Convert a nonnegative error return code to a negative one, as returned
elsewhere in the function.
A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)
// <smpl>
(
if@p1 (\(ret < 0\|ret != 0\))
{ ... return ret; }
|
ret@p1 = 0
)
... when != ret = e1
when != &ret
*if(...)
{
... when != ret = e2
when forall
return ret;
}
// </smpl>
Convert a nonnegative error return code to a negative one, as returned
elsewhere in the function.
A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)
// <smpl>
(
if@p1 (\(ret < 0\|ret != 0\))
{ ... return ret; }
|
ret@p1 = 0
)
... when != ret = e1
when != &ret
*if(...)
{
... when != ret = e2
when forall
return ret;
}
// </smpl>
[mchehab@redhat.com: Second hunk removed, as it doesn't apply anymore,
due to a merge conflict] Signed-off-by: Peter Senna Tschudin <peter.senna@gmail.com> Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
devm_kzalloc() has been used to simplify error handling.
While at it, the soc_camera_device_register function has been moved to
save a few lines of code and a variable.
Javier Martin [Mon, 20 Aug 2012 09:49:24 +0000 (06:49 -0300)]
[media] media: mx2_camera: Don't modify non volatile parameters in try_fmt
Signed-off-by: Javier Martin <javier.martin@vista-silicon.com>
[g.liakhovetski@gmx.de: also don't clear pcdev->resizing in .try_fmt()] Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
[media] soc-camera: Use new selection target definitions
Replace the deprecated V4L2_SEL_TGT_*_ACTIVE selection target
names with their new unified counterparts.
Compatibility definitions are already in linux/v4l2-common.h.
Antti Palosaari [Fri, 5 Oct 2012 21:50:12 +0000 (18:50 -0300)]
[media] cxd2820r: silence compiler warning
drivers/media/dvb-frontends/cxd2820r_core.c: In function 'cxd2820r_attach':
drivers/media/dvb-frontends/cxd2820r_core.c:691:10: warning: unused variable 'gpio' [-Wunused-variable]
Arun Kumar K [Thu, 4 Oct 2012 19:14:56 +0000 (16:14 -0300)]
[media] s5p-mfc: Set vfl_dir for encoder
The vfl_dir flag is presently set to VFL_DIR_M2M only for decoder.
The encoder is not working because of this. So adding this flag
to the encoder part also.
Jeongtae Park [Thu, 4 Oct 2012 01:19:11 +0000 (22:19 -0300)]
[media] s5p-mfc: Update MFC v4l2 driver to support MFC6.x
Multi Format Codec 6.x is a hardware video coding acceleration
module present in new Exynos5 SoC series. It is capable of
handling several new video codecs for decoding and encoding.
Arun Kumar K [Thu, 4 Oct 2012 01:19:09 +0000 (22:19 -0300)]
[media] s5p-mfc: Add MFC variant data to device context
MFC variant data replaces various macros used in the driver
which will change in a different version of MFC hardware.
Also does a cleanup of MFC context structure and common files.
Arun Kumar K [Thu, 4 Oct 2012 01:19:08 +0000 (22:19 -0300)]
[media] s5p-mfc: Update MFCv5 driver for callback based architecture
Modifies the driver to use a callback based architecture
for hardware dependent calls. This architecture is suitable
for supporting co-existence with newer versions of MFC hardware.
Arun Kumar K [Thu, 4 Oct 2012 01:19:07 +0000 (22:19 -0300)]
[media] s5p-mfc: Prepare driver for callback based re-architecture
The patch renames hardware specific opr and cmd files to
opr_v5 and cmd_v5 respectively. This is done for accomodating
firmware v6. Also the shared memory management files are removed
and the functionality is added to the opr_v5 file.
Arun Kumar K [Thu, 4 Oct 2012 01:19:05 +0000 (22:19 -0300)]
[media] v4l: Add fourcc definitions for new formats
Add the following new fourcc definitions, for multiplanar YCbCr:
V4L2_PIX_FMT_NV21M, V4L2_PIX_FMT_NV12MT_16X16 and compressed formats:
V4L2_PIX_FMT_H264_MVC, V4L2_PIX_FMT_VP8.
[media] s5p-csis: Allow to specify pixel clock's source through platform data
Depending on the sensor configuration it might be required to adjust
the CSIS's output pixel clock so it is greater than its input pixel
clock, in order to avoid the input data FIFO overflow.
Use platform data to select SCLK_CSIS clock from CMU as a source, rather
than CSI APB clock.
[media] s5p-fimc: Remove unused platform data structure fields
alignment, fixed_phy_vdd and phy_enable fields are now unused
so removed them. The data alignment is now derived directly
from media bus pixel code, phy_enable callback has been replaced
with direct function call and fixed_phy_vdd was dropped in commit 438df3ebe5f0ce408490a777a758d5905f0dd58f
"[media] s5p-csis: Handle all available power supplies".
[media] s5p-csis: Replace phy_enable platform data callback with direct call
The phy_enable callback is common for all Samsung SoC platforms,
replace it with direct function call so the MIPI-CSI2 DPHY control
is also possible on device tree instantiated platforms.
[media] ARM: samsung: Change __s5p_mipi_phy_control() function signature
Replace pdev argument __s5p_mipi_phy_control() helper with plain int
so MIPI-CSIS hardware instance index can be passed directly making
the function usable on platforms instantiated from device tree.
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com> Acked-by: Kukjin Kim <kgene.kim@samsung.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
[media] ARM: samsung: Remove unused fields from FIMC and CSIS platform data
The MIPI-CSI2 bus data alignment is now being derived from the media
bus pixel code, the drivers don't use the corresponding structure
fields, so remove them. Also remove the s5p_csis_phy_enable callback
which is now used directly by s5p-csis driver.
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com> Acked-by: Kukjin Kim <kgene.kim@samsung.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Hans Verkuil [Wed, 19 Sep 2012 13:14:41 +0000 (10:14 -0300)]
[media] v4l2-ioctl.c: handle PREPARE_BUF like QUERYBUF
The core code for PREPARE_BUF didn't take the multiplanar case into account,
which might cause page faults. Handle PREPARE_BUF just like QUERYBUF and
QBUF/DQBUF.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Hans Verkuil [Fri, 28 Sep 2012 09:12:53 +0000 (06:12 -0300)]
[media] videobuf2-core: move plane verification out of __fill_v4l2/vb_buffer
The plane verification should be done before actually queuing or
dequeuing buffers, so move it out of __fill_v4l2_buffer and __fill_vb_buffer
and call it as a separate step.
This also makes it possible to change the return type of __fill_v4l2/vb_buffer
to void.
The dqbuf case took some special care: before removing a buffer from the
done_list you have to verify that the receiving v4l2_buffer has enough room
for all the planes. The number of planes can differ between buffers due to the
fact that buffers for other formats can be prepared using VIDIOC_PREPARE_BUF.
[media] media: davinci: vpif: display: separate out subdev from output
vpif_display relied on a 1-1 mapping of output and subdev. This is not
necessarily the case. Separate the two. So there is a list of subdevs
and a list of outputs. Each output refers to a subdev and has routing
information. An output does not have to have a subdev.
The initial output for each channel is set to the fist output.
Currently missing is support for associating multiple subdevs with
an output.
Hans Verkuil [Thu, 20 Sep 2012 12:06:33 +0000 (09:06 -0300)]
[media] tvp514x: s_routing should just change routing, not try to detect a signal
The s_routing function should not try to detect a signal. It is a really
bad idea to try to detect a valid video signal and return an error if
you can't. Changing input should do just that and nothing more.
Also don't power on the ADCs on s_routing, instead do that on querystd.
Hans Verkuil [Thu, 20 Sep 2012 12:06:31 +0000 (09:06 -0300)]
[media] vpif_display: use a v4l2_subdev pointer to call a subdev
This makes it easier to have outputs without subdevs.
This needs more work. The way the outputs are configured should be identical
to how inputs are configured.
Hans Verkuil [Thu, 20 Sep 2012 12:06:30 +0000 (09:06 -0300)]
[media] vpif_capture: separate subdev from input
vpif_capture relied on a 1-1 mapping of input and subdev. This is not
necessarily the case. Separate the two. So there is a list of subdevs
and a list of inputs. Each input refers to a subdev and has routing
information. An input does not have to have a subdev.
The initial input for each channel is set to the fist input.
Currently missing is support for associating multiple subdevs with
an input.
Hans Verkuil [Thu, 20 Sep 2012 12:06:23 +0000 (09:06 -0300)]
[media] vpif_display: move output_id to channel_obj
The output index does not belong to video_obj, it belongs to
channel_obj. Also rename to output_idx to be consistent with
the input_idx name used in vpif_capture.
Hans Verkuil [Thu, 20 Sep 2012 12:06:22 +0000 (09:06 -0300)]
[media] vpif_capture: move input_idx to channel_obj
input_idx does not belong to video_obj. Move it where it belongs.
Also remove the bogus code in the open() function that suddenly
changes the input to 0 for no reason.
[media] ARM: davinci: da850 evm: Add EVM specific code for VPIF to work
Include the expander settings to select VPIF peripheral on
UI card and add registration call in EVM init. Also add platform
data to configure display and capture devices.
[media] media: davinci: vpif: add check for NULL handler
for da850/omap-l138, there is no need to setup_input_channel_mode()
and set_clock(), to avoid adding dummy code in board file just returning
zero add a check in the driver itself to call the handler only if its
not NULL.
Hans Verkuil [Tue, 4 Sep 2012 13:26:45 +0000 (10:26 -0300)]
[media] fsl-viu: fix compiler warning
drivers/media/platform/fsl-viu.c: In function 'vidioc_s_fbuf':
drivers/media/platform/fsl-viu.c:867:32: warning: initialization discards 'const' qualifier from pointer target type [enabled by default]
This is fall-out from this commit:
commit e6eb28c2207b9397d0ab56e238865a4ee95b7ef9
Author: Hans Verkuil <hans.verkuil@cisco.com>
[media] v4l2: make vidioc_s_fbuf const
[media] media: mt9p031/mt9t001/mt9v032: use V4L2_CID_TEST_PATTERN for test pattern control
V4L2_CID_TEST_PATTERN is now a standard control.
This patch replaces the user defined control for test
pattern to make use of standard control V4L2_CID_TEST_PATTERN.
[media] s5p-fimc: Add support for V4L2_PIX_FMT_S5C_UYVY_JPG fourcc
The V4L2_PIX_FMT_S5C_YUYV_JPG image formats consists of 2 planes, the
first containing interleaved JPEG/YUYV data and the second containing
meta data describing the interleaving method.
The image data is transferred with MIPI-CSI "User Defined Byte-Based
Data 1" type and is captured to memory by FIMC DMA engine.
The meta data is transferred using MIPI-CSI2 "Embedded 8-bit non Image
Data" and it is captured in the MIPI-CSI slave device and copied to
the bridge provided buffer.
To make sure the size of allocated buffers is correct for the subdevs
configuration when VIDIOC_STREAMON ioctl is invoked, an additional
check is added at the video pipeline validation function.
Flag FMT_FLAGS_COMPRESSED indicates the buffer size must be retrieved
from a sensor subdev.
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com> Signed-off-by: Andrzej Hajda <a.hajda@samsung.com> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
[media] s5p-csis: Add support for non-image data packets capture
MIPI-CSI has internal memory mapped buffers for the frame embedded
(non-image) data. There are two buffers, for even and odd frames which
need to be saved after an interrupt is raised. The packet data buffers
size is 4 KiB and there is no status register in the hardware where the
actual non-image data size can be read from. Hence the driver copies
whole packet data buffer into a buffer provided by the FIMC driver.
This will form a separate plane in the user buffer.
When FIMC DMA engine is stopped by the driver due the to user space
not keeping up with buffer de-queuing the MIPI-CSIS will still run,
however it must discard data which is not captured by FIMC. Which
frames are actually capture by MIPI-CSIS is determined by means of
the s_tx_buffer subdev callback. When it is not called after a single
embedded data frame has been captured and copied and before next
embedded data frame interrupt occurrs, subsequent embedded data frames
will be dropped.
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com> Signed-off-by: Andrzej Hajda <a.hajda@samsung.com> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This patch adds definition of the Samsung S5C73M3 camera specific
image format. V4L2_PIX_FMT_S5C_UYVY_JPG is a two-planar format,
the first plane contains interleaved UYVY and JPEG data followed
by meta-data. The second plane contains additional meta-data needed
for extracting JPEG and UYVY data stream from the first plane.
[media] V4L: Add V4L2_MBUS_FMT_S5C_UYVY_JPEG_1X8 media bus format
This patch adds media bus pixel code for the interleaved JPEG/UYVY
image format used by S5C73MX Samsung cameras. This interleaved image
data is transferred on MIPI-CSI2 bus as User Defined Byte-based Data.
It also defines an experimental vendor and device specific media bus
formats section and adds related DocBook documentation.
Add subdev callbacks for setting up parameters of the frame on media bus
that are not exposed to user space directly. This is just an initial,
mostly stub implementation. struct v4l2_mbus_frame_desc is intended
to be extended with sub-structures specific to a particular hardware media
bus. For now these new callbacks are used only to query or specify maximum
size of a compressed or hybrid (container) media bus frame in octets.
[media] V4L: Add s_rx_buffer subdev video operation
The s_rx_buffer callback allows the host to set buffer for a data being
received by the subdev, e.g. non-image frame (meta) data. This callback
can be implemented by subdevice like a MIPI CSI2 receiver, allowing the
host driver to gather additional data into frame buffer passed to user
space.
Hans Verkuil [Wed, 5 Sep 2012 08:10:48 +0000 (05:10 -0300)]
[media] s5p-fimc: fix compiler warning
drivers/media/platform/s5p-fimc/fimc-m2m.c:561:2: warning: passing argument 2 of 'fimc_m2m_try_crop' discards 'const' qualifier from pointer target type [enabled by default]
drivers/media/platform/s5p-fimc/fimc-m2m.c:502:12: note: expected 'struct v4l2_crop *' but argument is of type 'const struct v4l2_crop *'
This is fall-out from this commit:
commit 4f996594ceaf6c3f9bc42b40c40b0f7f87b79c86
Author: Hans Verkuil <hans.verkuil@cisco.com>
[media] v4l2: make vidioc_s_crop const
Hans Verkuil [Wed, 5 Sep 2012 08:10:48 +0000 (05:10 -0300)]
[media] s5p-g2d: fix compiler warning
drivers/media/platform/s5p-g2d/g2d.c:535:2: warning: passing argument 3 of 'vidioc_try_crop' discards 'const' qualifier from pointer target type [enabled by default]
drivers/media/platform/s5p-g2d/g2d.c:510:12: note: expected 'struct v4l2_crop *' but argument is of type 'const struct v4l2_crop *'
This is fall-out from this commit:
commit 4f996594ceaf6c3f9bc42b40c40b0f7f87b79c86
Author: Hans Verkuil <hans.verkuil@cisco.com>
[media] v4l2: make vidioc_s_crop const
[media] technisat-usb2: add a MODULE_DEVICE_TABLE for udev autoload
This patch adds a module-device-table-entry to the
technisat-usb2-driver which will help udev to on-demand load the
driver. This was obviously forgotten during initial commit.
Signed-off-by: Patrick Boettcher <pboettcher@kernellabs.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Lad, Prabhakar [Wed, 3 Oct 2012 05:13:28 +0000 (02:13 -0300)]
[media] media: davinci: vpbe: fix build warning
recent patch with commit id 4f996594ceaf6c3f9bc42b40c40b0f7f87b79c86
which makes vidioc_s_crop const, was causing a following build warning,
vpbe_display.c: In function 'vpbe_display_s_crop':
vpbe_display.c:640: warning: initialization discards qualifiers from pointer target type
This patch fixes the above build warning.
As spotted by Dan Carpenter <dan.carpenter@oracle.com> (thanks!), we have
improperly used an unsigned variable in a calculation that may result in a
negative number. This may cause an unintended underflow if the interface
frequency of the tuner is > approx. 40MHz.
This patch should resolve the issue, following an approach similar to what is
used in af9013.c.
[crope@iki.fi: add Reported-by] Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Hans-Frieder Vogt <hfvogt@gmx.net> Reviewed-by: Antti Palosaari <crope@iki.fi> Signed-off-by: Antti Palosaari <crope@iki.fi> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Lad, Prabhakar [Mon, 1 Oct 2012 11:52:48 +0000 (08:52 -0300)]
[media] media: davinci: vpfe: fix build error
recent patch with commit id 4f996594ceaf6c3f9bc42b40c40b0f7f87b79c86
which makes vidioc_s_crop const, was causing a following build error.
vpfe_capture.c: In function 'vpfe_s_crop':
vpfe_capture.c:1695: error: assignment of read-only location '*crop'
vpfe_capture.c:1706: warning: passing argument 1 of
'ccdc_dev->hw_ops.set_image_window' discards qualifiers from pointer target type
vpfe_capture.c:1706: note: expected 'struct v4l2_rect *' but argument is of
type 'const struct v4l2_rect *'
make[4]: *** [drivers/media/platform/davinci/vpfe_capture.o] Error 1
em28xx: Make all em28xx extensions to be initialized asynchronously
em28xx-dvb, em28xx-alsa and em28xx-ir are typically initialized
asyncrhronously. The exception for it is when those modules
are loaded before em28xx (or before an em28xx card insertion) or
when they're built in.
Make the extentions to always load asynchronously. That allows
having all DVB firmwares loaded synchronously with udev-182.
Antti tested it with the following hardware:
Hauppauge WinTV HVR 930C
MaxMedia UB425-TC
PCTV QuatroStick nano (520e)
Tested-by: Antti Palosaari <crope@iki.fi> Cc: stable@kernel.org # for Kernel 3.6 - please note that driver location has changed Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Michael Krufky [Tue, 2 Oct 2012 14:04:36 +0000 (11:04 -0300)]
[media] tda18271: properly report read errors in tda18271_get_id
Until now, if there is a read error in tda18271_get_id, the driver
reports "Unknown device..." Instead, check the return value of
tda18271_read_regs and display the appropriate error message.
Signed-off-by: Michael Krufky <mkrufky@linuxtv.org> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Michael Krufky [Tue, 2 Oct 2012 02:50:37 +0000 (23:50 -0300)]
[media] tda18271: delay IR & RF calibration until init() if delay_cal is set
if the configuration option 'delay_cal' is set, delay both IR & RF
calibration until init() is called.
both module option 'cal' or configuration option 'rf_cal_on_startup'
will override this delay. it makes no sense to mix 'delay_cal' with
'rf_cal_on_startup' as these options conflict with each other.
Signed-off-by: Michael Krufky <mkrufky@linuxtv.org> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
[media] s5p-tv: Report only multi-plane capabilities in vidioc_querycap
The mixer video node supports only multi-planar API so the driver
should not be setting V4L2_CAP_VIDEO_OUTPUT flags. Fix this and
also switch to device_caps. Additionally fix the VIDIOC_ENUM_FMT
ioctl handler which now works for V4L2_BUF_TYPE_CAPTURE, rather
than expected V4L2_BUF_TYPE_CAPTURE_MPLANE.
[media] s5p-mfc: Fix misplaced return statement in s5p_mfc_suspend()
return statement was wrongly placed before a code that needs to be
executed. Moved the return statement to the end of the function.
Tested suspend/resume on SMDK4412 board using 3.5-rc6 kernel.
[media] exynos-gsc: Add missing static storage class specifiers
Fixes the following warnings:
drivers/media/platform/exynos-gsc/gsc-core.c:313:5: warning:
symbol 'get_plane_info' was not declared. Should it be static?
drivers/media/platform/exynos-gsc/gsc-core.c:746:28: warning:
symbol 'gsc_ctrl_ops' was not declared. Should it be static?
drivers/media/platform/exynos-gsc/gsc-m2m.c:102:5: warning:
symbol 'gsc_fill_addr' was not declared. Should it be static?
drivers/media/platform/exynos-gsc/gsc-m2m.c:252:16: warning:
symbol 'gsc_m2m_qops' was not declared. Should it be static?
[media] s5k6aa: Fix possible NULL pointer dereference
Make sure __s5k6aa_get_crop_rect() always returns valid pointer,
as it is assumed at the callers.
crop->which is already validated when subdev set_crop and get_crop
callbacks are called from within the v4l2-core. If it ever happens
the crop operations are called directly for some reason in kernel
space, with incorrect crop->which argument, just log it with WARN
and return reference to the TRY crop.
Andrzej Hajda [Tue, 14 Aug 2012 09:13:40 +0000 (06:13 -0300)]
[media] s5p-mfc: optimized code related to working contextes
All code setting/clearing working context bits has been moved
to separate functions. set_bit/clear_bit have been replaced by
non-atomic variants - variable is already guarded by spin_lock.
Signed-off-by: Andrzej Hajda <a.hajda@samsung.com> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com> Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Andrzej Hajda [Tue, 21 Aug 2012 11:05:32 +0000 (08:05 -0300)]
[media] s5p-mfc: added support for end of stream handling in MFC encoder
s5p-mfc encoder after receiving V4L2_ENC_CMD_STOP command
will instruct MFC device to release all encoded frames.
After dequeuing last encoded frame driver will generate
V4L2_EVENT_EOS event.
Signed-off-by: Andrzej Hajda <a.hajda@samsung.com> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com> Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
[media] s5p-fimc: Keep local copy of sensors platform data
Use a copy of sensor platform data structure, rather than referencing
external platform data from the driver. This allows to fill the local
copy with values parsed from device tree when needed.
Add hardware event/error counters which can be dumped into the kernel
log through VIDIOC_LOG_STATUS ioctl. The counters are reset in each
s_stream(1) call. Any errors are logged after streaming is turned off.