libmediactl.c: add poor man's udev support
PATCH 1/1 libmediactl.c: add poor man's udev support
If libudev is not available (android!), then just try to find the device in /dev. It's a poor man's solution, but it is better than nothing.
utils/media-ctl/libmediactl.c +33 -4
@@ -441,7 +441,10 @@ static int media_get_devname_udev(struct udev *udev, return ret; } -#else /* HAVE_LIBUDEV */ +#else + +#include <dirent.h> +#include <sys/stat.h> struct udev ;
@@ -449,10 +452,36 @@ static inline int media_udev_open(struct udev **udev) { return 0; }static inline void media_udev_close (struct udev *udev) { } -static inline int media_get_devname_udev (struct udev *udev, -struct media_entity *entity) +static int media_get_devname_udev (struct udev *udev, +struct media_entity *entity) { -return -ENOTSUP ; +DIR *dp; +struct dirent *ep; +dev_t devnum; + + dp =opendir ("/dev" ); +if (dp ==NULL ) { +media_dbg (entity->media,"couldn't open /dev\n" ); +return -ENODEV ; + } + devnum =makedev (entity->info.v4l.major, entity->info.v4l.minor); +while ((ep =readdir (dp))) { +struct stat st; +char fname[256 ]; + +snprintf (fname,sizeof (fname) -1 ,"/dev/%s" , ep->d_name); + fname[sizeof (fname) -1 ] =0 ; +stat (fname, &st); +if ((st.st_mode &S_IFMT ) !=S_IFCHR ) +continue ; +if (st.st_rdev == devnum) { +strncpy (entity->devname, fname,sizeof (entity->devname)); + entity->devname[sizeof (entity->devname) -1 ] ='\0' ; +return 0 ; + } + } +closedir (dp); +return -ENODEV ; }#endif /* HAVE_LIBUDEV */
Hi Hans,
Thank you for the patch.
Quoted 1 line · On Thursday 10 December 2015 11:44:43 Hans Verkuil wrote:
If libudev is not available (android!),
It's time for Android to grow up and get rid of their LGPL hatred... That might actually happen as a result of the Chrome/Android merge that seems to be planned, as well as from the Android One project that will require something like libudev (although knowing the Android folks they might decide to reimplement it).
Quoted 2 lines
then just try to find the device in /dev. It's a poor man's solution, but it is better than nothing.
Can't we assume that, on such systems, the device will always be named /dev/video* or /dev/v4l-subdev* ? If so we could lower the runtime impact by only checking the /sys/class/video4linux/video* and /sys/class/video4linux/v4l-subdev* entries to match the device minor and major, and use the index from the sysfs entry to create the /dev path.
My enumeration library (which I need to revive) implements that, feel free to use the code if it can be useful.
git://git.ideasonboard.org/media-enum.git
Quoted 60 lines
Signed-off-by Hans Verkuil <[email protected]>utils/media-ctl/libmediactl.c +33
-4@@ -441,7 +441,10 @@ static int media_get_devname_udev(struct udev *udev, return ret; } -#else /* HAVE_LIBUDEV */ +#else + +#include <dirent.h> +#include <sys/stat.h> struct udev ;@@ -449,10 +452,36 @@ static inline int media_udev_open(struct udev **udev) {return 0 ; }static inline void media_udev_close (struct udev *udev) { } -static inline int media_get_devname_udev (struct udev *udev, -struct media_entity *entity) +static int media_get_devname_udev (struct udev *udev, +struct media_entity *entity) { -return -ENOTSUP ; +DIR *dp; +struct dirent *ep; +dev_t devnum; + + dp =opendir ("/dev" ); +if (dp ==NULL ) { +media_dbg (entity->media,"couldn't open /dev\n" ); +return -ENODEV ; + } + devnum =makedev (entity->info.v4l.major, entity->info.v4l.minor); +while ((ep =readdir (dp))) { +struct stat st; +char fname[256 ]; + +snprintf (fname,sizeof (fname) -1 ,"/dev/%s" , ep->d_name); + fname[sizeof (fname) -1 ] =0 ; +stat (fname, &st); +if ((st.st_mode &S_IFMT ) !=S_IFCHR ) +continue ; +if (st.st_rdev == devnum) { +strncpy (entity->devname, fname,sizeof (entity->devname)); + entity->devname[sizeof (entity->devname) -1 ] ='\0' ; +return 0 ; + } + } +closedir (dp); +return -ENODEV ; }#endif /*
HAVE_LIBUDEV*/
Signature · 3 lines
Regards, Laurent Pinchart