Supermount README
=================

supermount v0.6 for linux-2.0.32,  released 24 November 1997
Stephen Tweedie,  [email protected]

NEWS
----

* New: support for writable ext2fs media

* New: automatic CDRom door lock/unlock (IDE and SCSI drives only)

* Fixed initialisation of mount point inode data

* Fixed several races during unmount/remount code


What is it?
-----------

Supermount is a pseudo-filesystem which manages filesystems on
removable media like floppy disks and CD-ROMs.  It aims to make
management of removable media as easy as it is under DOS.

With supermount, you can change the disk in the drive whenever you
want (with the obvious exception that you shouldn't do it when the
filesystem is actively in use).  You don't need to "cd" out of the
directory first, and you don't need to tell the kernel what you're
doing --- supermount will detect the media change automatically.

Supermount will automatically detect whether the media you are
mounting is read-write or readonly, and if you mount a write-protected
disk, then the subfs will be mounted as a readonly filesystem.

Supermount detects when you have finished activity on the subfs, and
will flush all buffers to the disk before completing the operation.
So, if you copy a file onto a supermounted floppy disk, the data will
all be written to disk before the "cp" command finishes.  When the
command does complete, it will be safe to remove the disk.

It is worth while defining what I mean by "activity" here.  The subfs
is active if there are any processes running which have a handle on a
non-directory inode on the subfs, or which have a file open on the
subfs (even if only for reading).  There is one important case which
does NOT count as activity: if you "cd" to a directory or a
subdirectory under the supermount mount point, then that reference
does not make the subfs active, and you can safely remove the disk.

Yes, that's right.  You can "cd /floppy; ls" and get a listing of a
dos floppy.  Remove the disk, insert a new one, and "ls" will now list
the new contents.  Remove the disk altogether, and "ls" will give you
an I/O error.  Put in a new disk, and "ls" starts working again.  It
is NOT an error to remove a supermounted disk which is acting as the
current working directory for one or more processes!


The Superfilesystem and Subfilesystem concepts
----------------------------------------------

Normally, when you mount a filesystem, you create a direct connection
between a mount point in the directory tree and a filesystem on a
block device.  Supermount adds an extra layer in between; with
supermount, you explicitly mount the pseudo-filesystem (the
"superfilesystem") on the mount point, and supermount then
automatically mounts the real filesystem (the "subfilesystem") on the
block device when needed.

Running supermount
------------------

To run supermount, compile and install a kernel with the supermount
patches and select "Y" to the question

       Supermount removable media support (CONFIG_SUPERMOUNT) [Y/n/?]

when you run "make config".  You set up a supermount filesystem with
the normal mount command, using the syntax:

       mount -o <superfs-options>,--,<subfs-options> <mpt> <mpt>

where

       <superfs-options> are the options you want to pass to supermount
       itself.  These are described below.

       <subfs-options> are the options you want supermount to pass to the
       dismountable filesystem underneath.

       <mpt> is the mount point where you want your removable media to be
       mounted.

Notice that you do not directly specify the block device you are going
to mount on the mount command line.  This is because the supermount
filesystem is NOT connected to a block device; rather, supermount is
responsible for connecting a separate filesystem to the block device.
You specify the sub-filesystem and block device name by providing the
<superfs-options> field, where the following options are currently
recognised:

* subfs=<filesystem-type>               [default is "msdos"]

       Specify the subfilesystem type.  "msdos", "vfat", "ext2" and
"iso9660" have been tested, but any block-device filesystem should
work.

* dev=<block-device>                    [default is "/dev/fd0"]

       Specify the block device on which the subfs is to be mounted.

* debug

       Enable debugging code in the supermount filesystem, if
the debug option was enabled at compile time.  By default, debugging
code is compiled into the kernel but is disabled until a debug mount
option is seen.

* '--'

       All options after the option string '--' will be passed
directly to the subfilesystem when it gets mounted.


Here is an example of supermount options, taken directly out of my
current /etc/fstab:

/mnt/floppy     /mnt/floppy     supermount      fs=vfat,--,gid=51,conv=binary   0 0
/cd              /cd            supermount      ro,fs=iso9660,dev=/dev/hdd,--,conv=binary 0 0

This tells supermount to manage a vfat (msdos with long filename)
filesystem on /dev/fd0 mounted at /mnt/floppy (/dev/fd0 is the default
block device so does not need to be specified), with the vfat
filesystem getting the options "gid=51,conv=binary".  My cdrom on
/dev/hdd is similarly mounted on /cd.


Enjoy supermount.  I hope you find it useful --- I certainly find it
extremely convenient.  Send any comments, bug-fixes or bug-reports,
suggestions and success stories to [email protected].  Flames to
/dev/null, please!

Cheers,
Stephen.
--
Stephen Tweedie <[email protected]>


================================================================
Changes for v0.6:
Complete redesign of the shadow inode management.  Should be much more
reliable now.

Keep the subfilesystem mounted readonly by default.  Add support to
remount it read-write on demand, and back to readonly when
appropriate, so that filesystems such as ext2fs get a chance to update
the filesystem integrity data while the media is still present.

Added code to ext2fs to prevent it from incrementing the mount counts
in the superblock multiple times if the same filesystem is repeatedly
remounted readwrite then readonly without being unmounted.

Add a new VFS entry point into the block device system to allow the
kernel to forcibly lock and unlock a removable media block device.
Implemented in the IDE and SCSI cdrom drivers for now.

Changes for v0.5:
<Internal non-released version>

Changes for v0.4:
Performance tuning only.  Read-only operations like "find" should now
be MUCH faster.

Changes for v0.3:
Fixed supermount_create bug; now returns a properly attached
superinode.

Changes for v0.2:
Improved device invalidation code, so CD-ROMs work now.

You are viewing proxied material from ftp.icm.edu.pl. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.