From owner-cvs-sys  Fri Jun 27 13:23:19 1997
Return-Path: <owner-cvs-sys>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.5/8.8.5) id NAA13265
          for cvs-sys-outgoing; Fri, 27 Jun 1997 13:23:19 -0700 (PDT)
Received: (from wpaul@localhost)
          by hub.freebsd.org (8.8.5/8.8.5) id NAA13103;
          Fri, 27 Jun 1997 13:19:26 -0700 (PDT)
From: Bill Paul <wpaul>
Message-Id: <199706272019.NAA13103@hub.freebsd.org>
Subject: Re: cvs commit: src/sys/nfs nfs_vfsops.c
To: bde@zeta.org.au (Bruce Evans)
Date: Fri, 27 Jun 1997 13:19:25 -0700 (PDT)
Cc: cvs-all@freebsd.org, cvs-committers@freebsd.org, cvs-sys@freebsd.org
In-Reply-To: <199706271947.FAA04210@godzilla.zeta.org.au> from "Bruce Evans" at Jun 28, 97 05:47:51 am
X-Mailer: ELM [version 2.4 PL24]
Content-Type: text
Sender: owner-cvs-sys@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

> 
> >  This is a bit kludgy in that nfsmout is a label defined by the nfsm_subs.h
> >  macros, but these macros are themselves more than a little kludgy. This
> >  stops the machine from crashing, but does not fix the overall bug: 'error'
> >  somehow becomes 5 (EIO) when a statfs() is performed on the locally mounted
> >  NFS filesystem. This seems to only happen the first time the filesystem
> >  is accesed: on subsequent accesses, it seems to work fine again.
> 
> I think the following fixes the bogus errno.  The mkfifo bug also caused
> EIO instead of the correct errno.
> 
> Bruce
> 
> diff -c2 nfs_serv.c~ nfs_serv.c
> *** nfs_serv.c~	Sun Jun 15 01:24:55 1997
> --- nfs_serv.c	Sun Jun 15 01:25:14 1997
> ***************
> *** 1668,1672 ****
>   	}
>   	nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft);
> ! 	return (0);
>   nfsmout:
>   	if (dirp)
> --- 1669,1673 ----
>   	}
>   	nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft);
> ! 	return (error);
>   nfsmout:
>   	if (dirp)
> 

Nope; I just built a kernel with this change and it still does it. I
added a printf() nfs_vfsops() so I would check the value of 'error':

Jun 27 16:09:36 mbone /vmunix: ERROR: 0
Jun 27 16:09:46 mbone last message repeated 12 times
Jun 27 16:09:46 mbone /vmunix: ERROR: 5		<--- first time 'df' is run
Jun 27 16:09:57 mbone /vmunix: ERROR: 0		<--- second

I was wondering how a change to nfsrv_mknod() was supposed to fix this
since I never call mknod() on the filesystem.

That said, nfsrv_statfs() in the nfs_serv.c module also does things
like:

        if (v3)
                nfsm_srvpostop_attr(getret, &at);
        if (error)
                return (0);

Shouldn't this be:

	if (error)
		return(error);

Or am I missing something.

-Bill

-Bill