drm/sysfb: Add efidrm for EFI displays

Add support for screen_info setups with VIDEO_TYPE_EFI. Provide the
minimum functionality of reading modes, updating and clearing the display.

There is existing support for these displays provided by simpledrm with
CONFIG_SYSFB_SIMPLEFB=y. Using efidrm over simpledrm will allows for the
mapping of video memory with correct caching. Simpledrm always assumes WC
caching, while fully cached memory is possible with efidrm. Efidrm will
also allow for the use of additional functionality provided by EFI, such
as EDID information.

In addition to efidrm, add struct pixel_format plus initializer macros.
The type and macros describe pixel formats in a generic way on order to
find the DRM format from the screen_info settings. Similar existing code
in SIMPLEFB_FORMATS and fbdev is not really what is needed in efidrm,
but SIMPLEFB_FORMATS can later be converted to struct pixel_format.

v4:
- depend on CONFIG_EFI
- disallow module for now as efi_mem_desc_lookup() is not exported
v3:
- depend on !SYSFB_SIMPLEFB (Javier)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://lore.kernel.org/r/20250401094056.32904-15-tzimmermann@suse.de

firmware: sysfb: Move bpp-depth calculation into screen_info helper

Move the calculation of the bits per pixels for screen_info into a
helper function. This will make it available to other callers besides
the firmware code.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://lore.kernel.org/r/20250401094056.32904-14-tzimmermann@suse.de

drm/sysfb: ofdrm: Add EDID support

Add EDID support to sysfb connector helpers. Read the EDID property
from the OF node in ofdrm. Without EDID, this does nothing.

Some systems with OF display, such as 32-bit PPC Macintoshs, provide
the system display's EDID data as node property in their DT. Exporting
this information allows compositors to implement correct DPI and
meaningful color management.

v3:
- avoid parser error by clearing EDID extension field
v2:
- return errno codes on errors (Jani)
- simplify EDID read logic (Jani)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://lore.kernel.org/r/20250401094056.32904-13-tzimmermann@suse.de

drm/sysfb: Merge primary-plane functions

Merge the primary plane code of ofdrm and simpledrm. Replace the
plane implementation in each driver with the shared helpers. Set
up driver callbacks and format modifiers with initializer macros.

The plane code in ofdrm and simpledrm is very similar. Ofdrm has a
more sophisticated implementation of atomic_disable, which clears
individual scanlines. The code in simpledrm clears the whole buffer
at once. Take the ofdrm version.

Simpledrm supports get_scanout_buffer. Import it into the shared
helpers, which makes it available in ofdrm.

The supported formats are all native formats plus an optional enulated
XRGB8888 if that's not already a native format. Provide an initializer
macro that computes the size of the formats array.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://lore.kernel.org/r/20250401094056.32904-12-tzimmermann@suse.de

drm/sysfb: Merge CRTC functions

Merge the CRTC functions of ofdrm and simpledrm. Replace the code
in each driver with the shared helpers. Set up callbacks with
initializer macros.

Ofdrm supports a gamma LUT, while simpledrm does not. So far ofdrm's
LUT size has been hard-coded in the driver CRTC's atomic_check helper.
Now pass the size of the LUT to the sysfb device. Ofdrm's custom
atomic_flush is still required to apply changes to the LUT. Simpledrm
passes a LUT size of 0, which disables the gamma LUT.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://lore.kernel.org/r/20250401094056.32904-11-tzimmermann@suse.de

drm/sysfb: Maintain CRTC state in struct drm_sysfb_crtc_state

Move ofdrm's struct ofdrm_crtc_state plus functions to sysfb
helpers and rename everything to drm_sysfb_crtc_state.

The sysfb CRTC state is a regular CRTC state with information on
the primary plane's color format, as required for color management.
Helpers for sysfb planes will later set this up automatically.

In ofdrm and simpledrm, replace existing code with the new helpers.
Ofdrm continues to use the CRTC state for color management. This
has no effect on simpledrm.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://lore.kernel.org/r/20250401094056.32904-10-tzimmermann@suse.de

drm/sysfb: Merge connector functions

Merge the connector functions of ofdrm and simpledrm. Replace the
code in each driver with the shared helpers. Set up callbacks with
initializer macros.

No effective code changes. The sysfb connector only returns the
preconfigured display mode.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://lore.kernel.org/r/20250401094056.32904-9-tzimmermann@suse.de

drm/sysfb: Merge mode-config functions

Provide initializer to set struct drm_mode_config_funcs. Convert
ofdrm and simpledrm.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://lore.kernel.org/r/20250401094056.32904-8-tzimmermann@suse.de

drm/sysfb: Provide single mode-init helper

Merge the mode-init functions of ofdrm and simpledrm to the new helper
drm_sysfb_mode(). Also implement the DPI defaults there. Replace the
code in each driver with the shared helper.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://lore.kernel.org/r/20250401094056.32904-7-tzimmermann@suse.de

drm/sysfb: Add struct drm_sysfb_device

Add struct drm_sysfb_device that stores the system display's hardware
settings. Further helpers for the mode-setting pipeline will use these
fields. Convert ofdrm and simpledrm by embedding the sysfb device in
their device structs.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://lore.kernel.org/r/20250401094056.32904-6-tzimmermann@suse.de

drm: Move sysfb drivers into separate subdirectory

The ofdrm and simpledrm drivers are special as they operate on
externally provided framebuffers. Move them into their own sub-
directory. Will let them share common code.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://lore.kernel.org/r/20250401094056.32904-5-tzimmermann@suse.de

drm/simpledrm: Remove struct simpledrm_device.nformats

The field nformats is unused. Remove it.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://lore.kernel.org/r/20250401094056.32904-4-tzimmermann@suse.de

drm/ofdrm: Open-code drm_simple_encoder_init()

The helper drm_simple_encoder_init() is a trivial helper around
drm_encoder_init() and therefore deprecated. Open-code the function
and remove the dependency.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://lore.kernel.org/r/20250401094056.32904-3-tzimmermann@suse.de

drm/ofdrm: Remove struct ofdrm_device.pdev

The field pdev is unused. Remove it.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://lore.kernel.org/r/20250401094056.32904-2-tzimmermann@suse.de

drm/vc4: tests: Retry pv-muxing tests when EDEADLK

Some functions used by the HVS->PV muxing tests can return with EDEADLK,
meaning the entire sequence should be restarted. It's not a fatal error
and we should treat it as a recoverable error, and recover, instead of
failing the test like we currently do.

Fixes: 76ec18dc5afa ("drm/vc4: tests: Add unit test suite for the PV muxing")
Reviewed-by: Maíra Canal <mcanal@igalia.com>
Link: https://lore.kernel.org/r/20250403-drm-vc4-kunit-failures-v2-4-e09195cc8840@kernel.org
Signed-off-by: Maxime Ripard <mripard@kernel.org>

drm/vc4: tests: Stop allocating the state in test init

The vc4-pv-muxing-combinations and vc5-pv-muxing-combinations test
suites use a common test init function which, in part, allocates the
drm atomic state the test will use.

That allocation relies on  drm_kunit_helper_atomic_state_alloc(), and
thus requires a struct drm_modeset_acquire_ctx. This context will then
be stored in the allocated state->acquire_ctx field.

However, the context is local to the test init function, and is cleared
as soon as drm_kunit_helper_atomic_state_alloc() is done. We thus end up
with an dangling pointer to a cleared context in state->acquire_ctx for
our test to consumes.

We should really allocate the context and the state in the test
functions, so we can also control when we're done with it.

Fixes: 30188df0c387 ("drm/tests: Drop drm_kunit_helper_acquire_ctx_alloc()")
Reviewed-by: Maíra Canal <mcanal@igalia.com>
Link: https://lore.kernel.org/r/20250403-drm-vc4-kunit-failures-v2-3-e09195cc8840@kernel.org
Signed-off-by: Maxime Ripard <mripard@kernel.org>

drm/vc4: tests: Document output handling functions

vc4_mock_atomic_add_output() and vc4_mock_atomic_del_output() are public
but aren't documented. Let's provide the documentation.

In particular, special care should be taken to deal with EDEADLK.

Reviewed-by: Maíra Canal <mcanal@igalia.com>
Link: https://lore.kernel.org/r/20250403-drm-vc4-kunit-failures-v2-2-e09195cc8840@kernel.org
Signed-off-by: Maxime Ripard <mripard@kernel.org>

drm/vc4: tests: Use return instead of assert

The vc4_mock_atomic_add_output() and vc4_mock_atomic_del_output() assert
that the functions they are calling didn't fail. Since some of them can
return EDEADLK, we can't properly deal with it.

Since both functions are expected to return an int, and all caller check
the return value, let's just properly propagate the errors when they
occur.

Fixes: f759f5b53f1c ("drm/vc4: tests: Introduce a mocking infrastructure")
Fixes: 76ec18dc5afa ("drm/vc4: tests: Add unit test suite for the PV muxing")
Reviewed-by: Maíra Canal <mcanal@igalia.com>
Link: https://lore.kernel.org/r/20250403-drm-vc4-kunit-failures-v2-1-e09195cc8840@kernel.org
Signed-off-by: Maxime Ripard <mripard@kernel.org>

panel/panel-edp: Use refcounted allocation in place of devm_kzalloc()

Move to using the new API devm_drm_panel_alloc() to allocate the
panel.

Signed-off-by: Anusha Srivatsa <asrivats@redhat.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20250401-b4-drm-panel-mass-driver-convert-v1-10-cdd7615e1f93@redhat.com
Signed-off-by: Maxime Ripard <mripard@kernel.org>

panel/ebbg-ft8719: Use refcounted allocation in place of devm_kzalloc()

Move to using the new API devm_drm_panel_alloc() to allocate the
panel.

Signed-off-by: Anusha Srivatsa <asrivats@redhat.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20250401-b4-drm-panel-mass-driver-convert-v1-9-cdd7615e1f93@redhat.com
Signed-off-by: Maxime Ripard <mripard@kernel.org>

panel/dsi-cm: Use refcounted allocation in place of devm_kzalloc()

Move to using the new API devm_drm_panel_alloc() to allocate the
panel.

Signed-off-by: Anusha Srivatsa <asrivats@redhat.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20250401-b4-drm-panel-mass-driver-convert-v1-8-cdd7615e1f93@redhat.com
Signed-off-by: Maxime Ripard <mripard@kernel.org>

panel/boe-tv101wum-ll2: Use refcounted allocation in place of devm_kzalloc()

Move to using the new API devm_drm_panel_alloc() to allocate the
panel.

Signed-off-by: Anusha Srivatsa <asrivats@redhat.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20250401-b4-drm-panel-mass-driver-convert-v1-7-cdd7615e1f93@redhat.com
Signed-off-by: Maxime Ripard <mripard@kernel.org>

panel/th101mb31ig002-28a: Use refcounted allocation in place of devm_kzalloc()

Move to using the new API devm_drm_panel_alloc() to allocate the
panel.

Signed-off-by: Anusha Srivatsa <asrivats@redhat.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20250401-b4-drm-panel-mass-driver-convert-v1-6-cdd7615e1f93@redhat.com
Signed-off-by: Maxime Ripard <mripard@kernel.org>

panel/bf060y8m-aj0: Use refcounted allocation in place of devm_kzalloc()

Move to using the new API devm_drm_panel_alloc() to allocate the
panel.

Signed-off-by: Anusha Srivatsa <asrivats@redhat.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20250401-b4-drm-panel-mass-driver-convert-v1-5-cdd7615e1f93@redhat.com
Signed-off-by: Maxime Ripard <mripard@kernel.org>

panel/auo-a030jtn01: Use refcounted allocation in place of devm_kzalloc()

Move to using the new API devm_drm_panel_alloc() to allocate the
panel.

Signed-off-by: Anusha Srivatsa <asrivats@redhat.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20250401-b4-drm-panel-mass-driver-convert-v1-4-cdd7615e1f93@redhat.com
Signed-off-by: Maxime Ripard <mripard@kernel.org>

panel/z00t-tm5p5-n35596: Use refcounted allocation in place of devm_kzalloc()

Move to using the new API devm_drm_panel_alloc() to allocate the
panel.

Signed-off-by: Anusha Srivatsa <asrivats@redhat.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20250401-b4-drm-panel-mass-driver-convert-v1-3-cdd7615e1f93@redhat.com
Signed-off-by: Maxime Ripard <mripard@kernel.org>

panel/arm-versatile: Use the refcounted allocation in place of devm_kzalloc()

Move to using the new API devm_drm_panel_alloc() to allocate the
panel.

Signed-off-by: Anusha Srivatsa <asrivats@redhat.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20250401-b4-drm-panel-mass-driver-convert-v1-2-cdd7615e1f93@redhat.com
Signed-off-by: Maxime Ripard <mripard@kernel.org>

panel/abt-y030xx067a: Use the refcounted allocation in place of devm_kzalloc()

Move to using the new API devm_drm_panel_alloc() to allocate the
panel.

Signed-off-by: Anusha Srivatsa <asrivats@redhat.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20250401-b4-drm-panel-mass-driver-convert-v1-1-cdd7615e1f93@redhat.com
Signed-off-by: Maxime Ripard <mripard@kernel.org>

drm/format-helper: fix build

Build fails with:

error: multiple unsequenced modifications to 'sbuf32'
[-Werror,-Wunsequenced]
  264 |                         le32_to_cpup(sbuf32++),
      |                                            ^
  265 |                         le32_to_cpup(sbuf32++),
      |                                            ~~

With that move the increment of the sbuf32 pointer to the end of the
loop, instead of inside the array list initializer, where the
order/sequence of the sbuf32 pointer modifications is not defined.

Fixes: 58523a25cbf7 ("drm/format-helper: Optimize 32-to-24-bpp conversion")
Fixes: 3f31a017ddbc ("drm/format-helper: Optimize 32-to-16-bpp conversion")
Fixes: 65931bbc5177 ("drm/format-helper: Optimize 32-to-8-bpp conversion")
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Jocelyn Falempe <jfalempe@redhat.com>
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://lore.kernel.org/r/20250402104430.142398-2-matthew.auld@intel.com

drm: renesas: Extend RZ/G2L supported KMS formats

The RZ/G2L driver utilises the VSPD to read data from input sources.

The rzg2l_du_kms component lists a restricted subset of the capabilities
of the VSPD which prevents additional formats from being used for
display planes.

The supported display plane formats are mapped in rzg2l_du_vsp_formats[].

Extend the rzg2l_du_format_infos[] table with the corresponding mappings
between the supported DRM formats and the formats exposed by the VSP in
rzg2l_du_vsp_formats, maintaining the same ordering in both tables.

The RPF module on VSPD supports various format conversion and send the
image data to BRS(Blend ROP Sub Unit) for further processing.

Signed-off-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Tested-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Tested-by: Tommaso Merciai <tommaso.merciai.xr@bp.renesas.com>
Link: https://lore.kernel.org/r/20250330102357.56010-3-biju.das.jz@bp.renesas.com

drm: renesas: rz-du: Drop bpp variable from struct rzg2l_du_format_info

Drop the unused variable bpp from struct rzg2l_du_format_info.

Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Link: https://lore.kernel.org/r/20250330102357.56010-2-biju.das.jz@bp.renesas.com

drm: renesas: rz-du: Support dmabuf import

The rz-du driver uses GEM DMA helpers, but does not implement the
drm_driver .gem_prime_import_sg_table operation. This  prevents
importing dmabufs. Fix it by implementing the missing operation using
the DRM_GEM_DMA_DRIVER_OPS_WITH_DUMB_CREATE() helper macro.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Tested-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com> # RZ/V2H + DSI
Tested-by: Tommaso Merciai <tommaso.merciai.xr@bp.renesas.com>
Reviewed-by: Biju Das <biju.das.jz@bp.renesas.com>
Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250321104615.31809-1-laurent.pinchart+renesas@ideasonboard.com

drm: renesas: rz-du: Add Kconfig dependency between RZG2L_DU and RZG2L_MIPI_DSI

Add Kconfig dependency between RZG2L_DU and RZG2L_MIPI_DSI, so that
DSI module has functional dependency on DU. It is similar way that
the R-Car MIPI DSI encoder is handled.

While at it drop ARCH_RENESAS dependency as DRM_RZG2L_DU depend on
ARCH_RZG2L.

Suggested-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240827163727.108405-1-biju.das.jz@bp.renesas.com

drm/shmem-helper: Fix unsetting shmem vaddr while vmap refcount > 0

We switched to use refcount_t for vmaps and missed to change the vunmap
code to properly unset the vmap pointer, which is now cleared while vmap's
refcount > 0. Clear the cached vmap pointer only when refcounting drops to
zero to fix the bug.

Fixes: e1fc39a92332 ("drm/shmem-helper: Use refcount_t for vmap_use_count")
Reported-by: Lucas De Marchi <lucas.demarchi@intel.com>
Closes: https://lore.kernel.org/dri-devel/20250403105053.788b0f6e@collabora.com/T/#m3dca6d81bedc8d6146a56b82694624fbc6fa4c96
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Tested-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20250403142633.484660-1-dmitry.osipenko@collabora.com

drm/nouveau: disp: Avoid -Wflex-array-member-not-at-end warning

-Wflex-array-member-not-at-end was introduced in GCC-14, and we are
getting ready to enable it, globally.

Use the `DEFINE_RAW_FLEX()` helper for an on-stack definition of
a flexible structure where the size of the flexible-array member
is known at compile-time, and refactor the rest of the code,
accordingly.

So, with these changes, fix the following warning:

drivers/gpu/drm/nouveau/dispnv50/disp.c:779:47: warning: structure containing a flexible array member is not at the end of another structure [-Wflex-array-member-not-at-end]

Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://lore.kernel.org/r/Z-2zI55Qf88jTfNK@kspp

drm/nouveau: svm: Avoid -Wflex-array-member-not-at-end warning

-Wflex-array-member-not-at-end was introduced in GCC-14, and we are
getting ready to enable it, globally.

Use the `DEFINE_RAW_FLEX()` helper for an on-stack definition of
a flexible structure where the size of the flexible-array member
is known at compile-time, and refactor the rest of the code,
accordingly.

So, with these changes, fix the following warning:

drivers/gpu/drm/nouveau/nouveau_svm.c:724:44: warning: structure containing a flexible array member is not at the end of another structure [-Wflex-array-member-not-at-end]

Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://lore.kernel.org/r/Z-2uezeHt1aaHH6x@kspp

drm/nouveau: fence: Avoid -Wflex-array-member-not-at-end warning

-Wflex-array-member-not-at-end was introduced in GCC-14, and we are
getting ready to enable it, globally.

Use the `DEFINE_RAW_FLEX()` helper for an on-stack definition of
a flexible structure where the size of the flexible-array member
is known at compile-time, and refactor the rest of the code,
accordingly.

So, with these changes, fix the following warning:

drivers/gpu/drm/nouveau/nouveau_fence.c:188:38: warning: structure containing a flexible array member is not at the end of another structure [-Wflex-array-member-not-at-end]

Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://lore.kernel.org/r/Z-2r6v-Cji7vwOsz@kspp

drm/sti: fix inconsistent indenting warning

Fix below inconsistent indenting smatch warning.
smatch warnings:
drivers/gpu/drm/sti/sti_hda.c:696 sti_hda_bind() warn: inconsistent indenting

Signed-off-by: Charles Han <hanchunchao@inspur.com>
Acked-by: Raphaël Gallais-Pou <rgallaispou@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250305101641.2399-1-hanchunchao@inspur.com
Signed-off-by: Raphael Gallais-Pou <raphael.gallais-pou@foss.st.com>

drm/dp_mst: Use drm_dp_dpcd_write_data() to write GUID for non-root MST branch devices

drm_dp_dpcd_write_data() can be used to write the GUID for a non-root
MST branch device, similarly to writing the GUID to a root MST branch
device, do so.

Cc: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Cc: Lyude Paul <lyude@redhat.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://lore.kernel.org/r/20250401103846.686408-2-imre.deak@intel.com

drm/dp_mst: Fix GUID DPCD write to non-root MST branch devices

The return value on success of drm_dp_send_dpcd_write() called for
non-root MST branch devices from drm_dp_check_mstb_guid() is the number
of bytes transferred. Atm this return value (in case of a complete read)
will be regarded incorrectly as an error by the caller of
drm_dp_check_mstb_guid(). Fix this by converting the return value for a
complete read to the expected success code (0) and for a partial read to
a failure code (-EPROTO).

Fixes: 2554da0de3e8 ("drm/display: dp-mst-topology: use new DCPD access helpers")
Cc: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Cc: Lyude Paul <lyude@redhat.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://lore.kernel.org/r/20250401103846.686408-1-imre.deak@intel.com

drm/mode_config: Make drm_mode_config.(max|min)_(width|height) unsigned

It doesn't make much sense to allow devices to specify their min/max
resolution as signed integers, and in Rust with CONFIG_RUST_OVERFLOW_CHECKS
enabled this provides us actual over/underflow checks. Similarly, it
doesn't really make much sense for us to allow devices to specify their
minimum/maximum resolution as signed.

Signed-off-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Maxime Ripard <mripard@kernel.org>
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://lore.kernel.org/r/20250331222556.454334-3-lyude@redhat.com

drm/edid: Use unsigned int in drm_add_modes_noedid()

A negative resolution doesn't really make any sense, so let's make these
parameters unsigned. In C this doesn't make much of a difference, but Rust
is stricter about signed/unsigned casts and additionally can check for
arithmetic over/underflows if CONFIG_RUST_OVERFLOW_CHECKS is enabled.

Signed-off-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Maxime Ripard <mripard@kernel.org>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://lore.kernel.org/r/20250331222556.454334-2-lyude@redhat.com

drm/bridge: it6505: Switch to common helpers to power up/down dp link

Use the common dp link power up/down helpers to avoid duplicating code.

Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Link: https://lore.kernel.org/r/20250318063452.4983-5-andyshrk@163.com
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

drm/bridge: anx78xx: Switch to common helpers to power up/down dp link

Use the common dp link power up/down helpers to avoid duplicating code.

Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Link: https://lore.kernel.org/r/20250318063452.4983-4-andyshrk@163.com
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

drm/bridge: anx6345: Switch to common helpers to power up/down dp link

Use the common dp link power up/down helpers to avoid duplicating code.

Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Link: https://lore.kernel.org/r/20250318063452.4983-3-andyshrk@163.com
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

drm/bridge: cdns-mhdp8546: Switch to common helpers to power up/down dp link

Use the common dp link power up/down helpers to avoid duplicating code.

Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Link: https://lore.kernel.org/r/20250318063452.4983-2-andyshrk@163.com
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

drm/dp: Pull drm_dp_link_power_up/down from Tegra to common drm_dp_helper

The helper functions drm_dp_link_power_up/down were moved to Tegra
DRM in commit 9a42c7c647a9 ("drm/tegra: Move drm_dp_link helpers to Tegra DRM")".

Now since more and more users are duplicating the same code in their
own drivers, it's time to make them as DRM DP common helpers again.

Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Acked-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Link: https://lore.kernel.org/r/20250318063452.4983-1-andyshrk@163.com
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

MAINTAINERS: Add Dmitry Osipenko as drm/virtio co-maintainer

I was helping to co-maintain VirtIO-GPU driver in drm-misc with
permission from Gerd Hoffmann for past 2 years and would like to
receive new patches directly into my inbox. Add myself as co-maintainer.

Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Gurchetan Singh <gurchetansingh@chromium.org>
Link: https://lore.kernel.org/r/20250401130151.2238772-1-dmitry.osipenko@collabora.com

drm/syncobj: Extend EXPORT_SYNC_FILE for timeline syncobjs

Add support for exporting a dma_fence fd for a specific point on a
timeline.  This is needed for vtest/vpipe[1][2] to implement timeline
syncobj support, as it needs a way to turn a point on a timeline back
into a dma_fence fd.  It also closes an odd omission from the syncobj
UAPI.

[1] https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33433
[2] https://gitlab.freedesktop.org/virgl/virglrenderer/-/merge_requests/805

v2: Add DRM_SYNCOBJ_HANDLE_TO_FD_FLAGS_TIMELINE
v3: Add unstaged uabi header hunk
v4: Also handle IMPORT_SYNC_FILE case
v5: Address comments from Dmitry
v6: checkpatch.pl nits

Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250401155758.48855-1-robdclark@gmail.com
Signed-off-by: Christian König <christian.koenig@amd.com>

drm/panel/panel-simple: Use the new allocation in place of devm_kzalloc()

Start using the new helper that does the refcounted
allocations.

Reviewed-by: Luca Ceresoli <luca.ceresoli@bootlin.com>
Reviewed-by: Maxime Ripard <mripard@kernel.org>
Signed-off-by: Anusha Srivatsa <asrivats@redhat.com>
Link: https://lore.kernel.org/r/20250331-b4-panel-refcounting-v4-4-dad50c60c6c9@redhat.com
Signed-off-by: Maxime Ripard <mripard@kernel.org>

drm/panel: deprecate old-style panel allocation

Start moving to the new refcounted allocations using
the new API devm_drm_panel_alloc(). Deprecate any other
allocation.

Reviewed-by: Luca Ceresoli <luca.ceresoli@bootlin.com>
Reviewed-by: Maxime Ripard <mripard@kernel.org>
Signed-off-by: Anusha Srivatsa <asrivats@redhat.com>
Link: https://lore.kernel.org/r/20250331-b4-panel-refcounting-v4-3-dad50c60c6c9@redhat.com
Signed-off-by: Maxime Ripard <mripard@kernel.org>

drm/panel: Add refcount support

Allocate panel via reference counting. Add _get() and _put() helper
functions to ensure panel allocations are refcounted. Avoid use after
free by ensuring panel pointer is valid and can be usable till the last
reference is put.

Reviewed-by: Luca Ceresoli <luca.ceresoli@bootlin.com>
Reviewed-by: Maxime Ripard <mripard@kernel.org>
Signed-off-by: Anusha Srivatsa <asrivats@redhat.com>
Link: https://lore.kernel.org/r/20250331-b4-panel-refcounting-v4-2-dad50c60c6c9@redhat.com
Signed-off-by: Maxime Ripard <mripard@kernel.org>

drm/panel: Add new helpers for refcounted panel allocatons

Introduce reference counted allocations for panels to avoid
use-after-free. The patch adds the macro devm_drm_bridge_alloc()
to allocate a new refcounted panel. Followed the documentation for
drmm_encoder_alloc() and devm_drm_dev_alloc and other similar
implementations for this purpose.

Reviewed-by: Maxime Ripard <mripard@kernel.org>
Signed-off-by: Anusha Srivatsa <asrivats@redhat.com>
Link: https://lore.kernel.org/r/20250331-b4-panel-refcounting-v4-1-dad50c60c6c9@redhat.com
Signed-off-by: Maxime Ripard <mripard@kernel.org>

drm/format-helper: Optimize 32-to-8-bpp conversion

For ease of implementation, existing line-conversion functions
for 8-bit formats write each pixel individually. Optimize the
performance by writing multiple pixels in a single 32-bit store.

v2:
- simplify address calculation (Jani)
- fix typo in commit message (Jocelyn)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250328141709.217283-9-tzimmermann@suse.de

drm/format-helper: Optimize 32-to-16-bpp conversion

For ease of implementation, existing line-conversion functions
for 16-bit formats write each pixel individually. Optimize the
performance by writing multiple pixels in single 64-bit and 32-bit
stores.

v2:
- simplify address calculation (Jani)
- fix typo in commit message (Jocelyn)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250328141709.217283-8-tzimmermann@suse.de

drm/format-helper: Optimize 32-to-24-bpp conversion

For ease of implementation, existing line-conversion functions
for 24-bit formats write each byte individually. Optimize the
performance by writing 4 pixels in 3 32-bit stores.

v2:
- simplify address calculation (Jani)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250328141709.217283-7-tzimmermann@suse.de

drm/format-helper: Add generic conversion to 8-bit formats

Add drm_fb_xfrm_line_32to8() to implement conversion from 32-bit
pixels to 8-bit pixels. The pixel-conversion is specified by the
given callback parameter. Mark the helper as always_inline to avoid
overhead from function calls.

Then implement all existing line-conversion functions with the new
generic call and the respective pixel-conversion helper.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250328141709.217283-6-tzimmermann@suse.de

drm/format-helper: Add generic conversion to 16-bit formats

Add drm_fb_xfrm_line_32to16() to implement conversion from 32-bit
pixels to 16-bit pixels. The pixel-conversion is specified by the
given callback parameter. Mark the helper as always_inline to avoid
overhead from function calls.

Then implement all existing line-conversion functions with the new
generic call and the respective pixel-conversion helper. There's one
pixel-conversion helper that swaps output bytes. It is for gud and
requires refactoring, so don't move it into the header file.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250328141709.217283-5-tzimmermann@suse.de

drm/format-helper: Add generic conversion to 24-bit formats

Add drm_fb_xfrm_line_32to24() to implement conversion from 32-bit
pixels to 24-bit pixels. The pixel-conversion is specified by the
given callback parameter. Mark the helper as always_inline to avoid
overhead from function calls.

Then implement all existing line-conversion functions with the new
generic call and the respective pixel-conversion helper.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250328141709.217283-4-tzimmermann@suse.de

drm/format-helper: Add generic conversion to 32-bit formats

Add drm_fb_xfrm_line_32to32() to implement conversion from 32-bit
pixels to 32-bit pixels. The pixel-conversion is specified by the
given callback parameter. Mark the helper as always_inline to avoid
overhead from function calls.

Then implement all existing line-conversion functions with the new
generic call and the respective pixel-conversion helper.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250328141709.217283-3-tzimmermann@suse.de

drm/format-helper: Move helpers for pixel conversion to header file

The DRM draw helpers contain format-conversion helpers that operate
on individual pixels. Move them into an internal header file and adopt
them as individual API. Update the draw code accordingly. The pixel
helpers will also be useful for other format conversion helpers.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250328141709.217283-2-tzimmermann@suse.de

drm/cirrus-qemu: Remove custom plane state

Remove struct cirrus_primary_plane_state and its helpers, which
are all unused. Use struct drm_shadow_plane_state instead.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Gerd Hoffmann <kraxel@redhat.com>
Link: https://lore.kernel.org/r/20250328091821.195061-5-tzimmermann@suse.de

drm/cirrus-qemu: Use framebuffer format as-is, drop adjustments

Remove internal adjustments to framebuffer format from cirrus-qemu
driver. The driver did this to support higher resolutions by reducing
the per-pixel memory consumption. DRM has a policy of exporting formats
as they are implemented in hardware. So avoid internal adjustments if
possible.

Also remove the call to drm_fb_blit() from cirrus-qemu. The helper
is useful if source and destination format are not known beforehand.
This is not the case for cirrus-qemu.

This change effectively reduces the maximum available resolution to
800x600 at 32 bpp. A maximum scanline pitch of 4095 byte prevents
1024 pixels per scanline at 32 bpp. Higher resolutions are possible
at lower bit depths, but are currently not supported by userspace.

When cirrus-qemu currently reduced the internal bit depth to support
higher resolutions, it trades resolution for bit depth and CPU time.
Converting from 32-bit colors has a significant runtime overhead, as
outlined at [1]. Avoiding color-format adjustments also avoids this
tradeoff.

v2:
- expand commit message (Gerd)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Gerd Hoffmann <kraxel@redhat.com>
Link: https://lore.kernel.org/dri-devel/20250325110407.81107-1-tzimmermann@suse.de/
Link: https://lore.kernel.org/r/20250328091821.195061-4-tzimmermann@suse.de

drm/cirrus-qemu: Stricter mode validation

Implement strict checking of a display mode's minimum scanline
pitch in cirrus_mode_config_mode_valid(). Sort out all modes that
possibly overflow the maximum pitch.

The current validation only tests against a display mode's minimum
requirements for video memory. Only atomic_check later tests against
the pitch limit before programming the framebuffer.

The problem is that user-space compositors do not handle this
gracefully. If atomic_check fails to validate the scanline pitch
and returns an error, the compositor, namely Weston, does nothing
and the display remains stale.

Ruling out display modes that possibly overflow the pitch avoids
this problem. With only 4 MiB of video memory available, this
effectively limits horizontal resolution to 800 pixels. But with
cirrus-qemu being low-end and obsolete, this is probably not an
issue in practice. Better alternatives are available in qemu.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Gerd Hoffmann <kraxel@redhat.com>
Link: https://lore.kernel.org/r/20250328091821.195061-3-tzimmermann@suse.de

drm/cirrus-qemu: Fix pitch programming

Do not set CR1B[6] when programming the pitch. The bit effects VGA
text mode and is not interpreted by qemu. [1] It has no affect on
the scanline pitch.

The scanline bit that is set into CR1B[6] belongs into CR13[7], which
the driver sets up correctly.

This bug goes back to the driver's initial commit.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Gerd Hoffmann <kraxel@redhat.com>
Link: https://gitlab.com/qemu-project/qemu/-/blob/stable-9.2/hw/display/cirrus_vga.c?ref_type=heads#L1112
Fixes: f9aa76a85248 ("drm/kms: driver for virtual cirrus under qemu")
Cc: Adam Jackson <ajax@redhat.com>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <mripard@kernel.org>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: <stable@vger.kernel.org> # v3.5+
Link: https://lore.kernel.org/r/20250328091821.195061-2-tzimmermann@suse.de

drm/panfrost: Force AARCH64_4K page table format on MediaTek MT8192

MediaTek MT8192 SoC has an ARM Mali-G57 MC5 GPU (Valhall-JM). Now that
Panfrost supports AARCH64_4K page table format, let's enable it on this
SoC.

Running glmark2-es2-drm [0] benchmark, reported the same performance
score on both modes Mali LPAE (LEGACY) vs. AARCH64_4K, before and after
this commit. Tested on a Mediatek (MT8395) Genio 1200 EVK board.

[0] https://github.com/glmark2/glmark2

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Steven Price <steven.price@arm.com>
Link: https://lore.kernel.org/r/20250324185801.168664-7-ariel.dalessandro@collabora.com

drm/panfrost: Force AARCH64_4K page table format on MediaTek MT8188

MediaTek MT8188 SoC has an ARM Mali-G57 MC3 GPU (Valhall-JM), which
constantly faults with the current panfrost support.

For instance, running `glmark2-es2-drm` benchmark test:
```
[   79.617461] panfrost 13000000.gpu: js fault, js=1, status=JOB_BUS_FAULT, head=0xaadc380, tail=0xaadc380
[   80.119811] panfrost 13000000.gpu: gpu sched timeout, js=0, config=0x7300, status=0x58, head=0xaaca180, tail=0xaaca180, sched_job=000000002fd03ccc
[   80.129083] panfrost 13000000.gpu: Unhandled Page fault in AS0 at VA 0x0000000000000000
[   80.129083] Reason: TODO
[   80.129083] raw fault status: 0x1C2
[   80.129083] decoded fault status: SLAVE FAULT
[   80.129083] exception type 0xC2: TRANSLATION_FAULT_2
[   80.129083] access type 0x1: EXECUTE
[   80.129083] source id 0x0
```

Note that current panfrost mode (Mali LPAE - LEGACY) only allows to
specify write-cache or implementation-defined as the caching policy,
probably not matching the right configuration. As depicted in the source
code:

drivers/iommu/io-pgtable-arm.c:
```
* MEMATTR: Mali has no actual notion of a non-cacheable type, so the
* best we can do is mimic the out-of-tree driver and hope that the
* "implementation-defined caching policy" is good enough...
```

Now that Panfrost supports AARCH64_4K page table format, let's enable it
on Mediatek MT8188 and configure the cache/shareability policies
properly.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Steven Price <steven.price@arm.com>
Link: https://lore.kernel.org/r/20250324185801.168664-6-ariel.dalessandro@collabora.com

drm/panfrost: Add support for AARCH64_4K page table format

Currently, Panfrost only supports MMU configuration in "LEGACY" (as
Bifrost calls it) mode, a (modified) version of LPAE "Large Physical
Address Extension", which in Linux we've called "mali_lpae".

This commit adds support for conditionally enabling AARCH64_4K page
table format. To achieve that, a "GPU optional quirks" field was added
to `struct panfrost_features` with the related flag.

Note that, in order to enable AARCH64_4K mode, the GPU variant must have
the HW_FEATURE_AARCH64_MMU feature flag present.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Adrián Larumbe <adrian.larumbe@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Signed-off-by: Steven Price <steven.price@arm.com>
Link: https://lore.kernel.org/r/20250324185801.168664-5-ariel.dalessandro@collabora.com

drm/panfrost: Set HW_FEATURE_AARCH64_MMU feature flag on Bifrost models

Set this feature flag on all Mali Bifrost platforms as the MMU supports
AARCH64 4K page table format.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Reviewed-by: Adrián Larumbe <adrian.larumbe@collabora.com>
Signed-off-by: Steven Price <steven.price@arm.com>
Link: https://lore.kernel.org/r/20250324185801.168664-4-ariel.dalessandro@collabora.com

drm/panfrost: Use GPU_MMU_FEATURES_VA_BITS/PA_BITS macros

As done in panthor, define and use these GPU_MMU_FEATURES_* macros,
which makes code easier to read and reuse.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Reviewed-by: Adrián Larumbe <adrian.larumbe@collabora.com>
Signed-off-by: Steven Price <steven.price@arm.com>
Link: https://lore.kernel.org/r/20250324185801.168664-3-ariel.dalessandro@collabora.com

drm/panfrost: Set IOMMU_CACHE flag

Panfrost does not support uncached mappings, so flag them properly. Also
flag the pages that are mapped as response to a page fault as cached.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Adrián Larumbe <adrian.larumbe@collabora.com>
Signed-off-by: Steven Price <steven.price@arm.com>
Link: https://lore.kernel.org/r/20250324185801.168664-2-ariel.dalessandro@collabora.com

drm/bridge: sii902x: Set bridge type

This is a RGB to HDMI bridge, so set the bridge type accordingly.

Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Link: https://lore.kernel.org/r/20250326101124.4031874-1-alexander.stein@ew.tq-group.com
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

drm/panel: himax-hx8394: transition to mipi_dsi wrapped functions

Changes the himax-hx8394 panel to use multi style functions for
improved error handling.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Signed-off-by: Tejas Vipin <tejasvipin76@gmail.com>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20250325094707.961349-1-tejasvipin76@gmail.com
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

drm/panel/sharp-ls043t1le01: Use _multi variants

Move away from using deprecated API and use _multi variants
if available. Use mipi_dsi_msleep() and mipi_dsi_usleep_range()
instead of msleep() and usleep_range() respectively.

Used Coccinelle to find the _multi variant APIs,replacing
mpi_dsi_msleep() where necessary and for returning
dsi_ctx.accum_err in these functions. mipi_dsi_dcs_write()
does not have a corresponding _multi() variant. Replacing it with
mipi_dsi_dcs_write_seq_multi() instead. This change is manual.

The Coccinelle script is the same as the one in commit c8ba07caaecc
("drm/panel/synaptics-r63353: Use _multi variants")

v2: Use mipi_dsi_write_buffer_multi() in place of
mipi_dsi_dcs_write(). (Dmitry)

v3: add commit details where the same coccinelle script is
used and remove the actual script from commit log.
Use mipi_dsi_dcs_write_seq_multi() for mipi_dsi_dcs_write() (Doug)

Cc: Maxime Ripard <mripard@kernel.org>
Cc: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Cc: Tejas Vipin <tejasvipin76@gmail.com>
Cc: Doug Anderson <dianders@chromium.org>
Signed-off-by: Anusha Srivatsa <asrivats@redhat.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20250326-b4-panel-ls043t1le01-v3-1-96c554c0ea2b@redhat.com
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

drm/bridge: cdns-dsi: Move DSI mode check to _atomic_check()

At present, the DSI mode configuration check happens during the
_atomic_enable() phase, which is not really the best place for this.
Moreover, if the mode is not valid, the driver gives a warning and
continues the hardware configuration.

Move the DSI mode configuration check to _atomic_check() instead, which
can properly report back any invalid mode, before the _enable phase even
begins.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Tested-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Signed-off-by: Aradhya Bhatia <a-bhatia1@ti.com>
Signed-off-by: Aradhya Bhatia <aradhya.bhatia@linux.dev>
Link: https://lore.kernel.org/r/20250329113925.68204-10-aradhya.bhatia@linux.dev
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

drm/bridge: cdns-dsi: Add input format negotiation

Add support for the input format negotiation hook, that uses the helper
drm_mipi_dsi_get_input_bus_fmt() for dsi hosts, to figure out the
required input format.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Tested-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Signed-off-by: Aradhya Bhatia <a-bhatia1@ti.com>
Signed-off-by: Aradhya Bhatia <aradhya.bhatia@linux.dev>
Link: https://lore.kernel.org/r/20250329113925.68204-9-aradhya.bhatia@linux.dev
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

drm/mipi-dsi: Add helper to find input format

Add a helper API that can be used by the DSI hosts to find the required
input bus format for the given output dsi pixel format.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Tested-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Signed-off-by: Aradhya Bhatia <a-bhatia1@ti.com>
Signed-off-by: Aradhya Bhatia <aradhya.bhatia@linux.dev>
Link: https://lore.kernel.org/r/20250329113925.68204-8-aradhya.bhatia@linux.dev
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

drm/bridge: cdns-dsi: Move to devm_drm_of_get_bridge()

Instead of manually finding the next bridge/panel, and maintaining the
panel-bridge (in-case the next entity is a panel), switch to using the
automatically managing devm_drm_of_get_bridge() API.

Drop the drm_panel support completely from the driver while at it.

Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Tested-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Signed-off-by: Aradhya Bhatia <a-bhatia1@ti.com>
Signed-off-by: Aradhya Bhatia <aradhya.bhatia@linux.dev>
Link: https://lore.kernel.org/r/20250329113925.68204-7-aradhya.bhatia@linux.dev
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

drm/bridge: cdns-dsi: Wait for Clk and Data Lanes to be ready

Once the DSI Link and DSI Phy are initialized, the code needs to wait
for Clk and Data Lanes to be ready, before continuing configuration.
This is in accordance with the DSI Start-up procedure, found in the
Technical Reference Manual of Texas Instrument's J721E SoC[0] which
houses this DSI TX controller.

If the previous bridge (or crtc/encoder) are configured pre-maturely,
the input signal FIFO gets corrupt. This introduces a color-shift on the
display.

Allow the driver to wait for the clk and data lanes to get ready during
DSI enable.

[0]: See section 12.6.5.7.3 "Start-up Procedure" in J721E SoC TRM
Link: http://www.ti.com/lit/pdf/spruil1
Fixes: e19233955d9e ("drm/bridge: Add Cadence DSI driver")
Cc: stable@vger.kernel.org
Tested-by: Dominik Haller <d.haller@phytec.de>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Tested-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Signed-off-by: Aradhya Bhatia <a-bhatia1@ti.com>
Signed-off-by: Aradhya Bhatia <aradhya.bhatia@linux.dev>
Link: https://lore.kernel.org/r/20250329113925.68204-6-aradhya.bhatia@linux.dev
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

drm/bridge: cdns-dsi: Check return value when getting default PHY config

Check for the return value of the phy_mipi_dphy_get_default_config()
call, and in case of an error, return back the same.

Fixes: fced5a364dee ("drm/bridge: cdns: Convert to phy framework")
Cc: stable@vger.kernel.org
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Tested-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Signed-off-by: Aradhya Bhatia <a-bhatia1@ti.com>
Signed-off-by: Aradhya Bhatia <aradhya.bhatia@linux.dev>
Link: https://lore.kernel.org/r/20250329113925.68204-5-aradhya.bhatia@linux.dev
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

drm/bridge: cdns-dsi: Fix the clock variable for mode_valid()

The crtc_* mode parameters do not get generated (duplicated in this
case) from the regular parameters before the mode validation phase
begins.

The rest of the code conditionally uses the crtc_* parameters only
during the bridge enable phase, but sticks to the regular parameters
for mode validation. In this singular instance, however, the driver
tries to use the crtc_clock parameter even during the mode validation,
causing the validation to fail.

Allow the D-Phy config checks to use mode->clock instead of
mode->crtc_clock during mode_valid checks, like everywhere else in the
driver.

Fixes: fced5a364dee ("drm/bridge: cdns: Convert to phy framework")
Cc: stable@vger.kernel.org
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Tested-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Signed-off-by: Aradhya Bhatia <a-bhatia1@ti.com>
Signed-off-by: Aradhya Bhatia <aradhya.bhatia@linux.dev>
Link: https://lore.kernel.org/r/20250329113925.68204-4-aradhya.bhatia@linux.dev
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

drm/bridge: cdns-dsi: Fix phy de-init and flag it so

The driver code doesn't have a Phy de-initialization path as yet, and so
it does not clear the phy_initialized flag while suspending. This is a
problem because after resume the driver looks at this flag to determine
if a Phy re-initialization is required or not. It is in fact required
because the hardware is resuming from a suspend, but the driver does not
carry out any re-initialization causing the D-Phy to not work at all.

Call the counterparts of phy_init() and phy_power_on(), that are
phy_exit() and phy_power_off(), from _bridge_post_disable(), and clear
the flags so that the Phy can be initialized again when required.

Fixes: fced5a364dee ("drm/bridge: cdns: Convert to phy framework")
Cc: stable@vger.kernel.org
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Tested-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Signed-off-by: Aradhya Bhatia <a-bhatia1@ti.com>
Signed-off-by: Aradhya Bhatia <aradhya.bhatia@linux.dev>
Link: https://lore.kernel.org/r/20250329113925.68204-3-aradhya.bhatia@linux.dev
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

drm/bridge: cdns-dsi: Fix connecting to next bridge

Fix the OF node pointer passed to the of_drm_find_bridge() call to find
the next bridge in the display chain.

The code to find the next panel (and create its panel-bridge) works
fine, but to find the next (non-panel) bridge does not.

To find the next bridge in the pipeline, we need to pass "np" - the OF
node pointer of the next entity in the devicetree chain. Passing
"of_node" to of_drm_find_bridge (which is what the code does currently)
will fetch the bridge for the cdns-dsi which is not what's required.

Fix that.

Fixes: e19233955d9e ("drm/bridge: Add Cadence DSI driver")
Cc: stable@vger.kernel.org
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Tested-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Signed-off-by: Aradhya Bhatia <a-bhatia1@ti.com>
Signed-off-by: Aradhya Bhatia <aradhya.bhatia@linux.dev>
Link: https://lore.kernel.org/r/20250329113925.68204-2-aradhya.bhatia@linux.dev
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

drm/nouveau: fix hibernate on disabled GPU

Hibernate bricks the machine if a discrete GPU was disabled via

echo IGD > /sys/kernel/debug/vgaswitcheroo/switch

The freeze and thaw handler lacks checking the GPU power state,
as suspend and resume do.

This patch add the checks and fix this issue.

Signed-off-by: Christoph Rudorff <chris@rudorff.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
Link: https://lore.kernel.org/r/20250325-nouveau-fix-hibernate-v2-1-2bd5c13fb953@rudorff.com

accel/amdxdna: Add BO import and export

Add amdxdna_gem_prime_export() and amdxdna_gem_prime_import() for BO
import and export. Register mmu notifier for imported BO as well. When
MMU_NOTIFIER_UNMAP event is received, queue work to remove the notifier.

The same BO could be mapped multiple times if it is exported and imported
by an application. Use a link list to track VMAs the BO been mapped.

v2: Rebased and call get_dma_buf() before dma_buf_attach()
v3: Removed import_attach usage

Reviewed-by: Jeff Hugo <jeff.hugo@oss.qualcomm.com>
Signed-off-by: Lizhi Hou <lizhi.hou@amd.com>
Link: https://lore.kernel.org/r/20250325200105.2744079-1-lizhi.hou@amd.com

drm/nouveau/outp: Avoid -Wflex-array-member-not-at-end warning

-Wflex-array-member-not-at-end was introduced in GCC-14, and we are
getting ready to enable it, globally.

Use the `DEFINE_RAW_FLEX()` helper for an on-stack definition of
a flexible structure where the size of the flexible-array member
is known at compile-time, and refactor the rest of the code,
accordingly.

So, with these changes, fix the following warning:

drivers/gpu/drm/nouveau/nvif/outp.c:199:45: warning: structure containing a flexible array member is not at the end of another structure [-Wflex-array-member-not-at-end]

Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/Z-bFsmWjr5yZy6c6@kspp

drm/nouveau/conn: Avoid -Wflex-array-member-not-at-end warning

-Wflex-array-member-not-at-end was introduced in GCC-14, and we are
getting ready to enable it, globally.

Use the `DEFINE_RAW_FLEX()` helper for an on-stack definition of
a flexible structure where the size of the flexible-array member
is known at compile-time, and refactor the rest of the code,
accordingly.

So, with these changes, fix the following warning:

drivers/gpu/drm/nouveau/nvif/conn.c:34:38: warning: structure containing a flexible array member is not at the end of another structure [-Wflex-array-member-not-at-end]

Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/Z-a4meHAy-t58bcE@kspp

drm/virtio: Add capset definitions to UAPI

Since the context-type additions to the virtio-gpu spec, these have been
defined locally in guest user-space, and virtio-gpu backend library code.

Now, these capsets have been stabilized, and should be defined in a
common space, in both the virtio_gpu header, and alongside the virtgpu_drm
interface that they apply to.

Reviewed-by: Gurchetan Singh <gurchetansingh@chromium.org>
Signed-off-by: Aaron Ruby <aruby@qnx.com>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
[dmitry.osipenko@collabora.com: edit commit title]
Link: https://patchwork.freedesktop.org/patch/msgid/YT3PR01MB5857E808EDF6949F2DF517FDAFA12@YT3PR01MB5857.CANPRD01.PROD.OUTLOOK.COM

drm/bridge: anx7625: Remove redundant 'flush_workqueue()' calls

'destroy_workqueue()' already drains the queue before destroying it, so
there is no need to flush it explicitly.

Remove the redundant 'flush_workqueue()' calls.

This was generated with coccinelle:

@@
expression E;
@@
- flush_workqueue(E);
  destroy_workqueue(E);

Signed-off-by: Chen Ni <nichen@iscas.ac.cn>
Reviewed-by: Robert Foss <rfoss@kernel.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20250312090132.1624445-1-nichen@iscas.ac.cn

dt-bindings: display: panel: samsung,atna40yk20: document ATNA40YK20

The Samsung ATNA40YK20 panel is a 14" AMOLED eDP panel. It is
similar to the ATNA33XC20 except that it is larger and has a
different resolution.

Signed-off-by: Christopher Obbard <christopher.obbard@linaro.org>
Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20250325-wip-obbardc-qcom-t14s-oled-panel-v2-1-e9bc7c9d30cc@linaro.org

drm/panic: add missing Markdown code span

Add missing Markdown code span.

This was found using the Clippy `doc_markdown` lint, which we may want
to enable.

Fixes: cb5164ac43d0 ("drm/panic: Add a QR code panic screen")
Signed-off-by: Miguel Ojeda <ojeda@kernel.org>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Reviewed-by: Benno Lossin <benno.lossin@proton.me>
Signed-off-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250324210359.1199574-3-ojeda@kernel.org

drm/panic: add missing space

Add missing space in sentence.

This was found using the Clippy `doc_markdown` lint, which we may want
to enable.

Fixes: cb5164ac43d0 ("drm/panic: Add a QR code panic screen")
Signed-off-by: Miguel Ojeda <ojeda@kernel.org>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Reviewed-by: Benno Lossin <benno.lossin@proton.me>
Signed-off-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250324210359.1199574-2-ojeda@kernel.org

drm: Remove redundant statement in drm_crtc_helper_set_mode()

Commit dbbfaf5f2641a ("drm: Remove bridge support from legacy helpers")
removes the drm_bridge_mode_fixup() call in drm_crtc_helper_set_mode(),
which makes the subsequent "encoder_funcs = encoder->helper_private" be
redundant, so remove it.

Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Link: https://lore.kernel.org/r/20241111132149.1113736-1-chenhuacai@loongson.cn
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

accel/amdxdna: s/drm_gem_v[un]map_unlocked/drm_gem_v[un]map/

Commit 8f5c4871a014 ("drm/gem: Change locked/unlocked postfix of
drm_gem_v/unmap() function names") dropped the _unlocked suffix,
but accel drivers were left behind.

Fixes: 8f5c4871a014 ("drm/gem: Change locked/unlocked postfix of drm_gem_v/unmap() function names")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Cc: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Cc: Min Ma <min.ma@amd.com>
Cc: Lizhi Hou <lizhi.hou@amd.com>
Cc: Oded Gabbay <ogabbay@kernel.org>
Cc: dri-devel@lists.freedesktop.org
Reviewed-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Tested-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250327104300.1982058-3-boris.brezillon@collabora.com

accel/ivpu: s/drm_gem_shmem_v[un]map/drm_gem_shmem_v[un]map_locked/

Commit 954907f7147d ("drm/shmem-helper: Refactor locked/unlocked
functions") suffixed drm_gem_shmem_v[un]map with _locked to reflect
the fact these functions must be called with the GEM resv lock held,
but accel drivers were left behind.

Fixes: 954907f7147d ("drm/shmem-helper: Refactor locked/unlocked functions")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Cc: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Cc: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
Cc: Maciej Falkowski <maciej.falkowski@linux.intel.com>
Cc: Oded Gabbay <ogabbay@kernel.org>
Cc: dri-devel@lists.freedesktop.org
Reviewed-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Tested-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250327104300.1982058-2-boris.brezillon@collabora.com

accel/ivpu: pages_use_count is now a refcount_t

Commit 051b6646d36d ("drm/shmem-helper: Use refcount_t for
pages_use_count") changed the type of
drm_gem_shmem_object::pages_use_count but accel drivers were left
behind.

Fixes: 051b6646d36d ("drm/shmem-helper: Use refcount_t for pages_use_count")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Cc: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Cc: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
Cc: Maciej Falkowski <maciej.falkowski@linux.intel.com>
Cc: Oded Gabbay <ogabbay@kernel.org>
Cc: dri-devel@lists.freedesktop.org
Reviewed-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Tested-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250327104300.1982058-1-boris.brezillon@collabora.com

drm/shmem-helper: Use refcount_t for vmap_use_count

Use refcount_t helper for vmap_use_count to make refcounting consistent
with pages_use_count and pages_pin_count that use refcount_t. This also
makes vmapping to benefit from the refcount_t's overflow checks.

Acked-by: Maxime Ripard <mripard@kernel.org>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Suggested-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Thomas Zimmermann <tzimmermann@suse.d>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250322212608.40511-11-dmitry.osipenko@collabora.com

drm/shmem-helper: Switch drm_gem_shmem_vmap/vunmap to use pin/unpin

The vmapped pages shall be pinned in memory and previously get/put_pages()
were implicitly hard-pinning/unpinning the pages. This will no longer be
the case with addition of memory shrinker because pages_use_count > 0 won't
determine anymore whether pages are hard-pinned (they will be soft-pinned),
while the new pages_pin_count will do the hard-pinning. Switch the
vmap/vunmap() to use pin/unpin() functions in a preparation of addition
of the memory shrinker support to drm-shmem.

Acked-by: Maxime Ripard <mripard@kernel.org>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Thomas Zimmermann <tzimmermann@suse.d>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250322212608.40511-10-dmitry.osipenko@collabora.com

drm/shmem-helper: Use refcount_t for pages_use_count

Use atomic refcount_t helper for pages_use_count to optimize pin/unpin
functions by skipping reservation locking while GEM's pin refcount > 1.

Acked-by: Maxime Ripard <mripard@kernel.org>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Suggested-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Thomas Zimmermann <tzimmermann@suse.d>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250322212608.40511-9-dmitry.osipenko@collabora.com

drm/shmem-helper: Add and use pages_pin_count

Add separate pages_pin_count for tracking of whether drm-shmem pages are
moveable or not. With the addition of memory shrinker support to drm-shmem,
the pages_use_count will no longer determine whether pages are hard-pinned
in memory, but whether pages exist and are soft-pinned (and could be swapped
out). The pages_pin_count > 1 will hard-pin pages in memory.

Acked-by: Maxime Ripard <mripard@kernel.org>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Suggested-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Thomas Zimmermann <tzimmermann@suse.d>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250322212608.40511-8-dmitry.osipenko@collabora.com