The following reply was made to PR kern/36284; it has been noted by GNATS.

From: Kurt Schreiner <ks@ub.uni-mainz.de>
To: Geoff Wing <gcw@pobox.com>
Cc: gnats-bugs@NetBSD.org, David Laight <dsl@NetBSD.org>
Subject: Re: kern/36284
Date: Mon, 21 May 2007 09:49:22 +0200

 On Mon, May 21, 2007 at 03:37:54AM +0200, Geoff Wing wrote:
 > On Saturday 2007-05-19 14:25 +0000, Kurt Schreiner output:
 > :From: Kurt Schreiner <ks@ub.uni-mainz.de>
 > :To: gnats-bugs@NetBSD.org
 > :Cc: David Laight <dsl@NetBSD.org>
 > :Subject: Re: kern/36284
 > :Date: Sat, 19 May 2007 16:23:57 +0200
 > :
 > : this was introduced between 29.04.2007_23:59 and 30.04.2007_23:59.
 > : Below's a diff of ident <kernel_from_29...> and ident <kernel_from_30...>.
 > : Maybe some of the changes to vfs_syscalls causes the "shining through"
 > : of files from the lower layer?
 > : <      $NetBSD: vfs_syscalls.c,v 1.310 2007/04/22 08:30:01 dsl Exp $
 > : ---
 > : >      $NetBSD: vfs_syscalls.c,v 1.311 2007/04/30 08:32:14 dsl Exp $
 > 
 > Good pick.  There was a logic inversion introduced here in sys_fstatvfs1().
 > 
 > 	-	if ((error = dostatvfs(mp, sb, l, SCARG(uap, flags), 1)) != 0)
 > 	-		goto out;
 > 	-	error = copyout(sb, SCARG(uap, buf), sizeof(*sb));
 > 	- out:
 > 	-	FILE_UNUSE(fp, l);
 > 	+	error = do_sys_fstatvfs(l, SCARG(uap, fd), SCARG(uap, flags), sb);
 > 	+	if (error != 0)
 > 	+		error = copyout(sb, SCARG(uap, buf), sizeof(*sb));
 > 
 > Regards,
 > Geoff
 > 
 > --- sys/kern/vfs_syscalls.c.org	2007-05-21 11:22:13.000000000 +1000
 > +++ sys/kern/vfs_syscalls.c	2007-05-21 11:25:34.000000000 +1000
 > @@ -881,7 +881,7 @@
 >  
 >  	sb = STATVFSBUF_GET();
 >  	error = do_sys_fstatvfs(l, SCARG(uap, fd), SCARG(uap, flags), sb);
 > -	if (error != 0)
 > +	if (error == 0)
 >  		error = copyout(sb, SCARG(uap, buf), sizeof(*sb));
 >  	STATVFSBUF_PUT(sb);
 >  	return error;
 Yep! Applied your patch and no more "shining" doublicates ;-)
 
 
 Kurt