Date: Mon, 8 Jan 1996 12:48:48 -0700 (MST) From: Terry Lambert <terry@lambert.org> To: joerg_wunsch@uriah.heep.sax.de Cc: freebsd-hackers@FreeBSD.org, root@synthcom.com Subject: Re: DOS File system fixes Message-ID: <199601081948.MAA10201@phaeton.artisoft.com> In-Reply-To: <199601071046.LAA07666@uriah.heep.sax.de> from "J Wunsch" at Jan 7, 96 11:46:30 am
next in thread | previous in thread | raw e-mail | index | archive | help
> > >What is currently broken in the DOS file system for FreeBSD? I see there's an > > > > Deadlocks are possible for file lookup (locking is too strong/wrong). > > Locking in some other places is too weak. rename() sometimes > > (deterministically) updates the wrong directory entries. I/O is > > extremely inefficient, especially if the cluster size is small. > > There are several reports that enabling DOS f/s read/write can damage > other file systems. It seems that if you use FIPS or PRESIZE or "Partition Magic" to resize your partition, the cluster size will not be reduced. This results in crap being written out to the "phantom" locations as if it were a bigger disk. The problem is in the disk size determination and the cluster handling code. There is some indication that mark-for-update occurs on read-only devices, and that for MSDOSFS, this update actually takes place. In combination, this means that you can trash your BSD partition (the area usually immediately follows the MSDOSFS area on the disk, which is where you got space for it in the first place) when mounted read-only. Part of the problem is that code common to all file systems is duplicated in each file system instances instead of being done (correctly) once in the upper level code. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199601081948.MAA10201>