Scott Stevens wrote:
> Are you using the new filecore ? UnixFS accesses the disk using
> filecore operations. Also ensure theat the SWI base is correct for
> your card's DiscOp.

I'm trying to use UnixFS with the old filecore and a PowerTec SCSI2
card on a 270MB SCSI SyQuest with no RISC OS partition.  If I run
!BtRiscBSD.native./dev/sd0a I get:

Unknown partition type
Bombing out...
10
Internal error

It appears to (correctly) use 40980 as the SWI chunk base.

>  > bb_riscbsd
>  > ----------
>  > 
>  > This did not check correctly for SCSI when typing S. It always
>  > ended up with using ADFS. I just made a new copy of this program
>  > and set it to SCSI all the time.

The crucial bit is this:

     540 : IF filesys%=ASC("A")
     550 :   discop$="ADFS_DiscOp"
     560 : ELSE
     570 :   discop$="SCSI_DiscOp"
     580 : ENDIF

Let's just test this...

>NEW
>540 IF FALSE
>550   PRINT "But that's impossible!"
>560 ELSE
>570   PRINT "True. Phew."
>580 ENDIF
>RUN
But that's impossible!
>540 IF FALSE THEN
>RUN
True. Phew.
>

Because 540 doesn't end with "THEN", it's a single-line IF.  The
"ELSE" is read as "comment out this section of code silently please".

We noticed this when installing RiscBSD on a friend's machine.  His
IDE drive (all RISC OS) seems to be working okay despite him having
accidentally used bb_riscbsd to mark it as all BSD.  Are there likely
to be problems in the future?  (Of course, there's no backup bootblock
because we ran it more than once when it didn't seem to have any
effect on the SCSI drive...)

> Right this little proggy has only been tested on cumana and powertec
> cards AFAIK. These cards use SCSI_DiscOp. I'll change this. The
> case-sensitivity is a problem tho'.

My PowerTec uses SCSIFS_DiscOp.  I thought that my friend's Cumana did
too, but can't verify this right now.

We couldn't use UnixFS on his computer either.  RO3.6, Cumana SCSI2,
another 270MB SyQuest.  We tried both with the whole disc being used
for RiscBSD and with a minimal RISC OS partition.

On the subject of PowerTec cards: I think you (Scott) said recently
that the current RiscBSD driver always disables on-card termination.
You certainly know more about the subject than I do, but... I thought
that writing a 0 to the appropriate location on the card *enabled*
termination (i.e. the opposite of the *Configure SCSIFSTerminators
semantics), and that that's what the current driver's source (which I
can't check right now... HD failure... blah blah blah...) does.  Can
you just confirm that I'm completely wrong about this (which I
probably am)?

-- 
Aidan