bug#310: loopback mounting... b0rkt.

  I have determined for myself that loopback mouting appears to flunk 
  regardless of type of object I hope to loop-mount.  In other words the
  bug is confirmed.

If there is anyone among -devel who has ever succeeded at doing any
loopback mounts on any lnx-bbc pre flavor could you please add some
notes about it to bug #310.  Try to be as specific as possible. Thanks.

  My most notable test for this was to fetch one of the stack of .144
  floppy images I keep at my workstation*, then attempting loopmount.
	* in /usr/local/floppy, of course.
  The image was tested as successfully loopmounting on my normal kernel.

  The first time I attempted it I got a complaint about too many
  loop devices in use.  However, no amount of losetup would tell me what 
  was alleged to be using it (the "info" style of commandline said
  nothing), nor free the loops such that they were happy.  In fact 
  losetup keeps getting complaints leading with

	ioctl: LOOP_CLR_FD: (error message here)

  either "invalid argument" (when trying to mount) or 
  "no such device or address" (when trying to free loops for mount
  usage).

  I even tried the expanded mount commands where you specify which 
  loopdev you want to use.  No dice.

  Hitting up google for this and adding 2.4.18 to the keywords for good
  measure, finds threads from the crypto-linux guys mumphing that not 
  only is the mainline kernel loopmount a mess, but that submitting their
  patches (stripped of the crypto portion) had so far fallen on deaf 
  mailboxes.  The closest they'd gotten to mainstream as of that thread
  (http://mail.nl.linux.org/linux-crypto/2002-04/msg00029.html) was that 
  Andrea Arcangeli had merged them into 2.4.19pre7aa1.

  I *know* that at least for some circumstances 2.4.18 can loopback
  mount ... err, I take that back.  My trusty and dependable kernel
  of that vintage is actually a 2.4.18pre21.  My current trusty is a
  2.4.19pre10 and apparently just misses a nasty ext3 bug getting
  introduced, itself.  *sigh*

  I do find myself wondering if devfsd is mucking with things.  I have
  not yet tried loopmounting from a Knoppix disc, which also uses
  devfsd.

  If this worked, we should *merely* have the classic read-only-media 
  problem about /etc/mtab and /proc/mounts, defeatable with careful
  use of losetup.

  I have a dim recollection of loopback being visibly horked for some 
  portion of time, but I think that was around 2.4.14 or so.
  
  . | .   Heather Stern                  |         [email protected]
--->*<--- Starshine Technical Services - * - [email protected]
  ' | `   Sysadmin Support and Training  |        (800) 938-4078