Xref: utzoo comp.unix.xenix:6320 comp.unix.wizards:16891
Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!cs.utexas.edu!uunet!yale!spock!odicon!jlg
From: jlg@odicon.UUCP (John L. Grzesiak)
Newsgroups: comp.unix.xenix,comp.unix.wizards
Subject: Re: Kernel Pager Error
Summary: comp.unix.xenix
Keywords: system crashes -- help!
Message-ID: <14@odicon.UUCP>
Date: 11 Jun 89 22:31:11 GMT
References: <420@crplabs.UUCP>
Organization: Omega Dynamics ,Wallingford Ct . ODICON
Lines: 42

In article <420@crplabs.UUCP>, marcs@crplabs.UUCP (Marc Snyder) writes:
> Netlanders:
> software is SCO Xenix 2.3.1.  Recently, the system crashed (on two separate
> occasions) and these messages were placed into the /usr/adm/messages file:
> 
> >kernel: PANIC: non-recoverable kernel page fault
> >kernel: PANIC: non-recoverable kernel page fault
> 
>         I would assume that this is some type of memory (RAM) fault --
> although I may be completely off base.  I was wondering if there was some
> 
> Marc Snyder -- System Administrator
> Ciba-Geigy Corporation  Ardsley, NY


S.C.O. knows about this one, at least two of their tech support people told
  me that this is a problem with certain 386 computers. I had this problem
  with a 386 system we were evaluating. As soon as the system went out of
  paging and into swapping it started. Drove me nuts is all it did! It started
  with page to frame errors (and vice versa) . That one did prove to be a 
  memory error, but after solving the memory problem we starting getting the
  exact errors you are having. The cure was 1 part easy and 2 parts tough.
  1) Upgrade to 2.3.2. (S.C.O. says the paging algoritym in 2.3.1. is buggy)
  2) If this doesn't work - throw the system to a DOS application and get
     another box for XENIX . (This is what we had to do).
  3) If this is a bearer of BAD NEWS - I'm sorry , but I could'nt figure a way
     around it. (I think I tried everything in the book , from slowing the
     system down to increasing i/o speed with a ESDI drive, to using large
     amount of kernal buffer space) ((THE increased buffers did help , but
     did not remove the problem , just cut the frequency).




    +-------------------------+---------------------------------+
    + John L Grzesiak         | "Genius is the output of        +
    + 47 Spring Street        |   a mind twisted beyond         +
    + Wallingford Ct 06492    |    normal recognition"          +
    +                         |                                 +
    + !yale!spock!odicon!jlg  | Omega Dynamics - Wallingford Ct +
    +                         |                                 +
    +-------------------------+---------------------------------+