Re: loop-aes: make tests fails: Device busy

[Jens Lechtenboerger <lechten@xxxxxxxxxxxxxxxxxx>]

On Wed, Mar 18 2009, Jens Lechtenboerger wrote:

> [...]
> However, I realized that sometimes I get the following messages in
> syslog:
> __ratelimit: 10 callbacks suppressed
> Buffer I/O error on device loop7, logical block 0
> Buffer I/O error on device loop7, logical block 1
> ...
> Buffer I/O error on device loop7, logical block 9
>
> Sometimes, there are "30" or "50" callbacks instead of "10".
> The above messages do not appear every time (although make tests
> fails every time), and I fail to see a pattern.  I'll retry later
> today.

The problem is unrelated to loop-aes.  I executed the following
script (based on make tests):

#!/bin/bash
TLD=/dev/loop1
declare -i i=0
rm -f test-file*
dd if=/dev/zero of=test-file1 bs=1024 count=33
cp test-file1 test-file3
while true
do
    echo "Iteration $((i+=1))"
    /sbin/losetup /dev/loop1 test-file1
    dd if=test-file3 of=/dev/loop1 bs=1024 count=33 conv=notrunc
    if ! /sbin/losetup -d $TLD ; then exit 1 ; fi
    /sbin/losetup /dev/loop1 test-file1
    dd if=/dev/loop1 of=test-file4 bs=33792 count=1
    if ! /sbin/losetup -d $TLD ; then exit 1 ; fi
done

This script fails randomly after several executions of losetup -d
(the first as well as the second one) with:
ioctl: LOOP_CLR_FD: Device or resource busy
Sometimes with the above messages in syslog, sometimes without.
This also happens with plain Ubuntu Jaunty Alpha 6 (no loop-aes,
unpatched losetup).  If I invoke fuser or lsof before losetup -d,
then the script runs endlessly.

I guess I'll have to file a bug report, probably after testing a
vanilla kernel...

Thanks
Jens

-
Linux-crypto:  cryptography in and on the Linux system
Archive:       http://mail.nl.linux.org/linux-crypto/