md: restart recovery cleanly after device failure.

[safe/jmp/linux-2.6] / drivers / md / raid10.c

diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
index faf3d89..1de17da 100644 (file)
--- a/drivers/md/raid10.c
+++ b/drivers/md/raid10.c
@@ -1020,7 +1020,7 @@ static void error(mddev_t *mddev, mdk_rdev_t *rdev)
                /*
                 * if recovery is running, make sure it aborts.
                 */
-               set_bit(MD_RECOVERY_ERR, &mddev->recovery);
+               set_bit(MD_RECOVERY_INTR, &mddev->recovery);
        }
        set_bit(Faulty, &rdev->flags);
        set_bit(MD_CHANGE_DEVS, &mddev->flags);
@@ -1171,6 +1171,14 @@ static int raid10_remove_disk(mddev_t *mddev, int number)
                        err = -EBUSY;
                        goto abort;
                }
+               /* Only remove faulty devices in recovery
+                * is not possible.
+                */
+               if (!test_bit(Faulty, &rdev->flags) &&
+                   enough(conf)) {
+                       err = -EBUSY;
+                       goto abort;
+               }
                p->rdev = NULL;
                synchronize_rcu();
                if (atomic_read(&rdev->nr_pending)) {
@@ -1237,6 +1245,7 @@ static void end_sync_write(struct bio *bio, int error)
 
        if (!uptodate)
                md_error(mddev, conf->mirrors[d].rdev);
+
        update_head_pos(i, r10_bio);
 
        while (atomic_dec_and_test(&r10_bio->remaining)) {
@@ -1844,7 +1853,8 @@ static sector_t sync_request(mddev_t *mddev, sector_t sector_nr, int *skipped, i
                                        if (rb2)
                                                atomic_dec(&rb2->remaining);
                                        r10_bio = rb2;
-                                       if (!test_and_set_bit(MD_RECOVERY_ERR, &mddev->recovery))
+                                       if (!test_and_set_bit(MD_RECOVERY_INTR,
+                                                             &mddev->recovery))
                                                printk(KERN_INFO "raid10: %s: insufficient working devices for recovery.\n",
                                                       mdname(mddev));
                                        break;
@@ -2082,6 +2092,9 @@ static int run(mddev_t *mddev)
                goto out_free_conf;
        }
 
+       spin_lock_init(&conf->device_lock);
+       mddev->queue->queue_lock = &conf->device_lock;
+
        rdev_for_each(rdev, tmp, mddev) {
                disk_idx = rdev->raid_disk;
                if (disk_idx >= mddev->raid_disks
@@ -2103,7 +2116,6 @@ static int run(mddev_t *mddev)
 
                disk->head_position = 0;
        }
-       spin_lock_init(&conf->device_lock);
        INIT_LIST_HEAD(&conf->retry_list);
 
        spin_lock_init(&conf->resync_lock);