home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.arch:12420 comp.sys.hp:15370
- Newsgroups: comp.arch,comp.sys.hp
- Path: sparky!uunet!portal!sieler
- From: sieler@shell.portal.com (Stan Sieler)
- Subject: Re: HP-PA softw. virtual addr. translation HOW ?
- Message-ID: <C1JyD8.FK0@unix.portal.com>
- Followup-To: comp.arch,comp.sys.hp
- Sender: news@unix.portal.com
- Nntp-Posting-Host: jobe
- Organization: Portal Communications Company
- X-Newsreader: TIN [version 1.1 PL7]
- References: <1k6h9hINN6dm@news.cs.tu-berlin.de>
- Date: Thu, 28 Jan 1993 07:05:31 GMT
- Lines: 26
-
- Andre' Zehl (crusoe@cs.tu-berlin.de) wrote:
- : I'm working on a preparation on the HP-PA RISC Architecture.I'm desparately
- : trying to figure out, how you administer the virtual to physical address
- : translation.
- ...
- : The physical page number (ppn) is identical with the index from the
- :hashtable. But this only makes sense, if it's not the virtual page number (vpn)
- : which is hashed, but the physical page number, coming out of the
- : hardware translation? IS THIS SO ? Is the PDIR Index of the hashtable vpn
- the VPN is hashed.
- : the hashed ppn or the hashed ?
-
- Each entry in the hash table points to the FIRST PPN in a linked list of all
- physical pages whose VPN hash to the same value. Thus, you walk
- the linked list in the PDIR, examining each entry to see if its VPN matches
- the one you are looking for (presumably, the one you got a TLB miss on).
- If a match is found, you found it! (And the physical address is
- the PPN (which is based on the PDIR index as you noticed) plus the original
- bottom 11 or so bits (i.e.: virtual pages are aligned on physical page
- boundaries).
-
- (I emailed a little more to him)
-
- --
- Stan Sieler
- sieler@allegro.com
-