> cat /etc/fstab | grep floppy /floppy /floppy supermount fs=vfat,debug,dev=/dev/fd0,--,users,async,noatime 0 0 I insert a floppy freshly formatted by mtools. > df -h | grep floppy Filesystem Size Used Avail Use% Mounted on /floppy 1.4M 36K 1.4M 3% /floppy > cp * /floppy/ > ls -la /floppy total 795 drwxrwxrwx 2 root root 0 Nov 4 13:55 . drwxr-xr-x 28 root root 4096 Nov 4 10:44 .. -rwxr-xr-x 1 root root 631 Nov 4 14:21 doit -rwxr-xr-x 1 root root 808057 Nov 4 14:21 rescue.tar.gz > rm /floppy/* > ls -la /floppy total 4 drwxrwxrwx 2 root root 0 Nov 4 13:55 . drwxr-xr-x 28 root root 4096 Nov 4 10:44 .. No files are shown in the floppy directory listing but the occupied space is not freed up. > df -h | grep floppy /floppy 1.4M 827K 597K 59% /floppy To fix this, I have to do the following : > umount /floppy > dosfsck -vr /dev/fd0 dosfsck 2.8 (28 Feb 2001) dosfsck 2.8, 28 Feb 2001, FAT32, LFN Boot sector contents: System ID "MTOOL398" Media byte 0xf0 (5.25" or 3.5" HD floppy) 512 bytes per logical sector 512 bytes per cluster 1 reserved sector First FAT starts at byte 512 (sector 1) 2 FATs, 12 bit entries 4608 bytes per FAT (= 9 sectors) Root directory starts at byte 9728 (sector 19) 224 root directory entries Data area starts at byte 16896 (sector 33) 2847 data clusters (1457664 bytes) 18 sectors/track, 2 heads 0 hidden sectors 2880 sectors total Checking for unused clusters. Reclaimed 1581 unused clusters (809472 bytes). Perform changes ? (y/n) y /dev/fd0: 0 files, 72/2847 clusters AT THIS POINT, I TAKE THE FLOPPY OUT > mount /floppy > ls /floppy ls: /floppy: Input/output error AT THIS POINT, I PUT THE FLOPPY BACK INSIDE > ls -la /floppy total 4 drwxrwxrwx 1 root root 0 Nov 4 14:25 . drwxr-xr-x 28 root root 4096 Nov 4 10:44 .. > df -h | grep floppy /floppy 1.4M 36K 1.4M 3% /floppy Brandon, should I send you the debug output as well ? Aniruddha Shankar Bangalore, India
if this localized to the lolo sources or does this happen with a vanilla 2.4.20 kernel as well ?
the vanilla sources don't have supermount. From my experience, this is a problem with the supermount code as the supermount implementation in 2.4.18-6mdk suffers from the same deficiency. Juan Quintela's the maintainer of the supermount patch and the maintainer of the mandrake kernel. I've posted to the cooker mailing list but I got no response. Since this error does appear on a Gentoo ebuild, I thought that the more responsive Gentoo bugfix procedure might help produce a patch that could be transmitted upstream. Aniruddha Shankar Bangalore, India
test in lolo-sources-2.4.20.1_pre8 as of _pre5 a whole new kernel base is being used. Thanks.
problem persists on 2.4.20-lolo-r1_pre8 . should I file a new bugreport ? Aniruddha Shankar Bangalore, India
gurgle this must be a problem with supermount itself then... *twitch* hope they update the patch, because I'm not familiar with the code.
Should I file this as a bug with the supermount maintainer or will you ? Aniruddha Shankar Bangalore, India
problem persists in linux-2.4.20-gentoo-r1 linux-2.4.20-lolo-r1 linux-2.4.21-pre2-jp15 <== on this kernel, I get funny dmesg messages and the floppy does not show up when I do df -h dmesg output follows ===> floppy0: disk absent or changed during operation end_request: I/O error, cmd 1 dev 02:00 (floppy), sector 2879 VFS: Busy inodes after unmount. Self-destruct in 5 seconds. Have a nice day... VFS: revalidate called on non-open device. note: df[1626] exited with preempt_count -1 VFS: Busy inodes after unmount. Self-destruct in 5 seconds. Have a nice day... VFS: Busy inodes after unmount. Self-destruct in 5 seconds. Have a nice day... note: rm[1637] exited with preempt_count -1 note: rm[1641] exited with preempt_count -1 VFS: Busy inodes after unmount. Self-destruct in 5 seconds. Have a nice day... VFS: Busy inodes after unmount. Self-destruct in 5 seconds. Have a nice day...
Created attachment 7281 [details] kernel configuration for 2.4.21-pre2-jp15 I thought that since lolo-sources is based on the -jp kernel I'd send you the kernel config I used. Will try in a few days without preempt and lowlat - maybe that is the problem. ..... need sleep........ uurggh .
This bug fixed in latest cooker kernels by Juan Quintela. I have tested the fixed implementation and confirm that this is so. Please incorporate into lolo sources / gentoo sources. Aniruddha Shankar Bangalore, India
Im sure you already know this, but a new supermount version released on the 24th of january is available from http://people.mandrakesoft.com/~quintela/supermount/2.4.20/ Would be nice if this gets into a new gentoo-sources-2.4.20-r2 release. Loke
i wonder if this bug is related to to http://bugs.gentoo.org/show_bug.cgi?id=15422
fixed in latest lolo-sources thanks to -ck and the supermount folks. THanks for following this up so well.
would be nice if someone could attach a patch for supermount.c for 2.4.20-gentoo kernel