18 bit bus addresses. 16 bit bus addresses. I'd have to look up the RL11 to know if it could be used in a 22 bit machine, but if so, it'd be 17777404 and 17777400. De

On Tue, Apr 01, 2008 at 08:38:33AM +1100, Doug Jackson - pesonal email wrote: Yes. It's not a documentation error. The hardware doesn't throw away the upper bits... quite the opposite... What's happening is that you have a machine with 16-bit registers, but are working with, essentially, 18 address bits. Look at 174400... 16 bits, so it fits inside a PDP-11 register. 774404 takes 18 bits to represent, and _is_ what ends up on the Unibus. What's going on under the hood is that when you point to addresses near the top few K (2K? 8K?), what goes onto the Unibus _doesn't_ trigger an access to your RAM card - it adds two 1 bits on the Unibus and tickles I/O cards instead. In otherwords, if you write a program to scan all 64K that you can directly address with a register (and not fiddling with MMU registers), you'll hit whatever 64K bank of RAM is mapped in, until you hit the I/O page, then it's I/O registers to 177777. If you are at the pre-boot stage (i.e., no OS has set up your MMU or Unibus mapping registers), you should probably refer to the handbook for your exact processor for details about how to hit I/O registers nearly 25 years, ISTR slight differences between, say, an 11/04 and an 11/34 about this sort of thing (because the 11/34 is a richer- featured machine). The 11/24 has either an 18-bit or 22-bit bus depending if you have a KT24 installed or not The 11/70 has a 22-bit address space always, AFAIK. I don't know enough about the internals of the 11/44 or other models like the 11/40, 11/45, 11/60, but I'm sure there are plenty of large -11 experts here who can chime in. You didn't say what machine you have, so I've kept things quite generic, mostly thinking of how the 11/04 and/or 11/34 do things since they are the most common. At the very least, you should be able to hit the CSR and DAR from either your front panel or console ODT and see that the board shows up. You mention a "seek test"... is this with XXDP or are you toggling-in a standalone test or what? Is this caution, or will the machine not boot with a bootable pack? -ethan -- Ethan Dicks, A-333-S Current South Pole Weather at 31-Mar-2008 at 21:40 Z South Pole Station PSC 468 Box 400 Temp -59.8 F (-51.0 C) Windchill -100.6 F (-73.6 C) APO AP 96598 Wind 17.5 kts Grid 30 Barometer 691.5 mb (10197 ft) Ethan.Dicks at usap.gov http://penguincentral.com/penguincentral.html

Doug Jackson - pesonal email wrote: This is actually the closest to the truth when operating in unmapped mode. The hardware actually adds two extra bits, converting 16b program virtual addresses (what your code sees) to 18b UNIBUS physical addresses (what the controller sees). For this specific case 16b addresses in the range 160000-177777 are converted to 18b addresses 760000-777777. So the real answer to your question above is both are correct. It depends on where in the system you are looking. I would suggest you look at any one of the PDP-11 Processor handbooks to see how 16b to 18b address mapping works, and a discussion of the 'I/O page' architecture.