Add helper functions to make it easier to adapt drivers to vb2.
These helpers take care of core locking and check if the filehandle is the
owner of the queue.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This prepares struct video_device for easier integration with vb2.
It also introduces a new lock that protects the vb2_queue. It is up
to the driver to use it or not. And the driver can associate an owner
filehandle with the queue to check whether queuing requests are
permitted for the calling filehandle.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This also fixes incorrect error handling in create_bufs: the return code
of __vb2_queue_alloc is the number of allocated buffers, and not a
traditional error code.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Split off the memory and type validation. This is done both from reqbufs
and create_bufs, and will also be done by vb2 helpers in a later patch.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
v4l_i2c_print_ioctl wasn't used and v4l_print_ioctl could be replaced by
v4l_printk_ioctl.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Since this could theoretically change the debug value while in the middle
of v4l2-ioctl.c, we make a copy of vfd->debug to ensure consistent debug
behavior.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Implement the default case which finalizes the table conversion and allows
us to remove the last part of the switch.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Add the necessary plumbing to make it possible to replace the switch by a
table driven implementation.
The ioctls ops can either be called directly, or by calling a small function
that does some additional work if needed.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The switch statement that determines how much data should be copied from
userspace is replaced by a table lookup.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
A block of code is moved down in the code to make later changes easier.
Do just the move without other changes to keep the diff readable for the
upcoming patch.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Add ".has_radio = 1" for BTTV_BOARD_MAGICTVIEW061 because there are
some CHP05x/CHP06x boards with a radio tuner.
I still have an analog Askey Magic TView card (CHP051, PCI 144f:3002)
which I use for radio only. This worked fine with kernel 2.6.37
(openSUSE 11.4), but no longer works with kernel 3.1.10 (openSUSE 12.1).
The reason apparently is that ".has_radio = 1" is missing in
bttv-cards.c for BTTV_BOARD_MAGICTVIEW061; when I add this line and
recompile the kernel, radio is working again. The line is still missing
in the latest git tree, for which I have generated the attached patch.
I can't test what will happen for CHP05x/CHP06x devices without radio;
however, it seems that other cards also have the problem that there are
some boards with radio and some without, so I hope that this can be
sorted out.
Signed-off-by: Thomas Betker <thomas.betker@freenet.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Add ADV7393 I²C-based video encoder driver. This driver has been tested on
custom hardware. It has been tested for composite output. It is derived from the
ADV7343 driver.
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Remove variable and ignore return value of stv090x_chk_signal().
Tested by compilation only.
[mchehab@redhat.com: instead of reverting and applying V3,
applied just the diff patch, for the sake of a cleaner history]
Signed-off-by: Peter Senna Tschudin <peter.senna@gmail.com>
Reviewed-by: Manu Abraham <manu@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The alternate setting number is not equal to the alternate setting index
in the interface alternate settings table. Use the alternate setting
number from the interface descriptor when calling usb_set_interface().
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Ming Lei <ming.lei@canonical.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
When video endpoint is configured as bulk, a ZLP is sent after every
video frames with size as multiple of 512 bytes. This is done so that
host can detect end of transfer and pass data for processing.
Still, frames that are multiple of 16K in size gets dropped. The ZLP
sent by camera is ignored by uvc_video_decode_bulk(). The makes sure
that the ZLP is not part of a video frame before ignoring it. If ZLP
follows a video frame, then it triggers completion callback.
[mchehab@redhat.com: Fix a small CodingStyle issue]
Signed-off-by: Jayakrishnan Memana <jayakrishnan.memana@maxim-ic.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This patch simply fixes some logging calls:
- Use KERN_INFO when printing the chip status.
- Add a missing space when logging the drxk_gate_ctrl call.
- Use the same logging text as always if the scu_command in GetQAMLockStatus fails.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Currently there are two different implementations (in the firmware) for
the QAM demodulator command: one takes 4 and the other takes 2
parameters. The driver shows an error in dmesg When using the
4-parameter command with firmware that implements the 2-parameter
command. Unfortunately this happens every time when chaning the
frequency (on DVB-C).
This patch simply makes configurable, how many command parameters will
be used. All existing drxk_config instances using the "drxk_a3.mc" were
updated because this firmware is the only loadable firmware where the
QAM demodulator command takes 4 parameters. Some firmwares in the ROM
might also use it. The drxk instances in the em28xx-dvb driver were also
updated to silence the warnings.
If no qam_demod_parameter_count is given in the drxk_config struct, then
the correct number of parameters will be auto-detected.
[mchehab@redhat.com: Fix a small CodingStyle issue at one comment]
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Tested-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
IQM_AF_CLKNEG_CLKNEGDATA__M is 0x2 and
IQM_AF_CLKNEG_CLKNEGDATA_CLK_ADC_DATA_POS is 0. (clkNeg | 0x2) is never
equal to zero so the condition can never be true.
I consulted with Ralph Metzler and the '|' should be changed to a '&'.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
CC: Ralph Metzler <rjkm@metzlerbros.de>
Tested-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
When tuner-xc2028 is not compiled as a module, dracut will
need to copy the firmware inside the initfs image.
So, use MODULE_FIRMWARE() to indicate such need.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
USB_VIDEO_CLASS_INPUT_EVDEV should be dependent on the UVC
selection, as otherwise, when UVC is unselected, this dependent
config still appears.
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Better arrange the remote controller driver items to happen after the
core support, on their proper menus, and making clerarer what is media
core options and what is media driver options.
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Commit 7a6f6c29d2 (cx231xx: use
URB_NO_TRANSFER_DMA_MAP) was intended to avoid mapping the DMA buffer
for URB twice. This works for the URBs allocated with usb_alloc_urb(),
as those are allocated from cohernent DMA pools, but the flag was also
added for the VBI and audio URBs, which have a manually allocated area.
This leaves the random trash in the structure after allocation as the
DMA address, corrupting memory and preventing VBI and audio from
working. Letting the USB core map the buffers solves the problem.
Signed-off-by: David Dillow <dave@thedillows.org>
Cc: Sri Deevi <srinivasa.deevi@conexant.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Most calls to cx231xx_capture_start() already use the values from
TRANSFER_TYPE, but cx231xx_capture_start() and
cx231xx_initialize_stream_xfer() were hand coding the values.
Use the named values (81 is never passed in), and simplify
cx231xx_capture_start(), as the switch statements were identical and
pcb_config->config_num is a u8, so any non-zero config got the same
result.
Signed-off-by: David Dillow <dave@thedillows.org>
Cc: Sri Deevi <srinivasa.deevi@conexant.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
changeset 2b719d7baf (v4l: drop v4l2_buffer.input and V4L2_BUF_FLAG_INPUT)
broke compilation on x86_64:
v4l2-compat-ioctl32.c: In function 'get_v4l2_buffer32':
v4l2-compat-ioctl32.c:391:4: error: expected ')' before 'return'
...
v4l2-compat-ioctl32.c: In function 'put_v4l2_buffer32':
v4l2-compat-ioctl32.c:475:4: error: expected ')' before 'return'
...
Add the missing close parenthesis character.
Cc: Sakari Ailus <sakari.ailus@iki.fi>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
After this patch, the MEDIA*_SUPPORT will be used as a filter,
exposing only devices that are pertinent to one of the 3
V4L types: webcam/grabber, radio, analog TV.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
In the past, it was possible to have either DVB or V4L2 core
as module and the other as builtin. Such config never make much
sense, and created several issues in order to make the Kconfig
dependency to work, as all drivers that depend on both (most
TV drivers) would need to be compiled as 'm'. Due to that,
the VIDEO_MEDIA config option were added.
Instead of such weird approach, let's just use the MEDIA_SUPPORT
=y or =m to select if the media subsystem core will be either
builtin or module, simplifying the building system logic.
Also, fix the tuners configuration, by enabling them only if
a tuner is required. So, if just webcam/grabbers support is
selected, no tuner option will be selected. Also, if only digital
TV is selected, no analog tuner support is selected.
That removes the need of using EXPERT customise options, when
analog TV is not selected.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Change the main items to:
<m> Multimedia support --->
[ ] Cameras/video grabbers support
[ ] Analog TV support
[ ] Digital TV support
[ ] AM/FM radio receivers/transmitters support
[ ] Remote Controller support
This provides an interface that is clearer to end users that
are compiling the Kernel, and will allow the building system
to automatically unselect drivers for unused functions.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Remove input field in struct v4l2_buffer and flag V4L2_BUF_FLAG_INPUT which
tells the former is valid. The flag is used by no driver currently.
Also change the documentation accordingly.
Signed-off-by: Sakari Ailus <sakari.ailus@iki.fi>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
xc5000 is just a tuner, not a decoder, so both DMB-TH and ISDB-T should
work properly there: it is just a matter of teaching the driver what
saw filter should be used and how to calculate the center frequency.
Requested-by: Choi Wing Chan <chanchoiwing@gmail.com>
Cc: Steven Toth <stoth@linuxtv.org>
Cc: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Add horizontal and vertical flip to the mem2mem_testdev driver.
Flip modes can be enabled either separately or simultaneously.
Signed-off-by: Tomasz Moń <desowin@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Value 0x00 was written instead of 0x10 which was aim.
Bit4 is overcurrent disable timer - ODT. According to datasheet ODT
functions are always enabled but setting it 1 is still recommended.
Thanks to Aubin Constans to pointing that bug.
Reported-by: Aubin Constans <aconstans@wyplay.com>
Signed-off-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Dan asked "Btw, why are the sizes specified in hex instead of
decimal here?". After looking code, I agree also decimal
presentation is clearer. So lets change those.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Message length check was wrong which could cause garbage sent to LNB,
but only in case garbage got from the userspace application.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This only adds support for the Terratec Cinergy T Stick Black device.
Changes from previous patches:
- fixed compiler warnings
- added fc0013 tuner handling to this patch
Signed-off-by: Thomas Mair <thomas.mair86@googlemail.com>
Acked-by: Antti Palosaari <crope@iki.fi>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Instead of using a table for reg bitmask, use a macro. This should save
some data segment space.
Cc: Thomas Mair <thomas.mair86@googlemail.com>
Cc: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Changelog for ver. 0.5:
- fixed code style and naming errors
Changelog for ver. 0.4:
- removed statistics as their calculation was wrong
- fixed bug in Makefile
- indentation and code style improvements
Signed-off-by: Thomas Mair <thomas.mair86@googlemail.com>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
commit 9ef449c6b3 ("[media] rc: Postpone ISR
registration") fixed an early ISR registration on several drivers. It did
however also introduced a bug by moving the invocation of pnp_port_start()
to the end of the probe function.
This patch fixes this issue by moving the invocation of pnp_port_start() to
an earlier stage in the probe function.
Cc: stable@vger.kernel.org
Cc: Jarod Wilson <jarod@redhat.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Implement API support to return AFC frequency shift, as this device
supports it. The only other driver that implements it is tda9887,
and the frequency there is reported in Hz. So, use Hz also for this
tuner.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
There are several bugs at the signal strength algorithm:
- It is using logical OR, instead of bit OR;
- It doesn't wait up to 18 ms as it should;
- the strength range is not ok.
Rework on it, in order to make it work.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
If g_tuner is called and the tuner is able to return the signal strength
via has_signal(), call the tunner callback to retrieve such data for all
tuner types, not only for radio ones.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Change the firmware logic to use request_firmware_nowait(), and
to preserve the loaded firmwares in memory, to reduce the risk
of troubles with buggy userspace apps.
With this change, while the firmware is being loaded, the driver
will return -EAGAIN to any calls. If, for some reason, firmware
failed to be loaded from userspace, it will return -ENODEV.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
If firmware is not loaded for some reason, or if it is not ready
yet, it makes no sense to honour to any DVB callbacks.
So, return -EAGAIN, as the error condition may be temporary.
If the device doesn't initialize, either because it requires a
firmware or because there's an error during init_drxk, returns
-ENODEV, to indicate such error, on all DVB callbacks.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Don't allow other devices at the same I2C bus to use it during
firmware load, in order to prevent using the device while it is
not on a sane state.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
As it will be using the unlocked version of i2c_transfer during
firmware loads, make sure that the priv state routine will be
used on all I2C calls, in preparation for the next patch that
will implement an exclusive lock mode to be used during firmware
load, at drxk_init.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The firmware blob may not be available when the driver probes.
Instead of blocking the whole kernel use request_firmware_nowait() and
continue without firmware.
This shouldn't be that bad on drx-k devices, as they all seem to have an
internal firmware. So, only the firmware update will take a little longer
to happen.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
* v4l_for_linus: (44 commits)
[media] smia: Fix compile failures
[media] Fix VIDIOC_DQEVENT docbook entry
[media] s5p-fimc: Fix control creation function
[media] s5p-mfc: Fix checkpatch error in s5p_mfc_shm.h file
[media] s5p-mfc: Fix setting controls
[media] v4l/s5p-mfc: added image size align in VIDIOC_TRY_FMT
[media] v4l/s5p-mfc: corrected encoder v4l control definitions
[media] v4l: mem2mem_testdev: Fix race conditions in driver
[media] s5p-mfc: Bug fix of timestamp/timecode copy mechanism
[media] cxd2820r: Fix an incorrect modulation type bitmask
[media] em28xx: Show a warning if the board does not support remote controls
[media] em28xx: Add remote control support for Terratec's Cinergy HTC Stick HD
[media] USB: Staging: media: lirc: initialize spinlocks before usage
[media] Revert "[media] media: mx2_camera: Fix mbus format handling"
[media] bw-qcam: driver and pixfmt documentation fixes
[media] cx88: fix firmware load on big-endian systems
[media] cx18: support big-endian systems
[media] ivtv: fix support for big-endian systems
[media] tuner-core: return the frequency range of the correct tuner
[media] v4l2-dev.c: fix g_parm regression in determine_valid_ioctls()
...
Conflicts:
Documentation/DocBook/media/v4l/vidioc-create-bufs.xml
drivers/media/video/em28xx/em28xx-cards.c
Make it possible for gpio-ir-recv users to choose a map name.
Cc: Ravi Kumar V <kumarrav@codeaurora.org>
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Add missing information so that udev can create an entry for gpio-ir-recv under
/dev/input/by-path/ .
Cc: Ravi Kumar V <kumarrav@codeaurora.org>
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Convert radio-sf16fmi to use generic LM7000 driver.
Tested with SF16-FMI, SF16-FMP and SF16-FMD.
Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Convert radio-aimslab to use generic LM7000 driver.
Tested with Reveal RA300.
Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Add very simple driver for Sanyo LM7000 AM/FM tuner chip. Only FM is supported
as there is no known HW with AM implemented.
This will be used by radio-aimslab and radio-sf16fmi.
Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This reverts commit 4ecbb69414.
As requested by Hans Verkuil:
> You accidentally merged the wrong first version of the lm7000 patch series.
>
> These are the correct second version patches:
>
> http://patchwork.linuxtv.org/patch/11689/
> http://patchwork.linuxtv.org/patch/11690/
> http://patchwork.linuxtv.org/patch/11691/
>
> The second version is much simpler and doesn't require the creation of a whole
> new driver.
Requested-by: Hans Verkuil <hverkuil@xs4all.nl>
Cc: Ondrej Zary <linux@rainbow-software.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This reverts commit bece083a6b.
As requested by Hans Verkuil:
> You accidentally merged the wrong first version of the lm7000 patch series.
>
> These are the correct second version patches:
>
> http://patchwork.linuxtv.org/patch/11689/
> http://patchwork.linuxtv.org/patch/11690/
> http://patchwork.linuxtv.org/patch/11691/
>
> The second version is much simpler and doesn't require the creation of a whole
> new driver.
Requested-by: Hans Verkuil <hverkuil@xs4all.nl>
Cc: Ondrej Zary <linux@rainbow-software.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This reverts commit 1e70a6cf7e.
As requested by Hans Verkuil:
> You accidentally merged the wrong first version of the lm7000 patch series.
>
> These are the correct second version patches:
>
> http://patchwork.linuxtv.org/patch/11689/
> http://patchwork.linuxtv.org/patch/11690/
> http://patchwork.linuxtv.org/patch/11691/
>
> The second version is much simpler and doesn't require the creation of a whole
> new driver.
Requested-by: Hans Verkuil <hverkuil@xs4all.nl>
Cc: Ondrej Zary <linux@rainbow-software.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This is basically s/getmode/getrxsubchans/ and s/setmode/setaudmode/
with some whitespace adjustment in affected lines to please the eye.
The rename is done to point out their relation to the rxsubchans and
audmode fields of struct v4l2_tuner.
I also corrected a commented out call to v4l_dbg in one of the lines.
Signed-off-by: Daniel Glöckner <daniel-gl@gmx.net>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Stops the media entity pipeline which was started earlier
if fimc_pipeline_validate fails.
[s.nawrocki: reworked to not exceed 80 characters line length]
Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This patch is an update to changed media_entity_pipeline_start()
signature in commit af88be3887,
"media: Add link_validate() op to check links to the sink pad"
It fixes the following warning:
drivers/media/video/s5p-fimc/fimc-lite.c: In function ‘fimc_lite_streamon’:
drivers/media/video/s5p-fimc/fimc-lite.c:765:29: warning: ignoring return value
of ‘media_entity_pipeline_start’, declared with attribute warn_unused_result [-Wunused-result]
Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Commit 77e7c4e624
"v4l: Allow changing control handler lock" changed
the lock field of struct v4l2_ctrl_handler to a pointer
and this driver wasn't updated properly. This patch fixes
following warning:
drivers/media/video/s5p-fimc/fimc-core.c: In function ‘fimc_ctrls_activate’:
drivers/media/video/s5p-fimc/fimc-core.c:644: warning: passing argument 1 of ‘mutex_lock’ from incompatible pointer type
include/linux/mutex.h:152: note: expected ‘struct mutex *’ but argument is of type ‘struct mutex **’
drivers/media/video/s5p-fimc/fimc-core.c:663: warning: passing argument 1 of ‘mutex_unlock’ from incompatible pointer type
include/linux/mutex.h:169: note: expected ‘struct mutex *’ but argument is of type ‘struct mutex **’
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Shorten pixel format descriptions that exceed 32 characters
so they're not being truncated when queried from user space.
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Only suspend the video pipeline devices if they were active before
the pm.suspend() helper is called. This patch prevents following error:
/# echo mem > /sys/power/state
[ 34.965000] PM: Syncing filesystems ... done.
[ 35.035000] Freezing user space processes ... (elapsed 0.01 seconds) done.
...
[ 35.105000] dpm_run_callback(): platform_pm_suspend+0x0/0x5c returns -22
[ 35.105000] PM: Device exynos-fimc-lite.1 failed to suspend: error -22
[ 35.105000] PM: Some devices failed to suspend
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The camera clocks managed by the driver were improperly reference counted
and remained disabled when multiple video nodes were opened simultaneously.
It manifested itself with following warning:
[12.920000] WARNING: at drivers/media/video/s5p-fimc/fimc-mdevice.c:787 __fimc_md_set_camclk+0x1c0/0x1dc()
[13.005000] Modules linked in:
[13.005000] Backtrace:
[13.040000] [<c0013084>] (dump_backtrace+0x0/0x10c) from [<c0454b70>] (dump_stack+0x18/0x1c)
[13.070000] r7:00000009 r6:00000313 r5:c02d576c r4:00000000
[13.155000] [<c0454b58>] (dump_stack+0x0/0x1c) from [<c0022ec4>] (warn_slowpath_common+0x54/0x6c)
[13.285000] [<c0022e70>] (warn_slowpath_common+0x0/0x6c) from [<c0022f00>] (warn_slowpath_null+0x24/0x2c)
[13.360000] r9:e1981010 r8:00000000 r7:c061d3fc r6:e1981010 r5:e1981030
[13.430000] r4:00000000
[13.430000] [<c0022edc>] (warn_slowpath_null+0x0/0x2c) from [<c02d576c>] (__fimc_md_set_camclk+0x1c0/0x1dc)
[13.550000] [<c02d55ac>] (__fimc_md_set_camclk+0x0/0x1dc) from [<c02d57b0>] (fimc_md_set_camclk+0x28/0x2c)
[13.630000] [<c02d5788>] (fimc_md_set_camclk+0x0/0x2c) from [<c02d57e8>] (__fimc_pipeline_shutdown+0x34/0x50)
[13.705000] [<c02d57b4>] (__fimc_pipeline_shutdown+0x0/0x50) from [<c02d5844>] (fimc_pipeline_shutdown+0x40/0x58)
[13.765000] r5:e2391200 r4:e2357704
[13.805000] [<c02d5804>] (fimc_pipeline_shutdown+0x0/0x58) from [<c02d4754>] (fimc_capture_close+0xcc/0xe4)
[13.915000] r5:e1b396c0 r4:e2357410
[13.915000] [<c02d4688>] (fimc_capture_close+0x0/0xe4) from [<c02b2d5c>] (v4l2_release+0x5c/0x80)
[13.970000] r7:00000010 r6:e1d2d990 r5:e1b396c0 r4:e2394800
[14.000000] [<c02b2d00>] (v4l2_release+0x0/0x80) from [<c00b66cc>] (fput+0xc0/0x22c)
[14.015000] r5:c157ef30 r4:e1b396c0
[14.015000] [<c00b660c>] (fput+0x0/0x22c) from [<c00b2ca0>] (filp_close+0x60/0x80)
[14.080000] [<c00b2c40>] (filp_close+0x0/0x80) from [<c00b2d78>] (sys_close+0xb8/0xf4)
[14.125000] r7:00000001 r6:e1b396c0 r5:c1400340 r4:c1400300
[14.125000] [<c00b2cc0>] (sys_close+0x0/0xf4) from [<c000f300>] (ret_fast_syscall+0x0/0x30)
[14.205000] r7:00000006 r6:beee5b94 r5:00000003 r4:b6f64fac
Fix this, as well as potential memory leaks due to not calling
v4l2_fh_release() on some error paths.
Also remove some error logs printed for events that aren't critical and
are normal conditions for some system configurations.
Also check if the device have been properly run-time enabled during
video node open.
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The checks are already done at the v4l2 framework.
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Allow memory buffer size to be increased by means of
struct v4l2_pix_plane_format::sizeimage at VIDIOC_S_FMT ioctl.
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The driver is supposed to create active media link from sensor N
(or its corresponding s5p-mipi-csis entity) to FIMC.N by default.
Instead s5p-mipi-csis.N entity gets always connected by a default
active link to FIMC.N, regardless of there are parallel bus sensor
entities already connected to FIMC.N. Correct this.
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
When video pipeline initialization fails, the ST_CAPT_BUSY flag
needs to be cleared before pm_runtime_put_sync is called.
Otherwise the runtime suspend routine tries to suspend device,
rather than just turning it off. Also fix potential null pointer
dereference in fimc_pipeline_shutdown().
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The switch/case was setting clock_setting that is not being used. Both switch/case and the variable definition were removed.
Currently clock is being calculated by the formula:
(state->fclk/1000000 - 8)
Tested by compilation only.
Signed-off-by: Peter Senna Tschudin <peter.senna@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
In function saa7164_api_i2c_read variable regval was set but not used.
Tested by compilation only.
Signed-off-by: Peter Senna Tschudin <peter.senna@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
