- 19 Jul, 2020 40 commits
-
-
Laurent Pinchart authored
The sensor is connected to the CAMERARX. Store its pointer, as well as the sensor pixel rate and the endpoint, in the cal_camerarx. This prepares for decoupling the cal_camerarx and cal_ctx. Signed-off-by:
Laurent Pinchart <laurent.pinchart@ideasonboard.com> Reviewed-by:
Tomi Valkeinen <tomi.valkeinen@ti.com> Reviewed-by:
Benoit Parrot <bparrot@ti.com> Signed-off-by:
Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by:
Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
-
Laurent Pinchart authored
Prepare for passing the cal_camerarx pointer instead of the cal_ctx pointer to CAMERARX-related functions by adding print macros for cal_camerarx. Signed-off-by:
-
Laurent Pinchart authored
Use the dev_* print macros instead of the v4l2_* print macros. This prepares for a common print infrastructure that will also support the cal_camerarx instances. Signed-off-by:
-
Laurent Pinchart authored
Store a pointer to the cal_dev and the cal_camerarx instance number in the cal_camerarx structure. This prepares for passing a cal_camerarx pointer instead of a cal_ctx pointer to multiple functions that deal with the CAMERARX. Signed-off-by:
-
Laurent Pinchart authored
The cal_camerarx structure describes the PHY, there's no need for an internal structure named phy. Removed that level of indirection. Signed-off-by:
-
Laurent Pinchart authored
Name all variables that point to a cal_camerax instance 'phy' instead of 'cc'. The name 'cc' refers to Camera Core, but is not commonly used in the driver or in datasheets. Signed-off-by:
-
Laurent Pinchart authored
Name all variables pointing to a cal_dev instance 'cal', instead of a combination of 'caldev' and 'dev'. 'caldev' is only used in a few places and could have been renamed to 'dev', but 'dev' is confusing as it's also used for struct device instances. This generates lots of changes, but increases readability. Signed-off-by:
-
Laurent Pinchart authored
The cal_csi2_phy contains platform data for the CAMERARX blocks. Rename it to cal_camerarx_data. Signed-off-by:
-
Laurent Pinchart authored
The various data structures that describe the components of the camera access layer (CAL) are named without much consistency. Start cleaning this up by renaming the structure that describes the CAMERARX block, cc_data, to cal_camerarx. Signed-off-by:
-
Laurent Pinchart authored
Move the csi2_phy_config() function to avoid its forward declaration. No functional change is included. Signed-off-by:
-
Laurent Pinchart authored
The cal_dev flags field is a copy of data->flags. Use the latter and remove the former. Signed-off-by:
-
Laurent Pinchart authored
Remove structure fields that are never set, set and never read, or set to a fixed value. This allows removal of a global variable and a macro. Signed-off-by:
-
Laurent Pinchart authored
The global static const cal_regmap_config template is only used in a single location, to initialize 3 fields of a local variable. Two of those fields then get overwritten. Remove the template and set the last remaining field manually. Simplify the code by hardcoding the field values instead of calculating them for a variable that always has the same value. Signed-off-by:
-
Laurent Pinchart authored
Now that no platform data fields are modified at runtime, make the corresponding structures const. The const qualifier for integer fields is removed as it's not required anymore. Signed-off-by:
-
Laurent Pinchart authored
Several structure fields declared as int store positive values only. Make them unsigned. Signed-off-by:
-
Laurent Pinchart authored
The cal_csi2_phy structure has lost its 'fields' field that used to clash with the 'base_fields' field. Rename base_fields to fields. Signed-off-by:
-
Laurent Pinchart authored
The PHY regmap fields are created dynamically at runtime. They don't belong to the platform data that should be constant. Move them to the cc_data structure. Signed-off-by:
-
Laurent Pinchart authored
In preparation to constify cal_csi2_phy, avoid modifying the base_fields array at runtime. As the array now only needs to stored two values instead of a full struct reg_field instance, save memory by using a custom structure for the base_fields elements. Signed-off-by:
-
Laurent Pinchart authored
The base_fields field of cal_csi2_phy points to a fixed-size array of struct reg_field. This level of indirection isn't needed, embed the array directly in struct cal_csi2_phy. Signed-off-by:
-
Laurent Pinchart authored
Many loop indices only take positive values. Make them unsigned. Signed-off-by:
-
Laurent Pinchart authored
The found_port variable contains a boolean value, make it a bool. Signed-off-by:
-
Laurent Pinchart authored
Casting from a void pointer to a struct pointer doesn't require a cast. Remove them. Signed-off-by:
-
Laurent Pinchart authored
A few local variables are needlessly initialized. Fix them. Signed-off-by:
-
Laurent Pinchart authored
The static cal_formats array contains const data, make it const. This requires turning a few cal_fmt pointers into const pointers. Signed-off-by:
-
Laurent Pinchart authored
Turn the reg_(read|write)_field() macros into inline functions for additional type safety. Use the FIELD_GET() and FIELD_PREP() macros internally instead of reinventing the wheel. Signed-off-by:
-
Laurent Pinchart authored
The CAL_VERSION macro is used once only, in MODULE_VERSION(). This doesn't improve readability, inline it. Signed-off-by:
-
Laurent Pinchart authored
The cal_irq() function reads three IRQ status registers and stores their values in three different variables. As each value is processed right after reading the corresponding register, a single variable is enough. Signed-off-by:
-
Laurent Pinchart authored
The datasheet numbers the IRQ registers starting from 0. Update the numbering scheme to match the datasheet. Signed-off-by:
-
Laurent Pinchart authored
The datasheet numbers the CSI-2 PHYs (external to the CAL) starting from 1, but the PPI inputs to the CAL starting from 0. Update the numbering scheme for csi2_port to match the datasheet, which also simplifies the driver by removing the need to constantly subtract 1 from the value. Signed-off-by:
-
Laurent Pinchart authored
The CPORT mentioned in the CAL documentation has nothing to do with the CSI-2 ports. It's a remnant of CSI-3 support (or an attempt to achieve it), and now only serves as an arbitraty tag value used to associate data samples with contexts. Add a cport field to the cal_ctx structure and use it instead of the CSI-2 port number where a CPORT is required by the hardware. The value of the cport field is currently the same as the csi2_port field, so there's no functional change, but this will help reworking CPORT and CSI-2 port handling separately. Signed-off-by:
-
Laurent Pinchart authored
Move the cal_complete_ctx() function earlier in the file to avoid a foward declaration. There is no functional change. Signed-off-by:
-
Laurent Pinchart authored
Sorting headers alphabetically helps locating duplicates. Signed-off-by:
-
Florin Tudorache authored
If a device supports a resolution in 2 different formats, mode[0] is skipped on first iteration and a different format is selected. Example: 320x240/YUYV isn't reachable for OV534/ov772x Signed-off-by:
Florin Tudorache <florin_tudorache@live.com> Signed-off-by:
-
Christophe JAILLET authored
Memory allocated with 'v4l2_m2m_init()' must be freed by a corresponding call to 'v4l2_m2m_release()' Also reorder the code at the end of the probe function so that 'video_register_device()' is called last. Update the error handling path accordingly. Fixes: 5ce60d79 ("[media] s5p-g2d: Add DT based discovery support") Fixes: 91884734 ("[media] v4l: add G2D driver for s5p device family") Signed-off-by:
Christophe JAILLET <christophe.jaillet@wanadoo.fr> Signed-off-by:
-
Ezequiel Garcia authored
It's all too easy to get confused by the V4L2_TYPE_IS_OUTPUT macro, when it's used as !V4L2_TYPE_IS_OUTPUT. Reduce the risk of confusion with macro to explicitly check for the CAPTURE queue type case. This change does not affect functionality, and it's only intended to make the code more readable. Suggested-by:
Nicolas Dufresne <nicolas.dufresne@collabora.com> Signed-off-by:
Ezequiel Garcia <ezequiel@collabora.com> Signed-off-by:
-
Flavio Suligoi authored
Fix typo: "TRIGER" --> "TRIGGER" Signed-off-by:
Flavio Suligoi <f.suligoi@asem.it> Signed-off-by:
-
Alexander A. Klimov authored
PVRUSB2 VIDEO4LINUX DRIVER: Replace HTTP links with HTTPS ones Rationale: Reduces attack surface on kernel devs opening the links for MITM as HTTPS traffic is much harder to manipulate. Deterministic algorithm: For each file: If not .svg: For each line: If doesn't contain `\bxmlns\b`: If neither `\bgnu\.org/license`, nor `\bmozilla\.org/MPL\b`: If both the HTTP and HTTPS versions return 200 OK and serve the same content: Replace HTTP with HTTPS. Signed-off-by:Alexander A. Klimov <grandmaster@al2klimov.de> Signed-off-by:
-
Alexander A. Klimov authored
GSPCA USB WEBCAM DRIVER: Replace HTTP links with HTTPS ones Rationale: Reduces attack surface on kernel devs opening the links for MITM as HTTPS traffic is much harder to manipulate. Deterministic algorithm: For each file: If not .svg: For each line: If doesn't contain `\bxmlns\b`: If neither `\bgnu\.org/license`, nor `\bmozilla\.org/MPL\b`: If both the HTTP and HTTPS versions return 200 OK and serve the same content: Replace HTTP with HTTPS. Signed-off-by: -
Eugen Hristev authored
The component offsets were computed in a negative way: they were subtracted from the actual color component value. So, a higher offset was reducing the component value. This is not really desirable, as the offset is a 2's complements number with 1 bit for sign and 12 value bits, so we would like to be able to also add to the component, not only subtract. The reported number in v4l2 is fine, a range from -4095 to +4095. However when configuring a negative value for the offset, this would in fact not function, because with the old code, the number was subtracted from the max value. By setting something negative, it was overflowing in fact. Reworked the component offsets by placing the real value as the v4l2 ctrls. Now, the values are the real number that is added or subtracted from the component. The negative values received from v4l2 are already in 2's complements, so there is no need for conversion. This actually simplifies a lot the computation procedure, eliminating the need for the macros that convert from v4l2 values to ISC values and viceversa. Also the ZERO_VAL is eliminated, as 0 is now 0, as it's supposed to be. Example after this change: # v4l2-ctl --set-ctrl=red_component_offset=-150 -L 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=16 value=20 flags=slider white_balance_automatic 0x0098090c (bool) : default=1 value=0 flags=update do_white_balance 0x0098090d (button) : flags=inactive, write-only, execute-on-write gamma 0x00980910 (int) : min=0 max=3 step=1 default=3 value=3 flags=slider red_component_gain 0x009819c0 (int) : min=0 max=8191 step=1 default=512 value=512 flags=slider blue_component_gain 0x009819c1 (int) : min=0 max=8191 step=1 default=512 value=512 flags=slider green_red_component_gain 0x009819c2 (int) : min=0 max=8191 step=1 default=512 value=512 flags=slider green_blue_component_gain 0x009819c3 (int) : min=0 max=8191 step=1 default=512 value=512 flags=slider red_component_offset 0x009819c4 (int) : min=-4095 max=4095 step=1 default=0 value=-150 flags=slider blue_component_offset 0x009819c5 (int) : min=-4095 max=4095 step=1 default=0 value=0 flags=slider green_red_component_offset 0x009819c6 (int) : min=-4095 max=4095 step=1 default=0 value=0 flags=slider green_blue_component_offset 0x009819c7 (int) : min=-4095 max=4095 step=1 default=0 value=0 flags=slider The auto white balance algorithm is unchanged, but the obtained value to 'subtract' is now converted to negative and saved as a v4l2 control and displayed properly. Signed-off-by:Eugen Hristev <eugen.hristev@microchip.com> Signed-off-by:
-
Alexander A. Klimov authored
TI DAVINCI SERIES MEDIA DRIVER: Replace HTTP links with HTTPS ones Rationale: Reduces attack surface on kernel devs opening the links for MITM as HTTPS traffic is much harder to manipulate. Deterministic algorithm: For each file: If not .svg: For each line: If doesn't contain `\bxmlns\b`: If neither `\bgnu\.org/license`, nor `\bmozilla\.org/MPL\b`: If both the HTTP and HTTPS versions return 200 OK and serve the same content: Replace HTTP with HTTPS. Signed-off-by:Kieran Bingham <kieran.bingham@ideasonboard.com> Reviewed-by:
Lad Prabhakar <prabhakar.csengg@gmail.com> Signed-off-by:
-
