Re: cvs commit: src/sys/kern vfs_mount.c src/sys/vfs/ufs ffs_inode.c src/sys/vfs/msdosfs msdosfs_denode.c msdosfs_vfsops.c

[Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx>]

    I would appreciate it if people could test pulling out a USB memory
    stick with the mount still active, and even test with you CD'd into
    the mount.

    I am not prepared to say that it wouldn't crash the OS if you did it
    while the OS was in the middle of doing heavy activity, but it should
    be a lot more robust now then it was before.  Nothing should lockup,
    and it should be possible to unmount the filesystem with the memory
    stick pulled out (though of course you still need to CD out of the
    mount to unmount properly).

    --

    Pulling out a memory stick and putting it back in will not restore
    the mount, because the CAM device is still in-use by the old mount
    so CAM assigns a new disk device (e.g. da1) if you don't clear the
    mount before plugging it back in.

    But if someone *WANTS* to make it restore the mount and we can
    properly detect that the stick being inserted is the same one
    that was pulled (e.g. by looking at the serial number if there is one),
    then there is no reason why it couldn't reattach to the same SCSI disk
    device number.  Of course, even in the best of cases if we were to
    implement such a feature it would probably only work really well
    with read-only mounts.

					-Matt