On 2021-03-21 15:44, Tom Ivar Helbekkmo wrote:
Johnny Billquist <bqt%update.uu.se@localhost> writes:As observed/mentioned, with at least RSX, you would have been forced to do a full SYSGEN if you changed anything, if you ran -11M.Looking at DEC's list of devices and allocations, I think maybe someone there had that in mind. When I made my sysgen tool, I found a few places in the list where a non-existent device grabs a chunk of CSR space, and they just happen to be placed immediately behind some of the more likely candidates for expansion: MSCP disk, DHV11 asynchronous serial, and DSV11 synchronous serial. The result is that adding a device or two of those types isn't going to change the CSR allocation of anything else. Anything that's willing to probe by the DEC algorithm, and use any floating interrupt vectors that the probing reveals, would then just discover the new devices, and continue working. I wonder if the diagnostic software does it that way?
This is maybe in some sense getting pretty far from port-vax, but anyway. :-)
Diagnostics for the PDP-11 never seemed to do any probing and autoconfiguration, so at least on that side, no.
The VAX diagnostics do have a autoconfiguration tool, that does probe according to the convention (EVSBA). But it is also mentioned for every diagnostic how you attach it to the diagnostics monitor manually, which you need to do it you are not following the convention. EVSBA also allows you to modify the results of probing before it is applied to the diagnostics monitor.
And yes, for some devices, you can add some more device(s) without it affecting the address of the next device type, simply because of how the rules work. There might have been some attempt at planning for that to possibly benefit from it, but it's mostly just a side effect of how the system works.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt%softjar.se@localhost || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol