Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site mit-eddie.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!mit-eddie!franz!layer@kim.berkeley.edu
From: layer@kim.berkeley.edu
Newsgroups: net.emacs
Subject: Sun Hardware bug?
Message-ID: <993@mit-eddie.UUCP>
Date: Fri, 17-Jan-86 17:47:56 EST
Article-I.D.: mit-eddi.993
Posted: Fri Jan 17 17:47:56 1986
Date-Received: Sun, 19-Jan-86 03:44:32 EST
Sender: daemon@mit-eddi.UUCP
Organization: MIT, Cambridge, MA
Lines: 14

From: franz!layer@kim.berkeley.edu (Kevin Layer)
This is a little known bug which burned me badly for months.  The
vadvise() system call on Sun OS versions 1.3-2.2 trashes the
address space of user programs under some circumstances, when vadvise()
is used and there is lots of paging going on.  The result is the
message in daver's letter.

Sun finally admitted that it was their fault (after my giving them a 10
line program which pointed to vadvise()), and the fix should be in 3.0
(though not definitely).

	Kevin