Re: 3100/80: 4G disk limit?

[Mouse <mouse%Rodents-Montreal.ORG@localhost>]

>> I've got a 3100/80.  [...disk wrapping at 4G...]
> The 21-bit block addressing is just in the boot rom of *some* 3100
> models.  I don't even think the /80 is among those.  But as soon as
> the system have booted, you are not limited to that anyway, since
> then we're not using the boot rom drivers, but the ones in NetBSD.

Right.

> So I can't see any reason the hardware would be a part of this.

Neither can I, but I've seen enough peculiar surprises to be cautious,
and I'd rather not chase after a software problem only to discover it's
actually hardware.  I could easily imagine some hardware designer (less
likely at DEC, but, as I say, I've been surprised) thinking "oh, the
host is limited to 32 bits anyway, we don't need to support more than
23 bits of sector number".  It doesn't actually make sense, but I've
made equally nonsensical mistakes often enough (though admittedly not
in shipping product as far as I know).

> Besides, 4G is not even an obvious limit anywhere.  When we go to the
> 10-byte (or larger) SCSI commands, the block addresses are 32 bits,
> which means the limit would be about 2TB.

Right.  And we must be using 10-byte CDBs, or it would be wrapping
and/or erroring at 1G.

> But in software we are talking about disk addressing in bytes, so
> then 4G is an obvious point where it can break.

Right.  I suspect there is something somewhere that's carelessly
declared 32-bit instead of 64-bit.  If you aren't aware of any such
issue in the 3100/80 hardware, I'll assume there isn't any and, once I
get it running a fully-updated 1.6.2 in 4G of disk, start looking for a
software issue.

/~\ The ASCII				  Mouse
\ / Ribbon Campaign
 X  Against HTML		mouse%rodents-montreal.org@localhost
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B