Allow to set a crop rectangle on the encoder output queue to set the
visible resolution as required by the V4L2 codec API.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
The hardware codec is not colorspace aware. We should trust userspace to
set the correct colorimetry information on the OUTPUT queue and mirror
the exact same setting on the CAPTURE queue.
There is no reason to restrict colorspace to JPEG for JPEG images, if
userspace injects the correct colorspace information into the JPEG
headers after encoding.
Fixes: b14ac54568 ("[media] coda: improve colorimetry handling")
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
If new buffers are queued after streamoff, the flag will be cleared
anyway, so this is mostly for the purpose of correctness.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
The decoder profile/level controls initially can be used to determine
supported profiles and levels. The values are set for a given stream
once the headers are parsed.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
Commit f1a81afc98 ("[media] m2m: fix bad unlock balance")
removed the last use of v4l2_m2m_ops.lock and
v4l2_m2m_ops.unlock hooks. They are not actually
used anymore. Remove them.
Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
Do not set context colorimetry on the raw (OUTPUT) queue for encoders.
Always set colorimetry on the coded queue (CAPTURE for encoders, OUTPUT
for decoders).
This also skips propagation of capture queue format and selection
rectangle on S_FMT(OUTPUT) to the CAPTURE queue for encoders.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
The driver helpfully resets the capture queue format and selection
rectangle whenever output format is changed. This only works while
the capture queue is not busy.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
Instead of duplicating the same code, call into coda_s_fmt_vid_out to
propagate the output format to the capture queue.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
Add support for the CodaHx4 VPU used on i.MX51.
Decoding h.264, MPEG-4, and MPEG-2 video works, as well as encoding
h.264. MPEG-4 encoding is not enabled, it currently produces visual
artifacts.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
[hans.verkuil@cisco.com: fix (bogus) sparse warning about uninited me_bits]
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Since bytesperline always fulfills VDOA width requirements, detile the
whole buffer instead of limiting to visible width. This stops TRY_FMT
from returning -EINVAL for YUYV capture buffers that are not a multiple
of 16 wide.
An alternative would be to always round up width to stride, as we report
the valid image rectange via G_SELECTION (V4L2_SEL_TGT_COMPOSE_DEFAULT),
but that would require all applications to handle the compose default
rectangle properly.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
The current driver implementation expects at least one buffer on
all queues to start streaming. Properly signal this to the vb2
core, to avoid confusion when streamon is racing with qbuf.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
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>
Acked-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
The on-chip SRAM of i.MX6S is only 128 KiB. 4 KiB of that are allocated
for suspend to RAM since commit df595746fa ("ARM: imx: add suspend in
ocram support for i.mx6q"). Reduce the requested IRAM size to 124 KiB to
avoid an allocation failure that causes the coda driver to not use the
SRAM at all.
Signed-off-by: Jan Luebbe <jlu@pengutronix.de>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Commit a53e35db70 ("reset: Ensure drivers are explicit when requesting
reset lines") started to transition the reset control request API calls
to explicitly state whether the driver needs exclusive or shared reset
control behavior. Convert all drivers requesting exclusive resets to the
explicit API call so the temporary transition helpers can be removed.
No functional changes.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Disabling the BWB works around hangups observed while decoding. Since no
issues have been observed while encoding, and disabling BWB also reduces
encoding performance, reenable it for encoding.
Fixes: 89ed025d5c ("[media] coda: disable BWB for all codecs on CODA 960")
Reported-by: Ian Arkver <ian.arkver.dev@gmail.com>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Tested-by: Ian Arkver <ian.arkver.dev@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
If an encoder stop command is issued after the output queue has already
stopped streaming, the qsequence counter has been reset to 0. Always
wake up the capture queue if the output queue is not streaming.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
CodaDx6 only accepts GOP sizes up to 60 frames, but CODA960 can handle
up to 99 frames. If we disable automatic I frame generation altogether
by setting GOP size to 0, we can let an application produce arbitrarily
large I frame intervals using the V4L2_CID_MPEG_VIDEO_FORCE_KEY_FRAME
control.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
This smatch warning:
coda/coda-common.c:706 coda_s_fmt() warn: missing break? reassigning 'ctx->tiled_map_type'
can be silenced by moving the ctx->tiled_map_type assignment into the
breakout condition. That way the field is not reassigned when falling
through to the next switch statement.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
CODA7541 supports H.264 BP level 3/3.1 and MPEG-4 SP level 5/6.
CODA960 supports H.264 BP level 4.0 and MPEG-4 SP level 5/6.
Implement the necessary profile and level controls to let userspace know
this.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
I would have liked the the picture run timeout error handler to be renamed to
something a bit more descriptive in the original commit fb2be08f8c ("[media]
coda: first step at error recovery").
Somehow v1 [1] was merged instead of v2 [2].
[1] https://patchwork.kernel.org/patch/9663965/
[2] https://patchwork.kernel.org/patch/9774239/
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
This implements a simple handler for the case where decode did not finish
sucessfully. This might be helpful during normal streaming, but for now it
only handles the case where the context would deadlock with userspace,
i.e. userspace issued DEC_CMD_STOP and waits for EOS, but after the failed
decode run we would hold the context and wait for userspace to queue more
buffers.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Implement the V4L2_CID_MPEG_VIDEO_FORCE_KEY_FRAME control to force IDR
frames. This is useful to implement VFU (Video Fast Update) on RTP
transmissions.
We already force an IDR frame at the beginning of each GOP to work
around a firmware bug on i.MX27, use the same mechanism to service IDR
requests from userspace.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
The hardware codec is not colorspace aware. We should trust userspace to
set the correct colorimetry information on the OUTPUT queue and mirror
the exact same setting on the CAPTURE queue.
There is no reason to restrict colorspace to JPEG or REC709 only. Also,
set the default colorspace, as returned by calling VIDIOC_TRY/S_FMT with
V4L2_COLORSPACE_DEFAULT, initially.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
As of commit bb475230b8 ("reset: make optional functions really
optional"), the reset framework API calls use NULL pointers to
describe optional, non-present reset controls.
This allows to return errors from devm_reset_control_get_optional
without special cases and to call reset_control_reset unconditionally.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
TRY_FMT already disables the YUYV format if the VDOA is not available.
ENUM_FMT must do the same.
Fixes: d40e98c13b ("[media] coda: support YUYV output if VDOA is used")
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
The call to v4ls_m2m_get_vq is only used to get the return value
which is not being used, so it appears to be redundant and can
be removed.
Detected with CoverityScan, CID#1420674 ("Useless call")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
drivers/media/platform/coda/coda-common.c: In function ‘coda_alloc_aux_buf’:
./include/linux/kern_levels.h:4:18: warning: format ‘%u’ expects argument of type ‘unsigned int’, but argument 4 has type ‘size_t {aka long unsigned int}’ [-Wformat=]
#define KERN_SOH "\001" /* ASCII Start Of Header */
^
./include/media/v4l2-common.h:69:9: note: in definition of macro ‘v4l2_printk’
printk(level "%s: " fmt, (dev)->name , ## arg)
^~~~~
./include/linux/kern_levels.h:10:18: note: in expansion of macro ‘KERN_SOH’
#define KERN_ERR KERN_SOH "3" /* error conditions */
^~~~~~~~
./include/media/v4l2-common.h:72:14: note: in expansion of macro ‘KERN_ERR’
v4l2_printk(KERN_ERR, dev, fmt , ## arg)
^~~~~~~~
drivers/media/platform/coda/coda-common.c:1341:3: note: in expansion of macro ‘v4l2_err’
v4l2_err(&dev->v4l2_dev,
^~~~~~~~
Hans wrote a similar patch, but it was fold with a Kconfig change.
So, I opted to keep my version.
Signed-off-by Hans Verkuil <hans.verkuil@cisco.com>
Acked-by Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Some vars are not used, as warned by gcc:
drivers/media/platform/coda/coda-common.c: In function 'coda_buf_is_end_of_stream':
drivers/media/platform/coda/coda-common.c:816:20: warning: variable 'src_vq' set but not used [-Wunused-but-set-variable]
struct vb2_queue *src_vq;
^~~~~~
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Recently, an unfinished patch was merged that added a third entry to the
beginning of the array of firmware locations without changing the code
to also look at the third element, thus pushing an old firmware location
off the list.
Fixes: 8af7779f3c ("[media] coda: add Freescale firmware compatibility location")
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Acked-by: Baruch Siach <baruch@tkos.co.il>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
With reordering enabled, the sequence init in CODA960 firmware requests an
unreasonable number of internal frames for some baseline profile streams.
Disabling the reordering feature manually if baseline streams are detected
fixes this problem.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Keeping buffers filled into the bitstream on a temporary list instead of
immediately calling vb2_buffer_done on each of them immediately allows
start_streaming to correctly decide whether they should be marked as
done or requeued if an error occurs after the bitstream has been filled.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
streamon and streamoff are used as boolean values, not as bitfields.
Therefore, the logical && should be used to combine them.
Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
I don't know what the BWB unit is, I guess W is for write and one of the
Bs is for burst. All I know is that there repeatedly have been issues
with it hanging on certain streams (ENGR00223231, ENGR00293425), with
various firmware versions, sometimes blocking something related to the
GDI bus or the GDI AXI adapter. There are some error cases that we don't
know how to recover from without a reboot. Apparently this unit can be
disabled by setting bit 12 in the FRAME_MEM_CTRL mailbox register to
zero, so do that to avoid crashes.
Side effects are reduced burst lengths when writing out decoded frames
to memory, so there is an "enable_bwb" module parameter to turn it back
on.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
There is no need to call v4l2_m2m_try_schedule to kick off draining the
bitstream buffer for the encoder, but we have to wake up the destination
queue in case there are no new OUTPUT buffers to be encoded and userspace
is already polling for new CAPTURE buffers.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
The Freescale provided imx-vpu looks for firmware files under /lib/firmware/vpu
by default. Make coda look there for firmware files to ease the update path.
Cc: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
The VDOA is able to transform the NV12 custom macroblock tiled format of
the CODA to YUYV format. If and only if the VDOA is available, the
driver can also provide YUYV support.
While the driver is configured to produce YUYV output, the CODA must be
configured to produce NV12 macroblock tiled frames and the VDOA must
transform the intermediate result into the final YUYV output.
Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
If the CODA driver is configured to produce NV12 output and the VDOA is
available, the VDOA can be used to transform the custom macroblock tiled
format to a raster-ordered format for scanout.
In this case, set the output format of the CODA to the custom macroblock
tiled format, disable the rotator, and use the VDOA to write to the v4l2
buffer. The VDOA is synchronized with the CODA to always un-tile the
frame that the CODA finished in the previous run.
Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
In order to make the VDOA work correctly, the CODA must produce frames
in tiled format. Print this information in the debug output.
Also print the color format in fourcc instead of the numeric value.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Correctly store the rectangle of valid video data in the destination
q_data before rounding up to macroblock size. This fixes the output
of VIDIOC_G_SELECTION for the capture side compose rectangle.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
In the case of coda_firmware_request() failure, we should release the
prevously acquired resources.
Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Commit 39b4da71ca ("reset: use ENOTSUPP instead of ENOSYS") changed
return value for reset controller if it missing.
This patch changes the CODA driver to handle this value.
Signed-off-by: Alexander Shiyan <shc_work@mail.ru>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Make this a proper typed array. Drop the old allocate context code since
that is no longer used.
Note that the memops functions now get a struct device pointer instead of
the struct device ** that was there initially (actually a void pointer to
a struct containing only a struct device pointer).
This code is now a lot cleaner.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Sakari Ailus <sakari.ailus@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Stop using alloc_ctx and just fill in the device pointer.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Cc: Scott Jiang <scott.jiang.linux@gmail.com>
Acked-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
If we bail out this early, v4l2_device_register() has not been called
yet, so no need to call v4l2_device_unregister().
Fixes: b7bd660a51 ("[media] coda: Call v4l2_device_unregister() from a single location")
Reported-by: Michael Olbrich <m.olbrich@pengutronix.de>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Try loading the firmware from firmware files named vpu_fw_imx*.bin, as
they are originally distributed by NXP. Fall back to v4l-coda*-imx6*.bin.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Freescale distribute their VPU firmware files with a 16 byte header
in BIT processor native order. This patch allows to detect the header
and to reorder the firmware on the fly.
With this patch it should be possible to use the distributed
vpu_fw_imx{53,6q,6d}.bin files directly after renaming them to
v4l-coda*-imx{53,6q,6dl}.bin.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Hook up the MPEG-2 ES decoder.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
We could support start streaming with an empty output queue for the
BIT decoders due to the bitstream buffer which could still contain
data at this point, but there is really no reason for userspace to
expect this to work. Simplify the code by disallowing it.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Philipp Zabel