From visigoth@telemere.net  Wed Aug 23 19:01:11 2000
Return-Path: <visigoth@telemere.net>
Received: from mail.telemere.net (mail.telemere.net [63.224.9.4])
	by hub.freebsd.org (Postfix) with ESMTP id 8B0A237B422
	for <FreeBSD-gnats-submit@freebsd.org>; Wed, 23 Aug 2000 19:01:11 -0700 (PDT)
Received: by mail.telemere.net (Postfix, from userid 1001)
	id 5517420F01; Wed, 23 Aug 2000 21:04:18 -0500 (CDT)
Message-Id: <20000824020418.5517420F01@mail.telemere.net>
Date: Wed, 23 Aug 2000 21:04:18 -0500 (CDT)
From: visigoth@telemere.net
Reply-To: visigoth@telemere.net
To: FreeBSD-gnats-submit@freebsd.org
Subject: mount opt "intr" for nfs mounts undocumented in fstab.5
X-Send-Pr-Version: 3.2

>Number:         20814
>Category:       docs
>Synopsis:       mount opt "intr" for nfs mounts undocumented in fstab.5
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    sheldonh
>State:          closed
>Quarter:        
>Keywords:       
>Date-Required:  
>Class:          doc-bug
>Submitter-Id:   current-users
>Arrival-Date:   Wed Aug 23 19:10:01 PDT 2000
>Closed-Date:    Mon Aug 28 01:54:59 PDT 2000
>Last-Modified:  Mon Aug 28 01:55:18 PDT 2000
>Originator:     Visigoth (Damieon Stark)
>Release:        FreeBSD 4.1-STABLE i386
>Organization:
Telemere Technologies Inc
>Environment:

FreeBSD breakme.telemere.net 4.1-STABLE FreeBSD 4.1-STABLE #0: Thu Aug

>Description:

Nice option for fstab file allowing hung processes waiting on nfs i/o to
be killed and die with EINTR is not documented in fstab.5

>How-To-Repeat:

Type man fstab from your friendly neighborhood freebsd box and look for
how you enable the option for mount_nfs -i 

>Fix:

Trivial Patch as follows....

*** /usr/src/share/man/man5/fstab.5.old Wed Aug 23 20:33:13 2000
--- /usr/src/share/man/man5/fstab.5     Wed Aug 23 20:33:31 2000
***************
*** 151,156 ****
--- 151,164 ----
  userquota=/var/quotas/tmp.user
  .Ed
  .Pp
+ If the option ``intr'' is specified for a
+ .Pa fs_vfstype
+ .Pa nfs
+ filesystem, the mount is considered interruptible, which implies
+ that file system calls that are delayed due to an unresponsive
+ server will fail with EINTR when a termination signal is
+ posted for the process.
+ .Pp
  If the option ``noauto'' is specified, the filesystem will not be automatically
  mounted at system startup.
  .Pp




>Release-Note:
>Audit-Trail:

From: Sheldon Hearn <sheldonh@uunet.co.za>
To: visigoth@telemere.net
Cc: FreeBSD-gnats-submit@freebsd.org
Subject: Re: docs/20814: mount opt "intr" for nfs mounts undocumented in fstab.5 
Date: Thu, 24 Aug 2000 10:47:16 +0200

 On Wed, 23 Aug 2000 21:04:18 EST, visigoth@telemere.net wrote:
 
 > >Number:         20814
 > >Category:       docs
 > >Synopsis:       mount opt "intr" for nfs mounts undocumented in fstab.5
 
 I think that it's properly documented in mount_nfs(8).  Since fstab(5)
 already mentions mount_nfs(8), I think the only required change is to
 list mount_nfs(8) in the SEE ALSO section of the fstab(5) page.
 
 Ciao,
 Sheldon.
 

From: Visigoth <visigoth@telemere.net>
To: freebsd-gnats-submit@FreeBSD.org
Cc: sheldonh@uunet.co.za
Subject: Re: docs/20814: mount opt "intr" for nfs mounts undocumented in
 fstab.5
Date: Thu, 24 Aug 2000 09:47:09 -0500 (CDT)

 -----BEGIN PGP SIGNED MESSAGE-----
 Hash: SHA1
 
 
 
 	Not to dissagree with you sheldon, but I believe that mentioning
 that the word "intr" can be used in the fstab file is important
 considering the fact that mount_nfs.8 doesn't mention that the acitual
 word "intr" is what you would use.  I submitted it because it took quite a
 while to find exactly what option to put in the fstab(5) file to take
 advantage of that option.
 
 Thanks for hearing me out ;)  Either way...
 
 Laterz
 Visigoth
 
 
 Damieon Stark
 Sr. Unix Systems Administrator
 visigoth@telemere.net
 
 PGP Public Key: www.telemere.net/~visigoth/visigoth.asc
 
 ____________________________________________________________________________
 					|
 M$ -Where do you want to go today?	|
 Linux -Where do you want to go tomorrow?|   FreeBSD - The POWER to serve
 Freebsd -Are you guys coming or what?	|   http://www.freebsd.org
 					|
 					|
 - ----------------------------------------------------------------------------
 
 -----BEGIN PGP SIGNATURE-----
 Version: PGP 6.5.1i
 
 iQA/AwUBOaUnYTnmC/+RTnGeEQLQOACg1TYehweb5+y1Rcgzgwvx2u6gIFoAoLC7
 Gt7w8tOH8n88CyRZunqCmSKP
 =u4hc
 -----END PGP SIGNATURE-----
 
 

From: Sheldon Hearn <sheldonh@uunet.co.za>
To: Visigoth <visigoth@telemere.net>
Cc: freebsd-gnats-submit@freebsd.org
Subject: Re: docs/20814: mount opt "intr" for nfs mounts undocumented in fstab.5 
Date: Thu, 24 Aug 2000 16:52:57 +0200

 On Thu, 24 Aug 2000 09:47:09 EST, Visigoth wrote:
 
 > 	Not to dissagree with you sheldon, but I believe that mentioning
 > that the word "intr" can be used in the fstab file is important
 > considering the fact that mount_nfs.8 doesn't mention that the acitual
 > word "intr" is what you would use.
 
 mount_nfs(8):
      -i      Make the mount interruptible, which implies that file system
              calls that are delayed due to an unresponsive server will fail
              with EINTR when a termination signal is posted for the process.
 
 [...]
 
              intr         Same as -i.
 
 :-)
 
 Ciao,
 Sheldon.
 

From: Sheldon Hearn <sheldonh@uunet.co.za>
To: bmah@freebsd.org
Cc: freebsd-gnats-submit@freebsd.org
Subject: Re: docs/20814: mount opt "intr" for nfs mounts undocumented in fstab.5 
Date: Fri, 25 Aug 2000 09:57:10 +0200

 On Thu, 24 Aug 2000 12:25:36 MST, "Bruce A. Mah" wrote:
 
 > You snipped out the part of mount_nfs(8) that reads, "Use of these
 > options is deprecated, they are only mentioned for compatability with
 > historic versions of mount_nfs."
 
 I see.  In that case, the mount_nfs(8) manual page would appear to
 require some work.
 
 My concern is that we should not adjust the manual pages every time
 someone reads them too fast and becomes confused.  I'm well qualified to
 have an opinion on the subject because I've made this mistake many times
 myself. :-)
 
 I don't think that it's appropriate to mention mount options for every
 VFS in the fstab(5) manual page, especially when those options are well
 described in the manuals pertaining to those VFS.  Rather, only options
 common to all VFS should be included in the fstab(5) manual page.
 
 If you agree with me so far, then you may want to take a stab at fixing
 the mount_nfs(8) manual page.  I've taken a two minute look, and no
 elegant solution sprang to mind. :-)
 
 Ciao,
 Sheldon.
 

From: bmah@cisco.com (Bruce A. Mah)
To: Sheldon Hearn <sheldonh@uunet.co.za>
Cc: bmah@freebsd.org, freebsd-gnats-submit@freebsd.org
Subject: Re: docs/20814: mount opt "intr" for nfs mounts undocumented in fstab.5 
Date: Fri, 25 Aug 2000 09:53:37 -0700

 If memory serves me right, Sheldon Hearn wrote:
 
 > If you agree with me so far, then you may want to take a stab at fixing
 > the mount_nfs(8) manual page.  I've taken a two minute look, and no
 > elegant solution sprang to mind. :-)
 
 I was about to write up some changes, but then I found PR docs/735,
 which addressed this very same issue (and in fact, is what led to the
 "deprecated" options being listed in mount_nfs in the first place).
 After reading this (and the mount(8) manpage) very carefully, I've come 
 to realize that actually specifying "intr" in fstab is in fact *wrong*. 
 Instead, one should use the "dashed" option, e.g.:
 
 server:/remote/fs	/local/fs	nfs	ro,-i	0	0
 
 As I read the PR and manpage, the only reason we support "intr" is for 
 compatability with other versions of mount_nfs(8) (e.g. Solaris's 
 mount_nfs(8) uses this option syntax).  Using the "dashed" style 
 options is preferred, presumably because it's easy to tell at a glance 
 which options apply to all VFS types and which ones apply to only a 
 particular type of filesystem.
 
 So I've changed my stance on this.  In my opinion, only a small
 clarifying change is needed to the mount(8) manpage...to specifically
 tell people to look in filesystem-specific manpages to see the options
 for a given filesystem.  I suggest the patch below, with the caveat that
 I can't quite figure out how to get the .Xr macro to Do What I Want(TM)
 (i.e. I want part of a manpage cross-reference in emphasized text).  Do
 you know how to do this?
 
 Index: mount.8
 ===================================================================
 RCS file: /home/ncvs/src/sbin/mount/mount.8,v
 retrieving revision 1.32
 diff -c -r1.32 mount.8
 *** mount.8	2000/04/30 22:08:13	1.32
 --- mount.8	2000/08/25 16:46:49
 ***************
 *** 228,233 ****
 --- 228,243 ----
   .Bd -literal -offset indent
   /sbin/mount_mfs -o nosuid -N -s 4000 /dev/dk0b /tmp
   .Ed
 + .Pp
 + Options that pertain to a given filesystem type are described
 + in the 
 + .Xr mount_ Ns Em XXX 8
 + manual page, where 
 + .Em XXX 
 + is replaced by the type name.
 + For example, the options specific to NFS filesystems are described in the
 + .Xr mount_nfs 8
 + manual page.
   .It Fl p
   Print mount information in fstab format.
   Implies also the
 ***************
 *** 321,330 ****
   Verbose mode.
   .It Fl w
   The file system object is to be read and write.
 - .Pp
 - The options specific to NFS filesystems are described in the
 - .Xr mount_nfs 8
 - manual page.
   .Sh DIAGNOSTICS
   Various, most of them are self-explanatory.
   .Pp
 --- 331,336 ----
 
 Thanks,
 
 Bruce.
 
 
 
 

From: Sheldon Hearn <sheldonh@uunet.co.za>
To: bmah@cisco.com
Cc: bmah@freebsd.org, freebsd-gnats-submit@freebsd.org
Subject: Re: docs/20814: mount opt "intr" for nfs mounts undocumented in fstab.5 
Date: Sat, 26 Aug 2000 14:15:06 +0200

 On Fri, 25 Aug 2000 09:53:37 MST, Bruce A. Mah wrote:
 
 > After reading this (and the mount(8) manpage) very carefully, I've come 
 > to realize that actually specifying "intr" in fstab is in fact *wrong*. 
 > Instead, one should use the "dashed" option, e.g.:
 > 
 > server:/remote/fs	/local/fs	nfs	ro,-i	0	0
 
 Why's it wrong?  What negative impact does it have?  And what difference
 does it make whether one uses intr or -i, given that neither one is a
 standard mount(8) option.  :-)
 
 Ciao,
 Sheldon.
 

From: bmah@freebsd.org (Bruce A. Mah)
To: Sheldon Hearn <sheldonh@uunet.co.za>
Cc: bmah@freebsd.org, freebsd-gnats-submit@freebsd.org
Subject: Re: docs/20814: mount opt "intr" for nfs mounts undocumented in fstab.5 
Date: Sat, 26 Aug 2000 08:47:41 -0700

 --==_Exmh_1703988012P
 Content-Type: text/plain; charset=us-ascii
 
 If memory serves me right, Sheldon Hearn wrote:
 
 > On Fri, 25 Aug 2000 09:53:37 MST, Bruce A. Mah wrote:
 > 
 > > After reading this (and the mount(8) manpage) very carefully, I've come 
 > > to realize that actually specifying "intr" in fstab is in fact *wrong*. 
 > > Instead, one should use the "dashed" option, e.g.:
 > > 
 > > server:/remote/fs	/local/fs	nfs	ro,-i	0	0
 > 
 > Why's it wrong?  What negative impact does it have?  And what difference
 > does it make whether one uses intr or -i, given that neither one is a
 > standard mount(8) option.  :-)
 
 Because two manpages and a bunch of discussion surrounding a PR
 deprecates "intr"?  Look, I didn't write this part, and I'm not sure I
 even agree with it.  Maybe my comment was stronger than intended.  The
 take-away message is that I once wanted to change the mount_nfs(8)
 manpage, and now I don't feel there is a strong justification to do so
 based on what I've read.
 
 Please note that the patch I gave you *only* addresses the fact that the
 mount(8) manpage doesn't point users to the filesystem-specific manpages
 for filesystem-specific mount options.  I still think that patch is
 right, but I'm just going to shut up for now and leave it up to you to
 decide its fate.
 
 Cheers,
 
 Bruce.
 
 
 
 
 --==_Exmh_1703988012P
 Content-Type: application/pgp-signature
 
 -----BEGIN PGP SIGNATURE-----
 Version: PGPfreeware 5.0i for non-commercial use
 MessageID: IZ0B19fcPZQhkGSwyVUvycLDQvF2Bpaq
 
 iQA/AwUBOafmndjKMXFboFLDEQKIywCaA80mW8XrIE9XITu9oRA4tImJGLAAoOYP
 4B3lCaLIRUpFLQzKy58wM7Ab
 =s3v1
 -----END PGP SIGNATURE-----
 
 --==_Exmh_1703988012P--
 

From: Sheldon Hearn <sheldonh@uunet.co.za>
To: bmah@freebsd.org
Cc: freebsd-gnats-submit@freebsd.org
Subject: Re: docs/20814: mount opt "intr" for nfs mounts undocumented in fstab.5 
Date: Sun, 27 Aug 2000 11:17:55 +0200

 On Sat, 26 Aug 2000 08:47:41 MST, Bruce A. Mah wrote:
 
 > Please note that the patch I gave you *only* addresses the fact that the
 > mount(8) manpage doesn't point users to the filesystem-specific manpages
 > for filesystem-specific mount options.  I still think that patch is
 > right, but I'm just going to shut up for now and leave it up to you to
 > decide its fate.
 
 I agree with you that the mount(8) page should point folks at
 filesystem-specific pages.  I intend to commit the supplied patch on
 Monday.
 
 Ciao,
 Sheldon.
 
Responsible-Changed-From-To: freebsd-doc->sheldonh 
Responsible-Changed-By: sheldonh 
Responsible-Changed-When: Mon Aug 28 01:50:37 PDT 2000 
Responsible-Changed-Why:  
I'll commit a variant of Bruce's patch. 

http://www.freebsd.org/cgi/query-pr.cgi?pr=20814 
State-Changed-From-To: open->analyzed 
State-Changed-By: sheldonh 
State-Changed-When: Mon Aug 28 01:53:02 PDT 2000 
State-Changed-Why:  
A variation on the theme presented by Bruce has been 
committed to HEAD as rev 1.33. 

http://www.freebsd.org/cgi/query-pr.cgi?pr=20814 
State-Changed-From-To: analyzed->closed 
State-Changed-By: sheldonh 
State-Changed-When: Mon Aug 28 01:54:59 PDT 2000 
State-Changed-Why:  
Merged onto RELENG_4 as rev 1.31.2.2. 

http://www.freebsd.org/cgi/query-pr.cgi?pr=20814 
>Unformatted:
