svn commit: r324401 - stable/11/sys/geom/mirror

[Ngie Cooper]

Author: ngie
Date: Sat Oct  7 22:59:09 2017
New Revision: 324401
URL: https://svnweb.freebsd.org/changeset/base/324401

Log:
  MFC r306743,r317712:
  
  r306743 (by markj):
  
  gmirror: Bump the syncid if broken disks are found during startup.
  
  Consider a mirror with two components, m1 and m2. Suppose a hardware error
  results in the removal of m2, with m1's genid bumped. Suppose further that
  a replacement mirror component m3 is created and synchronized, after which
  the system is shut down uncleanly. During a subsequent bootup, if gmirror
  tastes m1 and m2 first, m2 will be removed from the mirror because it is
  broken, but the mirror will be started without bumping the syncid on m1
  because all elements of the mirror are accounted for. Then m3 will be
  added to the already-running mirror with the same syncid as m1, so the
  components will not be synchronized despite the unclean shutdown.
  
  Handle this scenario by bumping the syncid of healthy components if any
  broken mirrors are discovered during mirror startup.
  
  r317712 (by markj):
  
  Synchronize unclean mirrors before adding them to a running gmirror.
  
  During gmirror startup, if component mirrors are found to be dirty as is
  typical after a system crash, the mirrors are synchronized to the mirror
  with highest priority. However if a gmirror starts without all of its
  mirrors present, for example because of some transient delays during
  tasting, the remaining mirrors must be synchronized before they may become
  active.

Modified:
  stable/11/sys/geom/mirror/g_mirror.c
Directory Properties:
  stable/11/   (props changed)

Modified: stable/11/sys/geom/mirror/g_mirror.c
==============================================================================
--- stable/11/sys/geom/mirror/g_mirror.c	Sat Oct  7 21:13:54 2017	(r324400)
+++ stable/11/sys/geom/mirror/g_mirror.c	Sat Oct  7 22:59:09 2017	(r324401)
@@ -2195,7 +2195,9 @@ g_mirror_determine_state(struct g_mirror_disk *disk)
 	sc = disk->d_softc;
 	if (sc->sc_syncid == disk->d_sync.ds_syncid) {
 		if ((disk->d_flags &
-		    G_MIRROR_DISK_FLAG_SYNCHRONIZING) == 0) {
+		    G_MIRROR_DISK_FLAG_SYNCHRONIZING) == 0 &&
+		    (g_mirror_ndisks(sc, G_MIRROR_DISK_STATE_ACTIVE) == 0 ||
+		     (disk->d_flags & G_MIRROR_DISK_FLAG_DIRTY) == 0)) {
 			/* Disk does not need synchronization. */
 			state = G_MIRROR_DISK_STATE_ACTIVE;
 		} else {
@@ -2267,6 +2269,7 @@ g_mirror_update_device(struct g_mirror_softc *sc, bool
 	    {
 		struct g_mirror_disk *pdisk, *tdisk;
 		u_int dirty, ndisks, genid, syncid;
+		bool broken;
 
 		KASSERT(sc->sc_provider == NULL,
 		    ("Non-NULL provider in STARTING state (%s).", sc->sc_name));
@@ -2334,12 +2337,18 @@ g_mirror_update_device(struct g_mirror_softc *sc, bool
 		/*
 		 * Remove all disks without the biggest genid.
 		 */
+		broken = false;
 		LIST_FOREACH_SAFE(disk, &sc->sc_disks, d_next, tdisk) {
 			if (disk->d_genid < genid) {
 				G_MIRROR_DEBUG(0,
 				    "Component %s (device %s) broken, skipping.",
 				    g_mirror_get_diskname(disk), sc->sc_name);
 				g_mirror_destroy_disk(disk);
+				/*
+				 * Bump the syncid in case we discover a healthy
+				 * replacement disk after starting the mirror.
+				 */
+				broken = true;
 			}
 		}
 
@@ -2430,7 +2439,7 @@ g_mirror_update_device(struct g_mirror_softc *sc, bool
 		/* Reset hint. */
 		sc->sc_hint = NULL;
 		sc->sc_syncid = syncid;
-		if (force) {
+		if (force || broken) {
 			/* Remember to bump syncid on first write. */
 			sc->sc_bump_id |= G_MIRROR_BUMP_SYNCID;
 		}