platform/kernel/linux-rpi.git
4 years agomedia: vivid: fix incorrect PA assignment to HDMI outputs
Hans Verkuil [Fri, 20 Mar 2020 11:13:37 +0000 (12:13 +0100)]
media: vivid: fix incorrect PA assignment to HDMI outputs

The initial physical address was one too low for the outputs.
E.g. if 1.0.0.0 was expected, then it was set to 0.0.0.0, and
2.0.0.0 became 1.0.0.0.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Cc: Johan Korsnes <johan.korsnes@gmail.com>
Fixes: 4ee895e71abb ("media: vivid: reorder CEC allocation and control set-up")
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: hantro: Add linux-rockchip mailing list to MAINTAINERS
Ezequiel Garcia [Wed, 18 Mar 2020 13:21:08 +0000 (14:21 +0100)]
media: hantro: Add linux-rockchip mailing list to MAINTAINERS

The linux-rockchip mailing list is relevant for the
Hantro driver, given this support the VPU present
in Rockchip SoCs.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Reviewed-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: cedrus: h264: Fix 4K decoding on H6
Jernej Skrabec [Sun, 15 Mar 2020 09:54:46 +0000 (10:54 +0100)]
media: cedrus: h264: Fix 4K decoding on H6

Due to unknown reason, H6 needs larger intraprediction buffer for 4K
videos than other SoCs. This was discovered by playing 4096x2304 video,
which is maximum what H6 VPU is supposed to support.

Fixes: 03e612e701a6 ("media: cedrus: Fix H264 4k support")
Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: siano: Use scnprintf() for avoiding potential buffer overflow
Takashi Iwai [Wed, 11 Mar 2020 07:46:38 +0000 (08:46 +0100)]
media: siano: Use scnprintf() for avoiding potential buffer overflow

Since snprintf() returns the would-be-output size instead of the
actual output size, the succeeding calls may go beyond the given
buffer limit.  Fix it by replacing with scnprintf().

Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: rc: Use scnprintf() for avoiding potential buffer overflow
Takashi Iwai [Wed, 11 Mar 2020 07:47:06 +0000 (08:47 +0100)]
media: rc: Use scnprintf() for avoiding potential buffer overflow

Since snprintf() returns the would-be-output size instead of the
actual output size, the succeeding calls may go beyond the given
buffer limit.  Fix it by replacing with scnprintf().

Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: create new struct for channel parameters
Michael Tretter [Mon, 16 Mar 2020 15:26:38 +0000 (16:26 +0100)]
media: allegro: create new struct for channel parameters

Add a new struct for the channel parameters that is contained in the
CREATE_CHANNEL message. This is in preparation for newer firmwares that
pass the channel parameters in a dedicated buffer instead of embedding
the parameters into the CREATE_CHANNEL message.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: move mail definitions to separate file
Michael Tretter [Mon, 16 Mar 2020 15:26:37 +0000 (16:26 +0100)]
media: allegro: move mail definitions to separate file

Move the mail definitions from the driver core to a dedicated file.

The mails that are exchanged between driver and firmware are not stable
across firmware versions. This is in preparation to make the driver able
to handle multiple firmware version by having dedicated code for
handling mails.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: pass buffers through firmware
Michael Tretter [Mon, 16 Mar 2020 15:26:36 +0000 (16:26 +0100)]
media: allegro: pass buffers through firmware

As we know which buffers are processed by the codec from the address in
the ENCODE_FRAME response, we can queue multiple buffers in the firmware
and retrieve the buffer from the response of the firmware. This enables
the firmware to use the internal scheduling the codec and avoids round
trips through the driver when fetching the next frame.

Remove buffers that have been passed to the firmware from the m2m buffer
queue and put them into a shadow queue for tracking the buffer in the
driver. When we receive a ENCODE_FRAME response from the firmware, get
the buffer from the shadow queue and finish the buffer.

Furthermore, it is necessary to finish the job straight after passing
the buffer to the firmware to allow the V4L2 framework to send further
buffers to the driver.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: verify source and destination buffer in VCU response
Michael Tretter [Mon, 16 Mar 2020 15:26:35 +0000 (16:26 +0100)]
media: allegro: verify source and destination buffer in VCU response

The PUT_STREAM_BUFFER and ENCODE_FRAME request have fields that allow to
pass arbitrary 64 bit values through the firmware to the ENCODE_FRAME
response. Use these values to verify that the buffers when finishing the
frame are actually the same buffers that have been sent for encoding a
frame.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: handle dependency of bitrate and bitrate_peak
Michael Tretter [Mon, 16 Mar 2020 15:26:34 +0000 (16:26 +0100)]
media: allegro: handle dependency of bitrate and bitrate_peak

The peak bitrate must not be smaller than the configured bitrate. Update
the other control whenever one of the controls changes to reflect this
dependency.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: read bitrate mode directly from control
Michael Tretter [Mon, 16 Mar 2020 15:26:33 +0000 (16:26 +0100)]
media: allegro: read bitrate mode directly from control

There is no need to copy the bitrate mode to a field in the channel and
the value can be read directly from the control.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: make QP configurable
Michael Tretter [Mon, 16 Mar 2020 15:26:32 +0000 (16:26 +0100)]
media: allegro: make QP configurable

The V4L2_CID_MPEG_VIDEO_FRAME_RC_ENABLE control allows to enable/disable
rate control on a channel. When rate control is disabled, the driver
shall use constant QP, which are set by the application. Also implement
the controls for configuring the QP.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: make frame rate configurable
Michael Tretter [Mon, 16 Mar 2020 15:26:31 +0000 (16:26 +0100)]
media: allegro: make frame rate configurable

The allegro dvt codec adjust the encoding speed according to a
configured frame rate. Furthermore, the frame rate is written into the
coded stream.

Ensure that the coded video data has the correct frame rate by
implementing s_parm for setting the frame rate from user space.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: skip filler data if possible
Michael Tretter [Mon, 16 Mar 2020 15:26:30 +0000 (16:26 +0100)]
media: allegro: skip filler data if possible

The driver instructs the firmware to leave some space space in front of
the coded frame data for SPS/PPS data. If the driver receives an IDR, it
writes the SPS/PPS into that free space and fills the rest with filler
data. However, if there is no additional data, the driver can use the
plane offset to skip this space instead of adding filler data.

As the size of the SPS/PPS is only available after writing it, keep the
filler data between the SPS/PPS and the coded frame data.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: warn if response message has an unexpected size
Michael Tretter [Mon, 16 Mar 2020 15:26:29 +0000 (16:26 +0100)]
media: allegro: warn if response message has an unexpected size

The driver uses structs to parse the responses from the VCU and expects
a certain size of the responses. However, the size and format of the
mails is not stable across firmware versions. Therefore, print a warning
if the size does not match the expected size to warn the user that
strange things might happen.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: extract mcu and codec address calculation
Michael Tretter [Mon, 16 Mar 2020 15:26:28 +0000 (16:26 +0100)]
media: allegro: extract mcu and codec address calculation

The mcu and the codec use 32 bit addresses which are sent via the
firmware messages. Add helper functions for this address calculation to
make it obvious which address is used in the message.

As the mcu and the codec have a limited address space, print warnings if
the addresses are outside the respective address space.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: fix reset if WAKEUP has not been set properly
Michael Tretter [Mon, 16 Mar 2020 15:26:27 +0000 (16:26 +0100)]
media: allegro: fix reset if WAKEUP has not been set properly

The Zynq UltraScale+ Devices Register Reference states that the WAKEUP
bit "should be set to 0 after the MCU sleep status bit gets back to 0."
If this is not done, the mcu is not going to sleep on reset and fail the
reset.

Set WAKEUP to 0 before triggering a reset to make sure that the reset is
successful.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: fix calculation of CPB size
Michael Tretter [Mon, 16 Mar 2020 15:26:26 +0000 (16:26 +0100)]
media: allegro: fix calculation of CPB size

The cpb_size is given in kilobytes, but the bitrate is given in bits per
second. Therefore, the calculation of the initial removal delay and the
cpb size for the firmware were wrong.

Convert the bitrate to kilobytes before calculating the cpb size in 90
kHz units for sending it to the firmware. Also reuse the result for the
initial removal delay, to make it obvious that we are setting the
initial removal delay to the maximum value.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: start a GOP with an IDR frame
Michael Tretter [Mon, 16 Mar 2020 15:26:25 +0000 (16:26 +0100)]
media: allegro: start a GOP with an IDR frame

When creating a channel, freq_idr defines the number of frames between
IDR frames in the coded stream. In V4L2, the period between IDR frames
shall be taken from the GOP_SIZE control.

Set the IDR frame frequency equal to the GOP size and let every GOP
start with an IDR frame.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: remove unknown39 field from create_channel
Michael Tretter [Mon, 16 Mar 2020 15:26:24 +0000 (16:26 +0100)]
media: allegro: remove unknown39 field from create_channel

The subframe_latency and lda_control_mode fields directly follow
freq_golden_ref field and there is no unknown field in between. The
unknown field it at the end of the message.

Reorder the fields accordingly. This further allows to drop the hard
coded value from the lda_control_mode field and set the mode to 0.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: fix type of gop_length in channel_create message
Michael Tretter [Mon, 16 Mar 2020 15:26:23 +0000 (16:26 +0100)]
media: allegro: fix type of gop_length in channel_create message

The gop_length field is actually only u16 and there are two more u8
fields in the message:

- the number of consecutive b-frames
- frequency of golden frames

Fix the message and thus fix the configuration of the GOP length.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: fail encoding only on actual errors
Michael Tretter [Mon, 16 Mar 2020 15:26:22 +0000 (16:26 +0100)]
media: allegro: fail encoding only on actual errors

Only negative values are actual errors and positive values are used for
warnings. Warnings should not fail the encoding process.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: allegro: print message on mcu error
Michael Tretter [Mon, 16 Mar 2020 15:26:21 +0000 (16:26 +0100)]
media: allegro: print message on mcu error

The codec firmware uses error codes to report errors during the
configuration of a channel or while encoding a frame. Translate them
into human readable strings for debugging.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: staging: allegro: fix broken registration of controls
Michael Tretter [Thu, 19 Mar 2020 09:16:30 +0000 (10:16 +0100)]
media: staging: allegro: fix broken registration of controls

Since commit cc62c74749a3 ("media: allegro: add missed checks in
allegro_open()") the allegro device does provide v4l2 controls to user
space anymore. The reason is that v4l2_fh_init() initializes
fh->ctrl_handler to vdev->ctrl_handler, which invalidates the previous
driver override of the ctrl_handler.

Therefore, v4l2_fh_init() must be called before the driver overrides the
fh->ctrl_handler with its own handler.

Move the initialization of the fh back to the top, as the initialization
does not does not need to be reverted on errors, but it is enough to
free the channel.

Fixes: cc62c74749a3 ("media: allegro: add missed checks in allegro_open()")
Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: cec-notifier: make cec_notifier_get_conn() static
Hans Verkuil [Mon, 16 Mar 2020 09:36:48 +0000 (10:36 +0100)]
media: cec-notifier: make cec_notifier_get_conn() static

This function is no longer used by other drivers, so it can be
made static.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: cec-notifier: rename conn_name to port_name
Hans Verkuil [Mon, 16 Mar 2020 09:36:47 +0000 (10:36 +0100)]
media: cec-notifier: rename conn_name to port_name

This argument refers to a stable name for an HDMI port, mostly i915
(ACPI) specific. Since we'll be introducing a more generic 'name' argument
as well later, rename this now to avoid confusion.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: am437x-vpfe: Make use of to_vpfe() to get a pointer to vpfe_device
Lad Prabhakar [Sun, 15 Mar 2020 11:41:46 +0000 (12:41 +0100)]
media: am437x-vpfe: Make use of to_vpfe() to get a pointer to vpfe_device

This patch makes use of to_vpfe() to get a pointer to vpfe_device

Signed-off-by: Lad Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: v4l: Add 1X14 14-bit greyscale media bus code definition
Daniel Glöckner [Mon, 24 Feb 2020 17:52:22 +0000 (18:52 +0100)]
media: v4l: Add 1X14 14-bit greyscale media bus code definition

The code is called MEDIA_BUS_FMT_Y14_1X14 and behaves just like
MEDIA_BUS_FMT_Y12_1X12 with two more bits.

Signed-off-by: Daniel Glöckner <dg@emlix.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: v4l: Add 14-bit raw greyscale pixel format
Daniel Glöckner [Mon, 24 Feb 2020 17:52:21 +0000 (18:52 +0100)]
media: v4l: Add 14-bit raw greyscale pixel format

The new format is called V4L2_PIX_FMT_Y14. Like V4L2_PIX_FMT_Y10 and
V4L2_PIX_FMT_Y12 it is stored in two bytes per pixel but has only two
unused bits at the top.

Signed-off-by: Daniel Glöckner <dg@emlix.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: v4l: Add 14-bit raw bayer pixel formats
Sakari Ailus [Mon, 24 Feb 2020 17:52:20 +0000 (18:52 +0100)]
media: v4l: Add 14-bit raw bayer pixel formats

The formats added by this patch are:

V4L2_PIX_FMT_SBGGR14
V4L2_PIX_FMT_SGBRG14
V4L2_PIX_FMT_SGRBG14
V4L2_PIX_FMT_SRGGB14

Signed-off-by: Jouni Ukkonen <jouni.ukkonen@intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
[dg@emlix.com: rebased onto current media_tree]
Signed-off-by: Daniel Glöckner <dg@emlix.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agodevice property: Export fwnode_get_name()
Heikki Krogerus [Mon, 2 Mar 2020 13:53:51 +0000 (16:53 +0300)]
device property: Export fwnode_get_name()

This makes it possible to take advantage of the function in
the device drivers.

Signed-off-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://lore.kernel.org/r/20200302135353.56659-8-heikki.krogerus@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
4 years agomedia: imx: imx7-media-csi: Support clamping Y10 and Y12 to Y8
Laurent Pinchart [Tue, 10 Mar 2020 16:06:33 +0000 (17:06 +0100)]
media: imx: imx7-media-csi: Support clamping Y10 and Y12 to Y8

10-bit and 12-bit greyscale input data to the CSI can be written as
8-bit data to memory. Support this.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: imx: imx7-media-csi: Add Y10 and Y12 formats support
Laurent Pinchart [Tue, 10 Mar 2020 16:06:32 +0000 (17:06 +0100)]
media: imx: imx7-media-csi: Add Y10 and Y12 formats support

Support capturing the 10- and 12-bit greyscale formats.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: imx: imx6-media-csi: Support clamping Y10 and Y12 to Y8
Laurent Pinchart [Tue, 10 Mar 2020 16:06:31 +0000 (17:06 +0100)]
media: imx: imx6-media-csi: Support clamping Y10 and Y12 to Y8

10-bit and 12-bit greyscale input data to the CSI can be written as
8-bit data to memory. Support this.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Steve Longerbeam <slongerbeam@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: imx: imx6-media-csi: Replace Y16 with Y10 and Y12
Laurent Pinchart [Tue, 10 Mar 2020 16:06:30 +0000 (17:06 +0100)]
media: imx: imx6-media-csi: Replace Y16 with Y10 and Y12

The driver doesn't really support V4L2_PIX_FMT_Y16, as there's no 16-bit
greyscale media bus code defined by the kernel. It (ab)uses the format
to capture 10-bit and 12-bit greyscale formats. Fix it to properly
support MEDIA_BUS_FMT_Y10_1X10 and MEDIA_BUS_FMT_Y12_1X12 instead.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Acked-by: Steve Longerbeam <slongerbeam@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: imx: imx7-media-csi: Fix video field handling
Laurent Pinchart [Tue, 10 Mar 2020 16:06:29 +0000 (17:06 +0100)]
media: imx: imx7-media-csi: Fix video field handling

Commit 4791bd7d6adc ("media: imx: Try colorimetry at both sink and
source pads") reworked the way that formats are set on the sink pad of
the CSI subdevice, and accidentally removed video field handling.
Restore it by defaulting to V4L2_FIELD_NONE if the field value isn't
supported, with the only two supported value being V4L2_FIELD_NONE and
V4L2_FIELD_INTERLACED.

Fixes: 4791bd7d6adc ("media: imx: Try colorimetry at both sink and source pads")
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: imx: imx7-media-csi: Remove unneeded register read
Laurent Pinchart [Tue, 10 Mar 2020 16:06:28 +0000 (17:06 +0100)]
media: imx: imx7-media-csi: Remove unneeded register read

The imx7_csi_dma_reflash() function starts by reading the unrelated
register CSI_CSICR18 to then overwrite the read value with a read from
register CSI_CSICR3. This is likely due to a bad copy&paste, and is not
needed. Remove the extraneous read from register CSI_CSICR18.

Fixes: 9e5fa4e1e5b5b ("media: imx7-media-csi: Use u32 for storing register reads")
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: imx: imx7-mipi-csis: Filter on CSI-2 data type
Laurent Pinchart [Tue, 10 Mar 2020 16:06:27 +0000 (17:06 +0100)]
media: imx: imx7-mipi-csis: Filter on CSI-2 data type

Enable filtering based on CSI-2 data type. Otherwise sources that send
multiple data types (such as sensors that send embedded data) will not
be properly supported. They will both result in additional non-image
lines being captured, and in the receiver being confused as it doesn't
resynchronise on vsync.

Signed-off-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>
4 years agomedia: imx: imx7_mipi_csis: Add greyscale formats support
Laurent Pinchart [Tue, 10 Mar 2020 16:06:26 +0000 (17:06 +0100)]
media: imx: imx7_mipi_csis: Add greyscale formats support

Add support for the 8-, 10- and 12-bit greyscale media bus formats, and
map them to the CSI-2 RAW8, RAW10 and RAW12 formats respectively.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: imx: imx7_mipi_csis: Print the RESOL_CH0 register
Laurent Pinchart [Tue, 10 Mar 2020 16:06:25 +0000 (17:06 +0100)]
media: imx: imx7_mipi_csis: Print the RESOL_CH0 register

Add the RESOL_CH0 register to the list of registers printed through the
debugfs debug infrastructure for the driver, as it can be useful to
verify proper configuration of the CSI-2 receiver.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: imx: imx7_mipi_csis: Power off the source when stopping streaming
Laurent Pinchart [Tue, 10 Mar 2020 16:06:24 +0000 (17:06 +0100)]
media: imx: imx7_mipi_csis: Power off the source when stopping streaming

The .s_stream() implementation incorrectly powers on the source when
stopping the stream. Power it off instead.

Fixes: 7807063b862b ("media: staging/imx7: add MIPI CSI-2 receiver subdev for i.MX7")
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: mtk-mdp: Check return value of of_clk_get
Matthias Brugger [Thu, 13 Feb 2020 20:19:46 +0000 (21:19 +0100)]
media: mtk-mdp: Check return value of of_clk_get

Check the return value of of_clk_get and print an error
message if not EPROBE_DEFER.

Signed-off-by: Matthias Brugger <mbrugger@suse.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: platform: stm32: don't print an error on probe deferral
Etienne Carriere [Tue, 21 Jan 2020 10:22:49 +0000 (11:22 +0100)]
media: platform: stm32: don't print an error on probe deferral

Change stm32-cec driver to not print an error message when the
device probe operation is deferred.

Signed-off-by: Etienne Carriere <etienne.carriere@st.com>
Reviewed-by: Philippe Cornu <philippe.cornu@st.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: platform: stm32: defer probe for auxiliary clock
Etienne Carriere [Tue, 21 Jan 2020 10:22:32 +0000 (11:22 +0100)]
media: platform: stm32: defer probe for auxiliary clock

Change stm32-cec driver to defer probe when auxiliary clock
"hdmi-cec" is registered in the system not has not been probed yet.

Signed-off-by: Etienne Carriere <etienne.carriere@st.com>
Reviewed-by: Philippe Cornu <philippe.cornu@st.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: xirlink_cit: add missing descriptor sanity checks
Johan Hovold [Fri, 3 Jan 2020 16:35:11 +0000 (17:35 +0100)]
media: xirlink_cit: add missing descriptor sanity checks

Make sure to check that we have two alternate settings and at least one
endpoint before accessing the second altsetting structure and
dereferencing the endpoint arrays.

This specifically avoids dereferencing NULL-pointers or corrupting
memory when a device does not have the expected descriptors.

Note that the sanity check in cit_get_packet_size() is not redundant as
the driver is mixing looking up altsettings by index and by number,
which may not coincide.

Fixes: 659fefa0eb17 ("V4L/DVB: gspca_xirlink_cit: Add support for camera with a bcd version of 0.01")
Fixes: 59f8b0bf3c12 ("V4L/DVB: gspca_xirlink_cit: support bandwidth changing for devices with 1 alt setting")
Cc: stable <stable@vger.kernel.org> # 2.6.37
Cc: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: stv06xx: add missing descriptor sanity checks
Johan Hovold [Fri, 3 Jan 2020 16:35:10 +0000 (17:35 +0100)]
media: stv06xx: add missing descriptor sanity checks

Make sure to check that we have two alternate settings and at least one
endpoint before accessing the second altsetting structure and
dereferencing the endpoint arrays.

This specifically avoids dereferencing NULL-pointers or corrupting
memory when a device does not have the expected descriptors.

Note that the sanity checks in stv06xx_start() and pb0100_start() are
not redundant as the driver is mixing looking up altsettings by index
and by number, which may not coincide.

Fixes: 8668d504d72c ("V4L/DVB (12082): gspca_stv06xx: Add support for st6422 bridge and sensor")
Fixes: c0b33bdc5b8d ("[media] gspca-stv06xx: support bandwidth changing")
Cc: stable <stable@vger.kernel.org> # 2.6.31
Cc: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: ov519: add missing endpoint sanity checks
Johan Hovold [Fri, 3 Jan 2020 16:35:09 +0000 (17:35 +0100)]
media: ov519: add missing endpoint sanity checks

Make sure to check that we have at least one endpoint before accessing
the endpoint array to avoid dereferencing a NULL-pointer on stream
start.

Note that these sanity checks are not redundant as the driver is mixing
looking up altsettings by index and by number, which need not coincide.

Fixes: 1876bb923c98 ("V4L/DVB (12079): gspca_ov519: add support for the ov511 bridge")
Fixes: b282d87332f5 ("V4L/DVB (12080): gspca_ov519: Fix ov518+ with OV7620AE (Trust spacecam 320)")
Cc: stable <stable@vger.kernel.org> # 2.6.31
Cc: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: i2c: s5c73m3: Fix number in auto focus cluster
Sebastian Gross [Thu, 5 Dec 2019 10:31:08 +0000 (11:31 +0100)]
media: i2c: s5c73m3: Fix number in auto focus cluster

As of `s5c73m3.h` the auto focus cluster in `s5c73m3_ctrls` has 5 and not 6
controls.

Signed-off-by: Sebastian Gross <sebastian.gross@emlix.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: videobuf2-core: fix dprintk level
Alexandre Courbot [Tue, 26 Nov 2019 08:12:17 +0000 (09:12 +0100)]
media: videobuf2-core: fix dprintk level

This dprintk is supposed to be informational, not an error. Set it to
the same level as the other messages related to memory allocations so
the kernel log does not get filled by messages during normal operation.

Signed-off-by: Alexandre Courbot <acourbot@chromium.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: imx214: Remove redundant code
Ricardo Ribalda Delgado [Thu, 24 Oct 2019 13:48:32 +0000 (15:48 +0200)]
media: imx214: Remove redundant code

The pad is now checked by v4l2_subdev_call_wappers

call_g_frame_interval-> check_frame_interval-> check_pad

Signed-off-by: Ricardo Ribalda Delgado <ribalda@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: i2c: ov5695: Fix power on and off sequences
Dongchun Zhu [Wed, 11 Mar 2020 10:47:28 +0000 (11:47 +0100)]
media: i2c: ov5695: Fix power on and off sequences

From the measured hardware signal, OV5695 reset pin goes high for a
short period of time during boot-up. From the sensor specification, the
reset pin is active low and the DT binding defines the pin as active
low, which means that the values set by the driver are inverted and thus
the value requested in probe ends up high.

Fix it by changing probe to request the reset GPIO initialized to high,
which makes the initial state of the physical signal low.

In addition, DOVDD rising must occur before DVDD rising from spec., but
regulator_bulk_enable() API enables all the regulators asynchronously.
Use an explicit loops of regulator_enable() instead.

For power off sequence, it is required that DVDD falls first. Given the
bulk API does not give any guarantee about the order of regulators,
change the driver to use regulator_disable() instead.

The sensor also requires a delay between reset high and first I2C
transaction, which was assumed to be 8192 XVCLK cycles, but 1ms is
recommended by the vendor. Fix this as well.

Signed-off-by: Dongchun Zhu <dongchun.zhu@mediatek.com>
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: i2c: imx219: Add support for cropped 640x480 resolution
Lad Prabhakar [Tue, 10 Mar 2020 13:17:09 +0000 (14:17 +0100)]
media: i2c: imx219: Add support for cropped 640x480 resolution

This patch adds mode table entry for capturing cropped 640x480 resolution

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Acked-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: i2c: imx219: Add support for RAW8 bit bayer format
Lad Prabhakar [Tue, 10 Mar 2020 13:17:08 +0000 (14:17 +0100)]
media: i2c: imx219: Add support for RAW8 bit bayer format

IMX219 sensor is capable for RAW8/RAW10 modes. This commit adds support
for RAW8 bayer format.

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: i2c: imx219: Fix power sequence
Lad Prabhakar [Tue, 10 Mar 2020 13:17:07 +0000 (14:17 +0100)]
media: i2c: imx219: Fix power sequence

When supporting Rpi Camera v2 Module on the RZ/G2E, found the driver had
some issues with rcar mipi-csi driver. The sensor never entered into LP-11
state.

The powerup sequence in the datasheet[1] shows the sensor entering into
LP-11 in streaming mode, so to fix this issue transitions are performed
from "streaming -> standby" in the probe() after power up.

With this commit the sensor is able to enter LP-11 mode during power up,
as expected by some CSI-2 controllers.

[1] https://publiclab.org/system/images/photos/000/023/294/original/
RASPBERRY_PI_CAMERA_V2_DATASHEET_IMX219PQH5_7.0.0_Datasheet_XXX.PDF

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Acked-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: MAINTAINERS: adjust entry to refactoring smiapp.h
Lukas Bulwahn [Sat, 29 Feb 2020 20:15:31 +0000 (21:15 +0100)]
media: MAINTAINERS: adjust entry to refactoring smiapp.h

Commit 7cbeb2e1bf0c ("media: smiapp: Move definitions under driver
directory") moved the definitions in include/media/i2c/smiapp.h into other
files and removed include/media/i2c/smiapp.h.

Since then, ./scripts/get_maintainer.pl --self-test complains:

  warning: no file matches F: include/media/i2c/smiapp.h

Remove MAINTAINER file entry to this removed file and while at it, update
Sakari's email address in this section.

Signed-off-by: Lukas Bulwahn <lukas.bulwahn@gmail.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: mailmap: Add entry for my @iki.fi address
Sakari Ailus [Mon, 2 Mar 2020 10:43:28 +0000 (11:43 +0100)]
media: mailmap: Add entry for my @iki.fi address

Add entry for my @iki.fi e-mail address to the mailmap.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: staging: imgu: make imgu work on low frequency for low input
Bingbu Cao [Fri, 16 Aug 2019 10:20:48 +0000 (12:20 +0200)]
media: staging: imgu: make imgu work on low frequency for low input

Currently, imgu is working on 450MHz for all cases, however
in some cases (input frame less than 2048x1152), the imgu
did not need work in high frequency.
This patch make imgu work on 200MHz if the imgu input
frame is less than 2048x1152 to save power.

Signed-off-by: Bingbu Cao <bingbu.cao@intel.com>
Reviewed-by: Tomasz Figa <tfiga@chromium.org>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: Fix build failure due to missing select REGMAP_I2C
Ian Kumlien [Wed, 26 Feb 2020 14:34:05 +0000 (15:34 +0100)]
media: Fix build failure due to missing select REGMAP_I2C

While upgrading from 5.5.2 -> 5.5.6 I was surprised by:
ld: drivers/media/i2c/tvp5150.o: in function `tvp5150_probe':
tvp5150.c:(.text+0x11ac): undefined reference to
`__devm_regmap_init_i2c'
make: *** [Makefile:1078: vmlinux] Error 1

The fix was quick enough, make VIDEO_TVP5150 select REGMAP_I2C
And a quick grep showed that it was needed by more targets.

Signed-off-by: Ian Kumlien <ian.kumlien@gmail.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: tvp5150: make debug output more readable
Marco Felsch [Thu, 12 Mar 2020 10:31:56 +0000 (11:31 +0100)]
media: tvp5150: make debug output more readable

The debug output for tvp5150_selmux() isn't really intuitive. Register
values are printed decimal formatted and the input/output driver states
are printed as enum. Even more the "normal" output enum mapps to zero so
a active output will printing output=0 and a inactive output=1.

Change this by brinting the register values hex formatted and the states
as more readable string.

Signed-off-by: Marco Felsch <m.felsch@pengutronix.de>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: tvp5150: add support to limit sdtv standards
Marco Felsch [Thu, 12 Mar 2020 10:31:55 +0000 (11:31 +0100)]
media: tvp5150: add support to limit sdtv standards

The tvp5150 accepts NTSC(M,J,4.43), PAL (B,D,G,H,I,M,N) and SECAM video
data and is able to auto-detect the input signal. The auto-detection
does not work if the connector does not receive an input signal and the
tvp5150 might not be configured correctly. This misconfiguration leads
into wrong decoded video streams if the tvp5150 gets powered on before
the video signal is present.

Limit the supported sdtv standards according to the actual selected
connector to avoid a misconfiguration.

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>
4 years agomedia: dt-bindings: tvp5150: add optional sdtv standards documentation
Marco Felsch [Thu, 12 Mar 2020 10:31:54 +0000 (11:31 +0100)]
media: dt-bindings: tvp5150: add optional sdtv standards documentation

Document the optional binding to limit the possible sdtv standards on the
input connectors.

Signed-off-by: Marco Felsch <m.felsch@pengutronix.de>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: dt-bindings: tvp5150: cleanup bindings stlye
Marco Felsch [Thu, 12 Mar 2020 10:31:53 +0000 (11:31 +0100)]
media: dt-bindings: tvp5150: cleanup bindings stlye

Use underlines to highlight optional and required properties. This is
quite common for all bindings. Align descriptions and start sentence
with uppercase letter. Also reword the usage of the required
endpoint properties for the output port in case BT.656 should be used.

Signed-off-by: Marco Felsch <m.felsch@pengutronix.de>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: tvp5150: add subdev open/close callbacks
Marco Felsch [Thu, 12 Mar 2020 10:31:52 +0000 (11:31 +0100)]
media: tvp5150: add subdev open/close callbacks

Bring the device into a working state upon a open/close call. Currently
this involves only the interrupt enable/disable process but can be
extended in the future.

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>
4 years agomedia: tvp5150: add v4l2-event support
Marco Felsch [Thu, 12 Mar 2020 10:31:51 +0000 (11:31 +0100)]
media: tvp5150: add v4l2-event support

Currently the driver notifies internal subdevs if the signal is locked
or not. This information is also useful for userpace applications e.g. to
switch to another input device upon a signal lost event.

This commit adds the support for the userspace to subscribe to the
V4L2_EVENT_SOURCE_CHANGE and V4L2_EVENT_CTRL events.

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>
4 years agomedia: tvp5150: move irq en-/disable into runtime-pm ops
Marco Felsch [Thu, 12 Mar 2020 10:31:50 +0000 (11:31 +0100)]
media: tvp5150: move irq en-/disable into runtime-pm ops

As documented in [1] the runtime-pm ops are used to set the device into
a fully 'workable' state. Therefore it can be used to enable or disable
the irqs.

[1] https://www.kernel.org/doc/html/latest/power/runtime_pm.html

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>
4 years agomedia: tvp5150: initialize subdev before parsing device tree
Michael Tretter [Thu, 12 Mar 2020 10:31:49 +0000 (11:31 +0100)]
media: tvp5150: initialize subdev before parsing device tree

There are several debug prints in the tvp5150_parse_dt() function, which
do not print the prefix, because the v4l2_subdev is not initialized, yet.

Initialize the v4l2_subdev before parsing the device tree to fix the
debug messages.

Signed-off-by: Michael Tretter <m.tretter@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>
4 years agomedia: tvp5150: add FORMAT_TRY support for get/set selection handlers
Marco Felsch [Thu, 12 Mar 2020 10:31:48 +0000 (11:31 +0100)]
media: tvp5150: add FORMAT_TRY support for get/set selection handlers

Since commit 10d5509c8d50 ("[media] v4l2: remove g/s_crop from video ops")
the 'which' field for set/get_selection must be FORMAT_ACTIVE. There is
no way to try different selections. The patch adds a helper function to
select the correct selection memory space (sub-device file handle or
driver state) which will be set/returned.

The selection rectangle is updated if the format is FORMAT_ACTIVE and
the rectangle position and/or size differs from the current set
rectangle.

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>
4 years agomedia: tvp5150: fix set_selection rectangle handling
Marco Felsch [Thu, 12 Mar 2020 10:31:47 +0000 (11:31 +0100)]
media: tvp5150: fix set_selection rectangle handling

Currently a local copy of sel->r is made and adapted to the hardware
constraints. After the adaption the value is applied to the hardware but
the driver forgot to reflect the adapted value to the user space.

Drop the local copy and work directly on the requested rectangle
instead to fix this.

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>
4 years agomedia: dt-bindings: tvp5150: Add input port connectors DT bindings
Marco Felsch [Thu, 12 Mar 2020 10:31:46 +0000 (11:31 +0100)]
media: dt-bindings: tvp5150: Add input port connectors DT bindings

The TVP5150/1 decoders support different video input sources to their
AIP1A/B pins.

Possible configurations are as follows:
  - Analog Composite signal connected to AIP1A.
  - Analog Composite signal connected to AIP1B.
  - Analog S-Video Y (luminance) and C (chrominance)
    signals connected to AIP1A and AIP1B respectively.

This patch extends the device tree bindings documentation to describe
how the input connectors for these devices should be defined in a DT.

Signed-off-by: Marco Felsch <m.felsch@pengutronix.de>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: tvp5150: add input source selection of_graph support
Marco Felsch [Thu, 12 Mar 2020 10:31:45 +0000 (11:31 +0100)]
media: tvp5150: add input source selection of_graph support

This patch adds the of_graph support to describe the tvp input connections.
Physical the TVP5150 has three ports: AIP1A, AIP1B and YOUT. As result
of discussion [1],[2] the device-tree maps these ports 1:1. Look at the
Documentation for more information. Since the TVP5150 is a converter/bridge
the device-tree must contain at least 1-input and 1-output port. The
mc-connectors and mc-links are only created if the device-tree contains the
corresponding connector nodes. If more than one connector is available the
media_entity_operations.link_setup() callback ensures that only one
connector is active.

[1] https://www.spinics.net/lists/linux-media/msg138545.html
[2] https://www.spinics.net/lists/linux-media/msg138546.html

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>
4 years agomedia: partial revert of "[media] tvp5150: add HW input connectors support"
Javier Martinez Canillas [Thu, 12 Mar 2020 10:31:44 +0000 (11:31 +0100)]
media: partial revert of "[media] tvp5150: add HW input connectors support"

Commit f7b4b54e6364 ("[media] tvp5150: add HW input connectors support")
added input signals support for the tvp5150, but the approach was found
to be incorrect so the corresponding DT binding commit 82c2ffeb217a
("[media] tvp5150: document input connectors DT bindings") was reverted.

This left the driver with an undocumented (and wrong) DT parsing logic,
so lets get rid of this code as well until the input connectors support
is implemented properly.

It's a partial revert due other patches added on top of mentioned commit
not allowing the commit to be reverted cleanly anymore. But all the code
related to the DT parsing logic and input entities creation are removed.

Suggested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
[m.felsch@pengutronix.de: rm TVP5150_INPUT_NUM define]
Signed-off-by: Marco Felsch <m.felsch@pengutronix.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: v4l2-fwnode: add initial connector parsing support
Marco Felsch [Thu, 12 Mar 2020 10:31:43 +0000 (11:31 +0100)]
media: v4l2-fwnode: add initial connector parsing support

The patch adds the initial connector parsing code, so we can move from a
driver specific parsing code to a generic one. Currently only the
generic fields and the analog-connector specific fields are parsed. Parsing
the other connector specific fields can be added by a simple callbacks.

Signed-off-by: Marco Felsch <m.felsch@pengutronix.de>
[hverkuil-cisco@xs4all.nl: replace ; with break; in a empty case]
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: v4l2-fwnode: add v4l2_fwnode_connector
Marco Felsch [Thu, 12 Mar 2020 10:31:42 +0000 (11:31 +0100)]
media: v4l2-fwnode: add v4l2_fwnode_connector

Currently every driver needs to parse the connector endpoints by it self.
This is the initial work to make this generic. A generic connector has
common members and connector specific members. The common members are:
  - type
  - label (optional)
  - links
  - nr_of_links

The specific members are stored within a union, since only one of them
can be available at the time. Since this is the initial support the
patch adds only the analog-connector specific ones.

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>
4 years agomedia: v4l2-fwnode: add endpoint id field to v4l2_fwnode_link
Marco Felsch [Thu, 12 Mar 2020 10:31:41 +0000 (11:31 +0100)]
media: v4l2-fwnode: add endpoint id field to v4l2_fwnode_link

A link is between two endpoints not between two ports to be more
precise. Add the local_id/remote_id field which stores the endpoint
reg/port property to the link. Now the link holds all necessary
information about a link.

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>
4 years agomedia: v4l2-fwnode: simplify v4l2_fwnode_parse_link
Marco Felsch [Thu, 12 Mar 2020 10:31:40 +0000 (11:31 +0100)]
media: v4l2-fwnode: simplify v4l2_fwnode_parse_link

This helper was introduced before those helpers where awailable. Convert
it to cleanup the code and improbe readability.

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>
4 years agomedia: v4l2-fwnode: fix v4l2_fwnode_parse_link handling
Marco Felsch [Thu, 12 Mar 2020 10:31:39 +0000 (11:31 +0100)]
media: v4l2-fwnode: fix v4l2_fwnode_parse_link handling

Currently the driver differentiate the port number property handling for
ACPI and DT. This is wrong as because ACPI should use the "reg" val too
[1].

[1] https://patchwork.kernel.org/patch/11421985/

Fixes: ca50c197bd96 ("[media] v4l: fwnode: Support generic fwnode for parsing standardised properties")
Signed-off-by: Marco Felsch <m.felsch@pengutronix.de>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: v4l: link dt-bindings and uapi
Marco Felsch [Thu, 12 Mar 2020 10:31:38 +0000 (11:31 +0100)]
media: v4l: link dt-bindings and uapi

Since we expose the definition to the dt-bindings we need to keep those
definitions in sync. To address this the patch adds a simple cross
reference to the dt-bindings.

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>
4 years agomedia: dt-bindings: display: add sdtv-standards defines
Marco Felsch [Thu, 12 Mar 2020 10:31:37 +0000 (11:31 +0100)]
media: dt-bindings: display: add sdtv-standards defines

Add defines which can be included to easily describe the supported
standard tv norms 'sdtv-standards' within the device tree.

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>
4 years agomedia: dt-bindings: connector: analog: add sdtv standards property
Marco Felsch [Thu, 12 Mar 2020 10:31:36 +0000 (11:31 +0100)]
media: dt-bindings: connector: analog: add sdtv standards property

Some connectors no matter if in- or output supports only a limited
range of sdtv standards. It doesn't matter if the hardware behind that
connector supports more than the listed formats since the users are
restriced by a label e.g. to plug only a camera into this connector
which uses the PAL format.

This patch adds the capability to describe such limitation within the
firmware. There are no format restrictions if the property isn't
present, so it's completely backward compatible.

Signed-off-by: Marco Felsch <m.felsch@pengutronix.de>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: am437x-vpfe: exclude illegal values for enum when validing params from user...
Changming Liu [Tue, 10 Mar 2020 21:13:20 +0000 (22:13 +0100)]
media: am437x-vpfe: exclude illegal values for enum when validing params from user space

When calling ccdc_data_size_max_bit() to validate data_sz in
vpfe_ccdc_validate_param(), it's treated as an enumeration ranging from
0 to 7 while essentially it's an 32 bit unsigned integer directly from
user space. This can make the return value of ccdc_data_size_max_bit()
underflow and bypass the following check.

To fix this, an additional check is added to the following if clause to
keep this enumaration variable in range. And if its value is not legal,
return -EINVAL properly.

Signed-off-by: Changming Liu <charley.ashbringer@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: ti-vpe: cal: fix a kernel oops when unloading module
Benoit Parrot [Fri, 6 Mar 2020 13:08:39 +0000 (14:08 +0100)]
media: ti-vpe: cal: fix a kernel oops when unloading module

After the switch to use v4l2_async_notifier_add_subdev() and
v4l2_async_notifier_cleanup(), unloading the ti_cal module would cause a
kernel oops.

This was root cause to the fact that v4l2_async_notifier_cleanup() tries
to kfree the asd pointer passed into v4l2_async_notifier_add_subdev().

In our case the asd reference was from a statically allocated struct.
So in effect v4l2_async_notifier_cleanup() was trying to free a pointer
that was not kalloc.

So here we switch to using a kzalloc struct instead of a static one.
To achieve this we re-order some of the calls to prevent asd allocation
from leaking.

Fixes: d079f94c9046 ("media: platform: Switch to v4l2_async_notifier_add_subdev")
Cc: stable@vger.kernel.org
Signed-off-by: Benoit Parrot <bparrot@ti.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: mtk-vpu: avoid unaligned access to DTCM buffer.
Hsin-Yi Wang [Wed, 4 Mar 2020 02:58:51 +0000 (03:58 +0100)]
media: mtk-vpu: avoid unaligned access to DTCM buffer.

Previously, vpu->recv_buf and send_buf are forced cast from
void __iomem *tcm. vpu->recv_buf->share_buf is passed to
vpu_ipi_desc.handler(). It's not able to do unaligned access. Otherwise
kernel would crash due to unable to handle kernel paging request.

struct vpu_run {
u32 signaled;
char fw_ver[VPU_FW_VER_LEN];
unsigned int dec_capability;
unsigned int enc_capability;
wait_queue_head_t wq;
};

fw_ver starts at 4 byte boundary. If system enables
CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS, strscpy() will do
read_word_at_a_time(), which tries to read 8-byte: *(unsigned long *)addr

vpu_init_ipi_handler() calls strscpy(), which would lead to crash.

vpu_init_ipi_handler() and several other handlers (eg.
vpu_dec_ipi_handler) only do read access to this data, so they can be
const, and we can use memcpy_fromio() to copy the buf to another non iomem
buffer then pass to handler.

Fixes: 85709cbf1524 ("media: replace strncpy() by strscpy()")
Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: venus: support frame rate control
Jeffrey Kardatzke [Sat, 22 Feb 2020 00:33:11 +0000 (01:33 +0100)]
media: venus: support frame rate control

Add encoder control for enabling/disabling frame rate control via
V4L2_CID_MPEG_VIDEO_FRAME_RC_ENABLE. It is enabled by default.

Signed-off-by: Jeffrey Kardatzke <jkardatzke@google.com>
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: venus: vdec: Fix forgotten mutex unlock in start streaming
Stanimir Varbanov [Tue, 3 Mar 2020 11:15:32 +0000 (12:15 +0100)]
media: venus: vdec: Fix forgotten mutex unlock in start streaming

This fixes the following smatch warning in the error path:

drivers/media/platform/qcom/venus/vdec.c:968 vdec_start_streaming()
 warn: inconsistent returns 'mutex:&inst->lock'.
  Locked on:   line 952
  Unlocked on: line 963
               line 968

by goto mutex unlock.

Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: lmedm04: remove redundant assignment to variable gate
Colin Ian King [Mon, 10 Feb 2020 14:58:44 +0000 (15:58 +0100)]
media: lmedm04: remove redundant assignment to variable gate

The variable gate is being initialized and also checked and re-assigned
with values that are never read as it is being re-assigned later in a
for-loop with a new value.  The assignments are redundant and can be
removed.

Addresses Coverity ("Unused value")

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: tda10071: fix unsigned sign extension overflow
Colin Ian King [Mon, 10 Feb 2020 14:26:46 +0000 (15:26 +0100)]
media: tda10071: fix unsigned sign extension overflow

The shifting of buf[3] by 24 bits to the left will be promoted to
a 32 bit signed int and then sign-extended to an unsigned long. In
the unlikely event that the the top bit of buf[3] is set then all
then all the upper bits end up as also being set because of
the sign-extension and this affect the ev->post_bit_error sum.
Fix this by using the temporary u32 variable bit_error to avoid
the sign-extension promotion. This also removes the need to do the
computation twice.

Addresses-Coverity: ("Unintended sign extension")

Fixes: 267897a4708f ("[media] tda10071: implement DVBv5 statistics")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: dw2102: probe for demodulator i2c address
Brad Love [Sat, 1 Feb 2020 21:48:26 +0000 (22:48 +0100)]
media: dw2102: probe for demodulator i2c address

This is required to support the Terratec S2 USB Box Revision 4, which
reused usb vid:pid, but has a different demodulator (m88ds3103b) at
i2c address 0x6a.

[fixed checkpatch issues]

Signed-off-by: Michael Bunk <micha@freedict.org>
Signed-off-by: Brad Love <brad@nextdimension.cc>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: em28xx: Enable Hauppauge 461e rev2
Brad Love [Sat, 1 Feb 2020 21:48:25 +0000 (22:48 +0100)]
media: em28xx: Enable Hauppauge 461e rev2

Hauppauge 461e rev2 is a DVB-S/S2 usb device containing:
- m88ds3103b demod
- ts2022 tuner
- A8293 SEC

Device is the same as Hauppauge 461e,
except it contains updated m88ds3103b demod.

Signed-off-by: Brad Love <brad@nextdimension.cc>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: m88ds3103: Add support for ds3103b demod
Brad Love [Sat, 1 Feb 2020 21:48:24 +0000 (22:48 +0100)]
media: m88ds3103: Add support for ds3103b demod

The ds3103b demodulator identifies as an m88rs600, but requires different
clock settings and firmware, along with differences in register settings.

Changes were reverse engineered using an instrumented downstream GPLv2
driver to compare i2c traffic and clocking. The mclk functions are from
the downstream GPLv2 driver.

Signed-off-by: Brad Love <brad@nextdimension.cc>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: rc: make scancodes 64 bit
Sean Young [Fri, 17 Jan 2020 16:46:36 +0000 (17:46 +0100)]
media: rc: make scancodes 64 bit

There are many protocols that encode more than 32 bit. We want 64 bit
support so that BPF IR decoders can decode more than 32 bit. None of
the existing kernel IR decoders/encoders support 64 bit, for now.

The MSC_SCAN event can only contain 32 bit scancodes, so we only generate
MSC_SCAN events if the scancode fits into 32 bits. The full 64 bit
scancode can be read from the lirc chardev.

Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: meson: vdec: add VP9 decoder support
Maxime Jourdan [Wed, 4 Mar 2020 09:46:25 +0000 (10:46 +0100)]
media: meson: vdec: add VP9 decoder support

This adds VP9 decoding for the Amlogic GXL, G12A & SM1 SoCs, using
the commong "HEVC" HW decoder.

For G12A & SM1, it uses the IOMMU support from the firmware.

For 10bit decoding, the firmware can only decode in the proprietary
Amlogic Framebuffer Compression format, but can output in 8bit NV12
buffer while writing the decoded frame.

Signed-off-by: Maxime Jourdan <mjourdan@baylibre.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Tested-by: Kevin Hilman <khilman@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: meson: vdec: add VP9 input support
Maxime Jourdan [Wed, 4 Mar 2020 09:46:24 +0000 (10:46 +0100)]
media: meson: vdec: add VP9 input support

Amlogic VP9 decoder requires an additional 16-byte payload before every
frame header.

The source buffer is updated in-place, then given to the Parser FIFO DMA.

The FIFO DMA copies the blocks into the 16MiB parser ring buffer, then parses
and copies the slice into the decoder "workspace".

Signed-off-by: Maxime Jourdan <mjourdan@baylibre.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Tested-by: Kevin Hilman <khilman@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: meson: vdec: add common HEVC decoder support
Maxime Jourdan [Wed, 4 Mar 2020 09:46:23 +0000 (10:46 +0100)]
media: meson: vdec: add common HEVC decoder support

Add support for the HEVC & VP9 common decoder support, handling
Amlogic GXBB, GXL, G12A and SM1 platforms.

This handles the "HEVC" hw decoder used for HEVC and VP9, and will be
using in the new H264 multi-instance decoder for G12A & SM1 platforms.

Signed-off-by: Maxime Jourdan <mjourdan@baylibre.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Tested-by: Kevin Hilman <khilman@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: meson: vdec: add helpers for lossless framebuffer compression buffers
Maxime Jourdan [Wed, 4 Mar 2020 09:46:22 +0000 (10:46 +0100)]
media: meson: vdec: add helpers for lossless framebuffer compression buffers

Add helpers to support the lossless framebuffer compression format that
will be used in HEVC & VP9 decoders when decoding 10bit content for
downsampling to 8bit NV12 and later proper compressed buffer support.

Signed-off-by: Maxime Jourdan <mjourdan@baylibre.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Tested-by: Kevin Hilman <khilman@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: meson: vdec: align stride on 32 bytes
Neil Armstrong [Wed, 4 Mar 2020 09:46:21 +0000 (10:46 +0100)]
media: meson: vdec: align stride on 32 bytes

The HEVC/VP9 aligns the plane stride on 32, so align the planes stride
for all codecs to 32 to satisfy HEVC/VP9 decoding using the "HEVC" HW.

This fixes VP9 decoding of streams with following (not limited) widths:
- 264
 -288
- 350
- 352
- 472
- 480
- 528
- 600
- 720
- 800
- 848
- 1440

Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Tested-by: Kevin Hilman <khilman@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: meson: vdec: add H.264 decoding support
Maxime Jourdan [Tue, 3 Mar 2020 14:33:20 +0000 (15:33 +0100)]
media: meson: vdec: add H.264 decoding support

Add support for the H264 compressed format (V4L2_PIX_FMT_H264).

Signed-off-by: Maxime Jourdan <mjourdan@baylibre.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: meson: vdec: bring up to compliance
Maxime Jourdan [Tue, 3 Mar 2020 14:33:19 +0000 (15:33 +0100)]
media: meson: vdec: bring up to compliance

Add all the necessary bits to pass v4l2-compliance in stateful decoding
mode.

Mostly includes tracking the state of the decoder, allowing the OUTPUT
queue to stream while the CAPTURE queue is inactive, handling resolution
change events, draining with V4L2_DEC_CMD_STOP, copying more metadata
from the src buffers to the dst buffers, etc.

Signed-off-by: Maxime Jourdan <mjourdan@baylibre.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: vicodec: use v4l2-mem2mem draining, stopped and next-buf-is-last states handling
Neil Armstrong [Tue, 3 Mar 2020 14:33:18 +0000 (15:33 +0100)]
media: vicodec: use v4l2-mem2mem draining, stopped and next-buf-is-last states handling

Use the previously introduced v4l2-mem2mem core APIs to handle the drainig,
stopped and next-buf-is-last states.

With these changes, the v4l2-compliance still passes with the following
commands :
>>>><><><><><><><><><><><><><><><><>< 15.53 fps
 15.53 fps
><><><><><><><><><><><><>< 13.99 fps
 13.99 fps
><><><><><><><><><><><>< 13.52 fps
 13.52 fps
><><><><><><><><><><><><>< 13.41 fps
 13.41 fps
><><><><><><><><><><><><>< 13.21 fps
 13.21 fps
><><><><><><><><><><><>< 13.09 fps
 13.09 fps
><><><><><><><
STOP ENCODER
<<<
EOS EVENT

v4l2-compliance SHA: 7ead0e1856b89f2e19369af452bb03fd0cd16793, 64 bits
[...]
Total for vicodec device /dev/video0: 50, Succeeded: 50, Failed: 0, Warnings: 0

The full output is available at [1]

v4l2-compliance SHA: 7ead0e1856b89f2e19369af452bb03fd0cd16793, 64 bits
[...]
Total for vicodec device /dev/video1: 50, Succeeded: 50, Failed: 0, Warnings: 0

The full output is available at [2]

No functional changes should be noticed.

[1] https://termbin.com/25nn
[2] https://termbin.com/dza4

Suggested-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Suggested-by: Maxime Jourdan <mjourdan@baylibre.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: v4l2-mem2mem: handle draining, stopped and next-buf-is-last states
Neil Armstrong [Tue, 3 Mar 2020 14:33:17 +0000 (15:33 +0100)]
media: v4l2-mem2mem: handle draining, stopped and next-buf-is-last states

Since the draining and stop phase of the HW decoder mem2mem bahaviour is
now clearly defined, we can move handling of the following states to the
common v4l2-mem2mem core code:
- draining
- stopped
- next-buf-is-last

By introducing the following v4l2-mem2mem APIs:
- v4l2_m2m_encoder_cmd/v4l2_m2m_ioctl_encoder_cmd to handle start/stop command
- v4l2_m2m_decoder_cmd/v4l2_m2m_ioctl_decoder_cmd to handle start/stop command
- v4l2_m2m_update_start_streaming_state to update state on start of streaming
of the de/encoder queue
- v4l2_m2m_update_stop_streaming_state to update state on stop of streaming
of the de/encoder queue
- v4l2_m2m_last_buffer_done to make the current dest buffer as the last one

And inline helpers:
- v4l2_m2m_mark_stopped to mark the de/encoding process as stopped
- v4l2_m2m_clear_state to clear the de/encoding state
- v4l2_m2m_dst_buf_is_last to detect the current dequeued dst_buf is the last
- v4l2_m2m_has_stopped to detect the de/encoding stopped state
- v4l2_m2m_is_last_draining_src_buf to detect the current source buffer should
 be the last processing before stopping the de/encoding process

The special next-buf-is-last when min_buffers != 1 case is also handled
in v4l2_m2m_qbuf() by reusing the other introduced APIs.

This state management has been stolen from the vicodec implementation,
and is no-op for drivers not calling the v4l2_m2m_encoder_cmd or
v4l2_m2m_decoder_cmd and v4l2_m2m_update_start/stop_streaming_state.

The vicodec will be the first one to be converted as an example.

Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
4 years agomedia: v4l2-ctrls: v4l2_ctrl_request_complete() should always set ref->req
Hans Verkuil [Tue, 3 Mar 2020 10:34:48 +0000 (11:34 +0100)]
media: v4l2-ctrls: v4l2_ctrl_request_complete() should always set ref->req

When the request is completed, all controls are copied to the request object.
However, when VIDIOC_G_EXT_CTRLS attempts to read control values from the
request it will read the current value instead for any control reference that
has a NULL ref->req pointer. But that's wrong: after completing the request
*all* controls should have a non-NULL ref->req pointer since they are after
all copied to the request.

So set ref->req if it wasn't set already.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reported-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>