[media] cx231xx-cards: unregister IR earlier

Without this patch, a bug is issued when the module is removed:

[ 1417.425863] cx231xx 1-3.1.4:1.1: Cx231xx dvb Extension removed
[ 1417.571923] BUG: unable to handle kernel paging request at ffffffffc081a024
[ 1417.571962] IP: [<ffffffff813da854>] string+0x24/0x80
[ 1417.571987] PGD 1c09067 PUD 1c0b067 PMD 88e653067 PTE 0
[ 1417.572013] Oops: 0000 [#1] SMP
[ 1417.572026] Modules linked in: mb86a20s dvb_core cx231xx_alsa ir_kbd_i2c(-) tda18271 tea5767 tuner cx25840 cx231xx i2c_mux videobuf_vmalloc tveeprom cx2341x videobuf_core rc_core v4l2_common videodev media bnep usblp fuse xt_CHECKSUM iptable_mangle tun ebtable_filter ebtables ip6table_filter ip6_tables xt_physdev br_netfilter bridge nf_log_ipv4 nf_log_common xt_LOG xt_conntrack ipt_MASQUERADE nf_nat_masquerade_ipv4 iptable_nat nf_conntrack_ipv4 nf_defrag_ipv4 nf_nat_ipv4 nf_nat nf_conntrack cpufreq_stats vfat fat snd_hda_codec_hdmi snd_hda_codec_realtek snd_hda_codec_generic intel_rapl x86_pkg_temp_thermal intel_powerclamp coretemp kvm_intel snd_hda_intel snd_hda_codec kvm snd_hda_core snd_hwdep snd_seq snd_seq_device snd_pcm irqbypass crct10dif_pclmul iTCO_wdt crc32_pclmul nfsd hci_uart iTCO_vendor_support
[ 1417.572317]  snd_timer ghash_clmulni_intel btbcm intel_cstate btqca snd intel_uncore btintel intel_rapl_perf mei_me bluetooth mei shpchp soundcore pcspkr i2c_i801 auth_rpcgss wmi acpi_als kfifo_buf nfs_acl industrialio rfkill lockd pinctrl_sunrisepoint pinctrl_intel tpm_tis tpm intel_lpss_acpi intel_lpss acpi_pad grace sunrpc binfmt_misc hid_logitech_hidpp hid_logitech_dj 8021q garp stp llc mrp i915 i2c_algo_bit drm_kms_helper drm e1000e sdhci_pci sdhci mmc_core crc32c_intel ptp pps_core video i2c_hid fjes analog gameport joydev [last unloaded: rc_pixelview_002t]
[ 1417.572487] CPU: 4 PID: 24493 Comm: rmmod Tainted: G        W       4.7.0+ #2
[ 1417.572504] Hardware name:                  /NUC6i7KYB, BIOS KYSKLi70.86A.0041.2016.0817.1130 08/17/2016
[ 1417.572526] task: ffff880894b81e80 ti: ffff880896bdc000 task.ti: ffff880896bdc000
[ 1417.572544] RIP: 0010:[<ffffffff813da854>]  [<ffffffff813da854>] string+0x24/0x80
[ 1417.572564] RSP: 0018:ffff880896bdfbe8  EFLAGS: 00010286
[ 1417.572577] RAX: ffffffffc081a025 RBX: ffff8808935aa15c RCX: ffff0a00ffffff04
[ 1417.572594] RDX: ffffffffc081a024 RSI: ffffffffffffffff RDI: ffff8808935aa15c
[ 1417.572610] RBP: ffff880896bdfbe8 R08: fffffffffffffffe R09: ffff8808935aa91c
[ 1417.572628] R10: ffffffffc07b85d6 R11: 0000000000000000 R12: ffff8808935aa91c
[ 1417.572644] R13: 00000000000007c5 R14: ffffffffc07b85dd R15: ffffffffc07b85dd
[ 1417.572662] FS:  00007f5a5392d700(0000) GS:ffff8808bed00000(0000) knlGS:0000000000000000
[ 1417.572681] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 1417.572705] CR2: ffffffffc081a024 CR3: 0000000897188000 CR4: 00000000003406e0
[ 1417.572735] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 1417.572761] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[ 1417.572778] Stack:
[ 1417.572785]  ffff880896bdfc48 ffffffff813dcf77 0000000000000005 ffff8808935aa157
[ 1417.572806]  ffff880896bdfc58 ffff0a00ffffff04 000000009d27375e ffff8808935aa000
[ 1417.572829]  0000000000000800 ffff880896182800 0000000000000000 ffff88089e898ae0
[ 1417.572850] Call Trace:
[ 1417.572860]  [<ffffffff813dcf77>] vsnprintf+0x2d7/0x500
[ 1417.572873]  [<ffffffff813d3e12>] add_uevent_var+0x82/0x120
[ 1417.572890]  [<ffffffffc07b534d>] rc_dev_uevent+0x2d/0x60 [rc_core]
[ 1417.572907]  [<ffffffff81515969>] dev_uevent+0xd9/0x2d0
[ 1417.572921]  [<ffffffff813d4309>] kobject_uevent_env+0x2d9/0x4f0
[ 1417.572938]  [<ffffffff813d452b>] kobject_uevent+0xb/0x10
[ 1417.572954]  [<ffffffff81513a3f>] device_del+0x18f/0x260
[ 1417.572974]  [<ffffffff813d2db7>] ? kobject_put+0x27/0x50
[ 1417.572998]  [<ffffffffc07b5e25>] rc_unregister_device+0x75/0xb0 [rc_core]
[ 1417.573028]  [<ffffffffc07e6023>] ir_remove+0x23/0x30 [ir_kbd_i2c]
[ 1417.573055]  [<ffffffff8162bf88>] i2c_device_remove+0x58/0xb0
[ 1417.573078]  [<ffffffff81518191>] __device_release_driver+0xa1/0x160
[ 1417.573102]  [<ffffffff81518de6>] driver_detach+0xa6/0xb0
[ 1417.573122]  [<ffffffff81517b25>] bus_remove_driver+0x55/0xd0
[ 1417.573146]  [<ffffffff815195bc>] driver_unregister+0x2c/0x50
[ 1417.573168]  [<ffffffff8162cf62>] i2c_del_driver+0x22/0x50
[ 1417.573194]  [<ffffffffc07e6ba4>] ir_kbd_driver_exit+0x10/0x46c [ir_kbd_i2c]
[ 1417.573227]  [<ffffffff81126348>] SyS_delete_module+0x1b8/0x220
[ 1417.573254]  [<ffffffff817debf2>] entry_SYSCALL_64_fastpath+0x1a/0xa4
[ 1417.573279] Code: eb e9 76 ff ff ff 90 55 49 89 f1 48 89 ce 48 c1 fe 30 48 81 fa ff 0f 00 00 48 89 e5 4c 8d 46 ff 76 40 48 85 f6 74 4e 48 8d 42 01 <0f> b6 12 84 d2 74 43 49 01 c0 31 f6 eb 0c 48 83 c0 01 0f b6 50
[ 1417.573437] RIP  [<ffffffff813da854>] string+0x24/0x80
[ 1417.573455]  RSP <ffff880896bdfbe8>
[ 1417.573465] CR2: ffffffffc081a024
[ 1417.580053] ---[ end trace 4ca9e2eced326a62 ]---

Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] cx231xx: can't proceed if I2C bus register fails

The driver should not ignore errors while registering the I2C
bus, as this device can't even minimally work without the buses,
as it uses those buses internally to talk with the several IP
blocks inside the chip.

Cc: stable@vger.kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] cx231xx-i2c: handle errors with cx231xx_get_i2c_adap()

The cx231xx_get_i2c_adap() function should return the I2C
adapter that will be used to talk with a device. It should never
be NULL, as otherwise the driver will try to dereference a
null pointer.

We might instead fix the callers, but if this condition
ever happens, it is really a driver bug, because i2c_port
should always be a value from enum CX231XX_I2C_MASTER_PORT.

Found when checking the code due to this bug:

[   39.769021] BUG: unable to handle kernel NULL pointer dereference at 0000000000000002
[   39.769105] IP: [<ffffffff81638393>] i2c_master_send+0x13/0x70

Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] cx231xx-core: fix GPIO comments

The number of the cx231xx REQ for GPIO register set/get are wrong.
They should follow what's there at cx231xx-pcb-cfg.h.

Noticed while checking the cx231xx parser at the v4l-utils.

Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] cx231xx: prints error code if can't switch TV mode

If something bad happens when switching between digital
and analog mode, prints an error and outputs the returned code.

Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] cx231xx: fix GPIOs for Pixelview SBTVD hybrid

This device uses GPIOs: 28 to switch between analog and
digital modes: on digital mode, it should be set to 1.

The code that sets it on analog mode is OK, but it misses
the logic that sets it on digital mode.

Cc: stable@vger.kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] cx231xx: don't return error on success

The cx231xx_set_agc_analog_digital_mux_select() callers
expect it to return 0 or an error. Returning a positive value
makes the first attempt to switch between analog/digital to fail.

Cc: stable@vger.kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] mb86a20s: fix demod settings

With the current settings, only one channel locks properly.
That's likely because, when this driver was written, Brazil
were still using experimental transmissions.

Change it to reproduce the settings used by the newer drivers.
That makes it lock on other channels.

Tested with both PixelView SBTVD Hybrid (cx231xx-based) and
C3Tech Digital Duo HDTV/SDTV (em28xx-based) devices.

Cc: stable@vger.kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] mb86a20s: fix the locking logic

On this frontend, it takes a while to start output normal
TS data. That only happens on state S9. On S8, the TS output
is enabled, but it is not reliable enough.

However, the zigzag loop is too fast to let it sync.

As, on practical tests, the zigzag software loop doesn't
seem to be helping, but just slowing down the tuning, let's
switch to hardware algorithm, as the tuners used on such
devices are capable of work with frequency drifts without
any help from software.

Cc: stable@vger.kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] tea5767: use module prefix on printed messages

use pr_fmt() & friends for error messages to output like:

[    9.651721] tea5767: Chip ID is not zero. It is not a TEA5767

Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] tda18271: use prefix on all printk messages

Some messages have a hardcoded prefix; others not. Use the
pr_fmt() to ensure that all messages will use the same prefix.

Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

Merge branch 'docs-next' of git://git.lwn.net/linux into patchwork

* 'docs-next' of git://git.lwn.net/linux:
  doc-rst: define PDF's of the media folder
  doc-rst: generic way to build PDF of sub-folders
  docs: sphinx-extensions: add metadata parallel-safe
  docs-rst: kernel-doc: fix typedef output in RST format
  docs-rst: improve typedef parser
  docs: kernel-parameter: Improve the description of nr_cpus and maxcpus
  docs-rst: kernel-doc: better output struct members

Merge branch 'doc/4.9' into docs-next

doc-rst: define PDF's of the media folder

To build only the PDF of the media folder run::

  make SPHINXDIRS=media pdfdocs

Signed-off-by: Markus Heiser <markus.heiser@darmarIT.de>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>

doc-rst: generic way to build PDF of sub-folders

This extends the method to build only sub-folders to the targets
"latexdocs" and "pdfdocs". To do so, a conf.py in the sub-folder is
required, where the latex_documents of the sub-folder are
defined. E.g. to build only gpu's PDF add the following to the
Documentation/gpu/conf.py::

  +latex_documents = [
  +    ("index", "gpu.tex", "Linux GPU Driver Developer's Guide",
  +     "The kernel development community", "manual"),
  +]

and run:

  make SPHINXDIRS=gpu pdfdocs

Signed-off-by: Markus Heiser <markus.heiser@darmarIT.de>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>

docs: sphinx-extensions: add metadata parallel-safe

The setup() function of a Sphinx-extension can return a dictionary. This
is treated by Sphinx as metadata of the extension [1].

With metadata "parallel_read_safe = True" a extension is marked as
save for "parallel reading of source". This is needed if you want
build in parallel with N processes. E.g.:

  make SPHINXOPTS=-j4 htmldocs

will no longer log warnings like:

  WARNING: the foobar extension does not declare if it is safe for
  parallel reading, assuming it isn't - please ask the extension author
  to check and make it explicit.

Add metadata to extensions:

* kernel-doc
* flat-table
* kernel-include

[1] http://www.sphinx-doc.org/en/stable/extdev/#extension-metadata

Signed-off-by: Markus Heiser <markus.heiser@darmarIT.de>
Tested-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>

docs-rst: kernel-doc: fix typedef output in RST format

When using a typedef function like this one:
typedef bool v4l2_check_dv_timings_fnc (const struct v4l2_dv_timings * t, void * handle);

The Sphinx C domain expects it to create a c:type: reference,
as that's the way it creates the type references when parsing
a c:function:: declaration.

So, a declaration like:

.. c:function:: bool v4l2_valid_dv_timings (const struct v4l2_dv_timings * t, const struct v4l2_dv_timings_cap * cap, v4l2_check_dv_timings_fnc fnc, void * fnc_handle)

Will create a cross reference for :c:type:`v4l2_check_dv_timings_fnc`.

So, when outputting such typedefs in RST format, we need to handle
this special case, as otherwise it will produce those warnings:

./include/media/v4l2-dv-timings.h:43: WARNING: c:type reference target not found: v4l2_check_dv_timings_fnc
./include/media/v4l2-dv-timings.h:60: WARNING: c:type reference target not found: v4l2_check_dv_timings_fnc
./include/media/v4l2-dv-timings.h:81: WARNING: c:type reference target not found: v4l2_check_dv_timings_fnc

So, change the kernel-doc script to produce a RST output for the
above typedef as:
.. c:type:: v4l2_check_dv_timings_fnc

   **Typedef**: timings check callback

**Syntax**

  ``bool v4l2_check_dv_timings_fnc (const struct v4l2_dv_timings * t, void * handle);``

Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>

docs-rst: improve typedef parser

Improve the parser to handle typedefs like:

typedef bool v4l2_check_dv_timings_fnc(const struct v4l2_dv_timings *t, void *handle);

Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>

docs: kernel-parameter: Improve the description of nr_cpus and maxcpus

From the old description people still can't get what's the exact
difference between nr_cpus and maxcpus. Especially in kdump kernel
nr_cpus is always suggested if it's implemented in the ARCH. The
reason is nr_cpus is used to limit the max number of possible cpu
in system, the sum of already plugged cpus and hot plug cpus can't
exceed its value. However maxcpus is used to limit how many cpus
are allowed to be brought up during bootup.

Signed-off-by: Baoquan He <bhe@redhat.com>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>

[media] m2m-deinterlace: Fix error print during probe

v4l2_err() can not be used for printing error for missing interleaved
support in DMA as this point the pcdev->v4l2_dev is not valid.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] redrat3: fix sparse warning

Fix this sparse warning:

drivers/media/rc/redrat3.c:490:18: warning: incorrect type in assignment (different base types)
drivers/media/rc/redrat3.c:495:9: warning: cast to restricted __be32

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] drivers: hackrf: fixed a coding style issue

changed 'unsigned' to 'unsigned int' to obtain the coding style.

Signed-off-by: Jannik Becher <Becher.Jannik@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

docs-rst: kernel-doc: better output struct members

Right now, for a struct, kernel-doc produces the following output:

.. c:type:: struct v4l2_prio_state

   stores the priority states

**Definition**

::

  struct v4l2_prio_state {
    atomic_t prios[4];
  };

**Members**

``atomic_t prios[4]``
  array with elements to store the array priorities

Putting a member name in verbatim and adding a continuation line
causes the LaTeX output to generate something like:
item[atomic_t prios\[4\]] array with elements to store the array priorities

Everything inside "item" is non-breakable, with may produce
lines bigger than the column width.

Also, for function members, like:

        int (* rx_read) (struct v4l2_subdev *sd, u8 *buf, size_t count,ssize_t *num);

It puts the name of the member at the end, like:

        int (*) (struct v4l2_subdev *sd, u8 *buf, size_t count,ssize_t *num) read

With is very confusing.

The best is to highlight what really matters: the member name.
is a secondary information.

So, change kernel-doc, for it to produce the output on a different way:

**Members**

``prios[4]``

  array with elements to store the array priorities

Also, as the type is not part of LaTeX "item[]", LaTeX will split it into
multiple lines, if needed.

So, both LaTeX/PDF and HTML outputs will look good.

It should be noticed, however, that the way Sphinx LaTeX output handles
things like:

Foo
   bar

is different than the HTML output. On HTML, it will produce something
like:

**Foo**
   bar

While, on LaTeX, it puts both foo and bar at the same line, like:

**Foo** bar

Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>

[media] dvb_frontend: Use memdup_user() rather than duplicating its implementation

* Reuse existing functionality from memdup_user() instead of keeping
  duplicate source code.

  This issue was detected by using the Coccinelle software.

* Return directly if this copy operation failed.

* Replace the specification of data structures by pointer dereferences
  to make the corresponding size determination a bit safer according to
  the Linux coding style convention.

Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] vb2: Fix vb2_core_dqbuf() kernel-doc

The kernel-doc has the wrong function name and also the pindex
parameter is missing in the documentation.

Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: v4l2-ctrls: append missing h264 profile string

This appends missing "Stereo High" h264 profile string. Without
it the v4l2 compliance would crash kernel with NULL pointer
dereference at:

[   26.882278] [<ffff000008685cbc>] std_validate+0x378/0x42c
[   26.886967] [<ffff000008687424>] set_ctrl+0x8c/0x134
[   26.892521] [<ffff00000868755c>] v4l2_s_ctrl+0x90/0xf4
[   26.897555] [<ffff00000867f3b0>] v4l_s_ctrl+0x4c/0x110
[   26.902503] [<ffff00000867db04>] __video_do_ioctl+0x240/0x2b4
[   26.907625] [<ffff00000867d778>] video_usercopy+0x33c/0x46c
[   26.913441] [<ffff00000867d8bc>] video_ioctl2+0x14/0x1c
[   26.918822] [<ffff000008678878>] v4l2_ioctl+0xe0/0x110
[   26.924032] [<ffff0000081da898>] do_vfs_ioctl+0xb4/0x764
[   26.929238] [<ffff0000081dafcc>] SyS_ioctl+0x84/0x98
[   26.934707] [<ffff000008082f4c>] __sys_trace_return+0x0/0x4

Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media-types.rst: fix typo

Fix copy-and-paste error: the radio devices are /dev/radio, not /dev/vbi.

Signed-off-by: Hans Verkuil <<hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] helene: fix memory leak when heleno_x_pon fails

The error return path of failed calls to heleno_x_pon leak
memory because priv is not kfree'd.  Fix this by kfree'ing
priv before returning.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Abylay Ospan <aospan@netup.ru>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] staging: media: lirc: lirc_sasem: don't print error when allocating urb fails

kmalloc will print enough information in case of failure.

Signed-off-by: Wolfram Sang <wsa-dev@sang-engineering.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] staging: media: lirc: lirc_imon: don't print error when allocating urb fails

kmalloc will print enough information in case of failure.

Signed-off-by: Wolfram Sang <wsa-dev@sang-engineering.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] v4l2-async: remove unneeded .registered_async callback

The v4l2_subdev_core_ops .registered_async callback was added to notify
a subdev when its entity has been registered with the media device, to
allow for example to modify the media graph (i.e: adding entities/links).

But that's not needed since there is already a .registered callback in
struct v4l2_subdev_internal_ops that's called after the entity has been
registered with the media device in v4l2_device_register_subdev().

Suggested-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] tvp5150: use sd internal ops .registered instead .registered_async

The driver is using the struct v4l2_subdev_core_ops .registered_async
callback to register the connector entities and create the pad links
after the subdev entity has been registered with the media device.

But the .registered_async callback isn't needed since the v4l2 core
already calls the struct v4l2_subdev_internal_ops .registered callback
in v4l2_device_register_subdev(), after media_device_register_entity().

So, use the .registered() callback instead of the .registered_async()
that is going to be removed in a following patch since isn't needed.

Suggested-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] v4l: Do not allow re-registering sub-devices

Albeit not prohibited explicitly, re-registering sub-devices generated a
big, loud warning which quite likely soon was followed by a crash. What
followed was re-initialising a media entity, driver's registered() callback
being called and re-adding a list entry to a list.

Prevent this by returning an error if a sub-device is already registered.

[mchehab@s-opensource.com: reorder logic to check if !sd before dereferencing it]
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media-entity.h: remove redundant macro definition for gobj_to_pad()

The macro gobj_to_pad() is defined twice in media-entity.h.
Let's remove one.

Signed-off-by: Liu Ying <gnuiyl@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media-entity.h: Correct KernelDoc of media_entity_enum_empty()

The function media_entity_enum_empty() returns true when the bitmap
of the input parameter media entity enumeration is empty instead of marked.
This patch corrects the return value description of the function.

Signed-off-by: Liu Ying <gnuiyl@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media-entity.h: remove redundant macro definition for gobj_to_link()

The macro gobj_to_link() is defined twice in media-entity.h.
Let's remove one.

Signed-off-by: Liu Ying <gnuiyl@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] drxd_hard: Add missing error code assignment before test

It is likely that checking the result of the 2nd 'read16' is expected here.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] cx23885: Add support for Hauppauge WinTV quadHD ATSC version

Add support for the Hauppauge WinTV quadHD ATSC version.
IR support has not been provided, all 4 tuners, demodulators etc are working.
Further documentation can be found on Linux TV wiki.

Signed-off-by: Stephen Backway <stev391@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] rcar-fcp: Make sure rcar_fcp_enable() returns 0 on success

When resuming from suspend-to-RAM on r8a7795/salvator-x:

    dpm_run_callback(): pm_genpd_resume_noirq+0x0/0x90 returns 1
    PM: Device fe940000.fdp1 failed to resume noirq: error 1
    dpm_run_callback(): pm_genpd_resume_noirq+0x0/0x90 returns 1
    PM: Device fe944000.fdp1 failed to resume noirq: error 1
    dpm_run_callback(): pm_genpd_resume_noirq+0x0/0x90 returns 1
    PM: Device fe948000.fdp1 failed to resume noirq: error 1

According to its documentation, rcar_fcp_enable() returns 0 on success
or a negative error code if an error occurs.  Hence
fdp1_pm_runtime_resume() and vsp1_pm_runtime_resume() forward its return
value to their callers.

However, rcar_fcp_enable() forwards the return value of
pm_runtime_get_sync(), which can actually be 1 on success, leading to
the resume failure above.

To fix this, consider only negative values returned by
pm_runtime_get_sync() to be failures.

Fixes: 7b49235e83b2347c ("[media] v4l: Add Renesas R-Car FCP driver")

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: pci: pt3: don't print error when adding adapter fails

The core will do this for us now.

Signed-off-by: Wolfram Sang <wsa-dev@sang-engineering.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: usb: dvb-usb-v2: dvb_usb_core: don't print error when adding adapter fails

The core will do this for us now.

Signed-off-by: Wolfram Sang <wsa-dev@sang-engineering.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: pci: netup_unidvb: don't print error when adding adapter fails

The core will do this for us now.

Signed-off-by: Wolfram Sang <wsa-dev@sang-engineering.com>
Acked-by: Abylay Ospan <aospan@netup.ru>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] adv7511: fix error return code in adv7511_probe()

Fix to return error code -ENOMEM from the i2c client register error
handling case instead of 0, as done elsewhere in this function.

Signed-off-by: Wei Yongjun <weiyj.lk@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] VIDEO_MEDIATEK_VPU should depend on HAS_DMA

If NO_DMA=y:

    ERROR: "bad_dma_ops" [drivers/media/platform/mtk-vpu/mtk-vpu.ko] undefined!

Add a dependency on HAS_DMA to fix this.

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] exynos4-is: fimc-is-i2c: don't print error when adding adapter fails

The core will do this for us now.

Signed-off-by: Wolfram Sang <wsa-dev@sang-engineering.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] s5p-mfc: Remove deprecated create_singlethread_workqueue

alloc_workqueue replaces deprecated create_singlethread_workqueue().

The MFC device driver is a v4l2 driver which can encode/decode video
raw/elementary streams and has support for all popular video codecs.

The driver's watchdog_workqueue has been replaced with system_wq since
it queues a single work item, &dev->watchdog_work, which calls for no
ordering requirement. The work item is involved in running the watchdog
timer and is not being used on a memory reclaim path.

Work item has been flushed in s5p_mfc_remove() to ensure
that there are no pending tasks while disconnecting the driver.

Signed-off-by: Bhaktipriya Shridhar <bhaktipriya96@gmail.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: s5p-jpeg add missing blank lines after declarations

Missing blank lines after declarations are making it hard to read the
code. Fix them and also fix other checkpatch warnings at the same time.

Signed-off-by: Shuah Khan <shuahkh@osg.samsung.com>
Acked-by: Jacek Anaszewski <j.anaszewski@samsung.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: s5p-mfc remove unnecessary error messages

Removed unnecessary error message as appropriate error code is returned.
Changed error message into a debug.

Signed-off-by: Shuah Khan <shuahkh@osg.samsung.com>
Reviewed-by: Javier Martinez Canillas <javier@osg.samsung.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: s5p-mfc Fix misspelled error message and checkpatch errors

Fix misspelled error message and existing checkpatch errors in the
error message conditional.

WARNING: suspect code indent for conditional statements (8, 24)
        if (ctx->state != MFCINST_HEAD_PARSED &&
[...]
        mfc_err("Can not get crop information\n");

Signed-off-by: Shuah Khan <shuahkh@osg.samsung.com>
Reviewed-by: Javier Martinez Canillas <javier@osg.samsung.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] s5p-mfc: remove redundant return value check of platform_get_resource()

Remove unneeded error handling on the result of a call
to platform_get_resource() when the value is passed to
devm_ioremap_resource().

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: s5p-mfc remove void function return statement

Remove void function return statement

Signed-off-by: Shuah Khan <shuahkh@osg.samsung.com>
Reviewed-by: Javier Martinez Canillas <javier@osg.samsung.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: s5p-mfc fix invalid memory access from s5p_mfc_release()

If s5p_mfc_release() runs after s5p_mfc_remove(), the former will access
invalid s5p_mfc_dev pointer saved in the s5p_mfc_ctx and runs into kernel
paging request errors.

Clear ctx dev pointer in s5p_mfc_remove() and change s5p_mfc_release() to
avoid work that requires ctx->dev.

odroid kernel: Unable to handle kernel paging request at virtual address
    f17c1104
odroid kernel: pgd = ebca4000
odroid kernel: [f17c1104] *pgd=6e23d811, *pte=00000000, *ppte=00000000
odroid kernel: Internal error: Oops: 807 [#1] PREEMPT SMP ARM
odroid kernel: Modules linked in: cpufreq_userspace cpufreq_powersave
    cpufreq_conservative s5p_mfc s5p_jpeg v4l2_mem2mem
    videobuf2_dma_contig videobuf2_memops videobuf2_v4l2 videobuf2_core
    v4l2_common videodev media
odroid kernel: Hardware name: SAMSUNG EXYNOS (Flattened Device Tree)
odroid kernel: task: c2241400 ti: e7018000 task.ti: e7018000
odroid kernel: PC is at s5p_mfc_reset+0x40/0x28c [s5p_mfc]
odroid kernel: LR is at s5p_mfc_reset+0x34/0x28c [s5p_mfc]
odroid kernel: pc : [<bf15bfbc>]    lr : [<bf15bfb0>] psr: 60010013
odroid kernel: [<bf15bfbc>] (s5p_mfc_reset [s5p_mfc]) from [<bf15c62c>]
    (s5p_mfc_deinit_hw+0x14/0x3c [s5p_mfc])
odroid kernel: [<bf15c62c>] (s5p_mfc_deinit_hw [s5p_mfc]) from [<bf155958>]
    (s5p_mfc_release+0xac/0x1c4 [s5p_mfc])
odroid kernel: [<bf155958>] (s5p_mfc_release [s5p_mfc]) from [<bf021344>]
    (v4l2_release+0x38/0x74 [videodev])
odroid kernel: [<bf021344>] (v4l2_release [videodev]) from [<c01e4274>]
    (__fput+0x80/0x1c8)
odroid kernel: [<c01e4274>] (__fput) from [<c0135c58>]
    (task_work_run+0x94/0xc8)
odroid kernel: [<c0135c58>] (task_work_run) from [<c010a9d4>]
    (do_work_pending+0x7c/0xa4)
odroid kernel: [<c010a9d4>] (do_work_pending) from [<c0107794>]
    (slow_work_pending+0xc/0x20)
odroid kernel: Code: eb3edacc e5953078 e3a06000 e2833c11 (e5836004)

Signed-off-by: Shuah Khan <shuahkh@osg.samsung.com>
Tested-by: Luis de Bethencourt <luisbg@osg.samsung.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: Doc s5p-mfc add missing fields to s5p_mfc_dev structure definition

Add missing documentation for s5p_mfc_dev structure definition.

Signed-off-by: Shuah Khan <shuahkh@osg.samsung.com>
Reviewed-by: Javier Martinez Canillas <javier@osg.samsung.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] gsc-m2m: improve v4l2_capability driver and card fields

According to the V4L2 documentation the driver and card fields should be
used to identify the driver and the device but the gsc-m2m driver fills
those field using the platform device name, which in turn is the name of
the device DT node.

So not only the filled information isn't correct but also the same values
are used in the driver, card and bus_info fields.

Before this patch:

Driver Info (not using libv4l2):
        Driver name   : 13e00000.video-
        Card type     : 13e00000.video-scaler
        Bus info      : platform:13e00000.video-scaler
        Driver version: 4.7.0

After this patch:

Driver Info (not using libv4l2):
        Driver name   : exynos-gsc
        Card type     : exynos-gsc gscaler
        Bus info      : platform:13e00000.video-scaler
        Driver version: 4.7.0

Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] gsc-m2m: add device name sufix to bus_info capatiliby field

The driver doesn't set the device in the struct v4l2_capability bus_info
field so v4l2-compliance reports the following error for VIDIOC_QUERYCAP:

Required ioctls:
                fail: v4l2-compliance.cpp(537): missing bus_info prefix ('platform')
        test VIDIOC_QUERYCAP: FAIL

This patch fixes this by filling also the device besides the bus.

Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] s5p-jpeg: only fill driver's name in capabilities driver field

The driver fills in both the struct v4l2_capability driver and card fields
the same values, that is the driver's name plus the information if the dev
is a decoder or an encoder.

But the driver field has a fixed length of 16 bytes so the filled data is
truncated:

Driver Info (not using libv4l2):
        Driver name   : s5p-jpeg decode
        Card type     : s5p-jpeg decoder
        Bus info      : platform:11f50000.jpeg
        Driver version: 4.7.0

Also, this field should only contain the driver's name so use just that.
The information if the device is a decoder or an encoder is in the card
type field anyways.

Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>
Acked-by: Jacek Anaszewski <j.anaszewski@samsung.com>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Shuah Khan <shuahkh@osg.samsung.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] exynos4-is: Fix fimc_is_parse_sensor_config() nodes handling

The same struct device_node * is used for looking up the I2C sensor, OF
graph endpoint and port. So the reference count is incremented but not
decremented for the endpoint and port nodes.

Fix this by having separate pointers for each node looked up.

Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>
Tested-by: Nicolas Dufresne <nicoas.dufresne@collabora.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] s5p-jpeg: set capablity bus_info as required by VIDIOC_QUERYCAP

The driver doesn't set the struct v4l2_capability cap_info field so the
v4l2-compliance tool reports the following errors for VIDIOC_QUERYCAP:

Required ioctls:
                VIDIOC_QUERYCAP returned 0 (Success)
                fail: v4l2-compliance.cpp(304): string empty
                fail: v4l2-compliance.cpp(528): check_ustring(vcap.bus_info, sizeof(vcap.bus_info))
        test VIDIOC_QUERYCAP: FAIL

This patch fixes by setting the field in VIDIOC_QUERYCAP ioctl handler:

Required ioctls:
                VIDIOC_QUERYCAP returned 0 (Success)
        test VIDIOC_QUERYCAP: OK

Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>
Acked-by: Jacek Anaszewski <j.anaszewski@samsung.com>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Shuah Khan <shuahkh@osg.samsung.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] tw5864: remove two unused vars

Remove those two vars that aren't used, as reported by smatch:

drivers/media/pci/tw5864/tw5864-video.c: In function 'tw5864_prepare_frame_headers':
drivers/media/pci/tw5864/tw5864-video.c:1219:16: warning: variable 'space_before_sl_hdr' set but not used [-Wunused-but-set-variable]
  unsigned long space_before_sl_hdr;
                ^~~~~~~~~~~~~~~~~~~
drivers/media/pci/tw5864/tw5864-video.c:1218:6: warning: variable 'sl_hdr' set but not used [-Wunused-but-set-variable]
  u8 *sl_hdr;
      ^~~~~~

Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] tw5864: remove double irq lock code

As warned by smatch:
drivers/media/pci/tw5864/tw5864-core.c:160 tw5864_h264_isr() error: double lock 'irqsave:flags'
drivers/media/pci/tw5864/tw5864-core.c:174 tw5864_h264_isr() error: double unlock 'irqsave:flags'

Remove the IRQ duplicated lock.

Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] tw5864: add missing HAS_DMA dependency

Fix this warning:

warning: (VIDEO_TW5864 && VIDEO_MEDIATEK_VCODEC) selects VIDEOBUF2_DMA_CONTIG which has unmet direct dependencies (MEDIA_SUPPORT && HAS_DMA)

This driver depends on HAS_DMA.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] pci: Add tw5864 driver

Support for boards based on Techwell TW5864 chip which provides
multichannel video & audio grabbing and encoding (H.264, MJPEG,
ADPCM G.726).

This submission implements only H.264 encoding of all channels at D1
resolution.

Thanks to Mark Thompson <sw@jkqxz.net> for help, and for contribution of
H.264 startcode emulation prevention code.

Signed-off-by: Andrey Utkin <andrey.utkin@corp.bluecherry.net>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] extended-controls.rst: fix a build warning

/devel/v4l/patchwork/Documentation/media/uapi/v4l/extended-controls.rst:2116: WARNING: Inline literal start-string without end-string.

Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] vidioc-g-dv-timings.rst: document the v4l2_bt_timings reserved field

This field was never documented, and neither was it mentioned that
it should be zeroed by the application.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] vb2: add WARN_ONs checking if a valid struct device was passed

The dma-contig and dma-sg memops require a valid struct device for
the DMA to be handled correctly.

Call WARN_ON and return ERR_PTR(-EINVAL) if it was NULL.

Setting the correct device pointer was forgotten in several new driver
submissions. This was caught during code review, but it really should be
caught in the code.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] vb2: don't return NULL for alloc and get_userptr ops

Always return an ERR_PTR() instead of NULL.

This makes the behavior of alloc, get_userptr and attach_dmabuf the
same.

Update the documentation in videobuf2-core.h as well.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] vb2: remove TODO comment for dma-buf in QBUF

There is a TODO comment about the dma-buf being mapped in VIDIOC_QBUF
instead of doing it closer to when the actual DMA is going to happen
when the buffers are queued in the driver (i.e: __enqueue_in_driver).

But there is a reason to do it earlier in QBUF, and is that userspace
has no way to know if a exported dma-buf can be imported successfully
and so relies on QBUF succeeding as indication that the dma-buf mapped.

If QBUF fails, the application can fallback to another streaming I/O
method. But moving the dma-buf mapping later when queueing the buffers
can be too late for userspace to recover, since it may had dropped the
buffer(s) already when it knows that the dma-buf mapping failed.

So remove the TODO instead and change the comment to explain this.

Suggested-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] vb2: include lengths in dmabuf qbuf debug message

If the VIDIOC_QBUF ioctl fails due a wrong dmabuf length, it's
useful to get the invalid and minimum lengths as a debug info.

Before this patch:

vb2-core: __qbuf_dmabuf: invalid dmabuf length for plane 1

After this patch:

vb2-core: __qbuf_dmabuf: invalid dmabuf length 221248 for plane 1, minimum length 410880

Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] sur40: properly report a single frame rate of 60 FPS

The device hardware is always running at 60 FPS, so report this both via
PARM_IOCTL and ENUM_FRAMEINTERVALS.

[hans.verkuil@cisco.com: fix suspect indent checkpatch warning]
[mchehab@s-opensource.com: fix a trivial merge conflict]

Signed-off-by: Martin Kaltenbrunner <modin@yuri.at>
Signed-off-by: Florian Echtler <floe@butterbrot.org>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] Documentation: Fix V4L2_CTRL_FLAG_VOLATILE

V4L2_CTRL_FLAG_VOLATILE behaviour when V4L2_CTRL_FLAG_EXECUTE_ON_WRITE
is set was not properly explained.

Also set some hyperlink to ease the Documentation browsing.

Credit-to: Hans Verkuil <hansverk@cisco.com>

[mchehab@s-opensource.com: fix a trivial merge conflict]
Reported-by: Dimitrios Katsaros <patcherwork@gmail.com>
Signed-off-by: Ricardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: doc-rst: document ENODATA for cropping ioctls

Document that the cropping ioctls can return ENODATA if the operation
isn't supported for the current input or output.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] vivid: return -ENODATA if the current input doesn't support g/s_selection

Returning -EINVAL indicates wrong arguments, but that's not the case
here.

Returning -ENOTTY is also no option, since the ioctl is implemented, but
it just is not valid for this input.

So use -ENODATA instead. This is also used elsewhere when an ioctl isn't
valid for a specific input.

In this case G/S_SELECTION returned -EINVAL for the webcam input. That
input doesn't support cropping, instead it uses ENUM_FRAMESIZES to
enumerate a list of discrete frame sizes.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] v4l2-subdev: rename cropcap to g_pixelaspect

The old cropcap video op is now only used to pass the pixelaspect
ratio, so rename it.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] pvrusb2: convert g/s_crop to g/s_selection

This is part of a final push to convert all drivers to g/s_selection.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Cc: Mike Isely <isely@pobox.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] vpbe_display: convert g/s_crop to g/s_selection

This is part of a final push to convert all drivers to g/s_selection.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Cc: Prabhakar Lad <prabhakar.csengg@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] vpfe_capture: convert g/s_crop to g/s_selection

This is part of a final push to convert all drivers to g/s_selection.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Cc: Prabhakar Lad <prabhakar.csengg@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] zoran: convert g/s_crop to g/s_selection

This is part of a final push to convert all drivers to g/s_selection.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] saa7134: convert g/s_crop to g/s_selection

This is part of a final push to convert all drivers to g/s_selection.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] omap_vout: convert g/s_crop to g/s_selection

This is part of a final push to convert all drivers to g/s_selection.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] bttv: convert g/s_crop to g/s_selection

This is part of a final push to convert all drivers to g/s_selection.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] v4l2: remove g/s_crop from video ops

Replace all calls to g/s_crop by calls to the get/set_selection pad ops.

Remove the old g/s_crop video ops since they are now no longer used.

The cropcap video op is now only used to pass pixelaspect information,
and is only needed if the pixelaspect is not 1:1.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Cc: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: adv7180: add power pin control

Some targets control the ADV7180 power pin via a gpio, so add
optional support for "powerdown" pin control.

Signed-off-by: Steve Longerbeam <steve_longerbeam@mentor.com>
Tested-by: Tim Harvey <tharvey@gateworks.com>
Acked-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: adv7180: define more registers

Replace hard-coded addresses with new register macro defines. No
functional changes.

Signed-off-by: Steve Longerbeam <steve_longerbeam@mentor.com>
Acked-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] tw686x: Delete an unnecessary check before the function call "video_unregister_device"

The video_unregister_device() function tests whether its argument is NULL
and then returns immediately. Thus the test around the call is not needed.

This issue was detected by using the Coccinelle software.

Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] v4l2-common: Delete an unnecessary check before the function call "spi_unregister_device"

The spi_unregister_device() function tests whether its argument is NULL
and then returns immediately. Thus the test around the call is not needed.

This issue was detected by using the Coccinelle software.

Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] cec: Delete an unnecessary check before the function call "rc_free_device"

The rc_free_device() function tests whether its argument is NULL
and then returns immediately. Thus the test around the call is not needed.

This issue was detected by using the Coccinelle software.

Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] cobalt: support reduced fps

Add support for reduced fps (i.e. 59.94 Hz instead of 60 Hz) for the
HDMI output.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] ad9389b: Remove deprecated create_singlethread_workqueue

The workqueue work_queue is involved in EDID (Extended Display
Identification Data) handling.

It has a single work item(&state->edid_handler) and hence
doesn't require ordering. It is not being used on a memory reclaim path.
Hence, the singlethreaded workqueue has been replaced with
the use of system_wq.

&state->edid_handler is a self requeueing work item and it has been
been sync cancelled in ad9389b_remove() to ensure that nothing is
pending when the driver is disconnected.

The unused label err_unreg has also been dropped.

Signed-off-by: Bhaktipriya Shridhar <bhaktipriya96@gmail.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] gspca: finepix: Remove deprecated create_singlethread_workqueue

The workqueue "work_thread" is involved in streaming the camera data.
It has a single work item(&dev->work_struct) and hence doesn't require
ordering. Also, it is not being used on a memory reclaim path.
Hence, the singlethreaded workqueue has been replaced with the use of
system_wq.

System workqueues have been able to handle high level of concurrency
for a long time now and hence it's not required to have a singlethreaded
workqueue just to gain concurrency. Unlike a dedicated per-cpu workqueue
created with create_singlethread_workqueue(), system_wq allows multiple
work items to overlap executions even on the same CPU; however, a
per-cpu workqueue doesn't have any CPU locality or global ordering
guarantee unless the target CPU is explicitly specified and thus the
increase of local concurrency shouldn't make any difference.

Work item has been flushed in sd_stop0() to ensure that there are no
pending tasks while disconnecting the driver.

Signed-off-by: Bhaktipriya Shridhar <bhaktipriya96@gmail.com>
Acked-by: Tejun Heo <tj@kernel.org>
Acked-by: Frank Zago <frank@zago.net>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] gspca: jl2005bcd: Remove deprecated create_singlethread_workqueue

The workqueue "work_thread" is involved in streaming the camera data.
It has a single work item(&sd->work_struct) and hence doesn't require
ordering. Also, it is not being used on a memory reclaim path.
Hence, the singlethreaded workqueue has been replaced with the use of
system_wq.

System workqueues have been able to handle high level of concurrency
for a long time now and hence it's not required to have a singlethreaded
workqueue just to gain concurrency. Unlike a dedicated per-cpu workqueue
created with create_singlethread_workqueue(), system_wq allows multiple
work items to overlap executions even on the same CPU; however, a
per-cpu workqueue doesn't have any CPU locality or global ordering
guarantee unless the target CPU is explicitly specified and thus the
increase of local concurrency shouldn't make any difference.

Work item has been flushed in sd_stop0() to ensure that there are no
pending tasks while disconnecting the driver.

Signed-off-by: Bhaktipriya Shridhar <bhaktipriya96@gmail.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] gspca: vicam: Remove deprecated create_singlethread_workqueue

The workqueue "work_thread" is involved in streaming the camera data.
It has a single work item(&sd->work_struct) and hence doesn't require
ordering. Also, it is not being used on a memory reclaim path.
Hence, the singlethreaded workqueue has been replaced with the use of
system_wq.

System workqueues have been able to handle high level of concurrency
for a long time now and hence it's not required to have a singlethreaded
workqueue just to gain concurrency. Unlike a dedicated per-cpu workqueue
created with create_singlethread_workqueue(), system_wq allows multiple
work items to overlap executions even on the same CPU; however, a
per-cpu workqueue doesn't have any CPU locality or global ordering
guarantee unless the target CPU is explicitly specified and thus the
increase of local concurrency shouldn't make any difference.

Work item has been flushed in sd_stop0() to ensure that there are no
pending tasks while disconnecting the driver.

Signed-off-by: Bhaktipriya Shridhar <bhaktipriya96@gmail.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] gspca: sonixj: Remove deprecated create_singlethread_workqueue

The workqueue "work_thread" is involved in updating the JPEG quality
of the gspca_dev. It has a single work item(&sd->work) and hence doesn't
require ordering. Also, it is not being used on a memory reclaim path.
Hence, the singlethreaded workqueue has been replaced with the use of
system_wq.

System workqueues have been able to handle high level of concurrency
for a long time now and hence it's not required to have a singlethreaded
workqueue just to gain concurrency. Unlike a dedicated per-cpu workqueue
created with create_singlethread_workqueue(), system_wq allows multiple
work items to overlap executions even on the same CPU; however, a
per-cpu workqueue doesn't have any CPU locality or global ordering
guarantee unless the target CPU is explicitly specified and thus the
increase of local concurrency shouldn't make any difference.

Work item has been flushed in sd_stop0() to ensure that there are no
pending tasks while disconnecting the driver.

Signed-off-by: Bhaktipriya Shridhar <bhaktipriya96@gmail.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] pvrusb2: Remove deprecated create_singlethread_workqueue

The workqueue "workqueue" is involved in polling the pvrusb2 hardware
(pvr2_hdw).

It has a single work item(&hdw->workpoll) and hence doesn't require
ordering. Also, it is not being used on a memory reclaim path.
Hence, the singlethreaded workqueue has been replaced with the use of
system_wq.

System workqueues have been able to handle high level of concurrency
for a long time now and hence it's not required to have a singlethreaded
workqueue just to gain concurrency. Unlike a dedicated per-cpu workqueue
created with create_singlethread_workqueue(), system_wq allows multiple
work items to overlap executions even on the same CPU; however, a
per-cpu workqueue doesn't have any CPU locality or global ordering
guarantee unless the target CPU is explicitly specified and thus the
increase of local concurrency shouldn't make any difference.

Work item has been flushed in pvr2_hdw_destroy to ensure that there are no
pending tasks while disconnecting the driver.

Signed-off-by: Bhaktipriya Shridhar <bhaktipriya96@gmail.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: usb: zr364xx: zr364xx: don't print error when allocating urb fails

kmalloc will print enough information in case of failure.

Signed-off-by: Wolfram Sang <wsa-dev@sang-engineering.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: usb: usbvision: usbvision-core: don't print error when allocating urb fails

kmalloc will print enough information in case of failure.

Signed-off-by: Wolfram Sang <wsa-dev@sang-engineering.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: usb: tm6000: tm6000-video: don't print error when allocating urb fails

kmalloc will print enough information in case of failure.

Signed-off-by: Wolfram Sang <wsa-dev@sang-engineering.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: usb: tm6000: tm6000-dvb: don't print error when allocating urb fails

kmalloc will print enough information in case of failure.

Signed-off-by: Wolfram Sang <wsa-dev@sang-engineering.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: usb: stkwebcam: stk-webcam: don't print error when allocating urb fails

kmalloc will print enough information in case of failure.

Signed-off-by: Wolfram Sang <wsa-dev@sang-engineering.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: usb: stk1160: stk1160-video: don't print error when allocating urb fails

kmalloc will print enough information in case of failure.

Signed-off-by: Wolfram Sang <wsa-dev@sang-engineering.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>

[media] media: usb: s2255: s2255drv: don't print error when allocating urb fails

kmalloc will print enough information in case of failure.

Signed-off-by: Wolfram Sang <wsa-dev@sang-engineering.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>