VAX-11/730 and Emulex UC17 woes

[Josh Dersch]

> -----Original Message-----
> From: cctalk [mailto:cctalk-bounces at classiccmp.org] On Behalf Of Mike Ross
> Sent: Tuesday, May 24, 2016 2:15 PM
> To: General Discussion: On-Topic and Off-Topic Posts
> <cctalk at classiccmp.org>
> Subject: Re: VAX-11/730 and Emulex UC17 woes
> 
> On Wed, May 25, 2016 at 8:58 AM, Josh Dersch
> <JoshD at livingcomputermuseum.org> wrote:
> 
> > Interesting, the UC17 has the same firmware version (G143R) on the label
> of the EPROM.  I wonder if the contents are identical.  Could you send me a
> dump of your ROM so I can compare?
> >
> >>
> >> I dumped the memory for this code from the VAX and entered it into
> >> SIMH to use SIMH as an VAX unassembler and this is what it looks like.
> >> If the UC17 host resident VAX code is the same, any clues here as to
> >> what might be causing a halt on the 11/730?
>
> Here's what I found comparing my code with someone else's dump:
> 
> I've been working this with Glen Slick. Here's a snippet of a recent email with
> him. He provided a dump of the DMAed loader from his UC07:
> 
> The raw binary dumped on the VAX looked like this:
> 
> 
> <snipped for brevity>

Thanks.  Glen sent me his dump and I compared with mine.  I have the same three differences:

D 000002A0 01200880   // 0100F308
D 000002BC 0014688F   // FFF4688F
D 000002C0 65B45520   // 65B45500

(commented values are the ones that differ on my machine).

This changes the code as below:

-> 298:    MOVL #80000002,@#F30808   (was: 298:    MOVL #80000002,@#20088008)
2A3:    NOP
2A4:    MTPR #1F,#12
2A7:    MOVAL 400,R1
2AE:    MOVL #200,R2
2B5:    MOVL #0,(R1)+
2B8:    SOBGTR R2,2B5
-> 2BB:    MOVL #FFF468,R5   (was 2BB:    MOVL #20001468,R5)

I suspect these values are changed "on the fly" based on the machine type punched into the SA register as the last step before the "S 80" command; in particular, the source address for the MOVL instruction at 2BB now makes more sense, corresponding to a value in the UBA register set for the UC17.

However, F30808 doesn't make sense as the destination argument address for the MOVL at 298.  This looks like a map register for an 11/750 (as documented in the UC17 manual as being at F30800/F30804) and I wondered if it should be similar to the value for the 11/730 (F26800/F26804).

So I changed the instruction at 298 to:

MOVL #80000002,@#F26808

(which corresponds to a change in the value at 2A0 to 0100F268)

On the VAX-11/730 and did an "S 80" and lo and behold, it works!

I suspect either a typo in the manual for the machine type designator word, or a revision difference between firmware and documentation.

- Josh