Hi, earlier I mentioned here I'm working on the ability for filesystems like FUSE based, cifs and nfs. Recently I've created a patch for the kernel, which makes the calls inotify_add_watch and inotify_remove_watch send a message to userspace, containing (pid, fd, wd, [mask], [path]). (in case of add watch the values between [] are sent also, when remove not) My idea for a FUSE based fs is to listen to this messages (familiy GENERIC, group VFS_INOTIFY), filter messages out which do not apply (only those messages for the used mountpoint for fs) and then let the FUSE fs take proper action, which suits the backend. In case of a simple overlay fs, this means that it sets watch on the underlying fs, and send change events (which fits the mask) to the kernel. Recent FUSE versions have calls added which can notify the kernel about a change, without the need of a related request, like getattr, open, lookup etc. The fs has to remove this/stop this when a remove watch message is received. A proper adiminstration of current watches set is required. Can this also work for cifs?? I always thought that cifs is just like FUSE in userspace, but that isn't the case is it? Stef -- To unsubscribe from this list: send the line "unsubscribe linux-cifs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
