strange number corruption on pdp11/34

Pete Turnbull pete at dunnington.plus.com
Wed Jan 28 08:57:46 CST 2015


On 27/01/2015 19:32, Jacob Ritorto wrote:
> Hi,
>    I'm getting all zeroes nearly any time there's supposed to be a large
> number output to my terminal.  Does this sound familiar to anyone?
>
> This happened last night on rsx and is also happening today on xxdp.
>
> example xxdp testing rl02:

> BUS ADDRESS (O)  0 ?
> VECTOR (O)  0 ?

I agree with Johnny; even before I read his post I looked at this and 
thought here's a simple case of GIGO :-)

XXDP makes NO assumptions about things like CSRs for devices, and does 
almost no sanity checking on input.  There's no rule that says a 
controller has to be at a particular address, not even necessarily in 
the I/O page in Bank 7.  When you told XXDP the CSR is at address 0 (by 
not entering a different value), that's what it would believe, and it 
would happily test whatever is there.  In this case, that's real, 
existing, working memory, so it won't even fail with a bus error, but it 
will read back zeros after it's cleared - exactly as you saw.  Even the 
elapsed time is genuinely zero, as it's reporting immediately, and all 
the other values are the contents of a zeroed memory location instead of 
a register.

The fact that your CPU boots XXDP - never mind something more demanding 
like RSX or BSD - suggests it (and the memory, disks, etc) is pretty 
much working.  Retry the RL02 test with proper values.

-- 
Pete

Pete Turnbull


More information about the cctalk mailing list