Implement a notifier bound op to register media links from the remote
sub-device's source pad(s) to the video-mux sink pad(s).
Signed-off-by: Steve Longerbeam <slongerbeam@gmail.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Use v4l2_subdev_get_fwnode_pad_1_to_1() as the get_fwnode_pad operation.
The video mux maps fwnode port numbers and pad indexes 1:1.
Signed-off-by: Steve Longerbeam <slongerbeam@gmail.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Instead of using the convenience function
v4l2_async_register_fwnode_subdev(), parse the video-mux input endpoints
and set up the async sub-devices without using callbacks. The video-mux
knows which ports it must parse (the input ports) and how to handle
unconnected remotes, so it makes the code simpler to transfer control
of endpoint parsing to the driver.
Signed-off-by: Steve Longerbeam <slongerbeam@gmail.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Since its inclusion in v3.9, no users of the SuperH VEU mem2mem video
processing driver have appeared upstream. All VEU devices in SuperH
board code still bind to the "uio_pdrv_genirq" driver instead.
The original author marked the driver orphaned in v3.15.
Remove the driver; it can always be resurrected from git history when
needed.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Acked-by: Rob Landley <rob@landley.net>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Set the V4L2_CAP_IO_MC capability flag and remove the driver specific
vidioc_enum_input, vidioc_g_input and vidioc_s_input callbacks for the
media controller enabled part of the driver. Also add support mbus_code
filtering for format enumeration.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
It is spurious to call 'clk_disable_unprepare()' when
'clk_prepare_enable()' has not been called yet.
Re-order the error handling path to avoid it.
Fixes: a4260ea495 ("media: sun4i: Add H3 deinterlace driver")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Acked-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
[hverkuil-cisco@xs4all.nl: err_exlusive_rate -> err_exclusive_rate]
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
gcc warns about empty array declarations, which we get in this driver
when compile-testing without CONFIG_OF:
drivers/media/platform/ti-vpe/cal.c:2194:34: warning: array 'cal_of_match' assumed to have one element
2194 | static const struct of_device_id cal_of_match[];
Since all users of this driver do need CONFIG_OF anyway, there is no
point in making the array definition conditional to save space, so
just remove the #ifdef and move the array up a little.
Fixes: 343e89a792 ("[media] media: ti-vpe: Add CAL v4l2 camera capture driver")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
According to latest H/W manual v1.50, the description of channel
number in the VCDT/VCDT2 register is decremented by one.
Therefore, this patch fixes it about comment.
Signed-off-by: Koji Matsuoka <koji.matsuoka.xm@renesas.com>
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
According to the v4l spec s_fmt must return EBUSY while the
particular queue is streaming. Add such check in encoder and
decoder s_fmt methods.
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
According to stateful Codec API the decoder will process all
remaining buffers from before the source change event in
dynamic-resolution-change state and mark the last buffer with
V4L2_BUF_FLAG_LAST.
In Venus case the firmware doesn't mark that last buffer and
some mechanism have to be created in v4l decoder driver.
Fortunately the firmware interface (HFI) claims that the
decoder output buffers will be returned to v4l decoder
driver before it send the insufficient event.
In order to do that we save last queued in the driver capture
buffer in the event_notify and issue flush on output firmware
buffers queue. Once the saved buffer is returned (as a result of
flush command) we mark it as LAST. For all that possible we
extend HFI flush command with one more argument and one more
flush_done HFI driver callback.
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Presently the list initialization is done only in
dynamic-resolution-change state, which leads to list corruptions
and use-after-free. Init list_head unconditionally in
vdec_stop_capture called by vb2 stop_streaming without takeing
into account current codec state.
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Once the hfi_session_flush is issued by the vdec all queued
buffers to firmware should be returned to the v4l driver. Some
of those buffers are not processed at the time of flush command,
those buffers has filled len zero (no data). Catch that in
buffer_done callback and mark not filled capture buffers with
error state so that client can discard them.
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Currently calling venus_helper_buffers_done() will return buffers to
user for both capture and output queues in the same call. This is
wrong because both queues are really separate and calling
stop_streaming on one queue shouldn't return buffers for the other.
Solve this by add a new queue type argument and fix the clients of
the helper function.
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The array is not changed in the code, so make it const.
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The hfi_destroy function is called too early in remove method. It
destroys a mutex which is used later in the .remove from pmruntime.
Solve the issue by moving hfi_destroy after last usage of the mutex.
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Currently the encoder enables the rate control algorithms if the bitrate
control is non-zero. Implement the V4L2_CID_MPEG_VIDEO_FRAME_RC_ENABLE
and V4L2_CID_MPEG_VIDEO_MB_RC_ENABLE controls to allow userspace to
choose frame-level or macroblock-level rate control updates, or to
explicitly disable rate control. Both controls are initially enabled to
keep the current behavior.
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+huawei@kernel.org>
Enabling CONFIG_DMA_API_DEBUG=y and CONFIG_DMA_API_DEBUG_SG=y will
enable extra validation on DMA operations ensuring that the size
restraints are met.
When using the FCP in conjunction with the VSP1/DU, and display frames,
the size of the DMA operations is larger than the default maximum
segment size reported by the DMA core (64K). With the DMA debug enabled,
this produces a warning such as the following:
"DMA-API: rcar-fcp fea27000.fcp: mapping sg segment longer than device
claims to support [len=3145728] [max=65536]"
We have no specific limitation on the segment size which isn't already
handled by the VSP1/DU which actually handles the DMA allcoations and
buffer management, so define a maximum segment size of up to 4GB (a 32
bit mask).
Reported-by: Geert Uytterhoeven <geert+renesas@glider.be>
Fixes: 7b49235e83 ("[media] v4l: Add Renesas R-Car FCP driver")
Signed-off-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Tested-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
aliases property name must include only lowercase and '-'. Fix in dts
and driver.
Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org>
Reviewed-by: Matthias Brugger <matthias.bgg@gmail.com>
Reviewed-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The pointer 'common' is being assigned with a value that is never read,
the assignment is redundant and can be removed.
Addresses-Coverity: ("Unused value")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
As CEC support doesn't depend on MEDIA_SUPPORT, let's
place the platform drivers outside the media menu.
As a side effect, instead of depends on PCI, seco driver
can select it (and DMI).
Acked-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The stop-state timeout needs to be over 100us as per CSI spec. With the
CAL fclk of 266 MHZ on DRA76, with the current value the driver uses,
the timeout is 24us. Too small timeout will cause failure to enable the
streaming.
Also, the fclk can be different on other SoCs, as is the case with AM65x
where the fclk is 250 MHz.
This patch fixes the timeout by calculating it correctly based on the
fclk rate.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.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+huawei@kernel.org>
Sometimes there is a timeout when waiting for the Stop-State. Testing
shows that sometimes we need to wait more than what the current code
does. It is not clear how long this wait can be, but it is based on how
quickly the sensor provides a valid clock, and how quickly CAL syncs to
it.
Change the code to make it more obvious how long we'll wait, and set a
wider range for usleep_range. Increase the timeout to 750ms.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Sometimes there is a timeout when waiting for the 'ComplexIO Reset
Done'. Testing shows that sometimes we need to wait more than what the
current code does. It is not clear how long this wait can be, but it is
based on how quickly the sensor provides a valid clock, and how quickly
CAL syncs to it.
Change the code to make it more obvious how long we'll wait, and set a
wider range for usleep_range. Increase the timeout to 750ms.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
To make csi2_wait_for_phy() more readable, move code to separate
functions.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
[hverkuil-cisco@xs4all.nl: delete empty line before } ]
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Set DMA max seg size correctly to get rid of warnings on 64 bit
platforms:
DMA-API: cal 6f03000.cal: mapping sg segment longer than device claims to support [len=720896] [max=65536]
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
CAL_CSI2_CTX register has LINES field, which, according to the
documentation, should be programmed to the number of lines transmitted
by the camera. If the number of lines is unknown, it can be set to 0.
The driver sets the field to 0 for some reason, even if we know the
number of lines.
This patch sets the number of lines properly, which will allow the HW to
discard extra lines (if the sensor would send such for some reason),
and, according to documentation: "This leads to regular video timings
and avoids potential artifacts".
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
After ComplexIO reset, a dummy read to PHY is needed as per CAL spec to
finish the reset. Currently the driver reads a ComplexIO register, not
PHY register. Fix this.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Move the code to enable and disable ComplexIO power to its own function.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Simplify the code by using reg_write_field() where trivially possible.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Remove a bunch of IRQ defines, of which only CAL_HL_IRQ_ENABLE and
CAL_HL_IRQ_CLEAR are used, and these defines only end up obfuscating
code.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
These macros only obfuscate the code, so drop them.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Instead of having identical code block to handle irqs for the two CAL
ports, we can have a for loop and a single code block.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The driver does not print any errors on ComplexIO reset timeout or when
waiting for stop-state, making it difficult to debug and notice
problems.
Add error prints for these cases.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
CAL reports various errors via IRQs, which are not handled at all by the
current driver. Add code to enable and catch those IRQs and print
errors. This will make it much easier to notice and debug issues with
sensors.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
[hverkuil-cisco@xs4all.nl: fix: spaces preferred around that '-']
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Now that cal_runtime_get and cal_runtime_put are only direct wrappers to
pm_runtime_get/put, we can drop cal_runtime_get and cal_runtime_put.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
We need to do errata handling every time CAL is being enabled. The code
is currently in cal_runtime_get(), which is not the correct place for
it.
Move the code to cal_runtime_resume, which is called every time CAL is
enabled.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
i913_errata() sets a bit to 1 in PHY_REG10, but for some reason uses
CAL_CSI2_PHY_REG0_HSCLOCKCONFIG_DISABLE for the bit value. The value of
that macro is 1, so it works, but is still wrong.
Fix this to 1.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
IRQENABLE_SET registers are (usually) not meant to be read, only written
to. The current driver needlessly uses read-modify-write cycle to enable
IRQ bits.
The read-modify-write has no bad side effects here, but it's still
better to clean this up by only using write.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
When the CAL driver stops streaming, it will shut everything down
without waiting for the current frame to finish. This leaves the CAL DMA
in a slightly undefined state, and when CAL DMA is enabled when the
stream is started the next time, the old DMA transfer will continue.
It is not clear if the old DMA transfer continues with the exact
settings of the original transfer, or is it a mix of old and new
settings, but in any case the end result is memory corruption as the
destination memory address is no longer valid.
I could not find any way to ensure that any old DMA transfer would be
discarded, except perhaps full CAL reset. But we cannot do a full reset
when one port is getting enabled, as that would reset both ports.
This patch tries to make sure that the DMA transfer is finished properly
when the stream is being stopped. I say "tries", as, as mentioned above,
I don't see a way to force the DMA transfer to finish. I believe this
fixes the corruptions for normal cases, but if for some reason the DMA
of the final frame would stall a lot, resulting in timeout in the code
waiting for the DMA to finish, we'll again end up with unfinished DMA
transfer. However, I don't know what could cause such a timeout.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The Venus driver is voting Configuration NoC during .probe but not clear
voting in .suspend. Because of this NoC is up during shutdown also. As a
consequence the whole device could leak energy while in .suspend.
So correct this by moving voting in .resume and unvoting
in .suspend
Signed-off-by: Mansur Alisha Shaik <mansur@codeaurora.org>
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The current codebase makes use of the zero-length array language
extension to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:
struct foo {
int stuff;
struct boo array[];
};
By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on.
Also, notice that, dynamic memory allocations won't be affected by
this change:
"Flexible array members have incomplete type, and so the sizeof operator
may not be applied. As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]
This issue was found with the help of Coccinelle.
[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 7649773293 ("cxgb3/l2t: Fix undefined behaviour")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The current codebase makes use of the zero-length array language
extension to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:
struct foo {
int stuff;
struct boo array[];
};
By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on.
Also, notice that, dynamic memory allocations won't be affected by
this change:
"Flexible array members have incomplete type, and so the sizeof operator
may not be applied. As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]
This issue was found with the help of Coccinelle.
[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 7649773293 ("cxgb3/l2t: Fix undefined behaviour")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Implement pmruntime autosuspend in video decoder. This will
allow to save power while the userspace is inactive for some
reasonable period of time. Here we power-off venus core clocks
and power domain and don't touch vcodec because it is under
hardware control. The later decision is made to simplify the
code and avoid a mess in the power management code.
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Each jpeg can have the huffman tables optimized for its specific content
meaning that the table lenghts and values don't match the standard table
of substitutions so there's no reason to hardcode and expect the standard
lengths, otherwise we just end up rejecting optimized jpegs altogether.
Tested on CODA960.
Signed-off-by: Adrian Ratiu <adrian.ratiu@collabora.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Similar to commit 9ee50a9489 ("media: coda: lock capture queue wakeup
against encoder stop command"), make sure that a JPEG decoder stop
command running concurrently with a decoder finish_run always either
flags the last returned buffer or wakes up the capture queue to signal
the end of stream condition afterwards.
This was not necessary for BIT processor contexts because of the need to
release the bitstream buffer with the stream end condition. In contrast,
the JPEG decoder can be finished with decoding the image between the
time the application queues the last output buffer and the time it
issues the decoder stop command.
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+huawei@kernel.org>
If a JPEG decoding application queues the last capture and output
buffers, issues a decoder stop command after the decoding is already
done, and then dequeues the last capture buffer, it is not marked as
last. Detect this condition in the decoder stop command and mark the
last buffer on the capture done list.
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+huawei@kernel.org>
Split marking the last metadata entry into a helper function to simplify
coda_decoder_cmd.
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+huawei@kernel.org>
This patch adds JPEG decoding support for CODA960, handling the JPEG
hardware directly. A separate JPEG decoder video device is created due
to the separate hardware unit and different supported pixel formats.
While the hardware can not change subsampling on the fly, it can decode
4:2:2 subsampled JPEG images into YUV422P.
[m.felsch@pengutronix.de: fix qsequence counting by explicitly
checking for the !use_bit case]
[hverkuil-cisco@xs4all.nl: fix 'warning: missing braces around initializer']
[mchehab+huawei@kernel.org: fix 'BRACES: Unbalanced braces around else statement']
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Marco Felsch <m.felsch@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
