In message <Pine.SGI.3.91.970103171251.6145J-100000@physig4.ph.kcl.ac.uk>
          Mark Brinicombe <amb@physig.ph.kcl.ac.uk> wrote:

> NO. The arm32 port does use the MI ramdisk the ramdisk file in arm32/dev
> is just the hooks to the MI ramdisk.
> 
> There should be no problems with the change from rd->md. However due to 
> gateway problems I have not yet checked out the new code and verified the 
> compile. Any problems with using md should be fixed rather than trying to 
> revert the code.

Well, something was forgotten obviously during the change. I tried to revert
the change and this worked for me as /sys/dev/ramdisk.{c,h} is still there.
Now that you've made everything work again I'm naturally using the new scheme.
If I make changes I always keep up a backup of both the original and my patched
version so it's easy to change back.

> > permission fault in kernel by kernel while trying to probe the Cumana SCSI
> > II card.
> 
> Ok I'll do this. BTW. LKM's are getting closer and this should help.

Will this really keep on working when the MI SCSI code is changed, eg. to
support more targets on Wide SCSI controllers? Doesn't this also affect the
interface to the LKMs?
Anyway, should Cumana keep on punishing their customers by insisting on this
ridiculous NDA, I will sell my card. But I'm sure they won't because no
reasonable person can insist on keeping some addresses secret. At least I can't
imagine that.
But let's wait for their final decision.

MfG MB