media: vicodec: always return a valid format.

Rather than returning width/height values of 0, just default to
a format. Formats in V4L2 are always supposed to be valid, there
is no concept of an invalid format.

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

media: vicodec: move v4l2_ctrl_request_complete after spin_unlock

v4l2_ctrl_request_complete can sleep, so can't be called while
a spinlock is held.

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

media: MAINTAINERS: Add meson video decoder

Add an entry for the meson video decoder for amlogic SoCs.

Signed-off-by: Maxime Jourdan <mjourdan@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: meson: add v4l2 m2m video decoder driver

Amlogic SoCs feature a powerful video decoder unit able to
decode many formats, with a performance of usually up to 4k60.

This is a driver for this IP that is based around the v4l2 m2m framework.

It features decoding for:
- MPEG 1
- MPEG 2

Supported SoCs are: GXBB (S905), GXL (S905X/W/D), GXM (S912)

There is also a hardware bitstream parser (ESPARSER) that is handled here.

Signed-off-by: Maxime Jourdan <mjourdan@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: dt-bindings: media: add Amlogic Video Decoder Bindings

Add documentation for the meson vdec dts node.

Signed-off-by: Maxime Jourdan <mjourdan@baylibre.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: ttpci: use rc-core for the IR receiver

The IR protocol can now only be set via the rc protocols sysfs file rather
than via module parameters or a custom procfs file. So, it is no longer
necessary to periodically check for protocol changes.

The IR_RCMM protocol does not decode the Philips RC-MM protocol (12, 24 or
32 bit variants) or any protocol rc-core can encode, so this is marked
RC_PROTO_UNKNOWN.

Tested on Technotrend/Hauppauge WinTV Nexus-S rev2.1, which comes
with a small black hauppauge remote.

Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: stv090x: Implement probe/remove for stv090x

Move common code into a new function.

This provides the needed functionality to use dvb_module_probe() instead
of dvb_attach()!

[mchehab+samsung@kernel.org: fix an out of order error return code]
Signed-off-by: Tobias Klausmann <tobias.johannes.klausmann@mni.thm.de>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: stv6110x: Implement probe/remove for stv6110x

Refactor out the common parts of stv6110x_probe() and stv6110x_attach()
into separate functions.

This provides the needed functionality to use dvb_module_probe() instead
of dvb_attach()!

Signed-off-by: Tobias Klausmann <tobias.johannes.klausmann@mni.thm.de>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: Revert "media: staging/imx: Allow driver to build if COMPILE_TEST is enabled"

Commit 020bc7354a6e ("media: staging/imx: Allow driver to build
if COMPILE_TEST is enabled") incorrectly assumed that the imx media
driver had no build time dependency with IMX_IPUV3_CORE.

Building on x86_64 without IMX_IPUV3_CORE causes lots of build errors
such as:

All these definitions come from the imx ipu3 core driver, so make
sure that imx media depends on IMX_IPUV3_CORE.

This reverts commit 020bc7354a6ebec980e0aedf5bedf57b42f93aca.

Reported-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Fabio Estevam <festevam@gmail.com>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: v4l2: Fix the _MPLANE format check in v4l_enum_fmt()

CAP_M2M_MPLANE means the device supports _MPLANE formats for both
capture and output. Adjust the check to avoid EINVAL errors on
such devices.

Reported-by: Maxime Jourdan <mjourdan@baylibre.com>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Tomasz Figa <tfiga@chromium.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: MAINTAINERS: update email address

Use hverkuil-cisco@xs4all.nl instead of hans.verkuil@cisco.com.

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

media: Documentation: update email address

Use hverkuil-cisco@xs4all.nl instead of hans.verkuil@cisco.com.

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

media: pvrusb2: fix null-ptr-deref in class_unregister()

The class_ptr will be NULL if pvr2_sysfs_class_create() fails
in pvr_init(), when call pvr2_sysfs_class_destroy(), it will
lead to null-ptr-deref, fix it.

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: platform: ao-cec-g12a: remove spin_lock_irqsave() locking in meson_ao_cec_g12a_read/write

Since locking is handled by regmap, the spin_lock_irqsave() in the
meson_ao_cec_g12a_read/write() regmap callbacks is not needed.

Fixes: b7778c46683c ("media: platform: meson: Add Amlogic Meson G12A AO CEC Controller driver")

Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: imx7-media-csi: Remove unneeded error message

In case of ioremap failure, the core code will take care of printing
the error message, so there is no need for having a local error
message in the driver.

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: imx7-media-csi: Use u32 for storing register reads

The CSI registers are 32-bit, so using u32 type is more suitable
for storing the values from register reads.

Switch from 'unsigned long' to 'u32' type.

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: imx7-media-csi: Use functions instead of macros

Currently there is a macro for reading and another macro for writing
to the CSI registers.

Functions can do parameter type checking, which leads to a safer code,
so switch from macro to function implementation.

Signed-off-by: Fabio Estevam <festevam@gmail.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+samsung@kernel.org>

media: aspeed: use different delays for triggering VE H/W reset

In case of watchdog timeout detected while doing mode detection,
it's better triggering video engine hardware reset immediately so
this commit fixes code for the case. Other than the case, it will
trigger video engine hardware reset after RESOLUTION_CHANGE_DELAY.

Signed-off-by: Jae Hyun Yoo <jae.hyun.yoo@linux.intel.com>
Reviewed-by: Eddie James <eajames@linux.ibm.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: aspeed: remove source buffer allocation before mode detection

Mode detection doesn't require source buffer allocation so this
commit removes that.

Signed-off-by: Jae Hyun Yoo <jae.hyun.yoo@linux.intel.com>
Reviewed-by: Eddie James <eajames@linux.ibm.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: aspeed: refine interrupt handling logic

There are cases that interrupt bits are cleared by a 500ms delayed
work which causes unnecessary irq calls. Also, the current
interrupt handler returns IRQ_HANDLED always but it should return
IRQ_NONE if there is any unhandled interrupt. So this commit
refines the interrupt handling logic to fix these issues.

Signed-off-by: Jae Hyun Yoo <jae.hyun.yoo@linux.intel.com>
Reviewed-by: Andrew Jeffery <andrew@aj.id.au>
Reviewed-by: Eddie James <eajames@linux.ibm.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: aspeed: remove checking of VE_INTERRUPT_CAPTURE_COMPLETE

VE_INTERRUPT_CAPTURE_COMPLETE and VE_INTERRUPT_COMP_COMPLETE are
not set at the same time but the current interrupt handling
mechanism of this driver doesn't clear the interrupt flag until
both two are set, and this behavior causes unnecessary interrupt
handler calls. In fact, this driver provides JPEG format only so
taking care of the VE_INTERRUPT_COMP_COMPLETE is enough for getting
compressed image frame so this commit gets rid of the
VE_INTERRUPT_CAPTURE_COMPLETE checking logic to simplify the logic.
Handling of VE_INTERRUPT_CAPTURE_COMPLETE could be added back later
when it's actually needed.

Signed-off-by: Jae Hyun Yoo <jae.hyun.yoo@linux.intel.com>
Reviewed-by: Eddie James <eajames@linux.ibm.com>
Reviewed-by: Andrew Jeffery <andrew@aj.id.au>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: aspeed: reduce noisy log printing outs

Currently, this driver prints out too much log messages when a
mode change happens, video turned off by screen saver and etc.
Actually, all cases are reported to user space properly. Also,
these are not critical errors but recoverable things, so this
commit changes the log level of some noisy printing outs.

Signed-off-by: Jae Hyun Yoo <jae.hyun.yoo@linux.intel.com>
Reviewed-by: Eddie James <eajames@linux.ibm.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: aspeed: remove IRQF_SHARED flag

Video Engine has a dedicated interrupt line so this driver doesn't
need to use IRQF_SHARED flag so remove it. Also, it'd be good for
following what Thomas recommended in the IRQF_ONESHOT support
patch like below:

"Note that for now IRQF_ONESHOT cannot be used with IRQF_SHARED to
avoid complex accounting mechanisms."

Signed-off-by: Jae Hyun Yoo <jae.hyun.yoo@linux.intel.com>
Reviewed-by: Eddie James <eajames@linux.ibm.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: aspeed: change irq to threaded irq

Differently from other Aspeed drivers, this driver calls clock
control APIs in interrupt context. Since ECLK is coupled with a
reset bit in clk-aspeed module, aspeed_clk_enable will make 10ms of
busy waiting delay for triggering the reset and it will eventually
disturb other drivers' interrupt handling. To fix this issue, this
commit changes this driver's irq to threaded irq so that the delay
can be happened in a thread context.

Signed-off-by: Jae Hyun Yoo <jae.hyun.yoo@linux.intel.com>
Reviewed-by: Eddie James <eajames@linux.ibm.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: aspeed: refine clock control logic

Currently, this driver calls clk_prepare and clk_unprepare from
interrupt context too but these should be called from sleepable
context only. To fix this issue, this commit splits out
clk_enable/disable and clk_prepare/unprepare, and it places
clk_prepare/unprepare calls into the module probe/remove function.

Signed-off-by: Jae Hyun Yoo <jae.hyun.yoo@linux.intel.com>
Reviewed-by: Eddie James <eajames@linux.ibm.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: aspeed: fix a kernel warning on clk control

Video engine clock control can be double disabled and eventually
it causes a kernel warning with stack dump printing out like below:

[  515.540498] ------------[ cut here ]------------
[  515.545174] WARNING: CPU: 0 PID: 1310 at drivers/clk/clk.c:684 clk_core_unprepare+0x13c/0x170
[  515.553806] vclk-gate already unprepared
[  515.557841] CPU: 0 PID: 1310 Comm: obmc-ikvm Tainted: G        W         5.0.6-df66fbc97853fbba90a0bfa44de32f3d5f7602b4 #1
[  515.568973] Hardware name: Generic DT based system
[  515.573777] Backtrace:
[  515.576272] [<80107cdc>] (dump_backtrace) from [<80107f10>] (show_stack+0x20/0x24)
[  515.583930]  r7:803a5614 r6:00000009 r5:00000000 r4:9d88fe1c
[  515.589712] [<80107ef0>] (show_stack) from [<80690184>] (dump_stack+0x20/0x28)
[  515.597053] [<80690164>] (dump_stack) from [<80116044>] (__warn.part.3+0xb4/0xdc)
[  515.604557] [<80115f90>] (__warn.part.3) from [<801160d8>] (warn_slowpath_fmt+0x6c/0x90)
[  515.612734]  r6:000002ac r5:8080befc r4:80a07008
[  515.617463] [<80116070>] (warn_slowpath_fmt) from [<803a5614>] (clk_core_unprepare+0x13c/0x170)
[  515.626167]  r3:8080cdf4 r2:8080bfc0
[  515.629834]  r7:98d682a8 r6:9d8a9200 r5:9e5151a0 r4:97abd620
[  515.635530] [<803a54d8>] (clk_core_unprepare) from [<803a76a4>] (clk_unprepare+0x34/0x3c)
[  515.643812]  r5:9e5151a0 r4:97abd620
[  515.647529] [<803a7670>] (clk_unprepare) from [<804f36ec>] (aspeed_video_off+0x38/0x50)
[  515.655539]  r5:9e5151a0 r4:9e504000
[  515.659242] [<804f36b4>] (aspeed_video_off) from [<804f4358>] (aspeed_video_release+0x90/0x114)
[  515.668036]  r5:9e5044b0 r4:9e504000
[  515.671643] [<804f42c8>] (aspeed_video_release) from [<804d302c>] (v4l2_release+0xd4/0xe8)
[  515.679999]  r7:98d682a8 r6:9d087810 r5:9d8a9200 r4:9e504318
[  515.685695] [<804d2f58>] (v4l2_release) from [<80236454>] (__fput+0x98/0x1c4)
[  515.692914]  r5:9e51b608 r4:9d8a9200
[  515.696597] [<802363bc>] (__fput) from [<802365e8>] (____fput+0x18/0x1c)
[  515.703315]  r9:80a0700c r8:801011e4 r7:00000000 r6:80a64b9c r5:9d8e35a0 r4:9d8e38dc
[  515.711167] [<802365d0>] (____fput) from [<80131ca4>] (task_work_run+0x7c/0xa0)
[  515.718596] [<80131c28>] (task_work_run) from [<80106884>] (do_work_pending+0x4a8/0x578)
[  515.726777]  r7:801011e4 r6:80a07008 r5:9d88ffb0 r4:ffffe000
[  515.732466] [<801063dc>] (do_work_pending) from [<8010106c>] (slow_work_pending+0xc/0x20)
[  515.740727] Exception stack(0x9d88ffb0 to 0x9d88fff8)
[  515.745840] ffa0:                                     00000000 76f18094 00000000 00000000
[  515.754122] ffc0: 00000007 00176778 7eda4c20 00000006 00000000 00000000 48e20fa4 00000000
[  515.762386] ffe0: 00000002 7eda4b08 00000000 48f91efc 80000010 00000007
[  515.769097]  r10:00000000 r9:9d88e000 r8:801011e4 r7:00000006 r6:7eda4c20 r5:00176778
[  515.777006]  r4:00000007
[  515.779558] ---[ end trace 12c04aadef8afbbb ]---
[  515.784176] ------------[ cut here ]------------
[  515.788817] WARNING: CPU: 0 PID: 1310 at drivers/clk/clk.c:825 clk_core_disable+0x18c/0x204
[  515.797161] eclk-gate already disabled
[  515.800916] CPU: 0 PID: 1310 Comm: obmc-ikvm Tainted: G        W         5.0.6-df66fbc97853fbba90a0bfa44de32f3d5f7602b4 #1
[  515.811945] Hardware name: Generic DT based system
[  515.816730] Backtrace:
[  515.819210] [<80107cdc>] (dump_backtrace) from [<80107f10>] (show_stack+0x20/0x24)
[  515.826782]  r7:803a5900 r6:00000009 r5:00000000 r4:9d88fe04
[  515.832454] [<80107ef0>] (show_stack) from [<80690184>] (dump_stack+0x20/0x28)
[  515.839687] [<80690164>] (dump_stack) from [<80116044>] (__warn.part.3+0xb4/0xdc)
[  515.847170] [<80115f90>] (__warn.part.3) from [<801160d8>] (warn_slowpath_fmt+0x6c/0x90)
[  515.855247]  r6:00000339 r5:8080befc r4:80a07008
[  515.859868] [<80116070>] (warn_slowpath_fmt) from [<803a5900>] (clk_core_disable+0x18c/0x204)
[  515.868385]  r3:8080cdd0 r2:8080c00c
[  515.871957]  r7:98d682a8 r6:9d8a9200 r5:97abd560 r4:97abd560
[  515.877615] [<803a5774>] (clk_core_disable) from [<803a59a0>] (clk_core_disable_lock+0x28/0x34)
[  515.886301]  r7:98d682a8 r6:9d8a9200 r5:97abd560 r4:a0000013
[  515.891960] [<803a5978>] (clk_core_disable_lock) from [<803a7714>] (clk_disable+0x2c/0x30)
[  515.900216]  r5:9e5151a0 r4:9e515f60
[  515.903816] [<803a76e8>] (clk_disable) from [<804f36f8>] (aspeed_video_off+0x44/0x50)
[  515.911656] [<804f36b4>] (aspeed_video_off) from [<804f4358>] (aspeed_video_release+0x90/0x114)
[  515.920341]  r5:9e5044b0 r4:9e504000
[  515.923921] [<804f42c8>] (aspeed_video_release) from [<804d302c>] (v4l2_release+0xd4/0xe8)
[  515.932184]  r7:98d682a8 r6:9d087810 r5:9d8a9200 r4:9e504318
[  515.937851] [<804d2f58>] (v4l2_release) from [<80236454>] (__fput+0x98/0x1c4)
[  515.944980]  r5:9e51b608 r4:9d8a9200
[  515.948559] [<802363bc>] (__fput) from [<802365e8>] (____fput+0x18/0x1c)
[  515.955257]  r9:80a0700c r8:801011e4 r7:00000000 r6:80a64b9c r5:9d8e35a0 r4:9d8e38dc
[  515.963008] [<802365d0>] (____fput) from [<80131ca4>] (task_work_run+0x7c/0xa0)
[  515.970333] [<80131c28>] (task_work_run) from [<80106884>] (do_work_pending+0x4a8/0x578)
[  515.978421]  r7:801011e4 r6:80a07008 r5:9d88ffb0 r4:ffffe000
[  515.984086] [<801063dc>] (do_work_pending) from [<8010106c>] (slow_work_pending+0xc/0x20)
[  515.992247] Exception stack(0x9d88ffb0 to 0x9d88fff8)
[  515.997296] ffa0:                                     00000000 76f18094 00000000 00000000
[  516.005473] ffc0: 00000007 00176778 7eda4c20 00000006 00000000 00000000 48e20fa4 00000000
[  516.013642] ffe0: 00000002 7eda4b08 00000000 48f91efc 80000010 00000007
[  516.020257]  r10:00000000 r9:9d88e000 r8:801011e4 r7:00000006 r6:7eda4c20 r5:00176778
[  516.028072]  r4:00000007
[  516.030606] ---[ end trace 12c04aadef8afbbc ]---

To prevent this issue, this commit adds clock status checking
logic into the Aspeed video engine driver.

Signed-off-by: Jae Hyun Yoo <jae.hyun.yoo@linux.intel.com>
Reviewed-by: Eddie James <eajames@linux.ibm.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: stm32-dcmi: fix irq = 0 case

Manage the irq = 0 case, where we shall return an error.

Fixes: b5b5a27bee58 ("media: stm32-dcmi: return appropriate error codes during probe")

Signed-off-by: Fabien Dessenne <fabien.dessenne@st.com>
Reported-by: Pavel Machek <pavel@ucw.cz>
Acked-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: cxusb: Revert "media: cxusb: add raw mode support for, Medion MD95700"

This patch shouldn't have been included in the pull request as it adds
a non-standard raw mode that is for debugging only. So revert
commit ead14a70754f8d7f5dbcb0553c7f11eb0fc4a6ac.

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

media: platform: Fix Warning of Unneeded Semicolon reported by coccicheck

fixed the warning in the files below

drivers/media/platform/pxa_camera.c:1391:2-3: Unneeded semicolon
drivers/media/platform/qcom/venus/vdec_ctrls.c:78:2-3: Unneeded semicolon
drivers/media/platform/sti/c8sectpfe/c8sectpfe-dvb.c:146:3-4: Unneeded semicolon

Signed-off-by: Shobhit Kukreti <shobhitkukreti@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: vivid: fix incorrect assignment operation when setting video mode

The assigment of FB_VMODE_NONINTERLACE to var->vmode should be a
bit-wise or of FB_VMODE_NONINTERLACE instead of an assignment,
otherwise the previous clearing of the FB_VMODE_MASK bits of
var->vmode makes no sense and is redundant.

Addresses-Coverity: ("Unused value")
Fixes: ad4e02d5081d ("[media] vivid: add a simple framebuffer device for overlay testing")

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+samsung@kernel.org>

media: mtk-vcodec: support single-buffer frames

MT8183 will use a multi-planar format backed by a single buffer. Adapt
the existing code to be able to handle such frames instead of assuming
each frame is backed by two buffers.

Co-developed-by: Alexandre Courbot <acourbot@chromium.org>
[acourbot: refactor, cleanup and split]

Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
Signed-off-by: Alexandre Courbot <acourbot@chromium.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
[hverkuil-cisco@xs4all.nl: fix checkpatch alignment warning]
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: mtk-vcodec: constify formats

Formats are read-only internal memory structures, so make them const.

Signed-off-by: Alexandre Courbot <acourbot@chromium.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: mtk-vcodec: replace GPLv2 with SPDX

Replace the GPLv2 boilerplate with the corresponding SPDX reference.

Signed-off-by: Alexandre Courbot <acourbot@chromium.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: imx7-mipi-csis: Remove extra blank line

Checkpatch reports an extra blank line, so remove such unneeded
line.

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: imx7-mipi-csis: Remove unneeded 'ret' initialization

There is no need for initializing the 'ret' variable as it will
be assigned at:

ret = mipi_csis_parse_dt(pdev, state);

Remove the unneeded initialization.

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: imx7-mipi-csis: Propagate the error if clock enabling fails

Currently the return value from clk_bulk_prepare_enable() is checked,
but it is not propagate it in the case of failure.

Fix it and also move the error message to the caller of
mipi_csis_clk_enable().

Signed-off-by: Fabio Estevam <festevam@gmail.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+samsung@kernel.org>

media: vimc: Remove or modify stream checks

Change the way subdevices check if the stream is running. Verify the stream
pointer instead of src_frame. This makes easier to get rid of the void* and
u8* that points to frames in the subdevices structs.

Remove checks that s_stream does on subdevices. They are redundant since
the Media Controller Framework doesn't allow two streaming on the same
media pipeline at the same time.

Signed-off-by: André Almeida <andrealmeid@collabora.com>
Acked-by: Helen Koike <helen.koike@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: media/doc: Allow sizeimage to be set by v4l clients

This changes v4l2_pix_format and v4l2_plane_pix_format sizeimage
field description to allow v4l clients to set bigger image size
in case of variable length compressed data.

Presently s5p-mfc and mtk-vcodec codec drivers use that. Lets
make it obvious in the documentation.

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

media: docs: fix minor typos

Fix minor typos in the DVB demux documentation.

Signed-off-by: Marc Gonzalez <marc.w.gonzalez@free.fr>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: dvb: tag deprecated DVB APIs as such

There are three headers at DVB that should not be used on
future projects: audio.h, osd.h and video.h.

While this is already clear at the docs, make clear also at
the headers that those files should not be used on future
drivers.

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

media: do not use C++ style comments in uapi headers

Linux kernel tolerates C++ style comments these days. Actually, the
SPDX License tags for .c files start with //.

On the other hand, uapi headers are written in more strict C, where
the C++ comment style is forbidden.

[mchehab+samsung@kernel.org: fix a checkpatch --strict warning]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: vicodec: use new v4l2_m2m_ioctl_try_en/decoder_cmd funcs

Use the new helper functions for the try_de/decoder_cmd ioctls.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reviewed-by: Tomasz Figa <tfiga@chromium.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: v4l2-mem2mem: add try_en/decoder_cmd ioctl helpers

Most if not all codecs will need to implement these ioctls and
it is expected to be the same for all codecs. So add this to
the core v4l2-mem2mem framework so that this code can easily be
reused.

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

media: staging/media: set device_caps in struct video_device

Instead of filling in the struct v4l2_capability device_caps
field, fill in the struct video_device device_caps field.

That way the V4L2 core knows what the capabilities of the
video device are.

But this only really works if all drivers use this, so convert
all staging/media drivers in this patch.

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

media: vc04_services/bcm2835-camera: set device_caps in struct video_device

Instead of filling in the struct v4l2_capability device_caps
field, fill in the struct video_device device_caps field.

That way the V4L2 core knows what the capabilities of the
video device are.

But this only really works if all drivers use this, so convert
this driver accordingly.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: usb/gadget/f_uvc: set device_caps in struct video_device

Instead of filling in the struct v4l2_capability device_caps
field, fill in the struct video_device device_caps field.

That way the V4L2 core knows what the capabilities of the
video device are.

But this only really works if all drivers use this, so convert
this UVC gadget driver.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: rtl2832_sdr: set device_caps in struct video_device

Instead of filling in the struct v4l2_capability device_caps
field, fill in the struct video_device device_caps field.

That way the V4L2 core knows what the capabilities of the
video device are.

But this only really works if all drivers use this, so convert
this SDR driver in this patch.

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

media: media/usb: set device_caps in struct video_device

Instead of filling in the struct v4l2_capability device_caps
field, fill in the struct video_device device_caps field.

That way the V4L2 core knows what the capabilities of the
video device are.

But this only really works if all drivers use this, so convert
all usb drivers in this patch.

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

media: media/radio: set device_caps in struct video_device

Instead of filling in the struct v4l2_capability device_caps
field, fill in the struct video_device device_caps field.

That way the V4L2 core knows what the capabilities of the
video device are.

But this only really works if all drivers use this, so convert
all radio drivers in this patch.

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

media: v4l2: Get rid of ->vidioc_enum_fmt_vid_{cap, out}_mplane

Support for multiplanar and singleplanar formats is mutually exclusive,
at least in practice. In our attempt to unify support for support for
mplane and !mplane in v4l, let's get rid of the
->vidioc_enum_fmt_{vid,out}_cap_mplane() hooks and call
->vidioc_enum_fmt_{vid,out}_cap() instead.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
[hverkuil-cisco@xs4all.nl: fix typos: pirv -> priv and prov -> priv]
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: v4l2: Make sure all drivers set _MPLANE caps in vdev->device_caps

This is needed if we want the core to be able to check _MPLANE support
without having to call the ->vidioc_querycap() hook.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: dt-bindings: Fix vendor-prefixes YAML

Commit 8df39e16877f ("media: dt-bindings: media: Add vendor prefix for
allegro") introduced a new devicetree binding vendors, however with an
improper syntax making the resulting YAML impossible to parse. Let's fix
this.

Fixes: 8df39e16877f ("media: dt-bindings: media: Add vendor prefix for allegro")

Signed-off-by: Maxime Ripard <maxime.ripard@bootlin.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: staging/imx: fix two NULL vs IS_ERR() bugs

The imx_media_pipeline_pad() function return NULL pointers on error, it
never returns error pointers.

Fixes: 3ef46bc97ca2 ("media: staging/imx: Improve pipeline searching")

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: platform: ao-cec-g12a: disable regmap fast_io for cec bus regmap

With fast_io enabled, spinlock_irq is used for read/write operations,
thus leading to :
BUG: sleeping function called from invalid context at [snip]/ao-cec-g12a.c:379
 in_atomic(): 1, irqs_disabled(): 128, pid: 1451, name: irq/14-ff800280
[snip]
Call trace:
 dump_backtrace+0x0/0x180
 show_stack+0x14/0x1c
 dump_stack+0xa8/0xe0
 ___might_sleep+0xf4/0x104
 __might_sleep+0x4c/0x80
 meson_ao_cec_g12a_read+0x7c/0x164
 regmap_read+0x16c/0x1b0
 meson_ao_cec_g12a_irq_thread+0xcc/0x200
 irq_thread_fn+0x2c/0x60
 irq_thread+0x14c/0x1fc
 kthread+0x11c/0x12c
 ret_from_fork+0x10/0x18

Simply remove fast_io to use mutexes instead.

Fixes: b7778c46683c ("media: platform: meson: Add Amlogic Meson G12A AO CEC Controller driver")

Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: cx23885: remove redundant assignment to err

The variable err is assigned with a value that is never read and
it is re-assigned a new value later on.  The assignment is redundant
and can be removed.

Addresses-Coverity: ("Unused value")

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+samsung@kernel.org>

media: dvb-usb/cxusb-analog.c: fix coccinelle warning, use ktime.h

This patch fixes a coccinelle warning and includes ktime.h instead of
timekeeping.h. The first includes the latter, but the latter doesn't
exist before 3.17, causing problems for our compat build. It's easier
to just use ktime.h instead.

coccinelle warnings: (new ones prefixed by >>)

>> drivers/media/usb/dvb-usb/cxusb-analog.c:1498:41-42: WARNING: Use ARRAY_SIZE

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reported-by: kbuild test robot <lkp@intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: wl128x: Fix some error handling in fm_v4l2_init_video_device()

X-Originating-IP: [10.175.113.25]
X-CFilter-Loop: Reflected
The fm_v4l2_init_video_device() forget to unregister v4l2/video device
in the error path, it could lead to UAF issue, eg,

  BUG: KASAN: use-after-free in atomic64_read include/asm-generic/atomic-instrumented.h:836 [inline]
  BUG: KASAN: use-after-free in atomic_long_read include/asm-generic/atomic-long.h:28 [inline]
  BUG: KASAN: use-after-free in __mutex_unlock_slowpath+0x92/0x690 kernel/locking/mutex.c:1206
  Read of size 8 at addr ffff8881e84a7c70 by task v4l_id/3659

  CPU: 1 PID: 3659 Comm: v4l_id Not tainted 5.1.0 #8
  Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014
  Call Trace:
   __dump_stack lib/dump_stack.c:77 [inline]
   dump_stack+0xa9/0x10e lib/dump_stack.c:113
   print_address_description+0x65/0x270 mm/kasan/report.c:187
   kasan_report+0x149/0x18d mm/kasan/report.c:317
   atomic64_read include/asm-generic/atomic-instrumented.h:836 [inline]
   atomic_long_read include/asm-generic/atomic-long.h:28 [inline]
   __mutex_unlock_slowpath+0x92/0x690 kernel/locking/mutex.c:1206
   fm_v4l2_fops_open+0xac/0x120 [fm_drv]
   v4l2_open+0x191/0x390 [videodev]
   chrdev_open+0x20d/0x570 fs/char_dev.c:417
   do_dentry_open+0x700/0xf30 fs/open.c:777
   do_last fs/namei.c:3416 [inline]
   path_openat+0x7c4/0x2a90 fs/namei.c:3532
   do_filp_open+0x1a5/0x2b0 fs/namei.c:3563
   do_sys_open+0x302/0x490 fs/open.c:1069
   do_syscall_64+0x9f/0x450 arch/x86/entry/common.c:290
   entry_SYSCALL_64_after_hwframe+0x49/0xbe
  RIP: 0033:0x7f8180c17c8e
  ...
  Allocated by task 3642:
   set_track mm/kasan/common.c:87 [inline]
   __kasan_kmalloc.constprop.3+0xa0/0xd0 mm/kasan/common.c:497
   fm_drv_init+0x13/0x1000 [fm_drv]
   do_one_initcall+0xbc/0x47d init/main.c:901
   do_init_module+0x1b5/0x547 kernel/module.c:3456
   load_module+0x6405/0x8c10 kernel/module.c:3804
   __do_sys_finit_module+0x162/0x190 kernel/module.c:3898
   do_syscall_64+0x9f/0x450 arch/x86/entry/common.c:290
   entry_SYSCALL_64_after_hwframe+0x49/0xbe

  Freed by task 3642:
   set_track mm/kasan/common.c:87 [inline]
   __kasan_slab_free+0x130/0x180 mm/kasan/common.c:459
   slab_free_hook mm/slub.c:1429 [inline]
   slab_free_freelist_hook mm/slub.c:1456 [inline]
   slab_free mm/slub.c:3003 [inline]
   kfree+0xe1/0x270 mm/slub.c:3958
   fm_drv_init+0x1e6/0x1000 [fm_drv]
   do_one_initcall+0xbc/0x47d init/main.c:901
   do_init_module+0x1b5/0x547 kernel/module.c:3456
   load_module+0x6405/0x8c10 kernel/module.c:3804
   __do_sys_finit_module+0x162/0x190 kernel/module.c:3898
   do_syscall_64+0x9f/0x450 arch/x86/entry/common.c:290
   entry_SYSCALL_64_after_hwframe+0x49/0xbe

Add relevant unregister functions to fix it.

Cc: Hans Verkuil <hans.verkuil@cisco.com>
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: Staging: media: Release the correct resource in an error handling path

'res' is reassigned several times in the function and if we 'goto
error_unmap', its value is not the returned value of 'request_mem_region()'
anymore.

Introduce a new 'struct resource *' variable (i.e. res2) to keep a pointer
to the right resource, if needed in the error handling path.

Fixes: 4b4eda001704 ("Staging: media: Unmap and release region obtained by ioremap_nocache")

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: omap3isp: constify platform_device_id

platform_device_id are not supposed to change at runtime. All functions
working with platform_device_id provided by <linux/platform_device.h>
work with const platform_device_id. So mark the non-const structs as
const.

Signed-off-by: Arvind Yadav <arvind.yadav.cs@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: staging: media: davinci_vpfe: use __func__ for function names

Checkpatch reported warnings for use of embedded function names.
Use __func__ instead of embedded function names.

[mchehab+samsung@kernel.org: rebased on the top of upstream]
Signed-off-by: Diwakar Sharma <sharmalxmail@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: omap3isp: Call video_unregister_device() unconditionally

video_unregister_device() can be called on a never or an already
unregistered device. Drop the redundant check.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: omap3isp: Don't rely on devm for memory resource management

devm functions are fine for managing resources that are directly related
to the device at hand and that have no other dependencies. However, a
process holding a file handle to a device created by a driver for a device
may result in the file handle left behind after the device is long gone.
This will result in accessing released (and potentially reallocated)
memory.

Instead, manage the memory resources in the driver. Releasing the
resources can be later on bound to e.g. by releasing a reference.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: staging: media: davinci_vpfe: Replace a bit shift

This patch replaces bit shifting on 1 with the BIT(x) macro.
This was done with coccinelle:
@@
constant c;
@@

-1 << c
+BIT(c)

[mchehab+samsung@kernel.org: rebase on the top of upstream]
Signed-off-by: Arushi Singhal <arushisinghal19971997@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: staging: media: davinci_vpfe: - Fix for memory leak if decoder initialization fails.

Fix to avoid possible memory leak if the decoder initialization
got failed.Free the allocated memory for file handle object
before return in case decoder initialization fails.

Signed-off-by: Shailendra Verma <shailendra.v@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: videobuf-dma-contig: Use size of buffer in mmap not vma size

The size of the vma can be larger than the size of the backing buffer.
Use the buffer size over vma size to prevent exposing extra memory
to userspace.

Signed-off-by: Andrew F. Davis <afd@ti.com>
Reviewed-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: cx231xx-dvb: fix memory leak in dvb_fini()

In dvb_init(), dev->dvb is allocated by kzalloc.
Therefore, it must be freed being set to NULL.

Signed-off-by: Young Xiao <92siuyang@gmail.com>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: em28xx: use usb_to_input_id() rather than handrolling it

This also populates the version member correctly.

Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: em28xx: use common code for decoding nec scancodes

Tested on WinTV-HVR-930C.

Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: em28xx: give RC device proper name

Give the RC device the name of the board rather than "1-2:1.0 IR".

Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: sun6i: Support A83T variant

The A83T SoC has a camera sensor interface (known as CSI in Allwinner
lingo), which is similar to the one found on the A64 and H3. The only
difference seems to be that support of MIPI CSI through a connected
MIPI CSI-2 bridge.

Add support for this variant.

Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Acked-by: Maxime Ripard <maxime.ripard@bootlin.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: dt-bindings: media: sun6i-csi: Add compatible string for A83T variant

The A83T SoC has a camera sensor interface (known as CSI in Allwinner
lingo), which is similar to the one found on the A64 and H3. The only
difference seems to be that support of MIPI CSI through a connected
MIPI CSI-2 bridge.

Add a compatible string for this variant.

Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Acked-by: Maxime Ripard <maxime.ripard@bootlin.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: ov8856: modify register to fix test pattern

modify registers to fix bayer order in test pattern mode

Signed-off-by: Shawn Tu <shawnx.tu@intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: staging: imgu: Drop support for other page sizes

The hardware only supports 4ki pages; drop support for other sizes.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Tested-by: Rajmohan Mani <rajmohan.mani@intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: staging: imgu: Address compiler / checker warnings in MMU code

Address C compiler, sparse and smatch warnings and little style issues in
the IMGU MMU code.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Tested-by: Rajmohan Mani <rajmohan.mani@intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: staging: imgu: Remove redundant checks

Remove redundant checks for less than zero on unsigned variables.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Tested-by: Rajmohan Mani <rajmohan.mani@intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: staging: imgu: Address a compiler warning on alignment

Address a compiler warnings on alignment of struct ipu3_uapi_awb_fr_config_s
by adding __attribute__((aligned(32))) to a struct member of that type as
well.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Tested-by: Rajmohan Mani <rajmohan.mani@intel.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: rockchip/vpu: Add support for MPEG-2 decoding on RK3288

Add necessary bits to support MPEG2 decoding on RK3288.

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: rockchip/vpu: Add MPEG2 decoding support to RK3399

Add the necessary bits to support MPEG2 decoding on RK3399.

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: rockchip/vpu: Add infra to support MPEG-2 decoding

Only adds structs and helpers to allow supporting MPEG-2 decoding on
rockchip SoCs. Support for RK3399 and RK3288 will be added in separate
commits

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: rockchip/vpu: Add support for non-standard controls

Rework the way controls are registered by the driver,
so it can support non-standard controls, such as those
used by stateless codecs.

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

media: rockchip/vpu: Add decoder boilerplate

This commit adds the needed boilerplate code to support the VPU
in decoding operation. Two v4l2 interfaces are exposed, one for
encoding and one for decoding, but a single m2m device is shared
by them, so jobs are properly serialized.

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

media: rockchip/vpu: Prepare things to support decoders

The code in rockchip_vpu_v4l2 was hardcoded for encoder support.
Modify it more generic to support the decoder case so that we can
re-use the same vb2/v4l2 ops for both devices.

Co-developed-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: rockchip/vpu: Provide a helper to reset both src and dst formats

When initializing a context, the core wants to reset both src and dst
formats. Right now the order doesn't matter, but if we want to have a
valid default width/height on the non-coded/raw format side (src in
case of encoders, dst in case of decoders), we need to reset those
formats in the right order: first the coded-format side, then the other,
such that width and height on the raw format side can be taken from
the coded format.

Let's provide a helper that will reset both formats and make sure this
is done in the right order.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: rockchip/vpu: Move encoder logic to a common place

The V4L2/VB2 implementation for the encoder and decoder logic are very
similar, so let's rename rockchip_vpu_enc.c file into
rockchip_vpu_v4l2.c and remove the _enc_ part in objects/functions
exposed in rockchip_vpu_v4l2.h. We also rename the enc_queue_init()
function (in rockchip_vpu_drv.c) queue_init() since it will be used
to initialize both type of queues.

The implementation itself will be patched to support the decoding case
when decoder support is added.

Suggested-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: rockchip/vpu: Rename rockchip_vpu_common.h into rockchip_vpu_v4l2.h

We're about to add prototypes for the vb2/v4l2 helpers shared by the
encoder/decoder logic in this file, so let's pick a name that reflects
that (rockchip_vpu_common.h was a bit to generic).

Suggested-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: rockchip/vpu: Support the Request API

Introduce support for the Request API. Although the JPEG encoder
does not mandate using the Request API, it's perfectly possible to
use it, if the application wants to.

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

media: rockchip/vpu: Open-code media controller register

In preparation to support decoders, using a single memory-to-memory
device, we need to roll our own media controller entities registration.

To do that, we define a rockchip_vpu_func object that embeds the
video_device object plus all the elements that are needed to attach this
vdev to the media device.

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

media: rockchip/vpu: Use v4l2_apply_frmsize_constraints() where appropriate

Use the v4l2_apply_frmsize_constraints() helper instead of open-coding
it.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: v4l2-common: Add an helper to apply frmsize constraints

The rockchip VPU driver is open-coding this logic which seems pretty
generic. Let's provide an helper to apply the min/max and alignment
constraints on width/height.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: v4l2-common: Fix v4l2_fill_pixfmt[_mp]() prototypes

Width/height and 4CC formats are expressed using u32 types everywhere,
let's fix the v4l2_fill_pixfmt[_mp]() prototypes to do the same.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: cec-adap: fix regression in ping sanity check

Commit b6c96e156825 inadvertently also dropped the 'msg->len > 1' test
from the preceding sanity check. This caused compliance test failures.

Fixes: b6c96e156825 ("media: cec: allow any initiator for Ping and Image/Text View On")

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

media: saa7164: fix remove_proc_entry warning

if saa7164_proc_create() fails, saa7164_fini() will trigger a warning,

name 'saa7164'
WARNING: CPU: 1 PID: 6311 at fs/proc/generic.c:672 remove_proc_entry+0x1e8/0x3a0
  ? remove_proc_entry+0x1e8/0x3a0
  ? try_stop_module+0x7b/0x240
  ? proc_readdir+0x70/0x70
  ? rcu_read_lock_sched_held+0xd7/0x100
  saa7164_fini+0x13/0x1f [saa7164]
  __x64_sys_delete_module+0x30c/0x480
  ? __ia32_sys_delete_module+0x480/0x480
  ? __x64_sys_clock_gettime+0x11e/0x1c0
  ? __x64_sys_timer_create+0x1a0/0x1a0
  ? trace_hardirqs_off_caller+0x40/0x180
  ? do_syscall_64+0x18/0x450
  do_syscall_64+0x9f/0x450
  entry_SYSCALL_64_after_hwframe+0x49/0xbe

Fix it by checking the return of proc_create_single() before
calling remove_proc_entry().

Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
[hverkuil-cisco@xs4all.nl: use 0444 instead of S_IRUGO]
[hverkuil-cisco@xs4all.nl: use pr_info instead of KERN_INFO]
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: mc-device.c: don't memset __user pointer contents

You can't memset the contents of a __user pointer. Instead, call copy_to_user to
copy links.reserved (which is zeroed) to the user memory.

This fixes this sparse warning:

SPARSE:drivers/media/mc/mc-device.c drivers/media/mc/mc-device.c:521:16:  warning: incorrect type in argument 1 (different address spaces)

Fixes: f49308878d720 ("media: media_device_enum_links32: clean a reserved field")

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reviewed-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: media-ioc-enum-links.rst: fix incorrect reserved field documentation

The reserved field array for struct media_link_desc has length 2, not 4.

And the reserved field array of struct media_links_enum was never documented
at all.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: mtk-vcodec: fix leaked of_node references

The call to of_find_device_by_node returns a node pointer with refcount
incremented thus it must be explicitly decremented after the last
usage.

Detected by coccinelle with the following warnings:
drivers/media/platform/mtk-vcodec/mtk_vcodec_dec_pm.c:60:3-9: ERROR: missing of_node_put; acquired a node pointer with refcount incremented on line 38, but without a corresponding object release within this function.
drivers/media/platform/mtk-vcodec/mtk_vcodec_dec_pm.c:63:2-8: ERROR: missing of_node_put; acquired a node pointer with refcount incremented on line 38, but without a corresponding object release within this function.
drivers/media/platform/mtk-vcodec/mtk_vcodec_dec_pm.c:72:3-9: ERROR: missing of_node_put; acquired a node pointer with refcount incremented on line 38, but without a corresponding object release within this function.

Signed-off-by: Wen Yang <wen.yang99@zte.com.cn>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: mtk-vpu: fix leaked of_node references

The call to of_parse_phandle returns a node pointer with refcount
incremented thus it must be explicitly decremented after the last
usage.

Detected by coccinelle with the following warnings:
drivers/media/platform/mtk-vpu/mtk_vpu.c:477:1-7: ERROR: missing of_node_put; acquired a node pointer with refcount incremented on line 464, but without a corresponding object release within this function.

Signed-off-by: Wen Yang <wen.yang99@zte.com.cn>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

media: cx25840: Address several coding style issues

As we did a major change on this file, let's take the moment
to cleanup several coding style issues on it.

This patch was partially done with the help of two tools:

./scripts/checkpatch.pl --fix-inplace --strict
astyle --indent=tab=8 --style=linux

But manually adjusted in order to fit our style.

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

media: cxusb: fix several coding style issues

As this driver had a major change, let's take the opportunity
and do some coding style cleanup, in order to make it compliant
with Kernel's style.

This patch was partially done with the help of two tools:

./scripts/checkpatch.pl --fix-inplace --strict
astyle --indent=tab=8 --style=linux

But manually adjusted in order to fit our style.

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

media: cxusb-analog: Fix some coding style issues

This is a new file, so the best moment to make it to follow
Kernel coding style is now.

This patch was partially generated with:

./scripts/checkpatch.pl --fix-inplace --strict -f drivers/media/usb/dvb-usb/cxusb-analog.c

And manually checked and adjusted to avoid any warnings.

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

media: staging: allegro: cleanup two warnings

Sparse complains about two issues when building with i386
and COMPILE_TEST:

drivers/staging/media/allegro-dvt/allegro-core.c:1849:36: warning: constant 0xffffffff00000000UL is so big it is unsigned long long
drivers/staging/media/allegro-dvt/allegro-core.c:865:24: error: incompatible types in comparison expression (different type sizes)

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