platform/kernel/linux-starfive.git
2 years agomedia: Hantro: HEVC: Allows 10-bit bitstream
Benjamin Gaignard [Mon, 29 Aug 2022 16:21:57 +0000 (18:21 +0200)]
media: Hantro: HEVC: Allows 10-bit bitstream

Stop limiting HEVC support to 8-bits bitstreams also
accept 10-bits bitstreams.

Signed-off-by: Benjamin Gaignard <benjamin.gaignard@collabora.com>
Reviewed-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: hantro: postproc: Configure output regs to support 10bit
Benjamin Gaignard [Mon, 29 Aug 2022 16:21:56 +0000 (18:21 +0200)]
media: hantro: postproc: Configure output regs to support 10bit

Move output format setting in postproc and make sure that
8/10bit configuration is correctly set.

Signed-off-by: Benjamin Gaignard <benjamin.gaignard@collabora.com>
Reviewed-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: hantro: HEVC: Fix chroma offset computation
Benjamin Gaignard [Mon, 29 Aug 2022 16:21:55 +0000 (18:21 +0200)]
media: hantro: HEVC: Fix chroma offset computation

The chroma offset depends of the bitstream depth.
Make sure that ctx->bit_depth is used to compute it.

Signed-off-by: Benjamin Gaignard <benjamin.gaignard@collabora.com>
Reviewed-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: hantro: HEVC: Fix auxilary buffer size calculation
Benjamin Gaignard [Mon, 29 Aug 2022 16:21:54 +0000 (18:21 +0200)]
media: hantro: HEVC: Fix auxilary buffer size calculation

SAO and FILTER buffers size depend of the bit depth.
Make sure we have enough space for 10bit bitstreams.

Signed-off-by: Benjamin Gaignard <benjamin.gaignard@collabora.com>
Reviewed-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: hantro: Store HEVC bit depth in context
Benjamin Gaignard [Mon, 29 Aug 2022 16:21:53 +0000 (18:21 +0200)]
media: hantro: Store HEVC bit depth in context

Store HEVC bit depth in context.
Bit depth is equal to hevc sps bit_depth_luma_minus8 + 8.

Signed-off-by: Benjamin Gaignard <benjamin.gaignard@collabora.com>
Reviewed-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: meson: vdec: fix possible refcount leak in vdec_probe()
Hangyu Hua [Tue, 6 Sep 2022 07:46:30 +0000 (09:46 +0200)]
media: meson: vdec: fix possible refcount leak in vdec_probe()

v4l2_device_unregister need to be called to put the refcount got by
v4l2_device_register when vdec_probe fails or vdec_remove is called.

Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: platform: cros-ec: Add Kuldax to the match table
Rory Liu [Tue, 6 Sep 2022 03:30:16 +0000 (05:30 +0200)]
media: platform: cros-ec: Add Kuldax to the match table

The Google Kuldax device uses the same approach as the Google Brask
which enables the HDMI CEC via the cros-ec-cec driver.

Signed-off-by: Rory Liu <hellojacky0226@hotmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: MAINTAINERS: rectify entry in SAA7146 VIDEO4LINUX-2 DRIVER
Lukas Bulwahn [Mon, 5 Sep 2022 10:29:39 +0000 (12:29 +0200)]
media: MAINTAINERS: rectify entry in SAA7146 VIDEO4LINUX-2 DRIVER

Commit e33fdb5a0249 ("media: saa7146: deprecate hexium_gemini/orion, mxb
and ttpci") moves some media drivers to the staging subsystem, and
unfortunately only partially adjusts the entry in MAINTAINERS.

Hence, ./scripts/get_maintainer.pl --self-test=patterns complains about a
broken reference.

As the files matching include/media/drv-intf/saa7146* are moved to
drivers/staging/media/deprecated/saa7146/common, this directory is already
covered by the existing file entry
drivers/staging/media/deprecated/saa7146/.

Repair this file reference in SAA7146 VIDEO4LINUX-2 DRIVER.

Signed-off-by: Lukas Bulwahn <lukas.bulwahn@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: MAINTAINERS: adjust entry to zoran driver movement
Lukas Bulwahn [Mon, 5 Sep 2022 10:14:16 +0000 (12:14 +0200)]
media: MAINTAINERS: adjust entry to zoran driver movement

Commit 2a0c28063de2 ("media: zoran: move to mainline") moves the zoran
driver from the staging to the media subsystem, but does not adjust the
entry in MAINTAINERS.

Hence, ./scripts/get_maintainer.pl --self-test=patterns complains about a
broken reference.

Repair this file reference in ZR36067 VIDEO FOR LINUX DRIVER.

Signed-off-by: Lukas Bulwahn <lukas.bulwahn@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: tuners: Remove the unneeded result variable
ye xingchen [Mon, 5 Sep 2022 09:02:03 +0000 (11:02 +0200)]
media: tuners: Remove the unneeded result variable

Return the value xc_send_i2c_data() directly instead of storing it in
another redundant variable.

Signed-off-by: ye xingchen <ye.xingchen@zte.com.cn>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: platform: mtk-mdp3: fix PM reference leak in mdp_comp_clock_on()
Sun Ke [Fri, 2 Sep 2022 08:58:19 +0000 (10:58 +0200)]
media: platform: mtk-mdp3: fix PM reference leak in mdp_comp_clock_on()

mdp_comp_clock_on will increase runtime PM usage counter,
and mdp_comp_clock_off will decrease the runtime PM usage counter.
so, if mdp_comp_clock_on failed after increment runtime PM usage
counter, it should decrease it before return a error code.

pm_runtime_get_sync will increment pm usage counter even it failed.
Forgetting to putting operation will result in reference leak here.
Fix it by replacing it with pm_runtime_resume_and_get to keep usage
counter balanced.

And if failed to enable clk, add pm_runtime_put() to decrease the
runtime PM usage counter.

Fixes: 61890ccaefaf ("media: platform: mtk-mdp3: add MediaTek MDP3 driver")
Signed-off-by: Sun Ke <sunke32@huawei.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: platform: mtk-mdp3: fix error code in mdp_vpu_dev_init()
Dan Carpenter [Thu, 1 Sep 2022 14:47:48 +0000 (16:47 +0200)]
media: platform: mtk-mdp3: fix error code in mdp_vpu_dev_init()

Return a negative error code if mdp_vpu_shared_mem_alloc() fails.

Fixes: 61890ccaefaf ("media: platform: mtk-mdp3: add MediaTek MDP3 driver")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Reviewed-by: Matthias Brugger <matthias.bgg@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: radio-si476x: Remove the unneeded result variable
ye xingchen [Thu, 1 Sep 2022 07:45:07 +0000 (09:45 +0200)]
media: radio-si476x: Remove the unneeded result variable

Return the value v4l2_fh_release() directly instead of storing it in
another redundant variable.

Signed-off-by: ye xingchen <ye.xingchen@zte.com.cn>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: docs: libv4l-introduction.rst: Fix function signature and link
Daniel Lundberg Pedersen [Wed, 31 Aug 2022 14:54:59 +0000 (16:54 +0200)]
media: docs: libv4l-introduction.rst: Fix function signature and link

v4l2_mmap returns a void*, also link to mmap instead of munmap

Signed-off-by: Daniel Lundberg Pedersen <dlp@qtec.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: platform: mtk-mdp3: add pointer checks and use devm_kfree
Moudy Ho [Wed, 31 Aug 2022 08:56:04 +0000 (10:56 +0200)]
media: platform: mtk-mdp3: add pointer checks and use devm_kfree

Fix two errors reported by smatch:
    drivers/media/platform/mediatek/mdp3/mtk-mdp3-core.c:292
        mdp_probe() error: we previously assumed 'mdp' could be null
    drivers/media/platform/mediatek/mdp3/mtk-mdp3-cmdq.c:460
        mdp_cmdq_send() error: we previously assumed 'cmd' could be null

Also, avoid warnings reported by smatch:
    drivers/media/platform/mediatek/mdp3/mtk-mdp3-comp.c:872
        mdp_comp_create() warn: passing devm_ allocated variable to kfree. 'comp'

[hverkuil: fix devm_kfree call]

Signed-off-by: Moudy Ho <moudy.ho@mediatek.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: si4713: Use the right include
Linus Walleij [Tue, 30 Aug 2022 21:28:06 +0000 (23:28 +0200)]
media: si4713: Use the right include

The driver includes the legacy <linux/gpio.h> header but uses
<linux/gpio/consumer.h>.

Cc: Dinesh Ram <Dinesh.Ram@cern.ch>
Cc: Eduardo Valentin <edubezval@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: i2c: isl7998x: Use right include
Linus Walleij [Tue, 30 Aug 2022 21:11:10 +0000 (23:11 +0200)]
media: i2c: isl7998x: Use right include

This driver is using GPIO descriptors but uses the legacy
include header <linux/gpio.h>. Fix it by including the
intended <linux/gpio/consumer.h>.

Cc: Marek Vasut <marex@denx.de>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: cec: add support for Absolute Volume Control
Hans Verkuil [Tue, 30 Aug 2022 10:37:24 +0000 (12:37 +0200)]
media: cec: add support for Absolute Volume Control

Add support for this new CEC message. This was added in HDMI 2.1a.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: dvb-frontends/drxk: initialize err to 0
Hans Verkuil [Tue, 30 Aug 2022 05:59:24 +0000 (07:59 +0200)]
media: dvb-frontends/drxk: initialize err to 0

Fix a compiler warning:

drivers/media/dvb-frontends/drxk_hard.c: In function 'drxk_read_ucblocks':
drivers/media/dvb-frontends/drxk_hard.c:6673:21: warning: 'err' may be used uninitialized [-Wmaybe-uninitialized]
 6673 |         *ucblocks = (u32) err;
      |                     ^~~~~~~~~
drivers/media/dvb-frontends/drxk_hard.c:6663:13: note: 'err' was declared here
 6663 |         u16 err;
      |             ^~~

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: v4l2-ctrls: drop 'elems' argument from control type ops.
Hans Verkuil [Mon, 29 Aug 2022 14:05:14 +0000 (16:05 +0200)]
media: v4l2-ctrls: drop 'elems' argument from control type ops.

The type ops for equal, init and validate have an elems argument,
but this can be taken from struct v4l2_ctrl: ctrl->elems for
equal and init, and ctrl->new_elems for validate (since you are
validating a new control value).

So drop this argument and update all callers.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: amphion: release m2m ctx when releasing vpu instance
Ming Qian [Mon, 29 Aug 2022 05:33:16 +0000 (07:33 +0200)]
media: amphion: release m2m ctx when releasing vpu instance

release m2m ctx in the callback function that
release the vpu instance, then there is no need
to add lock around releasing m2m ctx.

Fixes: 3cd084519c6f ("media: amphion: add vpu v4l2 m2m support")
Signed-off-by: Ming Qian <ming.qian@nxp.com>
Reviewed-by: Tommaso Merciai <tommaso.merciai@amarulasolutions.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: coda: jpeg: drop coda9_jpeg_dec_huff_setup() return value
Philipp Zabel [Thu, 25 Aug 2022 10:41:49 +0000 (12:41 +0200)]
media: coda: jpeg: drop coda9_jpeg_dec_huff_setup() return value

coda9_jpeg_dec_huff_setup() never returns anything but 0. Drop
return value and superfluous error handling at the call site.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: pci/cx18: fix repeated words in comments
Jilin Yuan [Wed, 24 Aug 2022 13:27:42 +0000 (15:27 +0200)]
media: pci/cx18: fix repeated words in comments

Delete the redundant word 'is'.

Signed-off-by: Jilin Yuan <yuanjilin@cdjrlc.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: usb/dvb-usb-v2: fix repeated words in comments
Jilin Yuan [Wed, 24 Aug 2022 13:13:50 +0000 (15:13 +0200)]
media: usb/dvb-usb-v2: fix repeated words in comments

Delete the redundant word 'my'.

Signed-off-by: Jilin Yuan <yuanjilin@cdjrlc.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: usb/msi2500: fix repeated words in comments
Jilin Yuan [Wed, 24 Aug 2022 13:10:16 +0000 (15:10 +0200)]
media: usb/msi2500: fix repeated words in comments

Delete the redundant word 'for'.

Signed-off-by: Jilin Yuan <yuanjilin@cdjrlc.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: cros-ec-cec: limit msg.len to CEC_MAX_MSG_SIZE
Hans Verkuil [Wed, 24 Aug 2022 07:06:19 +0000 (09:06 +0200)]
media: cros-ec-cec: limit msg.len to CEC_MAX_MSG_SIZE

I expect that the hardware will have limited this to 16, but just in
case it hasn't, check for this corner case.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: s5p_cec: limit msg.len to CEC_MAX_MSG_SIZE
Hans Verkuil [Wed, 24 Aug 2022 07:02:42 +0000 (09:02 +0200)]
media: s5p_cec: limit msg.len to CEC_MAX_MSG_SIZE

I expect that the hardware will have limited this to 16, but just in
case it hasn't, check for this corner case.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rkisp1: Zero v4l2_subdev_format fields in when validating links
Laurent Pinchart [Tue, 23 Aug 2022 15:11:36 +0000 (17:11 +0200)]
media: rkisp1: Zero v4l2_subdev_format fields in when validating links

The local sd_fmt variable in rkisp1_capture_link_validate() has
uninitialized fields, which causes random failures when calling the
subdev .get_fmt() operation. Fix it by initializing the variable when
declaring it, which zeros all other fields.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rkisp1: Configure LSC after enabling the ISP
Laurent Pinchart [Tue, 16 Aug 2022 23:53:57 +0000 (01:53 +0200)]
media: rkisp1: Configure LSC after enabling the ISP

The ISP8000Nano v18.02 (found in the i.MX8MP) requires the ISP to be
enabled (as indicated by the ISP_CTRL.ISP_ENABLE bit) to configure the
lens shading table in internal RAM. The driver currently configures all
ISP initial parameters before enabling the ISP, which causes the LSC RAM
to not be initialized properly.

To fix this, split the rkisp1_params_configure() function into a
rkisp1_params_pre_configure() and a rkisp1_params_post_configure(). The
former configures all ISP parameters but LSC, while the latter
configures LSC. To implement this, the rkisp1_params_apply_params_cfg()
function is deconstructed, with two small helpers created to deal with
the parameters buffers, which are then used in rkisp1_params_isr(),
rkisp1_params_pre_configure() and rkisp1_params_post_configure().

While this initialization ordering is only needed for the ISP8000Nano
v18.02, it doesn't affect other ISP versions negatively, and can thus be
followed unconditionally.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rkisp1: Use correct macro for gradient registers
Laurent Pinchart [Tue, 16 Aug 2022 23:53:57 +0000 (01:53 +0200)]
media: rkisp1: Use correct macro for gradient registers

The rkisp1_lsc_config() function incorrectly uses the
RKISP1_CIF_ISP_LSC_SECT_SIZE() macro for the gradient registers. Replace
it with the correct macro, and rename it from
RKISP1_CIF_ISP_LSC_GRAD_SIZE() to RKISP1_CIF_ISP_LSC_SECT_GRAD() as the
corresponding registers store the gradients for each sector, not a size.
This doesn't cause any functional change as the two macros are defined
identically (the size and gradient registers store fields in the same
number of bits at the same positions).

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rkisp1: Simplify LSC x/y size and grad register macros
Laurent Pinchart [Tue, 16 Aug 2022 23:53:57 +0000 (01:53 +0200)]
media: rkisp1: Simplify LSC x/y size and grad register macros

The LSC module x/y size and grad configuration is stored in a set of 4
indexed registers each. The rkisp1-regs.h header defines all those
registers, but only the first one in each set is used, with manual
calculation of addresses of subsequent registers. Simplifies this by
merging all 4 register macros into one that takes the index as a
parameter. No functional change intended.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rkisp1: Store LSC register values in u32 variables
Laurent Pinchart [Tue, 16 Aug 2022 23:53:57 +0000 (01:53 +0200)]
media: rkisp1: Store LSC register values in u32 variables

Use the u32 type instead of unsigned int to store register values in the
LSC configuration code, to make the variables' size more explicit. No
functional change intended.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rkisp1: Clean up LSC configuration code
Laurent Pinchart [Tue, 16 Aug 2022 23:53:57 +0000 (01:53 +0200)]
media: rkisp1: Clean up LSC configuration code

Clean up the LSC configuration code to improve its readability by
shortening lines, using extra local variables and renaming long
variables. No functional change intended.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rkisp1: Allow setting color space on resizer sink pad
Laurent Pinchart [Fri, 12 Aug 2022 22:44:14 +0000 (00:44 +0200)]
media: rkisp1: Allow setting color space on resizer sink pad

The resizer doesn't deal with color spaces, so it can accept any color
space on its input, and propagates it unchanged to its output. When
operating with a Bayer input format (in pass-through mode) further
restrict the YCbCr encoding and quantization to Rec 601 and full range
respectively, as for raw data the former ought to be ignored and the
latter is always full range.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rkisp1: Initialize color space on resizer sink and source pads
Laurent Pinchart [Fri, 12 Aug 2022 22:44:14 +0000 (00:44 +0200)]
media: rkisp1: Initialize color space on resizer sink and source pads

Initialize the four color space fields on the sink and source video pads
of the resizer in the .init_cfg() operation. The resizer can't perform
any color space conversion, so set the sink and source color spaces to
the same defaults, which match the ISP source video pad default.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rkisp1: Configure CSM based on YCbCr encoding
Laurent Pinchart [Fri, 12 Aug 2022 22:44:14 +0000 (00:44 +0200)]
media: rkisp1: Configure CSM based on YCbCr encoding

The driver currently only implements the Rec. 601 YCbCr encoding, extend
it with support for the other encodings defined by V4L2 (Rec. 709, Rec.
2020 and SMPTE240m). The coefficients have been calculated by rounding
the floating point values to the nearest Q1.7 fixed-point value,
adjusting the rounding to ensure that the sum of each line in the matrix
is preserved to avoid overflows.

At the hardware level, the RGB to YUV conversion matrix is fully
configurable, custom encoding could be supported by extending the ISP
parameters if desired.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rkisp1: Don't pass the quantization to rkisp1_csm_config()
Laurent Pinchart [Fri, 12 Aug 2022 22:44:14 +0000 (00:44 +0200)]
media: rkisp1: Don't pass the quantization to rkisp1_csm_config()

The rkisp1_csm_config() function takes a pointer to the rkisp1_params
structure which contains the quantization value. There's no need to pass
it separately to the function. Drop it from the function parameters.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rkisp1: Configure quantization using ISP source pad
Laurent Pinchart [Fri, 12 Aug 2022 22:44:14 +0000 (00:44 +0200)]
media: rkisp1: Configure quantization using ISP source pad

The rkisp1_config_isp() function uses the format on the sink pad of the
ISP to configure quantization at the output of the ISP. This is
incorrect, as hinted by the src_frm variable name that stores the
format. Fix it by using the source pad.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rkisp1: Allow setting all color space fields on ISP source pad
Laurent Pinchart [Fri, 12 Aug 2022 22:56:22 +0000 (00:56 +0200)]
media: rkisp1: Allow setting all color space fields on ISP source pad

The ISP output color space is configured through the ISP source pad. At
the moment, only the quantization can be set. Extend it to the three
other color space fields:

- The ycbcr_enc field will be used to configure the RGB to YUV matrix
  (currently hardcoded to Rec. 601).

- The colorspace (which controls the color primaries) and xfer_func
  fields will not be used to configure the ISP, as the corresponding
  hardware blocks (the cross-talk RGB to RGB matrix and the tone mapping
  curve) are programmed directly by applications through ISP parameters.
  Nonetheless, those two fields should be set by applications to match
  the ISP configuration, in order to propagate the correct color space
  down the pipeline up to the capture video nodes.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rkisp1: Fix source pad format configuration
Laurent Pinchart [Fri, 12 Aug 2022 22:44:14 +0000 (00:44 +0200)]
media: rkisp1: Fix source pad format configuration

The ISP converts Bayer data to YUV when operating normally, and can also
operate in pass-through mode where the input and output formats must
match. Converting from YUV to Bayer isn't possible. If such an invalid
configuration is attempted, adjust it by copying the sink pad media bus
code to the source pad.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rkisp1: Allow setting color space on ISP sink pad
Laurent Pinchart [Fri, 12 Aug 2022 22:44:14 +0000 (00:44 +0200)]
media: rkisp1: Allow setting color space on ISP sink pad

The ISP accepts different color spaces on its input: for YUV input, it
doesn't set any restrictions, and for Bayer inputs, any color primaries
or transfer function can be accepted (YCbCr encoding isn't applicable
there, and quantization range can only be full).

Allow setting a color space on the ISP sink pad, with the aforementioned
restrictions. The settings don't influence hardware yet (only the YUV
quantization range will, anything else has no direct effect on the ISP
configuration), but can already be set to allow color space information
to be coherent across the ISP sink link.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rkisp1: Initialize color space on ISP sink and source pads
Laurent Pinchart [Fri, 12 Aug 2022 22:44:14 +0000 (00:44 +0200)]
media: rkisp1: Initialize color space on ISP sink and source pads

Initialize the four color space fields on the sink and source video pads
of the ISP in the .init_cfg() operation. As the main use case for the
ISP is to convert Bayer data to YUV, select a raw color space on the
sink pad and a limited range quantization of SYCC on the source pad by
default.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rockchip: rkisp1: Define macros for DPCC configurations in UAPI
Laurent Pinchart [Thu, 9 Jun 2022 10:31:13 +0000 (12:31 +0200)]
media: rockchip: rkisp1: Define macros for DPCC configurations in UAPI

Extend the UAPI rkisp1-config.h header with macros for all DPCC
configuration fields. While at it, clarify of fix issues in the DPCC
documentation.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rockchip: rkisp1: Mask invalid bits in DPCC parameters
Laurent Pinchart [Thu, 9 Jun 2022 10:31:13 +0000 (12:31 +0200)]
media: rockchip: rkisp1: Mask invalid bits in DPCC parameters

Restrict the DPCC configuration that can be set by userspace to valid
register bits. To do so, reorganize the related register macros to
define valid bitmasks, as well as bits of the DPCC mode register.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rockchip: rkisp1: Set DPCC methods enable bits inside loop
Laurent Pinchart [Thu, 9 Jun 2022 10:31:13 +0000 (12:31 +0200)]
media: rockchip: rkisp1: Set DPCC methods enable bits inside loop

The rkisp1_dpcc_config() function looks over methods sets to configure
them, but sets the RKISP1_CIF_ISP_DPCC_METHODS_SET_* registers outside
of the loop with hand-unrolled code. Move this to the loop to simplify
the code.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Dafna Hirschfeld <dafna@fastmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: mceusb: set timeout to at least timeout provided
Sean Young [Fri, 2 Sep 2022 10:32:21 +0000 (12:32 +0200)]
media: mceusb: set timeout to at least timeout provided

By rounding down, the actual timeout can be lower than requested. As a
result, long spaces just below the requested timeout can be incorrectly
reported as timeout and truncated.

Fixes: 877f1a7cee3f ("media: rc: mceusb: allow the timeout to be configurable")
Cc: stable@vger.kernel.org
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: imon: Remove the unneeded result variable
ye xingchen [Tue, 30 Aug 2022 08:30:27 +0000 (10:30 +0200)]
media: imon: Remove the unneeded result variable

Return the value send_packet() directly instead of storing it in another
redundant variable.

Reported-by: Zeal Robot <zealci@zte.com.cn>
Signed-off-by: ye xingchen <ye.xingchen@zte.com.cn>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: mceusb: Use new usb_control_msg_*() routines
Alan Stern [Fri, 26 Aug 2022 19:31:40 +0000 (21:31 +0200)]
media: mceusb: Use new usb_control_msg_*() routines

Automatic kernel fuzzing led to a WARN about invalid pipe direction in
the mceusb driver:

------------[ cut here ]------------
usb 6-1: BOGUS control dir, pipe 80000380 doesn't match bRequestType 40
WARNING: CPU: 0 PID: 2465 at drivers/usb/core/urb.c:410
usb_submit_urb+0x1326/0x1820 drivers/usb/core/urb.c:410
Modules linked in:
CPU: 0 PID: 2465 Comm: kworker/0:2 Not tainted 5.19.0-rc4-00208-g69cb6c6556ad #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS
1.13.0-1ubuntu1.1 04/01/2014
Workqueue: usb_hub_wq hub_event
RIP: 0010:usb_submit_urb+0x1326/0x1820 drivers/usb/core/urb.c:410
Code: 7c 24 40 e8 ac 23 91 fd 48 8b 7c 24 40 e8 b2 70 1b ff 45 89 e8
44 89 f1 4c 89 e2 48 89 c6 48 c7 c7 a0 30 a9 86 e8 48 07 11 02 <0f> 0b
e9 1c f0 ff ff e8 7e 23 91 fd 0f b6 1d 63 22 83 05 31 ff 41
RSP: 0018:ffffc900032becf0 EFLAGS: 00010282
RAX: 0000000000000000 RBX: ffff8881100f3058 RCX: 0000000000000000
RDX: ffffc90004961000 RSI: ffff888114c6d580 RDI: fffff52000657d90
RBP: ffff888105ad90f0 R08: ffffffff812c3638 R09: 0000000000000000
R10: 0000000000000005 R11: ffffed1023504ef1 R12: ffff888105ad9000
R13: 0000000000000040 R14: 0000000080000380 R15: ffff88810ba96500
FS: 0000000000000000(0000) GS:ffff88811a800000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007ffe810bda58 CR3: 000000010b720000 CR4: 0000000000350ef0
Call Trace:
<TASK>
usb_start_wait_urb+0x101/0x4c0 drivers/usb/core/message.c:58
usb_internal_control_msg drivers/usb/core/message.c:102 [inline]
usb_control_msg+0x31c/0x4a0 drivers/usb/core/message.c:153
mceusb_gen1_init drivers/media/rc/mceusb.c:1431 [inline]
mceusb_dev_probe+0x258e/0x33f0 drivers/media/rc/mceusb.c:1807

The reason for the warning is clear enough; the driver sends an
unusual read request on endpoint 0 but does not set the USB_DIR_IN bit
in the bRequestType field.

More importantly, the whole situation can be avoided and the driver
simplified by converting it over to the relatively new
usb_control_msg_recv() and usb_control_msg_send() routines.  That's
what this fix does.

Reported-and-tested-by: Rondreis <linhaoguo86@gmail.com>
Link: https://lore.kernel.org/all/CAB7eexLLApHJwZfMQ=X-PtRhw0BgO+5KcSMS05FNUYejJXqtSA@mail.gmail.com/
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Cc: stable@vger.kernel.org
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: remove reference to CONFIG_EMBEDDED in MEDIA_SUPPORT_FILTER
Lukas Bulwahn [Thu, 8 Sep 2022 10:43:34 +0000 (12:43 +0200)]
media: remove reference to CONFIG_EMBEDDED in MEDIA_SUPPORT_FILTER

The config EMBEDDED selects EXPERT, i.e., when EMBEDDED is enabled, EXPERT
is usually also enabled. Hence, it sufficient to have the option
MEDIA_SUPPORT_FILTER set to y if !EXPERT.

This way, MEDIA_SUPPORT_FILTER does not refer to CONFIG_EMBEDDED anymore
and allows us to remove CONFIG_EMBEDDED in the close future.

Remove the reference to CONFIG_EMBEDDED in MEDIA_SUPPORT_FILTER.

Link: https://lore.kernel.org/linux-media/20220908104337.11940-4-lukas.bulwahn@gmail.com
Signed-off-by: Lukas Bulwahn <lukas.bulwahn@gmail.com>
Reviewed-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: destage Hantro VPU driver
Ezequiel Garcia [Mon, 18 Jul 2022 21:41:21 +0000 (23:41 +0200)]
media: destage Hantro VPU driver

The Hantro mainline driver has been used in production
since several years and was only kept as a staging driver
due the stateless CODEC controls.

Now that all the stateless CODEC controls have been moved
out of staging, graduate the driver as well.

Signed-off-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: platform: mtk-mdp3: add MediaTek MDP3 driver
Moudy Ho [Tue, 23 Aug 2022 02:38:03 +0000 (04:38 +0200)]
media: platform: mtk-mdp3: add MediaTek MDP3 driver

This patch adds driver for MediaTek's Media Data Path ver.3 (MDP3).
It provides the following functions:
  color transform, format conversion, resize, crop, rotate, flip
  and additional image quality enhancement.

The MDP3 driver is mainly used for Google Chromebook products to
import the new architecture to set the HW settings as shown below:
  User -> V4L2 framework
    -> MDP3 driver -> SCP (setting calculations)
      -> MDP3 driver -> CMDQ (GCE driver) -> HW

Each modules' related operation control is sited in mtk-mdp3-comp.c
Each modules' register table is defined in file with "mdp_reg_" prefix
GCE related API, operation control  sited in mtk-mdp3-cmdq.c
V4L2 m2m device functions are implemented in mtk-mdp3-m2m.c
Probe, power, suspend/resume, system level functions are defined in
mtk-mdp3-core.c

[hverkuil: add 'depends on REMOTEPROC']

Signed-off-by: Ping-Hsun Wu <ping-hsun.wu@mediatek.com>
Signed-off-by: daoyuan huang <daoyuan.huang@mediatek.com>
Signed-off-by: Moudy Ho <moudy.ho@mediatek.com>
Tested-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: dt-binding: mediatek: add bindings for MediaTek CCORR and WDMA
Moudy Ho [Tue, 23 Aug 2022 02:38:01 +0000 (04:38 +0200)]
media: dt-binding: mediatek: add bindings for MediaTek CCORR and WDMA

This patch adds DT binding documentation for MediaTek's CCORR and
WDMA components.
These components exist in both MediaTek's Media Data Path 3(MDP3) and DRM,
and the bindings are placed under the folder "./soc/mediatek" to prevent
duplicate builds.

Signed-off-by: Moudy Ho <moudy.ho@mediatek.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: dt-binding: mediatek: add bindings for MediaTek MDP3 components
Moudy Ho [Tue, 23 Aug 2022 02:38:00 +0000 (04:38 +0200)]
media: dt-binding: mediatek: add bindings for MediaTek MDP3 components

This patch adds DT binding documents for Media Data Path 3 (MDP3)
a unit in multimedia system combined with several components and
used for scaling and color format convert.

Signed-off-by: Moudy Ho <moudy.ho@mediatek.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: xilinx: vipp: Fix refcount leak in xvip_graph_dma_init
Miaoqian Lin [Wed, 1 Jun 2022 04:25:14 +0000 (06:25 +0200)]
media: xilinx: vipp: Fix refcount leak in xvip_graph_dma_init

of_get_child_by_name() returns a node pointer with refcount
incremented, we should use of_node_put() on it when not need anymore.
Add missing of_node_put() to avoid refcount leak.

Fixes: df3305156f98 ("[media] v4l: xilinx: Add Xilinx Video IP core")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: xilinx: video: Add 1X12 greyscale format
Volodymyr Kharuk [Thu, 14 Jul 2022 08:34:44 +0000 (10:34 +0200)]
media: xilinx: video: Add 1X12 greyscale format

Extend the xilinx video driver with Y12_1X12 greyscale format

Signed-off-by: Volodymyr Kharuk <vkh@melexis.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: xilinx: csi2rxss: Add 1X12 greyscale format
Volodymyr Kharuk [Thu, 14 Jul 2022 08:34:43 +0000 (10:34 +0200)]
media: xilinx: csi2rxss: Add 1X12 greyscale format

Extend the csi2rxss with Y12_1X12 greyscale format

Signed-off-by: Volodymyr Kharuk <vkh@melexis.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: staging: media: imx: imx7-media-csi: Increase video mem limit
Petko Manolov [Sat, 9 Jul 2022 07:31:00 +0000 (09:31 +0200)]
media: staging: media: imx: imx7-media-csi: Increase video mem limit

Some high resolution (like Sony IMX492 47Mpix) sensors requre large amount space
for buffering.  64MB is far from sufficient so this patch increases the limit to
512MB.

Signed-off-by: Petko Manolov <petko.manolov@konsulko.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: uvcvideo: Limit power line control for Sonix Technology
huanglei [Sat, 20 Aug 2022 00:56:58 +0000 (02:56 +0200)]
media: uvcvideo: Limit power line control for Sonix Technology

The device does not implement the power line control correctly. Add a
corresponding control mapping override.

Bus 003 Device 003: ID 3277:0072 Sonix Technology Co., Ltd. USB 2.0 Camera
Device Descriptor:
  bLength                18
  bDescriptorType         1
  bcdUSB               2.00
  bDeviceClass          239 Miscellaneous Device
  bDeviceSubClass         2
  bDeviceProtocol         1 Interface Association
  bMaxPacketSize0        64
  idVendor           0x3277
  idProduct          0x0072
  bcdDevice            1.00
  iManufacturer           2 Sonix Technology Co., Ltd.
  iProduct                1 USB 2.0 Camera
  iSerial                 3 REV0001
  bNumConfigurations      1

Signed-off-by: huanglei <huanglei@kylinos.cn>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: uvcvideo: Use entity get_cur in uvc_ctrl_set
Yunke Cao [Thu, 7 Jul 2022 08:53:31 +0000 (10:53 +0200)]
media: uvcvideo: Use entity get_cur in uvc_ctrl_set

Entity controls should get_cur using an entity-defined function
instead of via a query. Fix this in uvc_ctrl_set.

Fixes: 65900c581d01 ("media: uvcvideo: Allow entity-defined get_info and get_cur")
Signed-off-by: Yunke Cao <yunkec@google.com>
Reviewed-by: Ricardo Ribalda <ribalda@chromium.org>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: uvcvideo: Fix typo 'the the' in comment
Slark Xiao [Fri, 22 Jul 2022 07:13:43 +0000 (09:13 +0200)]
media: uvcvideo: Fix typo 'the the' in comment

Replace 'the the' with 'the' in the comment.

Signed-off-by: Slark Xiao <slark_xiao@163.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: uvcvideo: Use indexed loops in uvc_ctrl_init_ctrl()
Laurent Pinchart [Mon, 18 Jul 2022 11:56:54 +0000 (13:56 +0200)]
media: uvcvideo: Use indexed loops in uvc_ctrl_init_ctrl()

As shown by the bug introduced in commit 86f7ef773156 ("media: uvcvideo:
Add support for per-device control mapping overrides"), the loop style
used by uvc_ctrl_init_ctrl() is error-prone. Rewrite the loops to use
indices instead.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Ricardo Ribalda <ribalda@chromium.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: uvcvideo: Fix memory leak in uvc_gpio_parse
José Expósito [Sat, 8 Jan 2022 17:04:39 +0000 (18:04 +0100)]
media: uvcvideo: Fix memory leak in uvc_gpio_parse

Previously the unit buffer was allocated before checking the IRQ for
privacy GPIO. In case of error, the unit buffer was leaked.

Allocate the unit buffer after the IRQ to avoid it.

Addresses-Coverity-ID: 1474639 ("Resource leak")

Fixes: 2886477ff987 ("media: uvcvideo: Implement UVC_EXT_GPIO_UNIT")
Signed-off-by: José Expósito <jose.exposito89@gmail.com>
Reviewed-by: Ricardo Ribalda <ribalda@chromium.org>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: renesas: vsp1: Add support for RZ/G2L VSPD
Biju Das [Sun, 28 Aug 2022 08:13:34 +0000 (10:13 +0200)]
media: renesas: vsp1: Add support for RZ/G2L VSPD

The RZ/G2L VSPD provides a single VSPD instance. It has the following
sub modules MAU, CTU, RPF, DPR, LUT, BRS, WPF and LIF.

The VSPD block on RZ/G2L SoCs does not have a version register, so
added a new compatible string "renesas,r9a07g044-vsp2" with a data
pointer containing the info structure. Also the reset line is shared
with the DU module.

Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Reviewed-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: renesas: vsp1: Add VSP1_HAS_NON_ZERO_LBA feature bit
Biju Das [Sun, 28 Aug 2022 08:13:33 +0000 (10:13 +0200)]
media: renesas: vsp1: Add VSP1_HAS_NON_ZERO_LBA feature bit

As per HW manual V3M and RZ/G2L SoCs has nonzero LIF buffer
attributes. So, introduce a feature bit for handling the same.

This patch also adds separate device info structure for V3M and V3H
SoCs, as both these SoCs share the same model ID, but V3H does not
have VSP1_HAS_NON_ZERO_LBA feature bit.

Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: renesas: vsp1: Add support for VSP software version
Biju Das [Sun, 28 Aug 2022 08:13:32 +0000 (10:13 +0200)]
media: renesas: vsp1: Add support for VSP software version

The VSPD block on RZ/G2L SoCs does not have a version register.

This patch adds support for adding VSP software version based on
device match.

Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: renesas: vsp1: Add support to deassert/assert reset line
Biju Das [Sun, 28 Aug 2022 08:13:31 +0000 (10:13 +0200)]
media: renesas: vsp1: Add support to deassert/assert reset line

As the resets DT property is mandatory, and is present in all .dtsi
in mainline, add support to perform deassert/assert using reference
counted reset handle.

Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Reviewed-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: dt-bindings: media: renesas,vsp1: Document RZ/G2L VSPD bindings
Biju Das [Sun, 28 Aug 2022 08:13:30 +0000 (10:13 +0200)]
media: dt-bindings: media: renesas,vsp1: Document RZ/G2L VSPD bindings

Document VSPD found in RZ/G2L SoC. VSPD block is similar to VSP2-D
found on R-Car SoC's, but it does not have a version register and
it has 3 clocks compared to 1 clock on vsp1 and vsp2.

This patch introduces a new compatible 'renesas,r9a07g044-vsp2' to
handle these differences.

Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Reviewed-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: meson: vdec: add missing clk_disable_unprepare on error in vdec_hevc_start()
Xu Qiang [Thu, 18 Aug 2022 06:57:53 +0000 (08:57 +0200)]
media: meson: vdec: add missing clk_disable_unprepare on error in vdec_hevc_start()

Add the missing clk_disable_unprepare() before return
from vdec_hevc_start() in the error handling case.

Fixes: 823a7300340e (“media: meson: vdec: add common HEVC decoder support”)
Signed-off-by: Xu Qiang <xuqiang36@huawei.com>
Reviewed-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: amphion: fix a bug that vpu core may not resume after suspend
Ming Qian [Thu, 18 Aug 2022 03:18:21 +0000 (05:18 +0200)]
media: amphion: fix a bug that vpu core may not resume after suspend

driver will enable the vpu core when request the first instance
on the core.
one vpu core can only support 8 streaming instances in the same
time, the instance won't be added to core's list before streamon.

so the actual instance count may be greater then the number in
the core's list.

in pm resume callback, driver will resume the core immediately if
core's list is not empty.
but this check is not accurate,
if suspend during one instance is requested, but not streamon,
then after suspend, the core won't be resume, and led to instance failure.

use the request_count instead of the core's list to check
whether is the core needed to resume immediately after suspend.

And it can make the pm suspend and resume callback more clear.

Fixes: 9f599f351e86 ("media: amphion: add vpu core driver")
Signed-off-by: Ming Qian <ming.qian@nxp.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: platform: fix some double free in meson-ge2d and mtk-jpeg and s5p-mfc
Hangyu Hua [Tue, 16 Aug 2022 08:58:19 +0000 (10:58 +0200)]
media: platform: fix some double free in meson-ge2d and mtk-jpeg and s5p-mfc

video_unregister_device will release device internally. There is no need to
call video_device_release after video_unregister_device.

Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: amphion: don't change the colorspace reported by decoder.
Ming Qian [Tue, 26 Jul 2022 03:02:29 +0000 (05:02 +0200)]
media: amphion: don't change the colorspace reported by decoder.

decoder will report the colorspace information
which is parsed from the sequence header,
if they are unspecified, just let application to determine it,
don't change it in driver.

Fixes: 6de8d628df6ef ("media: amphion: add v4l2 m2m vpu decoder stateful driver")
Signed-off-by: Ming Qian <ming.qian@nxp.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: amphion: adjust the encoder's value range of gop size
Ming Qian [Fri, 15 Jul 2022 07:38:00 +0000 (09:38 +0200)]
media: amphion: adjust the encoder's value range of gop size

adjust the value range of gop size from [0, 65535] to [1, 8000].
when the gop size is set to a too large value,
it may affect the encoded picture quality.
so constrain it to a reasonable range.

Fixes: 0401e659c1f92 ("media: amphion: add v4l2 m2m vpu encoder stateful driver")
Signed-off-by: Ming Qian <ming.qian@nxp.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: amphion: insert picture startcode after seek for vc1g format
Ming Qian [Fri, 15 Jul 2022 07:15:49 +0000 (09:15 +0200)]
media: amphion: insert picture startcode after seek for vc1g format

For format vc1, the amphion vpu requires driver to
help insert some custom startcode before sequence and frame.
the startcode is different for vc1l and vc1g format.

But the sequence startcode is only needed at the beginning,
and it's not expected after seek.
driver need to treat the codec header and the first frame after seek
as a normal frame, and insert picture startcode for it.

In previous patch, I just fix it for vc1l format,
and should fix the similar issue for vc1g too.

Fixes: e670f5d672ef (media: amphion: only insert the first sequence startcode for vc1l format)
Signed-off-by: Ming Qian <ming.qian@nxp.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: cedrus: Fix endless loop in cedrus_h265_skip_bits()
Dmitry Osipenko [Thu, 18 Aug 2022 20:33:08 +0000 (22:33 +0200)]
media: cedrus: Fix endless loop in cedrus_h265_skip_bits()

The busy status bit may never de-assert if number of programmed skip
bits is incorrect, resulting in a kernel hang because the bit is polled
endlessly in the code. Fix it by adding timeout for the bit-polling.
This problem is reproducible by setting the data_bit_offset field of
the HEVC slice params to a wrong value by userspace.

Cc: stable@vger.kernel.org
Fixes: 7678c5462680 (media: cedrus: Fix decoding for some HEVC videos)
Reported-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Signed-off-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: cedrus: Set the platform driver data earlier
Dmitry Osipenko [Thu, 18 Aug 2022 20:33:07 +0000 (22:33 +0200)]
media: cedrus: Set the platform driver data earlier

The cedrus_hw_resume() crashes with NULL deference on driver probe if
runtime PM is disabled because it uses platform data that hasn't been
set up yet. Fix this by setting the platform data earlier during probe.

Cc: stable@vger.kernel.org
Fixes: 50e761516f2b (media: platform: Add Cedrus VPU decoder driver)
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Signed-off-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Reviewed-by: Samuel Holland <samuel@sholland.org>
Acked-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: cedrus: Fix watchdog race condition
Nicolas Dufresne [Thu, 18 Aug 2022 20:33:06 +0000 (22:33 +0200)]
media: cedrus: Fix watchdog race condition

The watchdog needs to be scheduled before we trigger the decode
operation, otherwise there is a risk that the decoder IRQ will be
called before we have schedule the watchdog. As a side effect, the
watchdog would never be cancelled and its function would be called
at an inappropriate time.

This was observed while running Fluster with GStreamer as a backend.
Some programming error would cause the decoder IRQ to be call very
quickly after the trigger. Later calls into the driver would deadlock
due to the unbalanced state.

Cc: stable@vger.kernel.org
Fixes: 7c38a551bda1 ("media: cedrus: Add watchdog for job completion")
Signed-off-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Reviewed-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: videobuf2: Remove vb2_find_timestamp()
Ezequiel Garcia [Mon, 18 Jul 2022 22:02:11 +0000 (00:02 +0200)]
media: videobuf2: Remove vb2_find_timestamp()

Now that we've transitioned all users to vb2_find_buffer API,
remove the unused vb2_find_timestamp().

Signed-off-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
Acked-by: Tomasz Figa <tfiga@chromium.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: cedrus: Use vb2_find_buffer
Ezequiel Garcia [Mon, 18 Jul 2022 22:02:10 +0000 (00:02 +0200)]
media: cedrus: Use vb2_find_buffer

Use the newly introduced vb2_find_buffer API to get a vb2_buffer
given a buffer timestamp.

Cc: Maxime Ripard <mripard@kernel.org>
Cc: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Cc: Jernej Skrabec <jernej.skrabec@gmail.com>
Signed-off-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
Acked-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: mediatek: vcodec: Skip non CBR bitrate mode
Hirokazu Honda [Tue, 2 Aug 2022 04:42:42 +0000 (06:42 +0200)]
media: mediatek: vcodec: Skip non CBR bitrate mode

V4L2_MPEG_VIDEO_BITRATE_MODE_CBR is the only bitrate mode supported
by the mediatek driver. The other bitrates must be skipped in
QUERY_MENU.

Fixes: d8e8aa866ed8 ("media: mediatek: vcodec: Report supported bitrate modes")
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Chen-Yu Tsai <wenst@chromium.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: mediatek: vcodec: Add mt8188 encoder's chip name
Yunfei Dong [Wed, 27 Jul 2022 02:37:21 +0000 (04:37 +0200)]
media: mediatek: vcodec: Add mt8188 encoder's chip name

Getting mt8188's chip name according to encoder compatible name.

Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: mediatek: vcodec: add decoder compatible to support mt8188
Yunfei Dong [Wed, 27 Jul 2022 02:37:20 +0000 (04:37 +0200)]
media: mediatek: vcodec: add decoder compatible to support mt8188

1: add mt8188's compatible name: mediatek,mt8188-vcodec-dec.
2: mt8188 is lat single core architecture, using mtk_lat_sig_core_pdata to
   initialize private data.
3: Getting mt8188's chip name according to decoder compatible name.

Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: dt-bindings: media: mediatek: vcodec: add decoder dt-bindings for mt8188
Yunfei Dong [Wed, 27 Jul 2022 02:37:19 +0000 (04:37 +0200)]
media: dt-bindings: media: mediatek: vcodec: add decoder dt-bindings for mt8188

Add decoder document in dt-bindings yaml file for mt8188 platform.

Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: mediatek: vcodec: Use ctx vb2_queue mutex instead of device mutex
Irui Wang [Wed, 20 Jul 2022 08:57:31 +0000 (10:57 +0200)]
media: mediatek: vcodec: Use ctx vb2_queue mutex instead of device mutex

There is only one device mutex to lock vb2_queue when running
multi-instance encoding, it can be set by each encoder context.

[hverkuil: fix q_mutex documentation in the header]

Signed-off-by: Irui Wang <irui.wang@mediatek.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: mediatek: vcodec: Fix bitstream crop information error
Irui Wang [Wed, 20 Jul 2022 08:57:30 +0000 (10:57 +0200)]
media: mediatek: vcodec: Fix bitstream crop information error

Usually, the real bitstream width and height will set to driver
by vidioc_s_fmt, and vidioc_try_fmt() does align to get the
buffer width and height, driver calculate the encoded bitstream
crop information through them. The aligned resolution will be set
as real resolution now if user didn't set crop info by
V4L2_SEL_TGT_CROP, and the encoded bitstream may exist green line
because of crop information error.

Fixs: 'b6c57d313f5f8 ("media: mtk-vcodec: venc: remove redundant code")'

Signed-off-by: Irui Wang <irui.wang@mediatek.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: mediatek: vcodec: Remove encoder driver get IRQ resource
Irui Wang [Wed, 20 Jul 2022 08:57:29 +0000 (10:57 +0200)]
media: mediatek: vcodec: Remove encoder driver get IRQ resource

The "platform_get_resource(pdev, IORESOURCE_IRQ, 0)" is no longer
used after commit a1a2b7125e107("of/platform: Drop static setup of
IRQ resource from DT core"), so just remove the function in
encoder driver to avoid driver probe failed.

Signed-off-by: Irui Wang <irui.wang@mediatek.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: mediatek: vcodec: Add mt8188 encoder driver
Irui Wang [Wed, 20 Jul 2022 08:57:28 +0000 (10:57 +0200)]
media: mediatek: vcodec: Add mt8188 encoder driver

Add mt8188's compatible "mediatek,mt8188-vcodec-enc".
Add mt8188's device private data "mt8188_pdata".

Signed-off-by: Irui Wang <irui.wang@mediatek.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: dt-bindings: media: mediatek: vcodec: Add encoder dt-bindings for mt8188
Irui Wang [Wed, 20 Jul 2022 08:57:27 +0000 (10:57 +0200)]
media: dt-bindings: media: mediatek: vcodec: Add encoder dt-bindings for mt8188

Add encoder dt-bindings for mt8188.

Signed-off-by: Irui Wang <irui.wang@mediatek.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: mediatek: vcodec: Add encoder driver support for 34-bit iova
Irui Wang [Wed, 20 Jul 2022 08:57:26 +0000 (10:57 +0200)]
media: mediatek: vcodec: Add encoder driver support for 34-bit iova

Encoder driver got iova from IOMMU is 34-bit, for example:

Here is the sample code:
encoder input frame buffer dma address is:
frm_buf =
    vb2_dma_contig_plane_dma_addr(&vb2_v4l2_buffer->vb2_buf, 0);
the value of frm_buf is 0x1_ff30_0000.

encoder driver got the frm_buf and send the iova to SCP firmware
through SCP IPI message, then write to encoder hardware in SCP.
The iova is stored in IPI message as uint32_t data type, so the
value will be truncated from *0x1_ff30_0000* to *0xff30_0000*,
and then *0xff30_0000* will be written to encoder hardware, but
IOMMU will help to add the high *0x1_* bit back, so IOMMU can
translate the iova to PA correctly, encoder hardware can access
the correct memory for encoding.
Another reason to do this is the encoder hardware can't access
the 34-bit iova, IOMMU will help to add the remaining high bits
of iova. But for mt8188, encoder hardware can access 34-bit iova
directly, and encoder driver need write all 34 bits because
IOMMU can't help driver do this if the hardware support access
34-bit iova.
For the reasons above, this patch is added to support transfer
34-bit iova between kernel and SCP encoder driver. Use uint64_t
data type to store the iova, for compatibility with old chipsets,
add some new struct definitions for 34-bit.

[hverkuil: initialize wb and wb_34 to NULL]

Signed-off-by: Irui Wang <irui.wang@mediatek.com>
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: ti: cal: fix wdma irq for metadata
Tomi Valkeinen [Tue, 26 Apr 2022 07:06:18 +0000 (09:06 +0200)]
media: ti: cal: fix wdma irq for metadata

CAL HW interrupts are inherently racy. If we get both start and end
interrupts, we don't know what has happened: did the DMA for a single
frame start and end, or did one frame end and a new frame start?

Usually for normal pixel frames we get the interrupts separately. If
we do get both, we have to guess. The assumption in the code is that the
active vertical area is larger than the blanking vertical area, and thus
it is more likely that we get the end of the old frame and the start of
a new frame.

However, for embedded data, which is only a few lines high, we always
get both interrupts. Here the assumption is that we get both for the
same frame.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: ti: cal: combine wdma irq handling
Tomi Valkeinen [Tue, 26 Apr 2022 07:06:17 +0000 (09:06 +0200)]
media: ti: cal: combine wdma irq handling

Instead of handling the WDMA START and END interrupts separately, we
need to handle both at the same time to better manage the inherent race
conditions related to CAL interrupts.

Change the code so that we have a single function,
cal_irq_handle_wdma(), which gets two booleans, start and end, as
parameters, which allows us to manage the race conditions in the
following patch.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: ti: cal: use CSI-2 frame number for seq number
Tomi Valkeinen [Tue, 26 Apr 2022 07:06:16 +0000 (09:06 +0200)]
media: ti: cal: use CSI-2 frame number for seq number

The userspace needs a way to match received metadata buffers to pixel
data buffers. The obvious way to do this is to use the CSI-2 frame
number, as both the metadata and the pixel data have the same frame
number as they come from the same frame.

However, we don't have means to convey the frame number to userspace. We
do have the 'sequence' field, which with a few tricks can be used for
this purpose.

To achieve this, track the frame number for each virtual channel and
increase the sequence for each virtual channel by frame-number -
previous-frame-number, also taking into account the eventual wrap of the
CSI-2 frame number. If the CSI-2 peripheral does not support frame
numbers, CAL increases the frame number register by one each frame.

This way we get a monotonically increasing sequence number which is
common to all streams using the same virtual channel.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: ti: cal: rename sd_state to state
Tomi Valkeinen [Tue, 26 Apr 2022 07:06:15 +0000 (09:06 +0200)]
media: ti: cal: rename sd_state to state

Rename 'sd_state' parameters to 'state'. There are no other states, so
there is no ambiguity.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: ti: cal: fix useless variable init
Tomi Valkeinen [Tue, 26 Apr 2022 07:06:14 +0000 (09:06 +0200)]
media: ti: cal: fix useless variable init

'ret' is initialized needlessly in cal_legacy_try_fmt_vid_cap(). We can
also move the variable to the for block, which is the only place where
it is used.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: tc358743: limit msg.len to CEC_MAX_MSG_SIZE
Hans Verkuil [Wed, 24 Aug 2022 07:00:24 +0000 (09:00 +0200)]
media: tc358743: limit msg.len to CEC_MAX_MSG_SIZE

I expect that the hardware will have limited this to 16, but just in
case it hasn't, check for this corner case.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: media/cec: use CEC_MAX_MSG_SIZE instead of hardcoded 16
Hans Verkuil [Wed, 24 Aug 2022 06:57:26 +0000 (08:57 +0200)]
media: media/cec: use CEC_MAX_MSG_SIZE instead of hardcoded 16

Use the proper define for the maximum CEC message length instead of
hardcoding it.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: MAINTAINERS: change tc358743 maintainer
Hans Verkuil [Wed, 24 Aug 2022 06:20:46 +0000 (08:20 +0200)]
media: MAINTAINERS: change tc358743 maintainer

Move maintainer from Mats to Hans.

Add bindings file to the list of maintained files while we're at it.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mats Randgaard <matrandg@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: dw100: Fix an error handling path in dw100_probe()
Christophe JAILLET [Mon, 22 Aug 2022 09:40:43 +0000 (11:40 +0200)]
media: dw100: Fix an error handling path in dw100_probe()

After a successful call to media_device_init() it is safer to call
media_device_cleanup().

Add the missing call in the error handling path of the probe, as already
done in the remove function.

[hverkuil: fixed a typo in the commit log]

Fixes: bd090d4d995a ("media: dw100: Add i.MX8MP dw100 dewarper driver")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: cx23885: reset DMA on AMD Renior/Cezanne IOMMU due to RiSC engine stall
Daniel Lee Kruse [Fri, 19 Aug 2022 03:59:18 +0000 (05:59 +0200)]
media: cx23885: reset DMA on AMD Renior/Cezanne IOMMU due to RiSC engine stall

MythTv is unable to scan channels with APUs with the Renior IOMMU
that is also contained in the Cezanne line of APUs.  This issue
was discovered on the 5.15 version the kernel.  This patch adds
the IOMMU PCI ID to the broken_dev_id array.

This patch was developed with 5.19 of the media_tree repo.

[hverkuil: cleaned up the commit log a bit]

Signed-off-by: Daniel Lee Kruse <daniel.lee.kruse@proton.me>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: videodev2.h: drop V4L2_CAP_ASYNCIO
Hans Verkuil [Thu, 18 Aug 2022 07:51:40 +0000 (09:51 +0200)]
media: videodev2.h: drop V4L2_CAP_ASYNCIO

The V4L2_CAP_ASYNCIO capability was never implemented (and in fact
it isn't clear what it was supposed to do in the first place).

Drop it from the capabilities list. Keep it in videodev2.h with the
other defines under ifndef __KERNEL__ for backwards compatibility.

This will free up a capability bit for other future uses. And having
an unused and undefined I/O method is just plain confusing.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
2 years agomedia: rockchip: rga: Fix probe rga_parse_dt bugs
Ondrej Jirman [Mon, 15 Aug 2022 10:36:22 +0000 (12:36 +0200)]
media: rockchip: rga: Fix probe rga_parse_dt bugs

rga_parse_dt is missing a error return, so if some of the resources
return DEFER_PROBE, probe will succeed without these resources.

Signed-off-by: Ondrej Jirman <megi@xff.cz>
Co-developed-by: Jarrah Gosbell <kernel@undef.tools>
Signed-off-by: Jarrah Gosbell <kernel@undef.tools>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>