Bug 33647

Summary: Feedback for kern/32668
Product: Base System Reporter: Sheldon Hearn <sheldonh>
Component: kernAssignee: GNATS administrator <gnats-admin>
Status: Closed FIXED    
Severity: Affects Only Me    
Priority: Normal    
Version: 1.0-RELEASE   
Hardware: Any   
OS: Any   

Description Sheldon Hearn 2002-01-07 13:30:01 UTC
 The following is a reply from Ian Dowse to my enquiry as to whether or
 not a recent NFS fix from him had solved the problem.
 
 Ciao,
 Sheldon.
 
 ----- Original Message -----
 
 Date:  Tue, 01 Jan 2002 15:12:05 +0000
 From:  Ian Dowse <iedowse@maths.tcd.ie>
 To:  Sheldon Hearn <sheldonh@FreeBSD.org>
 cc:  cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org
 Subject:  Re: cvs commit: src/sys/nfs nfs_common.c 
 
 
 > In message <33575.1009791874@axl.seasidesoftware.co.za>, Sheldon Hearn writes:
 > >
 > >Sounds like this addresses PR kern/32668, although the PR reports a
 > >problem in -STABLE, where the NFS code is very different.  Hopefully,
 > >not so different in this case? :-)
 > 
 > No, this bug was specific to -current, as it was introduced with
 > Peter's cleanup and splitting of the NFS code.
 > 
 > The problem in kern/32668 is hard to fix in general because it is
 > a side-effect of the "silly rename" mechanism used by NFS to retain
 > unix semantics for open descriptors on deleted files. If there are
 > any open file descriptors on files in an NFS directory, then "rm
 > -rf" on that directory will fail.
 > 
 > The PR relates to the specific example of using portupgrade when
 > /var/db/pkg is NFS-mounted. It should be fairly easy to address
 > this case by finding out what process is holding the file descriptor
 > open, and changing something to ensure that the descriptor is closed
 > before the rm occurs.
 > 
 > Ian
 >
Comment 1 Mike Heffner freebsd_committer freebsd_triage 2002-01-07 21:41:07 UTC
State Changed
From-To: open->closed

Feedback for kern/32668.