- 23 May, 2019 15 commits
-
-
Lubomir Rintel authored
The commit d790b7ed ("[media] vb2-dma-sg: move dma_(un)map_sg here") left dma_desc_nent unset. It previously contained the number of DMA descriptors as returned from dma_map_sg(). We can now (since the commit referred to above) obtain the same value from the sg_table and drop dma_desc_nent altogether. Tested on OLPC XO-1.75 machine. Doesn't affect the OLPC XO-1's Cafe driver, since that one doesn't do DMA. [mchehab+samsung@kernel.org: fix a checkpatch warning] Fixes: d790b7ed ("[media] vb2-dma-sg: move dma_(un)map_sg here") Signed-off-by:
Lubomir Rintel <lkundrak@v3.sk> Signed-off-by:
Sakari Ailus <sakari.ailus@linux.intel.com> Signed-off-by:
Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
-
Lubomir Rintel authored
Add Marvell MMP2 camera host interface. Signed-off-by:
Rob Herring <robh@kernel.org> Signed-off-by:
-
Sakari Ailus authored
C-PHY doesn't use a clock lane, hence the test for the clock lane when there isn't one is faulty. Rework the test for the conflicting clock lane. Signed-off-by:
-
Akinobu Mita authored
If the v4l2 controls are set up for autofoo/foo-type handling by calling v4l2_ctrl_auto_cluster() with the last set_volatile argument setting true, each non-auto control doesn't need to be flagged V4L2_CTRL_FLAG_VOLATILE. Cc: Wenyou Yang <wenyou.yang@microchip.com> Cc: Eugen Hristev <eugen.hristev@microchip.com> Signed-off-by:
Akinobu Mita <akinobu.mita@gmail.com> Signed-off-by:
-
Akinobu Mita authored
Setting the value of the V4L2_CID_VFLIP control is currently ignored. Because V4L2_CID_HFLIP and V4L2_CID_VFLIP are independently controlled in s_ctrl() but these controls belong to the same cluster and the first control is V4L2_CID_HFLIP. Fix it by not clustering these controls. Also correct erroneous updating register bit for vertical flip. Cc: Wenyou Yang <wenyou.yang@microchip.com> Cc: Eugen Hristev <eugen.hristev@microchip.com> Signed-off-by:
-
Akinobu Mita authored
If the requested framesize by VIDIOC_SUBDEV_S_FMT is larger than supported framesizes, it causes an out of bounds array access and the resulting framesize is unexpected. Avoid out of bounds array access and select the default framesize. Cc: Wenyou Yang <wenyou.yang@microchip.com> Cc: Eugen Hristev <eugen.hristev@microchip.com> Signed-off-by:
-
Akinobu Mita authored
Avoid returning without decrement the usage count in s_ctrl(). Cc: Wenyou Yang <wenyou.yang@microchip.com> Cc: Eugen Hristev <eugen.hristev@microchip.com> Signed-off-by:
-
Hans Verkuil authored
Document which capability is required, rather than just saying that "root permissions" are required. Signed-off-by:
Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by:
-
Hans Verkuil authored
Some displays pull down the HPD when in standby, but CEC is still active and the display can be woken up by sending an Image View On or Text View On CEC command. The CEC specification doesn't tell you what the initiator should be for such a command (without a HPD it's unclear if the CEC adapter can claim a logical address). This patch allows any initiator value when there is no HPD for the Image/Text View On commands and for the Ping command. Signed-off-by:
-
Hans Verkuil authored
If this flag is set, then check for root permissions and skip all message checks expect for the core checks (i.e. validate the length etc.). Signed-off-by:
-
Hans Verkuil authored
Document this new cec_msg flag. Signed-off-by:
-
Hans Verkuil authored
This adds the userspace API to send raw unchecked CEC messages. This will require root permissions. Signed-off-by:
-
Hans Verkuil authored
This ensures all the cec_msg checks are done in the same place. Signed-off-by:
-
Hans Verkuil authored
The code that fills in the CEC_MSG_CDC_MESSAGE physical address is now done after the sanity checks. It also only does this if the message length is >= 4 (i.e. there is room for the physical address). Signed-off-by:
-
Hans Verkuil authored
Due to limitations in gpiolib it was impossible to disable the interrupt of an input gpio and then switch it to gpio output and drive it. The only way to achieve that was to free the interrupt first, then switch the direction. When going back to gpio input and using interrupts to read the gpio pin you had to request the irq again. This limitation was lifted in gpiolib in kernel 4.20, but the cec-gpio driver was still using the old workaround implementation. This patch updates the cec-gpio driver to just enable and disable the irq. Signed-off-by:
-
- 22 May, 2019 8 commits
-
-
Hans Verkuil authored
The cec device node can be used right after it was created, but that leaves a race condition where the device was created, but devnode->registered was still false. So an ioctl() would result in an error. So set it to true before calling cdev_device_add() and to false again if cdev_device_add returned an error. Signed-off-by:
-
Neil Armstrong authored
Remove comment and replace with the appropriate SPDX identifier. Signed-off-by:
Neil Armstrong <narmstrong@baylibre.com> Signed-off-by:
Sean Young <sean@mess.org> Signed-off-by:
-
Stefan Brüns authored
All error cases inside the function already report errors via dev_err(), and dvb_usb_v2_generic_rw also reports all error cases, so there is no silent code path when an error has occurred. Signed-off-by:
Stefan Brüns <stefan.bruens@rwth-aachen.de> Signed-off-by:
-
Stefan Brüns authored
Errors are already reported by the common code in dvb_usb_v2_generic_io (which dvbsky_usb_generic_rw is a wrapper of), so there is no reason report the error again. Signed-off-by:
-
Stefan Brüns authored
actual_length != wlen is the only error path which does not generate an error message. Adding an error message here allows to report a more specific error and to remove the error reporting from the call sites. Also clean up the error paths - in case of an error, the remaining code is skipped, and ret is returned. Skip setting ret and return immediately (no cleanup necessary). Signed-off-by:
-
Sean Young authored
au0828_usb_disconnect() gets the au0828_dev struct via usb_get_intfdata, so it needs to set up for the error paths. Reported-by: syzbot+357d86bcb4cca1a2f572@syzkaller.appspotmail.com Signed-off-by:
-
Oliver Neukum authored
dvb_usb_device_exit() frees and uses the device name in that order. Fix by storing the name in a buffer before freeing it. Signed-off-by:
Oliver Neukum <oneukum@suse.com> Reported-by: syzbot+26ec41e9f788b3eba396@syzkaller.appspotmail.com Signed-off-by:
-
Daniel Gomez authored
MODULE_DEVICE_TABLE(of, <of_match_table> should be called to complete DT OF mathing mechanism and register it. Before this patch: modinfo drivers/media/rc/ir-spi.ko | grep alias After this patch: modinfo drivers/media/rc/ir-spi.ko | grep alias alias: of:N*T*Cir-spi-ledC* alias: of:N*T*Cir-spi-led Reported-by:
Javier Martinez Canillas <javier@dowhile0.org> Signed-off-by:
Daniel Gomez <dagmcr@gmail.com> Signed-off-by:
-
- 21 May, 2019 16 commits
-
-
Brad Love authored
Minimum, maximum, and stepsize taken from Silicon Labs reference. Signed-off-by:
Brad Love <brad@nextdimension.cc> Signed-off-by:
-
Stanimir Varbanov authored
Venus v4 doesn't send ALLOC_MODE property and thus parser doesn't recognize it as dynamic buffer (for OUTPUT/OUTPUT2 type of buffers) make it obvious in the helper function. Signed-off-by:
Stanimir Varbanov <stanimir.varbanov@linaro.org> Signed-off-by:
-
Stanimir Varbanov authored
Add two more not-implemented properties for Venus v4. Signed-off-by:
-
Kelvin Lawson authored
Add support for V4L2 H265 controls: * V4L2_CID_MPEG_VIDEO_HEVC_PROFILE * V4L2_CID_MPEG_VIDEO_HEVC_LEVEL Signed-off-by:
Kelvin Lawson <klawson@lisden.com> Signed-off-by:
-
Alexandre Courbot authored
Patch commit de5a0baf ("media: venus: core: correct maximum hardware load for sdm845") meant to increase the maximum hardware load for sdm845, but ended up changing the one for msm8996 instead. Fixes: de5a0baf ("media: venus: core: correct maximum hardware load for sdm845") Signed-off-by:
Alexandre Courbot <acourbot@chromium.org> Signed-off-by:
-
Philipp Zabel authored
The header report return value from decoder sequence initialization is available on CodaHx4 and CODA7541 already. Use the profile and level identification values reported by the firmware to update codec specific profile and level controls after sequence initialization has succeeded. Signed-off-by:
Philipp Zabel <p.zabel@pengutronix.de> Signed-off-by:
-
Philipp Zabel authored
When a menu control is updated via s_ctrl, print the corresponding menu entry string in addition to the numerical value it is set to. Signed-off-by:
-
Philipp Zabel authored
Otherwise the default initialization would always swamp the debug log. Signed-off-by:
-
Philipp Zabel authored
Print a single line containing the following information: - which frame was decoded, including its type, - if no frame was decoded, the reason (code) why - which decoded frame was returned, copied out by either rotator or VODA, - if no frame was returned, the reason (code) why, and - the output queue sequence number, which is only useful in case each queued coded buffer corresponds to exactly one frame. Signed-off-by:
-
Philipp Zabel authored
Add a function to translate from V4L2 buffer flags to 'I'/'P'/'B' characters for debug output. Signed-off-by:
-
Philipp Zabel authored
Use the newly freed verbose debug level 2 for job ready debug messages. Signed-off-by:
-
Philipp Zabel authored
This allows to use coda_debug level 2 for verbose but not quite as verbose debug logging. Register access level logging is of limited use anyway, as this includes busy polling of status bits. Signed-off-by:
-
Eugen Hristev authored
In case the sensor refuses to set the format, avoid printing the error message that no compatible format was found. This means that the try_fmt will be less verbose. The error will be printed only if really a format cannot be found. Some application try all possible formats in a row (gstreamer e.g.) which will flood the console with error messages until a working one is found. Signed-off-by:
Eugen Hristev <eugen.hristev@microchip.com> Signed-off-by:
-
Eugen Hristev authored
This adds support for the 'button' control DO_WHITE_BALANCE This feature will enable the ISC to compute the white balance coefficients in a one time shot, at the user discretion. This can be used if a color chart/grey chart is present in front of the camera. The ISC will adjust the coefficients and have them fixed until next balance or until sensor mode is changed. This is particularly useful for white balance adjustment in different lighting scenarios, and then taking photos to similar scenery. The old auto white balance stays in place, where the ISC will adjust every 4 frames to the current scenery lighting, if the scenery is approximately grey in average, otherwise grey world algorithm fails. One time white balance adjustments needs streaming to be enabled, such that capture is enabled and the histogram has data to work with. Histogram without capture does not work in this hardware module. To start the one time white balance procedure: v4l2-ctl --set-ctrl=do_white_balance=1 This feature works only if the sensor is streaming RAW data, as the hardware supports a histogram only for RAW bayer components. If the auto white balance is enabled, do_white_balance does nothing. If the streaming is disabled, or the sensor does not output RAW data, the control is inactive. User controls now include the do_white_balance ctrl: User Controls brightness 0x00980900 (int) : min=-1024 max=1023 step=1 default=0 value=0 flags=slider contrast 0x00980901 (int) : min=-2048 max=2047 step=1 default=256 value=256 flags=slider white_balance_automatic 0x0098090c (bool) : default=1 value=0 do_white_balance 0x0098090d (button) : flags=write-only, execute-on-write gamma 0x00980910 (int) : min=0 max=2 step=1 default=2 value=2 flags=slider Signed-off-by: -
Eugen Hristev authored
Control DO_WHITE_BALANCE is a button, with read only and execute-on-write flags. Adding this control in the proper list in the fill function. After adding it here, we can see output of v4l2-ctl -L do_white_balance 0x0098090d (button) : flags=write-only, execute-on-write Signed-off-by:
-
Eugen Hristev authored
Reworked auto white balance feature (awb) to cope with all four channels. Implemented stretching and grey world algorithms. Using the histogram, the ISC will auto adjust the white balance during frame captures. Because each histogram needs a frame, it will take 4 frames for one adjustment. When the gains were updated by previous code, the registers for the gains were updated only on new streaming start. Now, after each full histogram the registers are updated with new gains. Also, on previous code, if the streaming stopped but not all 3 histograms finished, a new histogram was started either way. This used to lead to an error "timeout to update profile" when streaming was stopped. According to the hardware, histogram can only work together with the capture, not independently. Signed-off-by:
-
- 19 May, 2019 1 commit
-
-
Linus Torvalds authored
-
