Path: utzoo!mnetor!uunet!lll-winken!lll-lcc!mordor!sri-spam!rutgers!ll-xn!mit-eddie!uw-beaver!cornell!rochester!ritcv!cci632!ccicpg!felix!zemon
From: jpn@teddy.UUCP (John P. Nelson)
Newsgroups: comp.unix.ultrix
Subject: Re: Bug in 2.0 NFS
Message-ID: <17488@felix.UUCP>
Date: 30 Dec 87 21:58:03 GMT
References: <16842@felix.UUCP> <16925@felix.UUCP>
Sender: zemon@felix.UUCP
Reply-To: jpn@teddy.UUCP (John P. Nelson)
Organization: GenRad, Inc., Concord, Mass.
Lines: 26
Approved: zemon@felix.UUCP
Reply-Path:



I guess my first posting wasn't entirely clear.  matt@oddjob obviously
got the wrong idea, others might have as well.

)John P. Nelson writes:
) The bug is that the "utimes" system call returns an error if applied
) over NFS to a read-only file that resides on an ULTRIX 2.0 machine.
) ... if the same operation is done
) on the same file, but LOCALLY (i.e. not through NFS), then it succeeds.

>matt@oddjob.UUCP (Godfather to putty-tats) writes:
>
>Are you saying that utimes() can change the timestamps on a file on a
>local read-only file system?

I said nothing about read-only FILESYSTEMS.  I am talking about a FILE
which is mode 0444 on a filesystem mounted both R/W locally and R/W
over NFS.  The utimes() call functions properly locally, but fails
over NFS.

This is a real bug.  We have also been informed by DEC that they are
aware of the bug, and it is fixed in Ultrix 2.2.

John P. Nelson
   decvax!genrad!teddy!jpn
   mit-eddie!genrad!teddy!jpn
   ARPA!talcott!panda!teddy!jpn