md_check_recovery only locks a device and does stuff when it thinks there
is a real likelyhood that something needs doing.
So the test at the top must cover all possibilities.
But it didn't cover the possibility that the last outstanding request on
a failed device had finished and so the device needed to be removed.
As a result, a failed drive might not get removed from the personalities
perspective on the array, and so it could never be removed from the array as
a whole.
With this patch, whenever ->nr_pending hits zero on a faulty device,
MD_RECOVERY_NEEDED is set so that md_check_recovery will do stuff.
Signed-off-by: Neil Brown <neilb@xxxxxxxxxxxxxxx>
----------- Diffstat output ------------
./drivers/md/multipath.c | 2 +-
./drivers/md/raid1.c | 8 ++++----
./drivers/md/raid5.c | 4 ++--
./drivers/md/raid6main.c | 4 ++--
./include/linux/raid/md_k.h | 8 ++++++++
5 files changed, 17 insertions(+), 9 deletions(-)
diff ./drivers/md/multipath.c~current~ ./drivers/md/multipath.c
--- ./drivers/md/multipath.c~current~ 2004-05-28 15:47:47.000000000 +1000
+++ ./drivers/md/multipath.c 2004-05-28 15:48:33.000000000 +1000
@@ -131,7 +131,7 @@ int multipath_end_request(struct bio *bi
(unsigned long long)bio->bi_sector);
multipath_reschedule_retry(mp_bh);
}
- atomic_dec(&rdev->nr_pending);
+ rdev_dec_pending(rdev, conf->mddev);
return 0;
}
diff ./drivers/md/raid1.c~current~ ./drivers/md/raid1.c
--- ./drivers/md/raid1.c~current~ 2004-05-28 15:48:33.000000000 +1000
+++ ./drivers/md/raid1.c 2004-05-28 15:48:33.000000000 +1000
@@ -296,7 +296,7 @@ static int raid1_end_read_request(struct
reschedule_retry(r1_bio);
}
- atomic_dec(&conf->mirrors[mirror].rdev->nr_pending);
+ rdev_dec_pending(conf->mirrors[mirror].rdev, conf->mddev);
return 0;
}
@@ -343,7 +343,7 @@ static int raid1_end_write_request(struc
raid_end_bio_io(r1_bio);
}
- atomic_dec(&conf->mirrors[mirror].rdev->nr_pending);
+ rdev_dec_pending(conf->mirrors[mirror].rdev, conf->mddev);
return 0;
}
@@ -831,7 +831,7 @@ static int end_sync_read(struct bio *bio
conf->mirrors[r1_bio->read_disk].rdev);
else
set_bit(R1BIO_Uptodate, &r1_bio->state);
- atomic_dec(&conf->mirrors[r1_bio->read_disk].rdev->nr_pending);
+ rdev_dec_pending(conf->mirrors[r1_bio->read_disk].rdev, conf->mddev);
reschedule_retry(r1_bio);
return 0;
}
@@ -861,7 +861,7 @@ static int end_sync_write(struct bio *bi
md_done_sync(mddev, r1_bio->sectors, uptodate);
put_buf(r1_bio);
}
- atomic_dec(&conf->mirrors[mirror].rdev->nr_pending);
+ rdev_dec_pending(conf->mirrors[mirror].rdev, mddev);
return 0;
}
diff ./drivers/md/raid5.c~current~ ./drivers/md/raid5.c
--- ./drivers/md/raid5.c~current~ 2004-05-28 15:48:33.000000000 +1000
+++ ./drivers/md/raid5.c 2004-05-28 15:48:33.000000000 +1000
@@ -395,7 +395,7 @@ static int raid5_end_read_request (struc
md_error(conf->mddev, conf->disks[i].rdev);
clear_bit(R5_UPTODATE, &sh->dev[i].flags);
}
- atomic_dec(&conf->disks[i].rdev->nr_pending);
+ rdev_dec_pending(conf->disks[i].rdev, conf->mddev);
#if 0
/* must restore b_page before unlocking buffer... */
if (sh->bh_page[i] != bh->b_page) {
@@ -438,7 +438,7 @@ static int raid5_end_write_request (stru
if (!uptodate)
md_error(conf->mddev, conf->disks[i].rdev);
- atomic_dec(&conf->disks[i].rdev->nr_pending);
+ rdev_dec_pending(conf->disks[i].rdev, conf->mddev);
clear_bit(R5_LOCKED, &sh->dev[i].flags);
set_bit(STRIPE_HANDLE, &sh->state);
diff ./drivers/md/raid6main.c~current~ ./drivers/md/raid6main.c
--- ./drivers/md/raid6main.c~current~ 2004-05-28 15:48:33.000000000 +1000
+++ ./drivers/md/raid6main.c 2004-05-28 15:48:33.000000000 +1000
@@ -414,7 +414,7 @@ static int raid6_end_read_request (struc
md_error(conf->mddev, conf->disks[i].rdev);
clear_bit(R5_UPTODATE, &sh->dev[i].flags);
}
- atomic_dec(&conf->disks[i].rdev->nr_pending);
+ rdev_dec_pending(conf->disks[i].rdev, conf->mddev);
#if 0
/* must restore b_page before unlocking buffer... */
if (sh->bh_page[i] != bh->b_page) {
@@ -457,7 +457,7 @@ static int raid6_end_write_request (stru
if (!uptodate)
md_error(conf->mddev, conf->disks[i].rdev);
- atomic_dec(&conf->disks[i].rdev->nr_pending);
+ rdev_dec_pending(conf->disks[i].rdev, conf->mddev);
clear_bit(R5_LOCKED, &sh->dev[i].flags);
set_bit(STRIPE_HANDLE, &sh->state);
diff ./include/linux/raid/md_k.h~current~ ./include/linux/raid/md_k.h
--- ./include/linux/raid/md_k.h~current~ 2004-05-28 15:48:33.000000000 +1000
+++ ./include/linux/raid/md_k.h 2004-05-28 15:48:33.000000000 +1000
@@ -255,6 +255,14 @@ struct mddev_s
struct list_head all_mddevs;
};
+
+static inline void rdev_dec_pending(mdk_rdev_t *rdev, mddev_t *mddev)
+{
+ int faulty = rdev->faulty;
+ if (atomic_dec_and_test(&rdev->nr_pending) && faulty)
+ set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
+}
+
struct mdk_personality_s
{
char *name;
-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
