From: Ali Nasrollahi <a.nasrolahi01@gmail.com>
To: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: linux-media@vger.kernel.org, linux-kernel@vger.kernel.org,
mchehab@kernel.org, Frank.Li@nxp.com, hverkuil+cisco@kernel.org,
guoniu.zhou@oss.nxp.com
Subject: Re: [PATCH] media: v4l2-core: make number of video devices configurable
Date: Tue, 11 Aug 2026 07:59:47 +0330 [thread overview]
Message-ID: <anqic68F2nWjgA1p@zephyr> (raw)
In-Reply-To: <20260811040459.GE3046075@killaraus.ideasonboard.com>
On 26/08/11 07:04AM, Laurent Pinchart wrote:
> On Tue, Aug 11, 2026 at 07:29:36AM +0330, Ali Nasrollahi wrote:
> > On 26/08/10 07:28PM, Laurent Pinchart wrote:
> > > On Mon, Aug 10, 2026 at 07:54:07PM +0330, Ali Nasrollahi wrote:
> > > > The V4L2 core currently limits the number of registered video devices to
> > > > 256. This can become a restriction for workloads that create a large
> > > > number of virtual video devices, such as v4l2loopback instances.
> > >
> > > We don't support v4l2loopback upstream though.
> >
> > Yes, you are correct.
> >
> > But perhaps my description was a bit misleading. The main idea of this
> > patch is not to support out-of-tree modules by any means, but to allow a
> > configurable number of V4L2 devices to be created and used, whether
> > through loopback devices or other means.
>
> Do you have a use case where you're reaching the limit without
> out-of-tree drivers ?
>
My workloads depend on loopback devices which is how I've reached the
limit, however I haven't had a usecase with a such number of deivces
through physical hardware.
> > > > Make the maximum number of video devices configurable through a module
> > > > parameter. The default and minimum remain at the existing limit of 256,
> > > > while the maximum is capped at 4096.
> > > >
> > > > The parameter is only available when CONFIG_VIDEO_FIXED_MINOR_RANGES is
> > > > disabled, as the fixed minor allocation scheme relies on the existing
> > > > minor ranges.
> > > >
> > > > The device array is allocated according to the configured limit, while
> > > > the allocation bitmaps remain statically sized to the maximum supported
> > > > limit of 4096.
> > > >
> > > > Signed-off-by: Ali Nasrollahi <A.Nasrolahi01@gmail.com>
> > > > ---
> > > > drivers/media/v4l2-core/v4l2-dev.c | 70 +++++++++++++++++++++++-------
> > > > 1 file changed, 55 insertions(+), 15 deletions(-)
> > > >
> > > > diff --git a/drivers/media/v4l2-core/v4l2-dev.c b/drivers/media/v4l2-core/v4l2-dev.c
> > > > index fd267fb74905..5a526b577970 100644
> > > > --- a/drivers/media/v4l2-core/v4l2-dev.c
> > > > +++ b/drivers/media/v4l2-core/v4l2-dev.c
> > > > @@ -31,7 +31,8 @@
> > > > #include <media/v4l2-ioctl.h>
> > > > #include <media/v4l2-event.h>
> > > >
> > > > -#define VIDEO_NUM_DEVICES 256
> > > > +#define VIDEO_MIN_NUM_DEVICES 256
> > > > +#define VIDEO_MAX_NUM_DEVICES 4096
> > > > #define VIDEO_NAME "video4linux"
> > > >
> > > > #define dprintk(fmt, arg...) do { \
> > > > @@ -39,6 +40,27 @@
> > > > __func__, ##arg); \
> > > > } while (0)
> > > >
> > > > +static unsigned int video_nr_devices = VIDEO_MIN_NUM_DEVICES;
> > > > +
> > > > +#ifndef CONFIG_VIDEO_FIXED_MINOR_RANGES
> > > > +static int param_set_video_nr_devices(const char *val,
> > > > + const struct kernel_param *kp)
> > > > +{
> > > > + return param_set_uint_minmax(val, kp, VIDEO_MIN_NUM_DEVICES,
> > > > + VIDEO_MAX_NUM_DEVICES);
> > > > +}
> > > > +
> > > > +static const struct kernel_param_ops param_ops_video_nr_devices = {
> > > > + .set = param_set_video_nr_devices,
> > > > + .get = param_get_uint,
> > > > +};
> > > > +
> > > > +module_param_cb(video_nr_devices, ¶m_ops_video_nr_devices,
> > > > + &video_nr_devices, 0444);
> > > > +
> > > > +MODULE_PARM_DESC(video_nr_devices, "Maximum number of V4L2 devices (256-4096)");
> > > > +#endif
> > > > +
> > > > /*
> > > > * sysfs stuff
> > > > */
> > > > @@ -98,9 +120,14 @@ static struct dentry *v4l2_debugfs_root_dir;
> > > > /*
> > > > * Active devices
> > > > */
> > > > -static struct video_device *video_devices[VIDEO_NUM_DEVICES];
> > > > +#ifdef CONFIG_VIDEO_FIXED_MINOR_RANGES
> > > > +static struct video_device *video_devices[VIDEO_MIN_NUM_DEVICES];
> > > > +static DECLARE_BITMAP(devnode_nums[VFL_TYPE_MAX], VIDEO_MIN_NUM_DEVICES);
> > > > +#else
> > > > +static struct video_device **video_devices;
> > > > +static DECLARE_BITMAP(devnode_nums[VFL_TYPE_MAX], VIDEO_MAX_NUM_DEVICES);
> > > > +#endif
> > > > static DEFINE_MUTEX(videodev_lock);
> > > > -static DECLARE_BITMAP(devnode_nums[VFL_TYPE_MAX], VIDEO_NUM_DEVICES);
> > > >
> > > > /* Device node utility functions */
> > > >
> > > > @@ -504,7 +531,7 @@ static const struct file_operations v4l2_fops = {
> > > > * in the video_device array, but it was able to obtain a minor number.
> > > > *
> > > > * This means that we can always obtain a free stream index number since
> > > > - * the worst case scenario is that there are VIDEO_NUM_DEVICES - 1 slots in
> > > > + * the worst case scenario is that there are video_nr_devices - 1 slots in
> > > > * use of the video_device array.
> > > > *
> > > > * Returns a free index number.
> > > > @@ -513,19 +540,19 @@ static int get_index(struct video_device *vdev)
> > > > {
> > > > /* This can be static since this function is called with the global
> > > > videodev_lock held. */
> > > > - static DECLARE_BITMAP(used, VIDEO_NUM_DEVICES);
> > > > + static DECLARE_BITMAP(used, VIDEO_MAX_NUM_DEVICES);
> > > > int i;
> > > >
> > > > - bitmap_zero(used, VIDEO_NUM_DEVICES);
> > > > + bitmap_zero(used, VIDEO_MAX_NUM_DEVICES);
> > > >
> > > > - for (i = 0; i < VIDEO_NUM_DEVICES; i++) {
> > > > + for (i = 0; i < video_nr_devices; i++) {
> > > > if (video_devices[i] != NULL &&
> > > > video_devices[i]->v4l2_dev == vdev->v4l2_dev) {
> > > > __set_bit(video_devices[i]->index, used);
> > > > }
> > > > }
> > > >
> > > > - return find_first_zero_bit(used, VIDEO_NUM_DEVICES);
> > > > + return find_first_zero_bit(used, video_nr_devices);
> > > > }
> > > >
> > > > #define SET_VALID_IOCTL(ops, cmd, op) \
> > > > @@ -915,7 +942,7 @@ int __video_register_device(struct video_device *vdev,
> > > > int i = 0;
> > > > int ret;
> > > > int minor_offset = 0;
> > > > - int minor_cnt = VIDEO_NUM_DEVICES;
> > > > + int minor_cnt = video_nr_devices;
> > > > const char *name_base;
> > > >
> > > > /* A minor value of -1 marks this video device as never
> > > > @@ -1020,10 +1047,10 @@ int __video_register_device(struct video_device *vdev,
> > > > #else
> > > > /* The device node number and minor numbers are independent, so
> > > > we just find the first free minor number. */
> > > > - for (i = 0; i < VIDEO_NUM_DEVICES; i++)
> > > > + for (i = 0; i < video_nr_devices; i++)
> > > > if (video_devices[i] == NULL)
> > > > break;
> > > > - if (i == VIDEO_NUM_DEVICES) {
> > > > + if (i == video_nr_devices) {
> > > > mutex_unlock(&videodev_lock);
> > > > pr_err("could not get a free minor\n");
> > > > return -ENFILE;
> > > > @@ -1032,7 +1059,7 @@ int __video_register_device(struct video_device *vdev,
> > > > vdev->minor = i + minor_offset;
> > > > vdev->num = nr;
> > > >
> > > > - if (WARN_ON(vdev->minor >= VIDEO_NUM_DEVICES)) {
> > > > + if (WARN_ON(vdev->minor >= video_nr_devices)) {
> > > > mutex_unlock(&videodev_lock);
> > > > return -EINVAL;
> > > > }
> > > > @@ -1241,7 +1268,7 @@ static int __init videodev_init(void)
> > > > int ret;
> > > >
> > > > pr_info("Linux video capture interface: v2.00\n");
> > > > - ret = register_chrdev_region(dev, VIDEO_NUM_DEVICES, VIDEO_NAME);
> > > > + ret = register_chrdev_region(dev, video_nr_devices, VIDEO_NAME);
> > > > if (ret < 0) {
> > > > pr_warn("videodev: unable to get major %d\n",
> > > > VIDEO_MAJOR);
> > > > @@ -1250,11 +1277,21 @@ static int __init videodev_init(void)
> > > >
> > > > ret = class_register(&video_class);
> > > > if (ret < 0) {
> > > > - unregister_chrdev_region(dev, VIDEO_NUM_DEVICES);
> > > > + unregister_chrdev_region(dev, video_nr_devices);
> > > > pr_warn("video_dev: class_register failed\n");
> > > > return -EIO;
> > > > }
> > > >
> > > > +#ifndef CONFIG_VIDEO_FIXED_MINOR_RANGES
> > > > + video_devices = kzalloc_objs(*video_devices, video_nr_devices);
> > > > + if (!video_devices) {
> > > > + class_unregister(&video_class);
> > > > + unregister_chrdev_region(dev, video_nr_devices);
> > > > + pr_warn("video_dev: failed to allocate video_devices\n");
> > > > + return -ENOMEM;
> > > > + }
> > > > +#endif
> > > > +
> > > > return 0;
> > > > }
> > > >
> > > > @@ -1262,8 +1299,11 @@ static void __exit videodev_exit(void)
> > > > {
> > > > dev_t dev = MKDEV(VIDEO_MAJOR, 0);
> > > >
> > > > +#ifndef CONFIG_VIDEO_FIXED_MINOR_RANGES
> > > > + kvfree(video_devices);
> > > > +#endif
> > > > class_unregister(&video_class);
> > > > - unregister_chrdev_region(dev, VIDEO_NUM_DEVICES);
> > > > + unregister_chrdev_region(dev, video_nr_devices);
> > > > debugfs_remove_recursive(v4l2_debugfs_root_dir);
> > > > v4l2_debugfs_root_dir = NULL;
> > > > }
>
> --
> Regards,
>
> Laurent Pinchart
prev parent reply other threads:[~2026-08-11 4:29 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-08-10 16:24 [PATCH] media: v4l2-core: make number of video devices configurable Ali Nasrollahi
2026-08-10 16:28 ` Laurent Pinchart
2026-08-11 3:59 ` Ali Nasrollahi
2026-08-11 4:04 ` Laurent Pinchart
2026-08-11 4:29 ` Ali Nasrollahi [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=anqic68F2nWjgA1p@zephyr \
--to=a.nasrolahi01@gmail.com \
--cc=Frank.Li@nxp.com \
--cc=guoniu.zhou@oss.nxp.com \
--cc=hverkuil+cisco@kernel.org \
--cc=laurent.pinchart@ideasonboard.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-media@vger.kernel.org \
--cc=mchehab@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox