md/raid10: handle further errors during fix_read_error better.
authorNeilBrown <neilb@suse.de>
Thu, 28 Jul 2011 01:39:25 +0000 (11:39 +1000)
committerNeilBrown <neilb@suse.de>
Thu, 28 Jul 2011 01:39:25 +0000 (11:39 +1000)
If we find more read/write errors we should record a bad block before
failing the device.

Signed-off-by: NeilBrown <neilb@suse.de>
drivers/md/raid10.c

index fc9ebbab3f6277356ecfb4343d3c1298ef3003e5..8b29cd4f01c89f8213d7dcc43cd1aaf6dd5799c7 100644 (file)
@@ -1749,6 +1749,26 @@ static void check_decay_read_errors(mddev_t *mddev, mdk_rdev_t *rdev)
                atomic_set(&rdev->read_errors, read_errors >> hours_since_last);
 }
 
+static int r10_sync_page_io(mdk_rdev_t *rdev, sector_t sector,
+                           int sectors, struct page *page, int rw)
+{
+       sector_t first_bad;
+       int bad_sectors;
+
+       if (is_badblock(rdev, sector, sectors, &first_bad, &bad_sectors)
+           && (rw == READ || test_bit(WriteErrorSeen, &rdev->flags)))
+               return -1;
+       if (sync_page_io(rdev, sector, sectors << 9, page, rw, false))
+               /* success */
+               return 1;
+       if (rw == WRITE)
+               set_bit(WriteErrorSeen, &rdev->flags);
+       /* need to record an error - either for the block or the device */
+       if (!rdev_set_badblocks(rdev, sector, sectors, 0))
+               md_error(rdev->mddev, rdev);
+       return 0;
+}
+
 /*
  * This is a kernel thread which:
  *
@@ -1832,9 +1852,19 @@ static void fix_read_error(conf_t *conf, mddev_t *mddev, r10bio_t *r10_bio)
                rcu_read_unlock();
 
                if (!success) {
-                       /* Cannot read from anywhere -- bye bye array */
+                       /* Cannot read from anywhere, just mark the block
+                        * as bad on the first device to discourage future
+                        * reads.
+                        */
                        int dn = r10_bio->devs[r10_bio->read_slot].devnum;
-                       md_error(mddev, conf->mirrors[dn].rdev);
+                       rdev = conf->mirrors[dn].rdev;
+
+                       if (!rdev_set_badblocks(
+                                   rdev,
+                                   r10_bio->devs[r10_bio->read_slot].addr
+                                   + sect,
+                                   s, 0))
+                               md_error(mddev, rdev);
                        break;
                }
 
@@ -1855,10 +1885,10 @@ static void fix_read_error(conf_t *conf, mddev_t *mddev, r10bio_t *r10_bio)
 
                        atomic_inc(&rdev->nr_pending);
                        rcu_read_unlock();
-                       if (sync_page_io(rdev,
-                                        r10_bio->devs[sl].addr +
-                                        sect,
-                                        s<<9, conf->tmppage, WRITE, false)
+                       if (r10_sync_page_io(rdev,
+                                            r10_bio->devs[sl].addr +
+                                            sect,
+                                            s<<9, conf->tmppage, WRITE)
                            == 0) {
                                /* Well, this device is dead */
                                printk(KERN_NOTICE
@@ -1873,7 +1903,6 @@ static void fix_read_error(conf_t *conf, mddev_t *mddev, r10bio_t *r10_bio)
                                       "drive\n",
                                       mdname(mddev),
                                       bdevname(rdev->bdev, b));
-                               md_error(mddev, rdev);
                        }
                        rdev_dec_pending(rdev, mddev);
                        rcu_read_lock();
@@ -1893,11 +1922,12 @@ static void fix_read_error(conf_t *conf, mddev_t *mddev, r10bio_t *r10_bio)
 
                        atomic_inc(&rdev->nr_pending);
                        rcu_read_unlock();
-                       if (sync_page_io(rdev,
-                                        r10_bio->devs[sl].addr +
-                                        sect,
-                                        s<<9, conf->tmppage,
-                                        READ, false) == 0) {
+                       switch (r10_sync_page_io(rdev,
+                                            r10_bio->devs[sl].addr +
+                                            sect,
+                                            s<<9, conf->tmppage,
+                                                READ)) {
+                       case 0:
                                /* Well, this device is dead */
                                printk(KERN_NOTICE
                                       "md/raid10:%s: unable to read back "
@@ -1911,9 +1941,8 @@ static void fix_read_error(conf_t *conf, mddev_t *mddev, r10bio_t *r10_bio)
                                       "drive\n",
                                       mdname(mddev),
                                       bdevname(rdev->bdev, b));
-
-                               md_error(mddev, rdev);
-                       } else {
+                               break;
+                       case 1:
                                printk(KERN_INFO
                                       "md/raid10:%s: read error corrected"
                                       " (%d sectors at %llu on %s)\n",