home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.arch
- Path: sparky!uunet!utcsri!utzoo!henry
- From: henry@zoo.toronto.edu (Henry Spencer)
- Subject: Re: SparcStation Questions
- Message-ID: <BrutGJ.36H@zoo.toronto.edu>
- Date: Thu, 23 Jul 1992 18:17:05 GMT
- References: <1992Jul23.163423.19061@dcc.uchile.cl>
- Organization: U of Toronto Zoology
- Lines: 34
-
- In article <1992Jul23.163423.19061@dcc.uchile.cl> pchris@dcc.uchile.cl writes:
- >Looking through the /usr/include/sun4/param.h file on some Sun Sparcs, I came
- >across a few things I could use some help on...
-
- To get this sort of information from Sun -- or, unfortunately, almost any
- manufacturer today -- you need to convince them that you have a worthy
- project which needs to know. Then you sign a nondisclosure agreement.
- Then they ship you a bundle of Xeroxed sheets, missing page 17 and two
- of the appendixes, with hand-drawn diagrams and inked-in corrections,
- telling you not quite everything you need to know about the previous
- version of the hardware. And if you ask about the missing bits, you get
- told "oh, we never did get that stuff written down properly".
-
- >What does this 3 level mmu on the 470 look like? What does the normal mmu
- >on the other Suns look like?
-
- Details on Sun MMUs in particular are definitely proprietary nondisclosure
- material, although you can learn a lot from careful study of include files
- and the like.
-
- The outlines are not secret. (One wonders why Sun is so anal-retentive
- about details that would not be hard to reverse-engineer.) Some bits of
- the address plus bits from what is essentially a short process ID form
- an address into memory A. The number that comes out, plus some more bits
- from the address, form an address into memory B. Out of that comes a
- page-table entry, with an address in main memory plus some permission
- bits and such. This is essentially an associative TLB simulated with
- ordinary RAMs, using memory A as the associative-lookup mechanism.
-
- That was the situation about five years ago, anyway; there are probably
- some new wrinkles in the more recent designs.
- --
- There is nothing wrong with making | Henry Spencer @ U of Toronto Zoology
- mistakes, but... make *new* ones. -D.Sim| henry@zoo.toronto.edu utzoo!henry
-