt.net>
Fixes: 98830d91da08 ("drm/vc4: Add T-format scanout support.")
Link: https://patchwork.freedesktop.org/patch/msgid/20170927193209.11870-1-eric@anholt.net
Reviewed-by: Boris Brezillon <boris.brezillon@free-electrons.com>
7 years agoMerge branch 'etnaviv/next' of https://git.pengutronix.de/git/lst/linux into drm...
Dave Airlie [Fri, 13 Oct 2017 23:39:56 +0000 (09:39 +1000)]
Merge branch 'etnaviv/next' of https://git.pengutronix.de/git/lst/linux into drm-next

Most notable addition this time is the support for the GPU performance
counters by Christian. This has been in the making for some time and it
has matured a lot. Since this is adding UAPI, the corresponding WIP
userspace can be found at [1] mesa/libdrm repos. I expect that
Christian sends out the final userspace patches for this once you have
pulled the kernel bits.

Philipp optimized the probe path, so etnaviv gets out of the way for
systems that want to boot real quick.

I've done mostly cleanups, disentangling etnaviv from the IOMMU API,
with some MMUv1 optimizations on the way.

* 'etnaviv/next' of https://git.pengutronix.de/git/lst/linux: (36 commits)
  drm/etnaviv: remove unnecessary clock stabilization delay
  drm/etnaviv: reduce reset delay
  drm/etnaviv: remove unused function etnaviv_gem_new
  drm/etnaviv: remove stale comment
  drm/etnaviv: submit supports performance monitor requests
  drm/etnaviv: enable debug registers on demand
  drm/etnaviv: need to disable clock gating when doing profiling
  drm/etnaviv: add MC perf domain
  drm/etnaviv: add TX perf domain
  drm/etnaviv: add RA perf domain
  drm/etnaviv: add SE perf domain
  drm/etnaviv: add PA perf domain
  drm/etnaviv: add SH perf domain
  drm/etnaviv: add PE perf domain
  drm/etnaviv: add HI perf domain
  drm/etnaviv: use 'sync points' for performance monitor requests
  drm/etnaviv: clear alloced event
  drm/etnaviv: add 'sync point' support
  drm/etnaviv: add performance monitor request processing
  drm/etnaviv: copy pmrs from userspace
  ...

7 years agodrm/vc4: Move the DSI clock divider workaround closer to the clock call.
Eric Anholt [Tue, 15 Aug 2017 23:47:18 +0000 (16:47 -0700)]
drm/vc4: Move the DSI clock divider workaround closer to the clock call.

We want the adjusted_mode->clock to be the actual clock we're
expecting to program, so that consumers see the right values for clock
and vrefresh.

Signed-off-by: Eric Anholt <eric@anholt.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20170815234722.20700-1-eric@anholt.net
Reviewed-by: Boris Brezillon <boris.brezillon@free-electrons.com>