Re: CVS commit: src/sbin/newfs

[David Laight <david%l8s.co.uk@localhost>]

>    > i think it would be best if newfs Just Worked with vinum devices.
>    
>    No question.  Do you have any suggestion how?  I suppose one way would
>    be to ignore errors if you can't get the partition information
>    indicated by the last letter.  Another, tacky one would be to
>    recognize the path name (/dev/vinum) and go by that.  But that's even
>    worse than the -V flag.

The partition size could be got from stat, the label is only needed
for the sector size.  It is also used as defaults for the fragment/block
size and to validate the appleufs stuff.

IMHO the fragment/block sizes don't belong in the label!
And newfs shouldn't be writing the label....

> i see there is a DIOCGPART ioctl that isn't used anywhere in our
> tree but does appear to be supported by all the drivers...perhaps
> that could be tried on the as-is name?

Looks like you would have to open /dev/kmem and read the structure
from the reported address!

> (i had a quick look at the code in newfs, but it wasn't clear how
> all the partitioning stuff works.)

Maybe it doesn't!  The code is OTT for what it actually does.

        David

-- 
David Laight: david%l8s.co.uk@localhost