Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 30 Apr 1995 22:35:33 -0700 (PDT)
From:      Julian Elischer <julian@ref.tfs.com>
To:        bde@zeta.org.au (Bruce Evans)
Cc:        bde@freefall.cdrom.com, hackers@freefall.cdrom.com, julian@freefall.cdrom.com
Subject:   Re: early warning of DEVFS/SLICE mayhem
Message-ID:  <199505010535.WAA01686@ref.tfs.com>
In-Reply-To: <199505010516.PAA17351@godzilla.zeta.org.au> from "Bruce Evans" at May 1, 95 03:16:46 pm

next in thread | previous in thread | raw e-mail | index | archive | help
> 
> >I'm going to (as soon as it settles down) perform major mayhem on the slice
> >code.
> >...
> >the major change will be the breaking out of 'type handlers' for
> >each slice type. I need to write a type handler fro MACH2.5 type slices,
> >so I can cross mount my mach 2.5 partitions.. I believe this may 
> >also work for ISC sysV and other type 99 users as well. (but am unsure)
> 
> This will become very complicated.  There also needs to be a layer to
> support a type handler for each slicing type (dsinit() is only for DOS
> partitions and shouldn't be called directly).  There also needs to be
> a layer for unslicing (striping).
yes, but I think it's do-able..
and I NEED it..
also with DEVFS, each handler can name it's parts as it would expect :)
> 
> Note that slices are supposed to be machine-independent and support for
> foreign labels as labels was not planned for.

I have 16 partitions per MACH slice and they are definitly
not in a disklabel...
that's why my (example) definition in diskslice.h
changes sd_label to be an opaque data type..
each handler handles it's own structure..
we can't WRITE them, and DISKLABEL
will refuse to work with them, but that's ok..
it's to allow us to FIT IN with other systems that are already
present on a system

> I hoped to handle all
> foreign labelling schemes by converting them to slices.  This would
> require more slice bits, not more partition bits.

the trouble with this is that other unix 'slices' will have their own
bad-blocking methods.... we need a handler that can handle this or we might
as well not try..


> 
> Bruce
> 




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199505010535.WAA01686>