On 08/25/2017 05:35 PM, wm4 wrote:
+ + dirp = opendir("/dev"); + if (!dirp) + return AVERROR(errno); + + for (entry = readdir(dirp); entry; entry = readdir(dirp)) { + + if (strncmp(entry->d_name, "video", 5)) + continue; + + snprintf(node, sizeof(node), "/dev/%s", entry->d_name); + + av_log(log_ctx, AV_LOG_DEBUG, "probing device %s\n", node); + + s->devname = node; + ret = probe_v4l2_driver(s, log_ctx); + if (!ret) + break; + }This doesn't really look like a good idea. Even somehow trying to enumerate stuff in/sys/class/ sounds better. Just because device filename starts with "video" it doesn't need to be relevant, and poking random ioctl()s at arbitrary devices doesn't sound very reliable.
yes it might seem that way but this is pretty much the standard way of opening v4l2 devices (I am not reimplementing the API to my understanding). for an example already present in ffmpeg check libavdevice/v4l2.c v4l2_get_device_list.
_______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
