Re: panic in geom (spindown + eil + journal)

From: John Baldwin <jhb_at_freebsd.org>
Date: Thu, 26 Jun 2008 20:19:46 -0400
On Thursday 26 June 2008 09:59:56 am Eygene Ryabinkin wrote:
> Bjoern, good day.
> 
> Thu, Jun 26, 2008 at 12:26:42PM +0000, Bjoern A. Zeeb wrote:
> > I guess some pointer artimetic or anything went wrong here for some
> > reason. ad0 is a simple hdd with a few slices with different, mostly
> > FreeBSD, installations. ad8 has, as seen above, eli on the disk and
> > one standard gjournaled UFS on top of that.
> 
> May be provider's geom data was trashed in some way.
> 
> > Is there any information I could try to find out to find that bug.
> > I had done the same thing hundreds of times before, the only new thing
> > from yesterday on was the spindown on 2 of 3 disks.
> 
> Any chances to compile debug kernel with KDB/DDB
> 
>   
http://www.freebsd.org/doc/en/books/developers-handbook/kerneldebug-online-ddb.html
> 
> and when system will panic and debugger will be invoked, issue
> commands 'p *(bp->bio_offset)' and 'p *(cp->provider)' at the stack
> frame corresponding to the function g_io_request?

Note that DDB can't handle structure dereferences like that (nor evaluate 
symbols for local variables), you'll need to get a crash dump and use kgdb 
for that sort of thing.

-- 
John Baldwin
Received on Thu Jun 26 2008 - 22:41:13 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:32 UTC