- 01 Sep, 2020 12 commits
-
-
Jonathan Bakker authored
On some devices, there may be multiple camera sensors attached to the same port. Make sure we probe all of them, not just the first one. Signed-off-by:
Jonathan Bakker <xc-racer2@live.ca> Reviewed-by:
Sylwester Nawrocki <s.nawrocki@samsung.com> Signed-off-by:
Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by:
Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
-
Jonathan Bakker authored
Instead of keeping a local copy of how many sensors we've probed (which may not even properly represent the number of sensors probed if we have a port without a sensor), use the global num_sensors counter that has the actual number used. This will also make it easier to add support for multiple sensors being connected to the same port. Signed-off-by:
-
Jonathan Bakker authored
Commit ee7160e5 ("[media] s5p-fimc: Add support for JPEG capture") added support for JPEG capture, but missed setting a register when the parallel port was used rather than the CSIS device. Signed-off-by:
-
Jonathan Bakker authored
Commit bae45003 ("[media] exynos4-is: Add missing entity function initialization") tried to suppress the warnings such as s5p-fimc-md camera: Entity type for entity FIMC.0 was not initialized! However, this missed setting for the subdev. Set it now to avoid the ugly warnings on boot. Signed-off-by:
-
Jonathan Bakker authored
Not all devices use the CSIS device, some may use the FIMC directly in which case the CSIS device isn't registered. This leads to a nullptr exception when starting the stream as the CSIS device is always referenced. Instead, if getting the CSIS device fails, try getting the FIMC directly to check if we are using the subdev API Signed-off-by:
-
Tomasz Figa authored
On FIMC variants which don't have writeback channel, there is no need to access system registers. This patch makes the driver request sysreg regmap conditionally depending on whether writeback is supported. Signed-off-by:
Tomasz Figa <tomasz.figa@gmail.com> Signed-off-by:
Tomasz Figa <tfiga@chromium.org> Reviewed-by:
-
Jonathan Bakker authored
The S5PV210 platform only supports device tree based booting where the FIMC variant data is parsed directly from the device tree, hence the now unused static data can be removed. Signed-off-by:
-
Ezequiel Garcia authored
When the post-processor is enabled, the driver allocates "shadow buffers" which are used for the decoder core, and exposes the post-processed buffers to userspace. For this reason, extra motion vector space has to be allocated on the shadow buffers, which the driver wasn't doing. Fix it. This fix should address artifacts on high profile bitstreams. Fixes: 8c2d66b0 ("media: hantro: Support color conversion via post-processing") Signed-off-by:
Ezequiel Garcia <ezequiel@collabora.com> Reviewed-by:
Philipp Zabel <p.zabel@pengutronix.de> Signed-off-by:
-
Ezequiel Garcia authored
If the bitstream and the application are incorrectly configuring the reference pictures, the hardware will need to fallback to using some other reference picture. When the post-processor is enabled, the fallback buffer should be a shadow buffer (postproc.dec_q), and not a CAPTURE queue buffer, since the latter is post-processed and not really the output of the decoder core. Fixes: 8c2d66b0 ("media: hantro: Support color conversion via post-processing") Signed-off-by:
-
Julia Lawall authored
list_for_each_entry is able to handle an empty list. The only effect of avoiding the loop is not initializing the index variable. Drop list_empty tests in cases where these variables are not used. Note that list_for_each_entry is defined in terms of list_first_entry, which indicates that it should not be used on an empty list. But in list_for_each_entry, the element obtained by list_first_entry is not really accessed, only the address of its list_head field is compared to the address of the list head, so the list_first_entry is safe. The semantic patch that makes this change is as follows: (http://coccinelle.lip6.fr/) <smpl> @@ expression x,e; iterator name list_for_each_entry; statement S; identifier i; @@ -if (!(list_empty(x))) { list_for_each_entry(i,x,...) S - } ... when != i ? i = e </smpl> Signed-off-by:
Julia Lawall <Julia.Lawall@inria.fr> Signed-off-by:
-
Julia Lawall authored
list_for_each_safe is able to handle an empty list. The only effect of avoiding the loop is not initializing the index variable. Drop list_empty tests in cases where these variables are not used. The semantic patch that makes this change is as follows: (http://coccinelle.lip6.fr/) <smpl> @@ expression x,e; iterator name list_for_each_safe; statement S; identifier i,j; @@ -if (!(list_empty(x))) { list_for_each_safe(i,j,x) S - } ... when != i when != j ( i = e; | ? j = e; ) </smpl> Signed-off-by:
-
Julia Lawall authored
list_for_each_entry is able to handle an empty list. The only effect of avoiding the loop is not initializing the index variable. Drop list_empty tests in cases where these variables are not used. Note that list_for_each_entry is defined in terms of list_first_entry, which indicates that it should not be used on an empty list. But in list_for_each_entry, the element obtained by list_first_entry is not really accessed, only the address of its list_head field is compared to the address of the list head, so the list_first_entry is safe. The semantic patch that makes this change is as follows: (http://coccinelle.lip6.fr/) <smpl> @@ expression x,e; iterator name list_for_each_entry; statement S; identifier i; @@ -if (!(list_empty(x))) { list_for_each_entry(i,x,...) S - } ... when != i ? i = e </smpl> Signed-off-by:
-
- 29 Aug, 2020 28 commits
-
-
Gustavo A. R. Silva authored
Replace the existing /* fall through */ comments and its variants with the new pseudo-keyword macro fallthrough[1]. Also, remove unnecessary fall-through markings when it is the case. [1] https://www.kernel.org/doc/html/v5.7/process/deprecated.html?highlight=fallthrough#implicit-switch-case-fall-throughSigned-off-by:
Gustavo A. R. Silva <gustavoars@kernel.org> Acked-by:
Sakari Ailus <sakari.ailus@linux.intel.com> Signed-off-by:
-
Helen Koike authored
The driver access struct's fields one by one, they are not blindly copied to/from the hardware, so we can remove __packed from the structs. Tested on Scarlet Chromebook with ChromeOS stack and verified images with the Camera app. Signed-off-by:
Helen Koike <helen.koike@collabora.com> Signed-off-by:
-
Dafna Hirschfeld authored
In the irq handler 'rkisp1_params_isr', the lock 'config_lock' should be held as long as the current buffer is used. Otherwise the stop_streaming calback might remove it from the list and pass it to userspace while it is referenced in the irq handler. Signed-off-by:
Dafna Hirschfeld <dafna.hirschfeld@collabora.com> Acked-by:
-
Dafna Hirschfeld authored
replace a read and write of RKISP1_CIF_ISP_CTRL register in the isr with the 'rkisp1_param_set_bits' function that does that. Signed-off-by:
-
Dafna Hirschfeld authored
Currently the params isr is called and then returned when isp-frame interrupt is not set. This condition is already tested in the isp's isr so move the call under the condition in the isp's isr. Signed-off-by:
-
Dafna Hirschfeld authored
The fields in struct 'rkisp1_cif_isp_lsc_config' need to hold values of 12 bits, therefore it is enough to declare them as u16. Signed-off-by:
-
Dafna Hirschfeld authored
The two macros RKISP1_CIF_ISP_LSC_{GRAD/SIZE}_TBL_SIZE have the same value which is the number of sectors in the grads and size parameters. Therefore there is no need for two different macros. Replace them with one macro RKISP1_CIF_ISP_LSC_SECTORS_TBL_SIZE. Also use the macro when iterating the arrays instead of hardcoded '4'. Signed-off-by: -
Dafna Hirschfeld authored
Currently the fields '*_data_tbl' in the struct 'rkisp1_cif_isp_lsc_config' are one dimensional arrays of size 290. The entries are accessed in a nested loop by summing two indexes for the rows and columns. Therefore it fits better to define the arrays as two dimensional arrays of size 17x17. Signed-off-by:
-
Dafna Hirschfeld authored
The table of the flags of the structs v4l2_pix_format(_mplane) is currently in pixfmt-reserved.rst which is wrong, it should be in pixfmt-v4l2.rst Signed-off-by:
-
Dafna Hirschfeld authored
Add more detailed documentation of the structs and functions in rkisp1-common.h Signed-off-by:
-
Dafna Hirschfeld authored
In file rkisp1-common.h, group declaration of register/unregister functions together and group other functions together to make the code easier to read. Signed-off-by:
-
Dafna Hirschfeld authored
The file rkisp1-common.h has a function declarion 'rkisp1_params_isr_handler' that is not implemented. Remove it. Signed-off-by:
-
Dafna Hirschfeld authored
The different register/unregister functions receive different parameters. This patch unify them so they all receive just 'struct *rkisp1_device' as parameter. Signed-off-by:
-
Dafna Hirschfeld authored
The field vaddr in rkisp1_buffer struct is used only by the rkisp1-stats and rkisp1-params entities and they both use only vaddr[0] so there is no need to define this field as an array. Signed-off-by:
-
Dafna Hirschfeld authored
There is some missing documentation of structs and fields in rkisp1-config.h. This patch improves the documentation. Signed-off-by:
-
Dafna Hirschfeld authored
The flag RKISP1_CIF_ISP_STAT_AFM_FIN indicates userspace that auto-focus measurements were collected. Therefore the suffix _FIN in the flag's name does not fit. Signed-off-by:
-
Dafna Hirschfeld authored
The struct rkisp1_cif_isp_ctk_config contains 3 fields ct_offset_{rgb}. Replace them with one array field 'ct_offset[3]. Signed-off-by: -
Dafna Hirschfeld authored
The struct rkisp1_cif_isp_ctk_config has 9 fields 'coeff*' for the 3x3 color correction matrix. Replace these fields with one 3x3 array coeff[3][3] and document the field. Signed-off-by:
-
Sylwester Nawrocki authored
There is no need to request the "idle" pinctrl state in the driver as that is implemented in the driver core and the pinctrl_pm_* API can be used for switching between the default and the idle state. Simplify the pinctrl code to only request and check for the mandatory "default" pinctrl state. Switching between the default/idle pinctrl state is not yet implemented in the driver and this patch doesn't change that. Signed-off-by:
-
Sylwester Nawrocki authored
The "idle" pinctrl state is optional as documented in the DT binding. The change introduced by the commit being reverted makes that pinctrl state mandatory and breaks initialization of the whole media driver, since the "idle" state is not specified in any mainline dts. This reverts commit 18ffec75 ("media: exynos4-is: Add missed check for pinctrl_lookup_state()") to fix the regression. Fixes: 18ffec75 ("media: exynos4-is: Add missed check for pinctrl_lookup_state()") Signed-off-by:
-
Andrey Konovalov authored
Use the definitions introduced by commit "media: camss: add support for vidioc_enum_framesizes ioctl" instead of the hardcoded values. Signed-off-by:
Andrey Konovalov <andrey.konovalov@linaro.org> Signed-off-by:
-
Andrey Konovalov authored
VIDIOC_ENUM_FRAMESIZES support in the video capture driver is required by libcamera. Without this change libcamera errors out with: "ERROR V4L2 v4l2_videodevice.cpp:1059 /dev/video0[cap]: Unable to enumerate frame sizes: Inappropriate ioctl for device" Signed-off-by:
-
Andrey Konovalov authored
Implement mbus_code filtering for format enumeration. Without this patch libcamera errors out with: "ERROR V4L2 v4l2_videodevice.cpp:982 /dev/video0[cap]: Media bus code filtering not supported by the device" Signed-off-by:
-
Niklas Söderlund authored
The media device is not usable by userspace before all devices involved in capture are present in the system. Move registering of the media device to the async complete callback. Signed-off-by:
Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se> Signed-off-by:
-
Niklas Söderlund authored
If the VIN device is part of a group of VIN devices (all Gen3 boards) there is no reason to only unregister the group notifier if the VIN that registers the notifier is removed. The VIN that registers the notifier is always the last VIN device to be bound, so keeping the notifier around after any VIN is unbound creates an unbalanced state where no VIN in the group is operational. Fix this by unconditionally unregistering the notifier when any VIN device is unbound. Unregistering the notifier will lead to unbound() being called and all video devices exposed by any VIN instance to be removed. The lock was only needed to protect the check which VIN registers the notifier and is no longer needed. Signed-off-by:
-
Tom Rix authored
clang static analysis reports this problem tw5864-video.c:773:32: warning: The left expression of the compound assignment is an uninitialized value. The computed value will also be garbage fintv->stepwise.max.numerator *= std_max_fps; ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ stepwise.max is set with frameinterval, which comes from ret = tw5864_frameinterval_get(input, &frameinterval); fintv->stepwise.step = frameinterval; fintv->stepwise.min = frameinterval; fintv->stepwise.max = frameinterval; fintv->stepwise.max.numerator *= std_max_fps; When tw5864_frameinterval_get() fails, frameinterval is not set. So check the status and fix another similar problem. Signed-off-by:Tom Rix <trix@redhat.com> Signed-off-by:
-
Ismael Luceno authored
Remove broken bluecherry.net email Signed-off-by:
Ismael Luceno <ismael@iodev.co.uk> Signed-off-by:
-
Frederic Chen authored
We modified the type of media_device_request_alloc()'s second parameter from int* to void* so that it can match the interface defined in struct media_ioctl_info. [hverkuil: move #ifdef before variable to avoid compile warning] Signed-off-by:
Frederic Chen <frederic.chen@mediatek.com> Reviewed-by:
Laurent Pinchart <laurent.pinchart@ideasonboard.com> Signed-off-by:
-
