kernel.git/include/media/v4l2-ioctl.h, branch linux-rolling-stable

media: v4l2-ioctl: Stop passing fh pointer to ioctl handlers

2025-08-13T06:33:56Z

Now that all drivers access the v4l2_fh from the file structure, there is no need to pass it as an explicit argument to ioctl handlers. Set the argument to NULL in the __video_do_ioctl() function, and rename the 'fh' argument in the ioctl handler declarations to 'priv' indicate it does not contain a file handle. The argument could be removed altogether with a mechanical change (probably using coccinelle), but there are plans to pass a new argument to the ioctl handlers in the near future. The tree-wide change to remove the argument, only to add another one soon after, would be too much churn. While at it, fix argument alignment in vidioc_try_fmt_vid_out_overlay(). Signed-off-by: Laurent Pinchart Signed-off-by: Hans Verkuil

media: core: export v4l2_translate_cmd

2025-07-11T17:27:28Z

video_translate_cmd() can be useful for drivers to convert between the VIDIOC_*32 and VIDIOC_ defines. Let's export it. Now that the function is exported, use this opportunity to rename the function with the v4l2_ prefix, that is less ambiguous than video_ The VIDIOC_*32 defines are not accessible by the drivers, they live in v4l2-compat-ioctl32.c. Reviewed-by: Hans Verkuil Signed-off-by: Ricardo Ribalda Link: https://lore.kernel.org/r/20250701-uvc-grannular-invert-v4-6-8003b9b89f68@chromium.org Signed-off-by: Hans de Goede Signed-off-by: Hans Verkuil

media: v4l2: Remove vidioc_s_ctrl callback

2025-03-05T09:43:26Z

All the drivers either use the control framework or provide a vidiod_ext_ctrl. We can remove this callback. Signed-off-by: Ricardo Ribalda Signed-off-by: Hans Verkuil Signed-off-by: Mauro Carvalho Chehab

media: v4l2: Remove vidioc_g_ctrl callback

2025-03-05T09:43:26Z

All the drivers either use the control framework or provide a vidioc_g_ext_ctrls callback. We can remove this callback. Thanks for your service! Signed-off-by: Ricardo Ribalda Signed-off-by: Hans Verkuil Signed-off-by: Mauro Carvalho Chehab

media: v4l2: Remove vidioc_queryctrl callback

2025-03-05T09:43:25Z

All the drivers either use the control framework or provide a vidioc_query_ext_ctrl. We can remove this callback to reduce the temptation of new drivers to implement it. Signed-off-by: Ricardo Ribalda Signed-off-by: Hans Verkuil Signed-off-by: Mauro Carvalho Chehab

media: v4l2: Add REMOVE_BUFS ioctl

2024-03-25T11:00:44Z

VIDIOC_REMOVE_BUFS ioctl allows to remove buffers from a queue. The number of buffers to remove in given by count field of struct v4l2_remove_buffers and the range start at the index specified in the same structure. Signed-off-by: Benjamin Gaignard Reviewed-by: Mauro Carvalho Chehab Signed-off-by: Hans Verkuil [hverkuil: vidioc-remove-bufs.rst: mention no bufs are freed on error]

media: v4l2: prepare compat-ioctl rework

2020-11-16T09:31:05Z

The v4l2-compat-ioctl32() currently takes an extra round trip through user space pointers when converting the data structure formats. In particular, this involves using the compat_alloc_user_space() and copy_in_user() helpers that often lead to worse compat handlers compared to using in_compat_syscall() checks when copying the data. The native implementation already gained a simpler method to deal with the conversion for the time32 conversion. Hook into the same places to provide a location for reading and writing user space data from inside of the generic video_usercopy() helper. Hans Verkuil rewrote the video_get_user() function here to simplify the zeroing of the extra input fields and fixed a couple of bugs in the original implementation. [hverkuil: fix: CHECK: Please don't use multiple blank lines] Co-developed-by: Hans Verkuil Signed-off-by: Hans Verkuil Signed-off-by: Arnd Bergmann Signed-off-by: Mauro Carvalho Chehab

media: v4l2-core: fix v4l2_buffer handling for time64 ABI

2020-01-03T14:50:21Z

The v4l2_buffer structure contains a 'struct timeval' member that is defined by the user space C library, creating an ABI incompatibility when that gets updated to a 64-bit time_t. As in v4l2_event, handle this with a special case in video_put_user() and video_get_user() to replace the memcpy there. Since the structure also contains a pointer, there are now two native versions (on 32-bit systems) as well as two compat versions (on 64-bit systems), which unfortunately complicates the compat handler quite a bit. Duplicating the existing handlers for the new types is a safe conversion for now, but unfortunately this may turn into a maintenance burden later. A larger-scale rework of the compat code might be a better alternative, but is out of scope of the y2038 work. Sparc64 needs a special case because of their special suseconds_t definition. Signed-off-by: Arnd Bergmann Signed-off-by: Hans Verkuil Signed-off-by: Mauro Carvalho Chehab

media: v4l2-core: fix VIDIOC_DQEVENT for time64 ABI

2020-01-03T14:47:57Z

The v4l2_event structure contains a 'struct timespec' member that is defined by the user space C library, creating an ABI incompatibility when that gets updated to a 64-bit time_t. While passing a 32-bit time_t here would be sufficient for CLOCK_MONOTONIC timestamps, simply redefining the structure to use the kernel's __kernel_old_timespec would not work for any library that uses a copy of the linux/videodev2.h header file rather than including the copy from the latest kernel headers. This means the kernel has to be changed to handle both versions of the structure layout on a 32-bit architecture. The easiest way to do this is during the copy from/to user space. Signed-off-by: Arnd Bergmann Signed-off-by: Hans Verkuil Signed-off-by: Mauro Carvalho Chehab

media: v4l: ctrls: Add debug messages

2019-07-25T10:26:49Z

Currently, the v4l2 control code is a bit silent on errors. Add debug messages on (hopefully) most of the error paths. Signed-off-by: Ezequiel Garcia Signed-off-by: Hans Verkuil Signed-off-by: Mauro Carvalho Chehab