media: mediatek: vcodec: Initialize decoder parameters for each instance

The decoder parameters are stored in each instance's context data. This
needs to be initialized per-instance, but a previous fix incorrectly
changed it to only be initialized for the first opened instance. This
resulted in subsequent instances not correctly signaling the requirement
for the Requests API.

Fix this by calling the initializing function outside of the
v4l2_fh_is_singular() conditional block.

Fixes: faddaa735c20 ("media: mediatek: vcodec: Initialize decoder parameters after getting dec_capability")
Signed-off-by: Chen-Yu Tsai <wenst@chromium.org>
Reviewed-by: Yunfei Dong <yunfei.dong@mediatek.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: mediatek: vcodec: decoder: Embed framesize inside mtk_video_fmt

Right now the decoder maintains two separate lists for supported pixel
formats and frame sizes. Getting the supported frame sizes for the
current set format is a bit convoluted, requiring a search through
the separate frame size list. The frame sizes are used to clamp and
align requested resolutions.

Instead, the frame size structure could be embedded inside the pixel
format structure. Getting one also gets the other. And since the
the driver already keeps pointers to the current set format, getting
the frame sizes becomes straightforward.

Do just that. Move v4l2_frmsize_stepwise inside mtk_video_fmt, and get
rid of mtk_codec_framesizes.

Signed-off-by: Chen-Yu Tsai <wenst@chromium.org>
Tested-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: mediatek: vcodec: decoder: Drop max_{width,height} from mtk_vcodec_ctx

This partially reverts commit b018be06f3c7 ("media: mediatek: vcodec:
Read max resolution from dec_capability").

After the previous patches:

  - media: mediatek: vcodec: decoder: Fix 4K frame size enumeration
  - media: mediatek: vcodec: decoder: Skip alignment for default resolution
  - media: mediatek: vcodec: decoder: Fix resolution clamping in TRY_FMT

the max_{width,height} fields in |struct mtk_vcodec_ctx| no longer have
any real users. Remove them.

Fixes: b018be06f3c7 ("media: mediatek: vcodec: Read max resolution from dec_capability")
Signed-off-by: Chen-Yu Tsai <wenst@chromium.org>
Tested-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: mediatek: vcodec: decoder: Fix resolution clamping in TRY_FMT

In commit b018be06f3c7 ("media: mediatek: vcodec: Read max resolution
from dec_capability"), TRY_FMT clamps the resolution to the maximum
that was previously set either by default 1080p or the limit set by a
previous S_FMT call. This does not make sense when doing TRY_FMT for
the output side, which may have different capabilities.

Instead, for the output side, find the maximum resolution based on the
pixel format requested. For the capture side, find the maximum
resolution based on the currently set output format.

The maximum resolution is found from the list of per-format frame
sizes, so the patch "media: mediatek: vcodec: dec: Fix 4K frame size
enumeration" is needed.

Fixes: b018be06f3c7 ("media: mediatek: vcodec: Read max resolution from dec_capability")
Signed-off-by: Chen-Yu Tsai <wenst@chromium.org>
Tested-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: mediatek: vcodec: decoder: Skip alignment for default resolution

The default resolution of 64x64 is already aligned, according to the
call to v4l_bound_align_image() in mtk_vcodec_dec_set_default_params().

Drop the redundant v4l_bound_align_image() call. This also removes one
usage of ctx->max_{width,height}.

Signed-off-by: Chen-Yu Tsai <wenst@chromium.org>
Tested-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: mediatek: vcodec: decoder: Fix 4K frame size enumeration

This partially reverts commit b018be06f3c7 ("media: mediatek: vcodec:
Read max resolution from dec_capability"). In this commit, the maximum
resolution ended up being a function of both the firmware capability and
the current set format.

However, frame size enumeration for output (coded) formats should not
depend on the format set, but should return supported resolutions for
the format requested by userspace.

Fix this so that the driver returns the supported resolutions correctly,
even if the instance only has default settings, or if the output format
is currently set to VP8F, which does not support 4K.

This adds an copy of special casing for !VP8 and 4K support. The other
existing copy will be removed when .max_{width,height} are removed from
|struct mtk_vcodec_ctx| in a subsequent patch.

Fixes: b018be06f3c7 ("media: mediatek: vcodec: Read max resolution from dec_capability")
Signed-off-by: Chen-Yu Tsai <wenst@chromium.org>
Tested-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: mediatek: vcodec: decoder: Const-ify stepwise_fhd

stepwise_fhd is the reference framesize variable, and should not be
altered. Make it constant.

Fixes: ("76250b48de79 media: mediatek: vcodec: Getting supported decoder format types")
Signed-off-by: Chen-Yu Tsai <wenst@chromium.org>
Tested-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Drop usage of shared helpers

None of the shared helpers are used anymore. Make this official by
dropping inclusion of imx-media.h.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Drop V4L2 events support

The only event that the driver allows subscribing to,
V4L2_EVENT_IMX_FRAME_INTERVAL_ERROR, is never generated. Drop events
support.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Inline imx7_csi_enum_pixel_formats()

Inline the imx7_csi_enum_pixel_formats() function in its only caller.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Drop unneeded pixel format validation

The driver won't accept unsupported pixel formats, there's thus no need
to validate it in imx7_csi_pad_link_validate().

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Add V4L2_PIX_FMT_Y14 support

The device supports 14-bit greyscale formats the same way it supports
14-bit Bayer formats.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Fix list of supported formats

The list of supported formats comes from helpers shared with the i.MX6
IPUv3 and is incorrect in multiple regards:

- 10-, 12- and 14-bit Bayer formats are stored in memory as SBGGR10,
  SBGGR12 and SBGGR14 respectively (plus components permutations), not
  SBGGR16. Same thing for greyscale formats.
- 16-bit RAW formats are not supported by the hardware.

Fix the supported formats table.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Simplify default format in try_fmt

When trying a format on the video node, if the requested pixel format
isn't supported, fallback to the default with a simpler logic.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Inline imx7_csi_init_mbus_fmt()

Inline the imx7_csi_init_mbus_fmt() function in its only caller. This
allows simplifying the code.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Drop unneeded imx7_csi_pixfmt fields

The imx7_csi_pixfmt cycles field is never used nor set, the bayer field
is never used, and the planar field is never set. Drop them.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Drop YUV/RGB/BAYER format selectors

All the format lookup functions are called with a ANY selector, drop
selector support.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Simplify default mbus code in try_fmt

When trying a format on the subdev sink pad, if the requested media bus
code isn't supported, fallback to the default with a simpler logic.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Define macro for default mbus code

Define a macro for the default media bus code and use it through the
driver to replace a hardcoded value and a dynamic query from the
pixel_formats table.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Make default formats consistent

Use the same default image width and height for both the CSI subdev and
the video node to achieve a consistent default through the driver.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Drop unsupported YUV and RGB formats

A large number of formats defined in the pixel_formats array are not
supported, as shown by the switch-case check in
imx7_csi_pad_link_validate(). Drop them.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Drop IPU-only formats

The formats specific to the IPUv3 are not supported by the CSI bridge.
Drop them, along with the related code.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Drop IC support from imx7_csi_try_colorimetry()

The imx7_csi_try_colorimetry() function supports the unrelated image
converter hardware as it originates from shared helpers. Drop that.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Replace ipu_color_space with bool yuv field

Replace the enum ipu_color_space cs field in struct imx7_csi_pixfmt with
a bool yuv field. This decouples the driver from the unrelated IPUv3
headers.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Import format helpers

To prepare for code refactoring, copy the format helpers used by this
driver from imx-media-utils.c. Rename structures and functions to avoid
name clashes, no functional change intended.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Drop error message on alloc failure

The kernel memory allocators already complain loudly to the log when
allocation fails. There's no need for an additional message.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Decouple from shared macros

Decouple from the IMX7_CSI_VIDEO_DEF_PIX_WIDTH,
IMX7_CSI_VIDEO_DEF_PIX_HEIGHT and IMX_MEDIA_EOF_TIMEOUT macros defined
in shared helpers by duplicating them in the imx7-media-csi driver, with
a rename to avoid name clashes. No functional change intended.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Decouple from imx_media_dma_buf

Decouple from the imx_media_dma_buf structure defined in shared helpers
by duplicating it in the imx7-media-csi driver, along with the two small
alloc and free functions. No functional change intended.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Store imx7_csi in drv data

Replace the subdev pointer stored in platform drv data with the imx7_csi
pointer. Drop a now redundant dev_set_drvdata() call.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Fold imx_media_video_dev into imx7_csi

Now that the driver doesn't use any helper that relies on the
imx_media_video_dev structure anymore, merge its fields directly in the
imx7_csi structure. Update all the users accordingly, and drop the list
field that isn't used by the driver.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Decouple from imx_media_buffer

Decouple from the imx_media_buffer structure defined in shared helpers
by duplicating it in the imx7-media-csi driver. No functional change
intended.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Fold imx_media_dev into imx7_csi

Now that the driver doesn't use any helper that relies on the
imx_media_dev structure anymore, merge the three fields of the
imx_media_dev structure directly in the imx7_csi structure. Update all
the users accordingly.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Inline pipeline start/stop

The imx7_csi_media_pipeline_set_stream() function has separate branches
for the start and stop paths, surrounded by mutex lock/unlock. That is
very little shared code, inline the locking and corresponding branch in
each of the two callers.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Avoid unnecessary casts

Simplify the imx7_csi_media_pipeline_set_stream() function by passing it
the imx7_csi pointer, which avoids going from subdev to entity and back
to subdev.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Import imx_media_pipeline_set_stream()

To prepare for code refactoring, copy the
imx_media_pipeline_set_stream() helper used by this driver from
imx-media-utils.c. Rename the function to avoid name clashes, no
functional change intended.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Inline imx_media_pipeline_pad()

Inline the imx_media_pipeline_pad() helper, dropping all the code unused
by the imx7-media-csi driver.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Don't initialize unused fields

The imx_media_dev structure contains three fields that are not used by
this driver or any helper code that it calls. Don't initialize them.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Drop imx_media_add_video_device call

There's no need to call imx_media_add_video_device() anymore, as the
video devices list it manages is only used by the control inheritance
mechanism in the helpers, which this driver doesn't use.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Embed imx_media_dev in imx7_csi

There's no need anymore to allocate the imx_media_dev structure
separately from imx7_csi. Embed it.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Import imx_media_dev_init() helper

To prepare for code refactoring, copy the imx_media_dev_init() helper
used by this driver from imx-media-dev-common.c. Rename the function to
avoid name clashes, and leave the v4l2_device notify handler out as it
only serves to implement control inheritance, which this driver doesn't
use. No functional change intended.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Don't set subdev group id

The subdev group id isn't used by any of the remaining helpers used by
this driver, don't set it.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Ensure consistent function prefix

Rename all the functions and macros imported from imx-media-capture.c
with a consistent imx7_csi_video_ prefix. No functional change intended.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Fold capture_priv into imx7_csi

Merge the capture_priv fields in the imx7_csi structure as there's no
more need to have two separate structures. Update all the users
accordingly.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Reorganize imx7_csi structure

The struct imx7_csi fields are in a random order, reorganize them
logically to prepare for adding new fields.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Drop unused controls support

Drop the unused control handler from the driver.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Drop legacy video device support

Drop all the code related to the unused (by this driver) legacy video
device support.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Import video device helpers

To prepare for code refactoring, copy the video device helper code used
by this driver verbatim from imx-media-capture. Rename some functions to
avoid name clashes, and leave the legacy ioctls out as they're not used
by the imx7-media-csi driver. No functional change included.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Remove imx_media_of_add_csi

imx_media_of_add_csi() from imx-media-of.c validates that the fwnode for
the CSI bridge is enabled, and adds the fwnode to the async notifier of
the imxmd. The former is redundant, as if the CSI bridge driver probes,
clearly it is enabled in the DT. The latter is not needed as the imxmd
notifier isn't used anymore. The call is thus not needed and can be
dropped. This removes the dependency of imx7-media-csi.c on
imx-media-of.c.

Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Move format init to probe time

There's no need to wait until the subdev is registered to initialize the
active pad formats. Do so at probe time.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Drop unused frame_interval

The frame_interval field of the imx7_csi structure is set but never
used. Drop it.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Don't populate vdev lists

The imx_media_dev framework maintains a per-pad list of connected video
devices, created once all subdevs have been bound. This is used for two
purposes, updating V4L2 control inheritance when links change, and
relaying subdev events to video nodes. None of these are used by the
imx7-media-csi driver as it implements the MC-centric approach. Drop
them.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Move misc init out of probe()

There is a chunk of miscellaneous initializations related to the CSI
subdev and media pads directly in the probe function. Move them into the
imx7_csi_media_init() function to clean up the probe function.

Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Drop the imx_media notifier

The V4L2 async notifier embedded in the imx_media_dev structure is
triggered when the CSI bridge subdev is registered. We don't need an
async notifier for this, as we can call the .complete() handler directly
from the CSI bridge subdev .registered() handler.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Drop duplicate link creation

The imx_media_create_csi2_links() creates a media controller link
between the CSI-2 receiver and the next entity in the pipeline, which
can be either a video mux (handled by the video-mux driver) or the CSI
bridge itself. This isn't needed, as the link is already created either
by the video-mux driver or by the imx7-media-csi driver itself (in
imx7_csi_notify_bound()).

Drop imx_media_create_csi2_links(), which allows dropping the CSI bridge
subdev grp_id.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Import notifier helpers

To prepare for code refactoring, copy the V4L2 async notifier helper
code used by this driver verbatim from imx-media-dev-common.c. Rename
some functions to avoid name clashes. No functional change included.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Split imx_media_dev from probe()

Prepare for the decoupling of the imx7-media-csi driver from the
IPUv3-based drivers by moving the imx_media_dev handling from probe()
function to separate functions.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: imx: imx7-media-csi: Initialize locks early on

Group all the direct initialization of imx7_csi fields at the beginning
of the imx7_csi_probe() function instead of spreading them around.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rui Miguel Silva <rmfrfs@gmail.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: renesas: rcar_drif: Drop of_match_ptr()

The device_driver structure's of_match_table field exists
unconditionally, so there's no need for of_match_ptr(). This fixes a
compiler warning when test-compiling on non-OF platforms:

drivers/media/platform/renesas/rcar_drif.c:1470:34: warning: unused variable 'rcar_drif_of_table' [-Wunused-const-variable]
static const struct of_device_id rcar_drif_of_table[] = {
                                 ^

Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: xilinx: Drop unexpected word 'with' in comments

There is an unexpected word 'with' in the comments that need to be
dropped.

Signed-off-by: Jiang Jian <jiangjian@cdjrlc.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: mt9p031: Move open subdev op init code into init_cfg

The .open subdev op is not always called soon enough to initialize
the default crop settings. Move all this initialization into .init_cfg
op instead.

Signed-off-by: Marek Vasut <marex@denx.de>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: mt9p031: Add MT9P006 compatible

Add compatible for MT9P006 sensor, which is older variant of MT9P031
and compatible with this driver.

Signed-off-by: Marek Vasut <marex@denx.de>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: dt-bindings: mt9p031: Add MT9P006 compatible string

Add compatible for MT9P006 sensor, which is older variant of MT9P031
and compatible with this driver.

Signed-off-by: Marek Vasut <marex@denx.de>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging/media/atomisp/pci/atomisp: Fix typo in string

Remove the repeated ',' from string

Link: https://lore.kernel.org/linux-media/20220704022807.3215-1-zeming@nfschina.com
Signed-off-by: Li zeming <zeming@nfschina.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: clean up for-loop, remove redundant assignment to variable i

There is a for-loop that initializes variable i but does not use it; the
assignment is redundant and can be removed. The proceeding assignment to
pointer fw can also be moved into the for-loop to clean up the code.

Link: https://lore.kernel.org/linux-media/20220628152451.184416-1-colin.i.king@gmail.com
Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media/atomisp: use max() impl

fix a coccinelle warning by replacing opencoded max()
implementation with max()

WARNING opportunity for max()

Link: https://lore.kernel.org/linux-media/YrNFpF2qvTxpZP8e@karthik-strix-linux.karthek.com
Signed-off-by: Karthik Alapati <mail@karthek.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Fix typo in comments

Remove the repeated word 'and' from comments

file: drivers/staging/media/atomisp/pci/hive_isp_css_include/host/debug_public.h
line: 28

* The buffer has a remote and and a local store

changed to:

* The buffer has a remote and a local store

Link: https://lore.kernel.org/linux-media/20220622143442.16942-1-jiangjian@cdjrlc.com
Signed-off-by: Jiang Jian <jiangjian@cdjrlc.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp_gmin_platform: stop abusing efivar API

As the code comment already suggests, using the efivar API in this way
is not how it is intended, and so let's switch to the right one, which
is simply to call efi.get_variable() directly after checking whether or
not the GetVariable() runtime service is supported.

Link: https://lore.kernel.org/linux-media/20220620100819.1682995-1-ardb@kernel.org
Cc: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: fix -Wdangling-pointer warning

ia_css_rmgr_acq_vbuf() uses a local on stack
"struct ia_css_rmgr_vbuf_handle v" variable.

When this path using this is hit, either the rmgr_pop_handle() call
will make *handle point to another vbuf-handle, or because
v.count == 0, ia_css_rmgr_refcount_retain_vbuf() will alloc a new
vbuf-handle and make *handle point to it.

So on leaving the function *handle will never point to the on stack
vbuf-handle, but gcc does not know this and emits the following:

drivers/staging/media/atomisp/pci/runtime/rmgr/src/rmgr_vbuf.c: In function ‘ia_css_rmgr_acq_vbuf’:
drivers/staging/media/atomisp/pci/runtime/rmgr/src/rmgr_vbuf.c:276:33: warning: storing the address of local variable ‘h’ in ‘*handle’ [-Wdangling-pointer=]
  276 |                         *handle = &h;
      |                         ~~~~~~~~^~~~
drivers/staging/media/atomisp/pci/runtime/rmgr/src/rmgr_vbuf.c:257:40: note: ‘h’ declared here
  257 |         struct ia_css_rmgr_vbuf_handle h;
      |                                        ^
drivers/staging/media/atomisp/pci/runtime/rmgr/src/rmgr_vbuf.c:257:40: note: ‘handle’ declared here

Rework the code using a new_handle helper to suppress this
false-postive compiler warning.

Link: https://lore.kernel.org/linux-media/20220612160556.108264-4-hdegoede@redhat.com
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: fix uninitialized stack mem usage in ia_css_rmgr_acq_vbuf()

When ia_css_rmgr_acq_vbuf() enters the code path where it uses the local
"struct ia_css_rmgr_vbuf_handle v" on the stack it relies on v.count==0
so that ia_css_rmgr_refcount_retain_vbuf allocates a new handle.

Explicitly set v.count to 0 rather then it being whatever was on the stack.

Link: https://lore.kernel.org/linux-media/20220612160556.108264-3-hdegoede@redhat.com
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: revert "don't pass a pointer to a local variable"

The gcc is warning about returning a pointer to a local variable
is a false positive.

The type of handle is "struct ia_css_rmgr_vbuf_handle **" and
"h.vptr" is left to NULL, so the "if ((*handle)->vptr == 0x0)"
check always succeeds when the "*handle = &h;" statement which
gcc warns about executes. Leading to this statement being executed:

rmgr_pop_handle(pool, handle);

If that succeeds,  then *handle has been set to point to one of
the pre-allocated array of handles, so it no longer points to h.

If that fails the following statement will be executed:

/* Note that handle will change to an internally maintained one */
ia_css_rmgr_refcount_retain_vbuf(handle);

Which allocated a new handle from the array of pre-allocated handles
and then makes *handle point to this. So the address of h is actually
never returned.

The fix for the false-postive compiler warning actually breaks the code,
the new:

**handle = h;

is part of a "if (pool->copy_on_write) { ... }" which means that the
handle where *handle points to should be treated read-only, IOW
**handle must never be set, instead *handle must be set to point to
a new handle (with a copy of the contents of the old handle).

The old code correctly did this and the new fixed code gets this wrong.

Note there is another patch in this series, which fixes the warning
in another way.

Link: https://lore.kernel.org/linux-media/20220612160556.108264-2-hdegoede@redhat.com
Fixes: fa1451374ebf ("media: atomisp: don't pass a pointer to a local variable")
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp-mt9m114: Fix pointer dereferenced before checking

The info->data is dereferencing before null checking, so move
it after checking.

Link: https://lore.kernel.org/linux-media/1653897481-25681-1-git-send-email-baihaowen@meizu.com
Signed-off-by: Haowen Bai <baihaowen@meizu.com>
Reviewed-by: Tommaso Merciai <tommaso.merciai@amarulasolutions.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: [PATCH] pci: atomisp_cmd: fix three missing checks on list iterator

The three bugs are here:
__func__, s3a_buf->s3a_data->exp_id);
__func__, md_buf->metadata->exp_id);
__func__, dis_buf->dis_data->exp_id);

The list iterator 's3a_buf/md_buf/dis_buf' will point to a bogus
position containing HEAD if the list is empty or no element is found.
This case must be checked before any use of the iterator, otherwise
it will lead to a invalid memory access.

To fix this bug, add an check. Use a new variable '*_iter' as the
list iterator, while use the old variable '*_buf' as a dedicated
pointer to point to the found element.

Link: https://lore.kernel.org/linux-media/20220414041415.3342-1-xiam0nd.tong@gmail.com
Cc: stable@vger.kernel.org
Fixes: ad85094b293e4 ("Revert "media: staging: atomisp: Remove driver"")
Signed-off-by: Xiaomeng Tong <xiam0nd.tong@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: atomisp: Use kmap_local_page() in hmm_store()

The use of kmap() is being deprecated in favor of kmap_local_page()
where it is feasible. The same is true for kmap_atomic().

In file pci/hmm/hmm.c, function hmm_store() test if we are in atomic
context and, if so, it calls kmap_atomic(), if not, it calls kmap().

First of all, in_atomic() shouldn't be used in drivers. This macro
cannot always detect atomic context; in particular, it cannot know
about held spinlocks in non-preemptible kernels.

Notwithstanding what it is said above, this code doesn't need to care
whether or not it is executing in atomic context. It can simply use
kmap_local_page() / kunmap_local() that can instead do the mapping /
unmapping regardless of the context.

With kmap_local_page(), the mapping is per thread, CPU local and not
globally visible. Therefore, hmm_store()() is a function where the use
of kmap_local_page() in place of both kmap() and kmap_atomic() is
correctly suited.

Convert the calls of kmap() / kunmap() and kmap_atomic() /
kunmap_atomic() to kmap_local_page() / kunmap_local() and drop the
unnecessary tests which test if the code is in atomic context.

Link: https://lore.kernel.org/linux-media/20220413225531.9425-1-fmdefrancesco@gmail.com
Signed-off-by: Fabio M. De Francesco <fmdefrancesco@gmail.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: atomisp: Use kmap_local_page() in hmm_set()

The use of kmap() is being deprecated in favor of kmap_local_page()
where it is feasible. In file pci/hmm/hmm.c, function hmm_set() calls
kmap() / kunmap() where kmap_local_page() can instead do the mapping.

With kmap_local_page(), the mapping is per thread, CPU local and not
globally visible. Therefore, hmm_set()() is a function where the use
of kmap_local_page() in place of kmap() is correctly suited.

Convert the calls of kmap() / kunmap() to kmap_local_page() /
kunmap_local().

Link: https://lore.kernel.org/linux-media/20220413212210.18494-1-fmdefrancesco@gmail.com
Signed-off-by: Fabio M. De Francesco <fmdefrancesco@gmail.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: atomisp: Convert kmap() to kmap_local_page()

The use of kmap() is being deprecated in favor of kmap_local_page() where
it is feasible. With kmap_local_page(), the mapping is per thread, CPU
local and not globally visible.

load_and_flush_by_kmap() is a function where the use of kmap_local_page()
in place of kmap() is correctly suited.

Convert load_and_flush_by_kmap() from kmap() to kmap_local_page().

Link: https://lore.kernel.org/linux-media/20220408223129.3844-1-fmdefrancesco@gmail.com
Signed-off-by: Fabio M. De Francesco <fmdefrancesco@gmail.com>
Tested-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: atomisp: rework reading the id and revision values

Clang static analysis reports this representative issue
atomisp-ov2722.c:920:3: warning: 3rd function call
  argument is an uninitialized value
  dev_err(&client->dev, "sensor_id_high = 0x%x\n", high);
  ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

high and low are only set when ov2722_read_reg() is successful.
Reporting the high value when there is an error is not
meaningful.  The later read for low is not checked.  high
and low are or-ed together and checked against a non zero
value.

Remove the unneeded error reporting for high.  Initialize
high and low to 0 and use the id check to determine if
the reads were successful

The later read for revision is not checked.  If it
fails the old high value will be used and the revision
will be misreported.

Since the revision is only reported and not checked or
stored it is not necessary to return if the read with
successful.  This makes the ret variable unnecessary
so remove it.

Link: https://lore.kernel.org/linux-media/20220326191853.2914552-1-trix@redhat.com
Signed-off-by: Tom Rix <trix@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Add a notes.txt file

Add a files documenting what I've learned about the driver while
working on various cleanups.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-41-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: remove force argument from __destroy_[stream[s]|pipe[s]]()

The force argument to the __destroy_pipe[s]() and __destroy_stream[s]()
functions is always true. Remove the argument and remove the code necessary
to handle the false case.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-40-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: use atomisp_destroy_pipes_stream_force() in more places

Use atomisp_destroy_pipes_stream_force() in 4 more places,
instead of open coding it.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-39-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: use atomisp_css_update_stream() in more places

Use atomisp_css_update_stream() in 2 more places,
instead of open coding it.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-38-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: use atomisp_create_pipes_stream() in more places

Use atomisp_create_pipes_stream() in 2 more places,
instead of open coding it.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-37-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: add error logging to atomisp_destroy_pipes_stream_force()

__destroy_streams() and __destroy_pipes() may return an error.
Log a warning when either of them fails.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-36-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: add error checking to atomisp_create_pipes_stream()

The functions called by atomisp_create_pipes_stream() can fail,
add error checking for them.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-35-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: fix __get_frame_info() error handling

On ia_css_pipe_get_info() errors, destroy both the streams as well
as the pipes which were created.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-34-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: remove hmm_page_object

hmm_page_object only stores a struct page pointer, so we can just use
the hmm_bo.pages page pointer array everywhere.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-33-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: drop HMM_BO_SHARE type

HMM_BO_SHARE is not supported by the hmm_bo code at all, drop it.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-32-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: drop highmem var/arg from the hmm code

highmem is always false, drop it.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-31-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Simplify hmm_alloc() calls

Make hmm_alloc() only take size as a parameter and remove other parameters.
since all callers always pass the same flags.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-30-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: add hmm_create_from_userdata() helper

Most hmm_alloc() callers want BO_PRIVATE type memory.

Add a hmm_create_from_userdata() helper for other cases so that
the hmm_alloc() calls for all the callers who don't want this
can be simplied.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-29-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: remove unused hmm address translation functions

hmm_isp_vaddr_to_host_vaddr() and hmm_host_vaddr_to_hrt_vaddr()
are unused, remove them.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-28-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: drop unused ATOMISP_MAP_FLAG_* flags

Drop the ATOMISP_MAP_FLAG_CACHED flag, it is never set anywhere;
also drop the matching "cached" parameter to hmm[_bo]_alloc which
value was derived form the never set flag.

Drop the ATOMISP_MAP_FLAG_NOFLUSH, it is not used anywhere.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-27-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: drop ATOMISP_MAP_FLAG_CLEARED

This flag is only used by one hmm_alloc() caller, drop it and make
the caller call hmm_set(ptr, 0, size) itself to do the clearing.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-26-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: drop unused ATOMISP_ACC_FW_LOAD_* defines

Drop the ATOMISP_ACC_FW_LOAD_* defines, these are no longer used anywhere.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-25-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: remove atomisp_is_acc_enabled()

atomisp_is_acc_enabled() always returns false now, remove it.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-24-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: remove no longer used atomisp_css_acc_done() function

atomisp_css_acc_done() is no longer used anywhere, remove it.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-23-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: asc.acc.pipeline is always NULL

With the removal of the ACC ioctls and atomisp_acc.c asc.acc.pipeline
never gets set, so it is always NULL.

Remove asc.acc.pipeline and drop checks for it being NULL / !NULL.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-22-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: remove unused atomisp_*css_* functions

With the removal of the ACC ioctls and atomisp_acc.c a whole bunch
of atomisp_*css_* functions is no longer used, remove them.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-21-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: remove atomisp_acc.c

With the ACC ioctls removed sd->acc.fw is always empty turning
the atomisp_acc.c code into no-ops, remove it.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-20-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: remove private acceleration ioctls

These ioctls allow userspace to load custom programs into the ISP, which:

a) Seems dangerous
b) Cannot be used by opensource userspace since there is no FOSS code to
   create such programs
b) These seem to be unused even by the Android closed source camera code
   (they don't show up in a strace of the camera app)

So removing these seems be a good idea. Another reason to remove these is
that atomisp_acc_map() is the only user of the userptr functionality in
hmm_alloc(), so it gets in the way of further cleanups / simplification
of the hmm code.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-19-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: remove bogus comment above hmm_bo_allocated() prototype

The comment documenting hmm_bo_allocated() was copied (and not modified)
from the comment documenting hmm_bo_alloc(), so there are 2 copies
of the hmm_bo_alloc() documentation.

Remove the copy of the comment above the hmm_bo_allocated() prototype.

Link: https://lore.kernel.org/linux-media/20220615205037.16549-18-hdegoede@redhat.com
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>