media: vim2m: initialize the media device earlier

Before the video device node is registered, the v4l2_dev.mdev
pointer must be set in order to correctly associate the video
device with the media device. Move the initialization of the
media device up.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: vp8: Fix kernel-doc warnings

Fix following warnings:

./scripts/kernel-doc --none include/uapi/linux/v4l2-controls.h

        include/uapi/linux/v4l2-controls.h:1727: warning: bad line:
        include/uapi/linux/v4l2-controls.h:1853: warning: expecting prototype for struct v4l2_vp8_frame. Prototype was for struct v4l2_ctrl_vp8_frame instead
        include/uapi/linux/v4l2-controls.h:1853: warning: Function parameter or member 'segment' not described in 'v4l2_ctrl_vp8_frame'
        include/uapi/linux/v4l2-controls.h:1853: warning: Function parameter or member 'entropy' not described in 'v4l2_ctrl_vp8_frame'
        include/uapi/linux/v4l2-controls.h:1853: warning: Function parameter or member 'coder_state' not described in 'v4l2_ctrl_vp8_frame'

Reported-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: move VP8 stateless controls out of staging

Until now, the VP8 V4L2 API was not exported as a public API,
and only defined in a private media header (media/vp8-ctrls.h).

The reason for this was a concern about the API not complete
and ready to support VP8 decoding hardware accelerators.

After reviewing the VP8 specification in detail, and now
that the API is able to support Cedrus and Hantro G1,
we can consider this ready.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: vp8: Rename V4L2 VP8 specific macro to V4L2_VP8_

Before we move the VP8 controls to a stable public header,
rename this macro so it's clearly V4L2-specific.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: controls: Log VP8 stateless control in .std_log

Simply print the type of the control.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: Move the VP8 stateless control type out of staging

Move the VP8 stateless control types out of staging,
and re-number it to avoid any confusion.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: Move parsed VP8 pixel format out of staging

Since we are ready to stabilize the VP8 stateless API,
move the parsed VP8 pixel format.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: vp8: Add proper kernel-doc documentation

In preparation for making the interface public,
document all the structures.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: vp8: Rename v4l2_vp8_loopfilter to v4l2_vp8_loop_filter

Rename to struct v4l2_vp8_loop_filter, which is closer to the
VP8 syntax. This change is only cosmetics, but since this API
is exposed to userspace, it's worth the trouble to get it consistent.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: vp8: Remove "header" from symbol names and macros

It doesn't seem to add any clarity to have a "header" suffix in controls,
struct names and flags.

Since this just makes names too long without any benefit, just drop it.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: venus: core: Parse firmware-name DT property

On production devices the firmware could be located on different
places, this path could be provided by special firmware-name DT
property.

Here we check for existence of such DT property and if it exist
take the firmware path from there. Otherwise, if the property
is missing we fallback to the predefined path from driver resource
structure.

Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@somainline.org>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: venus: Include io.h for memremap()

This file uses memremap() now, so we should include io.h instead of
relying on any sort of implicit include elsewhere.

Cc: Dikshita Agarwal <dikshita@codeaurora.org>
Fixes: 0ca0ca980505 ("media: venus: core: add support to dump FW region")
Signed-off-by: Stephen Boyd <swboyd@chromium.org>
Reviewed-by: Fritz Koenig <frkoenig@chromium.org>
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: venus: core, venc, vdec: Fix probe dependency error

Commit aaaa93eda64b ("media] media: venus: venc: add video encoder files")
is the last in a series of three commits to add core.c vdec.c and venc.c
adding core, encoder and decoder.

The encoder and decoder check for core drvdata as set and return -EPROBE_DEFER
if it has not been set, however both the encoder and decoder rely on
core.v4l2_dev as valid.

core.v4l2_dev will not be valid until v4l2_device_register() has completed
in core.c's probe().

Normally this is never seen however, Dmitry reported the following
backtrace when compiling drivers and firmware directly into a kernel image.

[    5.259968] Hardware name: Qualcomm Technologies, Inc. Robotics RB5 (DT)
[    5.269850] sd 0:0:0:3: [sdd] Optimal transfer size 524288 bytes
[    5.275505] Workqueue: events deferred_probe_work_func
[    5.275513] pstate: 60400005 (nZCv daif +PAN -UAO -TCO BTYPE=--)
[    5.441211] usb 2-1: new SuperSpeedPlus Gen 2 USB device number 2 using xhci-hcd
[    5.442486] pc : refcount_warn_saturate+0x140/0x148
[    5.493756] hub 2-1:1.0: USB hub found
[    5.496266] lr : refcount_warn_saturate+0x140/0x148
[    5.500982] hub 2-1:1.0: 4 ports detected
[    5.503440] sp : ffff80001067b730
[    5.503442] x29: ffff80001067b730
[    5.592660] usb 1-1: new high-speed USB device number 2 using xhci-hcd
[    5.598478] x28: ffff6c6bc1c379b8
[    5.598480] x27: ffffa5c673852960 x26: ffffa5c673852000
[    5.598484] x25: ffff6c6bc1c37800 x24: 0000000000000001
[    5.810652] x23: 0000000000000000 x22: ffffa5c673bc7118
[    5.813777] hub 1-1:1.0: USB hub found
[    5.816108] x21: ffffa5c674440000 x20: 0000000000000001
[    5.820846] hub 1-1:1.0: 4 ports detected
[    5.825415] x19: ffffa5c6744f4000 x18: ffffffffffffffff
[    5.825418] x17: 0000000000000000 x16: 0000000000000000
[    5.825421] x15: 00000a4810c193ba x14: 0000000000000000
[    5.825424] x13: 00000000000002b8 x12: 000000000000f20a
[    5.825427] x11: 000000000000f20a x10: 0000000000000038
[    5.845447] usb 2-1.1: new SuperSpeed Gen 1 USB device number 3 using xhci-hcd
[    5.845904]
[    5.845905] x9 : 0000000000000000 x8 : ffff6c6d36fae780
[    5.871208] x7 : ffff6c6d36faf240 x6 : 0000000000000000
[    5.876664] x5 : 0000000000000004 x4 : 0000000000000085
[    5.882121] x3 : 0000000000000119 x2 : ffffa5c6741ef478
[    5.887578] x1 : 3acbb3926faf5f00 x0 : 0000000000000000
[    5.893036] Call trace:
[    5.895551]  refcount_warn_saturate+0x140/0x148
[    5.900202]  __video_register_device+0x64c/0xd10
[    5.904944]  venc_probe+0xc4/0x148
[    5.908444]  platform_probe+0x68/0xe0
[    5.912210]  really_probe+0x118/0x3e0
[    5.915977]  driver_probe_device+0x5c/0xc0
[    5.920187]  __device_attach_driver+0x98/0xb8
[    5.924661]  bus_for_each_drv+0x68/0xd0
[    5.928604]  __device_attach+0xec/0x148
[    5.932547]  device_initial_probe+0x14/0x20
[    5.936845]  bus_probe_device+0x9c/0xa8
[    5.940788]  device_add+0x3e8/0x7c8
[    5.944376]  of_device_add+0x4c/0x60
[    5.948056]  of_platform_device_create_pdata+0xbc/0x140
[    5.953425]  of_platform_bus_create+0x17c/0x3c0
[    5.958078]  of_platform_populate+0x80/0x110
[    5.962463]  venus_probe+0x2ec/0x4d8
[    5.966143]  platform_probe+0x68/0xe0
[    5.969907]  really_probe+0x118/0x3e0
[    5.973674]  driver_probe_device+0x5c/0xc0
[    5.977882]  __device_attach_driver+0x98/0xb8
[    5.982356]  bus_for_each_drv+0x68/0xd0
[    5.986298]  __device_attach+0xec/0x148
[    5.990242]  device_initial_probe+0x14/0x20
[    5.994539]  bus_probe_device+0x9c/0xa8
[    5.998481]  deferred_probe_work_func+0x74/0xb0
[    6.003132]  process_one_work+0x1e8/0x360
[    6.007254]  worker_thread+0x208/0x478
[    6.011106]  kthread+0x150/0x158
[    6.014431]  ret_from_fork+0x10/0x30
[    6.018111] ---[ end trace f074246b1ecdb466 ]---

This patch fixes by

- Only setting drvdata after v4l2_device_register() completes
- Moving v4l2_device_register() so that suspend/reume in core::probe()
  stays as-is
- Changes pm_ops->core_function() to take struct venus_core not struct
  device
- Minimal rework of v4l2_device_*register in probe/remove

Reported-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: venus: core: Fix some resource leaks in the error path of 'venus_probe()'

If an error occurs after a successful 'of_icc_get()' call, it must be
undone.

Use 'devm_of_icc_get()' instead of 'of_icc_get()' to avoid the leak.
Update the remove function accordingly and axe the now unneeded
'icc_put()' calls.

Fixes: 32f0a6ddc8c9 ("media: venus: Use on-chip interconnect API")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: venus: venc: Add support for AUD NALU control

Add support for Access Unit Delimiter control into encoder.

Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: v4l2-ctrls: Add control for AUD generation

Add a control to enable inserting of AUD NALU into encoded
bitstream.

Reviewed-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: docs: Deprecate mfc display delay controls

Deprecate mfc private display delay and display enable controls for
new clients and use the standard controls instead.

Acked-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: s5p-mfc: Use display delay and display enable std controls

Use the standard display_delay and display_delay_enable controls,
the legacy private MFC controls are kept for backward compatibility.

Acked-by: Marek Szyprowski <m.szyprowski@samsung.com>
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: venus: vdec: Add support for display delay and delay enable controls

Add support for display delay and display delay enable std controls.
With this we implement decoder decode output order (decode vs display).
Once firmware implement few new features the controls will be used
for other use-cases.

Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: v4l2-ctrl: Make display delay and display enable std controls

Make display delay and display delay enable MFC controls standard v4l
controls. This will allow reuse of the controls for other decoder
drivers. Also the new proposed controls are now codec agnostic because
they could be used for any codec.

Acked-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: add a subsystem profile documentation

Document the basic policies of the media subsystem profile.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

atomisp: don't let it go past pipes array

In practice, IA_CSS_PIPE_ID_NUM should never be used when
calling atomisp_q_video_buffers_to_css(), as the driver should
discover the right pipe before calling it.

Yet, if some pipe parsing issue happens, it could end using
it.

So, add a WARN_ON() to prevent such case.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_mipi_csis: Print shadow registers in mipi_csis_dump_regs()

Print the value of the ISP shadow registers in mipi_csis_dump_regs() as
this can help debugging.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_mipi_csis: Use register macros in mipi_csis_dump_regs()

Replace hardcoded register addresses in the MIPI_CSIS_DBG_CTRL()
function with macros. While at it, update the printed name of the
registers to match the datasheet, and short them by address.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_mipi_csis: Rename register macros to match datasheet

Rename several register macros to match the names from the
documentation.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_mipi_csis: Make ISP registers macros take channel ID

Replace the per-channel ISP registers macros with a single one that take
the channel as a parameter. Only channel 0 is supported for now, but
this will make support for multiple channels easier.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_mipi_csis: Define macros for DPHY_BCTRL_L fields

Define macros for the fields of the MIPI_CSIS_DPHY_BCTRL_L register,
based on the i.MX8MM datasheet. The numerical value written to the
register, taken from the i.MX7D code, is unchanged, and corresponds to a
20.0MHz Tx escape clock frequency according to the documentation of the
register from the i.MX8MM. Whether the other fields are applicable to
the i.MX7D is unknown.

While at it, rename the DPHY[BS]CTRL register names to DPHY_[BS]CTRL to
match the datasheet.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_mipi_csis: Fully initialize MIPI_CSIS_DPHYCTRL register

When setting the CSIS parameters, write the MIPI_CSIS_DPHYCTRL register
fully instead of modifying selected fields, as the register doesn't
contain any reserved fields that need to be preserved. This simplifies
initialization slightly, and ensures that the register value doesn't
depend on its previous state (before a warm reboot for instance).

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_mipi_csis: Turn register access macros into functions

Make it easier to instrument register access (for instance with
printk-based logging) by turning the macros into inline functions.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_mipi_csis: Calculate Ths_settle from source lane rate

The Ths_settle timing parameter depends solely on the lane data rate of
the source. Calculate it at runtime instead of requiring it to be
specified in the device tree.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_mipi_csis: Move link setup check out of locked section

Checking if the link setup operation is called for the sink or source
pad doesn't require any locking. Move it out of the section protected by
the mutex.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_mipi_csis: Inline mipi_csis_set_hsync_settle()

The mipi_csis_set_hsync_settle() is small, called from a single place,
and misnamed (HS stands for high speed, not horizontal sync). Inline it
in its only caller, and refactor the HSSETTLE register field macros
while at it.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_mipi_csis: Fix UYVY8 media bus format

CSI-2 uses, as a convention, 1Xn media bus formats. Replace
MEDIA_BUS_FMT_UYVY8_2X8 with MEDIA_BUS_FMT_UYVY8_1X16, and set the width
value accordingly.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_mipi_csis: Drop 10-bit YUV support

The data type specified for 10-bit YUV is the same as 8-bit YUV. This is
wrong, and has thus clearly not been tested. Instead of fixing data type
without being able to test it, drop it. It can be added back later when
someone will have access to a 10-bit YUV source to this it.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_mipi_csis: Avoid double get of wrap clock

The wrap clock is retrieved with the other clocks using
devm_clk_bulk_get(), and retrieved independently with a specific
devm_clk_get() call. The latter isn't needed, replace it with usage of
the wrap clock from the bulk array.

As the wrap clock is mandatory, simplify its usage further by removing
conditionals.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_mipi_csis: Make source .s_power() optional

The .s_power() subdev operation is optional, don't error out when the
source doesn't implement it.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_mipi_csis: Fix input size alignment

The alignments given to the v4l_bound_align_image() are expressed in
bytes, while the function expects them to be expressed as a power of
two. Fix it.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_mipi_csis: Acquire reset control without naming it

The device has a single reset line, there's thus no need to name it
explicitly when calling devm_reset_control_get_exclusive(). Drop the
name in preparation for the removal of the reset-names property in the
DT binding.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_media-csi: Create immutable link to source device

The CSI subdev has a single source, the input video mux. There's thus a
single link between the source and the CSI, which can be made immutable
and enabled to simplify configuration.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: v4l2-mc: Add link flags to v4l2_create_fwnode_links_to_pad()

Add a flags argument to the v4l2_create_fwnode_links_to_pad() function
to specify the link flags. This allows drivers to create immutable links
for instance.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7_media-csi: Add support for additional Bayer patterns

The CSI driver only supports the BGGR Bayer patterns currently. The
hardware supports all patterns (the only pattern-dependent hardware
operation is statistics calculation, as de-bayering isn't supported),
enable them in the driver too.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Don't enable SOF and EOF interrupts

The SOF and EOF interrupts are not used. Don't enable them.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Remove double reflash of DMA controller

The DMA controller needs to be reflashed after being configured. There
is however no need to do it twice, once in imx7_csi_configure() and once
in imx7_csi_sw_reset(), called from imx7_csi_enable(). Remove the
former.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Don't double-enable the RxFIFO

The RxFIFO is enabled by imx7_csi_dmareq_rff_enable() in both
imx7_csi_init() and imx7_csi_enable(). Enabling it once is enough,
remove the former.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Don't double-enable the CSI

The CSI is enabled at the end of imx7_csi_sw_reset(), called from
imx7_csi_enable(), as well as at the end of imx7_csi_enable(). The
latter is enough, remove the former.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Simplify imx7_csi_rx_fifo_clear()

There's no need to read the CSICR1 register before each write in
imx7_csi_rx_fifo_clear(). Simplify the function.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Set RFF burst type in imx7_csi_configure()

The RFF burst type is set in imx7_csi_dmareq_rff_enable(), which doesn't
match the function name. Move it to imx7_csi_configure().

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Clear all configurable CSICR18 fields

When configuring the CSI in imx7_csi_configure(), the CSICR18 is set
through a read-modify-write cycle to avoid affecting fields whose usage
isn't clear. The function then sets fields depending on whether the
input is interlaced or progressive, and from the parallel input or the
CSI-2 receiver.

Those bits are only set and never cleared. For instance, when switching
from a CSI-2 source to the parallel input, the BIT_DATA_FROM_MIPI will
stay set. Fix this issue by first clearing all the fields we need to
configure.

Add BIT_CSI_HW_ENABLE to the set of fields being cleared, as the CSI
needs to start in the disabled state. This allows dropping the call to
imx7_csi_hw_disable() in imx7_csi_sw_reset().

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Merge all config in imx7_csi_configure()

Merge all the configuration steps (imx7_csi_buf_stride_set(),
imx7_csi_deinterlace_enable(), imx7_csi_set_imagpara() and
imx7_csi_init_default()) in the imx7_csi_configure() function. This
simplifies the configuration procedure by removing most
read-update-write cycles, and makes the code generally more readable.

imx7_csi_init_default() is kept as it is also called from
imx7_csi_deinit(). This will be addressed separately.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Don't set the buffer stride when disabling

There's no need to set the buffer stride to 0 when disabling the CSI.
Remove the extraneous register write.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Set the MIPI data type based on the bus code

The MIPI CSI-2 data type depends solely on the media bus code on the
sink pad. Deriving it from the pixel format is confusing. Simplify the
code.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Merge hw_reset() with init_interface()

The imx7_csi_hw_reset() and imx7_csi_init_interface() functions are
always called together. Merge them. This allows simplifying the code by
avoiding duplicated register writes.

As the imx7_csi_hw_reset() function didn't perform a hardware reset by
initialized the registers to reset defaults (in addition to resetting
the frame counter), name the resulting function imx7_csi_init_default()
as it sets default values.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Merge streaming_start() with csi_enable()

The imx7_csi_streaming_start() function just wraps imx7_csi_enable().
Call the latter directly.

Similarly, merge imx7_csi_streaming_stop() and imx7_csi_disable() as
they're both called from a single location only.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Move CSI configuration before source start

There's no reason to delay the CSI configuration and the DMA setup after
starting the source. Move it before, simplifying error handling at
stream start.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Split imx7_csi_dma_stop()

The imx7_csi_dma_stop() function performs two operations, it waits for
DMA to be complete, and then cleans up and frees buffers. Split the
second part out to a new imx7_csi_dma_cleanup() function to match
imx7_csi_dma_setup(), in preparation for a stream start refactoring.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Rename imx7_csi_dma_start() to *_setup()

The imx7_csi_dma_start() doesn't start DMA. Rename it to
imx7_csi_dma_setup() to avoid confusion.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Don't lock access to is_csi2

The is_csi2 field can't be accessed concurrently by
imx7_csi_pad_link_validate() and imx7_csi_configure(), as the latter is
called from imx7_csi_s_stream(), which is called after link validation.
Drop the lock.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Fix source type identification

The code in imx7_csi_pad_link_validate() that checks the type of the
source incorrectly handles devices that have no parallel input. In that
case, the source entity of the CSI is the CSI-2 receiver, not the video
mux, and the driver will proceed to check the type of the source of the
CSI-2 receiver.

Make the code more explicit to fix this, by handling the three cases
(parallel input only, CSI-2 receiver only, and video mux) separately.

Note that the driver will not correctly handle the case where only a
parallel input is present, and the external entity connected to the
parallel input reports a MEDIA_ENT_F_VID_IF_BRIDGE or
MEDIA_ENT_F_VID_MUX function. This was broken already, and should be
fixed separately.

Fixes: f5ffb81f5137 ("media: imx7: csi: Fix pad link validation")
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Validate capture format in .link_validate()

Move capture format validation from imx7_csi_configure() to
.link_validate(). This simplifies error handling at stream on time, as
imx7_csi_configure) now becomes a void function.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Reorganize code in sections

To prepare for a major rework of the hardware initialization, and make
the code easier to read, reorder functions to group them in sections. No
functional change is included.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Tidy up register fields macros

Describe multi-bit fields with parameterized macros where applicable,
and use them to replace manual shifts and hardcoded numerical values.
Add macros for the CSICR2 fields used in the driver.

While at it, align the indentation for all macros, and use lower-case
hex numbers.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Replace CSICR*_RESET_VAL with values

The CSICR*_RESET_VAL macros only obfuscate code. Use their value
directly, replacing the numerical value with the macros that describe
register bits.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Create immutable link to capture device

The i.MX7 camera pipeline is simple, with the CSI connected directly to
the capture device without any other option. There's thus no need to
allow link configurable by userspace. Make it immutable.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Move (de)init from link setup to .s_stream()

There's no need to initialize the CSI every time a link is enabled (and
de-initialize it when a link is disabled). Move initialization to
.s_stream() instead.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Remove control handler

The control handler isn't used, drop it.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Support creating immutable link to capture device

When the subdevice connected to the capture device has a single possible
sink, there's no point in making the link mutable. Support creating
immutable links.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: imx7-media-csi: Disable legacy video node API

Support for the MC-centric API has been tested on the i.MX7. Enable it
for that platform. i.MX6 should be converted next.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Expose V4L2_CAP_IO_MC for the MC-centric API

Report to userspace that the MC-centric API is MC-centric by exposing
the V4L2_CAP_IO_MC. This requires adding support for mbus code filtering
in format enumeration, as required by V4L2_CAP_IO_MC.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Decouple video node from source with MC-centric API

When operating in MC-centric mode, the behaviour of video nodes shall
not be influenced by the active configuration of the source subdev. Add
a set of ioctl handlers that implement this mode, and select them when
support for the legacy API is not requested.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Simplify __capture_legacy_try_fmt()

The __capture_legacy_try_fmt() function returns two values through
pointer arguments. One is a compose rectangle, which duplicates
informationr returned through the subdev format argument, and can thus
be removed. The other is the imx_media_pixfmt, which can be returned
by value instead.

Simplify the implementation of __capture_legacy_try_fmt() by dropping
the retcc and compose arguments, and returning the imx_media_pixfmt by
value.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Reviewed-by: Steve Longerbeam <slongerbeam@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Simplify capture_validate_fmt() implementation

capture_validate_fmt() delegates the media bus format to pixel format
conversion to __capture_legacy_try_fmt(). It turns out that this can be
simplified quite a lot:

- The format lookup from media bus code can be performed by the recently
  added capture_find_format() function instead of receiving the
  information from a side effect of __capture_legacy_try_fmt().

- The validation of the pixel format size isn't needed, as they
  duplicate the validation of the compose rectangle. The pixel format
  size is directly derived from the size of the incoming stream, which
  is stored in the compose rectangle. If the compose rectangle is valid,
  the pixel format size will be valid too.

By removing the validation of the pixel format size and using
capture_find_format() to lookup the format, we can stop using
__capture_legacy_try_fmt() and simplify capture_validate_fmt().

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Extract format lookup from __capture_legacy_try_fmt

Extract the format lookup code from __capture_legacy_try_fmt() to a
separate function, it will be reused when validating the video node
format at stream start time.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Return -EPIPE from __capture_legacy_try_fmt()

The correct return code to report an invalid pipeline configuration is
-EPIPE. Return it instead of -EINVAL from __capture_legacy_try_fmt()
when the capture format doesn't match the media bus format of the
connected subdev.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Pass v4l2_pix_format to __capture_legacy_try_fmt()

The __capture_legacy_try_fmt() function only needs the v4l2_pix_format
embedded in the v4l2_format argument it receives. Pass it the
v4l2_pix_format directly, allowing on caller to allocate a smaller
structure on the stack.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Reviewed-by: Steve Longerbeam <slongerbeam@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Remove unneeded variable in __capture_legacy_try_fmt

The __capture_legacy_try_fmt() function doesn't need two
imx_media_pixfmt pointer variables. Remove one of them.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Reviewed-by: Steve Longerbeam <slongerbeam@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Add a mechanism to disable control inheritance

Add a parameter to the imx_media_capture_device_init() function to
select between the legacy and MC-centric API. When selecting the
MC-centric API, the video node doesn't inherit controls from subdevs
anymore. All callers are updated to use the legacy API for now,
preserving the existing behaviour.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Rename ioctl operations with legacy prefix

The i.MX media drivers implement a legacy video node API, where the
format of the video node is influenced by the active format of the
connected subdev (both for enumeration and for the get, set and try
format ioctls), and where controls exposed by the subdevs in the
pipeline are inherited by the video node.

At the same time, the drivers implement the media controller API and
expose subdev video nodes to userspace. Those two modes of operation are
incompatible and should not be exposed together. Furthermore, the legacy
API gets in the way of proper enumeration of pixel formats on the video
node, as it prevents compliance with the V4L2 specification.

As a first step towards fixing this, rename all V4L2 video node ioctl
handlers with a legacy prefix. This will allow implementing a new set of
ioctls in parallel and gradually switching drivers. Add a task to the
TODO file for the removal of the legacy API.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Rename querycap handler to capture_querycap

For consistency with all the other ioctl handlers, rename
vidioc_querycap() to capture_querycap().

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Reviewed-by: Steve Longerbeam <slongerbeam@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Move default format init to a separate function

To prepare for more complex default format initialization and keep
imx_media_capture_device_register() simple, extract the format
initialization code to a separate function.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Store v4l2_pix_format in imx_media_video_dev

The imx_media_video_dev structure stores the active format on the video
node in an instance of v4l2_format. This wastes memory when all we need
is the information contained in the smaller v4l2_pix_format
sub-structure. Replace v4l2_format by v4l2_pix_format to save memory.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Register the video device after completing init

When the video device is registered, it can get used by userspace
immediately. Its initialization must thus be fully complete. Ensure this
by registering the video device after all initialization steps.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Initialize video_device programmatically

Overwriting the whole video_device isn't future-proof as it would
overwrite any field initialized by video_device_alloc(). Furthermore,
the current implementation modifies the global template video_device as
if it were a per-instance structure, which is bad practice. To fix all
this, initialize the video device programmatically in
imx_media_capture_device_init().

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Move queue and ctrl handler init to init function

Move the initialization of the vb2 queue and the control handler to the
imx_media_capture_device_init() function. There's no need to delay them
until the registration time.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Remove capture_priv stop field

The stop field in the capture_priv structure is only set, never read.
Drop it.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Clean up capture_priv structure

Document all the fields of the capture_priv structure and group them by
purpose.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Handle errors from v4l2_fh_open()

If the call to v4l2_fh_open() fails, return the error code instead of
proceeding normally as if nothing happened.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Remove forward declaration of capture_qops

The forward declaration of capture_qops isn't needed, remove it.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Use device name to construct bus_info

The v4l2_capability bus_info field is meant to locate the device in the
system. Using a subdev name isn't a good fit, the device name is a much
better option. Use it.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: capture: Use dev_* instead of v4l2_* to log messages

The imx-media-capture helpers operate a video device node, but abuse the
v4l2_* subdev print functions to log messages. We have a struct device
pointer available, use it instead with the dev_* functions. This reduces
dependencies from imx-media-capture to the connected subdev.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: utils: Add ability to filter pixel formats by mbus code

Add a media bus code argument to the imx_media_enum_pixel_formats(). If
set to a non-zero value, the function will only consider pixel formats
that match the given media bus code.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: Set default sizes through macros in all drivers

All drivers use 640x480 as the default size, but they all hardcode those
values. Create two global macros named IMX_MEDIA_DEF_PIX_WIDTH and
IMX_MEDIA_DEF_PIX_HEIGHT to store the default size, and use them through
the code.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: Compile imx6-media-objs only for CONFIG_VIDEO_IMX_CSI

imx6-media-objs contains a set of objects that are specific to the i.MX6
IPU-based media subsystem. They're not needed for the i.MX7 CSI. Only
compile them if CONFIG_VIDEO_IMX_CSI is selected.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Reviewed-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: Drop manual dependency on VIDEO_IMX_MEDIA

The VIDEO_IMX_CSI and VIDEO_IMX7_CSI symbols are defined in a section
guarded by VIDEO_IMX_MEDIA. There's no need to duplicate that with a
"depends on" statement. Drop it.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: Move dependency on VIDEO_DEV to common Kconfig symbol

All the i.MX staging media drivers depend on VIDEO_DEV. Move the
dependency to the common VIDEO_IMX_MEDIA symbol to avoid repeating it.

While at it, sort the dependencies and selections alphabetically.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx: Drop dependency on I2C

The i.MX staging media drivers don't depend on I2C. Drop the dependency
from Kconfig.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: ngene: switch from 'pci_' to 'dma_' API

The wrappers in include/linux/pci-dma-compat.h should go away.

The patch has been generated with the coccinelle script below and has been
hand modified to replace GFP_ with a correct flag.
It has been compile tested.

When memory is allocated, GFP_KERNEL can be used because in all cases,
it is called from a probe function and no lock is taken in the between.

The call chain is:
  ngene_probe                       (probe function, used in ngene-cards.c)
    --> ngene_get_buffers
      --> AllocCommonBuffers                  (call dma_alloc_coherent)
        --> create_ring_buffer                (call dma_alloc_coherent)
        --> AllocateRingBuffers               (call dma_alloc_coherent)

@@
@@
-    PCI_DMA_BIDIRECTIONAL
+    DMA_BIDIRECTIONAL

@@
@@
-    PCI_DMA_TODEVICE
+    DMA_TO_DEVICE

@@
@@
-    PCI_DMA_FROMDEVICE
+    DMA_FROM_DEVICE

@@
@@
-    PCI_DMA_NONE
+    DMA_NONE

@@
expression e1, e2, e3;
@@
-    pci_alloc_consistent(e1, e2, e3)
+    dma_alloc_coherent(&e1->dev, e2, e3, GFP_)

@@
expression e1, e2, e3;
@@
-    pci_zalloc_consistent(e1, e2, e3)
+    dma_alloc_coherent(&e1->dev, e2, e3, GFP_)

@@
expression e1, e2, e3, e4;
@@
-    pci_free_consistent(e1, e2, e3, e4)
+    dma_free_coherent(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4;
@@
-    pci_map_single(e1, e2, e3, e4)
+    dma_map_single(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4;
@@
-    pci_unmap_single(e1, e2, e3, e4)
+    dma_unmap_single(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4, e5;
@@
-    pci_map_page(e1, e2, e3, e4, e5)
+    dma_map_page(&e1->dev, e2, e3, e4, e5)

@@
expression e1, e2, e3, e4;
@@
-    pci_unmap_page(e1, e2, e3, e4)
+    dma_unmap_page(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4;
@@
-    pci_map_sg(e1, e2, e3, e4)
+    dma_map_sg(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4;
@@
-    pci_unmap_sg(e1, e2, e3, e4)
+    dma_unmap_sg(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4;
@@
-    pci_dma_sync_single_for_cpu(e1, e2, e3, e4)
+    dma_sync_single_for_cpu(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4;
@@
-    pci_dma_sync_single_for_device(e1, e2, e3, e4)
+    dma_sync_single_for_device(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4;
@@
-    pci_dma_sync_sg_for_cpu(e1, e2, e3, e4)
+    dma_sync_sg_for_cpu(&e1->dev, e2, e3, e4)

@@
expression e1, e2, e3, e4;
@@
-    pci_dma_sync_sg_for_device(e1, e2, e3, e4)
+    dma_sync_sg_for_device(&e1->dev, e2, e3, e4)

@@
expression e1, e2;
@@
-    pci_dma_mapping_error(e1, e2)
+    dma_mapping_error(&e1->dev, e2)

@@
expression e1, e2;
@@
-    pci_set_dma_mask(e1, e2)
+    dma_set_mask(&e1->dev, e2)

@@
expression e1, e2;
@@
-    pci_set_consistent_dma_mask(e1, e2)
+    dma_set_coherent_mask(&e1->dev, e2)

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: drxj: remove redundant assignments to variable image_to_select

The variable image_to_select is being initialized with a value that
is never read and it is being updated later with a new value.  The
initialization is redundant and can be removed.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: i2c: adv7511: remove open coded version of SMBus block read

The open coded version differs from the one in the core in one way: the
buffer will be always copied back, even when the transfer failed. Be
more robust: use the block read from the I2C core and propagate a
potential errno further to the sanity checks.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: cx23885: Fix various Hauppauge device analog capture inputs

Reports indicated S-Video capture did not work on HVR1265_K4, so all
devices on hand were audited and corrected to become fully
functional.

ImpactVCB-e
- Remove extraneous composite inputs
- Fix S-Video inputs

HVR5525
- Add routing and config for composite capture
- Add routing and config for S-Video capture
- Add routing for audio on both composite/S-Video

HVR1265_K4
- Remove non-existent composite capture
- Add routing and config for S-Video capture

Signed-off-by: Brad Love <brad@nextdimension.cc>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: cx23885: add more quirks for reset DMA on some AMD IOMMU

The folowing AMD IOMMU are affected by the RiSC engine stall, requiring a
reset to maintain continual operation. After being added to the
broken_dev_id list the systems are functional long term.

0x1481 is the PCI ID for the IOMMU found on Starship/Matisse

0x1419 is the PCI ID for the IOMMU found on 15h (Models 10h-1fh) family

0x5a23 is the PCI ID for the IOMMU found on RD890S/RD990

Signed-off-by: Brad Love <brad@nextdimension.cc>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>