As shown by the bug introduced in commit 86f7ef77 ("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>

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: 2886477f ("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>

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>

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>

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>

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>

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>

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

Fixes: 823a7300 (“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>

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: 9f599f35 ("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>

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>

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: 6de8d628 ("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>

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: 0401e659 ("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>

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: e670f5d6 (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>

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: 7678c546 (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>

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: 50e76151 (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>

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: 7c38a551 ("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>

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>

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>

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: d8e8aa86 ("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>

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>

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>

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>

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>

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: 'b6c57d31 ("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>

The "platform_get_resource(pdev, IORESOURCE_IRQ, 0)" is no longer
used after commit a1a2b712("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>

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>

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>

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>

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>

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>

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>

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>

'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>

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>

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>

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>

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>

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>

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>

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>