platform/kernel/linux-exynos.git
9 years ago[media] tda10071: remove legacy media attach
Antti Palosaari [Mon, 20 Apr 2015 22:54:56 +0000 (19:54 -0300)]
[media] tda10071: remove legacy media attach

All users are now using I2C binding and old attach could be removed.
Use I2C client for proper logging at the same.

Signed-off-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] a8293: coding style issues
Antti Palosaari [Mon, 20 Apr 2015 21:57:03 +0000 (18:57 -0300)]
[media] a8293: coding style issues

Remove FSF address from license. Indent parameter correctly.

Signed-off-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] a8293: improve LNB register programming logic
Antti Palosaari [Mon, 20 Apr 2015 21:47:44 +0000 (18:47 -0300)]
[media] a8293: improve LNB register programming logic

On power-on LNB power supply voltage is disabled, due to that no
need to disable it during probe. Tone is supply is hard-coded as
external tone coming from the demodulator. Program both voltage
and tone on set_voltage(). Use register cache to prevent unneeded
programming.

Signed-off-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] a8293: use i2c_master_send / i2c_master_recv for I2C I/O
Antti Palosaari [Mon, 20 Apr 2015 21:16:19 +0000 (18:16 -0300)]
[media] a8293: use i2c_master_send / i2c_master_recv for I2C I/O

As driver is now proper I2C client driver, we could use correct
functions for I2C I/O. Also rename state from priv to dev. Fix
logging too.

Signed-off-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] a8293: remove legacy media attach
Antti Palosaari [Mon, 20 Apr 2015 20:39:33 +0000 (17:39 -0300)]
[media] a8293: remove legacy media attach

Remove legacy media attach as all users are on I2C bindings now.

Signed-off-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] em28xx: remove unused a8293 SEC config
Antti Palosaari [Fri, 17 Apr 2015 01:55:31 +0000 (22:55 -0300)]
[media] em28xx: remove unused a8293 SEC config

Devices that were using a8293 SEC are converted to I2C platform data
thus that old config structure is left unused.

Signed-off-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] bt819/saa7110/vpx3220: remove legacy control ops
Hans Verkuil [Mon, 20 Jul 2015 13:18:22 +0000 (10:18 -0300)]
[media] bt819/saa7110/vpx3220: remove legacy control ops

The zoran driver has now been converted to the control framework
which means that these three subdevice drivers no longer need to
support the legacy core control ops since the last bridge driver
that needed that has now been converted.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] zoran: convert to the control framework and to v4l2_fh
Hans Verkuil [Mon, 20 Jul 2015 13:18:21 +0000 (10:18 -0300)]
[media] zoran: convert to the control framework and to v4l2_fh

Switch this driver to the control framework and to v4l2_fh for
handling control events.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] zoran: use standard core lock
Hans Verkuil [Mon, 20 Jul 2015 13:18:20 +0000 (10:18 -0300)]
[media] zoran: use standard core lock

Use the standard core lock to take care of serializing ioctl calls and
to serialize file operations.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] zoran: remove unused read/write functions
Hans Verkuil [Mon, 20 Jul 2015 13:18:19 +0000 (10:18 -0300)]
[media] zoran: remove unused read/write functions

The zoran_read/write functions always return an error. Just remove them.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] zoran: remove unnecessary memset
Hans Verkuil [Mon, 20 Jul 2015 13:18:18 +0000 (10:18 -0300)]
[media] zoran: remove unnecessary memset

There is no need to zero the v4l2_capability struct, the v4l2 core has done
that already.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] fsl-viu: drop format names
Hans Verkuil [Mon, 20 Jul 2015 13:09:33 +0000 (10:09 -0300)]
[media] fsl-viu: drop format names

The names of the pixelformats is set by the core. So there no longer
is any need for drivers to fill it in.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] fsl-viu: small fixes
Hans Verkuil [Mon, 20 Jul 2015 13:09:32 +0000 (10:09 -0300)]
[media] fsl-viu: small fixes

- Fix an off-by-one index check in vidioc_enum_fmt()
- Fill in the pix.sizeimage field in vidioc_try_fmt_cap()
- Fix an off-by-one index check in vidioc_s_input()

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] fsl-viu: add control event support
Hans Verkuil [Mon, 20 Jul 2015 13:09:31 +0000 (10:09 -0300)]
[media] fsl-viu: add control event support

Convert the driver to use v4l2_fh in order to support control events.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] fsl-viu: fill in colorspace, always set field to interlaced
Hans Verkuil [Mon, 20 Jul 2015 13:09:30 +0000 (10:09 -0300)]
[media] fsl-viu: fill in colorspace, always set field to interlaced

- fill in the missing colorspace value.
- don't reject incorrect field values, always replace with a valid value.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] fsl-viu: fill in bus_info in vidioc_querycap
Hans Verkuil [Mon, 20 Jul 2015 13:09:29 +0000 (10:09 -0300)]
[media] fsl-viu: fill in bus_info in vidioc_querycap

The bus_info field was never filled.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] fsl-viu: convert to the control framework
Hans Verkuil [Mon, 20 Jul 2015 13:09:28 +0000 (10:09 -0300)]
[media] fsl-viu: convert to the control framework

Interestingly enough, the existing control handling code basically did
nothing. At least the new code will inherit the controls from the
saa7115 driver.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] usbvision: move init code to probe()
Hans Verkuil [Mon, 20 Jul 2015 12:59:38 +0000 (09:59 -0300)]
[media] usbvision: move init code to probe()

These things are only initialized if you start streaming video, but
they are also used in the disconnect function. So just init them
always during probe time.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] usbvision: fix standards for S-Video/Composite inputs
Hans Verkuil [Mon, 20 Jul 2015 12:59:37 +0000 (09:59 -0300)]
[media] usbvision: fix standards for S-Video/Composite inputs

The standards supported by S-Video and Composite inputs are not
limited by PAL, so make it more generic.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] usbvision: fix DMA from stack warnings
Hans Verkuil [Mon, 20 Jul 2015 12:59:36 +0000 (09:59 -0300)]
[media] usbvision: fix DMA from stack warnings

In various places the stack was used to provide buffers for USB data, but
this should be allocated memory.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] usbvision: fix locking error
Hans Verkuil [Mon, 20 Jul 2015 12:59:35 +0000 (09:59 -0300)]
[media] usbvision: fix locking error

If remove_pending is non-zero, then the v4l2_lock is never unlocked.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] usbvision: set field and colorspace
Hans Verkuil [Mon, 20 Jul 2015 12:59:34 +0000 (09:59 -0300)]
[media] usbvision: set field and colorspace

Set the colorspace and field in vidioc_try_fmt_vid_cap().

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] usbvision: frequency fixes
Hans Verkuil [Mon, 20 Jul 2015 12:59:33 +0000 (09:59 -0300)]
[media] usbvision: frequency fixes

- setup initial radio and tv frequencies.
- set/get the correct frequency (radio vs tv).
- disable tuner/freq ioctls if there is no tuner.
- fix some tuner index checks.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] usbvision: the radio device node has wrong caps
Hans Verkuil [Mon, 20 Jul 2015 12:59:32 +0000 (09:59 -0300)]
[media] usbvision: the radio device node has wrong caps

The radio device node had the same caps as the video node. Fix this.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] usbvision: remove g/s_audio and for radio remove enum/g/s_input
Hans Verkuil [Mon, 20 Jul 2015 12:59:31 +0000 (09:59 -0300)]
[media] usbvision: remove g/s_audio and for radio remove enum/g/s_input

The g/s_audio ioctls didn't do anything, so remove them all for both
video and radio nodes and remove V4L2_CAP_AUDIO.

The enum/g/s_input ioctls are invalid for radio nodes, so remove them
from the radio ioctl_ops.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] usbvision: return valid error in usbvision_register_video()
Hans Verkuil [Mon, 20 Jul 2015 12:59:30 +0000 (09:59 -0300)]
[media] usbvision: return valid error in usbvision_register_video()

Don't return -1, return a proper error code.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] usbvision: convert to the control framework
Hans Verkuil [Mon, 20 Jul 2015 12:59:29 +0000 (09:59 -0300)]
[media] usbvision: convert to the control framework

Convert this driver to the control framework and struct v4l2_fh
(needed for handling control events).

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] usbvision: remove power_on_at_open and timed power off
Hans Verkuil [Mon, 20 Jul 2015 12:59:28 +0000 (09:59 -0300)]
[media] usbvision: remove power_on_at_open and timed power off

This causes lots of problems and is *very* slow as well.

One of the main problems is that this prohibits the use of the control
framework since subdevs will be unloaded on power off which is not allowed
as long as they are used by a usb device.

Apparently the reason for doing this is to turn off a noisy tuner. My hardware
has no problem with that, and I wonder whether the hardware with that noisy
tuner wasn't just functioning improperly as I have never heard of noisy tuners.

Contact me if you have one of those devices and I can take a look whether the
tuner can't be powered off if necessary by letting the tuner subdevice go
into standby mode. Unloading the tuner module is just evil and is not the
right approach.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] DocBook: Fix typo in intro.xml
Masanari Iida [Mon, 13 Jul 2015 23:36:50 +0000 (20:36 -0300)]
[media] DocBook: Fix typo in intro.xml

This patch fix spelling typos in intro.xml.
This xml file is not created from comments within source,
I fix the xml file.

Signed-off-by: Masanari Iida <standby24x7@gmail.com>
[hans.verkuil@cisco.com: removed mention of obsolete devfs]
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] MAINTAINERS: V4L2: PLATFORM: Add entry for Renesas JPEG Processing Unit driver
Mikhail Ulyanov [Wed, 22 Jul 2015 11:23:05 +0000 (08:23 -0300)]
[media] MAINTAINERS: V4L2: PLATFORM: Add entry for Renesas JPEG Processing Unit driver

Update RENESAS JPU driver maintainer in MAINTAINERS file.

Signed-off-by: Mikhail Ulyanov <mikhail.ulyanov@cogentembedded.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] V4L2: platform: Add Renesas R-Car JPEG codec driver
Mikhail Ulyanov [Wed, 22 Jul 2015 11:23:03 +0000 (08:23 -0300)]
[media] V4L2: platform: Add Renesas R-Car JPEG codec driver

Here's the driver for the Renesas R-Car JPEG processing unit.

The driver is implemented within the V4L2 framework as a memory-to-memory
device.  It presents two video nodes to userspace, one for the encoding part,
and one for the decoding part.

It was found that the only working mode for encoding is no markers output, so we
generate markers with software. In the current version of driver we also use
software JPEG header parsing because with hardware parsing performance is lower
than desired.

>From a userspace point of view the process is typical (S_FMT, REQBUF,
optionally QUERYBUF, QBUF, STREAMON, DQBUF) for both the source and destination
queues. STREAMON can return -EINVAL in case of mismatch of output and capture
queues format. Also during decoding driver can return buffers if queued
buffer with JPEG image contains image with inappropriate subsampling (e.g.
4:2:0 in JPEG and 4:2:2 in capture).  If JPEG image and queue format dimensions
differ driver will return buffer on QBUF with VB2_BUF_STATE_ERROR flag.

During encoding the available formats are: V4L2_PIX_FMT_NV12M,
V4L2_PIX_FMT_NV12, V4L2_PIX_FMT_NV16, V4L2_PIX_FMT_NV16M for source and
V4L2_PIX_FMT_JPEG for destination.

During decoding the available formats are: V4L2_PIX_FMT_JPEG for source and
V4L2_PIX_FMT_NV12M, V4L2_PIX_FMT_NV16M, V4L2_PIX_FMT_NV12, V4L2_PIX_FMT_NV16
for destination.

Performance of current version:
1280x800 NV12 image encoding/decoding
decoding ~122 FPS
encoding ~191 FPS

Signed-off-by: Mikhail Ulyanov <mikhail.ulyanov@cogentembedded.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] devicetree: bindings: Document Renesas R-Car JPEG Processing Unit
Mikhail Ulyanov [Wed, 22 Jul 2015 11:23:04 +0000 (08:23 -0300)]
[media] devicetree: bindings: Document Renesas R-Car JPEG Processing Unit

Add Renesas R-Car JPEG processing unit driver device tree bindings
documentation.

Signed-off-by: Mikhail Ulyanov <mikhail.ulyanov@cogentembedded.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] v4l2-ioctl: Give more information when device_caps are missing
Laura Abbott [Thu, 23 Jul 2015 22:28:48 +0000 (19:28 -0300)]
[media] v4l2-ioctl: Give more information when device_caps are missing

Currently, the warning for missing device_caps gives a backtrace like so:

[<ffffffff8175c199>] dump_stack+0x45/0x57
[<ffffffff8109ad5a>] warn_slowpath_common+0x8a/0xc0
[<ffffffff8109ae8a>] warn_slowpath_null+0x1a/0x20
[<ffffffffa0237453>] v4l_querycap+0x43/0x80 [videodev]
[<ffffffffa0237734>] __video_do_ioctl+0x2a4/0x320 [videodev]
[<ffffffff812207e5>] ? do_last+0x195/0x1210
[<ffffffffa023a11e>] video_usercopy+0x22e/0x5b0 [videodev]
[<ffffffffa0237490>] ? v4l_querycap+0x80/0x80 [videodev]
[<ffffffffa023a4b5>] video_ioctl2+0x15/0x20 [videodev]
[<ffffffffa0233733>] v4l2_ioctl+0x113/0x150 [videodev]
[<ffffffff81225798>] do_vfs_ioctl+0x2f8/0x4f0
[<ffffffff8113b2d4>] ? __audit_syscall_entry+0xb4/0x110
[<ffffffff81022d7c>] ? do_audit_syscall_entry+0x6c/0x70
[<ffffffff81225a11>] SyS_ioctl+0x81/0xa0
[<ffffffff8113b526>] ? __audit_syscall_exit+0x1f6/0x2a0
[<ffffffff81763549>] system_call_fastpath+0x12/0x17

This indicates that device_caps are missing but doesn't give
much of a clue which driver is actually at fault. Improve
the warning output by showing the capabilities and the
responsible driver.

Signed-off-by: Laura Abbott <labbott@fedoraproject.org>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] tc358743: remove unused variable
Hans Verkuil [Mon, 20 Jul 2015 09:47:23 +0000 (06:47 -0300)]
[media] tc358743: remove unused variable

The bt pointer was never used, remove it.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] v4l2-mem2mem: drop lock in v4l2_m2m_fop_mmap
Hans Verkuil [Mon, 20 Jul 2015 07:58:24 +0000 (04:58 -0300)]
[media] v4l2-mem2mem: drop lock in v4l2_m2m_fop_mmap

The v4l2_m2m_fop_mmap function takes the core mutex, but this will result in a potential
circular locking dependency:

[  262.517164] ======================================================
[  262.517166] [ INFO: possible circular locking dependency detected ]
[  262.517169] 4.2.0-rc2-koryphon #844 Not tainted
[  262.517171] -------------------------------------------------------
[  262.517173] v4l2-compliance/1379 is trying to acquire lock:
[  262.517175]  (&dev->dev_mutex){+.+.+.}, at: [<ffffffffa000ddab>] v4l2_m2m_fop_mmap+0x2b/0x90 [v4l2_mem2mem]
[  262.517187]
               but task is already holding lock:
[  262.517189]  (&mm->mmap_sem){++++++}, at: [<ffffffff81159309>] vm_mmap_pgoff+0x69/0xc0
[  262.517199]
               which lock already depends on the new lock.

[  262.517202]
               the existing dependency chain (in reverse order) is:
[  262.517204]
               -> #1 (&mm->mmap_sem){++++++}:
[  262.517209]        [<ffffffff810d0e6b>] __lock_acquire+0x62b/0xe80
[  262.517215]        [<ffffffff810d2095>] lock_acquire+0x65/0x90
[  262.517218]        [<ffffffff811612e5>] __might_fault+0x75/0xa0
[  262.517222]        [<ffffffffa06dead9>] video_usercopy+0x3e9/0x4e0 [videodev]
[  262.517231]        [<ffffffffa06debe0>] video_ioctl2+0x10/0x20 [videodev]
[  262.517238]        [<ffffffffa06d8663>] v4l2_ioctl+0xc3/0xe0 [videodev]
[  262.517243]        [<ffffffff811a8cac>] do_vfs_ioctl+0x2fc/0x550
[  262.517248]        [<ffffffff811a8f74>] SyS_ioctl+0x74/0x80
[  262.517252]        [<ffffffff81a4d2ee>] entry_SYSCALL_64_fastpath+0x12/0x76
[  262.517258]
               -> #0 (&dev->dev_mutex){+.+.+.}:
[  262.517262]        [<ffffffff810cf464>] validate_chain.isra.38+0xd04/0x1170
[  262.517266]        [<ffffffff810d0e6b>] __lock_acquire+0x62b/0xe80
[  262.517270]        [<ffffffff810d2095>] lock_acquire+0x65/0x90
[  262.517273]        [<ffffffff81a48e3c>] mutex_lock_interruptible_nested+0x6c/0x4b0
[  262.517279]        [<ffffffffa000ddab>] v4l2_m2m_fop_mmap+0x2b/0x90 [v4l2_mem2mem]
[  262.517284]        [<ffffffffa06d80ff>] v4l2_mmap+0x4f/0x90 [videodev]
[  262.517288]        [<ffffffff8116b06c>] mmap_region+0x38c/0x5b0
[  262.517293]        [<ffffffff8116b585>] do_mmap_pgoff+0x2f5/0x3e0
[  262.517297]        [<ffffffff8115932a>] vm_mmap_pgoff+0x8a/0xc0
[  262.517300]        [<ffffffff81169bab>] SyS_mmap_pgoff+0x1cb/0x270
[  262.517304]        [<ffffffff8100876d>] SyS_mmap+0x1d/0x20
[  262.517309]        [<ffffffff81a4d2ee>] entry_SYSCALL_64_fastpath+0x12/0x76
[  262.517313]
               other info that might help us debug this:

[  262.517315]  Possible unsafe locking scenario:

[  262.517318]        CPU0                    CPU1
[  262.517319]        ----                    ----
[  262.517321]   lock(&mm->mmap_sem);
[  262.517324]                                lock(&dev->dev_mutex);
[  262.517327]                                lock(&mm->mmap_sem);
[  262.517329]   lock(&dev->dev_mutex);
[  262.517332]
                *** DEADLOCK ***

Since vb2_fop_mmap doesn't take the lock, neither should v4l2_m2m_fop_mmap.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Tested-by: Mikhail Ulyanov <mikhail.ulyanov@cogentembedded.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] s5p-jpeg: Eliminate double kfree()
Andrzej Pietrasiewicz [Fri, 3 Jul 2015 10:04:38 +0000 (07:04 -0300)]
[media] s5p-jpeg: Eliminate double kfree()

video_unregister_device() calls device_unregister(), which calls
put_device(), which calls kobject_put(), and if this is the last reference
then kobject_release() is called, which calls kobject_cleanup(), which
calls ktype's release method which happens to be device_release() in this
case, which calls dev->release(), which happens to be
v4l2_device_release() in this case, which calls vdev->release(), which
happens to be video_device_release(). But video_device_release() is
called explicitly both in error recovery path of s5p_jpeg_probe() and
in s5p_jpeg_remove(). The pointers in question are not nullified between
the two calls, so this is harmful.

This patch fixes the driver so that video_device_release() is not called
twice for the same object.

Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] s5p-mfc: add additional check for incorrect memory configuration
Marek Szyprowski [Wed, 3 Jun 2015 10:36:23 +0000 (07:36 -0300)]
[media] s5p-mfc: add additional check for incorrect memory configuration

MFC hardware is known to trash random memory if one tries to use a
buffer which has lower DMA addresses than the configured DMA base
address. This patch adds a check for this case and proper error
handling.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Acked-by: Kamil Debski <kamil@wypas.org>
[s.nawrocki@samsung.com: fixed typo (addres -> address]
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] s5p-mfc: add return value check in mfc_sys_init_cmd
Marek Szyprowski [Wed, 3 Jun 2015 10:36:22 +0000 (07:36 -0300)]
[media] s5p-mfc: add return value check in mfc_sys_init_cmd

alloc_dev_context_buffer method might fail, so add proper
return value check.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Acked-by: Kamil Debski <kamil@wypas.org>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] s5p-mfc: fix state check from encoder queue_setup
Seung-Woo Kim [Wed, 13 May 2015 07:25:25 +0000 (04:25 -0300)]
[media] s5p-mfc: fix state check from encoder queue_setup

MFCINST_GOT_INST state is set to encoder context with set_format
only for capture buffer. In queue_setup of encoder called during
reqbufs, it is checked MFCINST_GOT_INST state for both capture
and output buffer. So this patch fixes encoder to check
MFCINST_GOT_INST state only for capture buffer from queue_setup.

Signed-off-by: Seung-Woo Kim <sw0312.kim@samsung.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] s5p-tv: fix wait_event_timeout return handling
Nicholas Mc Guire [Wed, 10 Jun 2015 09:02:03 +0000 (06:02 -0300)]
[media] s5p-tv: fix wait_event_timeout return handling

event API conformance testing with coccinelle spatches are being
used to locate API usage inconsistencies this triggert with:
./drivers/media/platform/s5p-tv/mixer_reg.c:364
        incorrect check for negative return

Return type of wait_event_timeout is signed long not int and the
return type is >=0 always thus the negative check is unnecessary.
An appropriately named variable of type long is inserted and the
call fixed up aswell as the negative return check dropped.

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] s5p-tv: Drop owner assignment from i2c_driver
Krzysztof Kozlowski [Fri, 10 Jul 2015 06:19:45 +0000 (03:19 -0300)]
[media] s5p-tv: Drop owner assignment from i2c_driver

i2c_driver does not need to set an owner because i2c_register_driver()
will set it.

Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] dvb_core: Replace memset with eth_zero_addr
Vaishali Thakkar [Sat, 20 Jun 2015 02:58:17 +0000 (23:58 -0300)]
[media] dvb_core: Replace memset with eth_zero_addr

Use eth_zero_addr to assign the zero address to the given address
array instead of memset when second argument is address of zero.

The Coccinelle semantic patch that makes this change is as follows:

// <smpl>
@eth_zero_addr@
expression e;
@@

-memset(e,0x00,ETH_ALEN);
+eth_zero_addr(e);
// </smpl>

Signed-off-by: Vaishali Thakkar <vthakkar1994@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] staging: media: lirc: fix coding style error
Sunil Shahu [Sat, 20 Jun 2015 08:53:50 +0000 (05:53 -0300)]
[media] staging: media: lirc: fix coding style error

Fix code indentation error by replacing tab in place of spaces.

Signed-off-by: Sunil Shahu <shshahu@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] sh-veu: don't use COLORSPACE_JPEG
Hans Verkuil [Mon, 15 Jun 2015 11:33:29 +0000 (08:33 -0300)]
[media] sh-veu: don't use COLORSPACE_JPEG

COLORSPACE_JPEG should only be used for JPEGs. Use SMPTE170M instead,
which is how YCbCr images are usually encoded.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] media: ttpci: Use vsprintf %pM extension
Joe Perches [Mon, 15 Jun 2015 02:01:45 +0000 (23:01 -0300)]
[media] media: ttpci: Use vsprintf %pM extension

Format mac addresses with the normal kernel extension.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] tc358743: allow event subscription
Philipp Zabel [Fri, 17 Jul 2015 15:26:45 +0000 (12:26 -0300)]
[media] tc358743: allow event subscription

This is useful to subscribe to HDMI hotplug events via the
V4L2_CID_DV_RX_POWER_PRESENT control.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] tc358743: add direct interrupt handling
Philipp Zabel [Fri, 17 Jul 2015 14:02:56 +0000 (11:02 -0300)]
[media] tc358743: add direct interrupt handling

When probed from device tree, the i2c client driver can handle the interrupt
on its own.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] tc358743: support probe from device tree
Philipp Zabel [Fri, 17 Jul 2015 14:02:55 +0000 (11:02 -0300)]
[media] tc358743: support probe from device tree

Add support for probing the TC358743 subdevice from device tree.
The reference clock must be supplied using the common clock bindings.
MIPI CSI-2 specific properties are parsed from the OF graph endpoint
node and support for a non-continuous MIPI CSI-2 clock is added.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] tc358743: enable v4l2 subdevice devnode
Philipp Zabel [Fri, 17 Jul 2015 14:02:54 +0000 (11:02 -0300)]
[media] tc358743: enable v4l2 subdevice devnode

Add V4L2_SUBDEV_FL_HAS_DEVNODE to subdev flags, in order to
enable a subdev device node.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] tc358743: register v4l2 asynchronous subdevice
Philipp Zabel [Fri, 17 Jul 2015 14:02:53 +0000 (11:02 -0300)]
[media] tc358743: register v4l2 asynchronous subdevice

Add support for registering the sensor subdevice using the v4l2-async API.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] cobalt: allow fewer than 8 PCIe lanes
Hans Verkuil [Fri, 17 Jul 2015 11:45:22 +0000 (08:45 -0300)]
[media] cobalt: allow fewer than 8 PCIe lanes

Currently the cobalt driver refuses to load if fewer than 8 PCIe lanes
are assigned. This patch changes this and just issues a warning. The
only time it will refuse to load is if the number of assigned lanes is less
than what the PCIe host is capable of since this suggests that the card
isn't seated correctly in the slot.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] cobalt: accept unchanged timings when vb2_is_busy()
Hans Verkuil [Fri, 17 Jul 2015 11:29:40 +0000 (08:29 -0300)]
[media] cobalt: accept unchanged timings when vb2_is_busy()

When vb2_is_busy() it should still be possible to call S_DV_TIMINGS
provided the new timings are the same as the current timings.

For input 1 (test generator) the size is always 1080p, so just return
that.

Fixes a v4l2-compliance issue.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] media: am437x-vpfe: Fix a race condition during release
Benoit Parrot [Wed, 15 Jul 2015 21:00:06 +0000 (18:00 -0300)]
[media] media: am437x-vpfe: Fix a race condition during release

There was a race condition where during cleanup/release operation
on-going streaming would cause a kernel panic because the hardware
module was disabled prematurely with IRQ still pending.

Fixes: 417d2e507edc ("[media] media: platform: add VPFE capture driver support for AM437X")

Cc: <stable@vger.kernel.org> # v4.0+
Signed-off-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] videobuf2: add trace events
Philipp Zabel [Fri, 10 Jul 2015 13:49:26 +0000 (10:49 -0300)]
[media] videobuf2: add trace events

Add videobuf2 specific vb2_qbuf and vb2_dqbuf trace events that mirror the
v4l2_qbuf and v4l2_dqbuf trace events, only they include additional
information about queue fill state and are emitted right before the buffer
is enqueued in the driver or userspace is woken up. This allows to make
sense of the timeline of trace events in combination with others that might
be triggered by __enqueue_in_driver.

Also two new trace events vb2_buf_queue and vb2_buf_done are added,
allowing to trace the handover between videobuf2 framework and driver.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] v4l2-mem2mem: set the queue owner field just as vb2_ioctl_reqbufs does
Philipp Zabel [Fri, 10 Jul 2015 13:49:25 +0000 (10:49 -0300)]
[media] v4l2-mem2mem: set the queue owner field just as vb2_ioctl_reqbufs does

The queue owner will be used by videobuf2 trace events to determine and
record the device minor number. It is set in v4l2_m2m_reqbufs instead of
v4l2_m2m_ioctl_reqbufs because several drivers implement their own
vidioc_reqbufs handlers that still call v4l2_m2m_reqbufs directly.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] v4l2-dev: use event class to deduplicate v4l2 trace events
Philipp Zabel [Fri, 10 Jul 2015 13:49:24 +0000 (10:49 -0300)]
[media] v4l2-dev: use event class to deduplicate v4l2 trace events

Trace events with exactly the same parameters and trace output, such as
v4l2_qbuf and v4l2_dqbuf, are supposed to use the DECLARE_EVENT_CLASS and
DEFINE_EVENT macros instead of duplicated TRACE_EVENT macro calls.

Suggested-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] coda: make NV12 format default
Philipp Zabel [Thu, 16 Jul 2015 16:19:39 +0000 (13:19 -0300)]
[media] coda: make NV12 format default

The chroma interleaved NV12 format has higher memory bandwidth efficiency
because the chroma planes can be read/written with longer burst lengths.
Use NV12 as default format if available and consistently sort it first.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] coda: add macroblock tiling support
Philipp Zabel [Thu, 16 Jul 2015 16:19:38 +0000 (13:19 -0300)]
[media] coda: add macroblock tiling support

Storing internal frames in macroblock tiled order improves memory
access patterns by allowing increased burst sizes when transferring
the uncompressed macroblocks to or from main memory.
The translation logic only supports a single chroma base address,
so this is only supported for the chroma interleaved NV12 format.

Since the rotator used to copy the decoder output into the v4l2
capture buffers does not seem to support the tiled format correctly,
only enable it in the encoder for now.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] coda: move cache setup into coda9_set_frame_cache, also use it in start_encoding
Philipp Zabel [Thu, 16 Jul 2015 16:19:37 +0000 (13:19 -0300)]
[media] coda: move cache setup into coda9_set_frame_cache, also use it in start_encoding

The frame cache should be set up correctly to encode NV12 source frames.
This was not done before, so move the cache setup out of start_decoding
into its own function and call it from both start_encoding and
start_decoding.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] coda: Use S_PARM to set nominal framerate for h.264 encoder
Philipp Zabel [Thu, 16 Jul 2015 16:13:24 +0000 (13:13 -0300)]
[media] coda: Use S_PARM to set nominal framerate for h.264 encoder

The encoder needs to know the nominal framerate for the constant bitrate
control mechanism to work. Currently the only way to set the framerate is
by using VIDIOC_S_PARM on the output queue.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] coda: implement VBV delay and buffer size controls
Philipp Zabel [Fri, 10 Jul 2015 13:37:52 +0000 (10:37 -0300)]
[media] coda: implement VBV delay and buffer size controls

The encoder allows to specify the VBV model reference decoder's initial
delay and buffer size. Export the corresponding V4L2 controls.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] coda: reset CODA960 hardware after sequence end
Philipp Zabel [Fri, 10 Jul 2015 13:37:44 +0000 (10:37 -0300)]
[media] coda: reset CODA960 hardware after sequence end

On i.MX6, sometimes after decoding a stream, encoding will produce macroblock
errors caused by missing 8-byte sequences in the output stream. Until the cause
for this is found, reset the hardware after sequence end, which seems to help.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] coda: rework meta counting and add separate lock
Philipp Zabel [Thu, 9 Jul 2015 10:10:21 +0000 (07:10 -0300)]
[media] coda: rework meta counting and add separate lock

Keep count of number of buffer meta structures in the list and use
a separate spinlock for operations on this counted list instead
of reusing the bitstream mutex in some places and none at all in
others.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] coda: reuse src_bufs in coda_job_ready
Philipp Zabel [Thu, 9 Jul 2015 10:10:20 +0000 (07:10 -0300)]
[media] coda: reuse src_bufs in coda_job_ready

The v4l2_m2m_num_src_bufs_ready() function is called in multiple places
in coda_cob_ready, and there already is a variable src_bufs that is
assigned to its result. Move it to the beginning and use it everywhere.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] coda: use event class to deduplicate v4l2 trace events
Philipp Zabel [Thu, 9 Jul 2015 10:10:19 +0000 (07:10 -0300)]
[media] coda: use event class to deduplicate v4l2 trace events

Trace events with exactly the same parameters and trace output, such as
coda_enc_pic_run and coda_enc_pic_done, are supposed to use the
DECLARE_EVENT_CLASS and DEFINE_EVENT macros instead of duplicated
TRACE_EVENT macro calls.
This patch changes the order of parameters to coda_dec_rot_done and adds
a timestamp so it can share an event class with coda_bit_queue.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] coda: drop custom list of pixel format descriptions
Philipp Zabel [Thu, 9 Jul 2015 10:10:18 +0000 (07:10 -0300)]
[media] coda: drop custom list of pixel format descriptions

Since commit ba3002045f80 ("[media] v4l2-ioctl: fill in the description
for VIDIOC_ENUM_FMT"), all pixel formats are assigned their description
in a central place. We can now drop the custom list.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] coda: reset stream end in stop_streaming
Philipp Zabel [Thu, 9 Jul 2015 10:10:17 +0000 (07:10 -0300)]
[media] coda: reset stream end in stop_streaming

Otherwise a restarted stream won't queue buffers.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] coda: avoid calling SEQ_END twice
Philipp Zabel [Thu, 9 Jul 2015 10:10:16 +0000 (07:10 -0300)]
[media] coda: avoid calling SEQ_END twice

Allow coda_seq_end_work to be called multiple times, move the setting
of ctx->initialized from coda_start/stop_streaming() into
coda_start_encoding/decoding and coda_seq_end_work, respectively,
and skip the SEQ_END command in coda_seq_end_work if the context is
already deinitialized before.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] coda: keep buffers on the queue in bitstream end mode
Philipp Zabel [Thu, 9 Jul 2015 10:10:15 +0000 (07:10 -0300)]
[media] coda: keep buffers on the queue in bitstream end mode

In stream end mode the hardware will read the bitstream to its end,
overshooting the write pointer. Do not write additional data into
the bitstream in this mode.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] coda: fix bitstream preloading for MPEG4 decoding
Philipp Zabel [Thu, 9 Jul 2015 10:10:14 +0000 (07:10 -0300)]
[media] coda: fix bitstream preloading for MPEG4 decoding

All decoder instances using the BIT processor should preload buffers
into the bitstream ring buffer, including MPEG4 decoding. Fix this
by explicitly stating the above condition instead of listing all
relevant input formats.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] coda: fix mvcol buffer for MPEG4 decoding
Philipp Zabel [Thu, 9 Jul 2015 10:10:13 +0000 (07:10 -0300)]
[media] coda: fix mvcol buffer for MPEG4 decoding

The mvcol buffer is allocated at the end of the first internal buffer.
This patch fixes an out of bounds array access.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] coda: clamp frame sequence counters to 16 bit
Lucas Stach [Thu, 9 Jul 2015 10:10:12 +0000 (07:10 -0300)]
[media] coda: clamp frame sequence counters to 16 bit

This is already done for one side of the comparison with the expectation
that the HW counter rolls over at the 16 bit boundary. This is true when
decoding a h.264 stream, but doesn't hold for at least MJPEG. As we don't
know the exact wrap-around point for this format just clamp the HW counter
to the same 16 bits. This should be enough to detect most of the errors
and saves us from doing different comparisons based on the decoded format.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] v4l: vsp1: Don't sleep in atomic context
Laurent Pinchart [Mon, 13 Apr 2015 14:43:40 +0000 (11:43 -0300)]
[media] v4l: vsp1: Don't sleep in atomic context

The vsp1_entity_is_streaming() function is called in atomic context when
queuing buffers, and sleeps due to a mutex. As the mutex just protects
access to one structure field, fix this by replace the mutex with a
spinlock.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] v4l: vsp1: Fix plane stride and size checks
Laurent Pinchart [Fri, 19 Jun 2015 11:51:22 +0000 (08:51 -0300)]
[media] v4l: vsp1: Fix plane stride and size checks

The checks need to be performed on up to two planes, as the third plane,
if present, must have the same stride and size as the second plane.

The code incorrectly performs the checks on at least two planes instead
of at most two planes, fix it.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] v4l: xilinx: missing error code
Dan Carpenter [Wed, 24 Jun 2015 14:28:31 +0000 (11:28 -0300)]
[media] v4l: xilinx: missing error code

We should set "ret" on this error path instead of returning success.

Fixes: df3305156f98 ('[media] v4l: xilinx: Add Xilinx Video IP core')

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Hyun Kwon <hyun.kwon@xilinx.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] Driver for Toshiba TC358743 HDMI to CSI-2 bridge
Mats Randgaard [Thu, 9 Jul 2015 08:45:47 +0000 (05:45 -0300)]
[media] Driver for Toshiba TC358743 HDMI to CSI-2 bridge

The driver is tested on our hardware and all the implemented features
works as expected.

Missing features:
- CEC support
- HDCP repeater support
- IR support

Signed-off-by: Mats Randgaard <matrandg@cisco.com>
[hans.verkuil@cisco.com: updated copyright year to 2015]
[hans.verkuil@cisco.com: update confusing confctl_mutex comment]
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] bdisp: fix debug info memory access
Fabien Dessenne [Mon, 13 Jul 2015 09:54:11 +0000 (06:54 -0300)]
[media] bdisp: fix debug info memory access

bdisp_dev->dbg.copy_node shall be a copy of (and not point to)
bdisp_ctx->node, since this resource is freed upon driver release.

Signed-off-by: Fabien Dessenne <fabien.dessenne@st.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] bdisp: add debug info for RGB24 format
Fabien Dessenne [Fri, 10 Jul 2015 08:29:37 +0000 (05:29 -0300)]
[media] bdisp: add debug info for RGB24 format

Add this missing debug information

Signed-off-by: Fabien Dessenne <fabien.dessenne@st.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] bdisp: composing support
Fabien Dessenne [Fri, 10 Jul 2015 08:29:22 +0000 (05:29 -0300)]
[media] bdisp: composing support

Support the composing (at VIDEO_CAPTURE) with the _selection API.
v4l2-compliance successfully run ("test Composing: OK")

Signed-off-by: Fabien Dessenne <fabien.dessenne@st.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] stk1160: Add frame scaling support
Ezequiel Garcia [Fri, 3 Jul 2015 19:11:42 +0000 (16:11 -0300)]
[media] stk1160: Add frame scaling support

This commit implements frame decimation for stk1160, which allows
to support format changes instead of a static frame size.

The stk1160 supports independent row and column decimation, in two
different modes:
 * set a number of rows/columns units to skip for each unit sent.
 * set a number of rows/columns units to send for each unit skipped.

This effectively allows to achieve different frame scaling ratios.

The unit number can be set to either two row/columns sent/skipped,
or four row/columns sent/skipped. Since the video format (UYVY)
has 4-bytes, using a unit number of two row/columns results
in frame color 'shifting', so set to four row/columns sent/skipped.

Signed-off-by: Michael Stegemann <michael@stegemann.it>
Signed-off-by: Dale Hamel <dale.hamel@srvthe.net>
Signed-off-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] stk1160: Reduce driver verbosity
Ezequiel Garcia [Fri, 3 Jul 2015 19:11:41 +0000 (16:11 -0300)]
[media] stk1160: Reduce driver verbosity

These messages are not really informational, and just makes the driver's
output too verbose. This commit changes some messages to a debug level,
removes a really useless "driver loaded" message and finally undefines
the DEBUG macro.

Signed-off-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] media: am437x-vpfe: Requested frame size and fmt overwritten by current senso...
Benoit Parrot [Mon, 29 Jun 2015 21:19:06 +0000 (18:19 -0300)]
[media] media: am437x-vpfe: Requested frame size and fmt overwritten by current sensor setting

Upon a S_FMT the input/requested frame size and pixel format is
overwritten by the current sub-device settings.
Fix this so application can actually set the frame size and format.

Fixes: 417d2e507edc ("[media] media: platform: add VPFE capture driver support for AM437X")

Cc: <stable@vger.kernel.org> # v4.0+
Signed-off-by: Benoit Parrot <bparrot@ti.com>
Acked-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] adv7604/cobalt: Allow compile test if !GPIOLIB
Geert Uytterhoeven [Mon, 29 Jun 2015 13:45:56 +0000 (10:45 -0300)]
[media] adv7604/cobalt: Allow compile test if !GPIOLIB

The GPIO subsystem provides dummy GPIO consumer functions if GPIOLIB is
not enabled. Hence drivers that depend on GPIOLIB, but use GPIO consumer
functionality only, can still be compiled if GPIOLIB is not enabled.

Relax the dependency of VIDEO_ADV7604 and VIDEO_COBALT (the latter
selects the former) on GPIOLIB if COMPILE_TEST is enabled.

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] DocBook media: fix typo in V4L2_CTRL_FLAG_EXECUTE_ON_WRITE
Hans Verkuil [Wed, 8 Jul 2015 08:47:08 +0000 (05:47 -0300)]
[media] DocBook media: fix typo in V4L2_CTRL_FLAG_EXECUTE_ON_WRITE

Fix small typo (missing 'it') in the documentation for
V4L2_CTRL_FLAG_EXECUTE_ON_WRITE.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Ricardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] DocBook: fix media-ioc-device-info.xml type
Hans Verkuil [Thu, 2 Jul 2015 13:32:39 +0000 (10:32 -0300)]
[media] DocBook: fix media-ioc-device-info.xml type

The documentation had two media_version entries. The second one was a typo and
it should be driver_version instead. Correct this.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] adv7842: Deliver resolution change events to userspace
Lars-Peter Clausen [Wed, 24 Jun 2015 16:50:31 +0000 (13:50 -0300)]
[media] adv7842: Deliver resolution change events to userspace

Use the new v4l2_subdev_notify_event() helper function to deliver the
resolution change event to userspace via the v4l2 subdev event queue as
well as to the bridge driver using the callback notify mechanism.

This allows userspace applications to react to changes in resolution. This
is useful and often necessary for video pipelines where there is no direct
1-to-1 relationship between the subdevice converter and the video capture
device and hence it does not make sense to directly forward the event to
the video capture device node.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
[hans.verkuil@cisco.com: fix obvious mistake: v4l2_event_subdev_unsubscribe -> v4l2_ctrl_subdev_subscribe_event]
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] adv7604: Deliver resolution change events to userspace
Lars-Peter Clausen [Wed, 24 Jun 2015 16:50:30 +0000 (13:50 -0300)]
[media] adv7604: Deliver resolution change events to userspace

Use the new v4l2_subdev_notify_event() helper function to deliver the
resolution change event to userspace via the v4l2 subdev event queue as
well as to the bridge driver using the callback notify mechanism.

This allows userspace applications to react to changes in resolution. This
is useful and often necessary for video pipelines where there is no direct
1-to-1 relationship between the subdevice converter and the video capture
device and hence it does not make sense to directly forward the event to
the video capture device node.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
[hans.verkuil@cisco.com: fix obvious mistake: v4l2_event_subdev_unsubscribe -> v4l2_ctrl_subdev_subscribe_event]
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] Add helper function for subdev event notifications
Lars-Peter Clausen [Wed, 24 Jun 2015 16:50:29 +0000 (13:50 -0300)]
[media] Add helper function for subdev event notifications

Add a new helper function called v4l2_subdev_notify_event() which will
deliver the specified event to both the v4l2 subdev event queue as well as
to the notify callback. The former is typically used by userspace
applications to listen to notification events while the later is used by
bridge drivers. Combining both into the same function avoids boilerplate
code in subdev drivers.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] adv7842: Add support for control event notifications
Lars-Peter Clausen [Wed, 24 Jun 2015 16:50:28 +0000 (13:50 -0300)]
[media] adv7842: Add support for control event notifications

Allow userspace applications to subscribe to control change events. This
can e.g. be used to monitor the 5V detect control to be notified when a
source is connected or disconnected.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] adv7604: Add support for control event notifications
Lars-Peter Clausen [Wed, 24 Jun 2015 16:50:27 +0000 (13:50 -0300)]
[media] adv7604: Add support for control event notifications

Allow userspace applications to subscribe to control change events. This
can e.g. be used to monitor the 5V detect control to be notified when a
source is connected or disconnected.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years ago[media] v4l2-event: v4l2_event_queue: do nothing if vdev == NULL
Hans Verkuil [Tue, 23 Jun 2015 09:20:23 +0000 (06:20 -0300)]
[media] v4l2-event: v4l2_event_queue: do nothing if vdev == NULL

If the vdev pointer == NULL, then just return.

This makes it easier for subdev drivers to use this function without having to
check if the sd->devnode pointer is NULL or not.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years agoMerge tag 'v4.2-rc2' into patchwork
Mauro Carvalho Chehab [Fri, 17 Jul 2015 12:12:19 +0000 (09:12 -0300)]
Merge tag 'v4.2-rc2' into patchwork

Linux 4.2-rc2

We need to backport changeset 31f02455455d, with fixes a bug at
include/linux/compiler.h that breaks sparse.

* tag 'v4.2-rc2': (221 commits)
  Linux 4.2-rc2
  Revert "drm/i915: Use crtc_state->active in primary check_plane func"
  freeing unlinked file indefinitely delayed
  fix a braino in ovl_d_select_inode()
  9p: don't leave a half-initialized inode sitting around
  tick/broadcast: Prevent NULL pointer dereference
  selinux: fix mprotect PROT_EXEC regression caused by mm change
  parisc: Fix some PTE/TLB race conditions and optimize __flush_tlb_range based on timing results
  stifb: Implement hardware accelerated copyarea
  nfit: add support for NVDIMM "latch" flag
  nfit: update block I/O path to use PMEM API
  tools/testing/nvdimm: add mock acpi_nfit_flush_address entries to nfit_test
  tools/testing/nvdimm: fix return code for unimplemented commands
  tools/testing/nvdimm: mock ioremap_wt
  pmem: add maintainer for include/linux/pmem.h
  Revert "Input: synaptics - allocate 3 slots to keep stability in image sensors"
  arm64: entry32: remove pointless register assignment
  MIPS: O32: Use compat_sys_getsockopt.
  MIPS: c-r4k: Extend way_string array
  MIPS: Pistachio: Support CDMM & Fast Debug Channel
  ...

9 years ago[media] ttusb-dec: use swap() in swap_bytes()
Fabian Frederick [Wed, 10 Jun 2015 16:32:44 +0000 (13:32 -0300)]
[media] ttusb-dec: use swap() in swap_bytes()

Use kernel.h macro definition.

Thanks to Julia Lawall for Coccinelle scripting support.

Signed-off-by: Fabian Frederick <fabf@skynet.be>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years agoLinux 4.2-rc2 v4.2-rc2
Linus Torvalds [Sun, 12 Jul 2015 22:10:30 +0000 (15:10 -0700)]
Linux 4.2-rc2

9 years agoRevert "drm/i915: Use crtc_state->active in primary check_plane func"
Linus Torvalds [Sun, 12 Jul 2015 22:00:20 +0000 (15:00 -0700)]
Revert "drm/i915: Use crtc_state->active in primary check_plane func"

This reverts commit dec4f799d0a4c9edae20512fa60b0a36f3299ca2.

Jörg Otte reports a NULL pointder dereference due to this commit, as
'crtc_state' very much can be NULL:

        crtc_state = state->base.state ?
                intel_atomic_get_crtc_state(state->base.state, intel_crtc) : NULL;

So the change to test 'crtc_state->base.active' cannot possibly be
correct as-is.

There may be some other minimal fix (like just checking crtc_state for
NULL), but I'm just reverting it now for the rc2 release, and people
like Daniel Vetter who actually know this code will figure out what the
right solution is in the longer term.

Reported-and-bisected-by: Jörg Otte <jrg.otte@gmail.com>
Cc: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@intel.com>
CC: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Sun, 12 Jul 2015 21:09:36 +0000 (14:09 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/viro/vfs

Pull VFS fixes from Al Viro:
 "Fixes for this cycle regression in overlayfs and a couple of
  long-standing (== all the way back to 2.6.12, at least) bugs"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  freeing unlinked file indefinitely delayed
  fix a braino in ovl_d_select_inode()
  9p: don't leave a half-initialized inode sitting around

9 years agoMerge branch 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus
Linus Torvalds [Sun, 12 Jul 2015 20:55:24 +0000 (13:55 -0700)]
Merge branch 'upstream' of git://git.linux-mips.org/ralf/upstream-linus

Pull MIPS fixes from Ralf Baechle:
 "A fair number of 4.2 fixes also because Markos opened the flood gates.

   - Patch up the math used calculate the location for the page bitmap.

   - The FDC (Not what you think, FDC stands for Fast Debug Channel) IRQ
     around was causing issues on non-Malta platforms, so move the code
     to a Malta specific location.

   - A spelling fix replicated through several files.

   - Fix to the emulation of an R2 instruction for R6 cores.

   - Fix the JR emulation for R6.

   - Further patching of mindless 64 bit issues.

   - Ensure the kernel won't crash on CPUs with L2 caches with >= 8
     ways.

   - Use compat_sys_getsockopt for O32 ABI on 64 bit kernels.

   - Fix cache flushing for multithreaded cores.

   - A build fix"

* 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus:
  MIPS: O32: Use compat_sys_getsockopt.
  MIPS: c-r4k: Extend way_string array
  MIPS: Pistachio: Support CDMM & Fast Debug Channel
  MIPS: Malta: Make GIC FDC IRQ workaround Malta specific
  MIPS: c-r4k: Fix cache flushing for MT cores
  Revert "MIPS: Kconfig: Disable SMP/CPS for 64-bit"
  MIPS: cps-vec: Use macros for various arithmetics and memory operations
  MIPS: kernel: cps-vec: Replace KSEG0 with CKSEG0
  MIPS: kernel: cps-vec: Use ta0-ta3 pseudo-registers for 64-bit
  MIPS: kernel: cps-vec: Replace mips32r2 ISA level with mips64r2
  MIPS: kernel: cps-vec: Replace 'la' macro with PTR_LA
  MIPS: kernel: smp-cps: Fix 64-bit compatibility errors due to pointer casting
  MIPS: Fix erroneous JR emulation for MIPS R6
  MIPS: Fix branch emulation for BLTC and BGEC instructions
  MIPS: kernel: traps: Fix broken indentation
  MIPS: bootmem: Don't use memory holes for page bitmap
  MIPS: O32: Do not handle require 32 bytes from the stack to be readable.
  MIPS, CPUFREQ: Fix spelling of Institute.
  MIPS: Lemote 2F: Fix build caused by recent mass rename.

9 years agoMerge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 12 Jul 2015 17:02:38 +0000 (10:02 -0700)]
Merge branch 'x86-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Thomas Gleixner:

 - the high latency PIT detection fix, which slipped through the cracks
   for rc1

 - a regression fix for the early printk mechanism

 - the x86 part to plug irq/vector related hotplug races

 - move the allocation of the espfix pages on cpu hotplug to non atomic
   context.  The current code triggers a might_sleep() warning.

 - a series of KASAN fixes addressing boot crashes and usability

 - a trivial typo fix for Kconfig help text

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/kconfig: Fix typo in the CONFIG_CMDLINE_BOOL help text
  x86/irq: Retrieve irq data after locking irq_desc
  x86/irq: Use proper locking in check_irq_vectors_for_cpu_disable()
  x86/irq: Plug irq vector hotplug race
  x86/earlyprintk: Allow early_printk() to use console style parameters like '115200n8'
  x86/espfix: Init espfix on the boot CPU side
  x86/espfix: Add 'cpu' parameter to init_espfix_ap()
  x86/kasan: Move KASAN_SHADOW_OFFSET to the arch Kconfig
  x86/kasan: Add message about KASAN being initialized
  x86/kasan: Fix boot crash on AMD processors
  x86/kasan: Flush TLBs after switching CR3
  x86/kasan: Fix KASAN shadow region page tables
  x86/init: Clear 'init_level4_pgt' earlier
  x86/tsc: Let high latency PIT fail fast in quick_pit_calibrate()

9 years agoMerge branch 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 12 Jul 2015 16:36:59 +0000 (09:36 -0700)]
Merge branch 'timers-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull timer fixes from Thomas Gleixner:
 "This update from the timer departement contains:

   - A series of patches which address a shortcoming in the tick
     broadcast code.

     If the broadcast device is not available or an hrtimer emulated
     broadcast device, some of the original assumptions lead to boot
     failures.  I rather plugged all of the corner cases instead of only
     addressing the issue reported, so the change got a little larger.

     Has been extensivly tested on x86 and arm.

   - Get rid of the last holdouts using do_posix_clock_monotonic_gettime()

   - A regression fix for the imx clocksource driver

   - An update to the new state callbacks mechanism for clockevents.
     This is required to simplify the conversion, which will take place
     in 4.3"

* 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  tick/broadcast: Prevent NULL pointer dereference
  time: Get rid of do_posix_clock_monotonic_gettime
  cris: Replace do_posix_clock_monotonic_gettime()
  tick/broadcast: Unbreak CONFIG_GENERIC_CLOCKEVENTS=n build
  tick/broadcast: Handle spurious interrupts gracefully
  tick/broadcast: Check for hrtimer broadcast active early
  tick/broadcast: Return busy when IPI is pending
  tick/broadcast: Return busy if periodic mode and hrtimer broadcast
  tick/broadcast: Move the check for periodic mode inside state handling
  tick/broadcast: Prevent deep idle if no broadcast device available
  tick/broadcast: Make idle check independent from mode and config
  tick/broadcast: Sanity check the shutdown of the local clock_event
  tick/broadcast: Prevent hrtimer recursion
  clockevents: Allow set-state callbacks to be optional
  clocksource/imx: Define clocksource for mx27

9 years agoMerge branch 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 12 Jul 2015 16:15:02 +0000 (09:15 -0700)]
Merge branch 'irq-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull irq fix from Thomas Gleixner:
 "A single fix for a cpu hotplug race vs. interrupt descriptors:

  Prevent irq setup/teardown across the cpu starting/dying parts of cpu
  hotplug so that the starting/dying cpu has a stable view of the
  descriptor space.  This has been an issue for all architectures in the
  cpu dying phase, where interrupts are migrated away from the dying
  cpu.  In the starting phase its mostly a x86 issue vs the vector space
  update"

* 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  hotplug: Prevent alloc/free of irq descriptors during cpu up/down