drm/lima: fix recovering from PLBU out of memory

It looks like on PLBU_OUT_OF_MEM interrupt we need to resume from where we
stopped, i.e. new PLBU heap start is old end. Also update end address
in GP frame to grow heap on 2nd and subsequent out of memory interrupts.

Fixes: 2081e8dcf1ee ("drm/lima: recover task by enlarging heap buffer")
Signed-off-by: Vasily Khoruzhick <anarsoul@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200215035026.3180698-1-anarsoul@gmail.com

drm/pl111: Support Integrator IM-PD1 module

The last in-kernel user of the old framebuffer driver is the
IM-PD1 module for the Integrator/AP. Let's implement support for
this remaining user so we can migrate the last user over to
DRM and delete the old FB driver.

On the Integrator/AP the IM-PD1 system controller will exist
alongside the common Integrator system controller so make
sure to do a special lookup for the IM-PD1 syscon and make it
take precedence if found.

Tested on the Integrator/AP with the IM-PD1 mounted.

Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20200213124833.35545-1-linus.walleij@linaro.org

drm: drop unused drm_display_mode.private

drm_display_mode.private was only referenced in one place where
is was copied but never assigned.

Drop the copy and drop the field in drm_display_mode.
Adjust the comment of private_flags as is referred to the comment for
private.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Sean Paul <seanpaul@chromium.org>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200215183503.GA17310@ravnborg.org

drm: drop unused drm_crtc callback

struct drm_encoder_helper_funcs included a callback
named drm_crtc.

There are no users left - so drop it.
There was one reference in drm_crtc_helper.c,
which checked if the value was not NULL.
As it was never assigned this check could be dropped.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20200215173342.GA7458@ravnborg.org

drm/print: clean up RATELIMITED macros

Drop a few indirections, making the code simpler.
This also drops a RATELIMITED variant that is not in use.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200214175919.GA14492@ravnborg.org

drm/atomic-helper: fix kerneldoc

Just a tiny copypasta mistake.

Fixes: 751465913f04 ("drm/bridge: Add a drm_bridge_state object")
Cc: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Boris Brezillon <boris.brezillon@collabora.com>
Cc: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200214081340.2772853-1-daniel.vetter@ffwll.ch

drm/edid: Add a FIXME about DispID CEA data block revision

I don't understand what the DispID CEA data block revision
means. The spec doesn't say. I guess some DispID must have
a value of >= 3 in there or else we generally wouldn't
even parse the CEA data blocks. Or does all this code
actually not do anything?

Cc: Andres Rodriguez <andresx7@gmail.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200124200231.10517-6-ville.syrjala@linux.intel.com
Reviewed-by: Andres Rodriguez <andresx7@gmail.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>

drm/edid: Document why we don't bounds check the DispID CEA block start/end

After much head scratching I managed to convince myself that
for_each_displayid_db() has already done the bounds checks for
the DispID CEA data block. Which is why we don't need to repeat
them in cea_db_offsets(). To avoid having to go through that
pain again in the future add a comment which explains this fact.

Cc: Andres Rodriguez <andresx7@gmail.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200124200231.10517-5-ville.syrjala@linux.intel.com
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>

drm/edid: Clear out spurious whitespace

Nuke some whitespace that shouldn't be there.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200124200231.10517-4-ville.syrjala@linux.intel.com
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>

drm/edid: Introduce is_detailed_timing_descritor()

Let's introduce is_detailed_timing_descritor() as the opposite
counterpart of is_display_descriptor().

Cc: Allen Chen <allen.chen@ite.com.tw>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200124200231.10517-3-ville.syrjala@linux.intel.com
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>

drm/edid: Don't accept any old garbage as a display descriptor

Currently we assume any 18 byte descriptor to be a display descritor
if only the tag byte matches the expected value. But for detailed
timing descriptors that same byte is just the lower 8 bits of
hblank, and as such can match any display descriptor tag. To
properly validate that the 18 byte descriptor is in fact a
display descriptor we must also examine bytes 0-2 (just byte 1
should actually suffice but the spec does say that bytes 0 and
2 must also always be zero for display descriptors so we check
those too).

Unlike Allen's original proposed patch to just fix is_rb() we
roll this out across the board to fix everything.

Cc: Allen Chen <allen.chen@ite.com.tw>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200124200231.10517-2-ville.syrjala@linux.intel.com
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>

drm/edid: Check the number of detailed timing descriptors in the CEA ext block

CEA-861 says :
"d = offset for the byte following the reserved data block.
 If no data is provided in the reserved data block, then d=4.
 If no DTDs are provided, then d=0."

So let's not look for DTDs when d==0. In fact let's just make that
<4 since those values would just mean that he DTDs overlap the block
header. And let's also check that d isn't so big as to declare
the descriptors to live past the block end, although the code
does already survive that case as we'd just end up with a negative
number of descriptors and the loop would not do anything.

Cc: Allen Chen <allen.chen@ite.com.tw>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200124200231.10517-1-ville.syrjala@linux.intel.com
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>

drm/mcde: Fix Sphinx formatting

- Format the pipe diagram as a monospace block.
- Fix formatting of the list. Without the empty line, the first dash is
  not parsed as a bullet point.

Signed-off-by: Jonathan Neuschäfer <j.neuschaefer@gmx.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20200214163815.25442-1-j.neuschaefer@gmx.net

drm/print: Delete a few unused shouting macros

We want to go over to the new lowercase ones, encourage that a bit
more.

v2: Remove the accidentally included hunk from some WIP branch this
was based on (Jani&Sam).

Cc: Jani Nikula <jani.nikula@intel.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200214090428.2929833-1-daniel.vetter@ffwll.ch

drm/sun4i: dsi: Remove incorrect use of runtime PM

The driver currently uses runtime PM to perform some of the module
initialization and cleanup. This has three problems:

1) There is no Kconfig dependency on CONFIG_PM, so if runtime PM is
   disabled, the driver will not work at all, since the module will
   never be initialized.

2) The driver does not ensure that the device is suspended when
   sun6i_dsi_probe() fails or when sun6i_dsi_remove() is called. It
   simply disables runtime PM. From the docs of pm_runtime_disable():

      The device can be either active or suspended after its runtime PM
      has been disabled.

   And indeed, the device will likely still be active if sun6i_dsi_probe
   fails. For example, if the panel driver is not yet loaded, we have
   the following sequence:

   sun6i_dsi_probe()
      pm_runtime_enable()
      mipi_dsi_host_register()
         of_mipi_dsi_device_add(child)
            ...device_add()...
               __device_attach()
                 pm_runtime_get_sync(dev->parent) -> Causes resume
                 bus_for_each_drv()
                    __device_attach_driver() -> No match for panel
                 pm_runtime_put(dev->parent) -> Async idle request
      component_add()
         __component_add()
            try_to_bring_up_masters()
               try_to_bring_up_master()
                  sun4i_drv_bind()
                     component_bind_all()
                        component_bind()
                           sun6i_dsi_bind() -> Fails with -EPROBE_DEFER
      mipi_dsi_host_unregister()
      pm_runtime_disable()
         __pm_runtime_disable()
            __pm_runtime_barrier() -> Idle request is still pending
               cancel_work_sync()  -> DSI host is *not* suspended!

   Since the device is not suspended, the clock and regulator are never
   disabled. The imbalance causes a WARN at devres free time.

3) The driver relies on being suspended when sun6i_dsi_encoder_enable()
   is called. The resume callback has a comment that says:

      Some part of it can only be done once we get a number of
      lanes, see sun6i_dsi_inst_init

   And then part of the resume callback only runs if dsi->device is not
   NULL (that is, if sun6i_dsi_attach() has been called). However, as
   the above call graph shows, the resume callback is guaranteed to be
   called before sun6i_dsi_attach(); it is called before child devices
   get their drivers attached.

   Therefore, part of the controller initialization will only run if the
   device is suspended between the calls to mipi_dsi_host_register() and
   component_add() (which ends up calling sun6i_dsi_encoder_enable()).
   Again, as shown by the above call graph, this is not the case. It
   appears that the controller happens to work because it is still
   initialized by the bootloader.

   Because the connector is hardcoded to always be connected, the
   device's runtime PM reference is not dropped until system suspend,
   when sun4i_drv_drm_sys_suspend() ends up calling
   sun6i_dsi_encoder_disable(). However, that is done as a system sleep
   PM hook, and at that point the system PM core has already taken
   another runtime PM reference, so sun6i_dsi_runtime_suspend() is
   not called. Likewise, by the time the PM core releases its reference,
   sun4i_drv_drm_sys_resume() has already re-enabled the encoder.

   So after system suspend and resume, we have *still never called*
   sun6i_dsi_inst_init(), and now that the rest of the display pipeline
   has been reset, the DSI host is unable to communicate with the panel,
   causing VBLANK timeouts.

Fix all of these issues by inlining the runtime PM hooks into the
encoder enable/disable functions, which are guaranteed to run after a
panel is attached. This allows sun6i_dsi_inst_init() to be called
unconditionally. Furthermore, this causes the hardware to be turned off
during system suspend and reinitialized on resume, which was not
happening before.

Fixes: 133add5b5ad4 ("drm/sun4i: Add Allwinner A31 MIPI-DSI controller support")
Signed-off-by: Samuel Holland <samuel@sholland.org>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20200211072858.30784-4-samuel@sholland.org

drm/sun4i: dsi: Allow binding the host without a panel

Currently, the DSI host blocks binding the display pipeline until the
panel is available. This unnecessarily prevents other display outpus
from working, and adds logspam to dmesg when the panel driver is built
as a module (the component master is unsuccessfully brought up several
times during boot).

Flip the dependency, instead requiring the host to be bound before the
panel is attached. The panel driver provides no functionality outside of
the display pipeline anyway.

Since the panel is now probed after the DRM connector, we need a hotplug
event to turn on the connector after the panel is attached.

This has the added benefit of fixing panel module removal/insertion.
Previously, the panel would be turned off when its module was removed.
But because the connector state was hardcoded, nothing knew to turn the
panel back on when it was re-attached. Now, with hotplug events
available, the connector state will follow the panel module state, and
the panel will be re-enabled properly.

Fixes: 133add5b5ad4 ("drm/sun4i: Add Allwinner A31 MIPI-DSI controller support")
Signed-off-by: Samuel Holland <samuel@sholland.org>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20200211072858.30784-3-samuel@sholland.org

drm/sun4i: dsi: Use NULL to signify "no panel"

The continued use of an ERR_PTR to signify "no panel" outside of
sun6i_dsi_attach is confusing because it is a double negative. Because
the connector always reports itself as connected, there is also the
possibility of sending an ERR_PTR to drm_panel_get_modes(), which would
crash.

Solve both of these by only storing the panel pointer if it is valid.

Fixes: 133add5b5ad4 ("drm/sun4i: Add Allwinner A31 MIPI-DSI controller support")
Signed-off-by: Samuel Holland <samuel@sholland.org>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20200211072858.30784-2-samuel@sholland.org

drm/sun4i: dsi: Remove unused drv from driver context

This member is never used, so remove it.

Fixes: 133add5b5ad4 ("drm/sun4i: Add Allwinner A31 MIPI-DSI controller support")
Signed-off-by: Samuel Holland <samuel@sholland.org>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20200211072858.30784-1-samuel@sholland.org

drm/virtio: fix error check

The >= compare op must happen in cpu byte order, doing it in
little endian fails on big endian machines like s390.

Reported-by: Sebastian Mitterle <smitterl@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20200214080100.1273-1-kraxel@redhat.com

drm: Add support for DP 1.4 Compliance edid corruption test

Unlike DP 1.2 edid corruption test, DP 1.4 requires to calculate
real CRC value of the last edid data block, and write it back.
Current edid CRC calculates routine adds the last CRC byte,
and check if non-zero.

This behavior is not accurate; actually, we need to return
the actual CRC value when corruption is detected.
This commit changes this issue by returning the calculated CRC,
and initiate the required sequence.

Change since v7
- Fix for CI.CHECKPATCH

Change since v6
- Add return check

Change since v5
- Obtain real CRC value before dumping bad edid

Change since v4
- Fix for CI.CHECKPATCH

Change since v3
- Fix a minor typo.

Change since v2
- Rewrite checksum computation routine to avoid duplicated code.
- Rename to avoid confusion.

Change since v1
- Have separate routine for returning real CRC.

Signed-off-by: Jerry (Fangzhi) Zuo <Jerry.Zuo@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200211160832.24259-1-Jerry.Zuo@amd.com

drm: Remove legacy version of get_scanout_position()

The legacy version of get_scanout_position() was only useful while
drivers still used drm_driver.get_scanout_position(). With no such
drivers left, the related typedef and code can be removed

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-23-tzimmermann@suse.de

drm: Clean-up VBLANK-related callbacks in struct drm_driver

All non-legacy users of VBLANK functions in struct drm_driver have been
converted to use the respective interfaces in struct drm_crtc_funcs. The
remaining users of VBLANK callbacks in struct drm_driver are legacy drivers
with userspace modesetting.

All users of struct drm_driver.get_scanout_position() have been
converted to the respective CRTC helper function. Remove the callback
from struct drm_driver.

There are no users left of get_vblank_timestamp(), so the callback is
being removed. The other VBLANK callbacks are being moved to the legacy
section at the end of struct drm_driver.

Also removed is drm_calc_vbltimestamp_from_scanoutpos(). Callers of this
function have been converted to use the CRTC instead.

v4:
* more readable code for setting high_prec (Ville, Jani)
v2:
* merge with removal of struct drm_driver.get_scanout_position()
* remove drm_calc_vbltimestamp_from_scanoutpos()

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Tested-by: Yannick Fertré <yannick.fertre@st.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-22-tzimmermann@suse.de

drm/vmwgfx: Convert to CRTC VBLANK callbacks

VBLANK callbacks in struct drm_driver are deprecated in favor of
their equivalents in struct drm_crtc_funcs. Convert vmwgfx over.

v2:
* remove accidental whitespace fixes

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Thomas Hellstrom <thellstrom@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-21-tzimmermann@suse.de

drm/vkms: Convert to CRTC VBLANK callbacks

VBLANK callbacks in struct drm_driver are deprecated in favor of
their equivalents in struct drm_crtc_funcs. Convert vkms over.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Rodrigo Siqueira <rodrigosiqueira@gmail.com>
Tested-by: Rodrigo Siqueira <rodrigosiqueira@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-20-tzimmermann@suse.de

drm/vc4: Convert to CRTC VBLANK callbacks

VBLANK callbacks in struct drm_driver are deprecated in favor of
their equivalents in struct drm_crtc_funcs. Convert vc4 over.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Eric Anholt <eric@anholt.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-19-tzimmermann@suse.de

drm/vc4: Convert to struct drm_crtc_helper_funcs.get_scanout_position()

The callback struct drm_driver.get_scanout_position() is deprecated in
favor of struct drm_crtc_helper_funcs.get_scanout_position(). Convert vc4
over.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Eric Anholt <eric@anholt.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-18-tzimmermann@suse.de

drm/sti: Convert to CRTC VBLANK callbacks

VBLANK callbacks in struct drm_driver are deprecated in favor of
their equivalents in struct drm_crtc_funcs. Convert sti over.

v2:
* remove unnecessary include of sti_crtc.h from sti_drv.c

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Benjamin Gaignard <benjamin.gaignard@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-17-tzimmermann@suse.de

drm/stm: Convert to CRTC VBLANK callbacks

VBLANK callbacks in struct drm_driver are deprecated in favor of
their equivalents in struct drm_crtc_funcs. Convert stm over.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Tested-by: Yannick Fertré <yannick.fertre@st.com>
Acked-by: Philippe Cornu <philippe.cornu@st.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-16-tzimmermann@suse.de

drm/stm: Convert to struct drm_crtc_helper_funcs.get_scanout_position()

The callback struct drm_driver.get_scanout_position() is deprecated in
favor of struct drm_crtc_helper_funcs.get_scanout_position(). Convert stm
over.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Philippe Cornu <philippe.cornu@st.com>
Tested-by: Yannick Fertré <yannick.fertre@st.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-15-tzimmermann@suse.de

drm/msm: Convert to CRTC VBLANK callbacks

VBLANK callbacks in struct drm_driver are deprecated in favor of
their equivalents in struct drm_crtc_funcs. Convert msm over.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Tested-by: Yannick Fertré <yannick.fertre@st.com>
Acked-by: Rob Clark <robdclark@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-14-tzimmermann@suse.de

drm/msm: Convert to struct drm_crtc_helper_funcs.get_scanout_position()

The callback struct drm_driver.get_scanout_position() is deprecated in
favor of struct drm_crtc_helper_funcs.get_scanout_position(). Convert
msm over.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Rob Clark <robdclark@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-13-tzimmermann@suse.de

drm/radeon: Convert to CRTC VBLANK callbacks

VBLANK callbacks in struct drm_driver are deprecated in favor of
their equivalents in struct drm_crtc_funcs. Convert radeon over.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-12-tzimmermann@suse.de

drm/radeon: Convert to struct drm_crtc_helper_funcs.get_scanout_position()

The callback struct drm_driver.get_scanout_position() is deprecated in
favor of struct drm_crtc_helper_funcs.get_scanout_position(). Convert
radeon over.

v4:
* 80-character line fixes

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-11-tzimmermann@suse.de

drm/nouveau: Convert to CRTC VBLANK callbacks

VBLANK callbacks in struct drm_driver are deprecated in favor of
their equivalents in struct drm_crtc_funcs. Convert nouvean over.

v4:
* add argument names in function declaration

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-10-tzimmermann@suse.de

drm/nouveau: Convert to struct drm_crtc_helper_funcs.get_scanout_position()

The callback struct drm_driver.get_scanout_position() is deprecated in
favor of struct drm_crtc_helper_funcs.get_scanout_position(). Convert
nouveau over.

v4:
* add argument names in function declaration

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Ben Skeggs <bskeggs@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-9-tzimmermann@suse.de

drm/i915: Convert to CRTC VBLANK callbacks

VBLANK callbacks in struct drm_driver are deprecated in favor of their
equivalents in struct drm_crtc_funcs. Convert i915 over.

The callback struct drm_driver.get_scanout_position() is deprecated
in favor of struct drm_crtc_helper_funcs.get_scanout_position().
i915 doesn't use CRTC helpers. Instead pass i915's implementation of
get_scanout_position() to DRM core's
drm_crtc_vblank_helper_get_vblank_timestamp_internal().

v3:
* rename dcrtc to _crtc
* use intel_ prefix for i915_crtc_get_vblank_timestamp()
* update for drm_crtc_vblank_helper_get_vblank_timestamp_internal()
v2:
* use DRM's implementation of get_vblank_timestamp()
* simplify function names

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-8-tzimmermann@suse.de

drm/gma500: Convert to CRTC VBLANK callbacks

VBLANK callbacks in struct drm_driver are deprecated in favor of
their equivalents in struct drm_crtc_funcs. Convert gma500 over.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-7-tzimmermann@suse.de

drm/amdgpu: Convert to CRTC VBLANK callbacks

VBLANK callbacks in struct drm_driver are deprecated in favor of
their equivalents in struct drm_crtc_funcs. Convert amdgpu over.

v2:
* don't wrap existing functions; change signature instead

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-6-tzimmermann@suse.de

drm/amdgpu: Convert to struct drm_crtc_helper_funcs.get_scanout_position()

The callback struct drm_driver.get_scanout_position() is deprecated in
favor of struct drm_crtc_helper_funcs.get_scanout_position(). Convert
amdgpu over.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-5-tzimmermann@suse.de

drm: Add get_vblank_timestamp() to struct drm_crtc_funcs

The callback get_vblank_timestamp() is currently located in struct
drm_driver, but really belongs into struct drm_crtc_funcs. Add an
equivalent there. Driver will be converted in separate patches.

The default implementation is drm_calc_vbltimestamp_from_scanoutpos().
The patch adds drm_crtc_vblank_helper_get_vblank_timestamp(), which is
an implementation for the CRTC callback.

v4:
* more readable code for setting high_prec (Ville, Jani)
v3:
* use refactored timestamp calculation to minimize duplicated code
* do more checks for crtc != NULL to support legacy drivers
v2:
* rename helper to drm_crtc_vblank_helper_get_vblank_timestamp()
* replace drm_calc_vbltimestamp_from_scanoutpos() with
  drm_crtc_vblank_helper_get_vblank_timestamp() in docs

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-4-tzimmermann@suse.de

drm: Add get_scanout_position() to struct drm_crtc_helper_funcs

The new callback get_scanout_position() reads the current location
of the scanout process. The operation is currently located in struct
drm_driver, but really belongs to the CRTC. Drivers will be converted
in separate patches.

To help with the conversion, the timestamp calculation has been
moved from drm_calc_vbltimestamp_from_scanoutpos() to
drm_crtc_vblank_helper_get_vblank_timestamp_internal(). The helper
function supports the new and old interface of get_scanout_position().
drm_calc_vbltimestamp_from_scanoutpos() remains as a wrapper around
the new function.

Callback functions return the scanout position from the CRTC. The
legacy version of the interface receives the device and pipe index,
the modern version receives a pointer to the CRTC. We keep the
legacy version until all drivers have been converted.

v4:
* 80-character line fixes
v3:
* refactor drm_calc_vbltimestamp_from_scanoutpos() to minimize
  code duplication
* define types for get_scanout_position() callbacks
v2:
* fix logical op in drm_calc_vbltimestamp_from_scanoutpos()

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Tested-by: Yannick Fertré <yannick.fertre@st.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-3-tzimmermann@suse.de

drm: Remove internal setup of struct drm_device.vblank_disable_immediate

VBLANK interrupts can be disabled immediately or with a delay, where the
latter is the default. The former option can be selected by setting
get_vblank_timestamp and enabling vblank_disable_immediate in struct
drm_device. Simplify the code in preparation of the removal of struct
drm_device.get_vblank_timestamp.

v3:
* remove internal setup of vblank_disable_immediate

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200123135943.24140-2-tzimmermann@suse.de

drm/virtio: fix virtio_gpu_cursor_plane_update().

Add missing virtio_gpu_array_lock_resv() call.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20200211135047.22261-3-kraxel@redhat.com

drm/virtio: fix virtio_gpu_execbuffer_ioctl locking

Lockdep says we can't call vmemdup() while having objects reserved
because it needs the mmap semaphore.  So reorder the calls reserve
the objects later.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20200211135047.22261-2-kraxel@redhat.com

drm/qxl: replace zero-length array with flexible-array member

The current codebase makes use of the zero-length array language
extension to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:

struct foo {
        int stuff;
        struct boo array[];
};

By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertenly introduced[3] to the codebase from now on.

Also, notice that, dynamic memory allocations won't be affected by
this change:

"Flexible array members have incomplete type, and so the sizeof operator
may not be applied. As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]

This issue was found with the help of Coccinelle.

[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 76497732932f ("cxgb3/l2t: Fix undefined behaviour")

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20200212193344.GA27929@embeddedor
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm/hisilicon/hibmc: add gamma_set function

Add gamma_set function, and we can also use it to adjust the brightness
of the display.

Signed-off-by: Zhihui Chen <chenzhihui4@huawei.com>
Signed-off-by: Xinliang Liu <xinliang.liu@linaro.org>
Acked-by: Xinliang Liu <xinliang.liu@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20191223074910.1030-1-chenzhihui4@huawei.com

drm/hisilicon/hibmc: add DPMS on/off function

add DPMS function to turn on/off signal of monitor

Signed-off-by: Zhihui Chen <chenzhihui4@huawei.com>
Signed-off-by: Xinliang Liu <xinliang.liu@linaro.org>
Acked-by: Xinliang Liu <xinliang.liu@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20191220023112.2728-1-chenzhihui4@huawei.com

drm/hisilicon/hibmc: fix 'xset dpms force off' fail

both crtc_state->adjusted_mode.hdisplay and
crtc_state->adjusted_mode.vdisplay are 0 when switch dpms off,
return -EINVAL cause switch dpms off fail.

Signed-off-by: Zhihui Chen <chenzhihui4@huawei.com>
Signed-off-by: Xinliang Liu <xinliang.liu@linaro.org>
Acked-by: Xinliang Liu <xinliang.liu@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20191220023004.2658-1-chenzhihui4@huawei.com

drm/bridge: ti-sn65dsi86: Avoid invalid rates

Based on work by Bjorn Andersson <bjorn.andersson@linaro.org>,
Jeffrey Hugo <jeffrey.l.hugo@gmail.com>, and
Rob Clark <robdclark@chromium.org>.

Let's read the SUPPORTED_LINK_RATES and/or MAX_LINK_RATE (depending on
the eDP version of the sink) to figure out what eDP rates are
supported and pick the ideal one.

NOTE: I have only personally tested this code on eDP panels that are
1.3 or older.  Code reading SUPPORTED_LINK_RATES for DP 1.4+ was
tested by hacking the code to pretend that a table was there.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191218143416.v3.9.Ib59207b66db377380d13748752d6fce5596462c5@changeid

drm/bridge: ti-sn65dsi86: Train at faster rates if slower ones fail

If we fail training at a lower DP link rate let's now keep trying
until we run out of rates to try.  Basically the algorithm here is to
start at the link rate that is the theoretical minimum and then slowly
bump up until we run out of rates or hit the max rate of the sink.  We
query the sink using a DPCD read.

This is, in fact, important in practice.  Specifically at least one
panel hooked up to the bridge (AUO B116XAK01) had a theoretical min
rate more than 1.62 GHz (if run at 24 bpp) and fails to train at the
next rate (2.16 GHz).  It would train at 2.7 GHz, though.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Tested-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191218143416.v3.8.I251add713bc5c97225200894ab110ea9183434fd@changeid

drm/bridge: ti-sn65dsi86: Group DP link training bits in a function

We'll re-organize the ti_sn_bridge_enable() function a bit to group
together all the parts relating to link training and split them into a
sub-function.  This is not intended to have any functional change and
is in preparation for trying link training several times at different
rates.  One small side effect here is that if link training fails
we'll now leave the DP PLL disabled, but that seems like a sane thing
to do.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Tested-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191218143416.v3.7.I1fc75ad11db9048ef08cfe1ab7322753d9a219c7@changeid

drm/bridge: ti-sn65dsi86: Use 18-bit DP if we can

The current bridge driver always forced us to use 24 bits per pixel
over the DP link.  This is a waste if you are hooked up to a panel
that only supports 6 bits per color or fewer, since in that case you
can run at 18 bits per pixel and thus end up at a lower DP clock rate.

Let's support this.

While at it, let's clean up the math in the function to avoid rounding
errors (and round in the correct direction when we have to round).
Numbers are sufficiently small (because mode->clock is in kHz) that we
don't need to worry about integer overflow.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Tested-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
[narmstrong: s/ran/can/]
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191218143416.v3.6.Iaf8d698f4e5253d658ae283d2fd07268076a7c27@changeid

drm/bridge: ti-sn65dsi86: Read num lanes from the DP sink

At least one panel hooked up to the bridge (AUO B116XAK01) only
supports 1 lane of DP.  Let's read this information and stop
hardcoding 4 DP lanes.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Tested-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191218143416.v3.5.Idbd0051d0de53f7e9d18a291ea33011c0854fcc6@changeid

drm/bridge: ti-sn65dsi86: Config number of DP lanes Mo' Betta

The driver used to say that the value to program into bridge register
0x93 was dp_lanes - 1.  Looking at the datasheet for the bridge, this
is wrong.  The data sheet says:
* 1 = 1 lane
* 2 = 2 lanes
* 3 = 4 lanes

A more proper way to express this encoding is min(dp_lanes, 3).

At the moment this change has zero effect because we've hardcoded the
number of DP lanes to 4.  ...and (4 - 1) == min(4, 3).  How fortunate!
...but soon we'll stop hardcoding the number of lanes.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Tested-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191218143416.v3.4.If3e2d0493e7b6e8b510ea90d8724ff760379b3ba@changeid

drm/bridge: ti-sn65dsi86: Don't use MIPI variables for DP link

The ti-sn65dsi86 is a bridge from MIPI to DP and thus has two links:
the MIPI link and the DP link.  The two links do not need to have the
same format or number of lanes.  Stop using MIPI variables when
talking about the DP link.

This has zero functional change because:
* currently we are hardcoding the MIPI link as unpacked RGB888 which
  requires 24 bits and currently we are not changing the DP link rate
  from the bridge's default of 8 bits per pixel.
* currently we are hardcoding both the MIPI and DP as being 4 lanes.

This is all in prep for fixing some of the above.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Tested-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191218143416.v3.3.Ia6e05f4961adb0d4a0d32ba769dd7781ee8db431@changeid

drm/bridge: ti-sn65dsi86: zero is never greater than an unsigned int

When we iterate over ti_sn_bridge_dp_rate_lut, there's no reason to
start at index 0 which always contains the value 0.  0 is not a valid
link rate.

This change should have no real effect but is a small cleanup.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Tested-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191218143416.v3.2.Id445d0057bedcb0a190009e0706e9254c2fd48eb@changeid

drm/bridge: ti-sn65dsi86: Split the setting of the dp and dsi rates

These two things were in one function.  Split into two.  This looks
like it's duplicating some code, but don't worry.  This is is just in
preparation for future changes.

This is intended to have zero functional change and will just make
future patches easier to understand.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Tested-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191218143416.v3.1.Icb765d5799e9651e5249c0c27627ba33a9e411cf@changeid

drm/ttm: individualize resv objects before calling release_notify

This allows release_notify to add and remove fences from deleted objects.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: xinhui pan <xinhui.pan@amd.com>
Link: https://patchwork.freedesktop.org/patch/352750/

drm/ttm: replace dma_resv object on deleted BOs v3

When non-imported BOs are resurrected for delayed delete we replace
the dma_resv object to allow for easy reclaiming of the resources.

v2: move that to ttm_bo_individualize_resv
v3: add a comment to explain what's going on

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: xinhui pan <xinhui.pan@amd.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/352927/

drm/ttm: rework BO delayed delete. v2

This patch reworks the whole delayed deletion of BOs which aren't idle.

Instead of having two counters for the BO structure we resurrect the BO
when we find that a deleted BO is not idle yet.

This has many advantages, especially that we don't need to
increment/decrement the BOs reference counter any more when it
moves on the LRUs.

v2: remove duplicate ttm_tt_destroy, fix holde lock for LRU move

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: xinhui pan <xinhui.pan@amd.com>
Link: https://patchwork.freedesktop.org/patch/352912/

drm/virtio: add drm_driver.release callback.

Split virtio_gpu_deinit(), move the drm shutdown and release to
virtio_gpu_release().  Drop vqs_ready variable, instead use
drm_dev_{enter,exit,unplug} to avoid touching hardware after
device removal.  Tidy up here and there.

v4: add changelog.
v3: use drm_dev_*().

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20200211135805.24436-1-kraxel@redhat.com

drm/cirrus: add drm_driver.release callback.

Move final cleanups from cirrus_pci_remove() to the new callback.
Add drm_atomic_helper_shutdown() call to cirrus_pci_remove().

Use drm_dev_{enter,exit,unplug} to avoid touching hardware after
device removal.

v4: add changelog.
v3: use drm_dev*.
v2: stop touching hardware after pci_remove().

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20200211135522.23654-1-kraxel@redhat.com

drm/bochs: add drm_driver.release callback.

Call bochs_unload via drm_driver.release to make sure we release stuff
when it is safe to do so.  Use drm_dev_{enter,exit,unplug} to avoid
touching hardware after device removal.  Tidy up here and there.

v4: add changelog.
v3: use drm_dev_*().
v2: move hardware deinit to pci_remove().

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20200211135218.22871-1-kraxel@redhat.com

drm/irq: remove check on dev->dev_private

There is no real reason to require drivers to set and use
dev->dev_private. Indeed, the current recommendation, as documented in
drm_device.h, is to embed struct drm_device in the per-device struct
instead of using dev_private.

Remove the requirement for dev_private to have been set to indicate
driver initialization.

For background, quoting Daniel Vetter:

Now there might be some hilarious races this papers over, but:

- Proper drivers should only call drm_dev_register once everything is
  set up, including this stuff here. No race possible with anything else
  really.

- Slightly more wobbly drivers, including the legacy ones, all use
  drm_global_mutex. This was the former BKL, which means that it was
  impossible for soeone to go through the load/unload/reload (between
  lastclose and firstopen) paths and also run the ioctl. But the ioctl
  had to be made unlocked because blocking there killed X:

 commit 8f4ff2b06afcd6f151868474a432c603057eaf56
 Author: Ilija Hadzic <ihadzic@research.bell-labs.com>
 Date:   Mon Oct 31 17:46:18 2011 -0400

     drm: do not sleep on vblank while holding a mutex

  The even more legacy DRM_CONTROL ioctl stayed fully locked. But the
  file open/close paths are still fully locked, and that's the only
  place legacy drivers should call drm_irq_install/uninstall, so should
  all still be fully ordered and protected and happy.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel@ffwll.ch>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200211144753.3175-1-jani.nikula@intel.com

drm/panel: simple: Add Rocktech RK101II01D-CT panel

Add support for Rocktech RK101II01D-CT, 10.1" 1280x800 TFT with LVDS
interface, LED backlight and integrated Goodix GT928 capacitive touch
panel.

Signed-off-by: Jyri Sarha <jsarha@ti.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/b543b77d8af7cbbef852d3df4595f11ac1aa0046.1581423249.git.jsarha@ti.com

dt-bindings: panel-simple: Add rocktech,rk101ii01d-ct compatible

Add compatible to panel-simple for Rocktech Displays Limited
RK101II01D-CT 10.1" TFT 1280x800 Pixels with LVDS interface, LED
Backlight, and capacitive touch panel ("goodix,gt928" compatible).

Signed-off-by: Jyri Sarha <jsarha@ti.com>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/2ffedf45e65943a021ed57a0d174900e43663b27.1581423249.git.jsarha@ti.com

drm: Nerf drm_global_mutex BKL for good drivers

This catches the majority of drivers (unfortunately not if we take
users into account, because all the big drivers have at least a
lastclose hook).

With the prep patches out of the way all drm state is fully protected
and either prevents or can deal with the races from dropping the BKL
around open/close. The only thing left to audit are the various driver
hooks - by keeping the BKL around if any of them are set we have a
very simple cop-out!

Note that one of the biggest prep pieces to get here was making
dev->open_count atomic, which was done in

commit 7e13ad896484a0165a68197a2e64091ea28c9602
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Fri Jan 24 13:01:07 2020 +0000

    drm: Avoid drm_global_mutex for simple inc/dec of dev->open_count

v2:
- Rebase and fix locking in drm_open() (Chris)
- Indentation fix in drm_release
- Typo fix in the commit message (Sam)

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Sam Ravnborg <sam@ravnborg.org>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Tested-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200204150146.2006481-6-daniel.vetter@ffwll.ch

drm: Push drm_global_mutex locking in drm_open

We want to only take the BKL on crap drivers, but to know whether
we have a crap driver we first need to look it up. Split this shuffle
out from the main BKL-disabling patch, for more clarity. Historical
aside: When the kernel-wide BKL was removed, it was replaced by
drm_global_mutex within the scope of the drm subsystem hence why these
two things are (almost) interchangeable as concepts here.

Since the minors are refcounted drm_minor_acquire is purely internal
and this does not have a driver visible effect.

v2: Push the locking even further into drm_open(), suggested by Chris.
This gives us more symmetry with drm_release(), and maybe a futuer
avenue where we make drm_global_mutex locking (partially) opt-in like
with drm_release_noglobal().

v3:
- Actually push this stuff correctly, don't unlock twice (Chris)
- Fix typo on commit message, plus explain why BKL = drm_global_mutex
  (Sam)

Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Tested-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200204150146.2006481-5-daniel.vetter@ffwll.ch

drm/client: Rename _force to _locked

Plus extend the kerneldoc a bit to explain how this should be used.
With the previous patch to drop the force restore the main user of
this function is not emphasis on the "I hold the internal master lock
already" aspect, so rename the function to match.

Suggested by Noralf.

Cc: Noralf Trønnes <noralf@tronnes.org>
Reviewed-by: Noralf Trønnes <noralf@tronnes.org>
Tested-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200204150146.2006481-4-daniel.vetter@ffwll.ch

drm/fbdev-helper: don't force restores

Instead check for master status, in case we've raced.

This is the last exception to the general rule that we restore fbcon
only when there's no master active. Compositors are supposed to drop
their master status before they switch to a different console back to
text mode (or just switch to text mode directly, without a vt switch).

This is known to break some subtests of kms_fbcon_fbt in igt, but they're
just wrong - it does a graphics/text mode switch for the vt without
updating the master status.

Also add a comment to the drm_client->restore hook that this is expected
going forward from all clients (there's currently just one).

v2: Also drop the force in pan_display

v3: Restore the _force to pan_display, this actually means _locked in that
path. Spotted by Noralf.

Cc: Noralf Trønnes <noralf@tronnes.org>
Reviewed-by: Noralf Trønnes <noralf@tronnes.org>
Tested-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200204150146.2006481-3-daniel.vetter@ffwll.ch

drm/ttm: use RCU in ttm_bo_flush_all_fences

This allows it to call the function without the lock held.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: xinhui pan <xinhui.pan@amd.com>
Link: https://patchwork.freedesktop.org/patch/352742/

drm/ttm: cleanup ttm_buffer_object_transfer

The function is always called with deleted BOs.

While at it cleanup the indentation as well.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: xinhui pan <xinhui.pan@amd.com>
Link: https://patchwork.freedesktop.org/patch/352743/

drm/ttm: refine ghost BO resv criteria

Ghost BOs need to stick with the resv object only when the origin is imported.

This is a low hanging fruit to avoid OOM situations on evictions.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: xinhui pan <xinhui.pan@amd.com>
Link: https://patchwork.freedesktop.org/patch/352740/

drm/qxl: add drm_driver.release callback.

Move final cleanups to qxl_drm_release() callback.
Add drm_atomic_helper_shutdown() call to qxl_pci_remove().

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20200210113753.5614-3-kraxel@redhat.com

drm/qxl: reorder calls in qxl_device_fini().

Reorder calls in qxl_device_fini().  Cleaning up gem & ttm
might trigger qxl commands, so we should do that before
releaseing command rings.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20200210113753.5614-2-kraxel@redhat.com

drm/omap: fix possible object reference leak

The call to of_find_matching_node returns a node pointer with refcount
incremented thus it must be explicitly decremented after the last
usage.

Detected by coccinelle with the following warnings:
drivers/gpu/drm/omapdrm/dss/omapdss-boot-init.c:212:2-8: ERROR: missing of_node_put; acquired a node pointer with refcount incremented on line 209, but without a corresponding object release within this function.
drivers/gpu/drm/omapdrm/dss/omapdss-boot-init.c:237:1-7: ERROR: missing of_node_put; acquired a node pointer with refcount incremented on line 209, but without a corresponding object release within this function.

Signed-off-by: Wen Yang <wen.yang99@zte.com.cn>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Mukesh Ojha <mojha@codeaurora.org>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Sebastian Reichel <sebastian.reichel@collabora.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: dri-devel@lists.freedesktop.org
Cc: linux-kernel@vger.kernel.org
Cc: Markus Elfring <Markus.Elfring@web.de>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1554692313-28882-2-git-send-email-wen.yang99@zte.com.cn

drm/omap: dmm_tiler: Remove the dma_async_issue_pending() call

dma_sync_wait() is calling it so no need to call it in the driver.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190731094233.13890-3-peter.ujfalusi@ti.com

drm/omap: dmm_tiler: Use dmaengine_prep_dma_memcpy() for i878 workaround

Instead of dma_dev->device_prep_dma_memcpy() use the existing macro to
prepare the memcpy.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190731094233.13890-2-peter.ujfalusi@ti.com

drm/omapdrm: Fix trivial spelling

Fix trivial spelling identified while examining the code.

s/supprted./supported./

Signed-off-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191209123320.10186-1-kieran.bingham+renesas@ideasonboard.com

drm/omap: use true,false for bool variable

Fixes coccicheck warning:

drivers/gpu/drm/omapdrm/displays/panel-dsi-cm.c:681:1-15: WARNING: Assignment of 0/1 to bool variable
drivers/gpu/drm/omapdrm/displays/panel-dsi-cm.c:732:1-15: WARNING: Assignment of 0/1 to bool variable

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Zheng Bin <zhengbin13@huawei.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1579763123-62749-1-git-send-email-zhengbin13@huawei.com

drm/bridge: tfp410: add pclk limits

Add pixel clock limits to the driver as per TFP410 datasheet: min 25MHz,
max 165MHz.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200121094655.9092-1-tomi.valkeinen@ti.com

drm/virtio: move virtio_gpu_mem_entry initialization to new function

Introduce new virtio_gpu_object_shmem_init() helper function which will
create the virtio_gpu_mem_entry array, containing the backing storage
information for the host.  For the most path this just moves code from
virtio_gpu_object_attach().

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20200207074638.26386-5-kraxel@redhat.com
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>

drm/virtio: move mapping teardown to virtio_gpu_cleanup_object()

Stop sending DETACH_BACKING commands, that will happening anyway when
releasing resources via UNREF.  Handle guest-side cleanup in
virtio_gpu_cleanup_object(), called when the host finished processing
the UNREF command.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20200207074638.26386-4-kraxel@redhat.com
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>

drm/virtio: resource teardown tweaks

Add new virtio_gpu_cleanup_object() helper function for object cleanup.
Wire up callback function for resource unref, do cleanup from callback
when we know the host stopped using the resource.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20200207074638.26386-3-kraxel@redhat.com
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>

drm/virtio: simplify virtio_gpu_alloc_cmd

Just call virtio_gpu_alloc_cmd_resp with some fixed args
instead of duplicating most of the function body.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20200207074638.26386-2-kraxel@redhat.com
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>

drm/bochs: deinit bugfix

Check whenever mode_config was actually properly
initialized before trying to clean it up.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20200207124348.21641-1-kraxel@redhat.com

drm/bridge: panel: Fix typo in drm_panel_bridge_add docs

Fix the 'manged' typo with 'managed' in the drm_panel_bridge_add
kernel-doc documentation.

Signed-off-by: Enric Balletbo i Serra <enric.balletbo@collabora.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191218121223.30181-1-enric.balletbo@collabora.com

drm/bridge/synopsys: dsi: missing post disable

Sometime the post_disable function is missing (not registered).

Signed-off-by: Yannick Fertré <yannick.fertre@st.com>
Reviewed-by: Philippe Cornu <philippe.cornu@st.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1579602296-7683-1-git-send-email-yannick.fertre@st.com

drm/bridge: sii902x: Select SND_SOC_HDMI_CODEC if SND_SOC is configured

To enable HDMI audio the SND_SOC_HDMI_CODEC needs to be
configured. Enable HDMI audio by selecting SND_SOC_HDMI_CODEC if
SND_SOC is configured. SND_SOC_HDMI_CODEC has no config menu entry and
should be selected automatically by the drivers using it.

Signed-off-by: Jyri Sarha <jsarha@ti.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
[narmstrong: atomatically -> automatically]
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191129152342.29145-1-jsarha@ti.com

drm/bridge: Add tc358768 driver

Add basic support for the Toshiba TC358768 RGB to DSI bridge.
Not all the features of the TC358768 is implemented by the initial driver:
MIPI_DSI_MODE_VIDEO and MIPI_DSI_FMT_RGB888 is only supported and tested.

Only write is implemented for mipi_dsi_host_ops.transfer.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200131111553.472-3-peter.ujfalusi@ti.com

dt-bindings: display: bridge: Add documentation for Toshiba tc358768

TC358768/TC358778 is a Parallel RGB to MIPI DSI bridge.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200131111553.472-2-peter.ujfalusi@ti.com

dt-bindings: drm: bridge: adv7511: Add ADV7535 support

ADV7535 is a part compatible with ADV7533 but it supports 1080p@60hz and
v1p2 supply is fixed to 1.8V

Signed-off-by: Bogdan Togorean <bogdan.togorean@analog.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200121082719.27972-4-bogdan.togorean@analog.com

drm: bridge: adv7511: Add support for ADV7535

ADV7535 is a DSI to HDMI bridge chip like ADV7533 but it allows
1080p@60Hz. v1p2 is fixed to 1.8V on ADV7535.

Signed-off-by: Bogdan Togorean <bogdan.togorean@analog.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200121082719.27972-3-bogdan.togorean@analog.com

drm: bridge: adv7511: Remove DRM_I2C_ADV7533 Kconfig

This commit remove DRM_I2C_ADV7533 resulting a simpler driver and less
choices in Kconfig.

Signed-off-by: Bogdan Togorean <bogdan.togorean@analog.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200121082719.27972-2-bogdan.togorean@analog.com

drm/udl: Clear struct drm_connector_funcs.dpms

Atomic modesetting doesn't use struct drm_connector_funcs.dpms and
the set function, drm_helper_connector_dpms(), wouldn't support it
anyway. So keep the pointer to NULL.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20200207141602.4760-2-tzimmermann@suse.de

drm/bochs: Clear struct drm_connector_funcs.dpms

Atomic modesetting doesn't use struct drm_connector_funcs.dpms and
the set function, drm_helper_connector_dpms(), wouldn't support it
anyway. So keep the pointer to NULL.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20200207141602.4760-1-tzimmermann@suse.de

drm/tiny/st7735r: No need to set ->owner for spi_register_driver()

The spi_register_driver() will set the ->owner member to THIS_MODULE.

Cc: Noralf Trønnes <noralf@tronnes.org>
Cc: dri-devel@lists.freedesktop.org
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: David Lechner <david@lechnology.com>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20200131204923.48928-4-andriy.shevchenko@linux.intel.com

drm/tiny/st7735r: Make driver OF-independent

There is one OF call in the driver that limits its area of use.
Replace it to generic device_get_match_data() and get rid of OF dependency.

Cc: Noralf Trønnes <noralf@tronnes.org>
Cc: dri-devel@lists.freedesktop.org
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: David Lechner <david@lechnology.com>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20200131204923.48928-3-andriy.shevchenko@linux.intel.com

drm/tiny/repaper: No need to set ->owner for spi_register_driver()

The spi_register_driver() will set the ->owner member to THIS_MODULE.

Cc: Noralf Trønnes <noralf@tronnes.org>
Cc: dri-devel@lists.freedesktop.org
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: David Lechner <david@lechnology.com>
Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20200131204923.48928-2-andriy.shevchenko@linux.intel.com

drm/tiny/repaper: Make driver OF-independent

There is one OF call in the driver that limits its area of use.
Replace it to generic device_get_match_data() and get rid of OF dependency.

While here, cast SPI driver data to certain enumerator type.

Cc: Noralf Trønnes <noralf@tronnes.org>
Cc: dri-devel@lists.freedesktop.org
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: David Lechner <david@lechnology.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20200131204923.48928-1-andriy.shevchenko@linux.intel.com