drm/kirin: Don't set struct drm_device.irq_enabled

The field drm_device.irq_enabled is only used by legacy drivers
with userspace modesetting. Don't set it in kirin.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210625082222.3845-11-tzimmermann@suse.de

drm/exynos: Don't set struct drm_device.irq_enabled

The field drm_device.irq_enabled is only used by legacy drivers
with userspace modesetting. Don't set it in exynos.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210625082222.3845-10-tzimmermann@suse.de

drm/malidp: Don't set struct drm_device.irq_enabled

The field drm_device.irq_enabled is only used by legacy drivers
with userspace modesetting. Don't set it in malidp.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Liviu Dudau <liviu.dudau@arm.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210625082222.3845-9-tzimmermann@suse.de

drm/komeda: Don't set struct drm_device.irq_enabled

The field drm_device.irq_enabled is only used by legacy drivers
with userspace modesetting. Don't set it in komeda.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Liviu Dudau <liviu.dudau@arm.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210625082222.3845-8-tzimmermann@suse.de

drm/i915: Track IRQ state in local device state

Replace usage of struct drm_device.irq_enabled with the driver's
own state field struct drm_i915_private.irq_enabled. The field in
the DRM device structure is considered legacy and should not be
used by KMS drivers.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210625082222.3845-7-tzimmermann@suse.de

drm/armada: Don't set struct drm_device.irq_enabled

The field drm_device.irq_enabled is only used by legacy drivers
with userspace modesetting. Don't set it in armada.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210625082222.3845-6-tzimmermann@suse.de

drm: Don't test for IRQ support in VBLANK ioctls

For KMS drivers, replace the IRQ check in VBLANK ioctls with a check for
vblank support. IRQs might be enabled wthout vblanking being supported.

This change also removes the DRM framework's only dependency on IRQ state
for non-legacy drivers. For legacy drivers with userspace modesetting,
the original test remains in drm_wait_vblank_ioctl().

v4:
* avoid preprocessor ifdef in drm_wait_vblank_ioctl()
  (Jani, Thierry)
v3:
* optimize test in drm_wait_vblank_ioctl() for KMS case (Liviu)
* update docs for drm_irq_uninstall()
v2:
* keep the old test for legacy drivers in
  drm_wait_vblank_ioctl() (Daniel)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210625082222.3845-5-tzimmermann@suse.de

drm/radeon: Track IRQ state in local device state

Replace usage of struct drm_device.irq_enabled with the driver's
own state field struct radeon_device.irq.installed. The field in
the DRM device structure is considered legacy and should not be
used by KMS drivers.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210625082222.3845-4-tzimmermann@suse.de

drm/hibmc: Call drm_irq_uninstall() unconditionally

Remove the check around drm_irq_uninstall(). The same test is
done by the function internally. The tested state in irq_enabled
is considered obsolete and should not be used by KMS drivers.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Tian Tao <tiantao6@hisilicon.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210625082222.3845-3-tzimmermann@suse.de

drm/amdgpu: Track IRQ state in local device state

Replace usage of struct drm_device.irq_enabled with the driver's
own state field struct amdgpu_device.irq.installed. The field in
the DRM device structure is considered legacy and should not be
used by KMS drivers.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210625082222.3845-2-tzimmermann@suse.de

drm/sched: Declare entity idle only after HW submission

The panfrost driver tries to kill in-flight jobs on FD close after
destroying the FD scheduler entities. For this to work properly, we
need to make sure the jobs popped from the scheduler entities have
been queued at the HW level before declaring the entity idle, otherwise
we might iterate over a list that doesn't contain those jobs.

Suggested-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Cc: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210624140850.2229697-1-boris.brezillon@collabora.com

drm/tiny: drm_gem_simple_display_pipe_prepare_fb is the default

Goes through all the drivers and deletes the default hook since it's
the default now.

Acked-by: David Lechner <david@lechnology.com>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Acked-by: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Joel Stanley <joel@jms.id.au>
Cc: Andrew Jeffery <andrew@aj.id.au>
Cc: "Noralf Trønnes" <noralf@tronnes.org>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Emma Anholt <emma@anholt.net>
Cc: David Lechner <david@lechnology.com>
Cc: Kamlesh Gurudasani <kamlesh.gurudasani@gmail.com>
Cc: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Maxime Ripard <mripard@kernel.org>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: linux-aspeed@lists.ozlabs.org
Cc: linux-arm-kernel@lists.infradead.org
Cc: xen-devel@lists.xenproject.org
Link: https://patchwork.freedesktop.org/patch/msgid/20210622165511.3169559-14-daniel.vetter@ffwll.ch

drm/simple-helper: drm_gem_simple_display_pipe_prepare_fb as default

It's tedious to review this all the time, and my audit showed that
arcpgu actually forgot to set this.

Make this the default and stop worrying.

Again I sprinkled WARN_ON_ONCE on top to make sure we don't have
strange combinations of hooks: cleanup_fb without prepare_fb doesn't
make sense, and since simpler drivers are all new they better be GEM
based drivers.

v2: Warn and bail when it's _not_ a GEM driver (Noralf)

v3: It's neither ... nor, not not (Sam)

Acked-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Noralf Trønnes <noralf@tronnes.org>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <mripard@kernel.org>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210623162456.3373469-1-daniel.vetter@ffwll.ch

drm/omap: Follow implicit fencing in prepare_fb

I guess no one ever tried running omap together with lima or panfrost,
not even sure that's possible. Anyway for consistency, fix this.

Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Tomi Valkeinen <tomba@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210622165511.3169559-12-daniel.vetter@ffwll.ch

drm/vram-helpers: Create DRM_GEM_VRAM_PLANE_HELPER_FUNCS

Like we have for the shadow helpers too, and roll it out to drivers.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Tian Tao <tiantao6@hisilicon.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Hans de Goede <hdegoede@redhat.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <mripard@kernel.org>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Tian Tao <tiantao6@hisilicon.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210622165511.3169559-11-daniel.vetter@ffwll.ch

drm/armada: Remove prepare/cleanup_fb hooks

All they do is refcount the fb, which the atomic helpers already do.

This is was necessary with the legacy helpers and I guess just carry
over in the conversion. drm_plane_state always has a full reference
for its ->fb pointer during its entire lifetime,
see __drm_atomic_helper_plane_destroy_state()

Acked-by: Maxime Ripard <maxime@cerno.tech>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Russell King <linux@armlinux.org.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20210622165511.3169559-10-daniel.vetter@ffwll.ch

drm/<driver>: drm_gem_plane_helper_prepare_fb is now the default

No need to set it explicitly.

Acked-by: Philipp Zabel <p.zabel@pengutronix.de>
Acked-by: Heiko Stuebner <heiko@sntech.de>
Acked-by: Paul Cercueil <paul@crapouillou.net>
Acked-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Acked-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Acked-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Acked-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Acked-by: Philippe Cornu <philippe.cornu@foss.st.com>
Acked-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Laurentiu Palcu <laurentiu.palcu@oss.nxp.com>
Cc: Lucas Stach <l.stach@pengutronix.de>
Cc: Shawn Guo <shawnguo@kernel.org>
Cc: Sascha Hauer <s.hauer@pengutronix.de>
Cc: Pengutronix Kernel Team <kernel@pengutronix.de>
Cc: Fabio Estevam <festevam@gmail.com>
Cc: NXP Linux Team <linux-imx@nxp.com>
Cc: Philipp Zabel <p.zabel@pengutronix.de>
Cc: Paul Cercueil <paul@crapouillou.net>
Cc: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Cc: Matthias Brugger <matthias.bgg@gmail.com>
Cc: Neil Armstrong <narmstrong@baylibre.com>
Cc: Kevin Hilman <khilman@baylibre.com>
Cc: Jerome Brunet <jbrunet@baylibre.com>
Cc: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Cc: Marek Vasut <marex@denx.de>
Cc: Stefan Agner <stefan@agner.ch>
Cc: Sandy Huang <hjc@rock-chips.com>
Cc: "Heiko Stübner" <heiko@sntech.de>
Cc: Yannick Fertre <yannick.fertre@foss.st.com>
Cc: Philippe Cornu <philippe.cornu@foss.st.com>
Cc: Benjamin Gaignard <benjamin.gaignard@linaro.org>
Cc: Maxime Coquelin <mcoquelin.stm32@gmail.com>
Cc: Alexandre Torgue <alexandre.torgue@foss.st.com>
Cc: Maxime Ripard <mripard@kernel.org>
Cc: Chen-Yu Tsai <wens@csie.org>
Cc: Jernej Skrabec <jernej.skrabec@gmail.com>
Cc: Jyri Sarha <jyri.sarha@iki.fi>
Cc: Tomi Valkeinen <tomba@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-mips@vger.kernel.org
Cc: linux-mediatek@lists.infradead.org
Cc: linux-amlogic@lists.infradead.org
Cc: linux-rockchip@lists.infradead.org
Cc: linux-stm32@st-md-mailman.stormreply.com
Cc: linux-sunxi@lists.linux.dev
Link: https://patchwork.freedesktop.org/patch/msgid/20210622165511.3169559-9-daniel.vetter@ffwll.ch

drm/atomic-helper: make drm_gem_plane_helper_prepare_fb the default

There's a bunch of atomic drivers who don't do this quite correctly,
luckily most of them aren't in wide use or people would have noticed
the tearing.

By making this the default we avoid the constant audit pain and can
additionally remove a ton of lines from vfuncs for a bit more clarity
in smaller drivers.

While at it complain if there's a cleanup_fb hook but no prepare_fb
hook, because that makes no sense. I haven't found any driver which
violates this, but better safe than sorry.

Subsequent patches will reap the benefits.

v2: It's neither ... nor, not not (Sam)

Acked-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <mripard@kernel.org>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210623162200.3372056-1-daniel.vetter@ffwll.ch

dma-buf: Document dma-buf implicit fencing/resv fencing rules

Docs for struct dma_resv are fairly clear:

"A reservation object can have attached one exclusive fence (normally
associated with write operations) or N shared fences (read
operations)."

https://dri.freedesktop.org/docs/drm/driver-api/dma-buf.html#reservation-objects

Furthermore a review across all of upstream.

First of render drivers and how they set implicit fences:

- nouveau follows this contract, see in validate_fini_no_ticket()

nouveau_bo_fence(nvbo, fence, !!b->write_domains);

  and that last boolean controls whether the exclusive or shared fence
  slot is used.

- radeon follows this contract by setting

p->relocs[i].tv.num_shared = !r->write_domain;

  in radeon_cs_parser_relocs(), which ensures that the call to
  ttm_eu_fence_buffer_objects() in radeon_cs_parser_fini() will do the
  right thing.

- vmwgfx seems to follow this contract with the shotgun approach of
  always setting ttm_val_buf->num_shared = 0, which means
  ttm_eu_fence_buffer_objects() will only use the exclusive slot.

- etnaviv follows this contract, as can be trivially seen by looking
  at submit_attach_object_fences()

- i915 is a bit a convoluted maze with multiple paths leading to
  i915_vma_move_to_active(). Which sets the exclusive flag if
  EXEC_OBJECT_WRITE is set. This can either come as a buffer flag for
  softpin mode, or through the write_domain when using relocations. It
  follows this contract.

- lima follows this contract, see lima_gem_submit() which sets the
  exclusive fence when the LIMA_SUBMIT_BO_WRITE flag is set for that
  bo

- msm follows this contract, see msm_gpu_submit() which sets the
  exclusive flag when the MSM_SUBMIT_BO_WRITE is set for that buffer

- panfrost follows this contract with the shotgun approach of just
  always setting the exclusive fence, see
  panfrost_attach_object_fences(). Benefits of a single engine I guess

- v3d follows this contract with the same shotgun approach in
  v3d_attach_fences_and_unlock_reservation(), but it has at least an
  XXX comment that maybe this should be improved

- v4c uses the same shotgun approach of always setting an exclusive
  fence, see vc4_update_bo_seqnos()

- vgem also follows this contract, see vgem_fence_attach_ioctl() and
  the VGEM_FENCE_WRITE. This is used in some igts to validate prime
  sharing with i915.ko without the need of a 2nd gpu

- vritio follows this contract again with the shotgun approach of
  always setting an exclusive fence, see virtio_gpu_array_add_fence()

This covers the setting of the exclusive fences when writing.

Synchronizing against the exclusive fence is a lot more tricky, and I
only spot checked a few:

- i915 does it, with the optional EXEC_OBJECT_ASYNC to skip all
  implicit dependencies (which is used by vulkan)

- etnaviv does this. Implicit dependencies are collected in
  submit_fence_sync(), again with an opt-out flag
  ETNA_SUBMIT_NO_IMPLICIT. These are then picked up in
  etnaviv_sched_dependency which is the
  drm_sched_backend_ops->dependency callback.

- v4c seems to not do much here, maybe gets away with it by not having
  a scheduler and only a single engine. Since all newer broadcom chips than
  the OG vc4 use v3d for rendering, which follows this contract, the
  impact of this issue is fairly small.

- v3d does this using the drm_gem_fence_array_add_implicit() helper,
  which then it's drm_sched_backend_ops->dependency callback
  v3d_job_dependency() picks up.

- panfrost is nice here and tracks the implicit fences in
  panfrost_job->implicit_fences, which again the
  drm_sched_backend_ops->dependency callback panfrost_job_dependency()
  picks up. It is mildly questionable though since it only picks up
  exclusive fences in panfrost_acquire_object_fences(), but not buggy
  in practice because it also always sets the exclusive fence. It
  should pick up both sets of fences, just in case there's ever going
  to be a 2nd gpu in a SoC with a mali gpu. Or maybe a mali SoC with a
  pcie port and a real gpu, which might actually happen eventually. A
  bug, but easy to fix. Should probably use the
  drm_gem_fence_array_add_implicit() helper.

- lima is nice an easy, uses drm_gem_fence_array_add_implicit() and
  the same schema as v3d.

- msm is mildly entertaining. It also supports MSM_SUBMIT_NO_IMPLICIT,
  but because it doesn't use the drm/scheduler it handles fences from
  the wrong context with a synchronous dma_fence_wait. See
  submit_fence_sync() leading to msm_gem_sync_object(). Investing into
  a scheduler might be a good idea.

- all the remaining drivers are ttm based, where I hope they do
  appropriately obey implicit fences already. I didn't do the full
  audit there because a) not follow the contract would confuse ttm
  quite well and b) reading non-standard scheduler and submit code
  which isn't based on drm/scheduler is a pain.

Onwards to the display side.

- Any driver using the drm_gem_plane_helper_prepare_fb() helper will
  correctly. Overwhelmingly most drivers get this right, except a few
  totally dont. I'll follow up with a patch to make this the default
  and avoid a bunch of bugs.

- I didn't audit the ttm drivers, but given that dma_resv started
  there I hope they get this right.

In conclusion this IS the contract, both as documented and
overwhelmingly implemented, specically as implemented by all render
drivers except amdgpu.

Amdgpu tried to fix this already in

commit 049aca4363d8af87cab8d53de5401602db3b9999
Author: Christian König <christian.koenig@amd.com>
Date:   Wed Sep 19 16:54:35 2018 +0200

    drm/amdgpu: fix using shared fence for exported BOs v2

but this fix falls short on a number of areas:

- It's racy, by the time the buffer is shared it might be too late. To
  make sure there's definitely never a problem we need to set the
  fences correctly for any buffer that's potentially exportable.

- It's breaking uapi, dma-buf fds support poll() and differentitiate
  between, which was introduced in

commit 9b495a5887994a6d74d5c261d012083a92b94738
Author: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Date:   Tue Jul 1 12:57:43 2014 +0200

    dma-buf: add poll support, v3

- Christian König wants to nack new uapi building further on this
  dma_resv contract because it breaks amdgpu, quoting

  "Yeah, and that is exactly the reason why I will NAK this uAPI change.

  "This doesn't works for amdgpu at all for the reasons outlined above."

  https://lore.kernel.org/dri-devel/f2eb6751-2f82-9b23-f57e-548de5b729de@gmail.com/

  Rejecting new development because your own driver is broken and
  violates established cross driver contracts and uapi is really not
  how upstream works.

Now this patch will have a severe performance impact on anything that
runs on multiple engines. So we can't just merge it outright, but need
a bit a plan:

- amdgpu needs a proper uapi for handling implicit fencing. The funny
  thing is that to do it correctly, implicit fencing must be treated
  as a very strange IPC mechanism for transporting fences, where both
  setting the fence and dependency intercepts must be handled
  explicitly. Current best practices is a per-bo flag to indicate
  writes, and a per-bo flag to to skip implicit fencing in the CS
  ioctl as a new chunk.

- Since amdgpu has been shipping with broken behaviour we need an
  opt-out flag from the butchered implicit fencing model to enable the
  proper explicit implicit fencing model.

- for kernel memory fences due to bo moves at least the i915 idea is
  to use ttm_bo->moving. amdgpu probably needs the same.

- since the current p2p dma-buf interface assumes the kernel memory
  fence is in the exclusive dma_resv fence slot we need to add a new
  fence slot for kernel fences, which must never be ignored. Since
  currently only amdgpu supports this there's no real problem here
  yet, until amdgpu gains a NO_IMPLICIT CS flag.

- New userspace needs to ship in enough desktop distros so that users
  wont notice the perf impact. I think we can ignore LTS distros who
  upgrade their kernels but not their mesa3d snapshot.

- Then when this is all in place we can merge this patch here.

What is not a solution to this problem here is trying to make the
dma_resv rules in the kernel more clever. The fundamental issue here
is that the amdgpu CS uapi is the least expressive one across all
drivers (only equalled by panfrost, which has an actual excuse) by not
allowing any userspace control over how implicit sync is conducted.

Until this is fixed it's completely pointless to make the kernel more
clever to improve amdgpu, because all we're doing is papering over
this uapi design issue. amdgpu needs to attain the status quo
established by other drivers first, once that's achieved we can tackle
the remaining issues in a consistent way across drivers.

v2: Bas pointed me at AMDGPU_GEM_CREATE_EXPLICIT_SYNC, which I
entirely missed.

This is great because it means the amdgpu specific piece for proper
implicit fence handling exists already, and that since a while. The
only thing that's now missing is
- fishing the implicit fences out of a shared object at the right time
- setting the exclusive implicit fence slot at the right time.

Jason has a patch series to fill that gap with a bunch of generic
ioctl on the dma-buf fd:

https://lore.kernel.org/dri-devel/20210520190007.534046-1-jason@jlekstrand.net/

v3: Since Christian has fixed amdgpu now in

commit 8c505bdc9c8b955223b054e34a0be9c3d841cd20 (drm-misc/drm-misc-next)
Author: Christian König <christian.koenig@amd.com>
Date:   Wed Jun 9 13:51:36 2021 +0200

    drm/amdgpu: rework dma_resv handling v3

Use the audit covered in this commit message as the excuse to update
the dma-buf docs around dma_buf.resv usage across drivers.

Since dynamic importers have different rules also hammer these in
again while we're at it.

v4:
- Add the missing "through the device" in the dynamic section that I
  overlooked.
- Fix a kerneldoc markup mistake, the link didn't connect

v5:
- A few s/should/must/ to make clear what must be done (if the driver
  does implicit sync) and what's more a maybe (Daniel Stone)
- drop all the example api discussion, that needs to be expanded,
  clarified and put into a new chapter in drm-uapi.rst (Daniel Stone)

Cc: Daniel Stone <daniel@fooishbar.org>
Acked-by: Daniel Stone <daniel@fooishbar.org>
Reviewed-by: Dave Airlie <airlied@redhat.com> (v4)
Reviewed-by: Christian König <christian.koenig@amd.com> (v3)
Cc: mesa-dev@lists.freedesktop.org
Cc: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Cc: Dave Airlie <airlied@gmail.com>
Cc: Rob Clark <robdclark@chromium.org>
Cc: Kristian H. Kristensen <hoegsberg@google.com>
Cc: Michel Dänzer <michel@daenzer.net>
Cc: Daniel Stone <daniels@collabora.com>
Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: "Christian König" <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Deepak R Varma <mh12gx2825@gmail.com>
Cc: Chen Li <chenli@uniontech.com>
Cc: Kevin Wang <kevin1.wang@amd.com>
Cc: Dennis Li <Dennis.Li@amd.com>
Cc: Luben Tuikov <luben.tuikov@amd.com>
Cc: linaro-mm-sig@lists.linaro.org
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210624125246.166721-1-daniel.vetter@ffwll.ch

dma-buf: Switch to inline kerneldoc

Also review & update everything while we're at it.

This is prep work to smash a ton of stuff into the kerneldoc for
@resv.

v2: Move the doc for sysfs_entry.attachment_uid to the right place too
(Sam)

Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Christian König <christian.koenig@amd.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: "Christian König" <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Nirmoy Das <nirmoy.das@amd.com>
Cc: Deepak R Varma <mh12gx2825@gmail.com>
Cc: Chen Li <chenli@uniontech.com>
Cc: Kevin Wang <kevin1.wang@amd.com>
Cc: linux-media@vger.kernel.org
Cc: linaro-mm-sig@lists.linaro.org
Link: https://patchwork.freedesktop.org/patch/msgid/20210623161712.3370885-1-daniel.vetter@ffwll.ch

drm/gem: Tiny kernel clarification for drm_gem_fence_array_add

Spotted while trying to convert panfrost to these.

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: "Christian König" <christian.koenig@amd.com>
Cc: Lucas Stach <l.stach@pengutronix.de>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <mripard@kernel.org>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210622165511.3169559-15-daniel.vetter@ffwll.ch

drm/bridge: ti-sn65dsi86: Split connector creation to a function

To prepare for making connector creation option, move connector creation
out of ti_sn_bridge_attach to a separate function.

No functional change intended.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Stephen Boyd <swboyd@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210624000304.16281-7-laurent.pinchart+renesas@ideasonboard.com

drm/bridge: ti-sn65dsi86: Group code in sections

Reorganize the functions in sections, related to connector operations,
bridge operations, AUX adapter, GPIO controller and probe & remove.

This prepares for proper support of DRM_BRIDGE_ATTACH_NO_CONNECTOR that
will add more functions, to ensure that the code will stay readable.

No functional change intended.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Stephen Boyd <swboyd@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210624000304.16281-6-laurent.pinchart+renesas@ideasonboard.com

drm/bridge: ti-sn65dsi86: Wrap panel with panel-bridge

To simplify interfacing with the panel, wrap it in a panel-bridge and
let the DRM bridge helpers handle chaining of operations.

This also prepares for support of DRM_BRIDGE_ATTACH_NO_CONNECTOR, which
requires all components in the display pipeline to be represented by
bridges.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210624000304.16281-5-laurent.pinchart+renesas@ideasonboard.com

drm/bridge: ti-sn65dsi86: Use bitmask to store valid rates

The valid rates are stored in an array of 8 booleans. Replace it with a
bitmask to save space.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Stephen Boyd <swboyd@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210624000304.16281-4-laurent.pinchart+renesas@ideasonboard.com

drm/bridge: ti-sn65dsi86: Make enable GPIO optional

The enable signal may not be controllable by the kernel. Make it
optional.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Stephen Boyd <swboyd@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210624000304.16281-3-laurent.pinchart+renesas@ideasonboard.com

dt-bindings: drm/bridge: ti-sn65dsi8: Make enable GPIO optional

The SN65DSI86 EN pin can be hardwired to a high level, or connected to a
global reset signal, not controllable by the kernel. Make it optional in
those cases.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Stephen Boyd <swboyd@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210624000304.16281-2-laurent.pinchart+renesas@ideasonboard.com

drm/panfrost: Make sure MMU context lifetime is not bound to panfrost_priv

Jobs can be in-flight when the file descriptor is closed (either because
the process did not terminate properly, or because it didn't wait for
all GPU jobs to be finished), and apparently panfrost_job_close() does
not cancel already running jobs. Let's refcount the MMU context object
so it's lifetime is no longer bound to the FD lifetime and running jobs
can finish properly without generating spurious page faults.

Reported-by: Icecream95 <ixn@keemail.me>
Fixes: 7282f7645d06 ("drm/panfrost: Implement per FD address spaces")
Cc: <stable@vger.kernel.org>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210621133907.1683899-2-boris.brezillon@collabora.com

drm/panfrost: Fix implicit sync

Currently this has no practial relevance I think because there's not
many who can pull off a setup with panfrost and another gpu in the
same system. But the rules are that if you're setting an exclusive
fence, indicating a gpu write access in the implicit fencing system,
then you need to wait for all fences, not just the previous exclusive
fence.

panfrost against itself has no problem, because it always sets the
exclusive fence (but that's probably something that will need to be
fixed for vulkan and/or multi-engine gpus, or you'll suffer badly).
Also no problem with that against display.

With the prep work done to switch over to the dependency helpers this
is now a oneliner.

Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Rob Herring <robh@kernel.org>
Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Cc: Steven Price <steven.price@arm.com>
Cc: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: "Christian König" <christian.koenig@amd.com>
Cc: linux-media@vger.kernel.org
Cc: linaro-mm-sig@lists.linaro.org
Link: https://patchwork.freedesktop.org/patch/msgid/20210622165511.3169559-7-daniel.vetter@ffwll.ch

drm/panfrost: Use xarray and helpers for depedency tracking

More consistency and prep work for the next patch.

Aside: I wonder whether we shouldn't just move this entire xarray
business into the scheduler so that not everyone has to reinvent the
same wheels. Cc'ing some scheduler people for this too.

v2: Correctly handle sched_lock since Lucas pointed out it's needed.

v3: Rebase, dma_resv_get_excl_unlocked got renamed

v4: Don't leak job references on failure (Steven).

Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Cc: Lucas Stach <l.stach@pengutronix.de>
Cc: "Christian König" <christian.koenig@amd.com>
Cc: Luben Tuikov <luben.tuikov@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Lee Jones <lee.jones@linaro.org>
Cc: Steven Price <steven.price@arm.com>
Cc: Rob Herring <robh@kernel.org>
Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Cc: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: linux-media@vger.kernel.org
Cc: linaro-mm-sig@lists.linaro.org
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210622165511.3169559-6-daniel.vetter@ffwll.ch

drm/panfrost: Shrink sched_lock

drm/scheduler requires a lock between _init and _push_job, but the
reservation lock dance doesn't. So shrink the critical section a
notch.

v2: Lucas pointed out how this should really work, I got it all wrong
in v1.

Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Lucas Stach <l.stach@pengutronix.de>
Cc: Rob Herring <robh@kernel.org>
Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Cc: Steven Price <steven.price@arm.com>
Cc: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210622165511.3169559-5-daniel.vetter@ffwll.ch

dma-resv: Fix kerneldoc

Oversight from

commit 6edbd6abb783d54f6ac4c3ed5cd9e50cff6c15e9
Author: Christian König <christian.koenig@amd.com>
Date:   Mon May 10 16:14:09 2021 +0200

    dma-buf: rename and cleanup dma_resv_get_excl v3

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: "Christian König" <christian.koenig@amd.com>
Cc: linux-media@vger.kernel.org
Cc: linaro-mm-sig@lists.linaro.org
Link: https://patchwork.freedesktop.org/patch/msgid/20210622165511.3169559-2-daniel.vetter@ffwll.ch

drm/amdgpu: Fix BUG_ON assert

With added CPU domain to placement you can have
now 3 placemnts at once.

CC: stable@kernel.org
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210622162339.761651-5-andrey.grodzovsky@amd.com

drm/amdgpu: switch gtt_mgr to counting used pages

Change mgr->available into mgr->used (invert the value).

Makes more sense to do it this way since we don't need the spinlock any
more to double check the handling.

v3 (chk): separated from the TEMPOARAY FLAG change.

Signed-off-by: Lang Yu <Lang.Yu@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Acked-by: Nirmoy Das <nirmoy.das@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210622162339.761651-4-andrey.grodzovsky@amd.com

ydrm/amdgpu: always allow evicting to SYSTEM domain

When we run out of GTT we should still be able to evict VRAM->SYSTEM
with a bounce bufferdrm/amdgpu: always allow evicting to SYSTEM domain

Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Acked-by: Nirmoy Das <nirmoy.das@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210622162339.761651-3-andrey.grodzovsky@amd.com

drm/amdgpu: user temporary GTT as bounce buffer

Currently, we have a limitted GTT memory size and need a bounce buffer
when doing buffer migration between VRAM and SYSTEM domain.

The problem is under GTT memory pressure we can't do buffer migration
between VRAM and SYSTEM domain. But in some cases we really need that.
Eespecially when validating a VRAM backing store BO which resides in
SYSTEM domain.

v2: still account temporary GTT allocations
v3 (chk): revert to the simpler change for now

Signed-off-by: Lang Yu <Lang.Yu@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Acked-by: Nirmoy Das <nirmoy.das@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210622162339.761651-2-andrey.grodzovsky@amd.com

drm/ttm: add TTM_PL_FLAG_TEMPORARY flag v3

Sometimes drivers need to use bounce buffers to evict BOs. While those reside
in some domain they are not necessarily suitable for CS.

Add a flag so that drivers can note that a bounce buffers needs to be
reallocated during validation.

v2: add detailed comments
v3 (chk): merge commits and rework commit message

Suggested-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Lang Yu <Lang.Yu@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Acked-by: Nirmoy Das <nirmoy.das@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210622162339.761651-1-andrey.grodzovsky@amd.com

drm/ttm: Fix multihop assert on eviction.

Problem:
Under memory pressure when GTT domain is almost full multihop assert
will come up when trying to evict LRU BO from VRAM to SYSTEM.

Fix:
Don't assert on multihop error in evict code but rather do a retry
as we do in ttm_bo_move_buffer

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210622162339.761651-6-andrey.grodzovsky@amd.com

drm/vc4: hdmi: Register HDMI codec

The hdmi-codec brings a lot of advanced features, including the HDMI
channel mapping. Let's use it in our driver instead of our own codec.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Reviewed-by: Nicolas Saenz Julienne <nsaenz@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525132354.297468-11-maxime@cerno.tech

drm/vc4: hdmi: Remove firmware logic for MAI threshold setting

This was a workaround for bugs in hardware on earlier Pi models
and wasn't totally successful.

It makes audio quality worse on a Pi4 at the higher sample rates

Signed-off-by: Dom Cobley <popcornmix@gmail.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525132354.297468-10-maxime@cerno.tech

drm/vc4: hdmi: Set VC4_HDMI_MAI_CONFIG_FORMAT_REVERSE

Without this bit set, HDMI_MAI_FORMAT doesn't pick up
the format and samplerate from DVP_CFG_MAI0_FMT and you
can't get HDMI_HDMI_13_AUDIO_STATUS_1 to indicate HBR mode

Signed-off-by: Dom Cobley <popcornmix@gmail.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Reviewed-by: Nicolas Saenz Julienne <nsaenz@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525132354.297468-9-maxime@cerno.tech

drm/vc4: hdmi: Set HDMI_MAI_FMT

The hardware uses this for generating the right audio
data island packets when using formats other than PCM

Signed-off-by: Dom Cobley <popcornmix@gmail.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Reviewed-by: Nicolas Saenz Julienne <nsaenz@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525132354.297468-8-maxime@cerno.tech

drm/vc4: hdmi: Set HD_CTL_WHOLSMP and HD_CTL_CHALIGN_SET

Symptom is random switching of speakers when using multichannel.

Repeatedly running speakertest -c8 occasionally starts with
channels jumbled. This is fixed with HD_CTL_WHOLSMP.

The other bit looks beneficial and apears harmless in testing so
I'd suggest adding it too.

Documentation says: HD_CTL_WHILSMP_SET
Wait for whole sample. When this bit is set MAI transmit will start
only when there is at least one whole sample available in the fifo.

Documentation says: HD_CTL_CHALIGN_SET
Channel Align When Overflow. This bit is used to realign the audio
channels in case of an overflow.
If this bit is set, after the detection of an overflow, equal
amount of dummy words to the missing words will be written to fifo,
filling up the broken sample and maintaining alignment.

Signed-off-by: Dom Cobley <popcornmix@gmail.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Reviewed-by: Nicolas Saenz Julienne <nsaenz@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525132354.297468-7-maxime@cerno.tech

Merge tag 'asoc-hdmi-codec-improvements-v2' of git://git./linux/kernel/git/mripard/linux into drm-misc-next

Improvements to the hdmi-codec driver and ALSA infrastructure around it
to support the HDMI Channel Mapping and IEC958 controls

Signed-off-by: Maxime Ripard <maxime@cerno.tech>
# -----BEGIN PGP SIGNATURE-----
#
# iHUEABYIAB0WIQRcEzekXsqa64kGDp7j7w1vZxhRxQUCYMHitQAKCRDj7w1vZxhR
# xQ40AP49z0mUifkpbyUvYwdmrUVlg2JEWSTOWaH3tp0kke/dBQEA1vYxdMimhFu3
# SYKXxgtvlT7vL48vNYBxGbNuGQvzJw4=
# =IWna
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 10 Jun 2021 12:00:21 PM CEST
# gpg:                using EDDSA key 5C1337A45ECA9AEB89060E9EE3EF0D6F671851C5
# gpg: Good signature from "Maxime Ripard <maxime.ripard@anandra.org>" [unknown]
# gpg:                 aka "Maxime Ripard <mripard@kernel.org>" [unknown]
# gpg:                 aka "Maxime Ripard (Work Address) <maxime@cerno.tech>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: BE56 75C3 7E81 8C8B 5764  241C 254B CFC5 6BF6 CE8D
#      Subkey fingerprint: 5C13 37A4 5ECA 9AEB 8906  0E9E E3EF 0D6F 6718 51C5
From: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20210610122550.jnriewchqspdcrwk@gilmour

drm: Mention the power state requirement on side-channel operations

The drm_connector detect, drm_dp_aux transfer and mipi_dsi_host
operations typically require to access their underlying device to
perform what is expected of them.

However, there's no guarantee on the fact that the device has been
enabled through atomic_enable or similar that will usually power the
device. The access to an unpowered device is then an undefined behaviour
ranging from the access being ignored to a hard CPU hang.

Let's document that expectation to avoid as much as possible those
consequences.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210616141529.630719-3-maxime@cerno.tech

drm/dp_helper: Mention the concurrency requirement hw_mutex

Drivers that allow concurrent access over multiple DP channels need to
provide additional locking, even though the hw_mutex field might
indicate otherwise. Clarify it in the documentation.

Suggested-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210616141529.630719-2-maxime@cerno.tech

drm/dp_helper: Rework the drm_dp_aux documentation

Split the existing documentation to move the comments on particular
fields next to them.

Suggested-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210616141529.630719-1-maxime@cerno.tech

drm/virtio: Add the fence in resource_flush if present

If the framebuffer associated with the plane contains a fence, then
it is added to resource_flush and will be waited upon for a max of
50 msecs or until it is signalled by the Host.

Cc: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Vivek Kasireddy <vivek.kasireddy@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20210610232456.671905-4-vivek.kasireddy@intel.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm/virtio: Prepare resource_flush to accept a fence

A fence will be added to resource_flush for resources that
are guest blobs.

Cc: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Vivek Kasireddy <vivek.kasireddy@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20210610232456.671905-3-vivek.kasireddy@intel.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm/virtio: Add fences for Guest blobs

Add prepare and cleanup routines for primary planes as well
where a fence is added only if the BO/FB associated with the
plane is a guest blob.

Cc: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Vivek Kasireddy <vivek.kasireddy@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20210610232456.671905-2-vivek.kasireddy@intel.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

udmabuf: configurable list_limit and size_limit_mb

Default list_limit and size_limit_mb are not big enough to cover all
possible use cases. For example, list_limit could be well over its default,
1024 if only one or several pages are chained in all individual list entries
when creating dmabuf backed by >4MB buffer. list_limit and size_limit_mb are
now defined as module parameters so that those can be optionally configured
by root with proper values to remove these constraints.

Cc: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Dongwon Kim <dongwon.kim@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20210611212107.9876-1-dongwon.kim@intel.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

dma-buf: Document non-dynamic exporter expectations better

Christian and me realized we have a pretty massive disconnect about
different interpretations of what dma_resv is used for by different
drivers. The discussion is much, much bigger than this change here,
but this is an important one:

Non-dynamic exporters must guarantee that the memory they return is
ready for use. They cannot expect importers to wait for the exclusive
fence. Only dynamic importers are required to obey the dma_resv fences
strictly (and more patches are needed to define exactly what this
means).

Christian has patches to update nouvea, radeon and amdgpu. The only
other driver using both ttm and supporting dma-buf export is qxl,
which only uses synchronous ttm_bo_move.

v2: To hammer this in document that dynamic importers _must_ wait for
the exclusive fence after having called dma_buf_map_attachment.

Reviewed-by: Christian König <christian.koenig@amd.com>
Cc: Christian König <ckoenig.leichtzumerken@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210621151758.2347474-1-daniel.vetter@ffwll.ch

drm/amdgpu: rework dma_resv handling v3

Drop the workaround and instead implement a better solution.

Basically we are now chaining all submissions using a dma_fence_chain
container and adding them as exclusive fence to the dma_resv object.

This way other drivers can still sync to the single exclusive fence
while amdgpu only sync to fences from different processes.

v3: add the shared fence first before the exclusive one

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210614174536.5188-2-christian.koenig@amd.com

drm/amdgpu: unwrap fence chains in the explicit sync fence

Unwrap the explicit fence if it is a dma_fence_chain and
sync to the first fence not matching the owner rules.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210614174536.5188-1-christian.koenig@amd.com

drm: bridge: ti-sn65dsi83: Retrieve the display mode from the state

Instead of storing a copy of the display mode in the sn65dsi83
structure, retrieve it from the atomic state in
sn65dsi83_atomic_enable(). This allows the removal of the .mode_set()
operation, and completes the transition to the atomic API.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210621125518.13715-6-laurent.pinchart@ideasonboard.com

drm: bridge: ti-sn65dsi83: Retrieve output format from bridge state

The driver currently iterates over all connectors to get the bus format,
used to configure the LVDS output format. This causes several issues:

- If other connectors than the LVDS output are present, the format used
  by the driver may end up belonging to an entirely different output.

- The code can crash if some connectors are not connected, as bus_format
  may then be NULL.

- There's no guarantee that the bus format on the connector at the
  output of the pipeline matches the output of the sn65dsi83, as there
  may be other bridges in the pipeline.

Solve this by retrieving the format from the bridge state instead, which
provides the format corresponding to the output of the bridge.

The struct sn65dsi83 lvds_format_24bpp and lvds_format_jeida fields are
moved to local variables in sn65dsi83_atomic_enable() as they're now
used in that function only.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210621125518.13715-5-laurent.pinchart@ideasonboard.com

drm: bridge: ti-sn65dsi83: Switch to atomic operations

Use the atomic version of the enable/disable operations to continue the
transition to the atomic API, started with the introduction of
.atomic_get_input_bus_fmts(). This will be needed to access the mode
from the atomic state.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210621125518.13715-4-laurent.pinchart@ideasonboard.com

drm: bridge: ti-sn65dsi83: Pass mode explicitly to helper functions

Pass the display mode explicitly to the sn65dsi83_get_lvds_range() and
sn65dsi83_get_dsi_range() functions to prepare for its removal from the
sn65dsi83 structure. This is not meant to bring any functional change.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210621125518.13715-3-laurent.pinchart@ideasonboard.com

drm: bridge: ti-sn65dsi83: Move LVDS format selection to .mode_set()

The LVDS format is selected based on the bus format reported by the
connector. This is currently done in .mode_fixup(), but that's not the
right place, as the format should be selected when setting the mode.
Move it to .mode_set().

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210621125518.13715-2-laurent.pinchart@ideasonboard.com

drm/panel: Add bus_format and bus_flags for EDT ETM0430G0DH6

Add corresponding bus_format and bus_flags for the EDT ETM0430G0DH6
display.

Signed-off-by: Stefan Riedmueller <s.riedmueller@phytec.de>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210621150930.86617-3-s.riedmueller@phytec.de

drm/panel: Add connector_type for some EDT displays

The connector_type for following two EDT displays is missing:
 - EDT ETM0430G0DH6
 - EDT ETM0700G0BDH6

Both are parallel displays thus add the corresponding connector_type.

Signed-off-by: Stefan Riedmueller <s.riedmueller@phytec.de>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210621150930.86617-2-s.riedmueller@phytec.de

drm/panel: Add connector_type and bus_format for AUO G104SN02 V2 panel

The AUO G104SN02 V2 is an LVDS display which supports 6 and 8 bpc PSWG.
Add the corresponding connector type and 8 bpc as default bus_format.

Signed-off-by: Stefan Riedmueller <s.riedmueller@phytec.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210621150930.86617-1-s.riedmueller@phytec.de

drm/bridge: ti-sn65dsi83: Replace connector format patching with atomic_get_input_bus_fmts

Patching the connector format is causing various problematic
side effects. Implement .atomic_get_input_bus_fmts callback
instead, which sets up the input (DSI-end) format, and that
format can then be used in pipeline format negotiation between
the DSI-end of this bridge and the other component closer to
the scanout engine.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Adam Ford <aford173@gmail.com>
Cc: Douglas Anderson <dianders@chromium.org>
Cc: Frieder Schrempf <frieder.schrempf@kontron.de>
Cc: Jagan Teki <jagan@amarulasolutions.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Loic Poulain <loic.poulain@linaro.org>
Cc: Philippe Schenker <philippe.schenker@toradex.com>
Cc: Robert Foss <robert.foss@linaro.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Stephen Boyd <swboyd@chromium.org>
Cc: Valentin Raevsky <valentin@compulab.co.il>
Cc: dri-devel@lists.freedesktop.org
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210620224208.184719-1-marex@denx.de

drm/gma500/oaktrail_lvds: replace continue with break

Currently a loop scans through the connector list checking
for connectors that do not match a specific criteria. The
use of the continue statement is a little unintuitive and
can confuse static analysis checking.  Invert the criteria
matching logic and use a break to terminate the loop once
the first suitable connector has been found.

Thanks to Patrik Jakobsson for explaining the original
intent of the code and suggesting this change.

Addresses-Coverity: ("Continue has no effect")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210618183524.590186-1-colin.king@canonical.com

drm/amdgpu: fix amdgpu_preempt_mgr_new()

There is a reversed if statement in amdgpu_preempt_mgr_new() so it
always returns -ENOMEM.

Fixes: 09b020bb05a5 ("Merge tag 'drm-misc-next-2021-06-09' of git://anongit.freedesktop.org/drm/drm-misc into drm-next")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/YMxbQXg/Wqm0ACxt@mwanda
Signed-off-by: Christian König <christian.koenig@amd.com>

drm/bridge: anx7625: Make hpd workqueue freezable

There were still a race condition between hpd work and suspend, since
the workqueue work can still be run after anx7625 had powered off in
suspend.

Since we never want hpd work to run while suspending, and there's no
harm to delay them to be run after resume, mark the workqueue as
WQ_FREEZABLE so all works won't run while suspending.

Fixes: 409776fa3c42 ("drm/bridge: anx7625: add suspend / resume hooks")

Signed-off-by: Pi-Hsun Shih <pihsun@chromium.org>
Reviewed-by: Tzung-Bi Shih <tzungbi@google.com>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210615032937.2328468-1-pihsun@chromium.org

dma-buf: Document DMA_BUF_IOCTL_SYNC (v3)

This adds a new "DMA Buffer ioctls" section to the dma-buf docs and adds
documentation for DMA_BUF_IOCTL_SYNC.

v2 (Daniel Vetter):
 - Fix a couple typos
 - Add commentary about synchronization with other devices
 - Use item list format for describing flags

v3 (Pekka Paalanen):
 - Clarify stalling requirements.
 - Be more clear that that DMA_BUF_IOCTL_SYNC with SINC_END has to be
   called before more GPU work happens.

Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Christian König <christian.koenig@amd.com>
Acked-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Cc: Sumit Semwal <sumit.semwal@linaro.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210617194258.579011-1-jason@jlekstrand.net

fbdev/omap2: use DEFINE_SPINLOCK() instead of spin_lock_init().

spinlock can be initialized automatically with DEFINE_SPINLOCK()
rather than explicitly calling spin_lock_init().

Signed-off-by: Jiahua Yu <yujiahua1@huawei.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210616031713.24959-1-yujiahua1@huawei.com

drm/vmwgfx: Fix build issues in mksGuestStats discovered by the kernel test robot

Fixes for ARCH
i386
* printk format specifier warnings
* inconsistent operand constraints in an ‘asm’ errors
arm64
* not targeted by the commit being fixed

Reviewed-by: Zack Rusin <zackr@vmware.com>
Fixes: 7a7a933edd6c ("drm/vmwgfx: Introduce VMware mks-guest-stats")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210615182336.995192-5-zackr@vmware.com
Signed-off-by: Zack Rusin <zackr@vmware.com>

drm/vmwgfx: Fix a bad merge in otable batch takedown

Change
2ef4fb92363c ("drm/vmwgfx: Make sure bo's are unpinned before putting them back")
caused a conflict in one of the drm trees and the merge commit
68a32ba14177 ("Merge tag 'drm-next-2021-04-28' of git://anongit.freedesktop.org/drm/drm")
accidently re-added code that the original change was removing.
Fixed by removing the incorrect buffer unpin - it has already been unpinned
two lines above.

Fixes: 68a32ba14177 ("Merge tag 'drm-next-2021-04-28' of git://anongit.freedesktop.org/drm/drm")
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210615182336.995192-4-zackr@vmware.com

drm/vmwgfx: Fix a 64bit regression on svga3

Register accesses are always 4bytes, accidently this was changed to
a void pointer whwqich badly breaks 64bit archs when running on top
of svga3.

Fixes: 2cd80dbd3551 ("drm/vmwgfx: Add basic support for SVGA3")
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210615182336.995192-3-zackr@vmware.com

drm/vmwgfx: Update device headers

Historically our device headers have been forked versions of the
internal device headers, this has made maintaining them a bit
of a burden. To fix the situation, going forward, the device headers
will be verbatim copies of the internal headers.
To do that the driver code has to be adapted to use pristine
device headers. This will make future update to the device
headers trivial and automatic.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210615182336.995192-2-zackr@vmware.com

MAINTAINERS: update vmwgfx info

Roland will be focusing on lavapipe over the next few months and
won't have time for vmwgfx.
vmwgfx is now maintained within drm-misc.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210615182336.995192-1-zackr@vmware.com

drm/vc4: hdmi: Rely on interrupts to handle hotplug

DRM currently polls for the HDMI connector status every 10s, which can
be an issue when we connect/disconnect a display quickly or the device
on the other end only issues a hotplug pulse (for example on EDID
change).

Switch the driver to rely on the internal controller logic for the
BCM2711/RPi4.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210524132018.264396-1-maxime@cerno.tech

drivers: gpu: add missing MODULE_DEVICE_TABLE in anx7625.c

This patch adds missing MODULE_DEVICE_TABLE definition which generates
correct modalias for automatic loading of this driver when it is built
as an external module.

Signed-off-by: Yu Jiahua <yujiahua1@huawei.com>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Reviewed-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210616034448.34919-1-yujiahua1@huawei.com

dmabuf: Add the capability to expose DMA-BUF stats in sysfs

Overview
========
The patch adds DMA-BUF statistics to /sys/kernel/dmabuf/buffers. It
allows statistics to be enabled for each DMA-BUF in sysfs by enabling
the config CONFIG_DMABUF_SYSFS_STATS.

The following stats will be exposed by the interface:

/sys/kernel/dmabuf/buffers/<inode_number>/exporter_name
/sys/kernel/dmabuf/buffers/<inode_number>/size
/sys/kernel/dmabuf/buffers/<inode_number>/attachments/<attach_uid>/device
/sys/kernel/dmabuf/buffers/<inode_number>/attachments/<attach_uid>/map_counter

The inode_number is unique for each DMA-BUF and was added earlier [1]
in order to allow userspace to track DMA-BUF usage across different
processes.

Use Cases
=========
The interface provides a way to gather DMA-BUF per-buffer statistics
from production devices. These statistics will be used to derive DMA-BUF
per-exporter stats and per-device usage stats for Android Bug reports.
The corresponding userspace changes can be found at [2].
Telemetry tools will also capture this information(along with other
memory metrics) periodically as well as on important events like a
foreground app kill (which might have been triggered by Low Memory
Killer). It will also contribute to provide a snapshot of the system
memory usage on other events such as OOM kills and Application Not
Responding events.

Background
==========
Currently, there are two existing interfaces that provide information
about DMA-BUFs.
1) /sys/kernel/debug/dma_buf/bufinfo
debugfs is however unsuitable to be mounted in production systems and
cannot be considered as an alternative to the sysfs interface being
proposed.
2) proc/<pid>/fdinfo/<fd>
The proc/<pid>/fdinfo/<fd> files expose information about DMA-BUF fds.
However, the existing procfs interfaces can only provide information
about the buffers for which processes hold fds or have the buffers
mmapped into their address space. Since the procfs interfaces alone
cannot provide a full picture of all DMA-BUFs in the system, there is
the need for an alternate interface to provide this information on
production systems.

The patch contains the following major improvements over v1:
1) Each attachment is represented by its own directory to allow creating
a symlink to the importing device and to also provide room for future
expansion.
2) The number of distinct mappings of each attachment is exposed in a
separate file.
3) The per-buffer statistics are now in /sys/kernel/dmabuf/buffers
inorder to make the interface expandable in future.

All of the improvements above are based on suggestions/feedback from
Daniel Vetter and Christian König.

A shell script that can be run on a classic Linux environment to read
out the DMA-BUF statistics can be found at [3](suggested by John
Stultz).

[1]: https://lore.kernel.org/patchwork/patch/1088791/
[2]: https://android-review.googlesource.com/q/topic:%22dmabuf-sysfs%22+(status:open%20OR%20status:merged)
[3]: https://android-review.googlesource.com/c/platform/system/memory/libmeminfo/+/1549734

Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Hridya Valsaraju <hridya@google.com>
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210603214758.2955251-1-hridya@google.com

drm/ast: Add detect function support

The existence of the connector cannot be detected,
so add the detect function to support.

Signed-off-by: Ainux <ainux.wang@gmail.com>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210526111515.40015-1-ainux.wang@gmail.com

drm/panel: s6e63m0: Switch to DBI abstraction for SPI

The SPI access to s6e63m0 is using the DBI protocol, so switch
to using the elaborate DBI protocol implementation in the DRM
DBI helper library.

Acked-by: Noralf Trønnes <noralf@tronnes.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210614181135.1124445-2-linus.walleij@linaro.org

drm/dbi: Support DBI typec1 read operations

Implement SPI reads for typec1, for SPI controllers that
can support 9bpw in addition to 8bpw (such as GPIO bit-banged
SPI).

9bpw emulation is not supported but we have to start with
something.

This is used by s6e63m0 to read display MTP information
which is used by the driver for backlight control.

Reviewed-by: Noralf Trønnes <noralf@tronnes.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210614181135.1124445-1-linus.walleij@linaro.org

dma-buf: add dma_fence_chain_alloc/free v3

Add a common allocation helper. Cleaning up the mix of kzalloc/kmalloc
and some unused code in the selftest.

v2: polish kernel doc a bit
v3: polish kernel doc even a bit more

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210611120301.10595-3-christian.koenig@amd.com

dma-buf: some dma_fence_chain improvements

The callback and the irq work are never used at the same
time. Putting them into an union saves us 24 bytes and
makes the structure only 120 bytes in size.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210611120301.10595-2-christian.koenig@amd.com

drm/doc: document drm_mode_get_plane

Add a small description and document struct fields of
drm_mode_get_plane.

Signed-off-by: Leandro Ribeiro <leandro.ribeiro@collabora.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Signed-off-by: Simon Ser <contact@emersion.fr>
Link: https://patchwork.freedesktop.org/patch/msgid/20210611213516.77904-2-leandro.ribeiro@collabora.com

Merge tag 'tags/topic/i915-ttm-2021-06-11' into drm-misc-next

drm-misc and drm-intel pull request for topic/i915-ttm:
- Convert i915 lmem handling to ttm.
- Add a patch to temporarily add a driver_private member to vma_node.
- Use this to allow mixed object mmap handling for i915.

Merge drm/drm-next into drm-misc-next

Backmerge to prepare for i915-ttm topic branch.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>

drm/vmwgfx: Fix implicit declaration error

The declarations of ttm_range_man_init and ttm_range_man_fini
have been moved to ttm_range_manager.h so we have to add it
to the include list.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reported-by: Randy Dunlap <rdunlap@infradead.org>
Fixes: 3eb7d96e9415 ("drm/ttm: flip over the range manager to self allocated nodes")
Cc: Christian König <christian.koenig@amd.com>
Cc: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210609172307.131929-10-zackr@vmware.com

drm/vmwgfx: Remove vmw_chipset

vmw_chipset was duplicating pci_id. They are exactly the same
variable just with two different names. Becuase pci_id was
already used to detect the SVGA version, there's no point
in having vmw_chipset and thus we can remove it.

All references to vmw_chipset should use pci_id.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210609172307.131929-9-zackr@vmware.com

drm/vmwgfx: Refactor vmw_mksstat_remove_ioctl to expect pgid match with vmw_mksstat_add_ioctl to authorise removal.

Original vmw_mksstat_remove_ioctl expected pid to match the corresponding vmw_mksstat_add_ioctl.
That made impossible en-masse removals by one pid, which is a valid use case, so pid match was
discarded. Current change enforces a broader pgid match as a form of protection from arbitrary
processes interrupting an ongoing mks-guest-stats.

Reviewed-by: Zack Rusin <zackr@vmware.com>
Signed-off-by: Martin Krastev <krastevm@vmware.com>
Signed-off-by: Zack Rusin <zackr@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210609172307.131929-8-zackr@vmware.com

drm/vmwgfx: inline access to the pages from the piter

The indirection doesn't make sense because we always go through
the same function pointer. Instead of the extra indirection
lets inline the access to the current page.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210609172307.131929-7-zackr@vmware.com

drm/vmwgfx: remove code that was using physical page addresses

This code has been unused for a while now. When the explicit checks
for whether the driver is running on top of non-coherent swiotlb
have been deprecated we lost the ability to fallback to physical
mappings. Instead of trying to readd a module parameter to force
usage of physical addresses it's better to just force coherent
TTM pages via the force_coherent module parameter making this
code pointless.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210609172307.131929-6-zackr@vmware.com

drm/vmwgfx: Fix some static checker warnings

Fix some minor issues that Coverity spotted in the code. None
of that are serious but they're all valid concerns so fixing
them makes sense.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210609172307.131929-5-zackr@vmware.com

drm/vmwgfx: Fix subresource updates with new contexts

The has_dx variable was only set during the initialization which
meant that UPDATE_SUBRESOURCE was never used. We were emulating it
with UPDATE_GB_IMAGE but that's always been a stop-gap. Instead
of has_dx which has been deprecated a long time ago we need to check
for whether shader model 4.0 or newer is available to the device.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210609172307.131929-4-zackr@vmware.com

drm/vmwgfx: Introduce VMware mks-guest-stats

VMware mks-guest-stats mechanism allows the collection of performance stats from
guest userland GL contexts, as well as from vmwgfx kernelspace, via a set of sw-
defined performance counters. The userspace performance counters are (de)registerd
with vmware-vmx-stats hypervisor via new iocts. The vmwgfx kernelspace counters
are controlled at build-time via a new config DRM_VMWGFX_MKSSTATS.

* Add vmw_mksstat_{add|remove|reset}_ioctl controlling the tracking of
  mks-guest-stats in guest winsys contexts
* Add DRM_VMWGFX_MKSSTATS config to drivers/gpu/drm/vmwgfx/Kconfig controlling
  the instrumentation of vmwgfx for kernelspace mks-guest-stats counters
* Instrument vmwgfx vmw_execbuf_ioctl to collect mks-guest-stats according to
  DRM_VMWGFX_MKSSTATS

Signed-off-by: Martin Krastev <krastevm@vmware.com>
Reviewed-by: Zack Rusin <zackr@vmware.com>
Signed-off-by: Zack Rusin <zackr@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210609172307.131929-3-zackr@vmware.com

drm/vmwgfx: Simplify devcaps code

Make devcaps code self-contained so that it's easier to cache
and operate on them.
As the number of devcaps got bigger the code dealing with them
got more and more tricky. Lets create a central place to deal
with all the complexity. This lets us remove the lock we used
to require to deal with register write races because we only
read the devcaps at initialization.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210609172307.131929-2-zackr@vmware.com

arm64: dts: qcom: sc7180-trogdor: Move panel under the bridge chip

Putting the panel under the bridge chip (under the aux-bus node)
allows the panel driver to get access to the DP AUX bus, enabling all
sorts of fabulous new features.

While we're at this, get rid of a level of hierarchy for the panel
node. It doesn't need "ports / port" and can just have a "port" child.

For Linux, this patch has a hard requirement on the patches adding DP
AUX bus support to the ti-sn65dsi86 bridge chip driver. See the patch
("drm/bridge: ti-sn65dsi86: Add support for the DP AUX bus").

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210611101711.v10.11.Ibdb7735fb1844561b902252215a69526a14f9abd@changeid

drm/bridge: ti-sn65dsi86: Improve probe errors with dev_err_probe()

As I was testing to make sure that the DEFER path worked well with my
patch series, I got tired of seeing this scary message in my logs just
because the panel needed to defer:
  [drm:ti_sn_bridge_probe] *ERROR* could not find any panel node

Let's use dev_err_probe() which nicely quiets this error and also
simplifies the code a tiny bit. We'll also update other places in the
file which can use dev_err_probe().

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210611101711.v10.10.I24bba069e63b1eea84443eef0c8535fd032a6311@changeid

drm/bridge: ti-sn65dsi86: Don't read EDID blob over DDC

This is really just a revert of commit 58074b08c04a ("drm/bridge:
ti-sn65dsi86: Read EDID blob over DDC"), resolving conflicts.

The old code failed to read the EDID properly in a very important
case: before the bridge's pre_enable() was called. The way things need
to work:
1. Read the EDID.
2. Based on the EDID, decide on video settings and pixel clock.
3. Enable the bridge w/ the desired settings.

The way things were working:
1. Try to read the EDID but fail; fall back to hardcoded values.
2. Based on hardcoded values, decide on video settings and pixel clock.
3. Enable the bridge w/ the desired settings.
4. Try again to read the EDID, it works now!
5. Realize that the hardcoded settings weren't quite right.
6. Disable / reenable the bridge w/ the right settings.

The reasons for the failures were twofold:
a) Since we never ran the bridge chip's pre-enable then we never set
   the bit to ignore HPD. This meant the bridge chip didn't even _try_
   to go out on the bus and communicate with the panel.
b) Even if we fixed things to ignore HPD, the EDID still wouldn't read
   if the panel wasn't on.

Instead of reverting the code, we could fix it to set the HPD bit and
also power on the panel. However, it also works nicely to just let the
panel code read the EDID. Now that we've split the driver up we can
expose the DDC AUX channel bus to the panel node. The panel can take
charge of reading the EDID.

NOTE: in order for things to work, anyone that needs to read the EDID
will need to instantiate their panel using the new DP AUX bus (AKA by
listing their panel under the "aux-bus" node of the bridge chip in the
device tree).

In the future if we want to use the bridge chip to provide a full
external DP port (which won't have a panel) then we will have to
conditinally add EDID reading back in.

Suggested-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210611101711.v10.9.I9330684c25f65bb318eff57f0616500f83eac3cc@changeid

drm/bridge: ti-sn65dsi86: Add support for the DP AUX bus

We want to provide our panel with access to the DP AUX channel. The
way to do this is to let our panel be a child of ours using the fancy
new DP AUX bus support.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210611101711.v10.8.Ib5fe0638da85800141ce141bb8e441c5f25438d4@changeid

drm/bridge: ti-sn65dsi86: Promote the AUX channel to its own sub-dev

On its own, this change looks a little strange and doesn't do too much
useful. To understand why we're doing this we need to look forward to
future patches where we're going to probe our panel using the new DP
AUX bus. See the patch ("drm/bridge: ti-sn65dsi86: Add support for the
DP AUX bus").

Let's think about the set of steps we'll want to happen when we have
the DP AUX bus:

1. We'll create the DP AUX bus.
2. We'll populate the devices on the DP AUX bus (AKA our panel).
3. For setting up the bridge-related functions of ti-sn65dsi86 we'll
   need to get a reference to the panel.

If we do #1 - #3 in a single probe call things _mostly_ will work, but
it won't be massively robust. Let's explore.

First let's think of the easy case of no -EPROBE_DEFER. In that case
in step #2 when we populate the devices on the DP AUX bus it will
actually try probing the panel right away. Since the panel probe
doesn't defer then in step #3 we'll get a reference to the panel and
we're golden.

Second, let's think of the case when the panel returns
-EPROBE_DEFER. In that case step #2 won't synchronously create the
panel (it'll just add the device to the defer list to do it
later). Step #3 will fail to get the panel and the bridge sub-device
will return -EPROBE_DEFER. We'll depopulate the DP AUX bus. Later
we'll try the whole sequence again. Presumably the panel will
eventually stop returning -EPROBE_DEFER and we'll go back to the first
case where things were golden. So this case is OK too even if it's a
bit ugly that we have to keep creating / deleting the AUX bus over and
over.

So where is the problem? As I said, it's mostly about robustness. I
don't believe that step #2 (creating the sub-devices) is really
guaranteed to be synchronous. This is evidenced by the fact that it's
allowed to "succeed" by just sticking the device on the deferred
list. If anything about the process changes in Linux as a whole and
step #2 just kicks off the probe of the DP AUX endpoints (our panel)
in the background then we'd be in trouble because we might never get
the panel in step #3.

Adding an extra sub-device means we just don't need to worry about
it. We'll create the sub-device for the DP AUX bus and it won't go
away until the whole ti-sn65dsi86 driver goes away. If the bridge
sub-device defers (maybe because it can't find the panel) that won't
depopulate the DP AUX bus and so we don't need to worry about it.

NOTE: there's a little bit of a trick here. Though the AUX channel can
run without the MIPI-to-eDP bits of the code, the MIPI-to-eDP bits
can't run without the AUX channel. We could come up a complicated
signaling scheme (have the MIPI-to-eDP bits return EPROBE_DEFER for a
while or wait on some sort of completion), but it seems simple enough
to just not even bother creating the bridge device until the AUX
channel probes. That's what we'll do.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210611101711.v10.7.If89144992cb9d900f8c91a8d1817dbe00f543720@changeid

drm/panel: panel-simple: Stash DP AUX bus; allow using it for DDC

If panel-simple is instantiated as a DP AUX bus endpoint then we have
access to the DP AUX bus. Let's stash it in the panel-simple
structure, leaving it NULL for the cases where the panel is
instantiated in other ways.

If we happen to have access to the DP AUX bus and we weren't provided
the ddc-i2c-bus in some other manner, let's use the DP AUX bus for it.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210611101711.v10.6.I18e60221f6d048d14d6c50a770b15f356fa75092@changeid

drm/panel: panel-simple: Allow panel-simple be a DP AUX endpoint device

The panel-simple driver can already have devices instantiated as
platform devices or MIPI DSI devices. Let's add a 3rd way to
instantiate it: as DP AUX endpoint devices.

At the moment there is no benefit to instantiating it in this way,
but:
- In the next patch we'll give it access to the DDC channel via the DP
  AUX bus.
- Possibly in the future we may use this channel to configure the
  backlight.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210611101711.v10.5.Iada41f76a7342354bae929d0bb3ceba40f27f0ea@changeid