openwrt/staging/blogic.git
5 years agomedia: change au0828 to use Media Device Allocator API
Shuah Khan [Tue, 2 Apr 2019 00:40:20 +0000 (20:40 -0400)]
media: change au0828 to use Media Device Allocator API

Media Device Allocator API to allows multiple drivers share a media device.
This API solves a very common use-case for media devices where one physical
device (an USB stick) provides both audio and video. When such media device
exposes a standard USB Audio class, a proprietary Video class, two or more
independent drivers will share a single physical USB bridge. In such cases,
it is necessary to coordinate access to the shared resource.

Using this API, drivers can allocate a media device with the shared struct
device as the key. Once the media device is allocated by a driver, other
drivers can get a reference to it. The media device is released when all
the references are released.

Change au0828 to use Media Device Allocator API to allocate media device
with the parent usb struct device as the key, so it can be shared with the
snd_usb_audio driver.

Signed-off-by: Shuah Khan <shuah@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: Media Device Allocator API
Shuah Khan [Tue, 2 Apr 2019 00:40:19 +0000 (20:40 -0400)]
media: Media Device Allocator API

Media Device Allocator API to allows multiple drivers share a media device.
This API solves a very common use-case for media devices where one physical
device (an USB stick) provides both audio and video. When such media device
exposes a standard USB Audio class, a proprietary Video class, two or more
independent drivers will share a single physical USB bridge. In such cases,
it is necessary to coordinate access to the shared resource.

Using this API, drivers can allocate a media device with the shared struct
device as the key. Once the media device is allocated by a driver, other
drivers can get a reference to it. The media device is released when all
the references are released.

Signed-off-by: Shuah Khan <shuah@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: stm32-dcmi: fix crash when subdev do not expose any formats
Hugues Fruchet [Mon, 1 Apr 2019 08:56:09 +0000 (04:56 -0400)]
media: stm32-dcmi: fix crash when subdev do not expose any formats

Do not access sd_formats[] if num_of_sd_formats is zero, ie
subdev sensor didn't expose any formats.

Signed-off-by: Hugues Fruchet <hugues.fruchet@st.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: stm32-dcmi: do not emit error trace in case of few overrun
Hugues Fruchet [Mon, 1 Apr 2019 08:55:00 +0000 (04:55 -0400)]
media: stm32-dcmi: do not emit error trace in case of few overrun

Report overrun error only when it exceeds a given threshold.

Signed-off-by: Hugues Fruchet <hugues.fruchet@st.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: stm32-dcmi: fix error messages
Hugues Fruchet [Mon, 1 Apr 2019 08:54:59 +0000 (04:54 -0400)]
media: stm32-dcmi: fix error messages

Revisit and add some error messages.

Signed-off-by: Hugues Fruchet <hugues.fruchet@st.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: dvb: clean up redundant break statements
Colin Ian King [Sat, 30 Mar 2019 15:25:03 +0000 (11:25 -0400)]
media: dvb: clean up redundant break statements

There are several places where a break statement occurs before
a following break statement; these are unnecessary and can be
removed to clean up the code a little.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Mukesh Ojha <mojha@codeaurora.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: mtk-vcodec: enlarge struct vdec_pic_info fields
Yunfei Dong [Fri, 29 Mar 2019 08:37:14 +0000 (04:37 -0400)]
media: mtk-vcodec: enlarge struct vdec_pic_info fields

Enlarge the plane number to support more complex case
and add the support for fmt change case.

Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: dt-bindings: media: Renesas R-Car IMR bindings
Sergei Shtylyov [Tue, 26 Mar 2019 16:12:19 +0000 (12:12 -0400)]
media: dt-bindings: media: Renesas R-Car IMR bindings

The image renderer (IMR), or the distortion correction engine, is a
drawing processor with a simple instruction system capable of referencing
video capture data or data in an external memory as the 2D texture data
and performing texture mapping and drawing with respect to any shape that
is split into triangular objects.

Document  the device tree bindings for the image renderer light extended 4
(IMR-LX4) found in the R-Car gen3 SoCs...

Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: staging: media: imx: imx7-mipi-csis: simplify getting .driver_data
Wolfram Sang [Tue, 19 Mar 2019 16:36:22 +0000 (12:36 -0400)]
media: staging: media: imx: imx7-mipi-csis: simplify getting .driver_data

We should get 'driver_data' from 'struct device' directly. Going via
platform_device is an unneeded step back and forth.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vimc: propagate pixel format in the stream
Helen Fornazier [Wed, 13 Mar 2019 18:29:37 +0000 (14:29 -0400)]
media: vimc: propagate pixel format in the stream

Media bus codes were being mapped to pixelformats, which causes a
limitation on vimc because not all pixelformats can be mapped to media
bus codes.
Also, media bus codes are an internal configuration from the device.
Userspace only assures media bus codes matches between pads and expects
the image in a given pixelformat. So we can allow almost any media bus
format to be configured between pads, except for debayer that expects a
media bus code of type bayer in the sink pad.

[hverkuil-cisco@xs4all.nl: drop use of v4l2_get_fourcc_name: not yet available]
[hverkuil-cisco@xs4all.nl: made vimc_mbus_list static]

Signed-off-by: Helen Koike <helen.koike@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: v4l2-common: add bayer formats in v4l2_format_info
Helen Fornazier [Wed, 6 Mar 2019 22:42:40 +0000 (17:42 -0500)]
media: v4l2-common: add bayer formats in v4l2_format_info

Add bayer format information in struct v4l2_format_info table.

Signed-off-by: Helen Koike <helen.koike@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: Introduce helpers to fill pixel format structs
Ezequiel Garcia [Thu, 28 Mar 2019 18:07:04 +0000 (14:07 -0400)]
media: Introduce helpers to fill pixel format structs

Add two new API helpers, v4l2_fill_pixfmt and v4l2_fill_pixfmt_mp,
to be used by drivers to calculate plane sizes and bytes per lines.

Note that driver-specific padding and alignment are not
taken into account, and must be done by drivers using this API.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: rockchip/vpu: Rename pixel format helpers
Ezequiel Garcia [Thu, 28 Mar 2019 18:07:03 +0000 (14:07 -0400)]
media: rockchip/vpu: Rename pixel format helpers

The rockchip VPU driver uses generic names for its pixel format
helpers. We want to use the same names for generic versions
of these helpers, so rename the rockchip ones.

The driver will be switched to the generic helpers later.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vb2: Minor function naming fix
Ezequiel Garcia [Thu, 28 Mar 2019 16:16:30 +0000 (12:16 -0400)]
media: vb2: Minor function naming fix

Minor fix for helper function and comment, s/v4l2_vb2_buffer/vb2_v4l2_buffer.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: v4l2-ctrls.h: remove spurious text
Hans Verkuil [Mon, 1 Apr 2019 07:42:05 +0000 (03:42 -0400)]
media: v4l2-ctrls.h: remove spurious text

Somehow the string "Controls name" got pasted in two places
where it doesn't belong. Remove that text.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: platform: stm32: wait end of transmission
Yannick Fertre [Fri, 29 Mar 2019 12:39:42 +0000 (08:39 -0400)]
media: platform: stm32: wait end of transmission

It is mandatory to write CEC_CFGR only when CECEN=0. To protect
transmission, a check have been added to delayed logical address
modification. This patch is necessary tp pass all tests of compliance.

Signed-off-by: Yannick Fertré <yannick.fertre@st.com>
Reviewed-by: Benjamin Gaignard <benjamin.gaignard@linaro.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: atmel: atmel-isc: removed ARGB32 added ABGR32 and XBGR32
Eugen Hristev [Fri, 29 Mar 2019 07:38:31 +0000 (03:38 -0400)]
media: atmel: atmel-isc: removed ARGB32 added ABGR32 and XBGR32

ISC will output the "ARGB32" configuration in byte order: B, G, R, Alpha.
This is in fact the format BGRA, aka ABGR32.
If alpha is missing, the same format is equivalent to XBGR32.
Added both formats and removed ARGB32 which is wrong.

Signed-off-by: Eugen Hristev <eugen.hristev@microchip.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: atmel: atmel-isc: reworked driver and formats
Eugen Hristev [Fri, 29 Mar 2019 07:38:28 +0000 (03:38 -0400)]
media: atmel: atmel-isc: reworked driver and formats

This change is a redesign in the formats and the way the ISC is
configured w.r.t. sensor format and the output format from the ISC.
I have changed the splitting between sensor output (which is also ISC input)
and ISC output.
The sensor format represents the way the sensor is configured, and what ISC
is receiving.
The format configuration represents the way ISC is interpreting the data and
formatting the output to the subsystem.
Now it's much easier to figure out what is the ISC configuration for input, and
what is the configuration for output.
The non-raw format can be obtained directly from sensor or it can be done
inside the ISC. The controller format list will include a configuration for
each format.
The old supported formats are still in place, if we want to dump the sensor
format directly to the output, the try format routine will detect and
configure the pipeline accordingly.
This also fixes the previous issues when the raw format was NULL which
resulted in many crashes for sensors which did not have the expected/tested
formats.

Signed-off-by: Eugen Hristev <eugen.hristev@microchip.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: v4l: add I / P frame min max QP definitions
Fish Lin [Fri, 29 Mar 2019 03:20:46 +0000 (23:20 -0400)]
media: v4l: add I / P frame min max QP definitions

Add following V4L2 QP parameters for H.264:
* V4L2_CID_MPEG_VIDEO_H264_I_FRAME_MIN_QP
* V4L2_CID_MPEG_VIDEO_H264_I_FRAME_MAX_QP
* V4L2_CID_MPEG_VIDEO_H264_P_FRAME_MIN_QP
* V4L2_CID_MPEG_VIDEO_H264_P_FRAME_MAX_QP

These controls will limit QP range for intra and inter frame,
provide more manual control to improve video encode quality.

Signed-off-by: Fish Lin <linfish@google.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: index.rst: exclude Indexes section from latex/pdf output
Mauro Carvalho Chehab [Sat, 30 Mar 2019 13:33:54 +0000 (09:33 -0400)]
media: index.rst: exclude Indexes section from latex/pdf output

With LaTex, the Index at the end will always be present. Having
an extra chapter for it there will just add extra noise to the
document.

Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: pdf docs: adjust several tables for better display
Mauro Carvalho Chehab [Fri, 29 Mar 2019 16:13:50 +0000 (12:13 -0400)]
media: pdf docs: adjust several tables for better display

Avoid cell overlapping by changing some sizes, and changing the
font sizes when needed.

Tested with Sphinx 1.7.8.

Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: pixfmt-packed-yuv.rst: fix PDF output
Mauro Carvalho Chehab [Fri, 29 Mar 2019 15:58:14 +0000 (11:58 -0400)]
media: pixfmt-packed-yuv.rst: fix PDF output

Some lines inside this table has more than 34 columns, which isn't
what is expected, according to the .. tabularcolumns:: definition.

That causes the PDF output to fail.

This patch ensures that all columns will have exactly the same
size, merging the unused ones using :cpan:

Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: staging: use strscpy() instead of strlcpy()
Mauro Carvalho Chehab [Fri, 29 Mar 2019 15:28:45 +0000 (11:28 -0400)]
media: staging: use strscpy() instead of strlcpy()

There are a few left overs at staging which were still using the
deprecated strlcpy() function.

Reviewed-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reviewed-by: Mukesh Ojha <mojha@codeaurora.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: soc_camera: Remove leftover files, add TODO
Sakari Ailus [Tue, 26 Mar 2019 09:24:11 +0000 (05:24 -0400)]
media: soc_camera: Remove leftover files, add TODO

Remove a few files left over from the mt9t031 driver. While at it, add a
TODO file for the SoC camera framework as a whole.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: MAINTAINERS: Fix files for obsolete SoC camera framework
Sakari Ailus [Tue, 26 Mar 2019 09:19:28 +0000 (05:19 -0400)]
media: MAINTAINERS: Fix files for obsolete SoC camera framework

The SoC camera was moved to the staging tree but we missed updating
MAINTAINERS. Do that now.

Fixes: 280de94a6519 ("media: soc_camera: Move to the staging tree")
Reported-by: Joe Perches <joe@perches.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: ov7740: enable to get exposure control in autoexposure mode
Akinobu Mita [Mon, 18 Mar 2019 15:25:43 +0000 (11:25 -0400)]
media: ov7740: enable to get exposure control in autoexposure mode

The exposure control is clustered with the autoexposure control and
flagged as volatile, but the g_volatile_ctrl() doesn't handle
V4L2_CID_EXPOSURE_AUTO.  So, the value of the exposure control can't be
read in autoexposure mode.

This enables to get the exposure control in autoexposure mode by making
ov7740_get_volatile_ctrl() deal with V4L2_CID_EXPOSURE_AUTO.

This also sets the exposure control as volatile by specifying the
argument to v4l2_ctrl_auto_cluster() instead of manually flagging it.

Cc: Wenyou Yang <wenyou.yang@microchip.com>
Cc: Eugen Hristev <eugen.hristev@microchip.com>
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: bt8xx: avoid a needless u8 to char conversion
Mauro Carvalho Chehab [Fri, 29 Mar 2019 14:29:40 +0000 (10:29 -0400)]
media: bt8xx: avoid a needless u8 to char conversion

Instead of doing the cast, just change the type to char.

Suggested-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: replace strncpy() by strscpy()
Mauro Carvalho Chehab [Mon, 10 Sep 2018 12:19:16 +0000 (08:19 -0400)]
media: replace strncpy() by strscpy()

The strncpy() function is being deprecated upstream. Replace
it by the safer strscpy().

While here, replace a few occurences of strlcpy() that were
recently added to also use strscpy().

Reviewed-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vim2m: Fix RGB 565 BE/LE support
Mauro Carvalho Chehab [Thu, 28 Mar 2019 19:02:23 +0000 (15:02 -0400)]
media: vim2m: Fix RGB 565 BE/LE support

The support for those two formats are archtecture-dependent.
Use the endianness to CPU macros to do it right.

Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: wl128x: Fix some error handling in fmc_prepare()
Dan Carpenter [Thu, 28 Mar 2019 14:41:38 +0000 (10:41 -0400)]
media: wl128x: Fix some error handling in fmc_prepare()

The st_register() returns have changed over time, but these days it
never returns -1.  We should just be checking for any negative error
codes.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: wl128x: prevent two potential buffer overflows
Dan Carpenter [Tue, 26 Mar 2019 05:12:07 +0000 (01:12 -0400)]
media: wl128x: prevent two potential buffer overflows

Smatch marks skb->data as untrusted so it warns that "evt_hdr->dlen"
can copy up to 255 bytes and we only have room for two bytes.  Even
if this comes from the firmware and we trust it, the new policy
generally is just to fix it as kernel hardenning.

I can't test this code so I tried to be very conservative.  I considered
not allowing "evt_hdr->dlen == 1" because it doesn't initialize the
whole variable but in the end I decided to allow it and manually
initialized "asic_id" and "asic_ver" to zero.

Fixes: e8454ff7b9a4 ("[media] drivers:media:radio: wl128x: FM Driver Common sources")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: tw5864: Fix possible NULL pointer dereference in tw5864_handle_frame
YueHaibing [Fri, 15 Mar 2019 02:01:24 +0000 (22:01 -0400)]
media: tw5864: Fix possible NULL pointer dereference in tw5864_handle_frame

'vb' null check should be done before dereferencing it in
tw5864_handle_frame, otherwise a NULL pointer dereference
may occur.

Fixes: 34d1324edd31 ("[media] pci: Add tw5864 driver")
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: usbvision: fix a potential NULL pointer dereference
Kangjie Lu [Sat, 9 Mar 2019 07:42:26 +0000 (02:42 -0500)]
media: usbvision: fix a potential NULL pointer dereference

In case usb_alloc_coherent fails, the fix returns -ENOMEM to
avoid a potential NULL pointer dereference.

Signed-off-by: Kangjie Lu <kjlu@umn.edu>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: video-mux: fix null pointer dereferences
Kangjie Lu [Sat, 9 Mar 2019 07:20:56 +0000 (02:20 -0500)]
media: video-mux: fix null pointer dereferences

devm_kcalloc may fail and return a null pointer. The fix returns
-ENOMEM upon failures to avoid null pointer dereferences.

Signed-off-by: Kangjie Lu <kjlu@umn.edu>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: mtk-vcodec: remove ready_to_display
Alexandre Courbot [Tue, 26 Mar 2019 07:44:46 +0000 (03:44 -0400)]
media: mtk-vcodec: remove ready_to_display

This member is never read throughout the code, so remove it.

Signed-off-by: Alexandre Courbot <acourbot@chromium.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: mtk-vcodec: fix access to incorrect planes member
Alexandre Courbot [Tue, 26 Mar 2019 07:44:23 +0000 (03:44 -0400)]
media: mtk-vcodec: fix access to incorrect planes member

Commit 0650a91499e0 ("media: mtk-vcodec: Correct return type for mem2mem
buffer helpers") fixed the return types for mem2mem buffer helper
functions by changing a few local variables from vb2_buffer to
vb2_v4l2_buffer. However, it left a few accesses to vb2_buffer::planes
as-is, accidentally turning them into accesses to
vb2_v4l2_buffer::planes and resulting in values being read from/written
to the wrong place.

Fix this by inserting vb2_buf into these accesses so they mimic their
original behavior.

Fixes: 0650a91499e0 ("media: mtk-vcodec: Correct return type for mem2mem buffer helpers")
Signed-off-by: Alexandre Courbot <acourbot@chromium.org>
Reviewed-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: v4l2-ctrl: potential shift wrapping bugs
Dan Carpenter [Mon, 25 Mar 2019 09:06:26 +0000 (05:06 -0400)]
media: v4l2-ctrl: potential shift wrapping bugs

This code generates a static checker warning:

    drivers/media/v4l2-core/v4l2-ctrls.c:2921 v4l2_querymenu()
    warn: should '(1 << i)' be a 64 bit type?

The problem is that "ctrl->menu_skip_mask" is a u64 and we're only
testing the lower 32 bits.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: davinci-isif: avoid uninitialized variable use
Arnd Bergmann [Fri, 22 Mar 2019 14:34:22 +0000 (10:34 -0400)]
media: davinci-isif: avoid uninitialized variable use

clang warns about a possible variable use that gcc never
complained about:

drivers/media/platform/davinci/isif.c:982:32: error: variable 'frame_size' is uninitialized when used here
      [-Werror,-Wuninitialized]
                dm365_vpss_set_pg_frame_size(frame_size);
                                             ^~~~~~~~~~
drivers/media/platform/davinci/isif.c:887:2: note: variable 'frame_size' is declared here
        struct vpss_pg_frame_size frame_size;
        ^
1 error generated.

There is no initialization for this variable at all, and there
has never been one in the mainline kernel, so we really should
not put that stack data into an mmio register.

On the other hand, I suspect that gcc checks the condition
more closely and notices that the global
isif_cfg.bayer.config_params.test_pat_gen flag is initialized
to zero and never written to from any code path, so anything
depending on it can be eliminated.

To shut up the clang warning, just remove the dead code manually,
it has probably never been used because any attempt to do so
would have resulted in undefined behavior.

Fixes: 63e3ab142fa3 ("V4L/DVB: V4L - vpfe capture - source for ISIF driver on DM365")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Nathan Chancellor <natechancellor@gmail.com>
Acked-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: i2c: tda1997x: select V4L2_FWNODE
Koen Vandeputte [Mon, 18 Mar 2019 16:40:05 +0000 (12:40 -0400)]
media: i2c: tda1997x: select V4L2_FWNODE

Building tda1997x fails now unless V4L2_FWNODE is selected:

drivers/media/i2c/tda1997x.o: in function `tda1997x_parse_dt'
undefined reference to `v4l2_fwnode_endpoint_parse'

While at it, also sort the selections alphabetically

Fixes: 9ac0038db9a7 ("media: i2c: Add TDA1997x HDMI receiver driver")
Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
Cc: stable@vger.kernel.org # v4.17+
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+samsung@kernel.org>
5 years agomedia: staging: media: imx7-mipi-csis: fix debugfs compilation
Arnd Bergmann [Wed, 13 Mar 2019 21:17:32 +0000 (17:17 -0400)]
media: staging: media: imx7-mipi-csis: fix debugfs compilation

When CONFIG_DEBUGFS is enabled, we get a warning about an
incorrect section annotation that can lead to undefined
behavior:

WARNING: vmlinux.o(.text+0xd3c7c4): Section mismatch in reference from the function mipi_csis_probe() to the function .init.text:mipi_csis_debugfs_init()
The function mipi_csis_probe() references
the function __init mipi_csis_debugfs_init().
This is often because mipi_csis_probe lacks a __init
annotation or the annotation of mipi_csis_debugfs_init is wrong.

The same function for an unknown reason has a different
version for !CONFIG_DEBUGFS, which does not have this problem,
but behaves the same way otherwise (it does nothing when debugfs
is disabled).
Consolidate the two versions, using the correct section from
one version, and the implementation from the other.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: staging: media: davinci_vpfe: fix large stack usage with clang
Arnd Bergmann [Wed, 13 Mar 2019 21:10:29 +0000 (17:10 -0400)]
media: staging: media: davinci_vpfe: fix large stack usage with clang

clang is unable to optimize the isif_ioctl() in the same way that
gcc does, as it fails to prove that the local copy of
the 'struct vpfe_isif_raw_config' argument is unnecessary:

drivers/staging/media/davinci_vpfe/dm365_isif.c:622:13: error: stack frame size of 1344 bytes in function 'isif_ioctl' [-Werror,-Wframe-larger-than=]

Marking it as 'const' while passing the data down clearly shows us that
the copy is never modified, and we can skip copying it entirely, which
reduces the stack usage to just eight bytes.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: staging/imx: Fix inconsistent long line breaks
Lucas Leonardo Ciancaglini [Mon, 11 Mar 2019 05:34:38 +0000 (01:34 -0400)]
media: staging/imx: Fix inconsistent long line breaks

Small readability changes to make the breaking of some lines
cleaner.

Signed-off-by: Lucas Leonardo Ciancaglini <leociancalucas@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: gspca: Kill URBs on USB device disconnect
Ezequiel Garcia [Thu, 28 Feb 2019 15:28:34 +0000 (10:28 -0500)]
media: gspca: Kill URBs on USB device disconnect

In order to prevent ISOC URBs from being infinitely resubmitted,
the driver's USB disconnect handler must kill all the in-flight URBs.

While here, change the URB packet status message to a debug level,
to avoid spamming the console too much.

This commit fixes a lockup caused by an interrupt storm coming
from the URB completion handler.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: dvbdev: remove double-unlock
Mauro Carvalho Chehab [Thu, 28 Mar 2019 18:36:14 +0000 (14:36 -0400)]
media: dvbdev: remove double-unlock

As warned by smatch:
drivers/media/dvb-core/dvbdev.c: drivers/media/dvb-core/dvbdev.c:529 dvb_register_device() error: double unlock 'sem:&minor_rwsem'

Reported-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: hfi_parser: don't trick gcc with a wrong expected size
Mauro Carvalho Chehab [Mon, 25 Mar 2019 21:09:05 +0000 (17:09 -0400)]
media: hfi_parser: don't trick gcc with a wrong expected size

Smatch warns about small size on two structs:

drivers/media/platform/qcom/venus/hfi_parser.c:103 parse_profile_level() error: memcpy() 'proflevel' too small (8 vs 128)
drivers/media/platform/qcom/venus/hfi_parser.c: drivers/media/platform/qcom/venus/hfi_parser.c:129 parse_caps() error: memcpy() 'cap' too small (16 vs 512)

The reason is that the hfi_parser actually expects:
    - multiple data entries on hfi_capabilities
    - multiple profile_level on hfi_profile_level_supported

However, the structs trick gcc, making it to believe that
there's just one value for each.

Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: rcar-dma: p_set can't be NULL
Mauro Carvalho Chehab [Mon, 25 Mar 2019 20:47:13 +0000 (16:47 -0400)]
media: rcar-dma: p_set can't be NULL

The only way for p_set to be NULL would be if vin_coef_set would be an
empty array.

On such case, the driver will OOPS, as it would try to de-reference a
NULL value. So, the check if p_set is not NULL doesn't make any sense.

Solves those two smatch warnings:

drivers/media/platform/rcar-vin/rcar-dma.c:489 rvin_set_coeff() warn: variable dereferenced before check 'p_set' (see line 484)
drivers/media/platform/rcar-vin/rcar-dma.c:494 rvin_set_coeff() error: we previously assumed 'p_set' could be null (see line 489)

Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: sti/delta: remove uneeded check
Mauro Carvalho Chehab [Mon, 25 Mar 2019 20:32:19 +0000 (16:32 -0400)]
media: sti/delta: remove uneeded check

At the error logic, ipc_buf was already asigned to &ctx->ipc_buf_struct,
with can't be null, as warned by smatch:

drivers/media/platform/sti/delta/delta-ipc.c:223 delta_ipc_open() warn: variable dereferenced before check 'ctx->ipc_buf' (see line 183)

So, remove the uneeded check.

Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: pwc-ctl: pChoose can't be NULL
Mauro Carvalho Chehab [Mon, 25 Mar 2019 20:20:47 +0000 (16:20 -0400)]
media: pwc-ctl: pChoose can't be NULL

The way the code works, compression will be a valid value (less or equal to 3)
on both set_video_mode_foo() calls at the beginning of the while() loop.

So, the value for pChoose can't be NULL.

Solves those smatch warnings:

drivers/media/usb/pwc/pwc-ctrl.c: drivers/media/usb/pwc/pwc-ctrl.c:252 set_video_mode_Timon() warn: variable dereferenced before check 'pChoose' (see line 248)
drivers/media/usb/pwc/pwc-ctrl.c: drivers/media/usb/pwc/pwc-ctrl.c:302 set_video_mode_Kiara() warn: variable dereferenced before check 'pChoose' (see line 298)

and simplifies the code a little bit.

Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: cx2341x: replace badly designed macros
Mauro Carvalho Chehab [Mon, 25 Mar 2019 19:09:30 +0000 (15:09 -0400)]
media: cx2341x: replace badly designed macros

There are some macros at cx2341x_update() with seemed to
be introduced in order to ensure that lines would be less
than 80 columns.

Well, the thing is that they make the code harder to be analized,
not only by humans, but also for static code analyzers:

drivers/media/common/cx2341x.c:1116 cx2341x_update() error: we previously assumed 'old' could be null (see line 1047)

So, remove the "force" var, and replace the NEQ macro to a
better designed one that makes clearer about what it is doing.

While here, also remove the "temporal" var, as it is just another
way of doing the same type of check as the new CMP_FIELD() macro
already does.

Finally, fix coding style at the block code.
 remove such macros.

Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: imx214: don't de-reference a NULL pointer
Mauro Carvalho Chehab [Fri, 7 Dec 2018 11:43:03 +0000 (06:43 -0500)]
media: imx214: don't de-reference a NULL pointer

As warned by smatch:
drivers/media/i2c/imx214.c:591 imx214_set_format() warn: variable dereferenced before check 'format' (see line 589)

It turns that the code at imx214_set_format() has support for being
called with the format being NULL. I've no idea why, as it is only
called internally with the pointer set, and via subdev API (with
should also set it).

Also, the entire logic there depends on having format != NULL, so
just remove the bogus broken support for a null format.

Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
Reviewed-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reviewed-by: Ricardo Ribalda Delgado <ricardo@ribalda.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: v4l: rcar_fdp1: Fix indentation oddities
Laurent Pinchart [Sun, 22 Apr 2018 08:00:08 +0000 (04:00 -0400)]
media: v4l: rcar_fdp1: Fix indentation oddities

Indentation is odd in several places, especially when printing messages
to the kernel log. Fix it to match the usual coding style.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: set pixelformat to V4L2_PIX_FMT_FWHT_STATELESS for stateless decoder
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:43 +0000 (16:13 -0500)]
media: vicodec: set pixelformat to V4L2_PIX_FMT_FWHT_STATELESS for stateless decoder

for stateless decoder, set the output pixelformat
to V4L2_PIX_FMT_FWHT_STATELESS and the pix info to
pixfmt_stateless_fwht

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: Add support for stateless decoder.
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:42 +0000 (16:13 -0500)]
media: vicodec: Add support for stateless decoder.

Implement a stateless decoder for the new node.

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
[hverkuil-cisco@xs4all.nl: add 'return 0;' before default case in vicodec_try_ctrl()]
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: Register another node for stateless decoder
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:41 +0000 (16:13 -0500)]
media: vicodec: Register another node for stateless decoder

Add stateless decoder instance field to the dev struct and
register another node for the statelsess decoder.
The stateless API for the node will be implemented in further patches.

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
[hverkuil-cisco@xs4all.nl: fix typo: videdev-stateless-dec -> stateless-decoder]
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: v4l2-ioctl.c: add V4L2_PIX_FMT_FWHT_STATELESS to v4l_fill_fmtdesc
Hans Verkuil [Wed, 13 Mar 2019 14:44:20 +0000 (10:44 -0400)]
media: v4l2-ioctl.c: add V4L2_PIX_FMT_FWHT_STATELESS to v4l_fill_fmtdesc

Add V4L2_PIX_FMT_FWHT_STATELESS to the list of pixelformats that
v4l_fill_fmtdesc() understands.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: Introducing stateless fwht defs and structs
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:40 +0000 (16:13 -0500)]
media: vicodec: Introducing stateless fwht defs and structs

Add structs and definitions needed to implement stateless
decoder for fwht and add I/P-frames QP controls to the
public api.

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: add documentation to V4L2_PIX_FMT_FWHT_STATELESS
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:39 +0000 (16:13 -0500)]
media: vicodec: add documentation to V4L2_PIX_FMT_FWHT_STATELESS

add documentation to V4L2_PIX_FMT_FWHT_STATELESS
in pixfmt-compressed.rst

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: add documentation to V4L2_CID_MPEG_VIDEO_FWHT_PARAMS
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:38 +0000 (16:13 -0500)]
media: vicodec: add documentation to V4L2_CID_MPEG_VIDEO_FWHT_PARAMS

add documentation to V4L2_CID_MPEG_VIDEO_FWHT_PARAMS
control and its related 'v4l2_ctrl_fwht_params' struct

[mchehab+samsung@kernel.org: remove extra blank lines]
Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: add documentation to V4L2_CID_FWHT_I/P_FRAME_QP
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:37 +0000 (16:13 -0500)]
media: vicodec: add documentation to V4L2_CID_FWHT_I/P_FRAME_QP

add documentation to V4L2_CID_FWHT_I/P_FRAME_QP
controls in ext-ctrls-codec.rst

[mchehab+samsung@kernel.org: remove extra blank lines]
Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: add struct for encoder/decoder instance
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:36 +0000 (16:13 -0500)]
media: vicodec: add struct for encoder/decoder instance

Add struct 'vicodec_dev_instance' for the fields in vicodec_dev
that have have both decoder and encoder versions.

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: Handle the case that the reference buffer is NULL
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:35 +0000 (16:13 -0500)]
media: vicodec: Handle the case that the reference buffer is NULL

In the stateless decoder the reference buffer is null if the
frame is an I-frame (flagged with FWHT_FL_I_FRAME).
Make sure not to dereference it in that case.

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: rename v4l2_fwht_default_fmt to v4l2_fwht_find_nth_fmt
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:34 +0000 (16:13 -0500)]
media: vicodec: rename v4l2_fwht_default_fmt to v4l2_fwht_find_nth_fmt

Rename 'v4l2_fwht_default_fmt' to 'v4l2_fwht_find_nth_fmt'
and add a function 'v4l2_fwht_validate_fmt' to check if
a format info matches the parameters.
This function will also be used to validate the stateless
params when adding support for stateless codecs.

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: Validate version dependent header values in a separate function
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:33 +0000 (16:13 -0500)]
media: vicodec: Validate version dependent header values in a separate function

Move the code that validates version dependent header
values to a separate function 'validate_by_version'

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: keep the ref frame according to the format in decoder
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:32 +0000 (16:13 -0500)]
media: vicodec: keep the ref frame according to the format in decoder

In the decoder, save the inner reference frame in the same
format as the capture buffer.
The decoder writes directly to the capture buffer and then
the capture buffer is copied to the reference buffer.
This will simplify the stateless decoder.

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: add field 'buf' to fwht_raw_frame
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:31 +0000 (16:13 -0500)]
media: vicodec: add field 'buf' to fwht_raw_frame

Add the field 'buf' to fwht_raw_frame to indicate
the start of the raw frame buffer.
This field will be used to copy the capture buffer
to the reference buffer in the next patch.

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: Move raw frame preparation code to a function
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:30 +0000 (16:13 -0500)]
media: vicodec: Move raw frame preparation code to a function

Introduce 'prepare_raw_frame' function that fills the values
of a raw frame struct according to the format.

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: bugfix: free compressed_frame upon device release
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:29 +0000 (16:13 -0500)]
media: vicodec: bugfix: free compressed_frame upon device release

Free compressed_frame buffer upon device release.

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: bugfix - call v4l2_m2m_buf_copy_metadata also if decoding fails
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:28 +0000 (16:13 -0500)]
media: vicodec: bugfix - call v4l2_m2m_buf_copy_metadata also if decoding fails

The function 'v4l2_m2m_buf_copy_metadata' should
be called even if decoding/encoding ends with
status VB2_BUF_STATE_ERROR, so that the metadata
is copied from the source buffer to the dest buffer.

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: change variable name for the return value of v4l2_fwht_encode
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:27 +0000 (16:13 -0500)]
media: vicodec: change variable name for the return value of v4l2_fwht_encode

v4l2_fwht_encode returns either an error code on
failure or the size of the compressed frame on
success. So change the var assigned to it from
'ret' to 'comp_sz_or_errcode' to clarify that.

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: v4l2-ctrl: v4l2_ctrl_request_setup returns with error upon failure
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:26 +0000 (16:13 -0500)]
media: v4l2-ctrl: v4l2_ctrl_request_setup returns with error upon failure

If one of the controls fails to set,
then 'v4l2_ctrl_request_setup'
immediately returns with the error code.

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Reviewed-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: upon release, call m2m release before freeing ctrl handler
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:25 +0000 (16:13 -0500)]
media: vicodec: upon release, call m2m release before freeing ctrl handler

'v4l2_m2m_ctx_release' calls request complete
so it should be called before 'v4l2_ctrl_handler_free'.

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: fix g_selection: either handle crop or compose
Hans Verkuil [Tue, 12 Mar 2019 11:53:55 +0000 (07:53 -0400)]
media: vicodec: fix g_selection: either handle crop or compose

The logic of g_selection was wrong: encoders support crop,
decoders support compose, but the code allowed both.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: selection api should only check single buffer types
Dafna Hirschfeld [Wed, 6 Mar 2019 21:13:24 +0000 (16:13 -0500)]
media: vicodec: selection api should only check single buffer types

The selection api should check only single buffer types
because multiplanar types are converted to
single in drivers/media/v4l2-core/v4l2-ioctl.c

Signed-off-by: Dafna Hirschfeld <dafna3@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: media requests: return EBADR instead of EACCES
Hans Verkuil [Wed, 20 Mar 2019 14:31:18 +0000 (10:31 -0400)]
media: media requests: return EBADR instead of EACCES

If requests are used when they shouldn't, or not used when they should,
then return EBADR (Invalid request descriptor) instead of EACCES.

The reason for this change is that EACCES has more to do with permissions
(not being the owner of the resource), but in this case the request file
descriptor is just wrong for the current mode of the device.

Update the documentation accordingly.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: cedrus: set requires_requests
Hans Verkuil [Wed, 6 Mar 2019 21:13:23 +0000 (16:13 -0500)]
media: cedrus: set requires_requests

The cedrus stateless decoder requires the use of request, so
indicate this by setting requires_requests to 1.

Note that the cedrus driver never checked for this, and as far
as I can tell would just crash if an attempt was made to queue
a buffer without a request.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Acked-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vb2: add requires_requests bit for stateless codecs
Hans Verkuil [Wed, 6 Mar 2019 21:13:21 +0000 (16:13 -0500)]
media: vb2: add requires_requests bit for stateless codecs

Stateless codecs require the use of the Request API as opposed of it
being optional.

So add a bit to indicate this and let vb2 check for this.

If an attempt is made to queue a buffer without an associated request,
then the EBADR error is returned to userspace.

Doing this check in the vb2 core simplifies drivers, since they
don't have to check for this, they can just set this flag.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reviewed-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: Kconfig files: use the right help coding style
Mauro Carvalho Chehab [Wed, 20 Mar 2019 10:39:44 +0000 (06:39 -0400)]
media: Kconfig files: use the right help coding style

Checkpatch wants to use 'help' instead of '---help---':

WARNING: prefer 'help' over '---help---' for new help texts

Let's change it globally at the media subsystem, as otherwise people
would keep using the old way.

Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: v4l2-fwnode: Add a deprecation note in the old ACPI parsing example
Sakari Ailus [Tue, 19 Mar 2019 08:07:32 +0000 (04:07 -0400)]
media: v4l2-fwnode: Add a deprecation note in the old ACPI parsing example

This is not how ACPI tables are written. Add a deprecation note and refer
to the proper documentation.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: ipu3-cio2: Set CSI-2 receiver sub-device entity function
Sakari Ailus [Mon, 18 Mar 2019 15:44:32 +0000 (11:44 -0400)]
media: ipu3-cio2: Set CSI-2 receiver sub-device entity function

Set the entity function for the four CSI-2 receiver sub-devices the driver
creates. This avoids a kernel warning from each as well.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Reviewed-by: Bingbu Cao <bingbu.cao@intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: v4l: i2c: Regroup lens drivers under their own section
Sakari Ailus [Mon, 18 Mar 2019 15:21:13 +0000 (11:21 -0400)]
media: v4l: i2c: Regroup lens drivers under their own section

The lens drivers had ended up under the video decoder section; add a new
one just for them, between the camera sensors and flash drivers.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: ti-vpe: Parse local endpoint for properties, not the remote one
Sakari Ailus [Sat, 2 Dec 2017 20:36:01 +0000 (15:36 -0500)]
media: ti-vpe: Parse local endpoint for properties, not the remote one

ti-vpe driver parsed the remote endpoints for properties but ignored the
local ones. Fix this by parsing the local endpoint properties instead.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: pxa-camera: Match with device node, not the port node
Sakari Ailus [Sat, 2 Dec 2017 20:29:46 +0000 (15:29 -0500)]
media: pxa-camera: Match with device node, not the port node

V4L2 fwnode matching right now still works based on device nodes, not port
nodes. Fix this.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: v4l2-fwnode: The first default data lane is 0 on C-PHY
Sakari Ailus [Sat, 2 Mar 2019 15:23:12 +0000 (10:23 -0500)]
media: v4l2-fwnode: The first default data lane is 0 on C-PHY

C-PHY has no clock lanes. Therefore the first data lane is 0 by default.

Fixes: edc6d56c2e7e ("media: v4l: fwnode: Support parsing of CSI-2 C-PHY endpoints")
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: v4l2-fwnode: Defaults may not override endpoint configuration in firmware
Sakari Ailus [Fri, 1 Mar 2019 13:48:38 +0000 (08:48 -0500)]
media: v4l2-fwnode: Defaults may not override endpoint configuration in firmware

The lack of defaults provided by the caller to
v4l2_fwnode_endpoint_parse() signals the use of the default lane mapping.
The default lane mapping must not be used however if the firmmare contains
the lane mapping. Disable the default lane mapping in that case, and
improve the debug messages telling of the use of the defaults.

This was missed previously since the default mapping will only unsed in
this case if the bus type is set, and no driver did both while still
needing the lane mapping configuration.

Fixes: b4357d21d674 ("media: v4l: fwnode: Support default CSI-2 lane mapping for drivers")
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: ov7670: don't access registers when the device is powered off
Akinobu Mita [Mon, 11 Mar 2019 15:36:03 +0000 (11:36 -0400)]
media: ov7670: don't access registers when the device is powered off

Since commit 3d6a8fe25605 ("media: ov7670: hook s_power onto v4l2 core"),
the device is actually powered off while the video stream is stopped.

So now set_format and s_frame_interval could be called while the device
is powered off, but these callbacks try to change the register settings
at this time.

The frame format and framerate will be restored right after power-up, so
we can just postpone applying these changes at these callbacks if the
device is not powered up.

Fixes: 3d6a8fe25605 ("media: ov7670: hook s_power onto v4l2 core")
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Reviewed-by: Lubomir Rintel <lkundrak@v3.sk>
Tested-by: Lubomir Rintel <lkundrak@v3.sk>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: ov7670: restore default settings after power-up
Akinobu Mita [Mon, 11 Mar 2019 15:36:02 +0000 (11:36 -0400)]
media: ov7670: restore default settings after power-up

Since commit 3d6a8fe25605 ("media: ov7670: hook s_power onto v4l2 core"),
the device is actually powered off while the video stream is stopped.

The frame format and framerate are restored right after power-up, but
restoring the default register settings is forgotten.

Fixes: 3d6a8fe25605 ("media: ov7670: hook s_power onto v4l2 core")
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Reviewed-by: Lubomir Rintel <lkundrak@v3.sk>
Tested-by: Lubomir Rintel <lkundrak@v3.sk>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: staging/intel-ipu3: reduce kernel stack usage
Arnd Bergmann [Tue, 5 Mar 2019 13:26:29 +0000 (08:26 -0500)]
media: staging/intel-ipu3: reduce kernel stack usage

The imgu_css_queue structure is too large to be put on the kernel
stack, as we can see in 32-bit builds:

drivers/staging/media/ipu3/ipu3-css.c: In function 'imgu_css_fmt_try':
drivers/staging/media/ipu3/ipu3-css.c:1863:1: error: the frame size of 1172 bytes is larger than 1024 bytes [-Werror=frame-larger-than=]

By dynamically allocating this array, the stack usage goes down to an
acceptable 140 bytes for the same x86-32 configuration.

Fixes: f5f2e4273518 ("media: staging/intel-ipu3: Add css pipeline programming")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Cao, Bingbu <bingbu.cao@intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: staging/intel-ipu3: mark PM function as __maybe_unused
Arnd Bergmann [Mon, 4 Mar 2019 20:29:10 +0000 (15:29 -0500)]
media: staging/intel-ipu3: mark PM function as __maybe_unused

The imgu_rpm_dummy_cb() looks like an API misuse that is explained
in the comment above it. Aside from that, it also causes a warning
when power management support is disabled:

drivers/staging/media/ipu3/ipu3.c:794:12: error: 'imgu_rpm_dummy_cb' defined but not used [-Werror=unused-function]

The warning is at least easy to fix by marking the function as
__maybe_unused.

Fixes: 7fc7af649ca7 ("media: staging/intel-ipu3: Add imgu top level pci device driver")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: staging/intel-ipu3-v4l: reduce kernel stack usage
Arnd Bergmann [Mon, 4 Mar 2019 20:28:42 +0000 (15:28 -0500)]
media: staging/intel-ipu3-v4l: reduce kernel stack usage

The v4l2_pix_format_mplane structure is too large to be put on the kernel
stack, as we can see in 32-bit builds:

drivers/staging/media/ipu3/ipu3-v4l2.c: In function 'imgu_fmt':
drivers/staging/media/ipu3/ipu3-v4l2.c:753:1: error: the frame size of 1028 bytes is larger than 1024 bytes [-Werror=frame-larger-than=]

By dynamically allocating this array, the stack usage goes down to an
acceptable 272 bytes for the same x86-32 configuration.

Fixes: a0ca1627b450 ("media: staging/intel-ipu3: Add v4l2 driver based on media framework")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: reset last_src/dst_buf based on the IS_OUTPUT
Hans Verkuil [Wed, 13 Mar 2019 14:47:13 +0000 (10:47 -0400)]
media: vicodec: reset last_src/dst_buf based on the IS_OUTPUT

When start_streaming was called both last_src_buf and last_dst_buf
pointers were set to NULL, but this depends on whether the capture
or output queue starts streaming.

When decoding with resolution changes in between the capture queue
has to restart streaming whenever a resolution change occurs. And
that would reset last_src_buf as well, which causes a problem if
the decoder was stopped by the application. Since last_src_buf
is now NULL, the LAST flag is never set for the last capture
buffer.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: remove WARN_ON(1) from get_q_data()
Hans Verkuil [Wed, 13 Mar 2019 14:46:42 +0000 (10:46 -0400)]
media: vicodec: remove WARN_ON(1) from get_q_data()

Some functions like enum_fmt use the buffer type as was passed
from userspace, which might cause the switch to fall into the
default case. Just drop the WARN_ON(1) to avoid kernel log pollution.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: em28xx-input: make const array addr_list static
Colin Ian King [Fri, 8 Mar 2019 10:30:17 +0000 (05:30 -0500)]
media: em28xx-input: make const array addr_list static

Don't populate the array addr_list on the stack but instead make it
static. Makes the object code smaller by 20 bytes

Before:
   text    data     bss     dec     hex filename^M
  16929    3626     384   20939    51cb ../usb/em28xx/em28xx-input.o

After:
   text    data     bss     dec     hex filename^M
  16829    3706     384   20919    51b7 ../usb/em28xx/em28xx-input.o

(gcc version 8.3.0, aarch64)

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+samsung@kernel.org>
5 years agomedia: dvb: Add support for the Avermedia TD310
Jose Alberto Reguero [Thu, 7 Mar 2019 23:12:18 +0000 (18:12 -0500)]
media: dvb: Add support for the Avermedia TD310

This patch add support for Avermedia TD310 usb stick.

Signed-off-by: Jose Alberto Reguero <jose.alberto.reguero@gmail.com>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: dvb: init i2c already in it930x_frontend_attach
Andreas Kemnade [Thu, 7 Mar 2019 23:11:58 +0000 (18:11 -0500)]
media: dvb: init i2c already in it930x_frontend_attach

i2c bus is already needed when the frontend is probed, so init it already
in it930x_frontend_attach. That prevents errors like:

si2168: probe of 6-0067 failed with error -5

Signed-off-by: Andreas Kemnade <andreas@kemnade.info>
Signed-off-by: Jose Alberto Reguero <jose.alberto.reguero@gmail.com>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: si2165: fix a missing check of return value
Kangjie Lu [Fri, 21 Dec 2018 04:54:03 +0000 (23:54 -0500)]
media: si2165: fix a missing check of return value

si2165_readreg8() may fail. Looking into si2165_readreg8(), we will find
that "val_tmp" will be an uninitialized value when regmap_read() fails.
"val_tmp" is then assigned to "val". So if si2165_readreg8() fails,
"val" will be a random value. Further use will lead to undefined
behaviors. The fix checks if si2165_readreg8() fails, and if so, returns
its error code upstream.

Signed-off-by: Kangjie Lu <kjlu@umn.edu>
Reviewed-by: Matthias Schwarzott <zzam@gentoo.org>
Tested-by: Matthias Schwarzott <zzam@gentoo.org>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: serial_ir: Fix use-after-free in serial_ir_init_module
YueHaibing [Tue, 5 Mar 2019 05:40:26 +0000 (00:40 -0500)]
media: serial_ir: Fix use-after-free in serial_ir_init_module

Syzkaller report this:

BUG: KASAN: use-after-free in sysfs_remove_file_ns+0x5f/0x70 fs/sysfs/file.c:468
Read of size 8 at addr ffff8881dc7ae030 by task syz-executor.0/6249

CPU: 1 PID: 6249 Comm: syz-executor.0 Not tainted 5.0.0-rc8+ #3
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014
Call Trace:
 __dump_stack lib/dump_stack.c:77 [inline]
 dump_stack+0xfa/0x1ce lib/dump_stack.c:113
 print_address_description+0x65/0x270 mm/kasan/report.c:187
 kasan_report+0x149/0x18d mm/kasan/report.c:317
 ? 0xffffffffc1728000
 sysfs_remove_file_ns+0x5f/0x70 fs/sysfs/file.c:468
 sysfs_remove_file include/linux/sysfs.h:519 [inline]
 driver_remove_file+0x40/0x50 drivers/base/driver.c:122
 remove_bind_files drivers/base/bus.c:585 [inline]
 bus_remove_driver+0x186/0x220 drivers/base/bus.c:725
 driver_unregister+0x6c/0xa0 drivers/base/driver.c:197
 serial_ir_init_module+0x169/0x1000 [serial_ir]
 do_one_initcall+0xfa/0x5ca init/main.c:887
 do_init_module+0x204/0x5f6 kernel/module.c:3460
 load_module+0x66b2/0x8570 kernel/module.c:3808
 __do_sys_finit_module+0x238/0x2a0 kernel/module.c:3902
 do_syscall_64+0x147/0x600 arch/x86/entry/common.c:290
 entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x462e99
Code: f7 d8 64 89 02 b8 ff ff ff ff c3 66 0f 1f 44 00 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 bc ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007f9450132c58 EFLAGS: 00000246 ORIG_RAX: 0000000000000139
RAX: ffffffffffffffda RBX: 000000000073bf00 RCX: 0000000000462e99
RDX: 0000000000000000 RSI: 0000000020000100 RDI: 0000000000000003
RBP: 00007f9450132c70 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 00007f94501336bc
R13: 00000000004bcefa R14: 00000000006f6fb0 R15: 0000000000000004

Allocated by task 6249:
 set_track mm/kasan/common.c:85 [inline]
 __kasan_kmalloc.constprop.3+0xa0/0xd0 mm/kasan/common.c:495
 kmalloc include/linux/slab.h:545 [inline]
 kzalloc include/linux/slab.h:740 [inline]
 bus_add_driver+0xc0/0x610 drivers/base/bus.c:651
 driver_register+0x1bb/0x3f0 drivers/base/driver.c:170
 serial_ir_init_module+0xe8/0x1000 [serial_ir]
 do_one_initcall+0xfa/0x5ca init/main.c:887
 do_init_module+0x204/0x5f6 kernel/module.c:3460
 load_module+0x66b2/0x8570 kernel/module.c:3808
 __do_sys_finit_module+0x238/0x2a0 kernel/module.c:3902
 do_syscall_64+0x147/0x600 arch/x86/entry/common.c:290
 entry_SYSCALL_64_after_hwframe+0x49/0xbe

Freed by task 6249:
 set_track mm/kasan/common.c:85 [inline]
 __kasan_slab_free+0x130/0x180 mm/kasan/common.c:457
 slab_free_hook mm/slub.c:1430 [inline]
 slab_free_freelist_hook mm/slub.c:1457 [inline]
 slab_free mm/slub.c:3005 [inline]
 kfree+0xe1/0x270 mm/slub.c:3957
 kobject_cleanup lib/kobject.c:662 [inline]
 kobject_release lib/kobject.c:691 [inline]
 kref_put include/linux/kref.h:67 [inline]
 kobject_put+0x146/0x240 lib/kobject.c:708
 bus_remove_driver+0x10e/0x220 drivers/base/bus.c:732
 driver_unregister+0x6c/0xa0 drivers/base/driver.c:197
 serial_ir_init_module+0x14c/0x1000 [serial_ir]
 do_one_initcall+0xfa/0x5ca init/main.c:887
 do_init_module+0x204/0x5f6 kernel/module.c:3460
 load_module+0x66b2/0x8570 kernel/module.c:3808
 __do_sys_finit_module+0x238/0x2a0 kernel/module.c:3902
 do_syscall_64+0x147/0x600 arch/x86/entry/common.c:290
 entry_SYSCALL_64_after_hwframe+0x49/0xbe

The buggy address belongs to the object at ffff8881dc7ae000
 which belongs to the cache kmalloc-256 of size 256
The buggy address is located 48 bytes inside of
 256-byte region [ffff8881dc7ae000ffff8881dc7ae100)
The buggy address belongs to the page:
page:ffffea000771eb80 count:1 mapcount:0 mapping:ffff8881f6c02e00 index:0x0
flags: 0x2fffc0000000200(slab)
raw: 02fffc0000000200 ffffea0007d14800 0000000400000002 ffff8881f6c02e00
raw: 0000000000000000 00000000800c000c 00000001ffffffff 0000000000000000
page dumped because: kasan: bad access detected

Memory state around the buggy address:
 ffff8881dc7adf00: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 ffff8881dc7adf80: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
>ffff8881dc7ae000: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
                                     ^
 ffff8881dc7ae080: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
 ffff8881dc7ae100: fc fc fc fc fc fc fc fc 00 00 00 00 00 00 00 00

There are already cleanup handlings in serial_ir_init error path,
no need to call serial_ir_exit do it again in serial_ir_init_module,
otherwise will trigger a use-after-free issue.

Fixes: fa5dc29c1fcc ("[media] lirc_serial: move out of staging and rename to serial_ir")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: rc: remove unused including <linux/version.h>
YueHaibing [Thu, 21 Feb 2019 01:32:55 +0000 (20:32 -0500)]
media: rc: remove unused including <linux/version.h>

Remove including <linux/version.h> that don't need it.

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: cx23885: check allocation return
Nicholas Mc Guire [Sun, 20 Jan 2019 03:52:23 +0000 (22:52 -0500)]
media: cx23885: check allocation return

Checking of kmalloc() seems to have been committed - as
cx23885_dvb_register() is checking for != 0 return, returning
-ENOMEM should be fine here.  While at it address the coccicheck
suggestion to move to kmemdup rather than using kmalloc+memcpy.

Fixes: 46b21bbaa8a8 ("[media] Add support for DViCO FusionHDTV DVB-T Dual Express2")
Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: m88ds3103: serialize reset messages in m88ds3103_set_frontend
James Hutchinson [Sun, 13 Jan 2019 21:13:47 +0000 (16:13 -0500)]
media: m88ds3103: serialize reset messages in m88ds3103_set_frontend

Ref: https://bugzilla.kernel.org/show_bug.cgi?id=199323

Users are experiencing problems with the DVBSky S960/S960C USB devices
since the following commit:

9d659ae: ("locking/mutex: Add lock handoff to avoid starvation")

The device malfunctions after running for an indeterminable period of
time, and the problem can only be cleared by rebooting the machine.

It is possible to encourage the problem to surface by blocking the
signal to the LNB.

Further debugging revealed the cause of the problem.

In the following capture:
- thread #1325 is running m88ds3103_set_frontend
- thread #42 is running ts2020_stat_work

a> [1325] usb 1-1: dvb_usb_v2_generic_io: >>> 08 68 02 07 80
   [1325] usb 1-1: dvb_usb_v2_generic_io: <<< 08
   [42] usb 1-1: dvb_usb_v2_generic_io: >>> 09 01 01 68 3f
   [42] usb 1-1: dvb_usb_v2_generic_io: <<< 08 ff
   [42] usb 1-1: dvb_usb_v2_generic_io: >>> 08 68 02 03 11
   [42] usb 1-1: dvb_usb_v2_generic_io: <<< 07
   [42] usb 1-1: dvb_usb_v2_generic_io: >>> 09 01 01 60 3d
   [42] usb 1-1: dvb_usb_v2_generic_io: <<< 07 ff
b> [1325] usb 1-1: dvb_usb_v2_generic_io: >>> 08 68 02 07 00
   [1325] usb 1-1: dvb_usb_v2_generic_io: <<< 07
   [42] usb 1-1: dvb_usb_v2_generic_io: >>> 08 68 02 03 11
   [42] usb 1-1: dvb_usb_v2_generic_io: <<< 07
   [42] usb 1-1: dvb_usb_v2_generic_io: >>> 09 01 01 60 21
   [42] usb 1-1: dvb_usb_v2_generic_io: <<< 07 ff
   [42] usb 1-1: dvb_usb_v2_generic_io: >>> 08 68 02 03 11
   [42] usb 1-1: dvb_usb_v2_generic_io: <<< 07
   [42] usb 1-1: dvb_usb_v2_generic_io: >>> 09 01 01 60 66
   [42] usb 1-1: dvb_usb_v2_generic_io: <<< 07 ff
   [1325] usb 1-1: dvb_usb_v2_generic_io: >>> 08 68 02 03 11
   [1325] usb 1-1: dvb_usb_v2_generic_io: <<< 07
   [1325] usb 1-1: dvb_usb_v2_generic_io: >>> 08 60 02 10 0b
   [1325] usb 1-1: dvb_usb_v2_generic_io: <<< 07

Two i2c messages are sent to perform a reset in m88ds3103_set_frontend:

  a. 0x07, 0x80
  b. 0x07, 0x00

However, as shown in the capture, the regmap mutex is being handed over
to another thread (ts2020_stat_work) in between these two messages.

>From here, the device responds to every i2c message with an 07 message,
and will only return to normal operation following a power cycle.

Use regmap_multi_reg_write to group the two reset messages, ensuring
both are processed before the regmap mutex is unlocked.

Signed-off-by: James Hutchinson <jahutchinson99@googlemail.com>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: dvbsky: Avoid leaking dvb frontend
Stefan Brüns [Sun, 20 Jan 2019 01:30:04 +0000 (20:30 -0500)]
media: dvbsky: Avoid leaking dvb frontend

Commit 14f4eaeddabc ("media: dvbsky: fix driver unregister logic") fixed
a use-after-free by removing the reference to the frontend after deleting
the backing i2c device.

This has the unfortunate side effect the frontend device is never freed
in the dvb core leaving a dangling device, leading to errors when the
dvb core tries to register the frontend after e.g. a replug as reported
here: https://www.spinics.net/lists/linux-media/msg138181.html

media: dvbsky: issues with DVBSky T680CI

===
[  561.119145] sp2 8-0040: CIMaX SP2 successfully attached
[  561.119161] usb 2-3: DVB: registering adapter 0 frontend 0 (Silicon Labs
Si2168)...
[  561.119174] sysfs: cannot create duplicate filename '/class/dvb/
dvb0.frontend0'
===

The use after free happened as dvb_usbv2_disconnect calls in this order:
- dvb_usb_device::props->exit(...)
- dvb_usbv2_adapter_frontend_exit(...)
  + if (fe) dvb_unregister_frontend(fe)
  + dvb_usb_device::props->frontend_detach(...)

Moving the release of the i2c device from exit() to frontend_detach()
avoids the dangling pointer access and allows the core to unregister
the frontend.

This was originally reported for a DVBSky T680CI, but it also affects
the MyGica T230C. As all supported devices structure the registration/
unregistration identically, apply the change for all device types.

Signed-off-by: Stefan Brüns <stefan.bruens@rwth-aachen.de>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>