[media] exynos4-is: Extend link_notify handler to support fimc-is/lite pipelines
This patch corrects the link_notify handler to support more complex
pipelines, including fimc-lite and fimc-is entities.
After the FIMC-IS driver addition the assumptions made in the link_notify
callback are no longer valid, e.g. the link between fimc-lite subdev and
its video node is not immutable any more and there is more subdevs than
just sensor, MIPI-CSIS and FIMC(-LITE).
The graph is now walked and for each video node found a media pipeline
which ends at this node is disabled/enabled (the subdevs are powered
on/off).
[media] media: Change media device link_notify behaviour
Currently the media device link_notify callback is invoked before the
actual change of state of a link when the link is being enabled, and
after the actual change of state when the link is being disabled.
This doesn't allow a media device driver to perform any operations
on a full graph before a link is disabled, as well as performing
any tasks on a modified graph right after a link's state is changed.
This patch modifies signature of the link_notify callback. This
callback is now called always before and after a link's state change.
To distinguish the notifications a 'notification' argument is added
to the link_notify callback: MEDIA_DEV_NOTIFY_PRE_LINK_CH indicates
notification before link's state change and
MEDIA_DEV_NOTIFY_POST_LINK_CH corresponds to a notification after
link flags change.
[media] exynos4-is: Move __fimc_videoc_querycap() function to the common module
Move __fimc_videoc_querycap() function to the common exynos4-is-common.ko
module so it don't need to be reimplemented in multiple video node drivers
of the exynos4-is.
Remove duplicated code in the vidioc_try_fmt and vidioc_s_fmt handlers.
This is a pre-requsite to allow successful fimc.capture video open even
if its corresponding media entities are not linked into a complete
pipeline.
[media] exynos4-is: Remove leftovers of non-dt FIMC-LITE support
FIMC-LITE devices are never looked up by iterating over all platform
devices with bus_for_each_device() as these IP blocks are available
only on dt-only Exynos SoC platforms.
[media] exynos4-is: Add locking at fimc(-lite) subdev unregistered handler
Protect the fimc/fimc-lite video nodes unregistration with their video
lock. This prevents a kernel crash when e.g. udev opens a video node
right after the driver registers it and then the driver tries to
unregister it and defers its probing. Using video_is_unregistered()
together with the video mutex allows safe unregistration of the video
nodes at any time.
[media] exynos4-is: Remove WARN_ON() from __fimc_pipeline_close()
It's not a critical error to call __fimc_pipeline_close() with missing
sensor subdev entity. Replace WARN_ON() with pr_warn() and return 0
instead of -EINVAL to fix control flow in some conditions.
[media] exynos4-is: Do not use asynchronous runtime PM in release fop
Use pm_runtime_put_sync() instead of pm_runtime_put() to avoid races
in handling the 'state' bit flags when the fimc-capture drivers'
runtime_resume callback is called from the PM workqueue.
[media] exynos4-is: Media graph/video device locking rework
Remove driver private video node reference counters and use entity->use_count
instead. This makes the video pipelines power handling more similar to the
method used in omap3isp driver.
Now the graph mutex is taken always after the video mutex, as it is not
possible to ensure apposite order at the all modules.
[media] exynos4-is: Preserve state of controls between /dev/video open/close
This patch moves the code for inheriting subdev v4l2 controls on the
FIMC video capture nodes from open()/close() fops to the link setup
notification callback. This allows for the state of the FIMC controls
to be always kept, in opposite to the current situation when it is
lost when last process closes video device.
There is no visible change for the original V4L2 compliant interface.
For the MC aware applications (user_subdev_api == true) inheriting
of the controls is dropped, since there can be same controls on the
subdevs withing single pipeline, now when the ISP (FIMC-IS) is also
used.
This patch is a prerequisite to allow /dev/video device to be opened
without errors even if there is no media links connecting it to an
image source (sensor) subdev. This is required for a libv4l2 plugin
to be initialized while a video node is opened and it also should be
possible to always open the device to query the capabilities.
[media] s5c73m3: Do not ignore errors from regulator_enable()
This fixes following compilation warning:
drivers/media/i2c/s5c73m3/s5c73m3-core.c: In function ‘__s5c73m3_power_off’:
drivers/media/i2c/s5c73m3/s5c73m3-core.c:1389:19: warning: ignoring return value
of ‘regulator_enable’, declared with attribute warn_unused_result
These symbols are used only in their respective files and hence
should be made static.
[s.nawrocki@samsung.com: dropped the __fimc_is_hw_update_param()
function change]
Wei Yongjun [Mon, 13 May 2013 04:49:13 +0000 (01:49 -0300)]
[media] s5p-tv: fix error return code in mxr_acquire_video()
Fix to return a negative error code in the vb2_dma_contig_init_ctx()
error handling case instead of 0, as done elsewhere in this function.
Also vb2_dma_contig_init_ctx() return ERR_PTR() in case of error and
never return NULL, so use IS_ERR() replace IS_ERR_OR_NULL().
clk_unprepare checks for NULL pointer. Hence convert IS_ERR_OR_NULL
to IS_ERR only.
[s.nawrocki: added initialisation of the clock array to ERR_PTR() value]
[media] exynos4-is: Remove redundant NULL check in fimc-lite.c
clk_unprepare checks for NULL pointer. Hence convert IS_ERR_OR_NULL
to IS_ERR only.
[s.nawrocki: replaced initialisations to NULL with ERR_PTR() value]
dev_info is preferred to printk. Silences the related checkpatch
warning.
WARNING: Prefer netdev_info(netdev, ... then dev_info(dev, ...
then pr_info(... to printk(KERN_INFO ...
Laurent Pinchart [Mon, 11 Mar 2013 15:02:13 +0000 (12:02 -0300)]
[media] omap3isp: Defer probe when the IOMMU is not available
When the OMAP3 ISP driver is compiled in the kernel the device can be
probed before the corresponding IOMMU is available. Defer the probe in
that case, and fix a crash in the error path.
Reported-by: Javier Martin <javier.martin@vista-silicon.com> Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
[media] drxk_hard: Remove most 80-cols checkpatch warnings
X-Patchwork-Delegate: mchehab@redhat.com
There are a few cases where breaking the code into separate
lines make it worse to read. However, on several places,
breaking it to make checkpatch.pl happier is OK and improves
code readability.
So, break longer lines where that won't cause harm.
X-Patchwork-Delegate: mchehab@redhat.com
There are several places where: state->var = (some_var)
The parenthesis there are doing nothing but making it
harder to read and breaking the 80 columns soft limits.
Just get rid of it.
[media] drxk_hard.h: don't use more than 80 columns
X-Patchwork-Delegate: mchehab@redhat.com
Almost all 80-col warnings are related to comments. There's
one, however, that it is due to a one-line enum declaration
for enum agc_ctrl_mode.
Break it into one line per enumered data.
[media] drxk_hard.h: Remove some alien comment markups
X-Patchwork-Delegate: mchehab@redhat.com
The comments markup language used on Kernel is defined at:
Documentation/kernel-doc-nano-HOWTO.txt
Remove invalid markups from the header file.
X-Patchwork-Delegate: mchehab@redhat.com
Fixes the following checkpatch.pl warnings:
WARNING: msleep < 20ms can sleep for up to 20ms; see Documentation/timers/timers-howto.txt
+ msleep(10);
WARNING: msleep < 20ms can sleep for up to 20ms; see Documentation/timers/timers-howto.txt
+ msleep(1);
WARNING: msleep < 20ms can sleep for up to 20ms; see Documentation/timers/timers-howto.txt
+ msleep(1);
WARNING: msleep < 20ms can sleep for up to 20ms; see Documentation/timers/timers-howto.txt
+ msleep(1);
WARNING: msleep < 20ms can sleep for up to 20ms; see Documentation/timers/timers-howto.txt
+ msleep(1);
WARNING: msleep < 20ms can sleep for up to 20ms; see Documentation/timers/timers-howto.txt
+ msleep(1);
WARNING: msleep < 20ms can sleep for up to 20ms; see Documentation/timers/timers-howto.txt
+ msleep(1);
WARNING: msleep < 20ms can sleep for up to 20ms; see Documentation/timers/timers-howto.txt
+ msleep(1);
X-Patchwork-Delegate: mchehab@redhat.com
Thare are lots of CamelCase warnings produced by checkpatch.pl.
This weren't fixed at the time the driver got submitted due
to the lack of manpower do to such cleanup.
Now that I have one script that automates this task, cleans
it. That makes the driver almost checkpatch-compliant,
except for 80 column warnings.
X-Patchwork-Delegate: mchehab@redhat.com
Log10 routine is already defined at dvb_math.h and provides a good
enough approximation for 100 x log10(). So, instead of reinventing
the wheel, use the already existing function.
While here, don't use CamelCase on the function name.
[media] Add support for Crypto Redi PC50A device (rtl2832u + FC0012 tuner)
The device has been tested on a MIPSel box with kernel 3.1.1 and backported media_tree drivers
The kernel detects the device with the following output:
usbcore: registered new interface driver dvb_usb_rtl28xxu
usb 1-2: dvb_usb_v2: found a 'Crypto Redi PC50A' in warm state
usb 1-2: dvb_usb_v2: will pass the complete MPEG2 transport stream to the software demuxer
DVB: registering new adapter (Crypto Redi PC50A)
usb 1-2: DVB: registering adapter 1 frontend 0 (Realtek RTL2832 (DVB-T))...
i2c i2c-4: fc0012: Fitipower FC0012 successfully identified
usb 1-2: dvb_usb_v2: 'Crypto Redi PC50A' successfully initialized and connected
[media] rtl2832u: restore ir remote control support
Hi,
This patch uses the driver from openpli[1] as a template to restore the remote control support.
I had to divert from the original to use the in kernel rc protocol decoder. The key repetition does,
not seem to work but I cant find the problem in the driver. As a raw rc provider, no key table is
hardcoded.
Rodrigo.
[1]: https://aur.archlinux.org/packages/dvb-usb-rtl2832u-openpli/?comments=all
Andrzej Hajda [Mon, 3 Jun 2013 08:16:13 +0000 (05:16 -0300)]
[media] media: Rename media_entity_remote_source to media_entity_remote_pad
Function media_entity_remote_source actually returns the remote pad to
the given one, regardless if this is the source or the sink pad.
Name media_entity_remote_pad is more adequate for this function.
Signed-off-by: Andrzej Hajda <a.hajda@samsung.com> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com> Acked-by: Sakari Ailus <sakari.ailus@iki.fi> Acked-by: Sylwester Nawrocki <s.nawrocki@samsung.com> Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
[media] Add support for 'Digital Dual TV Receiver CTVDIGDUAL v2
Tested on a MIPSel box with 3.3.6 kernel
The kernel output when the device will be detected follows:
usbcore: registered new interface driver dvb_usb_it913x
it913x: Chip Version=01 Chip Type=9135
it913x: Remote propriety (raw) mode
it913x: Dual mode=3 Tuner Type=38
it913x: Chip Version=01 Chip Type=9135
usb 2-1: dvb_usb_v2: found a 'Digital Dual TV Receiver CTVDIGDUAL_V2' in cold state
usb 2-1: dvb_usb_v2: downloading firmware from file 'dvb-usb-it9137-01.fw'
it913x: FRM Starting Firmware Download
it913x: FRM Firmware Download Completed - Resetting Device
it913x: Chip Version=01 Chip Type=9135
it913x: Firmware Version 204147968
usb 2-1: dvb_usb_v2: found a 'Digital Dual TV Receiver CTVDIGDUAL_V2' in warm state
usb 2-1: dvb_usb_v2: will pass the complete MPEG2 transport stream to the software demuxer
DVB: registering new adapter (Digital Dual TV Receiver CTVDIGDUAL_V2)
it913x-fe: ADF table value :00
it913x-fe: Crystal Frequency :12000000 Adc Frequency :20250000 ADC X2: 00
it913x-fe: Tuner LNA type :38
usb 2-1: DVB: registering adapter 1 frontend 0 (Digital Dual TV Receiver CTVDIGDUAL_V2_1)...
usb 2-1: dvb_usb_v2: will pass the complete MPEG2 transport stream to the software demuxer
DVB: registering new adapter (Digital Dual TV Receiver CTVDIGDUAL_V2)
it913x-fe: ADF table value :00
it913x-fe: Crystal Frequency :12000000 Adc Frequency :20250000 ADC X2: 00
it913x-fe: Tuner LNA type :38
usb 2-1: DVB: registering adapter 2 frontend 0 (Digital Dual TV Receiver CTVDIGDUAL_V2_2)...
usb 2-1: dvb_usb_v2: 'Digital Dual TV Receiver CTVDIGDUAL_V2' successfully initialized and connected
RC part not tested
Zoran Turalija [Sun, 2 Jun 2013 17:56:33 +0000 (14:56 -0300)]
[media] stb0899: allow minimum symbol rate of 2000000
Looks like product datasheets for tuners containing STB0899 are
suggesting specification for min. symbol rate of 2MS/s.
Some specs found here, all suggesting 2MS/s for min. symbol rate:
Comtech DVBS2-6899
http://comtech.sg1002.myweb.hinet.net/pdf/dvbs2-6899.pdf
TechniSat SkyStar HD2
http://www.scaistar.com/skystar2/skystarhd2.htm
Azurewave AD-SP400
http://www.pulsat.com/products/AzureWave-AD%252dSP400-High-Definition-PC-Card.html
New patch:
This makes minimum symbol rate driver capabilities on par with some
accessible datasheet specifications*, and allows tuning on linux to
transponders that have symbol rate between 2000000-5000000, too.
Patch was tested successfully on Eutelsat 16A transponders that
became reachable with it (2000000 < symbol rate < 5000000):
* DVB/S 12507050 V 2532000 3/4
* DVB/S2 12574000 V 4355000 3/4 8PSK
* DVB/S 12593000 V 2500000 2/3
* DVB/S 12596940 V 2848000 2/3
* DVB/S 12600750 V 2500000 1/2
* DVB/S 12675590 H 4248000 3/4
(*) Datasheet: http://comtech.sg1002.myweb.hinet.net/pdf/dvbs2-6899.pdf
Maximum Symbol Rate
QPSK/LDPC/PCH: 20-30Mbps
8PSK/LDPC/BCH: 10-30Mbps
DVB: 2-45Mbps
^--------- min. symbol rate
Reinhard Nißl [Sun, 2 Jun 2013 17:52:43 +0000 (14:52 -0300)]
[media] stb0899: fix inversion enum values to match usage with CFR
Throughout the zig-zag-implementations, inversion is taken into
account when reading and writing the CFR register, which contains
the derotator frequency. As swapping IQ signals changes the sign
of that register for example, the idea is to compensate that sign
change by multiplying the register value with the inversion enum
value.
The current enum values 0 and 1 for IQ_SWAP_OFF and IQ_SWAP_ON
don't work in the case IQ_SWAP_OFF, due to the multiplication by
zero (I've only found a single device which actually uses
IQ_SWAP_OFF in it's config).
I've changed the enum values to +1 and -1 to accommodate to the
intended usage.
Reinhard Nißl [Sun, 2 Jun 2013 17:52:02 +0000 (14:52 -0300)]
[media] stb0899: enable auto inversion handling unconditionally
It seems that current inversion handling addresses only the signal
routing on the PCB, i. e. IQ signals are either swapped or not.
But when the device is operated in a Satellite Channel Router (SCR)
environment, an additional inversion is required due to the way how
the SCR works. Therefore it makes sense to me to always enable auto
inversion handling and drop the enum value IQ_SWAP_AUTO.
Zoran Turalija [Sun, 2 Jun 2013 17:40:51 +0000 (14:40 -0300)]
[media] stb0899: allow minimum symbol rate of 1000000
This makes minimum symbol rate driver capabilities on par with
windows driver, and allows tuning on linux to transponders that
have symbol rate below 5000000, too.
Patch was tested successfully on Eutelsat 16A transponders that
became reachable with it (1000000 < symbol rate < 5000000):
* DVB/S 12507050 V 2532000 3/4
* DVB/S2 12574000 V 4355000 3/4 8PSK
* DVB/S 12593000 V 2500000 2/3
* DVB/S 12596940 V 2848000 2/3
* DVB/S 12600750 V 2500000 1/2
* DVB/S 12675590 H 4248000 3/4
Reinhard Nißl [Sun, 2 Jun 2013 17:37:06 +0000 (14:37 -0300)]
[media] stb0899: sign extend raw CRL_FREQ value
Contrary to the chip's specs, the register's value is signed, so we
need to sign extend the value before using it in calculations like
when determining the offset frequency.
[media] dib8000: Fix dib8000_set_frontend() never setting ret
drivers/media/dvb-frontends/dib8000.c: In function ‘dib8000_set_frontend’:
drivers/media/dvb-frontends/dib8000.c:3556: warning: ‘ret’ is used uninitialized in this function
Remove the variable and return zero instead.
Maintaining the users count using an atomic variable makes sure that
access to the counter won't be racy, but doesn't serialize access to the
operations protected by the counter. This creates a race condition that
could result in the status URB being submitted multiple times.
Use a mutex to protect the users count and serialize access to the
status start and stop operations.
Joseph Salisbury [Wed, 15 May 2013 20:38:48 +0000 (17:38 -0300)]
[media] uvcvideo: quirk PROBE_DEF for Alienware X51 OmniVision webcam
BugLink: http://bugs.launchpad.net/bugs/1180409
OminiVision webcam 0x05a9:0x2643 needs the same UVC_QUIRK_PROBE_DEF
as other OmniVision models to work properly.
Kamal Mostafa [Mon, 15 Apr 2013 19:01:51 +0000 (16:01 -0300)]
[media] uvcvideo: quirk PROBE_DEF for Dell Studio / OmniVision webcam
BugLink: https://bugs.launchpad.net/bugs/1168430
OminiVision webcam 0x05a9:0x264a (in Dell Studio Hybrid 140g) needs the
same UVC_QUIRK_PROBE_DEF as other OmniVision model to be recognized
consistently.
Philipp Zabel [Thu, 23 May 2013 13:43:01 +0000 (10:43 -0300)]
[media] coda: do not call v4l2_m2m_job_finish from .job_abort
If we just declare the job finished here while the CODA is still
running, the call to v4l2_m2m_ctx_release in coda_release, which
is supposed to wait for a running job to finish, will return
immediately and free memory that the CODA is still using.
Just set the 'aborting' flag and let coda_irq_handler deal with it.
Philipp Zabel [Thu, 23 May 2013 13:43:00 +0000 (10:43 -0300)]
[media] coda: replace completion with mutex
Not only do we need to wait for job completion when we want to
call a command on the CODA in start/stop_streaming, we also need
to make sure that a new job doesn't start before the command
finished.
Use a mutex to lock the coda command handling. On timeout, the
device_run job must be marked as finished and the mutex released,
as otherwise coda_release will block.
Philipp Zabel [Thu, 23 May 2013 13:42:59 +0000 (10:42 -0300)]
[media] coda: add coda_encode_header helper function
In preparation for CODA7541 and CODA960 multi-stream support and for
replacement of the completion with a mutex lock, consolidate the
header encoding in a helper function.
Philipp Zabel [Thu, 23 May 2013 13:42:58 +0000 (10:42 -0300)]
[media] coda: per-product list of codecs instead of list of formats
This patch adds a list of supported codecs per device type which is used to
determine the allowed pixel formats and maximum frame sizes depending on the
possible codecs. It allows frame sizes larger than 720 x 576 on CODA7 and adds
support for the YVU420 (planar YUV 4:2:0 with switched Cb and Cr) format.
Other uncompressed formats that could be added in the future are the chroma
interleaved NV12 and NV21 formats and the multiplanar variants YUV420M,
YVU420M, NV12M, and NV21M.
Further, rawstreamon is renamed to streamon_out and compstreamon is renamed
to streamon_cap in preparation for decoding support. The maximum encoded
frame buffer size is increased to 1 MiB.
Instead of tracking inst_type and codec across S_FMT calls, set the codec
and inst_type in start_streaming.
Philipp Zabel [Tue, 21 May 2013 13:31:25 +0000 (10:31 -0300)]
[media] coda: fix error return value if v4l2_m2m_ctx_init fails
Use ret from the outer scope, instead of redefining it in the
conditional clause. That way the error value reaches the end of
the function as intended.
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@redhat.com>
Lad, Prabhakar [Tue, 14 May 2013 09:45:31 +0000 (06:45 -0300)]
[media] media: i2c: tvp7002: rearrange description of structure members
This patch rearranges the description of field members of
struct tvp7002_config. Also as the all the fields where accepting
a value either 0/1, made the members as bool.
projects / linux-beck.git / log