Re: Filesystems being eaten?

naddy_at_mips.inka.de (Christian Weisgerber) wrote:

> Last Friday my 8.0-CURRENT/amd64 box suffered a crash--no idea if
> it panicked, I was running X11--after which all filesystems were
> so badly corrupted that I had to reinstall the machine.  This was
> a vanilla setup, UFS2 and UFS2+S, -CURRENT about a week old at the
> time.

I had several hangs/panics/whatever, too, and reverted to:
8.0-CURRENT #39: Sat Mar  7 20:37:29 CET 2009 for now.

There were actually multiple issues:

1) The nullfs panic that supposedly has been fixed by kib in 189622.
2) A radeon-related panic that reproducible happened shortly after
   exiting Xorg (screenshot of the backtrace available, I think,
   but I meant to reproduce it with a different kernel when the
   other problems started, so I didn't keep the core files)
3) Something network-related that was reproducibly (as in 100% of
   two attempts) triggered when using ssh's -D option:

   (kgdb) f 12
   #12 0xc0722006 in in_control (so=0xc84e0188, cmd=1076655123, 
       data=0xc6760d40 "", ifp=0x0, td=0xc8481720)
       at /usr/src/sys/netinet/in.c:211
   warning: Source file is more recent than executable.

   Whole backtrace available but due to the warning I'm not
   sure how accurate it is.  I didn't try without the -D option.
4) Possibly something else that caused the "surprise" panics
   while running Xorg.

I lost some files (of which I had backups), but no file system yet.
I was using the userland from 2009-03-07 all the time. I'm using i386.
I intend to give it another try this weekend.

Fabian