FreeBSD 9.1 occasional kernel panic

[Andriy Gapon]

on 06/02/2013 10:29 Jens Jahnke said the following:
> Hi,
> 
> recently I've switched from 8.3 to 9.1 on my home box and are quite pleased
> but occasionally I get a kernel panic like this:
> 
> Feb  5 19:55:45 magni kernel: /dev: got error 6 while accessing filesystem 
> Feb  5 19:55:45 magni kernel: panic: softdep_deallocate_dependencies:
> unrecovered I/O error Feb  5 19:55:45 magni kernel: cpuid = 1 Feb  5 19:55:45
> magni kernel: KDB: stack backtrace: Feb  5 19:55:45 magni kernel: #0
> 0xffffffff809208a6 at kdb_backtrace+0x66 Feb  5 19:55:45 magni kernel: #1
> 0xffffffff808ea8be at panic+0x1ce Feb  5 19:55:45 magni kernel: #2
> 0xffffffff80b08d30 at clear_remove+0 Feb  5 19:55:45 magni kernel: #3
> 0xffffffff80966cd0 at brelse+0x60 Feb  5 19:55:45 magni kernel: #4
> 0xffffffff8096a14c at bufwrite+0xfc Feb  5 19:55:45 magni kernel: #5
> 0xffffffff80b0bd47 at softdep_process_journal+0x867 Feb  5 19:55:45 magni
> kernel: #6 0xffffffff80b0bf0a at jwait+0x9a Feb  5 19:55:45 magni kernel: #7
> 0xffffffff80b0c2b0 at flush_deplist+0x80 Feb  5 19:55:45 magni kernel: #8
> 0xffffffff80b0d531 at softdep_sync_metadata+0x1b1 Feb  5 19:55:45 magni
> kernel: #9 0xffffffff80b2236b at ffs_syncvnode+0x44b Feb  5 19:55:45 magni
> kernel: #10 0xffffffff80b22f91 at ffs_fsync+0x41 Feb  5 19:55:45 magni
> kernel: #11 0xffffffff8098c600 at sys_fsync+0x160 Feb  5 19:55:45 magni
> kernel: #12 0xffffffff80bd7ae6 at amd64_syscall+0x546 Feb  5 19:55:45 magni
> kernel: #13 0xffffffff80bc3447 at Xfast_syscall+0xf7
> 
> It does not happen very often but when it happens it mostly happens several
> times in a row. When I'm booting into single user while within such a
> "series" I seem to be able to reproduce the panic. If I mount my partitions
> one at a time the panic occurs several seconds after I mount my geli
> encrypted home partition.
> 
> System specs: Gigabyte GA-EP35-DS3P Intel Core 2 Duo E8500 4 GB DDR2-800 3
> HDDs (2x SATA, 1 SATA (Intel SSD))
> 
> System is installed on the SSD.
> 
> Any ideas where this could be originating from?

"error 6" is ENXIO, which usually corresponds to a disappearing device or some
such.  Do you get anything in logs (or other objective experience) that could
look like that?

-- 
Andriy Gapon