Re: amd64 -current crashs at boot

[Patrick Welche <prlw1%cam.ac.uk@localhost>]

On Tue, Dec 23, 2008 at 09:33:06AM -0700, Sverre Froyen wrote:
> > > spec_vnops.c:781 says:
> > >
> > > error = VFS_FSYNC(mp, vp, ap->a_flags | FSYNC_VFS);
> > >
> > > Does this, i.e., "| FSYNC_VFS" imply more crack smoking?

Funny thing about that is:

    if (vp->v_type == VBLK) {
        if ((mp = vp->v_specmountpoint) != NULL) {
            error = VFS_FSYNC(mp, vp, ap->a_flags | FSYNC_VFS);
            if (error != EOPNOTSUPP)
                return error;
        }
        vflushbuf(vp, (ap->a_flags & FSYNC_WAIT) != 0);
    }

If VBLK call VFS_FSYNC, so that seems to assume if VBLK set FSYNC_VFS.
(If FSYNC_VFS really means sync by calling vfs_fsync (of which I haven't
yet found an example))

Cheers,

Patrick

Index: vchain
===================================================================
RCS file: /cvsroot/src/sys/gdbscripts/vchain,v
retrieving revision 1.5
diff -u -r1.5 vchain
--- vchain      4 Nov 2006 20:33:17 -0000       1.5
+++ vchain      23 Dec 2008 23:19:58 -0000
@@ -8,7 +8,7 @@
 
        set $vp=(struct vnode *)$arg0
        while ($vp)
-               printf "vp: 0x%x freelist_next: 0x%x usecount: %d flags: 
0x%x\n", $vp, $vp->v_freelist.tqe_next, $vp->v_uobj.uo_refs, $vp->v_flag
+               printf "vp: 0x%x freelist_next: 0x%x usecount: %d vflags: 
0x%x\n", $vp, $vp->v_freelist.tqe_next, $vp->v_uobj.uo_refs, $vp->v_vflag
                set $num++
                set $vp = $vp->v_mntvnodes.tqe_next
        end