On Aug 27,  2:55pm, Mark Gooderum wrote:
> 
> One other point worth noting...
> 
> The write() from newfs that fails fails before it gets to sdstart(), so
> I'm really having a hard time figuring out where the EINVAL is coming
> from.

  That's probably because the bounds_check_with_label() is being called
in sdstrategy and being errored out before the I/O is even queued up.

  If the disk label is keeping the partition size in real block units
(which it what it needs to do so that newfs computes things correctly),
then the bounds_check is going to fail on raw I/O:  the block number
is going to be in 512-byte sectors and the partition size is going
to be in 1024-byte sectors.

Michael

-- 
Michael L. Hitch			INTERNET:  osymh@montana.edu
Computer Consultant
Office of Systems and Computing Services
Montana State University	Bozeman, MT	USA

------------------------------------------------------------------------------