- 19 Jul, 2020 37 commits
-
-
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 <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
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:
-
Alexander A. Klimov authored
Replace HTTP links with HTTPS ones: SOFTLOGIC 6x10 MPEG CODEC 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:Ismael Luceno <ismael@iodev.co.uk> Signed-off-by:
-
Randy Dunlap authored
Drop duplicated words in Documentation/userspace-api/media/. This addresses the words "struct" and "value". Signed-off-by:
Randy Dunlap <rdunlap@infradead.org> Cc: Jonathan Corbet <corbet@lwn.net> Signed-off-by:
-
Hans Verkuil authored
This driver didn't call _vb2_fop_release in the sun4i_csi_release function. Without that call the vb2 queue is not properly canceled and buffers may not be freed. Signed-off-by:
Maxime Ripard <mripard@kernel.org> Signed-off-by:
-
Hans Verkuil authored
For no obvious reason VB2_DMABUF was not enabled. Add this. Also both read and write file ops were set, but this was not enabled in the vb2 io_modes or in the device_caps capabilities. Drop this. Signed-off-by:
-
Hans Verkuil authored
This fixes a v4l2-compliance warning: Required ioctls: warn: v4l2-test-media.cpp(52): empty bus_info test MEDIA_IOC_DEVICE_INFO: OK Signed-off-by: -
Colin Ian King authored
The variable status is being initialized with a value that is never read and it is being updated later with a new value. The initialization is redundant and can be removed. Addresses-Coverity: ("Unused value") Signed-off-by:Colin Ian King <colin.king@canonical.com> Signed-off-by:
-
Ezequiel Garcia authored
While debugging, it's currently really hard to identify controls by their ID. Print the control name making the print more helpful. With this change, the print changes from: video1: VIDIOC_S_EXT_CTRLS: which=0xf010000, count=5, error_idx=4, request_fd=45, id/size=0x990ce8/1048, id/size=0x990ce9/12, id/size=0x990cea/480, id/size=0x990ceb/896, id/size=0x990cec/400 video1: VIDIOC_S_EXT_CTRLS: which=0xf010000, count=5, error_idx=4, request_fd=42, name=H264 Sequence Parameter Set, id/size=0x990ce8/1048, name=H264 Picture Parameter Set, id/size=0x990ce9/12, name=H264 Scaling Matrix, id/size=0x990cea/480, name=H264 Slice Parameters, id/size=0x990ceb/896, name=H264 Decode Parameters, id/size=0x990cec/400 For instance, this is specially helpful when the ioctl fails. Consider the following example: v4l2-ctrls: prepare_ext_ctrls: video1: pointer control id 0x990cec size too small, 400 bytes but 784 bytes needed v4l2-ctrls: try_set_ext_ctrls: video1: video1: try_set_ext_ctrls_common failed (-14) video1: VIDIOC_S_EXT_CTRLS: error -14: which=0xf010000, count=5, error_idx=5, request_fd=39, name=H264 Sequence Parameter Set, id/size=0x990ce8/1048, name=H264 Picture Parameter Set, id/size=0x990ce9/12, name=H264 Scaling Matrix, id/size=0x990cea/480, name=H264 Slice Parameters, id/size=0x990ceb/896, name=H264 Decode Parameters, id/size=0x990cec/400 Signed-off-by:
-
Hans Verkuil authored
Typo in comment block: v4l2_subdev_get_try_crop -> v4l2_subdev_get_try_compose Signed-off-by:
-
Hans Verkuil authored
Fix this smatch warning: drivers/media/cec/core/cec-api.c:156 cec_adap_g_log_addrs() warn: check that 'log_addrs' doesn't leak information (struct has a hole after 'features') Signed-off-by:
-
Oliver Neukum authored
A malicious USB device may lack endpoints the driver assumes to exist Accessing them leads to NULL pointer accesses. This patch introduces sanity checking. Reported-and-tested-by: syzbot+cabfa4b5b05ff6be4ef0@syzkaller.appspotmail.com Signed-off-by:
Oliver Neukum <oneukum@suse.com> Fixes: 866b8695 ("Staging: add the go7007 video driver") Signed-off-by:
-
Dmitry Osipenko authored
VDE partition is left turned ON after bootloader on most devices, hence let's ensure that it's turned OFF in order to lower power leakage while hardware is idling by turning it ON and OFF during of the driver's probe. Signed-off-by:
Dmitry Osipenko <digetx@gmail.com> Signed-off-by:
-
Dmitry Osipenko authored
On some devices bootloader isn't ready to a clamped VDE power, and thus, machine hangs on a warm reboot (CPU reset). The VDE power partition is turned ON by default on a cold boot, hence VDE driver should keep power partition enabled on system's reboot too. This fixes hang on a warm reboot on a Tegra20 Acer A500 device, which is handy if Embedded Controller driver is unavailable, i.e. cold reboot can't be performed. Signed-off-by:
Thierry Reding <treding@nvidia.com> Signed-off-by:
-
Dmitry Osipenko authored
Runtime PM is always available on Tegra nowadays since commit 40b2bb1b ("ARM: tegra: enforce PM requirement"), hence the case of unavailable RPM doesn't need to be handled. Signed-off-by:
-
- 04 Jul, 2020 3 commits
-
-
Dariusz Marcinkiewicz authored
Do not fail probing when device_init_wakeup fails. device_init_wakeup fails when the device is already enabled as wakeup device. Hence, the driver fails to probe the device if: - The device has already been enabled for wakeup (by e.g. sysfs) - The driver has been unloaded and is being loaded again. This goal of the patch is to fix the above cases. Overwhelming majority of the drivers do not check device_init_wakeup return code. Fixes: cd70de2d ("media: platform: Add ChromeOS EC CEC driver") Signed-off-by:
Dariusz Marcinkiewicz <darekm@google.com> Signed-off-by:
-
Paul Kocialkowski authored
Setting the output CSC mode is required for a YUV output, but must not be set when the input is also YUV. Doing this (as tested with a YUV420P to YUV420P conversion) results in wrong colors. Adapt the logic to only set the output CSC mode when the output is YUV and the input is RGB. Also add a comment to clarify the rationale. Fixes: f7e7b48e ("[media] rockchip/rga: v4l2 m2m support") Signed-off-by:
Paul Kocialkowski <paul.kocialkowski@bootlin.com> Reviewed-by:
-
Paul Kocialkowski authored
This introduces two macros: RGA_COLOR_FMT_IS_YUV and RGA_COLOR_FMT_IS_RGB which allow quick checking of the colorspace familily of a RGA color format. These macros are then used to refactor the logic for CSC mode selection. The two nested tests for input colorspace are simplified into a single one, with a logical and, making the whole more readable. Signed-off-by:
-
