[src/trunk]: src/sys/kern When called with WRITECLOSE vflush() must sync the ...

[hannken <hannken%NetBSD.org@localhost>]

details:   https://anonhg.NetBSD.org/src/rev/68a56894211a
branches:  trunk
changeset: 350928:68a56894211a
user:      hannken <hannken%NetBSD.org@localhost>
date:      Fri Jan 27 10:46:18 2017 +0000

description:
When called with WRITECLOSE vflush() must sync the vnode and take
care of unlinked but open vnodes.

PR kern/30525 remounting ffs read-only (mount -ur) does not sync metadata.

diffstat:

 sys/kern/vfs_mount.c |  139 +++++++++++++++++++++++++++++---------------------
 1 files changed, 80 insertions(+), 59 deletions(-)

diffs (184 lines):

diff -r bc9964cc76a2 -r 68a56894211a sys/kern/vfs_mount.c
--- a/sys/kern/vfs_mount.c      Fri Jan 27 10:45:11 2017 +0000
+++ b/sys/kern/vfs_mount.c      Fri Jan 27 10:46:18 2017 +0000
@@ -1,4 +1,4 @@
-/*     $NetBSD: vfs_mount.c,v 1.45 2017/01/13 10:10:32 hannken Exp $   */
+/*     $NetBSD: vfs_mount.c,v 1.46 2017/01/27 10:46:18 hannken Exp $   */
 
 /*-
  * Copyright (c) 1997-2011 The NetBSD Foundation, Inc.
@@ -67,7 +67,7 @@
  */
 
 #include <sys/cdefs.h>
-__KERNEL_RCSID(0, "$NetBSD: vfs_mount.c,v 1.45 2017/01/13 10:10:32 hannken Exp $");
+__KERNEL_RCSID(0, "$NetBSD: vfs_mount.c,v 1.46 2017/01/27 10:46:18 hannken Exp $");
 
 #include <sys/param.h>
 #include <sys/kernel.h>
@@ -478,87 +478,108 @@
 struct ctldebug debug1 = { "busyprt", &busyprt };
 #endif
 
-struct vflush_ctx {
-       const struct vnode *skipvp;
-       int flags;
-};
-
-static bool
-vflush_selector(void *cl, struct vnode *vp)
-{
-       struct vflush_ctx *c = cl;
-       /*
-        * Skip over a selected vnode.
-        */
-       if (vp == c->skipvp)
-               return false;
-       /*
-        * Skip over a vnodes marked VSYSTEM.
-        */
-       if ((c->flags & SKIPSYSTEM) && (vp->v_vflag & VV_SYSTEM))
-               return false;
-
-       /*
-        * If WRITECLOSE is set, only flush out regular file
-        * vnodes open for writing.
-        */
-       if ((c->flags & WRITECLOSE) && vp->v_type == VREG) {
-               if (vp->v_writecount == 0)
-                       return false;
-       }
-       return true;
-}
-
 static vnode_t *
-vflushnext(struct vnode_iterator *marker, void *ctx, int *when)
+vflushnext(struct vnode_iterator *marker, int *when)
 {
        if (hardclock_ticks > *when) {
                yield();
                *when = hardclock_ticks + hz / 10;
        }
-       return vfs_vnode_iterator_next1(marker, vflush_selector, ctx, true);
+       return vfs_vnode_iterator_next1(marker, NULL, NULL, true);
 }
 
+/*
+ * Flush one vnode.  Referenced on entry, unreferenced on return.
+ */
+static int
+vflush_one(vnode_t *vp, vnode_t *skipvp, int flags)
+{
+       int error;
+       struct vattr vattr;
+
+       if (vp == skipvp ||
+           ((flags & SKIPSYSTEM) && (vp->v_vflag & VV_SYSTEM))) {
+               vrele(vp);
+               return 0;
+       }
+       /*
+        * If WRITECLOSE is set, only flush out regular file
+        * vnodes open for writing or open and unlinked.
+        */
+       if ((flags & WRITECLOSE)) {
+               if (vp->v_type != VREG) {
+                       vrele(vp);
+                       return 0;
+               }
+               error = vn_lock(vp, LK_EXCLUSIVE);
+               if (error) {
+                       KASSERT(error == ENOENT);
+                       vrele(vp);
+                       return 0;
+               }
+               error = VOP_FSYNC(vp, curlwp->l_cred, FSYNC_WAIT, 0, 0);
+               if (error == 0)
+                       error = VOP_GETATTR(vp, &vattr, curlwp->l_cred);
+               VOP_UNLOCK(vp);
+               if (error) {
+                       vrele(vp);
+                       return error;
+               }
+               if (vp->v_writecount == 0 && vattr.va_nlink > 0) {
+                       vrele(vp);
+                       return 0;
+               }
+       }
+       /*
+        * First try to recycle the vnode.
+        */
+       if (vrecycle(vp))
+               return 0;
+       /*
+        * If FORCECLOSE is set, forcibly close the vnode.
+        */
+       if (flags & FORCECLOSE) {
+               vgone(vp);
+               return 0;
+       }
+       vrele(vp);
+       return EBUSY;
+}
 
 int
 vflush(struct mount *mp, vnode_t *skipvp, int flags)
 {
        vnode_t *vp;
        struct vnode_iterator *marker;
-       int busy = 0, when = 0;
-       struct vflush_ctx ctx;
+       int busy, error, when;
+
+       busy = error = when = 0;
 
        /* First, flush out any vnode references from deferred vrele list. */
        vfs_drainvnodes();
 
        vfs_vnode_iterator_init(mp, &marker);
 
-       ctx.skipvp = skipvp;
-       ctx.flags = flags;
-       while ((vp = vflushnext(marker, &ctx, &when)) != NULL) {
-               /*
-                * First try to recycle the vnode.
-                */
-               if (vrecycle(vp))
-                       continue;
-               /*
-                * If FORCECLOSE is set, forcibly close the vnode.
-                */
-               if (flags & FORCECLOSE) {
-                       vgone(vp);
-                       continue;
+       while ((vp = vflushnext(marker, &when)) != NULL) {
+               error = vflush_one(vp, skipvp, flags);
+               if (error == EBUSY) {
+                       error = 0;
+                       busy++;
+#ifdef DEBUG
+                       if (busyprt)
+                               vprint("vflush: busy vnode", vp);
+#endif
+               } else if (error != 0) {
+                       break;
                }
-#ifdef DEBUG
-               if (busyprt)
-                       vprint("vflush: busy vnode", vp);
-#endif
-               vrele(vp);
-               busy++;
        }
+
        vfs_vnode_iterator_destroy(marker);
+
+       if (error)
+               return error;
        if (busy)
-               return (EBUSY);
-
+               return EBUSY;
        return 0;
 }