Minor bug in 2.0.35 VFAT fs

John Campbell (jcampbel@lynn.ci-n.com)
Thu, 30 Jul 1998 16:28:28 -0400 (EDT)

I've just discovered that if I'm on a vfat partition and I issue the
command "cd ...", it accepts it, despite the fact that there is no directory
named "...". After this command, the current directory appears in the bash
prompt and to pwd as if there was a directory named "..." (i.e. if I'm in
"/win95/mps/civ2" and I "cd ...", it says I'm in "/win95/mps/civ2/..."), but
from ls output and such, it looks like the directory I'm actually in is the
".." directory ("/win95/mps", in the above example). The exception is if I'm
in the root directory of the vfat partition ("/win95") and I issue the
command, in which case I end up right where I started, and it still shows up
with the "..." (still in "/win95", but it shows up as "/win95/...").

This behaviour only exists on vfat, not on msdos, smbfs, hpfs, ext2,
proc, nfs, or ncpfs (though that last actually does support "cd ..." from
DOS clients, but not Linux clients, as a legitimate feature; "cd ..."
changes to the parent's parent, "cd ...." changes to the parent's parent's
parent, and so on 'til you run out of parents). The machine is 5 miles away
and hasn't got a CD in its drive at the moment, so I can't test iso9660.

Both bash and tcsh behave pretty much the same way (pwd can't figure
out the current directory at all in tcsh, but the $PWD environment variable
contains the "/win95/mps/civ2/..." directory), so I assume this is a
filesystem bug.

The machine is a 2.0.35 box with a non-FAT32 VFAT partition created
by Win95 OSR2 (I think... I'm not certain about the Win95 release number).
I'm not equipped at the moment to check if the behaviour's the same in other
kernel versions - 2.1 kernels, or the ones before FAT32 was added to 2.0,
for example.

John Campbell

QotD: Live fast, die young, leave a heavily modified corpse.

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html