[Simh] Fwd: 4.3BSD Quasijarus0c on the VAX-11/730

[Ethan Dicks]

On Wed, Jul 2, 2014 at 8:44 AM, Cory Smelosky <b4 at gewt.net> wrote:
>> root device? rb0*         <--- Enter "rb0*" here.
>> idc0: lost interrupt       <--- This error repeats infinitely.
>
> The only person I know with an 11/730 has a currently non-functional
> one...or it lacks a drive.

I have an 11/730 but it hasn't been fired up in over 10 years.  It's
in storage and I wouldn't be able to do anything with it until after
summer is over.

Looking at this, however, I am reminded of chasing down a bug with
SIMH over 15 years ago with the RP03 and 2BSD.  It had to do with the
interrupt probe routine that runs as the OS is first booting wanting
to force an interrupt of each device controller, but rp.c not properly
emulating the (poorly-documented) behavior of the real hardware when a
read-only register was written to.  Perhaps this is something similar?

> Boot
>: rb(0,1)/vmunix
> 326312+104440+130352 start 0x23b8
.
.
.
> idc0 at uba0 csr 175606 vec 250, ipl 15
> rb0 at idc0 slave 0
.
.
.
> root device? rb0*         <--- Enter "rb0*" here.
> idc0: lost interrupt       <--- This error repeats infinitely.

What's there up to the "lost interrupt" looks quite reasonable.  I
haven't grabbed the 4.3 sources to look at its driver, but I'd be
interested in knowing what code is firing that emits that message and
what conditions cause that to happen.  I'd also look closely at
whatever code in the driver is setting up the I/O and what bits in the
emulated CSR are getting twiddled.  I would be especially suspicious
if it did things differently than VMS was doing it.  Source for, say,
the VMS 4.4 RB730 driver might be easier to find on fische than
online, but I did spot the hardware docs on Bitsavers
(http://bitsavers.trailing-edge.com/pdf/dec/vax/730/EK-RB730-TD-001_VAX-11_730_IDC_Technical_Description_Sep82.pdf)

One should be able to attach an RA81 to the emulated 11/730, run from
that, then use the running system to debug problems with the RB730
code.  It would at least allow a controlled poking at the driver, and
even adding driver debug code.

-ethan