drm/virtio: Improve DMA API usage for shmem BOs

DRM API requires the DRM's driver to be backed with the device that can
be used for generic DMA operations. The VirtIO-GPU device can't perform
DMA operations if it uses PCI transport because PCI device driver creates
a virtual VirtIO-GPU device that isn't associated with the PCI. Use PCI's
GPU device for the DRM's device instead of the VirtIO-GPU device and drop
DMA-related hacks from the VirtIO-GPU driver.

Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20220630200726.1884320-8-dmitry.osipenko@collabora.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm/virtio: Simplify error handling of virtio_gpu_object_create()

Change the order of SHMEM initialization and reservation locking
to make code cleaner and to prepare for transitioning of the common
GEM SHMEM code to use the GEM's reservation lock instead of the
shmem.page_lock.

There is no need to lock reservation during allocation of the SHMEM pages
because the lock is needed only to avoid racing with the async host-side
allocation. Hence we can safely move the SHMEM initialization out of the
reservation lock.

Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20220630200726.1884320-7-dmitry.osipenko@collabora.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm/virtio: Use appropriate atomic state in virtio_gpu_plane_cleanup_fb()

Make virtio_gpu_plane_cleanup_fb() to clean the state which DRM core
wants to clean up and not the current plane's state. Normally the older
atomic state is cleaned up, but the newer state could also be cleaned up
in case of aborted commits.

Cc: stable@vger.kernel.org
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20220630200726.1884320-6-dmitry.osipenko@collabora.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm/virtio: Unlock reservations on dma_resv_reserve_fences() error

Unlock reservations on dma_resv_reserve_fences() error to fix recursive
locking of the reservations when this error happens.

Cc: stable@vger.kernel.org
Fixes: c8d4c18bfbc4 ("dma-buf/drivers: make reserving a shared slot mandatory v4")
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20220630200726.1884320-5-dmitry.osipenko@collabora.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm/virtio: Unlock reservations on virtio_gpu_object_shmem_init() error

Unlock reservations in the error code path of virtio_gpu_object_create()
to silence debug warning splat produced by ww_mutex_destroy(&obj->lock)
when GEM is released with the held lock.

Cc: stable@vger.kernel.org
Fixes: 30172efbfb84 ("drm/virtio: blob prep: refactor getting pages and attaching backing")
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20220630200726.1884320-4-dmitry.osipenko@collabora.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm/virtio: Check whether transferred 2D BO is shmem

Transferred 2D BO always must be a shmem BO. Add check for that to prevent
NULL dereference if userspace passes a VRAM BO.

Cc: stable@vger.kernel.org
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20220630200726.1884320-3-dmitry.osipenko@collabora.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm/virtio: Correct drm_gem_shmem_get_sg_table() error handling

Previous commit fixed checking of the ERR_PTR value returned by
drm_gem_shmem_get_sg_table(), but it missed to zero out the shmem->pages,
which will crash virtio_gpu_cleanup_object(). Add the missing zeroing of
the shmem->pages.

Fixes: c24968734abf ("drm/virtio: Fix NULL vs IS_ERR checking in virtio_gpu_object_shmem_init")
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20220630200726.1884320-2-dmitry.osipenko@collabora.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

fbdev: Remove conflict-handling code

Remove the call to do_remove_conflicting_framebuffers() from the
framebuffer registration. Aperture helpers take care of removing
conflicting devices. With all ownership information stored in the
aperture datastrcutures, remove remove_conflicting_framebuffers()
entirely.

This change also rectifies DRM generic-framebuffer registration, which
tried to unregister conflicting framebuffers, even though it's entirely
build on top of DRM.

v2:
* remove internal aperture-overlap helpers, which are
  now unused

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220718072322.8927-12-tzimmermann@suse.de

fbdev: Acquire framebuffer apertures for firmware devices

When registering a generic framebuffer, automatically acquire ownership
of the framebuffer's I/O range. The device will now be handled by the
aperture helpers. Fbdev-based conflict handling is no longer required.

v2:
* use fb_ prefix instead of fbm_ (Javier)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220718072322.8927-11-tzimmermann@suse.de

video/aperture: Remove conflicting VGA devices, if any

On the primary graphics adapter, a driver might conflict with a VGA
driver that controls the VGA framebuffer I/O range. Remove the VGA
driver from the aperture helpers. Until now, this case has been
hendled by fbdev, but it should work even with fbdev disabled.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220718072322.8927-10-tzimmermann@suse.de

video: Provide constants for VGA I/O range

Provide VGA_FB_ constants for the VGA framebuffer I/O range and convert
fbdev code. In the case of vga16fb, this is a rename of the existing
constants VGA_FB_PHYS and VGA_FB_PHYS_LEN.

v2:
* clarify relationship with old constants in vga16fb (Javier)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220718072322.8927-9-tzimmermann@suse.de

video/aperture: Disable and unregister sysfb devices via aperture helpers

Call sysfb_disable() before removing conflicting devices in aperture
helpers. Fixes sysfb state if fbdev has been disabled.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Fixes: fb84efa28a48 ("drm/aperture: Run fbdev removal before internal helpers")
Cc: Zack Rusin <zackr@vmware.com>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Javier Martinez Canillas <javierm@redhat.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Helge Deller <deller@gmx.de>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Zhen Lei <thunder.leizhen@huawei.com>
Cc: Changcheng Deng <deng.changcheng@zte.com.cn>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <mripard@kernel.org>
Cc: dri-devel@lists.freedesktop.org
Link: https://patchwork.freedesktop.org/patch/msgid/20220718072322.8927-8-tzimmermann@suse.de

fbdev: Remove conflicting devices on PCI bus

Remove firmware devices on the PCI bus, by calling
aperture_remove_conflicting_pci_devices() in the probe function of
each related fbdev driver. iSo far, most of these drivers did not
remove conflicting VESA or EFI devices, or outride failed for
resource conflicts (i.e., matroxfb.) This must have been broken
for quite some time.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220718072322.8927-7-tzimmermann@suse.de

fbdev: Convert drivers to aperture helpers

Convert fbdev drivers from fbdev's remove_conflicting_framebuffers() to
the framework-independent aperture_remove_conflicting_devices(). Calling
this function will also remove conflicting DRM drivers.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220718072322.8927-6-tzimmermann@suse.de

fbdev/core: Remove remove_conflicting_pci_framebuffers()

Remove remove_conflicting_pci_framebuffers() and implement similar
functionality in aperture_remove_conflicting_pci_device(), which was
the only caller. Removes an otherwise unused interface and streamlines
the aperture helper. No functional changes.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220718072322.8927-5-tzimmermann@suse.de

fbdev/vga16fb: Auto-generate module init/exit code

Move vgag16fb's option parsing into the driver's probe function and
generate the rest of the module's init/exit functions from macros.
Keep the options code, although there are no options defined.

v2:
* no options are supported, remove the code (Javier)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220718072322.8927-4-tzimmermann@suse.de

fbdev/vga16fb: Create EGA/VGA devices in sysfb code

Move the device-creation from vga16fb to sysfb code. The driver's
videomode checks are independent from device creation, so move them
into vga16fb's probe function. This will allow to create the module
init/exit code automatically.

The vga16fb driver requires a screen_info for type VIDEO_TYPE_VGAC
or VIDEO_TYPE_EGAC. Such code is nowhere present in the kernel, except
for some MIPS systems. It's not clear if the vga16fb driver actually
works in practice.

v2:
* keep driver name to "vga16fb" (Javier)
* give rational for moving mode checks (Javier)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220718072322.8927-3-tzimmermann@suse.de

fbdev: Remove trailing whitespaces

Fix coding style. No functional changes.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220718072322.8927-2-tzimmermann@suse.de

drm/panel/panel-sitronix-st7701: Split GIP and init sequences

The ST7701 initialization sequence is well parametrized, split the GIP
programming sequence, which is fully custom completely undocumented
TFT matrix specific magic register programming sequence into separate
callback so other TFT matrix definitions can add their own GIP sequence.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Guido Günther <agx@sigxcpu.org>
Cc: Jagan Teki <jagan@amarulasolutions.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220710194437.289042-9-marex@denx.de

drm/panel/panel-sitronix-st7701: Parametrize voltage and timing

Instead of hard-coding TFT matrix voltage and timing settings, which can
even lead to permanent TFT matrix damage, parametrize them in TFT matrix
descriptor.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Guido Günther <agx@sigxcpu.org>
Cc: Jagan Teki <jagan@amarulasolutions.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220710194437.289042-8-marex@denx.de

drm/panel/panel-sitronix-st7701: Infer horizontal pixel count from TFT mode

The horizontal pixel count is a property of the TFT matrix. Currently the
driver hard-codes content of this register to specific value which is
only compatible with one TFT matrix, likely the TS8550B one.

Calculate the horizontal pixel count from the mode instead.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Guido Günther <agx@sigxcpu.org>
Cc: Jagan Teki <jagan@amarulasolutions.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220710194437.289042-7-marex@denx.de

drm/panel/panel-sitronix-st7701: Adjust porch control bitfield name

Define DSI_CMD2_BK0_PORCTRL_VBP_MASK and DSI_CMD2_BK0_PORCTRL_VFP_MASK
and move the vertical back and front porch calculation from macros into
the st7701_init_sequence() function, so it is clear what this does.

No functional change.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Guido Günther <agx@sigxcpu.org>
Cc: Jagan Teki <jagan@amarulasolutions.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220710194437.289042-6-marex@denx.de

drm/panel/panel-sitronix-st7701: Infer vertical line count from TFT mode

The vertical line count is a property of the TFT matrix. Currently the
driver hard-codes content of this register to specific value which is
only compatible with one TFT matrix, likely the TS8550B one.

Calculate the vertical line count from the mode instead.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Guido Günther <agx@sigxcpu.org>
Cc: Jagan Teki <jagan@amarulasolutions.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220710194437.289042-5-marex@denx.de

drm/panel/panel-sitronix-st7701: Make gamma correction TFT specific

The gamma correction values are specific to the TFT which is attached to
the ST7701 TFT matrix driver, move the gamma correction values from what
incorrectly looks like common init sequence into TFT matrix specific
settings.

While doing so, add macros which defined fields within the gamma register
file and a macro which mimics FIELD_PREP except works with constant
expressions.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Guido Günther <agx@sigxcpu.org>
Cc: Jagan Teki <jagan@amarulasolutions.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220710194437.289042-4-marex@denx.de

drm/panel/panel-sitronix-st7701: Make voltage supplies common to ST7701

The ST7701 and ST7701S all have two voltage supplies, one for internal
logic and one for the TFT matrix driver. The supplies are not property
of the TFT matrix driver, so move them to common ST7701 code.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Guido Günther <agx@sigxcpu.org>
Cc: Jagan Teki <jagan@amarulasolutions.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220710194437.289042-3-marex@denx.de

drm/panel/panel-sitronix-st7701: Enable DSI burst mode, LPM, non-continuous clock

The ST7701(S) is capable of DSI burst mode, which is more energy
efficient than the non-burst modes. Make use of it.

The ST7701(S) is capable of DSI non-continuous clock, since it
sources the TFT matrix driver clock from internal clock source.
The DSI non-continuous clock further reduce power utilization.

The ST7701(S) uses DSI LPM for command transmissions, make sure
this is configured correctly in the DSI mode flags.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Guido Günther <agx@sigxcpu.org>
Cc: Jagan Teki <jagan@amarulasolutions.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220710194437.289042-2-marex@denx.de

drm/panel/panel-sitronix-st7701: Make DSI mode flags common to ST7701

The ST7701 and ST7701S are TFT matrix drivers with integrated multi
protocol decoder capable of DSI/DPI/SPI input and 480x360...864 line
TFT matrix output. Currently the only supported input is DSI.

The protocol decoder is separate from the TFT matrix driver and is
always capable of handling all of DSI non-burst mode with sync pulses
or sync events as well as DSI burst mode.

Move the DSI mode configuration from TFT matrix driver properties to
common ST7701 code, because this is common to all TFT matrices.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Guido Günther <agx@sigxcpu.org>
Cc: Jagan Teki <jagan@amarulasolutions.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Thierry Reding <thierry.reding@gmail.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220710194437.289042-1-marex@denx.de

drm:remove rebundant word "in" in comments

there is a repeated word "in", so remove it.

Signed-off-by: shaomin Deng <dengshaomin@cdjrlc.com>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20220718150418.7321-1-dengshaomin@cdjrlc.com

drm:delete the repeated word "the" in comments

remove the rebundant word "the" from comments.

Signed-off-by: shaomin Deng <dengshaomin@cdjrlc.com>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20220718151154.9628-1-dengshaomin@cdjrlc.com

dma-buf/dma_resv_usage: update explicit sync documentation

Make it clear that DMA_RESV_USAGE_BOOKKEEP can be used for explicit synced
user space submissions as well and document the rules around adding the
same fence with different usages.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220712131201.131475-1-christian.koenig@amd.com

drm/mgag200:remove rebundant word "or" in comments

there is a repeated word "or" in comments, so remove it.

Signed-off-by: shaomin Deng <dengshaomin@cdjrlc.com>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20220718145536.4866-1-dengshaomin@cdjrlc.com

drm/bridge: it6505: Modified video clock calculation and video debug message

Speed up video clock calculation and remove redundant video debug message.

Signed-off-by: Pin-Yen Lin <treapking@chromium.org>
Signed-off-by: Allen Chen <allen.chen@ite.com.tw>
Reviewed-by: Robert Foss <robert.foss@linaro.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220715024910.12578-4-allen.chen@ite.com.tw

drm/bridge: it6505: Add i2c api power on check

Use i2c bus to read/write when it6505 power off will occur i2c error.
Add this check will prevent i2c error when it6505 power off.

Signed-off-by: Pin-Yen Lin <treapking@chromium.org>
Signed-off-by: Allen Chen <allen.chen@ite.com.tw>
Reviewed-by: Robert Foss <robert.foss@linaro.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220715024910.12578-3-allen.chen@ite.com.tw

drm/bridge: it6505: Modified power sequence

Change power sequence to meet it6505 data sheet requirement when boot on.

Signed-off-by: Pin-Yen Lin <treapking@chromium.org>
Signed-off-by: Allen Chen <allen.chen@ite.com.tw>
Reviewed-by: Robert Foss <robert.foss@linaro.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220715024910.12578-2-allen.chen@ite.com.tw

drm/bridge: it6505: Power on downstream device in .atomic_enable

Send DPCD DP_SET_POWER_D0 command to the monitor in .atomic_enable
callback. Without this command, some monitors won't show up again after
changing the resolution.

Fixes: 46ca7da7f1e8 ("drm/bridge: it6505: Send DPCD SET_POWER to downstream")

Signed-off-by: Pin-Yen Lin <treapking@chromium.org>
Reviewed-by: Allen Chen <allen.chen@ite.com.tw>
Fixes: 46ca7da7f1e8 ("drm/bridge: it6505: Send DPCD SET_POWER to downstream")
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220714173715.v2.1.I85af54e9ceda74ec69f661852825845f983fc343@changeid

drm/ssd130x: Use new regmap bulk write support to drop custom bus

Data writes for the ssd130x 4-wire SPI protocol need special handling, due
the Data/Command control (D/C) pin having to be toggled prior to the write.

The regmap API only allowed drivers to provide .reg_{read,write} callbacks
to do per register read/write, but didn't provide a way for drivers to do
the same for bulk read/writes.

For this reason, a custom regmap bus was used by the driver just to define
a bulk write callback that implements the D/C pin toggling. But the regmap
API has been extended to support defining bulk read/write handlers, so the
custom regmap bus is not needed anymore and could just be dropped.

Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220618174338.2253021-1-javierm@redhat.com

drm/fourcc: Add formats for packed YUV 4:4:4 AVUY and XVUY permutations

Add FourCCs for two missing permutations of the packed YUV 4:4:4 color
components, namely AVUY and XVUY.

These formats are needed by the NXP i.MX8 ISI. While the ISI is
supported by a V4L2 device (corresponding formats have been submitted to
V4L2), it is handled in userspace by libcamera, which uses DRM FourCCs
for pixel formats.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20220616185210.22018-1-laurent.pinchart@ideasonboard.com

drm/via: Fix style issues in via_3d_reg header

- Fix comments using wrong style
- Drop repeated word

This fixes all checkpatch complains and makes the file a little
bit easier to read - as the eye is not distracted by the style
violations.

Suggested-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Kevin Brace <kevinbrace@bracecomputerlab.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220713170202.1798216-14-sam@ravnborg.org

drm/via: Make macros readable in the via_3d_reg header

The macros for texture 1 setting all used continuation on a new line,
resulting in a highly ureadable definition.
Merge the lines so they are each on a single line.

As a nice side-effect this fixes a number of checkpatch warnings:
"WARNING: please, no spaces at the start of a line".

Suggested-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Kevin Brace <kevinbrace@bracecomputerlab.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220713170202.1798216-13-sam@ravnborg.org

drm/via: Use SPDX tag for MIT license in via_3d_reg header

The license for the via_3d_reg header is MIT - so use the
shorter SPDX tag to identify the license.

Suggested-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Kevin Brace <kevinbrace@bracecomputerlab.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220713170202.1798216-12-sam@ravnborg.org

drm/via: Update to the latest via_3d_reg header

Updated the 3d_reg header file to match what is used by the openchrome
driver.
This verifies that the two drivers can use the same header file.

The file is a verbatim copy from the openchrome repo - a few style
issues will be fixed in following commits.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Kevin Brace <kevinbrace@bracecomputerlab.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220713170202.1798216-11-sam@ravnborg.org

drm/via: Embed via_drv.h in via_dri1

With this change the driver is now a signle file driver.
The only remaning heder file describes the HW and can be shared with the
new openchrome driver.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Kevin Brace <kevinbrace@bracecomputerlab.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220713170202.1798216-10-sam@ravnborg.org

drm/via: Embed via_verifier in via_dri1

Embed the header file in via_drv.h and the code in via_dri1.
All functions are made static as there are no more external users.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Kevin Brace <kevinbrace@bracecomputerlab.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220713170202.1798216-9-sam@ravnborg.org

drm/via: Embed via_dmablit in via_dri1

Embed some of the header file in via_drv.h and
the rest in via_dri1.c
While embedding deleted extra empty lines and functions that
has no external users are made static.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Kevin Brace <kevinbrace@bracecomputerlab.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220713170202.1798216-8-sam@ravnborg.org

drm/via: Embed via_irq in via_dri1

All functions are made static as there are no more external users.
The file had new copyrights that are kept.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Kevin Brace <kevinbrace@bracecomputerlab.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220713170202.1798216-7-sam@ravnborg.org

drm/via: Embed via_video in via_dri1

All functions are made static as there are no more external users.
The file had a new copyright that is kept.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Kevin Brace <kevinbrace@bracecomputerlab.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220713170202.1798216-6-sam@ravnborg.org

drm/via: Embed via_mm in via_dri1

All functions was made static as there are no external users.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Kevin Brace <kevinbrace@bracecomputerlab.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220713170202.1798216-5-sam@ravnborg.org

drm/via: Embed via_map in via_dri1

A few functions has no external use and are made static.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Kevin Brace <kevinbrace@bracecomputerlab.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220713170202.1798216-4-sam@ravnborg.org

drm/via: Embed via_dma in via_dri1

Moved the copyright notices so all copyrights are kept.
A few variables was made static as there are no more users outside this
file.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Kevin Brace <kevinbrace@bracecomputerlab.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220713170202.1798216-3-sam@ravnborg.org

drm/via: Rename via_drv to via_dri1

The via driver implements the DRI1 interface, and we have a new
implementation of the via driver coming that supports atomic
modesetting.

It is not acceptable just to replace the existing driver so
this is first step to make it a single-file implementation allowing
it to stay without interfering with the new driver.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Kevin Brace <kevinbrace@bracecomputerlab.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220713170202.1798216-2-sam@ravnborg.org

drm/nouveau/bios: set info only when the return is not 0

clang static analysis reports
drivers/gpu/drm/nouveau/nvkm/subdev/bios/pmu.c:68:17: warning: The right operand of '*' is a garbage value [core.UndefinedBinaryOperatorResult]
        switch (!!data * *ver) {
                       ^ ~~~~
A switch statement with only a default should be reduced to an if.

If nvbios_pmuEp() returns 0, via the data variable, the output info parameter
is not used.  So set info only when data is not 0.

The struct nvbios_pmuE only has the type and data elements.  Since both of these
are explicitly set, memset is not needed.  So remove it.

Signed-off-by: Tom Rix <trix@redhat.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220702153904.1696595-1-trix@redhat.com

drm/amdgpu: re-apply "move internal vram_mgr function into the C file""

This re-applys commit 708d19d9f362766147cab79eccae60912c6d3068.

The original problem this was reverted for was found and the correct fix
will be merged to drm-misc-next-fixes.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Arunpravin Paneer Selvam <Arunpravin.PaneerSelvam@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220714132315.587217-2-christian.koenig@amd.com

drm/amdgpu: reapply "fix start calculation in amdgpu_vram_mgr_new""

This re-applys commit 5e3f1e7729ec7a99e145e9d8ed58963d86cdfb98.

The original problem this was reverted for was found and the correct fix
will be merged to drm-misc-next-fixes.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Arunpravin Paneer Selvam <Arunpravin.PaneerSelvam@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220714132315.587217-1-christian.koenig@amd.com

drm: Prevent drm_copy_field() to attempt copying a NULL pointer

There are some struct drm_driver fields that are required by drivers since
drm_copy_field() attempts to copy them to user-space via DRM_IOCTL_VERSION.

But it can be possible that a driver has a bug and did not set some of the
fields, which leads to drm_copy_field() attempting to copy a NULL pointer:

[ +10.395966] Unable to handle kernel access to user memory outside uaccess routines at virtual address 0000000000000000
[  +0.010955] Mem abort info:
[  +0.002835]   ESR = 0x0000000096000004
[  +0.003872]   EC = 0x25: DABT (current EL), IL = 32 bits
[  +0.005395]   SET = 0, FnV = 0
[  +0.003113]   EA = 0, S1PTW = 0
[  +0.003182]   FSC = 0x04: level 0 translation fault
[  +0.004964] Data abort info:
[  +0.002919]   ISV = 0, ISS = 0x00000004
[  +0.003886]   CM = 0, WnR = 0
[  +0.003040] user pgtable: 4k pages, 48-bit VAs, pgdp=0000000115dad000
[  +0.006536] [0000000000000000] pgd=0000000000000000, p4d=0000000000000000
[  +0.006925] Internal error: Oops: 96000004 [#1] SMP
...
[  +0.011113] pstate: 80400005 (Nzcv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[  +0.007061] pc : __pi_strlen+0x14/0x150
[  +0.003895] lr : drm_copy_field+0x30/0x1a4
[  +0.004156] sp : ffff8000094b3a50
[  +0.003355] x29: ffff8000094b3a50 x28: ffff8000094b3b70 x27: 0000000000000040
[  +0.007242] x26: ffff443743c2ba00 x25: 0000000000000000 x24: 0000000000000040
[  +0.007243] x23: ffff443743c2ba00 x22: ffff8000094b3b70 x21: 0000000000000000
[  +0.007241] x20: 0000000000000000 x19: ffff8000094b3b90 x18: 0000000000000000
[  +0.007241] x17: 0000000000000000 x16: 0000000000000000 x15: 0000aaab14b9af40
[  +0.007241] x14: 0000000000000000 x13: 0000000000000000 x12: 0000000000000000
[  +0.007239] x11: 0000000000000000 x10: 0000000000000000 x9 : ffffa524ad67d4d8
[  +0.007242] x8 : 0101010101010101 x7 : 7f7f7f7f7f7f7f7f x6 : 6c6e6263606e7141
[  +0.007239] x5 : 0000000000000000 x4 : 0000000000000000 x3 : 0000000000000000
[  +0.007241] x2 : 0000000000000000 x1 : ffff8000094b3b90 x0 : 0000000000000000
[  +0.007240] Call trace:
[  +0.002475]  __pi_strlen+0x14/0x150
[  +0.003537]  drm_version+0x84/0xac
[  +0.003448]  drm_ioctl_kernel+0xa8/0x16c
[  +0.003975]  drm_ioctl+0x270/0x580
[  +0.003448]  __arm64_sys_ioctl+0xb8/0xfc
[  +0.003978]  invoke_syscall+0x78/0x100
[  +0.003799]  el0_svc_common.constprop.0+0x4c/0xf4
[  +0.004767]  do_el0_svc+0x38/0x4c
[  +0.003357]  el0_svc+0x34/0x100
[  +0.003185]  el0t_64_sync_handler+0x11c/0x150
[  +0.004418]  el0t_64_sync+0x190/0x194
[  +0.003716] Code: 92402c04 b200c3e8 f13fc09f 5400088c (a9400c02)
[  +0.006180] ---[ end trace 0000000000000000 ]---

Reported-by: Peter Robinson <pbrobinson@gmail.com>
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20220705100215.572498-3-javierm@redhat.com

drm: Use size_t type for len variable in drm_copy_field()

The strlen() function returns a size_t which is an unsigned int on 32-bit
arches and an unsigned long on 64-bit arches. But in the drm_copy_field()
function, the strlen() return value is assigned to an 'int len' variable.

Later, the len variable is passed as copy_from_user() third argument that
is an unsigned long parameter as well.

In theory, this can lead to an integer overflow via type conversion. Since
the assignment happens to a signed int lvalue instead of a size_t lvalue.

In practice though, that's unlikely since the values copied are set by DRM
drivers and not controlled by userspace. But using a size_t for len is the
correct thing to do anyways.

Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Tested-by: Peter Robinson <pbrobinson@gmail.com>
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20220705100215.572498-2-javierm@redhat.com

drm/vc4: v3d: Switch to devm_pm_runtime_enable

devm_pm_runtime_enable() simplifies the driver a bit since it will call
pm_runtime_disable() automatically through a device-managed action.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-70-maxime@cerno.tech

drm/vc4: v3d: Rework the runtime_pm setup

At bind time, vc4_v3d_bind() will read a register to retrieve the v3d
version and make sure it's a version we're compatible with.

However, the v3d has an optional clock that is enabled only after the
register read-out and a power domain that wasn't enabled at all in the bind
implementation. This was working fine at boot because both were enabled,
but resulted in the version check failing if we were unbinding and
rebinding the driver because the unbinding would have turned them off.

The fix isn't as easy as calling pm_runtime_resume_and_get() prior to the
register access to power up the power domain though.

Indeed, the runtime_resume implementation will enable the clock mentioned
above, call vc4_v3d_init_hw() and then vc4_irq_enable().

Prior to the previous patch, vc4_irq_enable() needed to occur after our
call to platform_get_irq() and vc4_irq_install(), since vc4_irq_enable()
used to call enable_irq() and vc4_irq_install() will call request_irq().

vc4_irq_install() will also do some register access, so needs the power
domain to be on. So we ended up in a situation where
vc4_v3d_runtime_resume() needed vc4_irq_install() to have been called
before, and vc4_irq_install() needed vc4_v3d_runtime_resume().

The previous patch removed the enable_irq() call in vc4_irq_enable() and
thus removed the dependency of vc4_v3d_runtime_resume() on
vc4_irq_install().

Thus, we can now rework our bind implementation to call
pm_runtime_resume_and_get() before our register access to make sure the
power domain is on. vc4_v3d_runtime_resume() also takes care of turning the
clock on and calling vc4_v3d_init_hw() so we can remove them from bind.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-69-maxime@cerno.tech

drm/vc4: v3d: Stop disabling interrupts

The vc4_irq_disable(), among other things, will call disable_irq() to
complete any in-flight interrupts.

This requires its counterpart, vc4_irq_enable(), to call enable_irq() which
causes issues addressed in a later patch.

However, vc4_irq_disable() is called by two callees: vc4_irq_uninstall()
and vc4_v3d_runtime_suspend().

vc4_irq_uninstall() also calls free_irq() which already disables the
interrupt line. We thus don't require an explicit disable_irq() for that
call site.

vc4_v3d_runtime_suspend() doesn't have any other code. However, the rest of
vc4_irq_disable() masks the interrupts coming from the v3d, so explictly
disabling the interrupt line is also redundant.

The only thing we really care about is thus to make sure we don't have any
handler in-flight, as suggested by the comment. We can thus replace
disable_irq() by synchronize_irq().

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-68-maxime@cerno.tech

drm/vc4: perfmon: Add missing mutex_destroy

vc4_perfmon_open_file() will instantiate a mutex for that file instance,
but we never call mutex_destroy () in vc4_perfmon_close_file().

Let's add that missing call.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-67-maxime@cerno.tech

drm/vc4: Switch to drmm_mutex_init

mutex_init is supposed to be balanced by a call to mutex_destroy that we
were never doing in the vc4 driver.

Since a DRM-managed mutex_init variant has been introduced, let's just
switch to it.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-66-maxime@cerno.tech

drm/vc4: debugfs: Simplify debugfs registration

The vc4 has a custom API to allow components to register a debugfs file
before the DRM driver has been registered and the debugfs_init hook has
been called.

However, the .late_register hook allows to have the debugfs file creation
deferred after that time already.

Let's remove our custom code to only register later our debugfs entries as
part of either debugfs_init or after it.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-65-maxime@cerno.tech

drm/vc4: debugfs: Return an error on failure

vc4_debugfs_add_file() can fail, so let's propagate its error code instead
of silencing it.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-64-maxime@cerno.tech

drm/vc4: debugfs: Protect device resources

Our current code now mixes some resources whose lifetime are tied to the
device (clocks, IO mappings, etc.) and some that are tied to the DRM device
(encoder, bridge).

The device one will be freed at unbind time, but the DRM one will only be
freed when the last user of the DRM device closes its file handle.

So we end up with a time window during which we can call the encoder hooks,
but we don't have access to the underlying resources and device.

Let's protect all those sections with drm_dev_enter() and drm_dev_exit() so
that we bail out if we are during that window.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-63-maxime@cerno.tech

drm/vc4: vec: Switch to devm_pm_runtime_enable

devm_pm_runtime_enable() simplifies the driver a bit since it will call
pm_runtime_disable() automatically through a device-managed action.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-62-maxime@cerno.tech

drm/vc4: vec: Protect device resources after removal

Whenever the device and driver are unbound, the main device and all the
subdevices will be removed by calling their unbind() method.

However, the DRM device itself will only be freed when the last user will
have closed it.

It means that there is a time window where the device and its resources
aren't there anymore, but the userspace can still call into our driver.

Fortunately, the DRM framework provides the drm_dev_enter() and
drm_dev_exit() functions to make sure our underlying device is still there
for the section protected by those calls. Let's add them to the VEC driver.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-61-maxime@cerno.tech

drm/vc4: vec: Switch to DRM-managed connector initialization

The current code will call drm_connector_unregister() and
drm_connector_cleanup() when the device is unbound. However, by then, there
might still be some references held to that connector, including by the
userspace that might still have the DRM device open.

Let's switch to a DRM-managed initialization to clean up after ourselves
only once the DRM device has been last closed.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-60-maxime@cerno.tech

drm/vc4: vec: Switch to DRM-managed encoder initialization

The current code will call drm_encoder_cleanup() when the device is
unbound. However, by then, there might still be some references held to
that encoder, including by the userspace that might still have the DRM
device open.

Let's switch to a DRM-managed initialization to clean up after ourselves
only once the DRM device has been last closed.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-59-maxime@cerno.tech

drm/vc4: vec: Remove call to drm_connector_unregister()

drm_connector_unregister() is only to be used for connectors that have been
registered through drm_connector_register() after drm_dev_register() has
been called. This is our case here so let's remove the call.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-58-maxime@cerno.tech

drm/vc4: vec: Switch to drmm_kzalloc

Our internal structure that stores the DRM entities structure is allocated
through a device-managed kzalloc.

This means that this will eventually be freed whenever the device is
removed. In our case, the most likely source of removal is that the main
device is going to be unbound, and component_unbind_all() is being run.

However, it occurs while the DRM device is still registered, which will
create dangling pointers, eventually resulting in use-after-free.

Switch to a DRM-managed allocation to keep our structure until the DRM
driver doesn't need it anymore.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-57-maxime@cerno.tech

drm/vc4: vec: Embed DRM structures into the private structure

The VC4 VEC driver private structure contains only a pointer to the
encoder and connector it implements. This makes the overall structure
somewhat inconsistent with the rest of the driver, and complicates its
initialisation without any apparent gain.

Let's embed the drm_encoder structure (through the vc4_encoder one) and
drm_connector into struct vc4_vec to fix both issues.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-56-maxime@cerno.tech

drm/vc4: vec: Remove vc4_dev vec pointer

There's no user for that pointer so let's just get rid of it.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-55-maxime@cerno.tech

drm/vc4: txp: Protect device resources

Our current code now mixes some resources whose lifetime are tied to the
device (clocks, IO mappings, etc.) and some that are tied to the DRM device
(encoder, bridge).

The device one will be freed at unbind time, but the DRM one will only be
freed when the last user of the DRM device closes its file handle.

So we end up with a time window during which we can call the encoder hooks,
but we don't have access to the underlying resources and device.

Let's protect all those sections with drm_dev_enter() and drm_dev_exit() so
that we bail out if we are during that window.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-54-maxime@cerno.tech

drm/vc4: txp: Remove call to drm_connector_unregister()

drm_connector_unregister() is only to be used for connectors that have been
registered through drm_connector_register() after drm_dev_register() has
been called. This is our case here so let's remove the call.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-53-maxime@cerno.tech

drm/vc4: txp: Switch to drmm_kzalloc

Our internal structure that stores the DRM entities structure is allocated
through a device-managed kzalloc.

This means that this will eventually be freed whenever the device is
removed. In our case, the most likely source of removal is that the main
device is going to be unbound, and component_unbind_all() is being run.

However, it occurs while the DRM device is still registered, which will
create dangling pointers, eventually resulting in use-after-free.

Switch to a DRM-managed allocation to keep our structure until the DRM
driver doesn't need it anymore.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-52-maxime@cerno.tech

drm/vc4: txp: Remove duplicate regset

There's already a regset in the vc4_crtc structure so there's no need to
duplicate it in vc4_txp.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-51-maxime@cerno.tech

drm/vc4: txp: Remove vc4_dev txp pointer

There's no user for that pointer so let's just get rid of it.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-50-maxime@cerno.tech

drm/vc4: hdmi: Switch to devm_pm_runtime_enable

devm_pm_runtime_enable() simplifies the driver a bit since it will call
pm_runtime_disable() automatically through a device-managed action.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-49-maxime@cerno.tech

drm/vc4: hdmi: Protect device resources after removal

Whenever the device and driver are unbound, the main device and all the
subdevices will be removed by calling their unbind() method.

However, the DRM device itself will only be freed when the last user will
have closed it.

It means that there is a time window where the device and its resources
aren't there anymore, but the userspace can still call into our driver.

Fortunately, the DRM framework provides the drm_dev_enter() and
drm_dev_exit() functions to make sure our underlying device is still there
for the section protected by those calls. Let's add them to the HDMI driver.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-48-maxime@cerno.tech

drm/vc4: hdmi: Move audio structure offset checks

The HDMI driver unbind hook doesn't have any ALSA-related code anymore, so
let's move the ALSA sanity checks and comments we have to some other part
of the driver dedicated to ALSA.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-47-maxime@cerno.tech

drm/vc4: hdmi: Use devm to register hotplug interrupts

Commit 776efe800fed ("drm/vc4: hdmi: Drop devm interrupt handler for
hotplug interrupts") dropped the device-managed interrupt registration
because it was creating bugs and races whenever an interrupt was coming in
while the device was removed.

However, our latest patches to the HDMI controller driver fix this as well,
so we can use device-managed interrupt handlers again.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-46-maxime@cerno.tech

drm/vc4: hdmi: Switch to DRM-managed kfree to build regsets

The current code to build the registers set later exposed in debugfs for
the HDMI controller relies on traditional allocations, that are later
free'd as part of the driver unbind hook.

Since krealloc doesn't have a DRM-managed equivalent, let's add an action
to free the buffer later on.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-45-maxime@cerno.tech

drm/vc4: hdmi: Use a device-managed action for DDC

The reference to the DDC controller device needs to be put back when we're
done with it. Let's use a device-managed action to simplify the driver.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-44-maxime@cerno.tech

drm/vc4: hdmi: Switch to device-managed CEC initialization

The current code to unregister our CEC device needs to be undone manually
when we remove the HDMI driver.

Since the CEC framework will allocate its main structure, and will defer
its deallocation to when the last user will have closed it, we don't really
need to take any particular measure to prevent any use-after-free and can
thus use any managed action.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-43-maxime@cerno.tech

drm/vc4: hdmi: Switch to device-managed ALSA initialization

The current code to unregister our ALSA device needs to be undone manually
when we remove the HDMI driver.

Since ALSA doesn't seem to support any mechanism to defer freeing something
until the last user of the ALSA device is gone, we can either use a
device-managed or a DRM-managed action.

The consistent way would be to use a DRM-managed one, just like pretty much
any framework-facing structure should be doing. However, ALSA does a lot of
allocation and registration using device-managed calls. Thus, if we're
going that way, by the time the DRM-managed action would run all of those
allocation would have been freed and we would end up with a use-after-free.

Thus, let's do a device-managed action. It's been tested with KASAN enabled
and doesn't seem to trigger any issue, so it's as good as anything.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-42-maxime@cerno.tech

drm/vc4: hdmi: Switch to DRM-managed connector initialization

The current code will call drm_connector_unregister() and
drm_connector_cleanup() when the device is unbound. However, by then, there
might still be some references held to that connector, including by the
userspace that might still have the DRM device open.

Let's switch to a DRM-managed initialization to clean up after ourselves
only once the DRM device has been last closed.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-41-maxime@cerno.tech

drm/vc4: hdmi: Switch to DRM-managed encoder initialization

The current code will call drm_encoder_cleanup() when the device is
unbound. However, by then, there might still be some references held to
that encoder, including by the userspace that might still have the DRM
device open.

Let's switch to a DRM-managed initialization to clean up after ourselves
only once the DRM device has been last closed.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-40-maxime@cerno.tech

drm/vc4: hdmi: Remove call to drm_connector_unregister()

drm_connector_unregister() is only to be used for connectors that have been
registered through drm_connector_register() after drm_dev_register() has
been called. This is our case here so let's remove the call.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-39-maxime@cerno.tech

drm/vc4: hdmi: Switch to drmm_kzalloc

Our internal structure that stores the DRM entities structure is allocated
through a device-managed kzalloc.

This means that this will eventually be freed whenever the device is
removed. In our case, the most likely source of removal is that the main
device is going to be unbound, and component_unbind_all() is being run.

However, it occurs while the DRM device is still registered, which will
create dangling pointers, eventually resulting in use-after-free.

Switch to a DRM-managed allocation to keep our structure until the DRM
driver doesn't need it anymore.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-38-maxime@cerno.tech

drm/vc4: dsi: Switch to devm_pm_runtime_enable

devm_pm_runtime_enable() simplifies the driver a bit since it will call
pm_runtime_disable() automatically through a device-managed action.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-37-maxime@cerno.tech

drm/vc4: dsi: Fix the driver structure lifetime

The vc4_dsi structure is currently allocated through a device-managed
allocation. This can lead to use-after-free issues however in the unbinding
path since the DRM entities will stick around, but the underlying structure
has been freed.

However, we can't just fix it by using a DRM-managed allocation like we did
for the other drivers since the DSI case is a bit more intricate.

Indeed, the structure will be allocated at probe time, when we don't have a
DRM device yet, to be able to register the DSI bus driver. We will then
reuse it at bind time to register our KMS entities in the framework.

In order to work around both constraints, we can use a kref to track the
users of the structure (DSI host, and KMS), and then put our structure when
the DSI host will have been unregistered, and through a DRM-managed action
that will execute once we won't need the KMS entities anymore.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-36-maxime@cerno.tech

drm/vc4: dsi: Switch to drmm_of_get_bridge

The current code uses a device-managed function to retrieve the next bridge
downstream.

However, that means that it will be removed at unbind time, where the DRM
device is still very much live and might still have some applications that
still have it open.

Switch to a DRM-managed variant to clean everything up once the DRM device
has been last closed.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-35-maxime@cerno.tech

drm/vc4: dsi: Switch to DRM-managed encoder initialization

The current code will call drm_encoder_cleanup() when the device is
unbound. However, by then, there might still be some references held to
that encoder, including by the userspace that might still have the DRM
device open.

Let's switch to a DRM-managed initialization to clean up after ourselves
only once the DRM device has been last closed.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-34-maxime@cerno.tech

drm/vc4: dsi: Embed DRM structures into the private structure

The VC4 DSI driver private structure contains only a pointer to the
encoder it implements. This makes the overall structure somewhat
inconsistent with the rest of the driver, and complicates its
initialisation without any apparent gain.

Let's embed the drm_encoder structure (through the vc4_encoder one) into
struct vc4_dsi to fix both issues.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-33-maxime@cerno.tech

drm/vc4: dpi: Protect device resources

Our current code now mixes some resources whose lifetime are tied to the
device (clocks, IO mappings, etc.) and some that are tied to the DRM device
(encoder, bridge).

The device one will be freed at unbind time, but the DRM one will only be
freed when the last user of the DRM device closes its file handle.

So we end up with a time window during which we can call the encoder hooks,
but we don't have access to the underlying resources and device.

Let's protect all those sections with drm_dev_enter() and drm_dev_exit() so
that we bail out if we are during that window.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-32-maxime@cerno.tech

drm/vc4: dpi: Switch to drmm_of_get_bridge

The current code uses a device-managed function to retrieve the next bridge
downstream.

However, that means that it will be removed at unbind time, where the DRM
device is still very much live and might still have some applications that
still have it open.

Switch to a DRM-managed variant to clean everything up once the DRM device
has been last closed.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-31-maxime@cerno.tech

drm/vc4: dpi: Switch to DRM-managed encoder initialization

The current code will call drm_encoder_cleanup() when the device is
unbound. However, by then, there might still be some references held to
that encoder, including by the userspace that might still have the DRM
device open.

Let's switch to a DRM-managed initialization to clean up after ourselves
only once the DRM device has been last closed.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-30-maxime@cerno.tech

drm/vc4: dpi: Add action to disable the clock

The DPI controller has two clocks called core and pixel, the core clock
being enabled at bind time.

Adding a device-managed action will make the error path easier, so let's
create one to disable it.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-29-maxime@cerno.tech

drm/vc4: dpi: Remove unnecessary drm_of_panel_bridge_remove call

Since we have a managed call to create our panel_bridge instance, the call
to drm_of_panel_bridge_remove() at unbind is both redundant and dangerous
since it might lead to a use-after-free.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-28-maxime@cerno.tech

drm/vc4: dpi: Return an error if we can't enable our clock

If we fail to enable the DPI clock, we just ignore the error and moves
forward. Let's return an error instead.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-27-maxime@cerno.tech

drm/vc4: dpi: Switch to drmm_kzalloc

Our internal structure that stores the DRM entities structure is allocated
through a device-managed kzalloc.

This means that this will eventually be freed whenever the device is
removed. In our case, the most likely source of removal is that the main
device is going to be unbound, and component_unbind_all() is being run.

However, it occurs while the DRM device is still registered, which will
create dangling pointers, eventually resulting in use-after-free.

Switch to a DRM-managed allocation to keep our structure until the DRM
driver doesn't need it anymore.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20220711173939.1132294-26-maxime@cerno.tech