don't set vnode type to VNON in error case of ufs_makeinode.

(backout rev.1.74)

it seems that there's no need to do it (anymore?) and LFS has trouble with it.
(VNON vnodes marked VDIROP will never reclaimed)

ok'ed by Frank van der Linden.
This commit is contained in:
yamt 2002-12-31 15:00:18 +00:00
parent 140a8e56ca
commit a5bf83bbfc

View File

@ -1,4 +1,4 @@
/* $NetBSD: ufs_vnops.c,v 1.88 2002/10/23 09:15:09 jdolecek Exp $ */
/* $NetBSD: ufs_vnops.c,v 1.89 2002/12/31 15:00:18 yamt Exp $ */
/*
* Copyright (c) 1982, 1986, 1989, 1993, 1995
@ -41,7 +41,7 @@
*/
#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: ufs_vnops.c,v 1.88 2002/10/23 09:15:09 jdolecek Exp $");
__KERNEL_RCSID(0, "$NetBSD: ufs_vnops.c,v 1.89 2002/12/31 15:00:18 yamt Exp $");
#include "opt_quota.h"
#include "fs_lfs.h"
@ -2068,7 +2068,6 @@ ufs_makeinode(int mode, struct vnode *dvp, struct vnode **vpp,
#endif
if (DOINGSOFTDEP(tvp))
softdep_change_linkcnt(ip);
tvp->v_type = VNON;
vput(tvp);
return (error);
}