diff options
author | NeilBrown <neilb@suse.de> | 2006-09-01 00:27:36 -0400 |
---|---|---|
committer | Linus Torvalds <torvalds@g5.osdl.org> | 2006-09-01 14:39:08 -0400 |
commit | ddac7c7e3a0fe9cfdcef0de24476b8d69f8cf3e7 (patch) | |
tree | b75845fa2424284342869566b3ed5233abecde92 /drivers | |
parent | df9ecaba3f152d1ea79f2a5e0b87505e03f47590 (diff) |
[PATCH] md: Fix issues with referencing rdev in md/raid1
We need to be careful when referencing mirrors[i].rdev. It can disappear
under us at various times.
So:
fix a couple of problem places.
comment a couple of non-problem places
move an 'atomic_add' which deferences rdev down a little
way to some where where it is sure to not be NULL.
Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Diffstat (limited to 'drivers')
-rw-r--r-- | drivers/md/raid1.c | 57 |
1 files changed, 35 insertions, 22 deletions
diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c index 87bfe9e7d8ca..3b4d69c05623 100644 --- a/drivers/md/raid1.c +++ b/drivers/md/raid1.c | |||
@@ -930,10 +930,13 @@ static void status(struct seq_file *seq, mddev_t *mddev) | |||
930 | 930 | ||
931 | seq_printf(seq, " [%d/%d] [", conf->raid_disks, | 931 | seq_printf(seq, " [%d/%d] [", conf->raid_disks, |
932 | conf->working_disks); | 932 | conf->working_disks); |
933 | for (i = 0; i < conf->raid_disks; i++) | 933 | rcu_read_lock(); |
934 | for (i = 0; i < conf->raid_disks; i++) { | ||
935 | mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev); | ||
934 | seq_printf(seq, "%s", | 936 | seq_printf(seq, "%s", |
935 | conf->mirrors[i].rdev && | 937 | rdev && test_bit(In_sync, &rdev->flags) ? "U" : "_"); |
936 | test_bit(In_sync, &conf->mirrors[i].rdev->flags) ? "U" : "_"); | 938 | } |
939 | rcu_read_unlock(); | ||
937 | seq_printf(seq, "]"); | 940 | seq_printf(seq, "]"); |
938 | } | 941 | } |
939 | 942 | ||
@@ -975,7 +978,6 @@ static void error(mddev_t *mddev, mdk_rdev_t *rdev) | |||
975 | static void print_conf(conf_t *conf) | 978 | static void print_conf(conf_t *conf) |
976 | { | 979 | { |
977 | int i; | 980 | int i; |
978 | mirror_info_t *tmp; | ||
979 | 981 | ||
980 | printk("RAID1 conf printout:\n"); | 982 | printk("RAID1 conf printout:\n"); |
981 | if (!conf) { | 983 | if (!conf) { |
@@ -985,14 +987,17 @@ static void print_conf(conf_t *conf) | |||
985 | printk(" --- wd:%d rd:%d\n", conf->working_disks, | 987 | printk(" --- wd:%d rd:%d\n", conf->working_disks, |
986 | conf->raid_disks); | 988 | conf->raid_disks); |
987 | 989 | ||
990 | rcu_read_lock(); | ||
988 | for (i = 0; i < conf->raid_disks; i++) { | 991 | for (i = 0; i < conf->raid_disks; i++) { |
989 | char b[BDEVNAME_SIZE]; | 992 | char b[BDEVNAME_SIZE]; |
990 | tmp = conf->mirrors + i; | 993 | mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev); |
991 | if (tmp->rdev) | 994 | if (rdev) |
992 | printk(" disk %d, wo:%d, o:%d, dev:%s\n", | 995 | printk(" disk %d, wo:%d, o:%d, dev:%s\n", |
993 | i, !test_bit(In_sync, &tmp->rdev->flags), !test_bit(Faulty, &tmp->rdev->flags), | 996 | i, !test_bit(In_sync, &rdev->flags), |
994 | bdevname(tmp->rdev->bdev,b)); | 997 | !test_bit(Faulty, &rdev->flags), |
998 | bdevname(rdev->bdev,b)); | ||
995 | } | 999 | } |
1000 | rcu_read_unlock(); | ||
996 | } | 1001 | } |
997 | 1002 | ||
998 | static void close_sync(conf_t *conf) | 1003 | static void close_sync(conf_t *conf) |
@@ -1008,20 +1013,20 @@ static int raid1_spare_active(mddev_t *mddev) | |||
1008 | { | 1013 | { |
1009 | int i; | 1014 | int i; |
1010 | conf_t *conf = mddev->private; | 1015 | conf_t *conf = mddev->private; |
1011 | mirror_info_t *tmp; | ||
1012 | 1016 | ||
1013 | /* | 1017 | /* |
1014 | * Find all failed disks within the RAID1 configuration | 1018 | * Find all failed disks within the RAID1 configuration |
1015 | * and mark them readable | 1019 | * and mark them readable. |
1020 | * Called under mddev lock, so rcu protection not needed. | ||
1016 | */ | 1021 | */ |
1017 | for (i = 0; i < conf->raid_disks; i++) { | 1022 | for (i = 0; i < conf->raid_disks; i++) { |
1018 | tmp = conf->mirrors + i; | 1023 | mdk_rdev_t *rdev = conf->mirrors[i].rdev; |
1019 | if (tmp->rdev | 1024 | if (rdev |
1020 | && !test_bit(Faulty, &tmp->rdev->flags) | 1025 | && !test_bit(Faulty, &rdev->flags) |
1021 | && !test_bit(In_sync, &tmp->rdev->flags)) { | 1026 | && !test_bit(In_sync, &rdev->flags)) { |
1022 | conf->working_disks++; | 1027 | conf->working_disks++; |
1023 | mddev->degraded--; | 1028 | mddev->degraded--; |
1024 | set_bit(In_sync, &tmp->rdev->flags); | 1029 | set_bit(In_sync, &rdev->flags); |
1025 | } | 1030 | } |
1026 | } | 1031 | } |
1027 | 1032 | ||
@@ -1237,7 +1242,7 @@ static void sync_request_write(mddev_t *mddev, r1bio_t *r1_bio) | |||
1237 | /* ouch - failed to read all of that. | 1242 | /* ouch - failed to read all of that. |
1238 | * Try some synchronous reads of other devices to get | 1243 | * Try some synchronous reads of other devices to get |
1239 | * good data, much like with normal read errors. Only | 1244 | * good data, much like with normal read errors. Only |
1240 | * read into the pages we already have so they we don't | 1245 | * read into the pages we already have so we don't |
1241 | * need to re-issue the read request. | 1246 | * need to re-issue the read request. |
1242 | * We don't need to freeze the array, because being in an | 1247 | * We don't need to freeze the array, because being in an |
1243 | * active sync request, there is no normal IO, and | 1248 | * active sync request, there is no normal IO, and |
@@ -1257,6 +1262,10 @@ static void sync_request_write(mddev_t *mddev, r1bio_t *r1_bio) | |||
1257 | s = PAGE_SIZE >> 9; | 1262 | s = PAGE_SIZE >> 9; |
1258 | do { | 1263 | do { |
1259 | if (r1_bio->bios[d]->bi_end_io == end_sync_read) { | 1264 | if (r1_bio->bios[d]->bi_end_io == end_sync_read) { |
1265 | /* No rcu protection needed here devices | ||
1266 | * can only be removed when no resync is | ||
1267 | * active, and resync is currently active | ||
1268 | */ | ||
1260 | rdev = conf->mirrors[d].rdev; | 1269 | rdev = conf->mirrors[d].rdev; |
1261 | if (sync_page_io(rdev->bdev, | 1270 | if (sync_page_io(rdev->bdev, |
1262 | sect + rdev->data_offset, | 1271 | sect + rdev->data_offset, |
@@ -1463,6 +1472,11 @@ static void raid1d(mddev_t *mddev) | |||
1463 | s = PAGE_SIZE >> 9; | 1472 | s = PAGE_SIZE >> 9; |
1464 | 1473 | ||
1465 | do { | 1474 | do { |
1475 | /* Note: no rcu protection needed here | ||
1476 | * as this is synchronous in the raid1d thread | ||
1477 | * which is the thread that might remove | ||
1478 | * a device. If raid1d ever becomes multi-threaded.... | ||
1479 | */ | ||
1466 | rdev = conf->mirrors[d].rdev; | 1480 | rdev = conf->mirrors[d].rdev; |
1467 | if (rdev && | 1481 | if (rdev && |
1468 | test_bit(In_sync, &rdev->flags) && | 1482 | test_bit(In_sync, &rdev->flags) && |
@@ -1486,7 +1500,6 @@ static void raid1d(mddev_t *mddev) | |||
1486 | d = conf->raid_disks; | 1500 | d = conf->raid_disks; |
1487 | d--; | 1501 | d--; |
1488 | rdev = conf->mirrors[d].rdev; | 1502 | rdev = conf->mirrors[d].rdev; |
1489 | atomic_add(s, &rdev->corrected_errors); | ||
1490 | if (rdev && | 1503 | if (rdev && |
1491 | test_bit(In_sync, &rdev->flags)) { | 1504 | test_bit(In_sync, &rdev->flags)) { |
1492 | if (sync_page_io(rdev->bdev, | 1505 | if (sync_page_io(rdev->bdev, |
@@ -1509,9 +1522,11 @@ static void raid1d(mddev_t *mddev) | |||
1509 | s<<9, conf->tmppage, READ) == 0) | 1522 | s<<9, conf->tmppage, READ) == 0) |
1510 | /* Well, this device is dead */ | 1523 | /* Well, this device is dead */ |
1511 | md_error(mddev, rdev); | 1524 | md_error(mddev, rdev); |
1512 | else | 1525 | else { |
1526 | atomic_add(s, &rdev->corrected_errors); | ||
1513 | printk(KERN_INFO "raid1:%s: read error corrected (%d sectors at %llu on %s)\n", | 1527 | printk(KERN_INFO "raid1:%s: read error corrected (%d sectors at %llu on %s)\n", |
1514 | mdname(mddev), s, (unsigned long long)(sect + rdev->data_offset), bdevname(rdev->bdev, b)); | 1528 | mdname(mddev), s, (unsigned long long)(sect + rdev->data_offset), bdevname(rdev->bdev, b)); |
1529 | } | ||
1515 | } | 1530 | } |
1516 | } | 1531 | } |
1517 | } else { | 1532 | } else { |
@@ -1787,19 +1802,17 @@ static sector_t sync_request(mddev_t *mddev, sector_t sector_nr, int *skipped, i | |||
1787 | for (i=0; i<conf->raid_disks; i++) { | 1802 | for (i=0; i<conf->raid_disks; i++) { |
1788 | bio = r1_bio->bios[i]; | 1803 | bio = r1_bio->bios[i]; |
1789 | if (bio->bi_end_io == end_sync_read) { | 1804 | if (bio->bi_end_io == end_sync_read) { |
1790 | md_sync_acct(conf->mirrors[i].rdev->bdev, nr_sectors); | 1805 | md_sync_acct(bio->bi_bdev, nr_sectors); |
1791 | generic_make_request(bio); | 1806 | generic_make_request(bio); |
1792 | } | 1807 | } |
1793 | } | 1808 | } |
1794 | } else { | 1809 | } else { |
1795 | atomic_set(&r1_bio->remaining, 1); | 1810 | atomic_set(&r1_bio->remaining, 1); |
1796 | bio = r1_bio->bios[r1_bio->read_disk]; | 1811 | bio = r1_bio->bios[r1_bio->read_disk]; |
1797 | md_sync_acct(conf->mirrors[r1_bio->read_disk].rdev->bdev, | 1812 | md_sync_acct(bio->bi_bdev, nr_sectors); |
1798 | nr_sectors); | ||
1799 | generic_make_request(bio); | 1813 | generic_make_request(bio); |
1800 | 1814 | ||
1801 | } | 1815 | } |
1802 | |||
1803 | return nr_sectors; | 1816 | return nr_sectors; |
1804 | } | 1817 | } |
1805 | 1818 | ||