The problem has been noticed on FreeBSD 6.2-RELEASE-p5 and FreeBSD 6.3

We first noted the problem when we wanted to backup our FreeBSD 6.2 to 6.3 using 
sh freebsd-update.sh -f freebsd-update.conf -r 6.3-RELEASE upgrade
as suggested in the 6.3 release Announcement.

After successfully completing a dump of a live file system using

dump -0Lf - /usr > /home/dumps/usr.dump

the system panic with

Jul 23 06:13:41 mail2 syslogd: kernel boot file is /boot/kernel/kernel
Jul 23 06:13:41 mail2 kernel: dev = da0s1d, block = 1, fs = /usr
Jul 23 06:13:41 mail2 kernel: panic: ffs_blkfree: freeing free block
Jul 23 06:13:41 mail2 kernel: cpuid = 0
Jul 23 06:13:41 mail2 kernel: Uptime: 44m25s
Jul 23 06:13:41 mail2 kernel: Cannot dump. No dump device defined.
Jul 23 06:13:41 mail2 kernel: Automatic reboot in 15 seconds - press a key on the console to abort
Jul 23 06:13:41 mail2 kernel: Rebooting...

This bug resembles a previously reported bug
http://www.jp.freebsd.org/cgi/query-pr.cgi?pr=kern/116913
for FreeBSD 7.0.

We believe this to be the same problem.  Our suspicion is that the panic happens when a large block of data on a file system mounted with soft updates is removed.  In our case the snap shot created by the -L option to dump cause the system to panic right after the dump completes and the system removes the snapshot file.

This thread seems to be discussing the same problem
http://www.mavetju.org/mail/view_message.php?list=freebsd-current&id=2546411

other links of interest
http://freebsd.monkey.org/freebsd-current/200710/threads.html#00196

Fix: 

A quick search of the internet suggests that the problem can be fixed by disabling soft updates for a file system.

after disabling softupdates for /usr the dump no longer causes a panic.

Thanks,

--luis
How-To-Repeat: 
dump -0Lf - /usr > /home/dumps/usr.dump

on a live system running the release version of FreeBSD 6.3