drm/nouveau: remove nouveau_chan.device

nouveau_chan.device is always the same as nouveau_chan.cli.device,
so there's no need to store it separately.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-33-bskeggs@nvidia.com

drm/nouveau: pass cli to nouveau_channel_new() instead of drm+device

Both of these are stored in nouveau_cli already, and also allows the
removal of some void casts.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-32-bskeggs@nvidia.com

drm/nouveau: pass drm to nv50_dmac_create(), rather than device+disp

Both of these are stored in nouveau_drm already.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-31-bskeggs@nvidia.com

drm/nouveau: pass drm to nouveau_mem_new(), instead of cli

The nouveau_cli pointer is only ever used to eventually access
nouveau_drm, so just store it directly.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-30-bskeggs@nvidia.com

drm/nouveau: add nvif_mmu to nouveau_drm

This allocates a new nvif_mmu in nouveau_drm, and uses it for TTM
backend memory allocations instead of nouveau_drm.master.mmu,
which is removed by a later commit that removes nouveau_drm.master
entirely.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-29-bskeggs@nvidia.com

drm/nouveau: move nvxx_* definitions to nouveau_drv.h

These are some dodgy "convenience" macros for the DRM driver to peek
into NVKM state.  They're still used in a few places, but don't belong
in nvif/device.h in any case.

Move them to nouveau_drv.h, and modify callers to pass a nouveau_drm
instead of an nvif_device.

v2:
- use drm->nvkm pointer for nvxx_*() macros, removing some void*
v3:
- add some explanation of the nvxx_*() macros

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-28-bskeggs@nvidia.com

drm/nouveau/nvif: remove disp chan rd/wr

There's no good reason the ioremap() that results from nvif_object_map()
should fail, so add a check that the map succeeded, and remove the rd/wr
methods from display channel objects.

As this was the last user of rd/wr methods, the nvif plumbing is removed
at the same time.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-27-bskeggs@nvidia.com

drm/nouveau/nvif: remove device rd/wr

The previous commit ensures the device is always mapped, so these
are unneeded.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-26-bskeggs@nvidia.com

drm/nouveau: always map device

The next commit removes the nvif rd/wr methods from nvif_device, which
were probably a bad idea, and mostly intended as a fallback if ioremap()
failed (or wasn't available, as was the case in some tools I once used).

The nv04 KMS driver already mapped the device, because it's mostly been
kept alive on life-support for many years and still directly bashes PRI
a lot for modesetting.

Post-nv50, I tried pretty hard to keep PRI accesses out of the DRM code,
but there's still a few random places where we do, and those were using
the rd/wr paths prior to this commit.

This allocates and maps a new nvif_device (which will replace the usage
of nouveau_drm.master.device later on), and replicates this pointer to
all other possible users.

This will be cleaned up by the end of another patch series, after it's
been made safe to do so.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-25-bskeggs@nvidia.com

drm/nouveau/nvif: remove device args

These were once used by used by userspace tools (with nvkm built as a
library), to access multiple GPUs from a single nvif_client.

The DRM code just uses the driver's default device, so remove the
arguments.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-24-bskeggs@nvidia.com

drm/nouveau/nvif: remove client fini

Does nothing.  Remove it.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-23-bskeggs@nvidia.com

drm/nouveau/nvif: remove client devlist

This was once used by userspace tools (with nvkm built as a library),
but is now unused.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-22-bskeggs@nvidia.com

drm/nouveau/nvif: remove client version

This is not, and has never, been used for anything.  Remove it.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-21-bskeggs@nvidia.com

drm/nouveau/nvif: remove client device arg

This was once used by userspace tools (with nvkm built as a library), as
a way to select a "default device".

The DRM code doesn't need this at all as clients only have access to a
single device already, so inherit the value from its parent.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-20-bskeggs@nvidia.com

drm/nouveau/nvif: remove driver keep/fini

These are remnants of code long gone.  Remove them.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-19-bskeggs@nvidia.com

drm/nouveau/nvif: remove nvxx_client()

Make use of nouveau_cli.name instead of nvkm_client.name.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-18-bskeggs@nvidia.com

drm/nouveau/nvif: remove nvxx_object()

This hasn't been used in a while.

Moves io accessors from nvkm/core/os.h to nvif/os.h at the same time to
fix a compile issue that results from <nvkm/core/object.h> no longer
being included.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-17-bskeggs@nvidia.com

drm/nouveau/nvif: remove route/token

These were a cludge used to prevent userspace's nvif ioctl from
accessing objects created by the kernel for the same client.

That interface was removed in a previous patch, so these are no
longer useful for anything.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-16-bskeggs@nvidia.com

drm/nouveau/nvif: remove support for userspace backends

The tools that used libnvkm no longer exist.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-15-bskeggs@nvidia.com

drm/nouveau/nvkm: remove nvkm_client_search()

Has been unused for a while now.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-14-bskeggs@nvidia.com

drm/nouveau/nvkm: remove perfmon

This has never really been used for anything, in part due to never
having reclocking stable enough in general to attempt to implement
dynamic clock changes based on load, etc.

To avoid having to rework its interfaces, remove it entirely.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-13-bskeggs@nvidia.com

drm/nouveau/nvkm: remove detect/mmio/subdev_mask from device args

All callers now pass "detect=true, mmio=true, subdev_mask=~0ULL",
so remove the function arguments, and associated code.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-12-bskeggs@nvidia.com

drm/nouveau: remove abi16->handles

Hasn't been needed since 2015...

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-11-bskeggs@nvidia.com

drm/nouveau: remove abi16->device

The previous commit removes the last remnants of userspace's own nvif
instance, so this isn't needed anymore to hide the abi16 objects from
userspace and we can use nouveau_cli.device instead.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-10-bskeggs@nvidia.com

drm/nouveau: handle limited nvif ioctl in abi16

nouveau_usif.c was already stripped right back a couple of years ago,
limiting what userspace could do with it.

A follow-on series removes the nvkm side of these interfaces entirely,
in order to make it less of a nightmare to add/change internal APIs in
the future.

Unfortunately.  Userspace uses some of this.

Fortunately, userspace only ever ended up using a fraction of the APIs,
so those are reimplemened here in a more direct manner, and return
-EINVAL to userspace for everything else.

v2:
- simplified struct nouveau_abi16_obj
- added a couple of comments
v3:
- comment harder

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-9-bskeggs@nvidia.com

drm/nouveau: add nouveau_cli to nouveau_abi16

Store a pointer to struct nouveau_cli in struct nouveau_abi16 to
avoid some dubious void casts.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-8-bskeggs@nvidia.com

drm/nouveau: move allocation of root client out of nouveau_cli_init()

drm->master isn't really a nouveau_cli, and is mostly just used to get
at an nvif_mmu object to implement a hack around issues with the ioctl
interface to nvkm.

Later patches in this series allocate nvif_device/mmu objects in
nouveau_drm directly, removing the need for master.

A pending series to remove the "ioctl" layer between DRM and NVKM
removes the need for the above-mentioned hack entirely.

The only other member of drm->master that's needed is the nvif_client,
and is a dependency of device/mmu.  So the first step is to move its
allocation out of code handling nouveau_cli init.

v2:
- modified slightly due to the addition of tegra/pci cleanup patches
v3:
- move nvif init below drm_dev_alloc() to avoid changing nouveau_name()

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-7-bskeggs@nvidia.com

drm/nouveau: store nvkm_device pointer in nouveau_drm

There's various different places in the drm code that get at the
nvkm_device via various creative (and not very type-safe) means.

One of those being via nvif_device.object.priv.

Another patch series is going to entirely remove the ioctl-like
interfaces beween the drm code and nvkm, and that field will no
longer exist.

This provides a safer replacement for accessing the nvkm_device,
and will used more in upcoming patches to cleanup other cases.

v2:
- fixup printk macros to not oops if used before client ctor

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-6-bskeggs@nvidia.com

drm/nouveau: create pci device once

HW isn't touched anymore (aside from detection) until the first
nvif_device has been allocated, so we no longer need a separate
probe-only step before kicking efifb (etc) off the HW.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-5-bskeggs@nvidia.com

drm/nouveau: replace drm_device* with nouveau_drm* as dev drvdata

We almost always want to cast the pointer from dev_get_drvdata() to
'struct nouveau_drm *', so just directly store that pointer instead,
simplifying callers, and fixing some clumsy naming of dev/drm_dev
variables at the same time.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-4-bskeggs@nvidia.com

drm/nouveau: handle pci/tegra drm_dev_{alloc, register} from common code

Unify some more of the PCI/Tegra DRM driver init, both as a general
cleanup, and because a subsequent commit changes the pointer stored
via dev_set_drvdata(), and this allows the change to be made in one
place.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-3-bskeggs@nvidia.com

drm/nouveau: move nouveau_drm_device_fini() above init()

The next commit wants to be able to call fini() from an init() failure
path to remove the need to duplicate a bunch of cleanup.

Moving fini() above init() avoids the need for a forward-declaration.

Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726043828.58966-2-bskeggs@nvidia.com

drm/vblank: add dynamic per-crtc vblank configuration support

We would like to be able to enable vblank_disable_immediate
unconditionally, however there are a handful of cases where a small off
delay is necessary (e.g. with PSR enabled). So, we would like to be able
to adjust the vblank off delay and disable imminent values dynamically
for a given CRTC. Since, it will allow drivers to apply static screen
optimizations more quickly and consequently allow users to benefit more
so from the power savings afforded by the aforementioned optimizations,
while avoiding issues in cases where an off delay is still warranted.
In particular, the PSR case requires a small off delay of 2 frames,
otherwise display firmware isn't able to keep up with all of the
requests made to amdgpu. So, introduce drm_crtc_vblank_on_config() which
is like drm_crtc_vblank_on(), but it allows drivers to specify the
vblank CRTC configuration before enabling vblanking support for a given
CRTC.

Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20240725205109.209743-1-hamza.mahfooz@amd.com

drm/panel-edp: Add CSW MNB601LS1-4

Add support for the CSW MNB601LS1-4, pleace the EDID here for
subsequent reference.
00 ff ff ff ff ff ff 00 0e 77 04 11 00 00 00 00
00 22 01 04 a5 1a 0e 78 03 a1 35 9b 5e 58 91 25
1c 50 54 00 00 00 01 01 01 01 01 01 01 01 01 01
01 01 01 01 01 01 09 1e 56 dc 50 00 28 30 30 20
36 00 00 90 10 00 00 1a 00 00 00 fd 00 28 3c 30
30 08 01 0a 20 20 20 20 20 20 00 00 00 fe 00 43
53 4f 54 20 54 39 0a 20 20 20 20 20 00 00 00 fe
00 4d 4e 42 36 30 31 4c 53 31 2d 34 0a 20 00 20

Signed-off-by: Haikun Zhou <zhouhaikun5@huaqin.corp-partner.google.com>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240725115229.3416028-1-zhouhaikun5@huaqin.corp-partner.google.com

accel/qaic: Remove the description of DRM_IOCTL_QAIC_PART_DEV

The partition device ioctl was removed during the development of the
initial version of qaic driver. Remove its description from the
documentation to avoid confusing readers.

Signed-off-by: Zenghui Yu <yuzenghui@huawei.com>
Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
Signed-off-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240716073036.453-1-yuzenghui@huawei.com

drm/ci: update link to Gitlab server

Before building an image, the build script looks to see if there are fixes
to apply from an upstream repository. The link for the upstream repository
git://anongit.freedesktop.org/drm/drm became obsolete with the move to
Gitlab server in March 2024. Until recently, this obsolete link was
harmless because anongit would at least respond that there were no such
fixes available. In the last few days anongit has stopped responding to
requests causing the build script to hang indefinitely.

Update the link from anongit to the Gitlab server to prevent the build
script from hanging indefinitely.

Signed-off-by: Deborah Brouwer <deborah.brouwer@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Acked-by: Helen Koike <helen.koike@collabora.com>
Signed-off-by: Helen Koike <helen.koike@collabora.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240717235221.64629-1-deborah.brouwer@collabora.com

drm/panel: nt36672e: Break some CMDS into helper functions

Break select page cmds and reload cmds into helper functions.

Signed-off-by: Cong Yang <yangcong5@huaqin.corp-partner.google.com>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20240710084715.1119935-4-yangcong5@huaqin.corp-partner.google.com
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240710084715.1119935-4-yangcong5@huaqin.corp-partner.google.com

drm/panel: nt35521: Break some CMDS into helper functions

Break select page cmds into helper functions.

Signed-off-by: Cong Yang <yangcong5@huaqin.corp-partner.google.com>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20240710084715.1119935-3-yangcong5@huaqin.corp-partner.google.com
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240710084715.1119935-3-yangcong5@huaqin.corp-partner.google.com

drm/panel: boe-tv101wum-nl6: Break some CMDS into helper functions

hj110iz-01a and tv110c9m-ll3 both nt36523 controller, and they have
some common cmds, so let's break them into helper functions.

Signed-off-by: Cong Yang <yangcong5@huaqin.corp-partner.google.com>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20240710084715.1119935-2-yangcong5@huaqin.corp-partner.google.com
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240710084715.1119935-2-yangcong5@huaqin.corp-partner.google.com

drm/test: use kunit action wrapper macro in the gem shmem test suite

Replace deferred action function wrappers with equivalent ones defined
using the macro introduced by commit 56778b49c9a2 ("kunit: Add a macro to
wrap a deferred action function")

Signed-off-by: Marco Pagani <marpagan@redhat.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20240723191036.131286-1-marpagan@redhat.com

drm/scheduler: remove full_recover from drm_sched_start

This was basically just another one of amdgpus hacks. The parameter
allowed to restart the scheduler without turning fence signaling on
again.

That this is absolutely not a good idea should be obvious by now since
the fences will then just sit there and never signal.

While at it cleanup the code a bit.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240722083816.99685-1-christian.koenig@amd.com

drm/bridge: it6505: Disable IRQ when powered off

When the bridge is powered off, disable the IRQ until the next power on
to workaround an interrupt storm on some badly-designed hardware.

Signed-off-by: Pin-yen Lin <treapking@chromium.org>
Reviewed-by: Robert Foss <rfoss@kernel.org>
Signed-off-by: Robert Foss <rfoss@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240719110623.1850021-1-treapking@chromium.org

drm/virtio: Add DRM capset definition

Define DRM native context capset in the VirtIO-GPU protocol header.

Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240714205502.3409718-1-dmitry.osipenko@collabora.com

drm/panel/panel-ilitek-ili9806e: Add Densitron DMT028VGHMCMI-1D TFT to ILI9806E DSI TCON driver

Add Densitron DMT028VGHMCMI-1D 480x640 TFT matrix 2.83 inch panel
attached to Ilitek ILI9806E DSI TCON into the ILI9806E driver.

Note that the Densitron panels use different TCONs, this driver is for
the later panel, use panel-ilitek-st7701.c for the former panel:
DMT028VGHMCMI-1A - ST7701
DMT028VGHMCMI-1D - ILI9806E

Signed-off-by: Marek Vasut <marex@denx.de>
Reviewed-by: Michael Walle <mwalle@kernel.org>
Link: https://lore.kernel.org/r/20240724005700.196073-2-marex@denx.de
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240724005700.196073-2-marex@denx.de

dt-bindings: display: panel: Document Densitron DMT028VGHMCMI-1D TFT on ILI9806E DSI TCON

Document Densitron DMT028VGHMCMI-1D 480x640 TFT matrix 2.83 inch panel
attached to Ilitek ILI9806E DSI TCON in the ILI9806E bindings.

Signed-off-by: Marek Vasut <marex@denx.de>
Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://lore.kernel.org/r/20240724005700.196073-1-marex@denx.de
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240724005700.196073-1-marex@denx.de

drm/panel: boe-th101mb31ig002 : using drm_connector_helper_get_modes_fixed()

Use public functions( drm_connector_helper_get_modes_fixed()) to
get porch parameters.

Signed-off-by: Zhaoxiong Lv <lvzhaoxiong@huaqin.corp-partner.google.com>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20240723062615.14796-3-lvzhaoxiong@huaqin.corp-partner.google.com
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240723062615.14796-3-lvzhaoxiong@huaqin.corp-partner.google.com

drm/panel: boe-th101mb31ig002 : Fix the way to get porch parameters

The current driver can only obtain the porch parameters
of boe-th101mb31ig002. Modify it to obtain the porch
parameters of the panel currently being used.

Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Zhaoxiong Lv <lvzhaoxiong@huaqin.corp-partner.google.com>
Link: https://lore.kernel.org/r/20240723062615.14796-2-lvzhaoxiong@huaqin.corp-partner.google.com
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240723062615.14796-2-lvzhaoxiong@huaqin.corp-partner.google.com

drm: panel: boe-bf060y8m-aj0: Enable prepare_prev_first

Enable prepare_prev_first flag for BF060Y8M-AJ0 panel so that the
previous controller should be prepared first before the prepare for
the panel is called.

This makes sure that the previous controller, likely to be a DSI host
controller should be initialized to LP-11 before the panel is powered up.

Suggested-by: Marijn Suijten <marijn.suijten@somainline.org>
Signed-off-by: Dang Huynh <danct12@riseup.net>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20240722-bf060y8m-aj0-prepare-prev-v1-1-0e51c0f59d22@riseup.net
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240722-bf060y8m-aj0-prepare-prev-v1-1-0e51c0f59d22@riseup.net

drm/dp_mst: Simplify character output in drm_dp_mst_dump_topology()

Single characters should be put into a sequence.
Thus use the corresponding function “seq_putc” for one selected call.

This issue was transformed by using the Coccinelle software.

Suggested-by: Christophe Jaillet <christophe.jaillet@wanadoo.fr>
Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/af602b2e-4d92-4c54-9884-7db84700aa93@web.de

drm/bridge: tc358767: Add configurable default preemphasis

Make the default DP port preemphasis configurable via new DT property
"toshiba,pre-emphasis". This is useful in case the DP link properties
are known and starting link training from preemphasis setting of 0 dB
is not useful. The preemphasis can be set separately for both DP lanes
in range 0=0dB, 1=3.5dB, 2=6dB .

Acked-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Marek Vasut <marex@denx.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20240708150130.54484-2-marex@denx.de

dt-bindings: display: bridge: tc358867: Document default DP preemphasis

Document default DP port preemphasis configurable via new DT property
"toshiba,pre-emphasis". This is useful in case the DP link properties
are known and starting link training from preemphasis setting of 0 dB
is not useful. The preemphasis can be set separately for both DP lanes
in range 0=0dB, 1=3.5dB, 2=6dB .

This is an endpoint property, not a port property, because the TC9595
datasheet does mention that the DP might operate in some sort of split
mode, where each DP lane is used to feed one display, so in that case
there might be two endpoints.

Signed-off-by: Marek Vasut <marex@denx.de>
Reviewed-by: Rob Herring (Arm) <robh@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240708150130.54484-1-marex@denx.de

drm/ttm: Allow direct reclaim to allocate local memory v2

Limiting the allocation of higher order pages to the closest NUMA node
and enabling direct memory reclaim provides not only failsafe against
situations when memory becomes too much fragmented and the allocator is
not able to satisfy the request from the local node but falls back to
remote pages (HUGEPAGE) but also offers performance improvement.
Accessing remote pages suffers due to bandwidth limitations and could be
avoided if memory becomes defragmented and in most cases without using
manual compaction. (/proc/sys/vm/compact_memory)

Note: On certain distros such as RHEL, the proactive compaction is
disabled. (https://tinyurl.com/4f32f7rs)

v2 (chk): drop __GFP_RECLAIM since that is already set by GFP_USER

Cc: Dave Airlie <airlied@redhat.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Rajneesh Bhardwaj <rajneesh.bhardwaj@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240708160636.1147308-1-rajneesh.bhardwaj@amd.com
Signed-off-by: Christian König <christian.koenig@amd.com>

dma-buf: heaps: Deduplicate docs and adopt common format

The docs for dma_heap_get_name were incorrect, and since they were
duplicated in the header they were wrong there too.

The docs formatting was inconsistent so I tried to make it more
consistent across functions since I'm already in here doing cleanup.

Remove multiple unused includes and alphabetize.

Signed-off-by: T.J. Mercier <tjmercier@google.com>
Signed-off-by: Yong Wu <yong.wu@mediatek.com>
[Yong: Just add a comment for "priv" to mute build warning]
Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240720071606.27930-5-yunfei.dong@mediatek.com
Signed-off-by: Christian König <christian.koenig@amd.com>

drm/nouveau: use GEM references instead of TTMs

TTM wants to get rid of the duplicate refcounting of the embedded GEM
object and its own reference count.

Hence, use of GEM object references where possible.

Also get rid of nouveau_bo_ref() and replace it with nouveau_bo_fini(),
which drops the initial reference we get from initializing a ttm_bo.

Reviewed-by: Ben Skeggs <bskeggs@nvidia.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240718165959.3983-4-dakr@kernel.org

drm/nouveau: bo: remove unused functions

nouveau_bo_new_pin_map() and nouveau_bo_unmap_unpin_unref() are unused,
hence remove them.

Reviewed-by: Ben Skeggs <bskeggs@nvidia.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240718165959.3983-3-dakr@kernel.org

drm/nouveau: prime: fix refcount underflow

Calling nouveau_bo_ref() on a nouveau_bo without initializing it (and
hence the backing ttm_bo) leads to a refcount underflow.

Instead of calling nouveau_bo_ref() in the unwind path of
drm_gem_object_init(), clean things up manually.

Fixes: ab9ccb96a6e6 ("drm/nouveau: use prime helpers")
Reviewed-by: Ben Skeggs <bskeggs@nvidia.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240718165959.3983-2-dakr@kernel.org

drm/panel-edp: Add 6 panels used by MT8186 Chromebooks

The raw EDIDs for each panel:

AUO
- B116XTN02.3
00 ff ff ff ff ff ff 00 06 af aa 73 00 00 00 00
00 21 01 04 95 1a 0e 78 02 6b f5 91 55 54 91 27
22 50 54 00 00 00 01 01 01 01 01 01 01 01 01 01
01 01 01 01 01 01 ce 1d 56 e2 50 00 1e 30 26 16
36 00 00 90 10 00 00 18 df 13 56 e2 50 00 1e 30
26 16 36 00 00 90 10 00 00 18 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 02
00 10 48 ff 0f 3c 7d 50 05 18 7d 20 20 20 00 67
- B116XAN06.1
00 ff ff ff ff ff ff 00 06 af 99 a1 00 00 00 00
00 1f 01 04 95 1a 0e 78 02 9e a5 96 59 58 96 28
1b 50 54 00 00 00 01 01 01 01 01 01 01 01 01 01
01 01 01 01 01 01 ce 1d 56 ea 50 00 1a 30 30 20
46 00 00 90 10 00 00 18 df 13 56 ea 50 00 1a 30
30 20 46 00 00 90 10 00 00 18 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 02
00 10 48 ff 0f 3c 7d 0c 0a 2a 7d 20 20 20 00 3a
- B116XAT04.1
00 ff ff ff ff ff ff 00 06 af b4 c4 00 00 00 00
12 22 01 04 95 1a 0e 78 02 9e a5 96 59 58 96 28
1b 50 54 00 00 00 01 01 01 01 01 01 01 01 01 01
01 01 01 01 01 01 ce 1d 56 ea 50 00 1a 30 30 20
46 00 00 90 10 00 00 18 df 13 56 ea 50 00 1a 30
30 20 46 00 00 90 10 00 00 18 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 02
00 10 48 ff 0f 3c 7d 0c 0a 2a 7d 20 20 20 00 e7

BOE
- NV116WHM-A4D
00 ff ff ff ff ff ff 00 09 e5 fa 0c 00 00 00 00
12 22 01 04 95 1a 0e 78 03 0b 55 9a 5f 58 95 28
1e 50 54 00 00 00 01 01 01 01 01 01 01 01 01 01
01 01 01 01 01 01 96 1d 56 c8 50 00 26 30 30 20
36 00 00 90 10 00 00 1a b9 13 56 c8 50 00 26 30
30 20 36 00 00 90 10 00 00 1a 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 02
00 0d 40 ff 0a 3c 7d 0f 0c 17 7d 00 00 00 00 1a

CMN
- N116BCA-EA2
00 ff ff ff ff ff ff 00 0d ae 5d 11 00 00 00 00
0f 21 01 04 95 1a 0e 78 03 67 75 98 59 53 90 27
1c 50 54 00 00 00 01 01 01 01 01 01 01 01 01 01
01 01 01 01 01 01 da 1d 56 e2 50 00 20 30 30 20
a6 00 00 90 10 00 00 1a e7 13 56 e2 50 00 20 30
30 20 a6 00 00 90 10 00 00 1a 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 02
00 0c 3d ff 0d 3c 7d 0d 0a 15 7d 00 00 00 00 0f
- N116BCP-EA2
00 ff ff ff ff ff ff 00 0d ae 61 11 00 00 00 00
0f 21 01 04 95 1a 0e 78 03 67 75 98 59 53 90 27
1c 50 54 00 00 00 01 01 01 01 01 01 01 01 01 01
01 01 01 01 01 01 da 1d 56 e2 50 00 20 30 30 20
a6 00 00 90 10 00 00 1a e7 13 56 e2 50 00 20 30
30 20 a6 00 00 90 10 00 00 1a 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 02
00 0c 3d ff 0d 3c 7d 0d 0a 15 7d 00 00 00 00 0b

Signed-off-by: Terry Hsiao <terry_hsiao@compal.corp-partner.google.com>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240721100449.8280-1-terry_hsiao@compal.corp-partner.google.com

drm/panel: Avoid warnings w/ panel-simple/panel-edp at shutdown

At shutdown if you've got a _properly_ coded DRM modeset driver then
you'll get these two warnings at shutdown time:

  Skipping disable of already disabled panel
  Skipping unprepare of already unprepared panel

These warnings are ugly and sound concerning, but they're actually a
sign of a properly working system. That's not great.

We're not ready to get rid of the calls to drm_panel_disable() and
drm_panel_unprepare() because we're not 100% convinced that all DRM
modeset drivers are properly calling drm_atomic_helper_shutdown() or
drm_helper_force_disable_all() at the right times. However, having the
warning show up for correctly working systems is bad.

As a bit of a workaround, add some "if" tests to try to avoid the
warning on correctly working systems. Also add some comments and
update the TODO items in the hopes that future developers won't be too
confused by what's going on here.

Suggested-by: Daniel Vetter <daniel@ffwll.ch>
Acked-by: Neil Armstrong <neil.armstrong@linaro.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240621134427.1.Ieb287c2c3ee3f6d3b0d5f49b29f746b93621749c@changeid

mtd: mtdoops: Fix kmsgdump parameter renaming.

When the kmsg_dumper callback parameter changed, the reason variable
in mtdoops_do_dump() was not updated accordingly.
This breaks the build with mtdoops.

Fixes: e1a261ba599e ("printk: Add a short description string to kmsg_dump()")
Suggested-by: Knop Ryszard <ryszard.knop@intel.com>
Signed-off-by: Jocelyn Falempe <jfalempe@redhat.com>
Acked-by: Richard Weinberger <richard@nod.at>
Link: https://patchwork.freedesktop.org/patch/msgid/20240719152542.1554440-1-jfalempe@redhat.com

fbcon: Use oops_in_progress instead of panic_cpu

Panic_cpu is not exported, so it can't be used if fbcon is used as
a module. Use oops_in_progress in this case, but non-fatal oops won't
be printed.

Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202407210203.2ISiIC9m-lkp@intel.com/
Signed-off-by: Jocelyn Falempe <jfalempe@redhat.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20240722114800.174558-1-jfalempe@redhat.com

drm/panic: Add missing static inline to drm_panic_is_enabled()

This breaks build if DRM_PANIC is not enabled.
Also add the missing include drm_crtc_internal.h in drm_panic.c

Fixes: 9f774c42a908 ("drm/panic: Add drm_panic_is_enabled()")
Signed-off-by: Jocelyn Falempe <jfalempe@redhat.com>
Reviewed-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240719122051.1507927-1-jfalempe@redhat.com

drm/stm: add COMMON_CLK dependency

The added lvds driver and a change in the dsi driver resulted in failed
builds when COMMON_CLK is disabled:

x86_64-linux-ld: drivers/gpu/drm/stm/dw_mipi_dsi-stm.o: in function `dw_mipi_dsi_stm_remove':
dw_mipi_dsi-stm.c:(.text+0x51e): undefined reference to `clk_hw_unregister'

x86_64-linux-ld: drivers/gpu/drm/stm/lvds.o: in function `lvds_remove':
lvds.c:(.text+0xe3): undefined reference to `of_clk_del_provider'
x86_64-linux-ld: lvds.c:(.text+0xec): undefined reference to `clk_hw_unregister'
x86_64-linux-ld: drivers/gpu/drm/stm/lvds.o: in function `lvds_pll_config':
lvds.c:(.text+0xb5d): undefined reference to `clk_hw_get_rate'
x86_64-linux-ld: drivers/gpu/drm/stm/lvds.o: in function `lvds_probe':
lvds.c:(.text+0x1476): undefined reference to `clk_hw_register'
x86_64-linux-ld: lvds.c:(.text+0x148b): undefined reference to `of_clk_hw_simple_get'
x86_64-linux-ld: lvds.c:(.text+0x1493): undefined reference to `of_clk_add_hw_provider'
x86_64-linux-ld: lvds.c:(.text+0x1535): undefined reference to `clk_hw_unregister'

Add this as a dependency for the stm driver itself, since it will be
required in practice anyway.

Fixes: 185f99b61442 ("drm/stm: dsi: expose DSI PHY internal clock")
Fixes: aca1cbc1c986 ("drm/stm: lvds: add new STM32 LVDS Display Interface Transmitter driver")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20240719075454.3595358-1-arnd@kernel.org
Signed-off-by: Raphael Gallais-Pou <raphael.gallais-pou@foss.st.com>

drm/panic: Remove build time dependency with FRAMEBUFFER_CONSOLE

Now that fbcon has the skip_panic option, there is no more conflicts
between drm_panic and fbcon.
Remove the build time dependency, so they can be both enabled.

Signed-off-by: Jocelyn Falempe <jfalempe@redhat.com>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20240717090102.968152-4-jfalempe@redhat.com

drm/fb-helper: Set skip_panic if the drm driver supports drm panic

fb_info->skip_panic flag prevents fbcon from writing to the
framebuffer during a kernel panic. So set it when the drm driver
implements drm panic, to avoid overwriting the panic screen.

Signed-off-by: Jocelyn Falempe <jfalempe@redhat.com>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20240717090102.968152-3-jfalempe@redhat.com

fbcon: Add an option to disable fbcon in panic

This is required to avoid conflict between DRM_PANIC, and fbcon. If
a drm device already handle panic with drm_panic, it should set
the skip_panic field in fb_info, so that fbcon will stay quiet, and
not overwrite the panic_screen.

Signed-off-by: Jocelyn Falempe <jfalempe@redhat.com>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20240717090102.968152-3-jfalempe@redhat.com

drm/panic: Add drm_panic_is_enabled()

It allows to check if the drm device supports drm_panic.
Prepare the work to have better integration with fbcon and vtconsole.

Signed-off-by: Jocelyn Falempe <jfalempe@redhat.com>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20240717090102.968152-2-jfalempe@redhat.com

drm/stm: ltdc: remove reload interrupt

The reload interrupt is not used by the driver. To avoid
unnecessary calls of the interrupt routine, don't enable it.
Solve small typo and add mask to simplify the driver.

Signed-off-by: Yannick Fertre <yannick.fertre@foss.st.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240712131453.98597-1-yannick.fertre@foss.st.com
Signed-off-by: Raphael Gallais-Pou <raphael.gallais-pou@foss.st.com>

drm/stm: ltdc: add mask for lxcr register

The purpose of this mask is to simplify writing to the lxcr
register and not to forget any fields.

Signed-off-by: Yannick Fertre <yannick.fertre@foss.st.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240712131423.98405-1-yannick.fertre@foss.st.com
Signed-off-by: Raphael Gallais-Pou <raphael.gallais-pou@foss.st.com>

drm/stm: ltdc: reset plane transparency after plane disable

The plane's opacity should be reseted while the plane
is disabled. It prevents from seeing a possible global
or layer background color set earlier.

Signed-off-by: Yannick Fertre <yannick.fertre@foss.st.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240712131344.98113-1-yannick.fertre@foss.st.com
Signed-off-by: Raphael Gallais-Pou <raphael.gallais-pou@foss.st.com>

drm/mgag200: Implement struct drm_crtc_funcs.get_vblank_timestamp

Implement struct drm_crtc_funcs.get_vblank_timestamp with the DRM
helper drm_crtc_vblank_helper_get_vblank_timestamp() with its helper
get_scanout_position. Read the scanout position from the MGAREG_VCOUNT
register.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240718104551.575912-8-tzimmermann@suse.de

drm/mgag200: Add vblank support

There's no VBLANK interrupt on Matrox chipsets. The workaround that is
being used here and in other free Matrox drivers is to program <linecomp>
to the value of <vblkstr> and enable the VLINE interrupt. This triggers
an interrupt at the time when VBLANK begins.

VLINE uses separate registers for enabling and clearing pending interrupts.
No extra synchronization between irq handler and the rest of the driver is
required.

v6:
- clear VLINE status bit before registering IRQ (Jocelyn)

v5:
- disable all interrupts before registering IRQ (Jocelyn)
- don't read from ICLEAR (Jocelyn)

v4:
- recreate patch on latest upstream
- use devm_request_irq() for managed cleanup
- fail if vblanking cannot be initialized
- rename register constants (Sam, Emil)
- clear interrupt before registering handler (Ville)
- move <linecomp> programming into separate commit
- set <linecomp> to <vblkstr>
- fix typo in commit message

v3:
- set <linecomp> to <vdisplay> + 1 to trigger at VBLANK
- expand comment on linecomp

v2:
- only signal vblank on CRTC 0
- use constants for registers and fields
- set VLINECLR before enabling interrupt
- test against STATUS and IEN in irq handler
- coding-style fixes

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Acked-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Tested-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240718104551.575912-7-tzimmermann@suse.de

drm/mgag200: Add dedicted variable for <linecomp> field

Replace the line-compare value of <vdispend> with a dedicated variable
<linecomp>. Improves readability and prepares for vblank support.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240718104551.575912-6-tzimmermann@suse.de

drm/mgag200: Add dedicated variables for blanking fields

Represent fields for horizontal and vertical blanking with <hblkstr>,
<hblkend>, <vblkstr> and <vblkend>. Aligns the code with the Matrox
programming manuals.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240718104551.575912-5-tzimmermann@suse.de

drm/mgag200: Use adjusted mode values for CRTCs

Use the values with the crtc_ prefix from struct drm_display_mode to
program hardware. The DRM core adjusted these values to the requirements
of CRTC hardware.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240718104551.575912-4-tzimmermann@suse.de

drm/mgag200: Align register field names with documentation

In mgag200_set_mode_regs(), align variable names with the field names
given in the Matrox programming manuals. Makes the code and docs grep-
able.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240718104551.575912-3-tzimmermann@suse.de

drm/mgag200: Use hexadecimal register indeces

In mgag200_set_mode_regs(), use hexadecimal indices for accessing
registers. Aligns the code with the register indices in the Matrox
programming manuals. Also convert to lower-case hexadecimal values.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240718104551.575912-2-tzimmermann@suse.de

drm/bridge: analogix: remove unused struct 'bridge_init'

commit 6a1688ae8794 ("drm/bridge: ptn3460: Convert to I2C driver model")
has dropped all the users of the struct bridge_init from the
exynos_dp_core, while retaining unused structure definition.
Later on the driver was reworked and the definition migrated
to the analogix_dp driver. Remove unused struct bridge_init definition.

Signed-off-by: Dr. David Alan Gilbert <linux@treblig.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240520125551.11534-1-linux@treblig.org
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>

drm: Add might_fault to drm_modeset_lock priming

We already teach lockdep that dma_resv nests within drm_modeset_lock,
but there's a lot more: All drm kms ioctl rely on being able to
put/get_user while holding modeset locks, so we really need a
might_fault in there too to complete the picture. Add it.

Motivated by a syzbot report that blew up on bcachefs doing an
unconditional console_lock way deep in the locking hierarchy, and
lockdep only noticing the depency loop in a drm ioctl instead of much
earlier. This annotation will make sure such issues have a much harder
time escaping.

References: https://lore.kernel.org/dri-devel/00000000000073db8b061cd43496@google.com/
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: 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
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20240710093120.732208-1-daniel.vetter@ffwll.ch

drm/bridge-connector: Fix double free in error handling paths

The recent switch to drmm allocation in drm_bridge_connector_init() may
cause double free on bridge_connector in some of the error handling
paths.

Drop the explicit kfree() calls on bridge_connector.

Fixes: c12907be57b1 ("drm/bridge-connector: switch to using drmm allocations")
Signed-off-by: Cristian Ciocaltea <cristian.ciocaltea@collabora.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Robert Foss <rfoss@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240711-bridge-connector-fix-dbl-free-v1-1-d558b2d0eb93@collabora.com

drm/bridge: Silence error messages upon probe deferral

When -EPROBE_DEFER is returned do not raise an error, but silently return
this error instead. Fixes error like this:
[drm:drm_bridge_attach] *ERROR* failed to attach bridge
  /soc@0/bus@30800000/mipi-dsi@30a00000 to encoder None-34: -517
[drm:drm_bridge_attach] *ERROR* failed to attach bridge
  /soc@0/bus@30800000/mipi-dsi@30a00000 to encoder None-34: -517

Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Reviewed-by: Robert Foss <rfoss@kernel.org>
Signed-off-by: Robert Foss <rfoss@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240703122715.4004765-1-alexander.stein@ew.tq-group.com

printk: Add a short description string to kmsg_dump()

kmsg_dump doesn't forward the panic reason string to the kmsg_dumper
callback.
This patch adds a new struct kmsg_dump_detail, that will hold the
reason and description, and pass it to the dump() callback.

To avoid updating all kmsg_dump() call, it adds a kmsg_dump_desc()
function and a macro for backward compatibility.

I've written this for drm_panic, but it can be useful for other
kmsg_dumper.
It allows to see the panic reason, like "sysrq triggered crash"
or "VFS: Unable to mount root fs on xxxx" on the drm panic screen.

v2:
 * Use a struct kmsg_dump_detail to hold the reason and description
   pointer, for more flexibility if we want to add other parameters.
   (Kees Cook)
 * Fix powerpc/nvram_64 build, as I didn't update the forward
   declaration of oops_to_nvram()

Signed-off-by: Jocelyn Falempe <jfalempe@redhat.com>
Acked-by: Petr Mladek <pmladek@suse.com>
Acked-by: Michael Ellerman <mpe@ellerman.id.au> (powerpc)
Acked-by: Kees Cook <kees@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240702122639.248110-1-jfalempe@redhat.com

drm/panel-edp: Add entry for BOE NV133WUM-N63 panel

This panel is found on some laptops e.g., variants of the Thinkpad X13s.

Signed-off-by: Clayton Craft <clayton@craftyguy.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20240520221836.16031-1-clayton@craftyguy.net
[DB: moved the entry to the proper place]
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>

drm: lcdif: Use adjusted_mode .clock instead of .crtc_clock

In case an upstream bridge modified the required clock frequency
in its .atomic_check callback by setting adjusted_mode.clock ,
make sure that clock frequency is generated by the LCDIFv3 block.

This is useful e.g. when LCDIFv3 feeds DSIM which feeds TC358767
with (e)DP output, where the TC358767 expects precise timing on
its input side, the precise timing must be generated by the LCDIF.

Signed-off-by: Marek Vasut <marex@denx.de>
Reviewed-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240531202813.277109-1-marex@denx.de

accel/ivpu: Add missing MODULE_FIRMWARE metadata

Modules that load firmware from various paths at runtime must declare
those paths at compile time, via the MODULE_FIRMWARE macro, so that the
firmware paths are included in the module's metadata.

The accel/ivpu driver loads firmware but lacks this metadata,
preventing dracut from correctly locating firmware files. Fix it.

Fixes: 9ab43e95f922 ("accel/ivpu: Switch to generation based FW names")
Fixes: 02d5b0aacd05 ("accel/ivpu: Implement firmware parsing and booting")
Signed-off-by: Alexander F. Lent <lx@xanderlent.com>
Reviewed-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240709-fix-ivpu-firmware-metadata-v3-1-55f70bba055b@xanderlent.com

drm/stm: ltdc: Remove unused function plane_to_ltdc

The function are defined in the ltdc.c file, but not called
anywhere, so delete the unused function.

drivers/gpu/drm/stm/ltdc.c:494:35: warning: unused function 'encoder_to_ltdc'.

Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Closes: https://bugzilla.openanolis.cn/show_bug.cgi?id=9403
Signed-off-by: Jiapeng Chong <jiapeng.chong@linux.alibaba.com>
Acked-by: Yannick Fertre <yannick.fertre@foss.st.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240624024113.54850-1-jiapeng.chong@linux.alibaba.com
Signed-off-by: Raphael Gallais-Pou <raphael.gallais-pou@foss.st.com>

drm/stm: Avoid use-after-free issues with crtc and plane

ltdc_load() calls functions drm_crtc_init_with_planes(),
drm_universal_plane_init() and drm_encoder_init(). These functions
should not be called with parameters allocated with devm_kzalloc()
to avoid use-after-free issues [1].

Use allocations managed by the DRM framework.

Found by Linux Verification Center (linuxtesting.org).

[1]
https://lore.kernel.org/lkml/u366i76e3qhh3ra5oxrtngjtm2u5lterkekcz6y2jkndhuxzli@diujon4h7qwb/

Signed-off-by: Katya Orlova <e.orlova@ispras.ru>
Acked-by: Raphaël Gallais-Pou <raphael.gallais-pou@foss.st.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240216125040.8968-1-e.orlova@ispras.ru
Signed-off-by: Raphael Gallais-Pou <raphael.gallais-pou@foss.st.com>

drm/stm: ltdc: check memory returned by devm_kzalloc()

devm_kzalloc() can fail and return NULL pointer. Check its return status.
Identified with Coccinelle (kmerr.cocci script).

Fixes: 484e72d3146b ("drm/stm: ltdc: add support of ycbcr pixel formats")
Signed-off-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Acked-by: Raphael Gallais-Pou <raphael.gallais-pou@foss.st.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230531072854.142629-1-claudiu.beznea@microchip.com
Signed-off-by: Raphael Gallais-Pou <raphael.gallais-pou@foss.st.com>

drm/mgag200: Rename BMC vidrst names

The BMC's scanout synchronization is only indirectly related to the
VIDRST functionality. Do some renaming.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240711072415.11831-4-tzimmermann@suse.de

drm/mgag200: Remove vidrst callbacks from struct mgag200_device_funcs

The callbacks disable_vidrst and enable_vidrst are obsolete. Remove
the fields from struct mgag200_device_funcs. Instead call their
implementations directly of the field 'has_vidrst' has been set in
struct mgag200_device_info.

Also change the logic slightly. The BMC used to start and stop scanout
during the CRTC's atomic_enable and atomic_disable. Plane updates were
done while the BMC scanned out the display. Now only stop once in
atomic_disable at the beginning of a modeset and then restart the
scanout at the end of a modeset in atomic_enable. While the modeset
takes place, the BMC does not scanout at all.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240711072415.11831-3-tzimmermann@suse.de

drm/mgag200: Only set VIDRST bits in CRTC modesetting

The VRSTEN and HRSTEN bits control whether a CRTC synchronizes its
display signal with an external source on the VIDRST pin. The G200WB
and G200EW3 models synchronize with a BMC chip, but different external
video encoders, such as the Matrox Maven, can also be attached to the
pin.

Only set VRSTEN and HRSTEN bits in the CRTC mode-setting code, so the
bits are independent from the BMC. Add the field set_vidrst to the CRTC
state for this purpose. Off by default, control the CRTC VIDRST setting
from the CRTC's atomic_check helper.

v3:
- don't clear bits unnecessary (Jocelyn)
v2:
- keep logic entirely in CRTC (Jocelyn)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240711072415.11831-2-tzimmermann@suse.de

drm/v3d: Expose memory stats through fdinfo

Use the common DRM function `drm_show_memory_stats()` to expose standard
fdinfo memory stats.

V3D exposes global GPU memory stats through debugfs. Those stats will be
preserved while the DRM subsystem doesn't have a standard solution to
expose global GPU stats.

Signed-off-by: Maíra Canal <mcanal@igalia.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240711142736.783816-1-mcanal@igalia.com

drm/v3d: Add some local variables in queries/extensions

Add some local variables to make the code a bit less verbose, with the
main benefit being pulling some lines to under 80 columns wide.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Reviewed-by: Maíra Canal <mcanal@igalia.com>
Signed-off-by: Maíra Canal <mcanal@igalia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240711135340.84617-12-tursulin@igalia.com

drm/v3d: Prefer get_user for scalar types

It makes it just a tiny bit more obvious what is going on.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Reviewed-by: Maíra Canal <mcanal@igalia.com>
Signed-off-by: Maíra Canal <mcanal@igalia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240711135340.84617-11-tursulin@igalia.com

drm/v3d: Move perfmon init completely into own unit

Now that the build time dependencies on various array sizes have been
removed, we can move the perfmon init completely into its own compilation
unit and remove the hardcoded defines.

This improves on the temporary fix quickly delivered in commit
9c3951ec27b9 ("drm/v3d: Fix perfmon build error/warning").

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
References: 9c3951ec27b9 ("drm/v3d: Fix perfmon build error/warning")
Reviewed-by: Maíra Canal <mcanal@igalia.com>
Signed-off-by: Maíra Canal <mcanal@igalia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240711135340.84617-10-tursulin@igalia.com

drm/v3d: Do not use intermediate storage when copying performance query results

Removing the intermediate buffer removes the last use of the
V3D_MAX_COUNTERS define, which will enable further driver cleanup.

While at it pull the 32 vs 64 bit copying decision outside the loop in
order to reduce the number of conditional instructions.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Maíra Canal <mcanal@igalia.com>
Signed-off-by: Maíra Canal <mcanal@igalia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240711135340.84617-9-tursulin@igalia.com

drm/v3d: Size the kperfmon_ids array at runtime

Instead of statically reserving pessimistic space for the kperfmon_ids
array, make the userspace extension code allocate the exactly required
amount of space.

Apart from saving some memory at runtime, this also removes the need for
the V3D_MAX_PERFMONS macro whose removal will benefit further driver
cleanup.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Reviewed-by: Maíra Canal <mcanal@igalia.com>
Signed-off-by: Maíra Canal <mcanal@igalia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240711135340.84617-8-tursulin@igalia.com

drm/v3d: Move part of copying of reset/copy performance extension to a helper

The loop which looks up the syncobj and copies the kperfmon ids is
identical so lets move it to a helper.

The only change is replacing copy_from_user with get_user when copying a
scalar.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Reviewed-by: Maíra Canal <mcanal@igalia.com>
Signed-off-by: Maíra Canal <mcanal@igalia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240711135340.84617-7-tursulin@igalia.com

drm/v3d: Validate passed in drm syncobj handles in the performance extension

If userspace provides an unknown or invalid handle anywhere in the handle
array the rest of the driver will not handle that well.

Fix it by checking handle was looked up successfully or otherwise fail the
extension by jumping into the existing unwind.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Fixes: bae7cb5d6800 ("drm/v3d: Create a CPU job extension for the reset performance query job")
Cc: Maíra Canal <mcanal@igalia.com>
Cc: Iago Toral Quiroga <itoral@igalia.com>
Cc: stable@vger.kernel.org # v6.8+
Reviewed-by: Maíra Canal <mcanal@igalia.com>
Signed-off-by: Maíra Canal <mcanal@igalia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240711135340.84617-6-tursulin@igalia.com

drm/v3d: Validate passed in drm syncobj handles in the timestamp extension

If userspace provides an unknown or invalid handle anywhere in the handle
array the rest of the driver will not handle that well.

Fix it by checking handle was looked up successfully or otherwise fail the
extension by jumping into the existing unwind.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Fixes: 9ba0ff3e083f ("drm/v3d: Create a CPU job extension for the timestamp query job")
Cc: Maíra Canal <mcanal@igalia.com>
Cc: Iago Toral Quiroga <itoral@igalia.com>
Cc: stable@vger.kernel.org # v6.8+
Reviewed-by: Maíra Canal <mcanal@igalia.com>
Signed-off-by: Maíra Canal <mcanal@igalia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240711135340.84617-5-tursulin@igalia.com

drm/v3d: Fix potential memory leak in the performance extension

If fetching of userspace memory fails during the main loop, all drm sync
objs looked up until that point will be leaked because of the missing
drm_syncobj_put.

Fix it by exporting and using a common cleanup helper.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Fixes: bae7cb5d6800 ("drm/v3d: Create a CPU job extension for the reset performance query job")
Cc: Maíra Canal <mcanal@igalia.com>
Cc: Iago Toral Quiroga <itoral@igalia.com>
Cc: stable@vger.kernel.org # v6.8+
Signed-off-by: Maíra Canal <mcanal@igalia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240711135340.84617-4-tursulin@igalia.com