This patch adds extension to V4L2 api. A new ioctl VIDIOC_EXPBUF is added.  The
ioctl is used to export an mmap buffer as a DMABUF file descriptor.
Signed-off-by: NTomasz Stanislawski <t.stanislaws@samsung.com>
Signed-off-by: NKyungmin Park <kyungmin.park@samsung.com>
Acked-by: NHans Verkuil <hans.verkuil@cisco.com>
Tested-by: NMauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

This makes it easier to read and also ties in more closely with the
profile concept.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The ioctl validity checks have been improved and now take vfl_type
and vfl_dir into account.
During the 2012 Media Workshop it was decided that these improved
v4l2 core checks should be added as they simplified drivers and
made drivers behave consistently.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Currently the core code looks at tvnorms to see whether ENUMSTD
or G_PARM should be enabled. This is not a good check for drivers
that support the STD API on one input and the DV Timings API on another.
In that case tvnorms may be 0.
Instead check whether s_std is present (for ENUMSTD) or whether g_std or
current_norm is present for g_parm.
Also, in the enumstd core function return ENODATA if tvnorms is 0,
because in that case the current input does not support the STD API
and ENUMSTD should return ENODATA for that.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: NSakari Ailus <sakari.ailus@iki.fi>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

This buffer type isn't used at all, and since it is effectively undefined
what it should do it is deprecated. The define still exists, but any
internal support for such buffers is removed.
The decisions to deprecate this was taken during the 2012 Media Workshop.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Currently, the v4l2 core is mixed together with other non-core drivers.
Move them into a separate directory.
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

All drivers that needed V4L2_FL_LOCK_ALL_FOPS have been converted,
so remove this flag altogether.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

It is possible that video_put() releases video_device struct,
provoking a panic when debug printk wants to get video_device node name.
Signed-off-by: NEzequiel Garcia <elezegarcia@gmail.com>
Acked-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

This adds the usual core support code for this new ioctl.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

G_PARM should only be enabled if:

- vidioc_g_parm is present
- or: it is a video node and vidioc_g_std or tvnorms are set.

Without this additional check v4l2-compliance would complain about
being able to use g_parm when it didn't expect it.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The VIDIOC_DV_TIMINGS_CAP ioctl check wasn't added to determine_valid_ioctls().
This caused this ioctl to always return -ENOTTY.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Cc: stable@kernel.org
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The VIDIOC_DV_TIMINGS_CAP ioctl check wasn't added to determine_valid_ioctls().
This caused this ioctl to always return -ENOTTY.

The cause for this was that for 3.5 two patch series were merged, one
changing V4L2 core ioctl handling and one adding new functionality, and
some of the new functionality wasn't handled by the new V4L2 core code.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
[ Taking it directly due to vacations  - Linus ]
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

This prepares struct video_device for easier integration with vb2.

It also introduces a new lock that protects the vb2_queue. It is up
to the driver to use it or not. And the driver can associate an owner
filehandle with the queue to check whether queuing requests are
permitted for the calling filehandle.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Since this could theoretically change the debug value while in the middle
of v4l2-ioctl.c, we make a copy of vfd->debug to ensure consistent debug
behavior.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The field current_norm does not have to be set for g_parm to be a valid
ioctl. Remove that check, but add a check whether this is a video node
instead as g_parm only makes sense for those nodes.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Now query/enum_dv_timings finally work again.

The timings API patches and the core ioctl changes clearly sailed right past
each other without realizing that both needed to adapt to the other.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Rename the function v4l2_dont_use_lock to v4l2_disable_ioctl_locking,
and rename v4l2_dont_use_cmd to v4l2_disable_ioctl.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

This used to be the default if the lock pointer was set, but now that lock is by
default only used for ioctl serialization. Those drivers that already used
core locking have this flag set explicitly, except for some drivers where
it was obvious that there was no need to serialize any file operations other
than ioctl.

The drivers that didn't need this flag were:

drivers/media/radio/dsbr100.c
drivers/media/radio/radio-isa.c
drivers/media/radio/radio-keene.c
drivers/media/radio/radio-miropcm20.c
drivers/media/radio/radio-mr800.c
drivers/media/radio/radio-tea5764.c
drivers/media/radio/radio-timb.c
drivers/media/video/vivi.c
sound/i2c/other/tea575x-tuner.c

The other drivers that use core locking and where it was not immediately
obvious that this flag wasn't needed were changed so that the flag is set
together with a comment that that driver needs work to avoid having to
set that flag. This will often involve taking the core lock in the fops
themselves.

Eventually this flag should go and it should not be used in new drivers.

There are a few reasons why we want to avoid core locking of non-ioctl
fops: in the case of mmap this can lead to a deadlock in rare situations
since when mmap is called the mmap_sem is held and it is possible for
other parts of the code to take that lock as well (copy_from_user()/copy_to_user()
perform a down_read(&mm->mmap_sem) when a page fault occurs).

It is very unlikely that that happens since the core lock serializes all
fops, but the kernel warns about it if lock validation is turned on.

For poll it is also undesirable to take the core lock as that can introduce
increased latency. The same is true for read/write.

While it was possible to make flags or something to turn on/off taking the
core lock for each file operation, in practice it is much simpler to just
not take it at all except for ioctl and leave it to the driver to take the
lock. There are only a handful fops compared to the zillion ioctls we have.

I also wanted to make it obvious which drivers still take the lock for all
fops, so that's why I chose to have drivers set it explicitly.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Acked-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Rather than testing whether an ioctl is implemented in the driver or not
every time the ioctl is called, do it upfront when the device is registered.

This also allows a driver to disable certain ioctls based on the capabilities
of the detected board, something you can't do today without creating separate
v4l2_ioctl_ops structs for each new variation.

For the most part it is pretty straightforward, but for control ioctls a flag
is needed since it is possible that you have per-filehandle controls, and that
can't be determined upfront of course.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Acked-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Using the V4L2 core lock is a very robust method that is usually very good
at doing the right thing. But some drivers, particularly USB drivers, may
want to prevent the core from taking the lock for specific ioctls, particularly
buffer queuing ioctls.

The reason is that certain commands like S_CTRL can take a long time to process
over USB and all the time the core has the lock, preventing VIDIOC_DQBUF from
proceeding, even though a frame may be ready in the queue.

This introduces unwanted latency.

Since the buffer queuing commands often have their own internal lock it is
often not necessary to take the core lock. Drivers can now say that they don't
want the core to take the lock for specific ioctls.

As it is a specific opt-out it makes it clear to the reviewer that those
ioctls will need more care when reviewing.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Acked-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Remove all #inclusions of asm/system.h preparatory to splitting and killing
it.  Performed with the following command:

perl -p -i -e 's!^#\s*include\s*<asm/system[.]h>.*\n!!' `grep -Irl '^#\s*include\s*<asm/system[.]h>' *`
Signed-off-by: NDavid Howells <dhowells@redhat.com>

As the V4L2 based UVC webcam gadget (g_webcam) expects the
'videodev' to present when the 'webcam_bind' routine is called,
so 'videodev' should be available as early as possible.

Now, when 'g_webcam' is built as a module (i.e. not a part of
kernel) the late availability of 'videodev' is OK, but if
'g_webcam' is built statically as a part of the kernel,
the kernel crashes (a sample crash dump using Designware 2.0 UDC
is provided below).

To solve the same, this patch makes 'videodev_init' as a subsys initcall.

Kernel Crash Dump:

------------------

designware_udc designware_udc: Device Synopsys UDC probed csr 90810000: plug 90812000
g_webcam gadget: uvc_function_bind
Unable to handle kernel NULL pointer dereference at virtual address 000000e4
pgd = 80004000
[000000e4] *pgd=00000000
Internal error: Oops: 5 [#1] SMP
Modules linked in:
CPU: 0    Not tainted  (3.3.0-rc3-13888-ge774c03-dirty #20)
PC is at do_raw_spin_lock+0x10/0x16c
LR is at _raw_spin_lock+0x10/0x14
pc : [<8019e344>]    lr : [<804095c0>]    psr: 60000013
sp : 8f839d20  ip : 8f839d50  fp : 8f839d4c
r10: 80760a94  r9 : 8042de98  r8 : 00000154
r7 : 80760e94  r6 : 805cfc10  r5 : 8fb6a008  r4 : 8fb6a008
r3 : 805dd0c8  r2 : 8f839d48  r1 : 805cfc08  r0 : 000000e0
Flags: nZCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment kernel
Control: 10c5387d  Table: 0000404a  DAC: 00000015
Process swapper/0 (pid: 1, stack limit = 0x8f8382f0)
Stack: (0x8f839d20 to 0x8f83a000)
9d20: ffffffff ffffffff 8fb6a008 8fb6a008 805cfc10 80760e94 00000154 8042de98
9d40: 8f839d5c 8f839d50 804095c0 8019e340 8f839d7c 8f839d60 80222b28 804095bc
9d60: 8fb12b80 8fb6a008 8fb6a010 805cfc08 8f839dc4 8f839d80 80223db8 80222adc
9d80: 8f839dac 8f839d90 8022baa0 8019e2e8 8fb6a008 8075e7f4 8fb6a008 8fb6a008
9da0: 00000000 8fb6a008 80760e94 00000154 8042de98 80760a94 8f839ddc 8f839dc8
9dc0: 802242a8 80223d1c 8fb12b80 8fb6a000 8f839e1c 8f839de0 8030132c 80224298
9de0: 80223ce8 803f2ee8 00000001 804f7750 8f839e1c 8f824008 805cff20 8f824000
9e00: 8fb6a000 ffffffff 00000000 8f8d4880 8f839e4c 8f839e20 80562e3c 80301100
9e20: 00000000 8fb13140 8f824008 805cff20 8042aa68 8f824000 8042aa8c 805e4d40
9e40: 8f839e64 8f839e50 802d20c4 80562ba8 805d0058 805cff20 8f839e8c 8f839e68
9e60: 80563034 802d206c 8042aa8c 805cff20 8f8d4880 00000000 805cfc08 8fb12a40
9e80: 8f839e9c 8f839e90 805630c4 80562ec4 8f839ebc 8f839ea0 802d2364 805630b0
9ea0: 805cfeac 8f8d4880 805cfbe8 807605e8 8f839ed4 8f839ec0 80562b3c 802d22cc
9ec0: 80562ac4 8f8d4880 8f839f04 8f839ed8 802d0b40 80562ad0 8f839ef4 805cff90
9ee0: 805cff90 805cfb98 00000000 00000000 805cfbe8 805e4d40 8f839f3c 8f839f08
9f00: 802cd078 802d0a18 00000000 802d0a0c 00000000 8fb9ba00 802d0a0c 805cff90
9f20: 00000013 00000000 00000000 805e4d40 8f839f5c 8f839f40 802cf390 802ccff0
9f40: 00000003 00000003 804fb598 00000000 8f839f74 8f839f60 802d2554 802cf2a0
9f60: 8f838000 8057731c 8f839f84 8f839f78 80562b90 802d24d0 8f839fdc 8f839f88
9f80: 800085d4 80562b84 805af2ac 805af2ac 80562b78 00000000 00000013 00000000
9fa0: 00000000 00000000 8f839fc4 8f839fb8 80043dd0 8057706c 8057731c 8002875c
9fc0: 00000013 00000000 00000000 00000000 8f839ff4 8f839fe0 805468d4 800085a0
9fe0: 00000000 80546840 00000000 8f839ff8 8002875c 8054684c 51155555 55545555
Backtrace:
[<8019e334>] (do_raw_spin_lock+0x0/0x16c) from [<804095c0>] (_raw_spin_lock+0x10/0x14)
 r9:8042de98 r8:00000154 r7:80760e94 r6:805cfc10 r5:8fb6a008
r4:8fb6a008
[<804095b0>] (_raw_spin_lock+0x0/0x14) from [<80222b28>] (get_device_parent+0x58/0x1c0)
[<80222ad0>] (get_device_parent+0x0/0x1c0) from [<80223db8>] (device_add+0xa8/0x57c)
 r6:805cfc08 r5:8fb6a010 r4:8fb6a008 r3:8fb12b80
[<80223d10>] (device_add+0x0/0x57c) from [<802242a8>] (device_register+0x1c/0x20)
[<8022428c>] (device_register+0x0/0x20) from [<8030132c>] (__video_register_device+0x238/0x484)
 r4:8fb6a000 r3:8fb12b80
[<803010f4>] (__video_register_device+0x0/0x484) from [<80562e3c>] (uvc_function_bind+0x2a0/0x31c)
[<80562b9c>] (uvc_function_bind+0x0/0x31c) from [<802d20c4>] (usb_add_function+0x64/0x118)
[<802d2060>] (usb_add_function+0x0/0x118) from [<80563034>] (uvc_bind_config+0x17c/0x1ec)
 r5:805cff20 r4:805d0058
[<80562eb8>] (uvc_bind_config+0x0/0x1ec) from [<805630c4>] (webcam_config_bind+0x20/0x28)
 r8:8fb12a40 r7:805cfc08 r6:00000000 r5:8f8d4880 r4:805cff20
r3:8042aa8c
[<805630a4>] (webcam_config_bind+0x0/0x28) from [<802d2364>] (usb_add_config+0xa4/0x124)
[<802d22c0>] (usb_add_config+0x0/0x124) from [<80562b3c>] (webcam_bind+0x78/0xb4)
 r6:807605e8 r5:805cfbe8 r4:8f8d4880 r3:805cfeac
[<80562ac4>] (webcam_bind+0x0/0xb4) from [<802d0b40>] (composite_bind+0x134/0x308)
 r4:8f8d4880 r3:80562ac4
[<802d0a0c>] (composite_bind+0x0/0x308) from [<802cd078>] (dw_udc_start+0x94/0x2bc)
[<802ccfe4>] (dw_udc_start+0x0/0x2bc) from [<802cf390>] (usb_gadget_probe_driver+0xfc/0x180)
[<802cf294>] (usb_gadget_probe_driver+0x0/0x180) from [<802d2554>] (usb_composite_probe+0x90/0xb4)
 r6:00000000 r5:804fb598 r4:00000003 r3:00000003
[<802d24c4>] (usb_composite_probe+0x0/0xb4) from [<80562b90>] (webcam_init+0x18/0x24)
 r5:8057731c r4:8f838000
[<80562b78>] (webcam_init+0x0/0x24) from [<800085d4>] (do_one_initcall+0x40/0x184)
[<80008594>] (do_one_initcall+0x0/0x184) from [<805468d4>] (kernel_init+0x94/0x134)
[<80546840>] (kernel_init+0x0/0x134) from [<8002875c>] (do_exit+0x0/0x6f8)
 r5:80546840 r4:00000000
Code: e1a0c00d e92ddbf0 e24cb004 e24dd008 (e5902004)
---[ end trace 7ecca37f36fbdc06 ]---
Signed-off-by: NBhupesh Sharma <bhupesh.sharma@st.com>
Tested-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

When enumerating an entity, assign the entire entity specification
instead of only the first two words. (This requires giving the
specification union a name.)

So far, no driver actually uses more than two words, but this will
be needed for ALSA entities.
Signed-off-by: NClemens Ladisch <clemens@ladisch.de>
[laurent.pinchart@ideasonboard.com: Rename specification to info]
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Improve the use of the WARN_ON() macro and use a local variable, instead
of reduntantly dereferencing a pointer in v4l2-dev.c
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The change in 8280b662 does not cover the case when v4l2_dev is already
NULL, fix that.

With a Kinect sensor, seen as an USB camera using GSPCA in this context,
a NULL pointer dereference BUG can be triggered by just unplugging the
device after the camera driver has been loaded.
Signed-off-by: NAntonio Ospite <ospite@studenti.unina.it>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Drivers that have no v4l2_device release callback might free the
v4l2_device instance in the video_device release callback. Make sure we
don't access the v4l2_device instance after it gets freed.
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Entities associated with video device nodes are unregistered in
video_unregister_device(). This destroys the entity even though it can
still be accessed through open video device nodes.

Move the media_device_unregister_entity() call from
video_unregister_device() to v4l2_device_release() to ensure that the
entity isn't unregistered until the last reference to the video device
is released.

Also remove the media_entity_get()/put() calls from v4l2-dev.c. Those
functions were designed for subdevs, to avoid a parent module from being
removed while still accessible through board code. They're not currently
needed for video device nodes, and will oops when a hotpluggable device
is disconnected during streaming, as media_entity_put() called in
v4l2_device_release() tries to access entity->parent->dev->driver which
is set to NULL when the device is disconnected.
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: NSakari Ailus <sakari.ailus@iki.fi>
Acked-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

This reverts commit c29fcff3.
Signed-off-by: NBob Liu <lliubbo@gmail.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Subdevs already have their own entity, don't register as second one when
registering the subdev device node.
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: NSakari Ailus <sakari.ailus@maxwell.research.nokia.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Rather than guess which driver supports core priority handling, require drivers
that do to explicitly set the V4L2_FL_USE_FH_PRIO flag in video_device.

Updated the core prio handling accordingly and set the flag in the three
drivers that do.
Signed-off-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The video_device struct has proper ref counting and its release function
will be called when the last user releases it. But no such support was
available for struct v4l2_device. This made it hard to determine when a
USB driver can release the device if it has multiple device nodes.

With one device node it is easy of course, since when the device node is
released, the whole device can be released.

This patch adds refcounting to v4l2_device. When registering device nodes
the v4l2_device refcount will be increased, when releasing device nodes
it will be decreased. The (optional) release function will be called when
the last device node was released.
Signed-off-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Integrate the v4l2_prio_state into the core, ready for use.

One struct v4l2_prio_state is added to v4l2_device and a pointer
to a prio state is added to video_device.
Signed-off-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

We are going to move priority handling into the v4l2 core. As a consequence
the v4l2_prio helper functions need to be moved into the core videodev
module as well to prevent circular dependencies.
Signed-off-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

V4L2 devices are media entities. As such they need to inherit from
(include) the media_entity structure.

When registering/unregistering the device, the media entity is
automatically registered/unregistered. The entity is acquired on device
open and released on device close.
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NSakari Ailus <sakari.ailus@iki.fi>
Acked-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Create a device node named subdevX for every registered subdev.

As the device node is registered before the subdev core::s_config
function is called, return -EGAIN on open until initialization
completes.
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NVimarsh Zutshi <vimarsh.zutshi@gmail.com>
Acked-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Zeroing video_device.dev causes a memory leak if video_set_drvdata
was called before video_register_device was called. video_set_drvdata
calls dev_set_drvdata which allocates video_device.dev.p.

memsetting this will prevent freeing of that memory.
Signed-off-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The unregister function had a race condition with the v4l2_open
function. Ensure that both functions test and clear the REGISTER
flag from within a critical section.

Thanks to Laurent Pinchart for finding this race.
Signed-off-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The BKL replacement mutex had some serious performance side-effects on
V4L drivers. It is replaced by a better heuristic that works around the
worst of the side-effects.

Read the v4l2-dev.c comments for the whole sorry story. This is a
temporary measure only until we can convert all v4l drivers to use
unlocked_ioctl.
Signed-off-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Where reasonable use mutex_lock_interruptible instead of mutex_lock.
Also fix the poll, read and write error codes when called with an unregistered
device (e.g. after a USB device was disconnected). Poll must return
POLLERR|POLLHUP and read/write must return -ENODEV.
Signed-off-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>