aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorNeilBrown <neilb@suse.de>2005-11-09 00:39:31 -0500
committerLinus Torvalds <torvalds@g5.osdl.org>2005-11-09 10:56:38 -0500
commitba22dcbf106338a5c46d6979f9b19564faae3d49 (patch)
tree4caae4532cda0412d2446eca8b036d4e0df447f8
parent007583c9253fed363a0bd71b039e9b40a0f6855e (diff)
[PATCH] md: improvements to raid5 handling of read errors
Two refinements to the 'attempt-overwrite-on-read-error' mechanism. 1/ If the array is read-only, don't attempt an over-write. 2/ If there are more than max_nr_stripes read errors on a device with no success, fail the drive. This will make sure a dead drive will be eventually kicked even when we aren't trying to rewrite (which would normally kick a dead drive more quickly. Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
-rw-r--r--drivers/md/md.c1
-rw-r--r--drivers/md/raid5.c25
-rw-r--r--include/linux/raid/md_k.h3
3 files changed, 21 insertions, 8 deletions
diff --git a/drivers/md/md.c b/drivers/md/md.c
index 3db5c3513072..3fb80397f8aa 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -1582,6 +1582,7 @@ static mdk_rdev_t *md_import_device(dev_t newdev, int super_format, int super_mi
1582 rdev->in_sync = 0; 1582 rdev->in_sync = 0;
1583 rdev->data_offset = 0; 1583 rdev->data_offset = 0;
1584 atomic_set(&rdev->nr_pending, 0); 1584 atomic_set(&rdev->nr_pending, 0);
1585 atomic_set(&rdev->read_errors, 0);
1585 1586
1586 size = rdev->bdev->bd_inode->i_size >> BLOCK_SIZE_BITS; 1587 size = rdev->bdev->bd_inode->i_size >> BLOCK_SIZE_BITS;
1587 if (!size) { 1588 if (!size) {
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 08a1620b9f8c..77610b98d4e0 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -420,21 +420,29 @@ static int raid5_end_read_request(struct bio * bi, unsigned int bytes_done,
420 clear_bit(R5_ReadError, &sh->dev[i].flags); 420 clear_bit(R5_ReadError, &sh->dev[i].flags);
421 clear_bit(R5_ReWrite, &sh->dev[i].flags); 421 clear_bit(R5_ReWrite, &sh->dev[i].flags);
422 } 422 }
423 if (atomic_read(&conf->disks[i].rdev->read_errors))
424 atomic_set(&conf->disks[i].rdev->read_errors, 0);
423 } else { 425 } else {
426 int retry = 0;
424 clear_bit(R5_UPTODATE, &sh->dev[i].flags); 427 clear_bit(R5_UPTODATE, &sh->dev[i].flags);
425 if (conf->mddev->degraded) { 428 atomic_inc(&conf->disks[i].rdev->read_errors);
429 if (conf->mddev->degraded)
426 printk("R5: read error not correctable.\n"); 430 printk("R5: read error not correctable.\n");
427 clear_bit(R5_ReadError, &sh->dev[i].flags); 431 else if (test_bit(R5_ReWrite, &sh->dev[i].flags))
428 clear_bit(R5_ReWrite, &sh->dev[i].flags);
429 md_error(conf->mddev, conf->disks[i].rdev);
430 } else if (test_bit(R5_ReWrite, &sh->dev[i].flags)) {
431 /* Oh, no!!! */ 432 /* Oh, no!!! */
432 printk("R5: read error NOT corrected!!\n"); 433 printk("R5: read error NOT corrected!!\n");
434 else if (atomic_read(&conf->disks[i].rdev->read_errors)
435 > conf->max_nr_stripes)
436 printk("raid5: Too many read errors, failing device.\n");
437 else
438 retry = 1;
439 if (retry)
440 set_bit(R5_ReadError, &sh->dev[i].flags);
441 else {
433 clear_bit(R5_ReadError, &sh->dev[i].flags); 442 clear_bit(R5_ReadError, &sh->dev[i].flags);
434 clear_bit(R5_ReWrite, &sh->dev[i].flags); 443 clear_bit(R5_ReWrite, &sh->dev[i].flags);
435 md_error(conf->mddev, conf->disks[i].rdev); 444 md_error(conf->mddev, conf->disks[i].rdev);
436 } else 445 }
437 set_bit(R5_ReadError, &sh->dev[i].flags);
438 } 446 }
439 rdev_dec_pending(conf->disks[i].rdev, conf->mddev); 447 rdev_dec_pending(conf->disks[i].rdev, conf->mddev);
440#if 0 448#if 0
@@ -1328,7 +1336,8 @@ static void handle_stripe(struct stripe_head *sh)
1328 /* If the failed drive is just a ReadError, then we might need to progress 1336 /* If the failed drive is just a ReadError, then we might need to progress
1329 * the repair/check process 1337 * the repair/check process
1330 */ 1338 */
1331 if (failed == 1 && test_bit(R5_ReadError, &sh->dev[failed_num].flags) 1339 if (failed == 1 && ! conf->mddev->ro &&
1340 test_bit(R5_ReadError, &sh->dev[failed_num].flags)
1332 && !test_bit(R5_LOCKED, &sh->dev[failed_num].flags) 1341 && !test_bit(R5_LOCKED, &sh->dev[failed_num].flags)
1333 && test_bit(R5_UPTODATE, &sh->dev[failed_num].flags) 1342 && test_bit(R5_UPTODATE, &sh->dev[failed_num].flags)
1334 ) { 1343 ) {
diff --git a/include/linux/raid/md_k.h b/include/linux/raid/md_k.h
index 4169c11e5451..200c69e34fc0 100644
--- a/include/linux/raid/md_k.h
+++ b/include/linux/raid/md_k.h
@@ -134,6 +134,9 @@ struct mdk_rdev_s
134 * only maintained for arrays that 134 * only maintained for arrays that
135 * support hot removal 135 * support hot removal
136 */ 136 */
137 atomic_t read_errors; /* number of consecutive read errors that
138 * we have tried to ignore.
139 */
137}; 140};
138 141
139typedef struct mdk_personality_s mdk_personality_t; 142typedef struct mdk_personality_s mdk_personality_t;