Skip to contentSkip to conversation

Introduction

PATCH New NeilBrown 1 participant 4 messages Sep 6, 2005 MD RAID

lore
Share & tools
lore

NeilBrown

author
#0 lore raw

PATCH 1/1 Introduction

Here are three patches for md in 2.6.13-mm1

The first (embarassment) gets sb_size initialisation REALLY right.
I did more testing with mdadm-v1 and found hot-add wasn't working either. This patch definitely covers all cases.

The other two fix problems with raid10 when enough drives fails that not all block are available. It was particularly a problem as recovery would BUG-out, but other issues are fixed to.

These patches are tested and suitable for early release to Linus.

Thanks,
NeilBrown

[PATCH md 001 of 3] Really get sb_size setting right in all cases. [PATCH md 002 of 3] Fix raid10 assembly when too many devices are missing. [PATCH md 003 of 3] Fix BUG when raid10 rebuilds without enough drives.

author
#1 lore raw

PATCH 1/1 Really get sb_size setting right in all cases.

There was another case where sb_size wasn't being set, so instead do the sensible thing and set if when filling in the content of a superblock. That ensures that whenever we write a superblock, the sb_size MUST be set.

### Diffstat output

 ./drivers/md/md.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/md/md.c renamed from drivers/md/md.c~current~ +2 -2
diff ./drivers/md/md.c~current~ ./drivers/md/md.c
@@ -712,6 +712,8 @@ static void super_90_sync(mddev_t *mddev
 	int i;
 	int active=0, working=0,failed=0,spare=0,nr_disks=0;
 
+	rdev->sb_size = MD_SB_BYTES;
+
 	sb = (mdp_super_t*)page_address(rdev->sb_page);
 
 	memset(sb, 0, sizeof(*sb));
@@ -2303,8 +2305,6 @@ static int add_new_disk(mddev_t * mddev,
 		else
 			rdev->in_sync = 0;
 
-		rdev->sb_size = MD_SB_BYTES;
-
 		if (info->state & (1<<MD_DISK_WRITEMOSTLY))
 			set_bit(WriteMostly, &rdev->flags);
 
author
#2 lore raw

PATCH 1/1 Fix raid10 assembly when too many devices are missing.

In reply to NeilBrown

if you try to assemble an array with too many missing devices, raid10 will now reject the attempt, instead of allowing it.

Also check when hot-adding a drive and refuse the hot-add if the array is beyond hope.

### Diffstat output

 ./drivers/md/raid10.c |   30 +++++++++++++++++++++++++++---
 1 file changed, 27 insertions(+), 3 deletions(-)
drivers/md/raid10.c renamed from drivers/md/raid10.c~current~ +27 -3
diff ./drivers/md/raid10.c~current~ ./drivers/md/raid10.c
@@ -906,6 +906,27 @@ static void close_sync(conf_t *conf)
 	conf->r10buf_pool = NULL;
 }
 
+/* check if there are enough drives for
+ * every block to appear on atleast one
+ */
+static int enough(conf_t *conf)
+{
+	int first = 0;
+
+	do {
+		int n = conf->copies;
+		int cnt = 0;
+		while (n--) {
+			if (conf->mirrors[first].rdev)
+				cnt++;
+			first = (first+1) % conf->raid_disks;
+		}
+		if (cnt == 0)
+			return 0;
+	} while (first != 0);
+	return 1;
+}
+
 static int raid10_spare_active(mddev_t *mddev)
 {
 	int i;
@@ -944,6 +965,8 @@ static int raid10_add_disk(mddev_t *mdde
 		 * very different from resync
 		 */
 		return 0;
+	if (!enough(conf))
+		return 0;
 
 	for (mirror=0; mirror < mddev->raid_disks; mirror++)
 		if ( !(p=conf->mirrors+mirror)->rdev) {
@@ -1684,9 +1707,10 @@ static int run(mddev_t *mddev)
 	init_waitqueue_head(&conf->wait_idle);
 	init_waitqueue_head(&conf->wait_resume);
 
-	if (!conf->working_disks) {
-		printk(KERN_ERR "raid10: no operational mirrors for %s\n",
-			mdname(mddev));
+	/* need to check that every block has at least one working mirror */
+	if (!enough(conf)) {
+		printk(KERN_ERR "raid10: not enough operational mirrors for %s\n",
+		       mdname(mddev));
 		goto out_free_conf;
 	}
 
author
#3 lore raw

PATCH 1/1 Fix BUG when raid10 rebuilds without enough drives.

In reply to NeilBrown

This shouldn't be a BUG. We should cope.

### Diffstat output

 ./drivers/md/raid10.c |    8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
drivers/md/raid10.c renamed from drivers/md/raid10.c~current~ +7 -1
diff ./drivers/md/raid10.c~current~ ./drivers/md/raid10.c
@@ -1474,7 +1474,13 @@ static sector_t sync_request(mddev_t *md
 					}
 				}
 				if (j == conf->copies) {
-					BUG();
+					/* Cannot recover, so abort the recovery */
+					put_buf(r10_bio);
+					r10_bio = rb2;
+					if (!test_and_set_bit(MD_RECOVERY_ERR, &mddev->recovery))
+						printk(KERN_INFO "raid10: %s: insufficient working devices for recovery.\n",
+						       mdname(mddev));
+					break;
 				}
 			}
 		if (biolist == NULL) {