home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!stanford.edu!bcm!convex!darwin.sura.net!paladin.american.edu!auvm!psuvm!frmop11!barilvm!p85025
- Organization: Bar-Ilan University Computing Center, Israel
- Date: Wed, 11 Nov 1992 11:18:42 IST
- From: Doron Shikmoni <P85025@BARILVM.BITNET>
- Message-ID: <92316.111842P85025@BARILVM.BITNET>
- Newsgroups: bit.listserv.ibm-main
- Subject: Re: STORAGE BELOW 16M LINE
- References: <921109.120437.CET.REICHETZ@AWIIMC12.IMC.UniVie.AC.AT>
- Lines: 26
-
- In article <921109.120437.CET.REICHETZ@AWIIMC12.IMC.UniVie.AC.AT>,
- Christian.J.Reichetzeder@AWIIMC12.IMC.UNIVIE.AC.AT says:
- >
- >On Sun, 8 Nov 92 23:43:00 PST Leonard D Woren said:
- >>You would think so, but page 0 in a V=R machine is a special case.
- >Indeed prefixing is applied to page 0. The matter is quite complicated and
- >even more so with a V=R guest, who expects page 0 to be a real page 0 frame.
- >Thus VM has to move it's own real page 0 elsewhere. It may well be that
- >certain flavours of VM goofed in some aspects (dunno whether MVS does a LRA
- >for page 0 for example).
-
- I may be missing something (I joined this thread in the middle). However:
- Of course page 0 is prefixed. Even on native MVS (non VM) machines, page 0
- is prefixed (prefix register, remember). This is over and above DAT, as the
- 370/390 architecture requires that every processor sees its private version
- of page zero when it incurs an interrupt (think of old/new PSW and reg save
- areas). Obviously VM (or VM assist, in the past 7 years or so) maintains
- a virtual pref-reg for every virtual processor. And obviously, you'll see
- this when you LRA on page 0 (you will probably see different result on each
- processor).
-
- So: There are as many page zero frames in MVS as there are processors.
- And then, there is the unprefixed-page-zero frame - the real, real, real
- absolute zero..
-
- Doron
-