On Wed, 2010-09-15 at 11:05 -0700, Mike Riley wrote:
> On 09/15/10 07:45, Garrett D'Amore wrote:
> > On Tue, 2010-09-14 at 22:37 -0700, Mike Riley wrote:
> >
> >>
> >> There was a project to take the handling of partitions, and disk geometry,
> >> completely out of sd
> >> and impose it as a layer above it. That way the new layer would handle
> >> knowledge of disk
> >> layouts and provide sd with a simple starting sector, sector size, and
> >> total sectors info to
> >> work with. Again, it was being worked on back when I was with Sun, but I
> >> am not sure it ever
> >> got anywhere. I know several of the people working on it at the time are
> >> no longer with Sun, so
> >> my guess is that it died. Pity, because it is really needed, but it also
> >> is fraught with the
> >> possibility of bugs being introduced. However, sd is so full of bugs
> >> right now I am not sure a
> >> re-write/re-design would be any worse.
> >
> > That project is basically done. The code has been moved into "cmlb",
> > which handles all the details of partitioning, and labeling, etc. It is
> > a library with functions that the target driver (sd, or cmdk, or other
> > similar drivers) can call.
>
> Glad to hear it. But did they change sd, cmdk, etc. to use it yet?
Yes.
>
> I know they were dreading trying to do that and no one group at Sun wanted to
> take
> responsibility for sd. I had to get votes of approval from at least 3 groups
> to get a change in
> that I did to it so that larger drives could be handled. That fix was only
> supposed to be a
> band-aid until the code was changed to eliminate the need for disk geometry
> in sd.
>
Its long since done. :-) I think Shidokht played a major role there.
- Garrett
_______________________________________________
Discuss mailing list
[email protected]
http://lists.illumos.org/m/listinfo/discuss
