linux-2.6-microblaze.git
5 years agomedia: tda7432: fix spelling mistake "maximium" -> "maximum"
Colin Ian King [Wed, 28 Nov 2018 15:34:09 +0000 (10:34 -0500)]
media: tda7432: fix spelling mistake "maximium" -> "maximum"

There is a spelling mistake in the module description as well
as a comment. Fix them.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vivid: add req_validate error injection
Hans Verkuil [Wed, 28 Nov 2018 09:11:52 +0000 (04:11 -0500)]
media: vivid: add req_validate error injection

Add a new vivid button control to inject an error into the
req_validate request callback.

This will help testing with v4l2-compliance.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: adv7604 added include of linux/interrupt.h
Jasmin Jessich [Mon, 26 Nov 2018 22:01:09 +0000 (17:01 -0500)]
media: adv7604 added include of linux/interrupt.h

On older Kernels (prior to 4.15) irqreturn_t and devm_request_threaded_irq
is not defined when compiling adv7604.c. It seems more recent Kernels
include it via another header which is included by adv7604.c.
Now we include linux/interrupt.h explicitly to get the type also defined
for Kernels prior to 4.15.

Signed-off-by: Jasmin Jessich <jasmin@anw.at>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: v4l2-pci-skeleton: depend on CONFIG_SAMPLES
Michael Tretter [Mon, 26 Nov 2018 18:01:23 +0000 (13:01 -0500)]
media: v4l2-pci-skeleton: depend on CONFIG_SAMPLES

Commit 0185f8501762 ("[media] samples: v4l: from Documentation to
samples directory") moved the v4l2-pci-skeleton driver to the samples
directory. The samples are only be built, if CONFIG_SAMPLES is enabled.

Therefore, VIDEO_PCI_SKELETON is not enough to build the
v4l2-pci-skeleton driver, but SAMPLES needs to be enabled, too. Let
VIDEO_PCI_SKELETON depend on SAMPLES.

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+samsung@kernel.org>
5 years agomedia: v4l2-pci-skeleton: replace vb2_buffer with vb2_v4l2_buffer
Michael Tretter [Mon, 26 Nov 2018 18:01:24 +0000 (13:01 -0500)]
media: v4l2-pci-skeleton: replace vb2_buffer with vb2_v4l2_buffer

Commit 2d7007153f0c ("[media] media: videobuf2: Restructure vb2_buffer")
replaced vb2_buffer with vb2_v4l2_buffer in all v4l2 drivers. The
restructuring skipped the v4l2-pci-skeleton, probably because it resides
outside the drivers directory.

The v4l2_buf_ops assume that the passed buffer is a vb2_v4l2_buffer.
This is not the case if the skel_buffer is based on vb2_buffer instead
of vb2_v4l2_buffer.

Replace vb2_buffer with vb2_v4l2_buffer in the skeleton to make sure
that future drivers that are based on the skeleton use vb2_v4l2_buffer.

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+samsung@kernel.org>
5 years agomedia: v4l: ioctl: Allow drivers to fill in the format description
Sakari Ailus [Mon, 26 Nov 2018 11:07:29 +0000 (06:07 -0500)]
media: v4l: ioctl: Allow drivers to fill in the format description

The v4l_fill_fmtdesc() function does a good job in filling in pixelformat
description. While generally all drivers should depend on this function
doing the job, staging drivers that use their own formats may not.

Allow staging drivers to fill in their own formats by checking whether the
description begins with a non-nil character before issuing the warning.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Suggested-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: v4l: Fix MPEG-2 slice Intra DC Precision validation
Jonas Karlman [Sun, 25 Nov 2018 15:21:42 +0000 (10:21 -0500)]
media: v4l: Fix MPEG-2 slice Intra DC Precision validation

intra_dc_precision is a 2-bit integer [1]
allow use of all valid options, 8 - 11 bits precision

[1] ISO/IEC 13818-2 Table 6-13

Fixes: c27bb30e7b6d ("media: v4l: Add definitions for MPEG-2 slice format and metadata")

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Acked-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: DaVinci-VPBE: fix error handling in vpbe_initialize()
Alexey Khoroshilov [Fri, 23 Nov 2018 21:56:26 +0000 (16:56 -0500)]
media: DaVinci-VPBE: fix error handling in vpbe_initialize()

If vpbe_set_default_output() or vpbe_set_default_mode() fails,
vpbe_initialize() returns error code without releasing resources.

The patch adds error handling for that case.

Found by Linux Driver Verification project (linuxtesting.org).

Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vivid: fix smatch warnings
Hans Verkuil [Fri, 23 Nov 2018 12:07:33 +0000 (07:07 -0500)]
media: vivid: fix smatch warnings

Reorganize code to fix two smatch warnings:

drivers/media/platform/vivid/vivid-core.c: drivers/media/platform/vivid/vivid-core.c:889 vivid_create_instance() warn: potentially one past the end of array
'dev->query_dv_timings_qmenu[dev->query_dv_timings_size]'
drivers/media/platform/vivid/vivid-core.c: drivers/media/platform/vivid/vivid-core.c:889 vivid_create_instance() warn: potentially one past the end of array
'dev->query_dv_timings_qmenu[dev->query_dv_timings_size]'

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 VA from encoder frame buffers
Tomasz Figa [Wed, 21 Nov 2018 06:37:59 +0000 (01:37 -0500)]
media: mtk-vcodec: Remove VA from encoder frame buffers

The encoder driver has no need to do any CPU access to the source frame
buffers. Use a separate structure for holding DMA addresses and sizes
for those and remove, so we do not end up introducing any erroneous
dereferences of those VAs.

This fixes DMA-buf import from exporters that do not provide contiguous
kernel mappings, which includes the MTK DRM driver.

Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: camss: Take in account sensor skip frames
Todor Tomov [Tue, 13 Nov 2018 14:03:07 +0000 (09:03 -0500)]
media: camss: Take in account sensor skip frames

When streaming is starting ask the sensor for its skip frames value.
Max supported frame skip is 29 frames, so clip it if it is higher.

Signed-off-by: Todor Tomov <todor.tomov@linaro.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: adv7180: add g_skip_frames support
Tim Harvey [Wed, 24 Oct 2018 20:25:51 +0000 (16:25 -0400)]
media: adv7180: add g_skip_frames support

The adv7180 produces 1 to 2 frames of garbage before proper sync is
established. This allows V4L2 drivers and apps to skip those.

Signed-off-by: Tim Harvey <tharvey@gateworks.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: em28xx: fix spelling mistake, "Cinnergy" -> "Cinergy"
Colin Ian King [Fri, 12 Oct 2018 15:37:17 +0000 (11:37 -0400)]
media: em28xx: fix spelling mistake, "Cinnergy" -> "Cinergy"

Fix name of the Hybrid T USB XS em28xx card, should be Cinergy.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: exynos4-is: fix spelling mistake ACTURATOR -> ACTUATOR
Colin Ian King [Fri, 12 Oct 2018 14:48:32 +0000 (10:48 -0400)]
media: exynos4-is: fix spelling mistake ACTURATOR -> ACTUATOR

Trivial fix to spelling mistake in macro name and text string
ERROR_SENSOR_ACTURATOR_INIT_FAIL -> ERROR_SENSOR_ACTUATOR_INIT_FAIL

Signed-off-by: Colin Ian King <colin.king@canonical.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 CONFIG_HDMI
Arnd Bergmann [Tue, 2 Oct 2018 21:12:43 +0000 (17:12 -0400)]
media: i2c: TDA1997x: select CONFIG_HDMI

Without CONFIG_HDMI, we get a link error for this driver:

drivers/media/i2c/tda1997x.o: In function `tda1997x_parse_infoframe':
tda1997x.c:(.text+0x2195): undefined reference to `hdmi_infoframe_unpack'
tda1997x.c:(.text+0x21b6): undefined reference to `hdmi_infoframe_log'
drivers/media/i2c/tda1997x.o: In function `tda1997x_log_infoframe':
tda1997x.c:(.text.unlikely+0x13d3): undefined reference to `hdmi_infoframe_unpack'
tda1997x.c:(.text.unlikely+0x1426): undefined reference to `hdmi_infoframe_log'

All other drivers in this directory that use HDMI select CONFIG_HDMI,
so do the same here:

Fixes: 9ac0038db9a7 ("media: i2c: Add TDA1997x HDMI receiver driver")

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: uvcvideo: Refactor teardown of uvc on USB disconnect
Daniel Axtens [Sun, 23 Apr 2017 04:53:49 +0000 (00:53 -0400)]
media: uvcvideo: Refactor teardown of uvc on USB disconnect

Currently, disconnecting a USB webcam while it is in use prints out a
number of warnings, such as:

WARNING: CPU: 2 PID: 3118 at /build/linux-ezBi1T/linux-4.8.0/fs/sysfs/group.c:237 sysfs_remove_group+0x8b/0x90
sysfs group ffffffffa7cd0780 not found for kobject 'event13'

This has been noticed before. [0]

This is because of the order in which things are torn down.

If there are no streams active during a USB disconnect:

 - uvc_disconnect() is invoked via device_del() through the bus
   notifier mechanism.

 - this calls uvc_unregister_video().

 - uvc_unregister_video() unregisters the video device for each
   stream,

 - because there are no streams open, it calls uvc_delete()

 - uvc_delete() calls uvc_status_cleanup(), which cleans up the status
   input device.

 - uvc_delete() calls media_device_unregister(), which cleans up the
   media device

 - uvc_delete(), uvc_unregister_video() and uvc_disconnect() all
   return, and we end up back in device_del().

 - device_del() then cleans up the sysfs folder for the camera with
   dpm_sysfs_remove(). Because uvc_status_cleanup() and
   media_device_unregister() have already been called, this all works
   nicely.

If, on the other hand, there *are* streams active during a USB disconnect:

 - uvc_disconnect() is invoked

 - this calls uvc_unregister_video()

 - uvc_unregister_video() unregisters the video device for each
   stream,

 - uvc_unregister_video() and uvc_disconnect() return, and we end up
   back in device_del().

 - device_del() then cleans up the sysfs folder for the camera with
   dpm_sysfs_remove(). Because the status input device and the media
   device are children of the USB device, this also deletes their
   sysfs folders.

 - Sometime later, the final stream is closed, invoking uvc_release().

 - uvc_release() calls uvc_delete()

 - uvc_delete() calls uvc_status_cleanup(), which cleans up the status
   input device. Because the sysfs directory has already been removed,
   this causes a WARNing.

 - uvc_delete() calls media_device_unregister(), which cleans up the
   media device. Because the sysfs directory has already been removed,
   this causes another WARNing.

To fix this, we need to make sure the devices are always unregistered
before the end of uvc_disconnect(). To this, move the unregistration
into the disconnect path:

 - split uvc_status_cleanup() into two parts, one on disconnect that
   unregisters and one on delete that frees.

 - move v4l2_device_unregister() and media_device_unregister() into
   the disconnect path.

[0]: https://lkml.org/lkml/2016/12/8/657

[Renamed uvc_input_cleanup() to uvc_input_unregister()]

Signed-off-by: Daniel Axtens <dja@axtens.net>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: uvcvideo: Add support for the CNF4 format
Sergey Dorodnicov [Wed, 12 Sep 2018 06:42:07 +0000 (02:42 -0400)]
media: uvcvideo: Add support for the CNF4 format

Register the GUID used by Intel RealSense cameras with fourcc CNF4,
encoding depth sensor confidence information for every pixel.

Signed-off-by: Sergey Dorodnicov <sergey.dorodnicov@intel.com>
Signed-off-by: Evgeni Raikhel <evgeni.raikhel@intel.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: v4l: Add 4bpp packed depth confidence format CNF4
Sergey Dorodnicov [Wed, 12 Sep 2018 06:42:06 +0000 (02:42 -0400)]
media: v4l: Add 4bpp packed depth confidence format CNF4

Adding new fourcc CNF4 for 4 bit-per-pixel packed depth confidence
information provided by Intel RealSense cameras. Every two consecutive
pixels are packed into a single byte.

Signed-off-by: Sergey Dorodnicov <sergey.dorodnicov@intel.com>
Signed-off-by: Evgeni Raikhel <evgeni.raikhel@intel.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: xilinx: fix typo in formats table
Andrea Merello [Fri, 28 Sep 2018 07:32:13 +0000 (03:32 -0400)]
media: xilinx: fix typo in formats table

In formats table the entry for CFA pattern "rggb" has GRBG fourcc.
This patch fixes it.

Signed-off-by: Mirco Di Salvo <mirco.disalvo@iit.it>
Signed-off-by: Andrea Merello <andrea.merello@gmail.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: xilinx: Use SPDX-License-Identifier
Dhaval Shah [Fri, 8 Dec 2017 12:35:37 +0000 (07:35 -0500)]
media: xilinx: Use SPDX-License-Identifier

SPDX-License-Identifier is used for the Xilinx Video IP and
related drivers.

[Added drivers/media/platform/xilinx/Kconfig]
[Added drivers/media/platform/xilinx/Makefile]
[Added include/dt-bindings/media/xilinx-vip.h]

Signed-off-by: Dhaval Shah <dhaval23031987@gmail.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: sony-cxd2880: add optional vcc regulator to bindings
Neil Armstrong [Thu, 8 Nov 2018 12:50:10 +0000 (07:50 -0500)]
media: sony-cxd2880: add optional vcc regulator to bindings

This patchset adds an optional VCC regulator to the bindings of the Sony
CXD2880 DVB-T2/T tuner + demodulator adapter.

Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Acked-by: Yasunari Takiguchi <Yasunari.Takiguchi@sony.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: cxd2880-spi: Add optional vcc regulator
Neil Armstrong [Thu, 8 Nov 2018 12:50:09 +0000 (07:50 -0500)]
media: cxd2880-spi: Add optional vcc regulator

This patchset adds an optional VCC regulator to the driver probe function
to make sure power is enabled to the module before starting attaching to
the device.

Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Acked-by: Yasunari Takiguchi <Yasunari.Takiguchi@sony.com>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: dib0900: fix smatch warnings
Hans Verkuil [Thu, 22 Nov 2018 10:03:01 +0000 (05:03 -0500)]
media: dib0900: fix smatch warnings

drivers/media/dvb-frontends/dib0090.c: drivers/media/dvb-frontends/dib0090.c:1075 dib0090_pwm_gain_reset() warn: '*&bb_ramp_pwm_normal' 2590696709486571520 can't fit into 65535 '*bb_ramp'
drivers/media/dvb-frontends/dib0090.c: drivers/media/dvb-frontends/dib0090.c:1083 dib0090_pwm_gain_reset() warn: '*&bb_ramp_pwm_normal_socs' 2590696709486571520 can't fit into 65535 '*bb_ramp'
drivers/media/dvb-frontends/dib0090.c: drivers/media/dvb-frontends/dib0090.c:1085 dib0090_pwm_gain_reset() warn: '*&rf_ramp_pwm_cband_8090' 2590696709486571520 can't fit into 65535 '*rf_ramp'
drivers/media/dvb-frontends/dib0090.c: drivers/media/dvb-frontends/dib0090.c:1089 dib0090_pwm_gain_reset() warn: '*&rf_ramp_pwm_cband_7090e_sensitivity' 2590696709486571520 can't fit into 65535 '*rf_ramp'
drivers/media/dvb-frontends/dib0090.c: drivers/media/dvb-frontends/dib0090.c:1093 dib0090_pwm_gain_reset() warn: '*&rf_ramp_pwm_cband_7090p' 2590696709486571520 can't fit into 65535 '*rf_ramp'
drivers/media/dvb-frontends/dib0090.c: drivers/media/dvb-frontends/dib0090.c:1096 dib0090_pwm_gain_reset() warn: '*&rf_ramp_pwm_cband' 2590696709486571520 can't fit into 65535 '*rf_ramp'
drivers/media/dvb-frontends/dib0090.c: drivers/media/dvb-frontends/dib0090.c:1101 dib0090_pwm_gain_reset() warn: '*&bb_ramp_pwm_normal_socs' 2590696709486571520 can't fit into 65535 '*bb_ramp'
drivers/media/dvb-frontends/dib0090.c: drivers/media/dvb-frontends/dib0090.c:1104 dib0090_pwm_gain_reset() warn: '*&rf_ramp_pwm_vhf' 2590696709486571520 can't fit into 65535 '*rf_ramp'
drivers/media/dvb-frontends/dib0090.c: drivers/media/dvb-frontends/dib0090.c:1107 dib0090_pwm_gain_reset() warn: '*&bb_ramp_pwm_normal_socs' 2590696709486571520 can't fit into 65535 '*bb_ramp'
drivers/media/dvb-frontends/dib0090.c: drivers/media/dvb-frontends/dib0090.c:1109 dib0090_pwm_gain_reset() warn: '*&rf_ramp_pwm_uhf_8090' 2590696709486571520 can't fit into 65535 '*rf_ramp'
drivers/media/dvb-frontends/dib0090.c: drivers/media/dvb-frontends/dib0090.c:1111 dib0090_pwm_gain_reset() warn: '*&rf_ramp_pwm_uhf_7090' 2590696709486571520 can't fit into 65535 '*rf_ramp'
drivers/media/dvb-frontends/dib0090.c: drivers/media/dvb-frontends/dib0090.c:1113 dib0090_pwm_gain_reset() warn: '*&rf_ramp_pwm_uhf' 2590696709486571520 can't fit into 65535 '*rf_ramp'
drivers/media/dvb-frontends/dib0090.c: drivers/media/dvb-frontends/dib0090.c:1419 dib0090_update_rframp_7090() warn: '*&rf_ramp_pwm_cband_7090e_sensitivity' 2590696709486571520 can't fit into 65535
'*state->rf_ramp'
drivers/media/dvb-frontends/dib0090.c: drivers/media/dvb-frontends/dib0090.c:1421 dib0090_update_rframp_7090() warn: '*&rf_ramp_pwm_cband_7090e_aci' 2590696709486571520 can't fit into 65535
'*state->rf_ramp'

For no apparent reason this code casts away the const of the const u16 arrays, and it
also takes the address of an array. While that's ignored in C I think smatch gets confused
by it.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: mxl5xx: constify dvb_frontend_ops structure
Julia Lawall [Sat, 27 Oct 2018 11:46:11 +0000 (07:46 -0400)]
media: mxl5xx: constify dvb_frontend_ops structure

The dvb_frontend_ops structure is only copied into the ops field
of a dvb_frontend structure, so it can be const.

Done with the help of Coccinelle.

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: dvb-frontends: Use kmemdup instead of duplicating its function
zhong jiang [Wed, 19 Sep 2018 08:16:09 +0000 (04:16 -0400)]
media: dvb-frontends: Use kmemdup instead of duplicating its function

kmemdup has implemented the function that kmalloc() + memcpy().
We prefer to kmemdup rather than code opened implementation.

Signed-off-by: zhong jiang <zhongjiang@huawei.com>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: usb: Use kmemdup instead of duplicating its function.
zhong jiang [Tue, 18 Sep 2018 15:25:25 +0000 (11:25 -0400)]
media: usb: Use kmemdup instead of duplicating its function.

kmemdup has implemented the function that kmalloc() + memcpy().
We prefer to kmemdup rather than code opened implementation.

Signed-off-by: zhong jiang <zhongjiang@huawei.com>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: cxd2880-spi: fix probe when dvb_attach fails
Neil Armstrong [Thu, 8 Nov 2018 12:44:48 +0000 (07:44 -0500)]
media: cxd2880-spi: fix probe when dvb_attach fails

When dvb_attach fails, probe returns 0, and remove crashes afterwards.
This patch sets the return value to -ENODEV when attach fails.

Fixes: bd24fcddf6b8 ("media: cxd2880-spi: Add support for CXD2880 SPI interface")

Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Acked-by: Yasunari Takiguchi <Yasunari.Takiguchi@sony.com>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: dvb: Use WARM definition from identify_state()
Victor Toso [Tue, 30 Oct 2018 16:14:49 +0000 (12:14 -0400)]
media: dvb: Use WARM definition from identify_state()

Device should be either COLD or WARM.
This change only make usage of the existing definition.

Signed-off-by: Victor Toso <me@victortoso.com>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: af9033: Remove duplicated switch statement
Victor Toso [Tue, 30 Oct 2018 16:14:48 +0000 (12:14 -0400)]
media: af9033: Remove duplicated switch statement

The switch before set is_af9035 or is_it9135 which makes the second
switch redundant. Keeping the comment as to avoid sleep on IT9135.

Signed-off-by: Victor Toso <me@victortoso.com>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: dib0700: fix spelling mistake "Amplifyer" -> "Amplifier"
Colin Ian King [Sun, 25 Nov 2018 22:51:14 +0000 (17:51 -0500)]
media: dib0700: fix spelling mistake "Amplifyer" -> "Amplifier"

There is a spelling mistake in the MODULE_PARM_DESC text, fix it.

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-usb-v2: Fix incorrect use of transfer_flags URB_FREE_BUFFER
Malcolm Priestley [Mon, 26 Nov 2018 20:18:25 +0000 (15:18 -0500)]
media: dvb-usb-v2: Fix incorrect use of transfer_flags URB_FREE_BUFFER

commit 1a0c10ed7bb1 ("media: dvb-usb-v2: stop using coherent memory for
URBs") incorrectly adds URB_FREE_BUFFER after every urb transfer.

It cannot use this flag because it reconfigures the URBs accordingly
to suit connected devices. In doing a call to usb_free_urb is made and
invertedly frees the buffers.

The stream buffer should remain constant while driver is up.

Signed-off-by: Malcolm Priestley <tvboxspy@gmail.com>
CC: stable@vger.kernel.org # v4.18+
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: saa7134: rc-core maintains users count, no need to duplicate
Sean Young [Sun, 25 Nov 2018 17:59:48 +0000 (12:59 -0500)]
media: saa7134: rc-core maintains users count, no need to duplicate

This simplifies the code a little. Tested with suspend and resume.

Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: venus: fix reported size of 0-length buffers
Alexandre Courbot [Tue, 13 Nov 2018 09:30:48 +0000 (04:30 -0500)]
media: venus: fix reported size of 0-length buffers

The last buffer is often signaled by an empty buffer with the
V4L2_BUF_FLAG_LAST flag set. Such buffers were returned with the
bytesused field set to the full size of the OPB, which leads
user-space to believe that the buffer actually contains useful data. Fix
this by passing the number of bytes reported used by the firmware.

Signed-off-by: Alexandre Courbot <acourbot@chromium.org>
Acked-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: venus: dynamic handling of bitrate
Malathi Gottam [Fri, 2 Nov 2018 12:41:16 +0000 (08:41 -0400)]
media: venus: dynamic handling of bitrate

Any request for a change in bitrate after both planes
are streamed on is handled by setting the target bitrate
property to hardware.

Signed-off-by: Malathi Gottam <mgottam@codeaurora.org>
Acked-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: venus: handle peak bitrate set property
Malathi Gottam [Mon, 22 Oct 2018 12:05:12 +0000 (08:05 -0400)]
media: venus: handle peak bitrate set property

Max bitrate property is not supported for venus version 4xx.
Return unsupported from packetization layer. Handle it in
hfi_venus layer to exit gracefully to venc layer.

Signed-off-by: Malathi Gottam <mgottam@codeaurora.org>
Tested-by: Alexandre Courbot <acourbot@chromium.org>
Acked-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: venus: add support for USERPTR to queue
Malathi Gottam [Tue, 9 Oct 2018 07:52:37 +0000 (03:52 -0400)]
media: venus: add support for USERPTR to queue

Add USERPTR to queue access methods by adding this
support to io_modes on both the planes.

Signed-off-by: Malathi Gottam <mgottam@codeaurora.org>
Tested-by: Alexandre Courbot <acourbot@chromium.org>
Acked-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: dvb-pll: don't re-validate tuner frequencies
Mauro Carvalho Chehab [Fri, 23 Nov 2018 17:10:57 +0000 (12:10 -0500)]
media: dvb-pll: don't re-validate tuner frequencies

The dvb_frontend core already checks for the frequencies. No
need for any additional check inside the driver.

It is part of the fixes for the following bug:
  https://bugzilla.opensuse.org/show_bug.cgi?id=1116374

Fixes: a3f90c75b833 ("media: dvb: convert tuner_info frequencies to Hz")
Reported-by: Stakanov Schufter <stakanov@eclipso.eu>
Reported-by: Takashi Iwai <tiwai@suse.de>
Cc: stable@vger.kernel.org # For 4.19
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: dvb-pll: fix tuner frequency ranges
Mauro Carvalho Chehab [Tue, 20 Nov 2018 10:19:36 +0000 (05:19 -0500)]
media: dvb-pll: fix tuner frequency ranges

Tuners should report frequencies in Hz. That works fine on most
drivers, but, in the case of dvb-pll, some settings are for
satellite tuners, while others are for terrestrial/cable ones.

The code was trying to solve it at probing time, but that doesn't
work, as, when _attach is called, the delivery system may be wrong.

Fix it by ensuring that all frequencies are in Hz at the per-tuner
max/min values.

While here, add a debug message, as this would help to debug any
issues there.

It partially fixes the following bug:
  https://bugzilla.opensuse.org/show_bug.cgi?id=1116374

Fixes: a3f90c75b833 ("media: dvb: convert tuner_info frequencies to Hz")
Reported-by: Stakanov Schufter <stakanov@eclipso.eu>
Reported-by: Takashi Iwai <tiwai@suse.de>
Cc: stable@vger.kernel.org # For 4.19
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: dvb_frontend: add debug message for frequency intervals
Mauro Carvalho Chehab [Tue, 20 Nov 2018 10:19:35 +0000 (05:19 -0500)]
media: dvb_frontend: add debug message for frequency intervals

As we did an internal change inside the subsystem to always
represent min/max frequencies in Hz, add a debug message, as this
would help to discover bugs on drivers, if any.

Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: dvb_frontend: don't print function names twice
Mauro Carvalho Chehab [Tue, 20 Nov 2018 10:19:34 +0000 (05:19 -0500)]
media: dvb_frontend: don't print function names twice

The dvb_frontend dprintk() macro already prints __func__. So,
we don't need to add it again at the printed message.

Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vb2: be sure to unlock mutex on errors
Mauro Carvalho Chehab [Fri, 23 Nov 2018 12:05:58 +0000 (07:05 -0500)]
media: vb2: be sure to unlock mutex on errors

As reported by smatch:
drivers/media/common/videobuf2/videobuf2-core.c: drivers/media/common/videobuf2/videobuf2-core.c:2159 vb2_mmap() warn: inconsistent returns 'mutex:&q->mmap_lock'.
  Locked on:   line 2148
  Unlocked on: line 2100
               line 2108
               line 2113
               line 2118
               line 2156
               line 2159

There is one error condition that doesn't unlock a mutex.

Fixes: cd26d1c4d1bc ("media: vb2: vb2_mmap: move lock up")
Reviewed-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: seco-cec: declare ops as static const
Mauro Carvalho Chehab [Fri, 23 Nov 2018 11:59:41 +0000 (06:59 -0500)]
media: seco-cec: declare ops as static const

As warned by smatch:

drivers/media/platform/seco-cec/seco-cec.c:338:21:  warning: symbol 'secocec_cec_adap_ops' was not declared. Should it be static?

This struct should be static. Also, it is const, so declare it
as such.

Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: seco-cec: fix Makefile
Hans Verkuil [Fri, 23 Nov 2018 11:53:58 +0000 (06:53 -0500)]
media: seco-cec: fix Makefile

Fix the incorrect obj-y.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: videobuf2-v4l2: drop WARN_ON in vb2_warn_zero_bytesused()
Hans Verkuil [Mon, 19 Nov 2018 15:33:44 +0000 (10:33 -0500)]
media: videobuf2-v4l2: drop WARN_ON in vb2_warn_zero_bytesused()

Userspace shouldn't set bytesused to 0 for output buffers.
vb2_warn_zero_bytesused() warns about this (only once!), but it also
calls WARN_ON(1), which is confusing since it is not immediately clear
that it warns about a 0 value for bytesused.

Just drop the WARN_ON as it serves no purpose.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Acked-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vb2: Allow reqbufs(0) with "in use" MMAP buffers
John Sheu [Thu, 15 Nov 2018 15:57:16 +0000 (10:57 -0500)]
media: vb2: Allow reqbufs(0) with "in use" MMAP buffers

Videobuf2 presently does not allow VIDIOC_REQBUFS to destroy outstanding
buffers if the queue is of type V4L2_MEMORY_MMAP, and if the buffers are
considered "in use".  This is different behavior than for other memory
types and prevents us from deallocating buffers in following two cases:

1) There are outstanding mmap()ed views on the buffer. However even if
   we put the buffer in reqbufs(0), there will be remaining references,
   due to vma .open/close() adjusting vb2 buffer refcount appropriately.
   This means that the buffer will be in fact freed only when the last
   mmap()ed view is unmapped.

2) Buffer has been exported as a DMABUF. Refcount of the vb2 buffer
   is managed properly by VB2 DMABUF ops, i.e. incremented on DMABUF
   get and decremented on DMABUF release. This means that the buffer
   will be alive until all importers release it.

Considering both cases above, there does not seem to be any need to
prevent reqbufs(0) operation, because buffer lifetime is already
properly managed by both mmap() and DMABUF code paths. Let's remove it
and allow userspace freeing the queue (and potentially allocating a new
one) even though old buffers might be still in processing.

To let userspace know that the kernel now supports orphaning buffers
that are still in use, add a new V4L2_BUF_CAP_SUPPORTS_ORPHANED_BUFS
to be set by reqbufs and create_bufs.

[p.zabel@pengutronix.de: added V4L2_BUF_CAP_SUPPORTS_ORPHANED_BUFS,
 updated documentation, and added back debug message]

Signed-off-by: John Sheu <sheu@chromium.org>
Reviewed-by: Pawel Osciak <posciak@chromium.org>
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
[hverkuil-cisco@xs4all.nl: added V4L2-BUF-CAP-SUPPORTS-ORPHANED-BUFS ref]
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: cedrus: Get rid of interrupt bottom-half
Ezequiel Garcia [Thu, 18 Oct 2018 18:02:24 +0000 (14:02 -0400)]
media: cedrus: Get rid of interrupt bottom-half

Now that the mem2mem framework guarantees that .device_run
won't be called from interrupt context, it is safe to call
v4l2_m2m_job_finish directly in the top-half.

So this means the bottom-half is no longer needed and we
can get rid of it.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Acked-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: v4l2-mem2mem: Avoid calling .device_run in v4l2_m2m_job_finish
Ezequiel Garcia [Thu, 18 Oct 2018 18:02:23 +0000 (14:02 -0400)]
media: v4l2-mem2mem: Avoid calling .device_run in v4l2_m2m_job_finish

v4l2_m2m_job_finish() is typically called when
DMA operations complete, in interrupt handlers or DMA
completion callbacks. Calling .device_run from v4l2_m2m_job_finish
creates a nasty re-entrancy path into the driver.

Moreover, some implementation of .device_run might need to sleep,
as is the case for drivers supporting the Request API,
where controls are applied via v4l2_ctrl_request_setup,
which takes the ctrl handler mutex.

This commit adds a deferred context that calls v4l2_m2m_try_run,
and gets scheduled by v4l2_m2m_job_finish().

Before this change, device_run would be called from these
paths:

vb2_m2m_request_queue, or
v4l2_m2m_streamon, or
v4l2_m2m_qbuf
  v4l2_m2m_try_schedule
    v4l2_m2m_try_run
      .device_run

v4l2_m2m_job_finish
  v4l2_m2m_try_run
    .device_run

After this change, the latter is now gone and instead:

v4l2_m2m_device_run_work
  v4l2_m2m_try_run
    .device_run

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-mem2mem: Simplify exiting the function in __v4l2_m2m_try_schedule
Sakari Ailus [Thu, 18 Oct 2018 18:02:22 +0000 (14:02 -0400)]
media: v4l2-mem2mem: Simplify exiting the function in __v4l2_m2m_try_schedule

The __v4l2_m2m_try_schedule function acquires and releases multiple
spinlocks. Simplify unlocking the job lock by adding labels to unlock
the lock and exit the function.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
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-ioctl.c: Simplify locking for m2m devices
Ezequiel Garcia [Thu, 18 Oct 2018 18:02:21 +0000 (14:02 -0400)]
media: v4l2-ioctl.c: Simplify locking for m2m devices

Now that the mutexes for output and capture vb2 queues match,
it is possible to refer to the context q_lock as the
m2m lock for a given m2m context.

Remove the output/capture lock selection.

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: mem2mem: Require capture and output mutexes to match
Ezequiel Garcia [Thu, 18 Oct 2018 18:02:20 +0000 (14:02 -0400)]
media: mem2mem: Require capture and output mutexes to match

Currently, all the mem2mem driver either use a single mutex
to lock the capture and output videobuf2 queues, or don't
set any mutex.

This means the mutexes match, and so the mem2mem framework
is able to set the m2m context lock.

Enforce this by making it mandatory for drivers to set
the same capture and output mutex, or not set any mutex at all.

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: dt-bindings: media: Document bindings for venus firmware device
Vikash Garodia [Wed, 17 Oct 2018 13:18:23 +0000 (09:18 -0400)]
media: dt-bindings: media: Document bindings for venus firmware device

Add devicetree binding documentation for firmware loader for video
hardware running on qualcomm chip.

Signed-off-by: Vikash Garodia <vgarodia@codeaurora.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Acked-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Reviewed-by: Alexandre Courbot <acourbot@chromium.org>
Tested-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: venus: firmware: add no TZ boot and shutdown routine
Vikash Garodia [Wed, 17 Oct 2018 13:18:22 +0000 (09:18 -0400)]
media: venus: firmware: add no TZ boot and shutdown routine

Video hardware is mainly comprised of vcodec subsystem and video
control subsystem. Video control has ARM9 which executes the video
firmware instructions whereas vcodec does the video frame processing.
This change adds support to load the video firmware and bring ARM9
out of reset for platforms which does not have trustzone.
An iommu domain is associated and managed with the firmware device.

Signed-off-by: Vikash Garodia <vgarodia@codeaurora.org>
Acked-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Reviewed-by: Alexandre Courbot <acourbot@chromium.org>
Tested-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: venus: firmware: register separate platform_device for firmware loader
Stanimir Varbanov [Wed, 17 Oct 2018 13:18:21 +0000 (09:18 -0400)]
media: venus: firmware: register separate platform_device for firmware loader

This registers a firmware platform_device and associate it with
video-firmware DT subnode. Then calls dma configure to initialize
dma and iommu.

Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Reviewed-by: Alexandre Courbot <acourbot@chromium.org>
Tested-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: venus: firmware: move load firmware in a separate function
Vikash Garodia [Wed, 17 Oct 2018 13:18:20 +0000 (09:18 -0400)]
media: venus: firmware: move load firmware in a separate function

Separate firmware loading part into a new function.

Signed-off-by: Vikash Garodia <vgarodia@codeaurora.org>
Acked-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Reviewed-by: Alexandre Courbot <acourbot@chromium.org>
Tested-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: venus: firmware: add routine to reset ARM9
Vikash Garodia [Wed, 17 Oct 2018 13:18:19 +0000 (09:18 -0400)]
media: venus: firmware: add routine to reset ARM9

Add routine to reset the ARM9 and brings it out of reset. Also
abstract the Venus CPU state handling with a new function. This
is in preparation to add PIL functionality in venus driver.

Signed-off-by: Vikash Garodia <vgarodia@codeaurora.org>
Acked-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Reviewed-by: Alexandre Courbot <acourbot@chromium.org>
Tested-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: vivid: free bitmap_cap when updating std/timings/etc.
Hans Verkuil [Fri, 9 Nov 2018 13:37:44 +0000 (08:37 -0500)]
media: vivid: free bitmap_cap when updating std/timings/etc.

When vivid_update_format_cap() is called it should free any overlay
bitmap since the compose size will change.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reported-by: syzbot+0cc8e3cc63ca373722c6@syzkaller.appspotmail.com
Cc: <stable@vger.kernel.org> # for v3.18 and up
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: v4l2-tpg: array index could become negative
Hans Verkuil [Thu, 8 Nov 2018 16:12:47 +0000 (11:12 -0500)]
media: v4l2-tpg: array index could become negative

text[s] is a signed char, so using that as index into the font8x16 array
can result in negative indices. Cast it to u8 to be safe.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reported-by: syzbot+ccf0a61ed12f2a7313ee@syzkaller.appspotmail.com
Cc: <stable@vger.kernel.org> # for v4.7 and up
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: MAINTAINERS fixups
Hans Verkuil [Thu, 8 Nov 2018 13:23:58 +0000 (08:23 -0500)]
media: MAINTAINERS fixups

Update file paths in MAINTAINERS.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reported-by: Joe Perches <joe@perches.com>
Reviewed-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vb2: check memory model for VIDIOC_CREATE_BUFS
Hans Verkuil [Thu, 8 Nov 2018 12:23:37 +0000 (07:23 -0500)]
media: vb2: check memory model for VIDIOC_CREATE_BUFS

vb2_core_create_bufs did not check if the memory model for newly added
buffers is the same as for already existing buffers. It should return an
error if they aren't the same.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reported-by: syzbot+e1fb118a2ebb88031d21@syzkaller.appspotmail.com
Cc: <stable@vger.kernel.org> # for v4.16 and up
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: adv*/tc358743/ths8200: fill in min width/height/pixelclock
Hans Verkuil [Thu, 8 Nov 2018 09:51:51 +0000 (04:51 -0500)]
media: adv*/tc358743/ths8200: fill in min width/height/pixelclock

The v4l2_dv_timings_cap struct is used to do sanity checks when setting and
enumerating DV timings, ensuring that only valid timings as per the HW
capabilities are allowed.

However, many drivers just filled in 0 for the minimum width, height or
pixelclock frequency. This can cause timings with e.g. 0 as width and height
to be accepted, which will in turn lead to a potential division by zero.

Fill in proper values are minimum boundaries. 640x350 was chosen since it is
the smallest resolution in v4l2-dv-timings.h. Same for 13 MHz as the lowest
pixelclock frequency (it's slightly below the minimum of 13.5 MHz in the
v4l2-dv-timings.h header).

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vim2m: use cancel_delayed_work_sync instead of flush_schedule_work
Hans Verkuil [Wed, 7 Nov 2018 14:04:54 +0000 (09:04 -0500)]
media: vim2m: use cancel_delayed_work_sync instead of flush_schedule_work

The use of flush_schedule_work() made no sense and caused a syzkaller error.
Replace with the correct cancel_delayed_work_sync().

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reported-by: syzbot+69780d144754b8071f4b@syzkaller.appspotmail.com
Cc: <stable@vger.kernel.org> # for v4.20 and up
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vivid: fill in media_device bus_info
Hans Verkuil [Tue, 6 Nov 2018 14:06:41 +0000 (09:06 -0500)]
media: vivid: fill in media_device bus_info

If you create multiple vivid instances, each with their own media
device, then there was no way to tell them apart.

Fill in the bus_info so each instance has a unique bus_info string.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: rockchip/rga: constify video_device structure
Julia Lawall [Fri, 2 Nov 2018 14:48:15 +0000 (10:48 -0400)]
media: rockchip/rga: constify video_device structure

The video_device structure is only copied into another structure, so
it can be const.

Done with the help of Coccinelle.

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vivid: set min width/height to a value > 0
Hans Verkuil [Mon, 29 Oct 2018 17:32:38 +0000 (13:32 -0400)]
media: vivid: set min width/height to a value > 0

The capture DV timings capabilities allowed for a minimum width and
height of 0. So passing a timings struct with 0 values is allowed
and will later cause a division by zero.

Ensure that the width and height must be >= 16 to avoid this.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reported-by: syzbot+57c3d83d71187054d56f@syzkaller.appspotmail.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vivid: fix error handling of kthread_run
Hans Verkuil [Mon, 29 Oct 2018 10:15:31 +0000 (06:15 -0400)]
media: vivid: fix error handling of kthread_run

kthread_run returns an error pointer, but elsewhere in the code
dev->kthread_vid_cap/out is checked against NULL.

If kthread_run returns an error, then set the pointer to NULL.

I chose this method over changing all kthread_vid_cap/out tests
elsewhere since this is more robust.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reported-by: syzbot+53d5b2df0d9744411e2e@syzkaller.appspotmail.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vimc: constify structures stored in fields of v4l2_subdev_ops structure
Julia Lawall [Sat, 27 Oct 2018 12:49:05 +0000 (08:49 -0400)]
media: vimc: constify structures stored in fields of v4l2_subdev_ops structure

The fields of a v4l2_subdev_ops structure are all const, so the
structures that are stored there and are not used elsewhere can be
const as well.

Done with the help of Coccinelle.

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Acked-by: Helen Koike <helen.koike@collabora.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: rockchip/rga: constify v4l2_m2m_ops structure
Julia Lawall [Sat, 27 Oct 2018 12:30:32 +0000 (08:30 -0400)]
media: rockchip/rga: constify v4l2_m2m_ops structure

The v4l2_m2m_ops structure can be const as it is only passed to
v4l2_m2m_init whose parameter is const.

Done with the help of Coccinelle.

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vicodec: constify v4l2_ctrl_ops structure
Julia Lawall [Sat, 27 Oct 2018 12:16:39 +0000 (08:16 -0400)]
media: vicodec: constify v4l2_ctrl_ops structure

The v4l2_ctrl_ops structure is only stored in the ops field of a
v4l2_ctrl_config structure, and this field is const, or passed as the
second argument of v4l2_ctrl_new_std, and the corresponding parameter
is declared as const.  Accordingly, the structure can also be const.

Done with the help of Coccinelle.

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: v4l uapi docs: few minor corrections and typos
Sean Young [Fri, 26 Oct 2018 12:18:33 +0000 (08:18 -0400)]
media: v4l uapi docs: few minor corrections and typos

Do a series of minor corrections at the V4L2 uAPI documentation.

Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: cec: keep track of outstanding transmits
Hans Verkuil [Fri, 19 Oct 2018 07:55:34 +0000 (03:55 -0400)]
media: cec: keep track of outstanding transmits

I noticed that repeatedly running 'cec-ctl --playback' would occasionally
select 'Playback Device 2' instead of 'Playback Device 1', even though there
were no other Playback devices in the HDMI topology. This happened both with
'real' hardware and with the vivid CEC emulation, suggesting that this was an
issue in the core code that claims a logical address.

What 'cec-ctl --playback' does is to first clear all existing logical addresses,
and immediately after that configure the new desired device type.

The core code will poll the logical addresses trying to find a free address.
When found it will issue a few standard messages as per the CEC spec and return.
Those messages are queued up and will be transmitted asynchronously.

What happens is that if you run two 'cec-ctl --playback' commands in quick
succession, there is still a message of the first cec-ctl command being transmitted
when you reconfigure the adapter again in the second cec-ctl command.

When the logical addresses are cleared, then all information about outstanding
transmits inside the CEC core is also cleared, and the core is no longer aware
that there is still a transmit in flight.

When the hardware finishes the transmit it calls transmit_done and the CEC core
thinks it is actually in response of a POLL messages that is trying to find a
free logical address. The result of all this is that the core thinks that the
logical address for Playback Device 1 is in use, when it is really an earlier
transmit that ended.

The main transmit thread looks at adap->transmitting to check if a transmit
is in progress, but that is set to NULL when the adapter is unconfigured.
adap->transmitting represents the view of userspace, not that of the hardware.
So when unconfiguring the adapter the message is marked aborted from the point
of view of userspace, but seen from the PoV of the hardware it is still ongoing.

So introduce a new bool transmit_in_progress that represents the hardware state
and use that instead of adap->transmitting. Now the CEC core waits until the
hardware finishes the transmit before starting a new transmit.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Cc: <stable@vger.kernel.org> # for v4.18 and up
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: cec: add debug_phys_addr module option
Hans Verkuil [Wed, 17 Oct 2018 11:05:41 +0000 (07:05 -0400)]
media: cec: add debug_phys_addr module option

If debug_phys_addr is set, then CEC_CAP_PHYS_ADDR is added to the CEC
adapter capabilities.

This allows for testing CEC even if the physical address isn't set. This
makes it possible to connect two HDMI outputs together and still use CEC.
Very useful for testing CEC if you don't have access to an HDMI receiver
under linux.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: cec: report Vendor ID after initialization
Hans Verkuil [Tue, 16 Oct 2018 07:44:20 +0000 (03:44 -0400)]
media: cec: report Vendor ID after initialization

The CEC specification requires that the Vendor ID (if any) is reported
after a logical address was claimed.

This was never done, so add support for this.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: adv7604: add CEC support for adv7611/adv7612
Hans Verkuil [Fri, 12 Oct 2018 11:30:02 +0000 (07:30 -0400)]
media: adv7604: add CEC support for adv7611/adv7612

The CEC IP is very similar between the three HDMI receivers, but
not identical. Add support for all three variants.

Tested with an adv7604 and an adv7612.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reviewed-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: vb2: vb2_mmap: move lock up
Hans Verkuil [Tue, 13 Nov 2018 14:06:46 +0000 (09:06 -0500)]
media: vb2: vb2_mmap: move lock up

If a filehandle is dup()ped, then it is possible to close it from one fd
and call mmap from the other. This creates a race condition in vb2_mmap
where it is using queue data that __vb2_queue_free (called from close())
is in the process of releasing.

By moving up the mutex_lock(mmap_lock) in vb2_mmap this race is avoided
since __vb2_queue_free is called with the same mutex locked. So vb2_mmap
now reads consistent buffer data.

Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Reported-by: syzbot+be93025dd45dccd8923c@syzkaller.appspotmail.com
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: pulse8-cec: return 0 when invalidating the logical address
Hans Verkuil [Wed, 14 Nov 2018 13:25:53 +0000 (08:25 -0500)]
media: pulse8-cec: return 0 when invalidating the logical address

Return 0 when invalidating the logical address. The cec core produces
a warning for drivers that do this.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reported-by: Torbjorn Jansson <torbjorn.jansson@mbox200.swipnet.se>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: cec-pin: fix broken tx_ignore_nack_until_eom error injection
Hans Verkuil [Wed, 14 Nov 2018 08:37:53 +0000 (03:37 -0500)]
media: cec-pin: fix broken tx_ignore_nack_until_eom error injection

If the tx_ignore_nack_until_eom error injection was activated,
then tx_nacked was never set instead of setting it when the last
byte of the message was transmitted.

As a result the transmit was marked as OK, when it should have
been NACKed.

Modify the condition so that it always sets tx_nacked when the
last byte of the message was transmitted.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Cc: <stable@vger.kernel.org> # for v4.17 and up
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: doc-rst: Fix broken references
Ricardo Ribalda Delgado [Fri, 9 Nov 2018 15:25:41 +0000 (10:25 -0500)]
media: doc-rst: Fix broken references

Documentation and code was linking the old documentation at:
http://v4l2spec.bytesex.org/spec/x1904.htm

Signed-off-by: Ricardo Ribalda Delgado <ricardo@ribalda.com>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: platform: fix platform_no_drv_owner.cocci warnings
kbuild test robot [Tue, 6 Nov 2018 11:33:19 +0000 (06:33 -0500)]
media: platform: fix platform_no_drv_owner.cocci warnings

drivers/staging/media/sunxi/cedrus/cedrus.c:421:3-8: No need to set .owner here. The core will do it.

 Remove .owner field if calls are used which set it automatically

Generated by: scripts/coccinelle/api/platform_no_drv_owner.cocci

Fixes: 50e761516f2b ("media: platform: Add Cedrus VPU decoder driver")

CC: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Signed-off-by: kbuild test robot <fengguang.wu@intel.com>
Acked-by: Maxime Ripard <maxime.ripard@bootlin.com>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: rcar-csi2: Handle per-SoC number of channels
Jacopo Mondi [Tue, 6 Nov 2018 10:54:27 +0000 (05:54 -0500)]
media: rcar-csi2: Handle per-SoC number of channels

The R-Car CSI-2 interface has a number of selectable 'channels' that
provides pixel data to the VINs during image acquisition.

Each channel can be used to match a CSI-2 data type and a CSI-2 virtual
channel to be routed to output path.

Different SoCs have different number of channels, with R-Car E3 being the
notable exception supporting only 2 of them.

Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: rcar: rcar-csi2: Update V3M/E3 PHTW tables
Jacopo Mondi [Tue, 6 Nov 2018 10:54:26 +0000 (05:54 -0500)]
media: rcar: rcar-csi2: Update V3M/E3 PHTW tables

Update PHTW tables for V3M and E3 SoCs to the latest datasheet release
(R-Car Series, 3rd Generation manual rev1.00 20181017).

Signed-off-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: rcar-csi2: Add R8A77990 support
Jacopo Mondi [Tue, 6 Nov 2018 10:54:25 +0000 (05:54 -0500)]
media: rcar-csi2: Add R8A77990 support

Add support for R-Car E3 R8A77965 to R-Car CSI-2 driver.
Based on the experimental patch from Magnus Damm.

Signed-off-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: dt-bindings: rcar-csi2: Add R8A77990
Jacopo Mondi [Tue, 6 Nov 2018 10:54:24 +0000 (05:54 -0500)]
media: dt-bindings: rcar-csi2: Add R8A77990

Add compatible string for R-Car E3 R8A77990 to the list of supported SoCs.

Signed-off-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: rcar-vin: Add support for R-Car R8A77990
Jacopo Mondi [Tue, 6 Nov 2018 10:54:23 +0000 (05:54 -0500)]
media: rcar-vin: Add support for R-Car R8A77990

Add R-Car E3 R8A77990 SoC to the rcar-vin supported ones.
Based on the experimental patch from Magnus Damm.

Signed-off-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: dt-bindings: rcar-vin: Add R8A77990 support
Jacopo Mondi [Tue, 6 Nov 2018 10:54:22 +0000 (05:54 -0500)]
media: dt-bindings: rcar-vin: Add R8A77990 support

Add compatible string for R-Car E3 R8A77990 to the list of SoCs supported by
rcar-vin driver.

Signed-off-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: imx-pxp: Improve pxp_soft_reset() error message
Fabio Estevam [Tue, 6 Nov 2018 10:16:05 +0000 (05:16 -0500)]
media: imx-pxp: Improve pxp_soft_reset() error message

Improve the pxp_soft_reset() error message by moving it to the
caller function, associating it with a proper device and also
by displaying the error code.

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: imx-pxp: Check for pxp_soft_reset() error
Fabio Estevam [Tue, 6 Nov 2018 10:16:04 +0000 (05:16 -0500)]
media: imx-pxp: Check for pxp_soft_reset() error

pxp_soft_reset() may fail with a timeout, so it is better to propagate
the error in this case.

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: imx-pxp: Check the return value from clk_prepare_enable()
Fabio Estevam [Tue, 6 Nov 2018 10:16:03 +0000 (05:16 -0500)]
media: imx-pxp: Check the return value from clk_prepare_enable()

clk_prepare_enable() may fail, so we should better check its return value
and propagate it in the case of error.

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: coda: debug output when setting visible size via crop selection
Philipp Zabel [Mon, 5 Nov 2018 15:25:13 +0000 (10:25 -0500)]
media: coda: debug output when setting visible size via crop selection

In addition to the S_FMT debug output, S_SELECTION (SEL_TGT_CROP) is
relevant to determine encoded size. Add debug output for it.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: coda: normalise debug output
Philipp Zabel [Mon, 5 Nov 2018 15:25:12 +0000 (10:25 -0500)]
media: coda: normalise debug output

Consistently add the context index to debug output, which otherwise is
impossible to make sense of when two contexts are running concurrently.
For this purpose, add a convenience macro coda_dbg(). Use the function
name with the coda_ prefix stripped as keyword where applicable, and
consistently use vid-out and vid-cap names for the queues. Add sequence
counters to the decoder job finished message and correctly indicate B
frames. Add a start streaming message to complement the stop streaming
message and a start encoding message to complement the existing start
decoding message.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: coda: improve queue busy error message
Philipp Zabel [Mon, 5 Nov 2018 15:25:11 +0000 (10:25 -0500)]
media: coda: improve queue busy error message

Use v4l2_type_names to indicate which of the two queues is busy.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: coda: print SEQ_INIT error code as hex value
Michael Tretter [Mon, 5 Nov 2018 15:25:10 +0000 (10:25 -0500)]
media: coda: print SEQ_INIT error code as hex value

The error code looks much more like a bit field than an error value.
Print it as hex rather than decimal.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: coda: fail S_SELECTION for read-only targets
Philipp Zabel [Mon, 5 Nov 2018 15:25:09 +0000 (10:25 -0500)]
media: coda: fail S_SELECTION for read-only targets

v4l2-compliance complains if S_SELECTION returns 0 for read-only targets.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: coda: never set infinite timeperframe
Philipp Zabel [Mon, 5 Nov 2018 15:25:08 +0000 (10:25 -0500)]
media: coda: never set infinite timeperframe

v4l2-compliance complains if G_PARM returns 0 in the denominator.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: coda: implement ENUM_FRAMEINTERVALS
Philipp Zabel [Mon, 5 Nov 2018 15:25:07 +0000 (10:25 -0500)]
media: coda: implement ENUM_FRAMEINTERVALS

v4l2-compliance complains about S_PARM being supported, but not
ENUM_FRAMEINTERVALS.
Report a continuous frame interval even though the hardware only
supports 16-bit numerator and denominator, with min/max values
that can be programmed into the mailbox registers.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: coda: set V4L2_CAP_TIMEPERFRAME flag in coda_s_parm
Philipp Zabel [Mon, 5 Nov 2018 15:25:06 +0000 (10:25 -0500)]
media: coda: set V4L2_CAP_TIMEPERFRAME flag in coda_s_parm

The flag is already set in coda_g_parm, but v4l2-compliance complains
about it not being set during S_PARM.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: coda: don't disable IRQs across buffer meta handling
Lucas Stach [Mon, 5 Nov 2018 15:25:05 +0000 (10:25 -0500)]
media: coda: don't disable IRQs across buffer meta handling

The CODA driver uses threaded IRQs only, so there is nothing happening
in hardirq context that could interfere with the buffer meta handling.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: coda: remove unused instances list
Philipp Zabel [Mon, 5 Nov 2018 15:25:04 +0000 (10:25 -0500)]
media: coda: remove unused instances list

The per-device instance list is unused, remove it.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: coda: reduce minimum frame size to 48x16 pixels.
Philipp Zabel [Mon, 5 Nov 2018 15:25:03 +0000 (10:25 -0500)]
media: coda: reduce minimum frame size to 48x16 pixels.

Three macroblocks seem to be the minimum resolution that can be encoded
and decoded by the CODA960 h.264 codec. Picture run commands fail for
smaller resolutions.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: coda: limit queueing into internal bitstream buffer
Lucas Stach [Mon, 5 Nov 2018 15:25:02 +0000 (10:25 -0500)]
media: coda: limit queueing into internal bitstream buffer

The ringbuffer used to hold the bitstream is very conservatively sized,
as keyframes can get very large and still need to fit into this buffer.
This means that the buffer is way oversized for the average stream to
the extend that it will hold a few hundred frames when the video data
is compressing well.

The current strategy of queueing as much bitstream data as possible
leads to large delays when draining the decoder. In order to keep the
drain latency to a reasonable bound, try to only queue a full reorder
window of buffers. We can't always hit this low target for very well
compressible video data, as we might end up with less than the minimum
amount of data that needs to be available to the bitstream prefetcher,
so we must take this into account and allow more buffers to be queued
in this case.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
5 years agomedia: coda: always hold back decoder jobs until we have enough bitstream payload
Philipp Zabel [Mon, 5 Nov 2018 15:25:01 +0000 (10:25 -0500)]
media: coda: always hold back decoder jobs until we have enough bitstream payload

The bitstream prefetch unit reads data in 256 byte blocks with some kind
of queueing. For the decoder to see data up to a desired position in the
next run, the bitstream has to be filled for 2 256 byte blocks past that
position aligned up to the next 256 byte boundary.
This should make sure we never run into a buffer underrun condition if
userspace does not supply new input buffers fast enough.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>