commit fc1e2c8ea85e109acf09e74789e9b852f6eed251 upstream.

Commit 367e8560e8d7a62d96e9b1d644028a3816e04206 introduced a bug
in fbtft-core where fps is always 0, this is because variable
update_time is not assigned correctly.

Signed-off-by: Ksenija Stanojevic <ksenija.stanojevic@gmail.com>
Fixes: 367e8560e8d7 ("Staging: fbtbt: Replace timespec with ktime_t")
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Pull media fixes from Mauro Carvalho Chehab:

 - several fixes for new drivers added for Kernel 4.8 addition (cec
   core, pulse8 cec driver and Mediatek vcodec)

 - a regression fix for cx23885 and saa7134 drivers

 - an important fix for rcar-fcp, making rcar_fcp_enable() return 0 on
   success

* tag 'media/v4.8-7' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media: (25 commits)
  [media] cx23885/saa7134: assign q->dev to the PCI device
  [media] rcar-fcp: Make sure rcar_fcp_enable() returns 0 on success
  [media] cec: fix ioctl return code when not registered
  [media] cec: don't Feature Abort broadcast msgs when unregistered
  [media] vcodec:mediatek: Refine VP8 encoder driver
  [media] vcodec:mediatek: Refine H264 encoder driver
  [media] vcodec:mediatek: change H264 profile default to profile high
  [media] vcodec:mediatek: Add timestamp and timecode copy for V4L2 Encoder
  [media] vcodec:mediatek: Fix visible_height larger than coded_height issue in s_fmt_out
  [media] vcodec:mediatek: Fix fops_vcodec_release flow for V4L2 Encoder
  [media] vcodec:mediatek:code refine for v4l2 Encoder driver
  [media] cec-funcs.h: add missing vendor-specific messages
  [media] cec-edid: check for IEEE identifier
  [media] pulse8-cec: fix error handling
  [media] pulse8-cec: set correct Signal Free Time
  [media] mtk-vcodec: add HAS_DMA dependency
  [media] cec: ignore messages when log_addr_mask == 0
  [media] cec: add item to TODO
  [media] cec: set unclaimed addresses to CEC_LOG_ADDR_INVALID
  [media] cec: add CEC_LOG_ADDRS_FL_ALLOW_UNREG_FALLBACK flag
  ...

Don't return the confusing -EIO error code when the device is not registered,
instead return -ENODEV which is the proper thing to do in this situation.

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

If the adapter is configured as 'Unregistered', then cec_receive_notify
incorrectly thinks that broadcast messages are directed messages. The
destination for broadcast messages is 0xf, and the logical address
assigned to Unregistered devices is also 0xf and the logic didn't handle
that correctly.

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

Jonathan writes:

First round of IIO fixes for the 4.8 cycle.

This is somewhat of a bumper set due to my unavailabilty earlier in the
month.  The only slightly unusual ones are the dts updates for the
rockchip adc reset.  The fix in the driver only makes sense with these
and the rockchip maintainer is happy with them going through IIO to
reach mainline.

Core stuff

* Fix an issue with a blocking op when !TASK_RUNNING. This been there
  a while and snuck in with seemingly minor additions to some core
  code paths.
* Tools
  - generic_buffer failed to initialize the channel array pointer thus
    in the case of no channels blows up trying to free a random memory
    address.
* sw-trigger:
  - Fix config group initialization when configfs is built as a module.

Drivers

* ad5933
  - Fix an incorrect overwrite of an error value.
* ad799x
  - A missed assignment of the update_scan_mode callback means buffered mode
    doesn't work on the ad7991, ad7995 or ad7999.
* ads1015
  - wrong pointer returned from i2c_get_clientdata (missmatch of assumptions)
* am2315
  - Timestamps are reported, but never actually acquired from anywhere
    (so always 0)
  - missing buffer selects in Kconfig
* am335x adc
  - Protect fifo1 from concurrent access.
  - Increase timeout waiting for ADC to be long enough in all cases.
* as3935
  - Timestamps are reported, but never actually acquired from anywhere
    (so always 0)
* at91
  - Fix reading of channel 3.
* atlas-ph-sensor
  - Typo means that the scale of electrical conductivity readings is way off.
* bma220
  - Timestamps are reported, but never actually acquired from anywhere
    (so always 0)
  - Missing buffer selects in Kconfig
* bmp280
  - pass the write pointer to PTR_ERR (i.e. the one that was just checked
    with IS_ERR).
  - suspend /resume crash due to wrong assumption about what dev_get_drvdata
    would return.
* hdc100x
  - It superficially appeared that smbus_read_byte commands would allow
  reading of the outputs in two goes.  In reality it doesn't work, but
  instead returns the same for the upper and lower bytes (nice catch from
  Alison!)
* kxsd9
  - Fix raw read return value to ensure it actually reports the value rather
  than a blank string.
* max44000
  - Missing buffer selects in Kconfig
* rockchip_saradc
  - Add use of reset controller to enforce a clean state of the ADC.
    Some bootloaders can leave it in an 'intersting' state and effectively
    frozen without this. A couple of associated dts updates.
* stk8ba50
  - Missing buffer selects in Kconfig
* stx104
  - Fix a possible race due to use of devm_iio_device_register when there
    was other stuff in the remove function.

Support more error codes and fix a bug where MSGCODE_TRANSMIT_FAILED_LINE
was mapped to CEC_TX_STATUS_ARB_LOST, which is wrong.

Thanks to Pulse-Eight for providing me with the information needed
to handle this correctly (I hope).

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

Don't hardcode the signal free time to 3 bit periods, instead use
the value for the signal free time as passed in by the CEC framework.

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

Most CEC adapters will still receive broadcast messages, even if no logical
addresses are claimed. But those messages should only be passed on for
monitoring purposes, but not for processing by either kernel or userspace
if userspace didn't call CEC_ADAP_S_LOG_ADDRS first.

So if adap->log_addrs.log_addr_mask is 0, then just return before passing
the received message on to the processing code.

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

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

Up to 4 logical addresses can be claimed. Make sure that any
unclaimed logical addresses are set to CEC_LOG_ADDR_INVALID as
per the documentation.

Take special care in the unregistered case: when falling back to
unregistered num_log_addrs may be > 1, so mark those as invalid.

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