From 506c9e44a85f6a79fc0643f2d2498ab6cda3d3f8 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:26 +1100 Subject: md: allow non-privileged uses to GET_*_INFO about raid arrays. The info is already available in /proc/mdstat and /sys/block in an accessible form so there is no point in putting a road-block in the ioctl for information gathering. Signed-off-by: NeilBrown --- drivers/md/md.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/md.c b/drivers/md/md.c index f47f1f8ac44b..004d8e5b2a57 100644 --- a/drivers/md/md.c +++ b/drivers/md/md.c @@ -6054,8 +6054,15 @@ static int md_ioctl(struct block_device *bdev, fmode_t mode, struct mddev *mddev = NULL; int ro; - if (!capable(CAP_SYS_ADMIN)) - return -EACCES; + switch (cmd) { + case RAID_VERSION: + case GET_ARRAY_INFO: + case GET_DISK_INFO: + break; + default: + if (!capable(CAP_SYS_ADMIN)) + return -EACCES; + } /* * Commands dealing with the RAID driver but not any -- cgit v1.2.3 From 2e61ebbcc45438899235d7a39f17291cb24e746c Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:50 +1100 Subject: md/bitmap: daemon_work cleanup. We have a variable 'mddev' in this function, but repeatedly get the same value by dereferencing bitmap->mddev. There is room for simplification here... Signed-off-by: NeilBrown --- drivers/md/bitmap.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/bitmap.c b/drivers/md/bitmap.c index 6d03774b176e..048eec751474 100644 --- a/drivers/md/bitmap.c +++ b/drivers/md/bitmap.c @@ -1149,12 +1149,12 @@ void bitmap_daemon_work(struct mddev *mddev) return; } if (time_before(jiffies, bitmap->daemon_lastrun - + bitmap->mddev->bitmap_info.daemon_sleep)) + + mddev->bitmap_info.daemon_sleep)) goto done; bitmap->daemon_lastrun = jiffies; if (bitmap->allclean) { - bitmap->mddev->thread->timeout = MAX_SCHEDULE_TIMEOUT; + mddev->thread->timeout = MAX_SCHEDULE_TIMEOUT; goto done; } bitmap->allclean = 1; @@ -1206,7 +1206,7 @@ void bitmap_daemon_work(struct mddev *mddev) * sure that events_cleared is up-to-date. */ if (bitmap->need_sync && - bitmap->mddev->bitmap_info.external == 0) { + mddev->bitmap_info.external == 0) { bitmap_super_t *sb; bitmap->need_sync = 0; sb = kmap_atomic(bitmap->sb_page, KM_USER0); @@ -1270,8 +1270,8 @@ void bitmap_daemon_work(struct mddev *mddev) done: if (bitmap->allclean == 0) - bitmap->mddev->thread->timeout = - bitmap->mddev->bitmap_info.daemon_sleep; + mddev->thread->timeout = + mddev->bitmap_info.daemon_sleep; mutex_unlock(&mddev->bitmap_info.mutex); } -- cgit v1.2.3 From 908f4fbd265733310c17ecc906299846b5dac44a Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:50 +1100 Subject: md/raid5: be more thorough in calculating 'degraded' value. When an array is being reshaped to change the number of devices, the two halves can be differently degraded. e.g. one could be missing a device and the other not. So we need to be more careful about calculating the 'degraded' attribute. Instead of just inc/dec at appropriate times, perform a full re-calculation examining both possible cases. This doesn't happen often so it not a big cost, and we already have most of the code to do it. Signed-off-by: NeilBrown --- drivers/md/raid5.c | 63 ++++++++++++++++++++++++++++-------------------------- 1 file changed, 33 insertions(+), 30 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 858fdbb7eb07..a68c49250ed5 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -370,12 +370,10 @@ static struct stripe_head *__find_stripe(struct r5conf *conf, sector_t sector, * of the two sections, and some non-in_sync devices may * be insync in the section most affected by failed devices. */ -static int has_failed(struct r5conf *conf) +static int calc_degraded(struct r5conf *conf) { - int degraded; + int degraded, degraded2; int i; - if (conf->mddev->reshape_position == MaxSector) - return conf->mddev->degraded > conf->max_degraded; rcu_read_lock(); degraded = 0; @@ -399,14 +397,14 @@ static int has_failed(struct r5conf *conf) degraded++; } rcu_read_unlock(); - if (degraded > conf->max_degraded) - return 1; + if (conf->raid_disks == conf->previous_raid_disks) + return degraded; rcu_read_lock(); - degraded = 0; + degraded2 = 0; for (i = 0; i < conf->raid_disks; i++) { struct md_rdev *rdev = rcu_dereference(conf->disks[i].rdev); if (!rdev || test_bit(Faulty, &rdev->flags)) - degraded++; + degraded2++; else if (test_bit(In_sync, &rdev->flags)) ; else @@ -416,9 +414,22 @@ static int has_failed(struct r5conf *conf) * almost certainly hasn't. */ if (conf->raid_disks <= conf->previous_raid_disks) - degraded++; + degraded2++; } rcu_read_unlock(); + if (degraded2 > degraded) + return degraded2; + return degraded; +} + +static int has_failed(struct r5conf *conf) +{ + int degraded; + + if (conf->mddev->reshape_position == MaxSector) + return conf->mddev->degraded > conf->max_degraded; + + degraded = calc_degraded(conf); if (degraded > conf->max_degraded) return 1; return 0; @@ -1724,18 +1735,15 @@ static void error(struct mddev *mddev, struct md_rdev *rdev) { char b[BDEVNAME_SIZE]; struct r5conf *conf = mddev->private; + unsigned long flags; pr_debug("raid456: error called\n"); - if (test_and_clear_bit(In_sync, &rdev->flags)) { - unsigned long flags; - spin_lock_irqsave(&conf->device_lock, flags); - mddev->degraded++; - spin_unlock_irqrestore(&conf->device_lock, flags); - /* - * if recovery was running, make sure it aborts. - */ - set_bit(MD_RECOVERY_INTR, &mddev->recovery); - } + spin_lock_irqsave(&conf->device_lock, flags); + clear_bit(In_sync, &rdev->flags); + mddev->degraded = calc_degraded(conf); + spin_unlock_irqrestore(&conf->device_lock, flags); + set_bit(MD_RECOVERY_INTR, &mddev->recovery); + set_bit(Blocked, &rdev->flags); set_bit(Faulty, &rdev->flags); set_bit(MD_CHANGE_DEVS, &mddev->flags); @@ -4852,8 +4860,7 @@ static int run(struct mddev *mddev) dirty_parity_disks++; } - mddev->degraded = (max(conf->raid_disks, conf->previous_raid_disks) - - working_disks); + mddev->degraded = calc_degraded(conf); if (has_failed(conf)) { printk(KERN_ERR "md/raid:%s: not enough operational devices" @@ -5025,7 +5032,7 @@ static int raid5_spare_active(struct mddev *mddev) } } spin_lock_irqsave(&conf->device_lock, flags); - mddev->degraded -= count; + mddev->degraded = calc_degraded(conf); spin_unlock_irqrestore(&conf->device_lock, flags); print_raid5_conf(conf); return count; @@ -5286,8 +5293,7 @@ static int raid5_start_reshape(struct mddev *mddev) * pre and post number of devices. */ spin_lock_irqsave(&conf->device_lock, flags); - mddev->degraded += (conf->raid_disks - conf->previous_raid_disks) - - added_devices; + mddev->degraded = calc_degraded(conf); spin_unlock_irqrestore(&conf->device_lock, flags); } mddev->raid_disks = conf->raid_disks; @@ -5356,12 +5362,9 @@ static void raid5_finish_reshape(struct mddev *mddev) revalidate_disk(mddev->gendisk); } else { int d; - mddev->degraded = conf->raid_disks; - for (d = 0; d < conf->raid_disks ; d++) - if (conf->disks[d].rdev && - test_bit(In_sync, - &conf->disks[d].rdev->flags)) - mddev->degraded--; + spin_lock_irq(&conf->device_lock); + mddev->degraded = calc_degraded(conf); + spin_unlock_irq(&conf->device_lock); for (d = conf->raid_disks ; d < conf->raid_disks - mddev->delta_disks; d++) { -- cgit v1.2.3 From 915c420ddfa3eb22a0dbdb7a8e0ecf020c31961f Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:51 +1100 Subject: md/bitmap: be more consistent when setting new bits in memory bitmap. For each active region corresponding to a bit in the bitmap with have a 14bit counter (and some flags). This counts number of active writes + bit in the on-disk bitmap + delay-needed. The "delay-needed" is because we always want a delay before clearing a bit. So the number here is normally number of active writes plus 2. If there have been no writes for a while, we drop to 1. If still no writes we clear the bit and drop to 0. So for consistency, when setting bit from the on-disk bitmap or by request from user-space it is best to set the counter to '2' to start with. In particular we might also set the NEEDED_MASK flag at this time, and in all other cases NEEDED_MASK is only set when the counter is 2 or more. Signed-off-by: NeilBrown --- drivers/md/bitmap.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/md') diff --git a/drivers/md/bitmap.c b/drivers/md/bitmap.c index 048eec751474..cdf36b1e9aa6 100644 --- a/drivers/md/bitmap.c +++ b/drivers/md/bitmap.c @@ -1587,7 +1587,7 @@ static void bitmap_set_memory_bits(struct bitmap *bitmap, sector_t offset, int n } if (!*bmc) { struct page *page; - *bmc = 1 | (needed ? NEEDED_MASK : 0); + *bmc = 2 | (needed ? NEEDED_MASK : 0); bitmap_count_page(bitmap, offset, 1); page = filemap_get_page(bitmap, offset >> CHUNK_BLOCK_SHIFT(bitmap)); set_page_attr(bitmap, page, BITMAP_PAGE_PENDING); -- cgit v1.2.3 From 476a7abb9b00adfe6bf70e82800367319ab8078b Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:51 +1100 Subject: md: remove test for duplicate device when setting slot number. When setting the slot number on a device in an active array we currently check that the number is not already in use. We then call into the personality's hot_add_disk function which performs the same test and returns the same error. Thus the common test is not needed. As we will shortly be changing some personalities to allow duplicates in some cases (to support hot-replace), the common test will become inconvenient. So remove the common test. Reviewed-by: Dan Williams Signed-off-by: NeilBrown --- drivers/md/md.c | 5 ----- 1 file changed, 5 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/md.c b/drivers/md/md.c index 004d8e5b2a57..d51c688d9c94 100644 --- a/drivers/md/md.c +++ b/drivers/md/md.c @@ -2697,7 +2697,6 @@ slot_store(struct md_rdev *rdev, const char *buf, size_t len) set_bit(MD_RECOVERY_NEEDED, &rdev->mddev->recovery); md_wakeup_thread(rdev->mddev->thread); } else if (rdev->mddev->pers) { - struct md_rdev *rdev2; /* Activating a spare .. or possibly reactivating * if we ever get bitmaps working here. */ @@ -2711,10 +2710,6 @@ slot_store(struct md_rdev *rdev, const char *buf, size_t len) if (rdev->mddev->pers->hot_add_disk == NULL) return -EINVAL; - list_for_each_entry(rdev2, &rdev->mddev->disks, same_set) - if (rdev2->raid_disk == slot) - return -EEXIST; - if (slot >= rdev->mddev->raid_disks && slot >= rdev->mddev->raid_disks + rdev->mddev->delta_disks) return -ENOSPC; -- cgit v1.2.3 From b8321b68d1445f308324517e45fb0a5c2b48e271 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:51 +1100 Subject: md: change hot_remove_disk to take an rdev rather than a number. Soon an array will be able to have multiple devices with the same raid_disk number (an original and a replacement). So removing a device based on the number won't work. So pass the actual device handle instead. Reviewed-by: Dan Williams Signed-off-by: NeilBrown --- drivers/md/md.c | 6 +++--- drivers/md/md.h | 2 +- drivers/md/multipath.c | 7 +++---- drivers/md/raid1.c | 7 +++---- drivers/md/raid10.c | 7 +++---- drivers/md/raid5.c | 10 +++++----- 6 files changed, 18 insertions(+), 21 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/md.c b/drivers/md/md.c index d51c688d9c94..0e2288824938 100644 --- a/drivers/md/md.c +++ b/drivers/md/md.c @@ -2689,7 +2689,7 @@ slot_store(struct md_rdev *rdev, const char *buf, size_t len) if (rdev->mddev->pers->hot_remove_disk == NULL) return -EINVAL; err = rdev->mddev->pers-> - hot_remove_disk(rdev->mddev, rdev->raid_disk); + hot_remove_disk(rdev->mddev, rdev); if (err) return err; sysfs_unlink_rdev(rdev->mddev, rdev); @@ -7340,7 +7340,7 @@ static int remove_and_add_spares(struct mddev *mddev) ! test_bit(In_sync, &rdev->flags)) && atomic_read(&rdev->nr_pending)==0) { if (mddev->pers->hot_remove_disk( - mddev, rdev->raid_disk)==0) { + mddev, rdev) == 0) { sysfs_unlink_rdev(mddev, rdev); rdev->raid_disk = -1; } @@ -7477,7 +7477,7 @@ void md_check_recovery(struct mddev *mddev) test_bit(Faulty, &rdev->flags) && atomic_read(&rdev->nr_pending)==0) { if (mddev->pers->hot_remove_disk( - mddev, rdev->raid_disk)==0) { + mddev, rdev) == 0) { sysfs_unlink_rdev(mddev, rdev); rdev->raid_disk = -1; } diff --git a/drivers/md/md.h b/drivers/md/md.h index cf742d9306ec..670c10e6b484 100644 --- a/drivers/md/md.h +++ b/drivers/md/md.h @@ -428,7 +428,7 @@ struct md_personality */ void (*error_handler)(struct mddev *mddev, struct md_rdev *rdev); int (*hot_add_disk) (struct mddev *mddev, struct md_rdev *rdev); - int (*hot_remove_disk) (struct mddev *mddev, int number); + int (*hot_remove_disk) (struct mddev *mddev, struct md_rdev *rdev); int (*spare_active) (struct mddev *mddev); sector_t (*sync_request)(struct mddev *mddev, sector_t sector_nr, int *skipped, int go_faster); int (*resize) (struct mddev *mddev, sector_t sectors); diff --git a/drivers/md/multipath.c b/drivers/md/multipath.c index 5899246fa37e..a222f516660e 100644 --- a/drivers/md/multipath.c +++ b/drivers/md/multipath.c @@ -292,17 +292,16 @@ static int multipath_add_disk(struct mddev *mddev, struct md_rdev *rdev) return err; } -static int multipath_remove_disk(struct mddev *mddev, int number) +static int multipath_remove_disk(struct mddev *mddev, struct md_rdev *rdev) { struct mpconf *conf = mddev->private; int err = 0; - struct md_rdev *rdev; + int number = rdev->raid_disk; struct multipath_info *p = conf->multipaths + number; print_multipath_conf(conf); - rdev = p->rdev; - if (rdev) { + if (rdev == p->rdev) { if (test_bit(In_sync, &rdev->flags) || atomic_read(&rdev->nr_pending)) { printk(KERN_ERR "hot-remove-disk, slot %d is identified" diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c index ede2461e79c5..4317f8de143a 100644 --- a/drivers/md/raid1.c +++ b/drivers/md/raid1.c @@ -1327,16 +1327,15 @@ static int raid1_add_disk(struct mddev *mddev, struct md_rdev *rdev) return err; } -static int raid1_remove_disk(struct mddev *mddev, int number) +static int raid1_remove_disk(struct mddev *mddev, struct md_rdev *rdev) { struct r1conf *conf = mddev->private; int err = 0; - struct md_rdev *rdev; + int number = rdev->raid_disk; struct mirror_info *p = conf->mirrors+ number; print_conf(conf); - rdev = p->rdev; - if (rdev) { + if (rdev == p->rdev) { if (test_bit(In_sync, &rdev->flags) || atomic_read(&rdev->nr_pending)) { err = -EBUSY; diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index 685ddf325ee4..f5088dda4dca 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c @@ -1385,16 +1385,15 @@ static int raid10_add_disk(struct mddev *mddev, struct md_rdev *rdev) return err; } -static int raid10_remove_disk(struct mddev *mddev, int number) +static int raid10_remove_disk(struct mddev *mddev, struct md_rdev *rdev) { struct r10conf *conf = mddev->private; int err = 0; - struct md_rdev *rdev; + int number = rdev->raid_disk; struct mirror_info *p = conf->mirrors+ number; print_conf(conf); - rdev = p->rdev; - if (rdev) { + if (rdev == p->rdev) { if (test_bit(In_sync, &rdev->flags) || atomic_read(&rdev->nr_pending)) { err = -EBUSY; diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index a68c49250ed5..6b9fc58e8f2d 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -5038,16 +5038,15 @@ static int raid5_spare_active(struct mddev *mddev) return count; } -static int raid5_remove_disk(struct mddev *mddev, int number) +static int raid5_remove_disk(struct mddev *mddev, struct md_rdev *rdev) { struct r5conf *conf = mddev->private; int err = 0; - struct md_rdev *rdev; + int number = rdev->raid_disk; struct disk_info *p = conf->disks + number; print_raid5_conf(conf); - rdev = p->rdev; - if (rdev) { + if (rdev == p->rdev) { if (number >= conf->raid_disks && conf->reshape_progress == MaxSector) clear_bit(In_sync, &rdev->flags); @@ -5369,7 +5368,8 @@ static void raid5_finish_reshape(struct mddev *mddev) d < conf->raid_disks - mddev->delta_disks; d++) { struct md_rdev *rdev = conf->disks[d].rdev; - if (rdev && raid5_remove_disk(mddev, d) == 0) { + if (rdev && + raid5_remove_disk(mddev, rdev) == 0) { sysfs_unlink_rdev(mddev, rdev); rdev->raid_disk = -1; } -- cgit v1.2.3 From 2d78f8c451785f030ac1676a18691896b59c69d8 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:51 +1100 Subject: md: create externally visible flags for supporting hot-replace. hot-replace is a feature being added to md which will allow a device to be replaced without removing it from the array first. With hot-replace a spare can be activated and recovery can start while the original device is still in place, thus allowing a transition from an unreliable device to a reliable device without leaving the array degraded during the transition. It can also be use when the original device is still reliable but it not wanted for some reason. This will eventually be supported in RAID4/5/6 and RAID10. This patch adds a super-block flag to distinguish the replacement device. If an old kernel sees this flag it will reject the device. It also adds two per-device flags which are viewable and settable via sysfs. "want_replacement" can be set to request that a device be replaced. "replacement" is set to show that this device is replacing another device. The "rd%d" links in /sys/block/mdXx/md only apply to the original device, not the replacement. We currently don't make links for the replacement - there doesn't seem to be a need. Signed-off-by: NeilBrown --- Documentation/md.txt | 22 ++++++++++--- drivers/md/md.c | 55 +++++++++++++++++++++++++++++++- drivers/md/md.h | 80 ++++++++++++++++++++++++++++------------------- include/linux/raid/md_p.h | 7 +++-- 4 files changed, 125 insertions(+), 39 deletions(-) (limited to 'drivers/md') diff --git a/Documentation/md.txt b/Documentation/md.txt index fc94770f44ab..993fba37b7d1 100644 --- a/Documentation/md.txt +++ b/Documentation/md.txt @@ -357,14 +357,14 @@ Each directory contains: written to, that device. state - A file recording the current state of the device in the array + A file recording the current state of the device in the array which can be a comma separated list of faulty - device has been kicked from active use due to - a detected fault or it has unacknowledged bad - blocks + a detected fault, or it has unacknowledged bad + blocks in_sync - device is a fully in-sync member of the array writemostly - device will only be subject to read - requests if there are no other options. + requests if there are no other options. This applies only to raid1 arrays. blocked - device has failed, and the failure hasn't been acknowledged yet by the metadata handler. @@ -374,6 +374,13 @@ Each directory contains: This includes spares that are in the process of being recovered to write_error - device has ever seen a write error. + want_replacement - device is (mostly) working but probably + should be replaced, either due to errors or + due to user request. + replacement - device is a replacement for another active + device with same raid_disk. + + This list may grow in future. This can be written to. Writing "faulty" simulates a failure on the device. @@ -386,6 +393,13 @@ Each directory contains: Writing "in_sync" sets the in_sync flag. Writing "write_error" sets writeerrorseen flag. Writing "-write_error" clears writeerrorseen flag. + Writing "want_replacement" is allowed at any time except to a + replacement device or a spare. It sets the flag. + Writing "-want_replacement" is allowed at any time. It clears + the flag. + Writing "replacement" or "-replacement" is only allowed before + starting the array. It sets or clears the flag. + This file responds to select/poll. Any change to 'faulty' or 'blocked' causes an event. diff --git a/drivers/md/md.c b/drivers/md/md.c index 0e2288824938..be569eb41a93 100644 --- a/drivers/md/md.c +++ b/drivers/md/md.c @@ -1714,6 +1714,8 @@ static int super_1_validate(struct mddev *mddev, struct md_rdev *rdev) } if (sb->devflags & WriteMostly1) set_bit(WriteMostly, &rdev->flags); + if (le32_to_cpu(sb->feature_map) & MD_FEATURE_REPLACEMENT) + set_bit(Replacement, &rdev->flags); } else /* MULTIPATH are always insync */ set_bit(In_sync, &rdev->flags); @@ -1767,6 +1769,9 @@ static void super_1_sync(struct mddev *mddev, struct md_rdev *rdev) sb->recovery_offset = cpu_to_le64(rdev->recovery_offset); } + if (test_bit(Replacement, &rdev->flags)) + sb->feature_map |= + cpu_to_le32(MD_FEATURE_REPLACEMENT); if (mddev->reshape_position != MaxSector) { sb->feature_map |= cpu_to_le32(MD_FEATURE_RESHAPE_ACTIVE); @@ -2560,6 +2565,15 @@ state_show(struct md_rdev *rdev, char *page) len += sprintf(page+len, "%swrite_error", sep); sep = ","; } + if (test_bit(WantReplacement, &rdev->flags)) { + len += sprintf(page+len, "%swant_replacement", sep); + sep = ","; + } + if (test_bit(Replacement, &rdev->flags)) { + len += sprintf(page+len, "%sreplacement", sep); + sep = ","; + } + return len+sprintf(page+len, "\n"); } @@ -2628,6 +2642,42 @@ state_store(struct md_rdev *rdev, const char *buf, size_t len) } else if (cmd_match(buf, "-write_error")) { clear_bit(WriteErrorSeen, &rdev->flags); err = 0; + } else if (cmd_match(buf, "want_replacement")) { + /* Any non-spare device that is not a replacement can + * become want_replacement at any time, but we then need to + * check if recovery is needed. + */ + if (rdev->raid_disk >= 0 && + !test_bit(Replacement, &rdev->flags)) + set_bit(WantReplacement, &rdev->flags); + set_bit(MD_RECOVERY_NEEDED, &rdev->mddev->recovery); + md_wakeup_thread(rdev->mddev->thread); + err = 0; + } else if (cmd_match(buf, "-want_replacement")) { + /* Clearing 'want_replacement' is always allowed. + * Once replacements starts it is too late though. + */ + err = 0; + clear_bit(WantReplacement, &rdev->flags); + } else if (cmd_match(buf, "replacement")) { + /* Can only set a device as a replacement when array has not + * yet been started. Once running, replacement is automatic + * from spares, or by assigning 'slot'. + */ + if (rdev->mddev->pers) + err = -EBUSY; + else { + set_bit(Replacement, &rdev->flags); + err = 0; + } + } else if (cmd_match(buf, "-replacement")) { + /* Similarly, can only clear Replacement before start */ + if (rdev->mddev->pers) + err = -EBUSY; + else { + clear_bit(Replacement, &rdev->flags); + err = 0; + } } if (!err) sysfs_notify_dirent_safe(rdev->sysfs_state); @@ -6717,8 +6767,11 @@ static int md_seq_show(struct seq_file *seq, void *v) if (test_bit(Faulty, &rdev->flags)) { seq_printf(seq, "(F)"); continue; - } else if (rdev->raid_disk < 0) + } + if (rdev->raid_disk < 0) seq_printf(seq, "(S)"); /* spare */ + if (test_bit(Replacement, &rdev->flags)) + seq_printf(seq, "(R)"); sectors += rdev->sectors; } diff --git a/drivers/md/md.h b/drivers/md/md.h index 670c10e6b484..44c63dfeeb2b 100644 --- a/drivers/md/md.h +++ b/drivers/md/md.h @@ -72,34 +72,7 @@ struct md_rdev { * This reduces the burden of testing multiple flags in many cases */ - unsigned long flags; -#define Faulty 1 /* device is known to have a fault */ -#define In_sync 2 /* device is in_sync with rest of array */ -#define WriteMostly 4 /* Avoid reading if at all possible */ -#define AutoDetected 7 /* added by auto-detect */ -#define Blocked 8 /* An error occurred but has not yet - * been acknowledged by the metadata - * handler, so don't allow writes - * until it is cleared */ -#define WriteErrorSeen 9 /* A write error has been seen on this - * device - */ -#define FaultRecorded 10 /* Intermediate state for clearing - * Blocked. The Fault is/will-be - * recorded in the metadata, but that - * metadata hasn't been stored safely - * on disk yet. - */ -#define BlockedBadBlocks 11 /* A writer is blocked because they - * found an unacknowledged bad-block. - * This can safely be cleared at any - * time, and the writer will re-check. - * It may be set at any time, and at - * worst the writer will timeout and - * re-check. So setting it as - * accurately as possible is good, but - * not absolutely critical. - */ + unsigned long flags; /* bit set of 'enum flag_bits' bits. */ wait_queue_head_t blocked_wait; int desc_nr; /* descriptor index in the superblock */ @@ -152,6 +125,44 @@ struct md_rdev { sector_t size; /* in sectors */ } badblocks; }; +enum flag_bits { + Faulty, /* device is known to have a fault */ + In_sync, /* device is in_sync with rest of array */ + WriteMostly, /* Avoid reading if at all possible */ + AutoDetected, /* added by auto-detect */ + Blocked, /* An error occurred but has not yet + * been acknowledged by the metadata + * handler, so don't allow writes + * until it is cleared */ + WriteErrorSeen, /* A write error has been seen on this + * device + */ + FaultRecorded, /* Intermediate state for clearing + * Blocked. The Fault is/will-be + * recorded in the metadata, but that + * metadata hasn't been stored safely + * on disk yet. + */ + BlockedBadBlocks, /* A writer is blocked because they + * found an unacknowledged bad-block. + * This can safely be cleared at any + * time, and the writer will re-check. + * It may be set at any time, and at + * worst the writer will timeout and + * re-check. So setting it as + * accurately as possible is good, but + * not absolutely critical. + */ + WantReplacement, /* This device is a candidate to be + * hot-replaced, either because it has + * reported some faults, or because + * of explicit request. + */ + Replacement, /* This device is a replacement for + * a want_replacement device with same + * raid_disk number. + */ +}; #define BB_LEN_MASK (0x00000000000001FFULL) #define BB_OFFSET_MASK (0x7FFFFFFFFFFFFE00ULL) @@ -482,15 +493,20 @@ static inline char * mdname (struct mddev * mddev) static inline int sysfs_link_rdev(struct mddev *mddev, struct md_rdev *rdev) { char nm[20]; - sprintf(nm, "rd%d", rdev->raid_disk); - return sysfs_create_link(&mddev->kobj, &rdev->kobj, nm); + if (!test_bit(Replacement, &rdev->flags)) { + sprintf(nm, "rd%d", rdev->raid_disk); + return sysfs_create_link(&mddev->kobj, &rdev->kobj, nm); + } else + return 0; } static inline void sysfs_unlink_rdev(struct mddev *mddev, struct md_rdev *rdev) { char nm[20]; - sprintf(nm, "rd%d", rdev->raid_disk); - sysfs_remove_link(&mddev->kobj, nm); + if (!test_bit(Replacement, &rdev->flags)) { + sprintf(nm, "rd%d", rdev->raid_disk); + sysfs_remove_link(&mddev->kobj, nm); + } } /* diff --git a/include/linux/raid/md_p.h b/include/linux/raid/md_p.h index 9e65d9e20662..6f6df86f1ae5 100644 --- a/include/linux/raid/md_p.h +++ b/include/linux/raid/md_p.h @@ -277,7 +277,10 @@ struct mdp_superblock_1 { */ #define MD_FEATURE_RESHAPE_ACTIVE 4 #define MD_FEATURE_BAD_BLOCKS 8 /* badblock list is not empty */ - -#define MD_FEATURE_ALL (1|2|4|8) +#define MD_FEATURE_REPLACEMENT 16 /* This device is replacing an + * active device with same 'role'. + * 'recovery_offset' is also set. + */ +#define MD_FEATURE_ALL (1|2|4|8|16) #endif -- cgit v1.2.3 From 671488cc25f7c194c7c7a9f258bab1df17a6ff69 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:52 +1100 Subject: md/raid5: allow each slot to have an extra replacement device Just enhance data structures to record a second device per slot to be used as a 'replacement' device, replacing the original. We also have a second bio in each slot in each stripe_head. This will only be used when writing to the array - we need to write to both the original and the replacement at the same time, so will need two bios. For now, only try using the replacement drive for aligned-reads. In this case, we prefer the replacement if it has been recovered far enough, otherwise use the original. This includes a small enhancement. Previously we would only do aligned reads if the target device was fully recovered. Now we also do them if it has recovered far enough. Reviewed-by: Dan Williams Signed-off-by: NeilBrown --- drivers/md/raid5.c | 15 ++++++++++++-- drivers/md/raid5.h | 57 +++++++++++++++++++++++++++++++----------------------- 2 files changed, 46 insertions(+), 26 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 6b9fc58e8f2d..94bc35ba4c81 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -3594,6 +3594,7 @@ static int chunk_aligned_read(struct mddev *mddev, struct bio * raid_bio) int dd_idx; struct bio* align_bi; struct md_rdev *rdev; + sector_t end_sector; if (!in_chunk_boundary(mddev, raid_bio)) { pr_debug("chunk_aligned_read : non aligned\n"); @@ -3618,9 +3619,19 @@ static int chunk_aligned_read(struct mddev *mddev, struct bio * raid_bio) 0, &dd_idx, NULL); + end_sector = align_bi->bi_sector + (align_bi->bi_size >> 9); rcu_read_lock(); - rdev = rcu_dereference(conf->disks[dd_idx].rdev); - if (rdev && test_bit(In_sync, &rdev->flags)) { + rdev = rcu_dereference(conf->disks[dd_idx].replacement); + if (!rdev || test_bit(Faulty, &rdev->flags) || + rdev->recovery_offset < end_sector) { + rdev = rcu_dereference(conf->disks[dd_idx].rdev); + if (rdev && + (test_bit(Faulty, &rdev->flags) || + !(test_bit(In_sync, &rdev->flags) || + rdev->recovery_offset >= end_sector))) + rdev = NULL; + } + if (rdev) { sector_t first_bad; int bad_sectors; diff --git a/drivers/md/raid5.h b/drivers/md/raid5.h index e10c5531f9c5..43106f01862d 100644 --- a/drivers/md/raid5.h +++ b/drivers/md/raid5.h @@ -226,8 +226,11 @@ struct stripe_head { #endif } ops; struct r5dev { - struct bio req; - struct bio_vec vec; + /* rreq and rvec are used for the replacement device when + * writing data to both devices. + */ + struct bio req, rreq; + struct bio_vec vec, rvec; struct page *page; struct bio *toread, *read, *towrite, *written; sector_t sector; /* sector of this page */ @@ -252,29 +255,35 @@ struct stripe_head_state { int handle_bad_blocks; }; -/* Flags */ -#define R5_UPTODATE 0 /* page contains current data */ -#define R5_LOCKED 1 /* IO has been submitted on "req" */ -#define R5_OVERWRITE 2 /* towrite covers whole page */ +/* Flags for struct r5dev.flags */ +enum r5dev_flags { + R5_UPTODATE, /* page contains current data */ + R5_LOCKED, /* IO has been submitted on "req" */ + R5_OVERWRITE, /* towrite covers whole page */ /* and some that are internal to handle_stripe */ -#define R5_Insync 3 /* rdev && rdev->in_sync at start */ -#define R5_Wantread 4 /* want to schedule a read */ -#define R5_Wantwrite 5 -#define R5_Overlap 7 /* There is a pending overlapping request on this block */ -#define R5_ReadError 8 /* seen a read error here recently */ -#define R5_ReWrite 9 /* have tried to over-write the readerror */ + R5_Insync, /* rdev && rdev->in_sync at start */ + R5_Wantread, /* want to schedule a read */ + R5_Wantwrite, + R5_Overlap, /* There is a pending overlapping request + * on this block */ + R5_ReadError, /* seen a read error here recently */ + R5_ReWrite, /* have tried to over-write the readerror */ -#define R5_Expanded 10 /* This block now has post-expand data */ -#define R5_Wantcompute 11 /* compute_block in progress treat as - * uptodate - */ -#define R5_Wantfill 12 /* dev->toread contains a bio that needs - * filling - */ -#define R5_Wantdrain 13 /* dev->towrite needs to be drained */ -#define R5_WantFUA 14 /* Write should be FUA */ -#define R5_WriteError 15 /* got a write error - need to record it */ -#define R5_MadeGood 16 /* A bad block has been fixed by writing to it*/ + R5_Expanded, /* This block now has post-expand data */ + R5_Wantcompute, /* compute_block in progress treat as + * uptodate + */ + R5_Wantfill, /* dev->toread contains a bio that needs + * filling + */ + R5_Wantdrain, /* dev->towrite needs to be drained */ + R5_WantFUA, /* Write should be FUA */ + R5_WriteError, /* got a write error - need to record it */ + R5_MadeGood, /* A bad block has been fixed by writing to it */ + R5_ReadRepl, /* Will/did read from replacement rather than orig */ + R5_MadeGoodRepl,/* A bad block on the replacement device has been + * fixed by writing to it */ +}; /* * Write method */ @@ -344,7 +353,7 @@ enum { struct disk_info { - struct md_rdev *rdev; + struct md_rdev *rdev, *replacement; }; struct r5conf { -- cgit v1.2.3 From ede7ee8b4d007f308aa033be676b1a048f99e9db Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:52 +1100 Subject: md/raid5: raid5.h cleanup Remove some #defines that are no longer used, and replace some others with an enum. And remove an unused field. Reviewed-by: Dan Williams Signed-off-by: NeilBrown --- drivers/md/raid5.h | 27 +++++++++------------------ 1 file changed, 9 insertions(+), 18 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid5.h b/drivers/md/raid5.h index 43106f01862d..4cfd8016010e 100644 --- a/drivers/md/raid5.h +++ b/drivers/md/raid5.h @@ -27,7 +27,7 @@ * The possible state transitions are: * * Empty -> Want - on read or write to get old data for parity calc - * Empty -> Dirty - on compute_parity to satisfy write/sync request.(RECONSTRUCT_WRITE) + * Empty -> Dirty - on compute_parity to satisfy write/sync request. * Empty -> Clean - on compute_block when computing a block for failed drive * Want -> Empty - on failed read * Want -> Clean - on successful completion of read request @@ -284,15 +284,6 @@ enum r5dev_flags { R5_MadeGoodRepl,/* A bad block on the replacement device has been * fixed by writing to it */ }; -/* - * Write method - */ -#define RECONSTRUCT_WRITE 1 -#define READ_MODIFY_WRITE 2 -/* not a write method, but a compute_parity mode */ -#define CHECK_PARITY 3 -/* Additional compute_parity mode -- updates the parity w/o LOCKING */ -#define UPDATE_PARITY 4 /* * Stripe state @@ -320,13 +311,14 @@ enum { /* * Operation request flags */ -#define STRIPE_OP_BIOFILL 0 -#define STRIPE_OP_COMPUTE_BLK 1 -#define STRIPE_OP_PREXOR 2 -#define STRIPE_OP_BIODRAIN 3 -#define STRIPE_OP_RECONSTRUCT 4 -#define STRIPE_OP_CHECK 5 - +enum { + STRIPE_OP_BIOFILL, + STRIPE_OP_COMPUTE_BLK, + STRIPE_OP_PREXOR, + STRIPE_OP_BIODRAIN, + STRIPE_OP_RECONSTRUCT, + STRIPE_OP_CHECK, +}; /* * Plugging: * @@ -359,7 +351,6 @@ struct disk_info { struct r5conf { struct hlist_head *stripe_hashtbl; struct mddev *mddev; - struct disk_info *spare; int chunk_sectors; int level, algorithm; int max_degraded; -- cgit v1.2.3 From 995c4275a7e14b8752f301e4570831a108ae4303 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:52 +1100 Subject: md/raid5: remove redundant bio initialisations. We current initialise some fields of a bio when preparing a stripe_head, and again just before submitting the request. Remove the duplication by only setting the fields that lower level devices don't touch in raid5_build_block, and only set the changeable fields in ops_run_io. Reviewed-by: Dan Williams Signed-off-by: NeilBrown --- drivers/md/raid5.c | 9 +-------- 1 file changed, 1 insertion(+), 8 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 94bc35ba4c81..2d2aaf6e98e2 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -574,10 +574,7 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s) atomic_inc(&sh->count); bi->bi_sector = sh->sector + rdev->data_offset; bi->bi_flags = 1 << BIO_UPTODATE; - bi->bi_vcnt = 1; - bi->bi_max_vecs = 1; bi->bi_idx = 0; - bi->bi_io_vec = &sh->dev[i].vec; bi->bi_io_vec[0].bv_len = STRIPE_SIZE; bi->bi_io_vec[0].bv_offset = 0; bi->bi_size = STRIPE_SIZE; @@ -1720,12 +1717,8 @@ static void raid5_build_block(struct stripe_head *sh, int i, int previous) dev->req.bi_io_vec = &dev->vec; dev->req.bi_vcnt++; dev->req.bi_max_vecs++; - dev->vec.bv_page = dev->page; - dev->vec.bv_len = STRIPE_SIZE; - dev->vec.bv_offset = 0; - - dev->req.bi_sector = sh->sector; dev->req.bi_private = sh; + dev->vec.bv_page = dev->page; dev->flags = 0; dev->sector = compute_blocknr(sh, i, previous); -- cgit v1.2.3 From 14a75d3e07c784c004b4b44b34af996b8e4ac453 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:52 +1100 Subject: md/raid5: preferentially read from replacement device if possible. If a replacement device is present and has been recovered far enough, then use it for reading into the stripe cache. If we get an error we don't try to repair it, we just fail the device. A replacement device that gives errors does not sound sensible. This requires removing the setting of R5_ReadError when we get a read error during a read that bypasses the cache. It was probably a bad idea anyway as we don't know that every block in the read caused an error, and it could cause ReadError to be set for the replacement device, which is bad. Signed-off-by: NeilBrown --- drivers/md/raid5.c | 69 ++++++++++++++++++++++++++++++++++++++++++------------ 1 file changed, 54 insertions(+), 15 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 2d2aaf6e98e2..2ae63c5b1c25 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -524,7 +524,11 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s) bi->bi_end_io = raid5_end_read_request; rcu_read_lock(); - rdev = rcu_dereference(conf->disks[i].rdev); + if (rw == READ && + test_bit(R5_ReadRepl, &sh->dev[i].flags)) + rdev = rcu_dereference(conf->disks[i].replacement); + else + rdev = rcu_dereference(conf->disks[i].rdev); if (rdev && test_bit(Faulty, &rdev->flags)) rdev = NULL; if (rdev) @@ -1605,11 +1609,18 @@ static void raid5_end_read_request(struct bio * bi, int error) BUG(); return; } + if (test_bit(R5_ReadRepl, &sh->dev[i].flags)) + rdev = conf->disks[i].replacement; + else + rdev = conf->disks[i].rdev; if (uptodate) { set_bit(R5_UPTODATE, &sh->dev[i].flags); if (test_bit(R5_ReadError, &sh->dev[i].flags)) { - rdev = conf->disks[i].rdev; + /* Note that this cannot happen on a + * replacement device. We just fail those on + * any error + */ printk_ratelimited( KERN_INFO "md/raid:%s: read error corrected" @@ -1622,16 +1633,24 @@ static void raid5_end_read_request(struct bio * bi, int error) clear_bit(R5_ReadError, &sh->dev[i].flags); clear_bit(R5_ReWrite, &sh->dev[i].flags); } - if (atomic_read(&conf->disks[i].rdev->read_errors)) - atomic_set(&conf->disks[i].rdev->read_errors, 0); + if (atomic_read(&rdev->read_errors)) + atomic_set(&rdev->read_errors, 0); } else { - const char *bdn = bdevname(conf->disks[i].rdev->bdev, b); + const char *bdn = bdevname(rdev->bdev, b); int retry = 0; - rdev = conf->disks[i].rdev; clear_bit(R5_UPTODATE, &sh->dev[i].flags); atomic_inc(&rdev->read_errors); - if (conf->mddev->degraded >= conf->max_degraded) + if (test_bit(R5_ReadRepl, &sh->dev[i].flags)) + printk_ratelimited( + KERN_WARNING + "md/raid:%s: read error on replacement device " + "(sector %llu on %s).\n", + mdname(conf->mddev), + (unsigned long long)(sh->sector + + rdev->data_offset), + bdn); + else if (conf->mddev->degraded >= conf->max_degraded) printk_ratelimited( KERN_WARNING "md/raid:%s: read error not correctable " @@ -1665,7 +1684,7 @@ static void raid5_end_read_request(struct bio * bi, int error) md_error(conf->mddev, rdev); } } - rdev_dec_pending(conf->disks[i].rdev, conf->mddev); + rdev_dec_pending(rdev, conf->mddev); clear_bit(R5_LOCKED, &sh->dev[i].flags); set_bit(STRIPE_HANDLE, &sh->state); release_stripe(sh); @@ -3036,7 +3055,19 @@ static void analyse_stripe(struct stripe_head *sh, struct stripe_head_state *s) } if (dev->written) s->written++; - rdev = rcu_dereference(conf->disks[i].rdev); + /* Prefer to use the replacement for reads, but only + * if it is recovered enough and has no bad blocks. + */ + rdev = rcu_dereference(conf->disks[i].replacement); + if (rdev && !test_bit(Faulty, &rdev->flags) && + rdev->recovery_offset >= sh->sector + STRIPE_SECTORS && + !is_badblock(rdev, sh->sector, STRIPE_SECTORS, + &first_bad, &bad_sectors)) + set_bit(R5_ReadRepl, &dev->flags); + else { + rdev = rcu_dereference(conf->disks[i].rdev); + clear_bit(R5_ReadRepl, &dev->flags); + } if (rdev && test_bit(Faulty, &rdev->flags)) rdev = NULL; if (rdev) { @@ -3078,17 +3109,26 @@ static void analyse_stripe(struct stripe_head *sh, struct stripe_head_state *s) set_bit(R5_Insync, &dev->flags); if (rdev && test_bit(R5_WriteError, &dev->flags)) { - clear_bit(R5_Insync, &dev->flags); - if (!test_bit(Faulty, &rdev->flags)) { + /* This flag does not apply to '.replacement' + * only to .rdev, so make sure to check that*/ + struct md_rdev *rdev2 = rcu_dereference( + conf->disks[i].rdev); + if (rdev2 == rdev) + clear_bit(R5_Insync, &dev->flags); + if (rdev2 && !test_bit(Faulty, &rdev2->flags)) { s->handle_bad_blocks = 1; - atomic_inc(&rdev->nr_pending); + atomic_inc(&rdev2->nr_pending); } else clear_bit(R5_WriteError, &dev->flags); } if (rdev && test_bit(R5_MadeGood, &dev->flags)) { - if (!test_bit(Faulty, &rdev->flags)) { + /* This flag does not apply to '.replacement' + * only to .rdev, so make sure to check that*/ + struct md_rdev *rdev2 = rcu_dereference( + conf->disks[i].rdev); + if (rdev2 && !test_bit(Faulty, &rdev2->flags)) { s->handle_bad_blocks = 1; - atomic_inc(&rdev->nr_pending); + atomic_inc(&rdev2->nr_pending); } else clear_bit(R5_MadeGood, &dev->flags); } @@ -4220,7 +4260,6 @@ static int retry_aligned_read(struct r5conf *conf, struct bio *raid_bio) return handled; } - set_bit(R5_ReadError, &sh->dev[dd_idx].flags); if (!add_stripe_bio(sh, raid_bio, dd_idx, 0)) { release_stripe(sh); raid5_set_bi_hw_segments(raid_bio, scnt); -- cgit v1.2.3 From 657e3e4d88461a5ab660dd87f8f773f55e748da4 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:52 +1100 Subject: md/raid5: allow removal for failed replacement devices. Enhance raid5_remove_disk to be able to remove ->replacement as well as ->rdev. Reviewed-by: Dan Williams Signed-off-by: NeilBrown --- drivers/md/raid5.c | 58 ++++++++++++++++++++++++++++++------------------------ 1 file changed, 32 insertions(+), 26 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 2ae63c5b1c25..e5795d39d418 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -5086,36 +5086,42 @@ static int raid5_remove_disk(struct mddev *mddev, struct md_rdev *rdev) struct r5conf *conf = mddev->private; int err = 0; int number = rdev->raid_disk; + struct md_rdev **rdevp; struct disk_info *p = conf->disks + number; print_raid5_conf(conf); - if (rdev == p->rdev) { - if (number >= conf->raid_disks && - conf->reshape_progress == MaxSector) - clear_bit(In_sync, &rdev->flags); + if (rdev == p->rdev) + rdevp = &p->rdev; + else if (rdev == p->replacement) + rdevp = &p->replacement; + else + return 0; - if (test_bit(In_sync, &rdev->flags) || - atomic_read(&rdev->nr_pending)) { - err = -EBUSY; - goto abort; - } - /* Only remove non-faulty devices if recovery - * isn't possible. - */ - if (!test_bit(Faulty, &rdev->flags) && - mddev->recovery_disabled != conf->recovery_disabled && - !has_failed(conf) && - number < conf->raid_disks) { - err = -EBUSY; - goto abort; - } - p->rdev = NULL; - synchronize_rcu(); - if (atomic_read(&rdev->nr_pending)) { - /* lost the race, try later */ - err = -EBUSY; - p->rdev = rdev; - } + if (number >= conf->raid_disks && + conf->reshape_progress == MaxSector) + clear_bit(In_sync, &rdev->flags); + + if (test_bit(In_sync, &rdev->flags) || + atomic_read(&rdev->nr_pending)) { + err = -EBUSY; + goto abort; + } + /* Only remove non-faulty devices if recovery + * isn't possible. + */ + if (!test_bit(Faulty, &rdev->flags) && + mddev->recovery_disabled != conf->recovery_disabled && + !has_failed(conf) && + number < conf->raid_disks) { + err = -EBUSY; + goto abort; + } + *rdevp = NULL; + synchronize_rcu(); + if (atomic_read(&rdev->nr_pending)) { + /* lost the race, try later */ + err = -EBUSY; + *rdevp = rdev; } abort: -- cgit v1.2.3 From 977df36255ab0ea78b048cbc9055300c586dcc91 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:53 +1100 Subject: md/raid5: writes should get directed to replacement as well as original. When writing, we need to submit two writes, one to the original, and one to the replacement - if there is a replacement. If the write to the replacement results in a write error, we just fail the device. We only try to record write errors to the original. When writing for recovery, we shouldn't write to the original. This will be addressed in a subsequent patch that generally addresses recovery. Reviewed-by: Dan Williams Signed-off-by: NeilBrown --- drivers/md/raid5.c | 119 ++++++++++++++++++++++++++++++++++++++++++----------- drivers/md/raid5.h | 1 + 2 files changed, 97 insertions(+), 23 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index e5795d39d418..14878a9ae09d 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -503,8 +503,8 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s) for (i = disks; i--; ) { int rw; - struct bio *bi; - struct md_rdev *rdev; + struct bio *bi, *rbi; + struct md_rdev *rdev, *rrdev = NULL; if (test_and_clear_bit(R5_Wantwrite, &sh->dev[i].flags)) { if (test_and_clear_bit(R5_WantFUA, &sh->dev[i].flags)) rw = WRITE_FUA; @@ -516,27 +516,36 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s) continue; bi = &sh->dev[i].req; + rbi = &sh->dev[i].rreq; /* For writing to replacement */ bi->bi_rw = rw; - if (rw & WRITE) + rbi->bi_rw = rw; + if (rw & WRITE) { bi->bi_end_io = raid5_end_write_request; - else + rbi->bi_end_io = raid5_end_write_request; + } else bi->bi_end_io = raid5_end_read_request; rcu_read_lock(); - if (rw == READ && - test_bit(R5_ReadRepl, &sh->dev[i].flags)) + rdev = rcu_dereference(conf->disks[i].rdev); + if (rw & WRITE) + rrdev = rcu_dereference(conf->disks[i].replacement); + else if (test_bit(R5_ReadRepl, &sh->dev[i].flags)) rdev = rcu_dereference(conf->disks[i].replacement); - else - rdev = rcu_dereference(conf->disks[i].rdev); + if (rdev && test_bit(Faulty, &rdev->flags)) rdev = NULL; if (rdev) atomic_inc(&rdev->nr_pending); + if (rrdev && test_bit(Faulty, &rrdev->flags)) + rrdev = NULL; + if (rrdev) + atomic_inc(&rrdev->nr_pending); rcu_read_unlock(); /* We have already checked bad blocks for reads. Now - * need to check for writes. + * need to check for writes. We never accept write errors + * on the replacement, so we don't to check rrdev. */ while ((rw & WRITE) && rdev && test_bit(WriteErrorSeen, &rdev->flags)) { @@ -583,8 +592,32 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s) bi->bi_io_vec[0].bv_offset = 0; bi->bi_size = STRIPE_SIZE; bi->bi_next = NULL; + if (rrdev) + set_bit(R5_DOUBLE_LOCKED, &sh->dev[i].flags); generic_make_request(bi); - } else { + } + if (rrdev) { + if (s->syncing || s->expanding || s->expanded) + md_sync_acct(rrdev->bdev, STRIPE_SECTORS); + + set_bit(STRIPE_IO_STARTED, &sh->state); + + rbi->bi_bdev = rrdev->bdev; + pr_debug("%s: for %llu schedule op %ld on " + "replacement disc %d\n", + __func__, (unsigned long long)sh->sector, + rbi->bi_rw, i); + atomic_inc(&sh->count); + rbi->bi_sector = sh->sector + rrdev->data_offset; + rbi->bi_flags = 1 << BIO_UPTODATE; + rbi->bi_idx = 0; + rbi->bi_io_vec[0].bv_len = STRIPE_SIZE; + rbi->bi_io_vec[0].bv_offset = 0; + rbi->bi_size = STRIPE_SIZE; + rbi->bi_next = NULL; + generic_make_request(rbi); + } + if (!rdev && !rrdev) { if (rw & WRITE) set_bit(STRIPE_DEGRADED, &sh->state); pr_debug("skip op %ld on disc %d for sector %llu\n", @@ -1695,14 +1728,23 @@ static void raid5_end_write_request(struct bio *bi, int error) struct stripe_head *sh = bi->bi_private; struct r5conf *conf = sh->raid_conf; int disks = sh->disks, i; + struct md_rdev *uninitialized_var(rdev); int uptodate = test_bit(BIO_UPTODATE, &bi->bi_flags); sector_t first_bad; int bad_sectors; + int replacement = 0; - for (i=0 ; idev[i].req) + for (i = 0 ; i < disks; i++) { + if (bi == &sh->dev[i].req) { + rdev = conf->disks[i].rdev; break; - + } + if (bi == &sh->dev[i].rreq) { + rdev = conf->disks[i].replacement; + replacement = 1; + break; + } + } pr_debug("end_write_request %llu/%d, count %d, uptodate: %d.\n", (unsigned long long)sh->sector, i, atomic_read(&sh->count), uptodate); @@ -1711,21 +1753,30 @@ static void raid5_end_write_request(struct bio *bi, int error) return; } - if (!uptodate) { - set_bit(WriteErrorSeen, &conf->disks[i].rdev->flags); - set_bit(R5_WriteError, &sh->dev[i].flags); - } else if (is_badblock(conf->disks[i].rdev, sh->sector, STRIPE_SECTORS, - &first_bad, &bad_sectors)) - set_bit(R5_MadeGood, &sh->dev[i].flags); + if (replacement) { + if (!uptodate) + md_error(conf->mddev, rdev); + else if (is_badblock(rdev, sh->sector, + STRIPE_SECTORS, + &first_bad, &bad_sectors)) + set_bit(R5_MadeGoodRepl, &sh->dev[i].flags); + } else { + if (!uptodate) { + set_bit(WriteErrorSeen, &rdev->flags); + set_bit(R5_WriteError, &sh->dev[i].flags); + } else if (is_badblock(rdev, sh->sector, + STRIPE_SECTORS, + &first_bad, &bad_sectors)) + set_bit(R5_MadeGood, &sh->dev[i].flags); + } + rdev_dec_pending(rdev, conf->mddev); - rdev_dec_pending(conf->disks[i].rdev, conf->mddev); - - clear_bit(R5_LOCKED, &sh->dev[i].flags); + if (!test_and_clear_bit(R5_DOUBLE_LOCKED, &sh->dev[i].flags)) + clear_bit(R5_LOCKED, &sh->dev[i].flags); set_bit(STRIPE_HANDLE, &sh->state); release_stripe(sh); } - static sector_t compute_blocknr(struct stripe_head *sh, int i, int previous); static void raid5_build_block(struct stripe_head *sh, int i, int previous) @@ -1739,6 +1790,13 @@ static void raid5_build_block(struct stripe_head *sh, int i, int previous) dev->req.bi_private = sh; dev->vec.bv_page = dev->page; + bio_init(&dev->rreq); + dev->rreq.bi_io_vec = &dev->rvec; + dev->rreq.bi_vcnt++; + dev->rreq.bi_max_vecs++; + dev->rreq.bi_private = sh; + dev->rvec.bv_page = dev->page; + dev->flags = 0; dev->sector = compute_blocknr(sh, i, previous); } @@ -3132,6 +3190,15 @@ static void analyse_stripe(struct stripe_head *sh, struct stripe_head_state *s) } else clear_bit(R5_MadeGood, &dev->flags); } + if (test_bit(R5_MadeGoodRepl, &dev->flags)) { + struct md_rdev *rdev2 = rcu_dereference( + conf->disks[i].replacement); + if (rdev2 && !test_bit(Faulty, &rdev2->flags)) { + s->handle_bad_blocks = 1; + atomic_inc(&rdev2->nr_pending); + } else + clear_bit(R5_MadeGoodRepl, &dev->flags); + } if (!test_bit(R5_Insync, &dev->flags)) { /* The ReadError flag will just be confusing now */ clear_bit(R5_ReadError, &dev->flags); @@ -3404,6 +3471,12 @@ finish: STRIPE_SECTORS); rdev_dec_pending(rdev, conf->mddev); } + if (test_and_clear_bit(R5_MadeGoodRepl, &dev->flags)) { + rdev = conf->disks[i].replacement; + rdev_clear_badblocks(rdev, sh->sector, + STRIPE_SECTORS); + rdev_dec_pending(rdev, conf->mddev); + } } if (s.ops_request) diff --git a/drivers/md/raid5.h b/drivers/md/raid5.h index 4cfd8016010e..f6faaa16a565 100644 --- a/drivers/md/raid5.h +++ b/drivers/md/raid5.h @@ -259,6 +259,7 @@ struct stripe_head_state { enum r5dev_flags { R5_UPTODATE, /* page contains current data */ R5_LOCKED, /* IO has been submitted on "req" */ + R5_DOUBLE_LOCKED,/* Cannot clear R5_LOCKED until 2 writes complete */ R5_OVERWRITE, /* towrite covers whole page */ /* and some that are internal to handle_stripe */ R5_Insync, /* rdev && rdev->in_sync at start */ -- cgit v1.2.3 From 9a3e1101b827a59ac9036a672f5fa8d5279d0fe2 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:53 +1100 Subject: md/raid5: detect and handle replacements during recovery. During recovery we want to write to the replacement but not the original. So we have two new flags - R5_NeedReplace if this stripe has a replacement that needs to be written at some stage - R5_WantReplace if NeedReplace, and the data is available, and a 'sync' has been requested on this stripe. We also distinguish between 'sync and replace' which need to read all other devices, and 'replace' which only needs to read the devices being replaced. Note that during resync we always write to any replacement device. It might not need to be written to, but as we don't read to compare, we have to write to be sure. Signed-off-by: NeilBrown --- drivers/md/raid5.c | 123 ++++++++++++++++++++++++++++++++++++++++------------- drivers/md/raid5.h | 13 +++++- 2 files changed, 106 insertions(+), 30 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 14878a9ae09d..516baf49a1fa 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -503,6 +503,7 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s) for (i = disks; i--; ) { int rw; + int replace_only = 0; struct bio *bi, *rbi; struct md_rdev *rdev, *rrdev = NULL; if (test_and_clear_bit(R5_Wantwrite, &sh->dev[i].flags)) { @@ -512,7 +513,11 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s) rw = WRITE; } else if (test_and_clear_bit(R5_Wantread, &sh->dev[i].flags)) rw = READ; - else + else if (test_and_clear_bit(R5_WantReplace, + &sh->dev[i].flags)) { + rw = WRITE; + replace_only = 1; + } else continue; bi = &sh->dev[i].req; @@ -528,10 +533,15 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s) rcu_read_lock(); rdev = rcu_dereference(conf->disks[i].rdev); - if (rw & WRITE) - rrdev = rcu_dereference(conf->disks[i].replacement); - else if (test_bit(R5_ReadRepl, &sh->dev[i].flags)) - rdev = rcu_dereference(conf->disks[i].replacement); + rrdev = rcu_dereference(conf->disks[i].replacement); + if (rw & WRITE) { + if (replace_only) + rdev = NULL; + } else { + if (test_bit(R5_ReadRepl, &sh->dev[i].flags)) + rdev = rrdev; + rrdev = NULL; + } if (rdev && test_bit(Faulty, &rdev->flags)) rdev = NULL; @@ -575,7 +585,8 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s) } if (rdev) { - if (s->syncing || s->expanding || s->expanded) + if (s->syncing || s->expanding || s->expanded + || s->replacing) md_sync_acct(rdev->bdev, STRIPE_SECTORS); set_bit(STRIPE_IO_STARTED, &sh->state); @@ -597,7 +608,8 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s) generic_make_request(bi); } if (rrdev) { - if (s->syncing || s->expanding || s->expanded) + if (s->syncing || s->expanding || s->expanded + || s->replacing) md_sync_acct(rrdev->bdev, STRIPE_SECTORS); set_bit(STRIPE_IO_STARTED, &sh->state); @@ -2440,8 +2452,9 @@ handle_failed_sync(struct r5conf *conf, struct stripe_head *sh, md_done_sync(conf->mddev, STRIPE_SECTORS, 0); clear_bit(STRIPE_SYNCING, &sh->state); s->syncing = 0; + s->replacing = 0; /* There is nothing more to do for sync/check/repair. - * For recover we need to record a bad block on all + * For recover/replace we need to record a bad block on all * non-sync devices, or abort the recovery */ if (!test_bit(MD_RECOVERY_RECOVER, &conf->mddev->recovery)) @@ -2451,12 +2464,18 @@ handle_failed_sync(struct r5conf *conf, struct stripe_head *sh, */ for (i = 0; i < conf->raid_disks; i++) { struct md_rdev *rdev = conf->disks[i].rdev; - if (!rdev - || test_bit(Faulty, &rdev->flags) - || test_bit(In_sync, &rdev->flags)) - continue; - if (!rdev_set_badblocks(rdev, sh->sector, - STRIPE_SECTORS, 0)) + if (rdev + && !test_bit(Faulty, &rdev->flags) + && !test_bit(In_sync, &rdev->flags) + && !rdev_set_badblocks(rdev, sh->sector, + STRIPE_SECTORS, 0)) + abort = 1; + rdev = conf->disks[i].replacement; + if (rdev + && !test_bit(Faulty, &rdev->flags) + && !test_bit(In_sync, &rdev->flags) + && !rdev_set_badblocks(rdev, sh->sector, + STRIPE_SECTORS, 0)) abort = 1; } if (abort) { @@ -2465,6 +2484,22 @@ handle_failed_sync(struct r5conf *conf, struct stripe_head *sh, } } +static int want_replace(struct stripe_head *sh, int disk_idx) +{ + struct md_rdev *rdev; + int rv = 0; + /* Doing recovery so rcu locking not required */ + rdev = sh->raid_conf->disks[disk_idx].replacement; + if (rdev + && !test_bit(Faulty, &rdev->flags) + && !test_bit(In_sync, &rdev->flags) + && (rdev->recovery_offset <= sh->sector + || rdev->mddev->recovery_cp <= sh->sector)) + rv = 1; + + return rv; +} + /* fetch_block - checks the given member device to see if its data needs * to be read or computed to satisfy a request. * @@ -2484,6 +2519,7 @@ static int fetch_block(struct stripe_head *sh, struct stripe_head_state *s, (dev->toread || (dev->towrite && !test_bit(R5_OVERWRITE, &dev->flags)) || s->syncing || s->expanding || + (s->replacing && want_replace(sh, disk_idx)) || (s->failed >= 1 && fdev[0]->toread) || (s->failed >= 2 && fdev[1]->toread) || (sh->raid_conf->level <= 5 && s->failed && fdev[0]->towrite && @@ -3037,22 +3073,18 @@ static void handle_stripe_expansion(struct r5conf *conf, struct stripe_head *sh) } } - /* * handle_stripe - do things to a stripe. * - * We lock the stripe and then examine the state of various bits - * to see what needs to be done. + * We lock the stripe by setting STRIPE_ACTIVE and then examine the + * state of various bits to see what needs to be done. * Possible results: - * return some read request which now have data - * return some write requests which are safely on disc + * return some read requests which now have data + * return some write requests which are safely on storage * schedule a read on some buffers * schedule a write of some buffers * return confirmation of parity correctness * - * buffers are taken off read_list or write_list, and bh_cache buffers - * get BH_Lock set before the stripe lock is released. - * */ static void analyse_stripe(struct stripe_head *sh, struct stripe_head_state *s) @@ -3061,10 +3093,10 @@ static void analyse_stripe(struct stripe_head *sh, struct stripe_head_state *s) int disks = sh->disks; struct r5dev *dev; int i; + int do_recovery = 0; memset(s, 0, sizeof(*s)); - s->syncing = test_bit(STRIPE_SYNCING, &sh->state); s->expanding = test_bit(STRIPE_EXPAND_SOURCE, &sh->state); s->expanded = test_bit(STRIPE_EXPAND_READY, &sh->state); s->failed_num[0] = -1; @@ -3082,7 +3114,8 @@ static void analyse_stripe(struct stripe_head *sh, struct stripe_head_state *s) dev = &sh->dev[i]; pr_debug("check %d: state 0x%lx read %p write %p written %p\n", - i, dev->flags, dev->toread, dev->towrite, dev->written); + i, dev->flags, + dev->toread, dev->towrite, dev->written); /* maybe we can reply to a read * * new wantfill requests are only permitted while @@ -3123,6 +3156,8 @@ static void analyse_stripe(struct stripe_head *sh, struct stripe_head_state *s) &first_bad, &bad_sectors)) set_bit(R5_ReadRepl, &dev->flags); else { + if (rdev) + set_bit(R5_NeedReplace, &dev->flags); rdev = rcu_dereference(conf->disks[i].rdev); clear_bit(R5_ReadRepl, &dev->flags); } @@ -3210,9 +3245,25 @@ static void analyse_stripe(struct stripe_head *sh, struct stripe_head_state *s) if (s->failed < 2) s->failed_num[s->failed] = i; s->failed++; + if (rdev && !test_bit(Faulty, &rdev->flags)) + do_recovery = 1; } } spin_unlock_irq(&conf->device_lock); + if (test_bit(STRIPE_SYNCING, &sh->state)) { + /* If there is a failed device being replaced, + * we must be recovering. + * else if we are after recovery_cp, we must be syncing + * else we can only be replacing + * sync and recovery both need to read all devices, and so + * use the same flag. + */ + if (do_recovery || + sh->sector >= conf->mddev->recovery_cp) + s->syncing = 1; + else + s->replacing = 1; + } rcu_read_unlock(); } @@ -3254,7 +3305,7 @@ static void handle_stripe(struct stripe_head *sh) if (unlikely(s.blocked_rdev)) { if (s.syncing || s.expanding || s.expanded || - s.to_write || s.written) { + s.replacing || s.to_write || s.written) { set_bit(STRIPE_HANDLE, &sh->state); goto finish; } @@ -3280,7 +3331,7 @@ static void handle_stripe(struct stripe_head *sh) sh->reconstruct_state = 0; if (s.to_read+s.to_write+s.written) handle_failed_stripe(conf, sh, &s, disks, &s.return_bi); - if (s.syncing) + if (s.syncing + s.replacing) handle_failed_sync(conf, sh, &s); } @@ -3311,7 +3362,9 @@ static void handle_stripe(struct stripe_head *sh) */ if (s.to_read || s.non_overwrite || (conf->level == 6 && s.to_write && s.failed) - || (s.syncing && (s.uptodate + s.compute < disks)) || s.expanding) + || (s.syncing && (s.uptodate + s.compute < disks)) + || s.replacing + || s.expanding) handle_stripe_fill(sh, &s, disks); /* Now we check to see if any write operations have recently @@ -3373,7 +3426,20 @@ static void handle_stripe(struct stripe_head *sh) handle_parity_checks5(conf, sh, &s, disks); } - if (s.syncing && s.locked == 0 && test_bit(STRIPE_INSYNC, &sh->state)) { + if (s.replacing && s.locked == 0 + && !test_bit(STRIPE_INSYNC, &sh->state)) { + /* Write out to replacement devices where possible */ + for (i = 0; i < conf->raid_disks; i++) + if (test_bit(R5_UPTODATE, &sh->dev[i].flags) && + test_bit(R5_NeedReplace, &sh->dev[i].flags)) { + set_bit(R5_WantReplace, &sh->dev[i].flags); + set_bit(R5_LOCKED, &sh->dev[i].flags); + s.locked++; + } + set_bit(STRIPE_INSYNC, &sh->state); + } + if ((s.syncing || s.replacing) && s.locked == 0 && + test_bit(STRIPE_INSYNC, &sh->state)) { md_done_sync(conf->mddev, STRIPE_SECTORS, 1); clear_bit(STRIPE_SYNCING, &sh->state); } @@ -4262,7 +4328,6 @@ static inline sector_t sync_request(struct mddev *mddev, sector_t sector_nr, int return sync_blocks * STRIPE_SECTORS; /* keep things rounded to whole stripes */ } - bitmap_cond_end_sync(mddev->bitmap, sector_nr); sh = get_active_stripe(conf, sector_nr, 0, 1, 0); diff --git a/drivers/md/raid5.h b/drivers/md/raid5.h index f6faaa16a565..8d8e13934a48 100644 --- a/drivers/md/raid5.h +++ b/drivers/md/raid5.h @@ -242,7 +242,13 @@ struct stripe_head { * for handle_stripe. */ struct stripe_head_state { - int syncing, expanding, expanded; + /* 'syncing' means that we need to read all devices, either + * to check/correct parity, or to reconstruct a missing device. + * 'replacing' means we are replacing one or more drives and + * the source is valid at this point so we don't need to + * read all devices, just the replacement targets. + */ + int syncing, expanding, expanded, replacing; int locked, uptodate, to_read, to_write, failed, written; int to_fill, compute, req_compute, non_overwrite; int failed_num[2]; @@ -284,6 +290,11 @@ enum r5dev_flags { R5_ReadRepl, /* Will/did read from replacement rather than orig */ R5_MadeGoodRepl,/* A bad block on the replacement device has been * fixed by writing to it */ + R5_NeedReplace, /* This device has a replacement which is not + * up-to-date at this stripe. */ + R5_WantReplace, /* We need to update the replacement, we have read + * data in, and now is a good time to write it out. + */ }; /* -- cgit v1.2.3 From dd054fce88d33da1aa81d018db75b91b102a6959 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:53 +1100 Subject: md/raid5: handle activation of replacement device when recovery completes. When recovery completes - as reported by a call to ->spare_active, we clear In_sync on the original and set it on the replacement. Then when the original gets removed we move the replacement from 'replacement' to 'rdev'. This could race with other code that is looking at these pointers, so we use memory barriers and careful ordering to ensure that a reader might see one device twice, but never no devices. Then the readers guard against using both devices, which could only happen when writing. Signed-off-by: NeilBrown --- drivers/md/raid5.c | 69 ++++++++++++++++++++++++++++++++++++++++++++++++------ 1 file changed, 62 insertions(+), 7 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 516baf49a1fa..b443cd2459df 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -532,13 +532,21 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s) bi->bi_end_io = raid5_end_read_request; rcu_read_lock(); - rdev = rcu_dereference(conf->disks[i].rdev); rrdev = rcu_dereference(conf->disks[i].replacement); + smp_mb(); /* Ensure that if rrdev is NULL, rdev won't be */ + rdev = rcu_dereference(conf->disks[i].rdev); + if (!rdev) { + rdev = rrdev; + rrdev = NULL; + } if (rw & WRITE) { if (replace_only) rdev = NULL; + if (rdev == rrdev) + /* We raced and saw duplicates */ + rrdev = NULL; } else { - if (test_bit(R5_ReadRepl, &sh->dev[i].flags)) + if (test_bit(R5_ReadRepl, &sh->dev[i].flags) && rrdev) rdev = rrdev; rrdev = NULL; } @@ -1640,7 +1648,7 @@ static void raid5_end_read_request(struct bio * bi, int error) int disks = sh->disks, i; int uptodate = test_bit(BIO_UPTODATE, &bi->bi_flags); char b[BDEVNAME_SIZE]; - struct md_rdev *rdev; + struct md_rdev *rdev = NULL; for (i=0 ; idev[i].flags)) + /* If replacement finished while this request was outstanding, + * 'replacement' might be NULL already. + * In that case it moved down to 'rdev'. + * rdev is not removed until all requests are finished. + */ rdev = conf->disks[i].replacement; - else + if (!rdev) rdev = conf->disks[i].rdev; if (uptodate) { @@ -1753,7 +1766,14 @@ static void raid5_end_write_request(struct bio *bi, int error) } if (bi == &sh->dev[i].rreq) { rdev = conf->disks[i].replacement; - replacement = 1; + if (rdev) + replacement = 1; + else + /* rdev was removed and 'replacement' + * replaced it. rdev is not removed + * until all requests are finished. + */ + rdev = conf->disks[i].rdev; break; } } @@ -3539,6 +3559,9 @@ finish: } if (test_and_clear_bit(R5_MadeGoodRepl, &dev->flags)) { rdev = conf->disks[i].replacement; + if (!rdev) + /* rdev have been moved down */ + rdev = conf->disks[i].rdev; rdev_clear_badblocks(rdev, sh->sector, STRIPE_SECTORS); rdev_dec_pending(rdev, conf->mddev); @@ -5204,7 +5227,25 @@ static int raid5_spare_active(struct mddev *mddev) for (i = 0; i < conf->raid_disks; i++) { tmp = conf->disks + i; - if (tmp->rdev + if (tmp->replacement + && tmp->replacement->recovery_offset == MaxSector + && !test_bit(Faulty, &tmp->replacement->flags) + && !test_and_set_bit(In_sync, &tmp->replacement->flags)) { + /* Replacement has just become active. */ + if (!tmp->rdev + || !test_and_clear_bit(In_sync, &tmp->rdev->flags)) + count++; + if (tmp->rdev) { + /* Replaced device not technically faulty, + * but we need to be sure it gets removed + * and never re-added. + */ + set_bit(Faulty, &tmp->rdev->flags); + sysfs_notify_dirent_safe( + tmp->rdev->sysfs_state); + } + sysfs_notify_dirent_safe(tmp->replacement->sysfs_state); + } else if (tmp->rdev && tmp->rdev->recovery_offset == MaxSector && !test_bit(Faulty, &tmp->rdev->flags) && !test_and_set_bit(In_sync, &tmp->rdev->flags)) { @@ -5250,6 +5291,7 @@ static int raid5_remove_disk(struct mddev *mddev, struct md_rdev *rdev) if (!test_bit(Faulty, &rdev->flags) && mddev->recovery_disabled != conf->recovery_disabled && !has_failed(conf) && + (!p->replacement || p->replacement == rdev) && number < conf->raid_disks) { err = -EBUSY; goto abort; @@ -5260,7 +5302,20 @@ static int raid5_remove_disk(struct mddev *mddev, struct md_rdev *rdev) /* lost the race, try later */ err = -EBUSY; *rdevp = rdev; - } + } else if (p->replacement) { + /* We must have just cleared 'rdev' */ + p->rdev = p->replacement; + clear_bit(Replacement, &p->replacement->flags); + smp_mb(); /* Make sure other CPUs may see both as identical + * but will never see neither - if they are careful + */ + p->replacement = NULL; + clear_bit(WantReplacement, &rdev->flags); + } else + /* We might have just removed the Replacement as faulty- + * clear the bit just in case + */ + clear_bit(WantReplacement, &rdev->flags); abort: print_raid5_conf(conf); -- cgit v1.2.3 From 17045f52ac76d9cd1a120e52af5d83b570af4ba8 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:53 +1100 Subject: md/raid5: recognise replacements when assembling array. If a Replacement is seen, file it as such. If we see two replacements (or two normal devices) for the one slot, abort. Reviewed-by: Dan Williams Signed-off-by: NeilBrown --- drivers/md/raid5.c | 37 +++++++++++++++++++++++++++++++------ 1 file changed, 31 insertions(+), 6 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index b443cd2459df..c80f8c2471cc 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -4847,7 +4847,15 @@ static struct r5conf *setup_conf(struct mddev *mddev) continue; disk = conf->disks + raid_disk; - disk->rdev = rdev; + if (test_bit(Replacement, &rdev->flags)) { + if (disk->replacement) + goto abort; + disk->replacement = rdev; + } else { + if (disk->rdev) + goto abort; + disk->rdev = rdev; + } if (test_bit(In_sync, &rdev->flags)) { char b[BDEVNAME_SIZE]; @@ -4936,6 +4944,7 @@ static int run(struct mddev *mddev) int dirty_parity_disks = 0; struct md_rdev *rdev; sector_t reshape_offset = 0; + int i; if (mddev->recovery_cp != MaxSector) printk(KERN_NOTICE "md/raid:%s: not clean" @@ -5025,12 +5034,25 @@ static int run(struct mddev *mddev) conf->thread = NULL; mddev->private = conf; - /* - * 0 for a fully functional array, 1 or 2 for a degraded array. - */ - list_for_each_entry(rdev, &mddev->disks, same_set) { - if (rdev->raid_disk < 0) + for (i = 0; i < conf->raid_disks && conf->previous_raid_disks; + i++) { + rdev = conf->disks[i].rdev; + if (!rdev && conf->disks[i].replacement) { + /* The replacement is all we have yet */ + rdev = conf->disks[i].replacement; + conf->disks[i].replacement = NULL; + clear_bit(Replacement, &rdev->flags); + conf->disks[i].rdev = rdev; + } + if (!rdev) continue; + if (conf->disks[i].replacement && + conf->reshape_progress != MaxSector) { + /* replacements and reshape simply do not mix. */ + printk(KERN_ERR "md: cannot handle concurrent " + "replacement and reshape.\n"); + goto abort; + } if (test_bit(In_sync, &rdev->flags)) { working_disks++; continue; @@ -5064,6 +5086,9 @@ static int run(struct mddev *mddev) dirty_parity_disks++; } + /* + * 0 for a fully functional array, 1 or 2 for a degraded array. + */ mddev->degraded = calc_degraded(conf); if (has_failed(conf)) { -- cgit v1.2.3 From 7bfec5f35c68121e7b1849f3f4166dd96c8da5b3 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:53 +1100 Subject: md/raid5: If there is a spare and a want_replacement device, start replacement. When attempting to add a spare to a RAID[456] array, also consider adding it as a replacement for a want_replacement device. This requires that common md code attempt hot_add even when the array is not formally degraded. Reviewed-by: Dan Williams Signed-off-by: NeilBrown --- drivers/md/md.c | 30 ++++++++++++++---------------- drivers/md/raid5.c | 16 ++++++++++++++-- 2 files changed, 28 insertions(+), 18 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/md.c b/drivers/md/md.c index be569eb41a93..1c1c562f63dc 100644 --- a/drivers/md/md.c +++ b/drivers/md/md.c @@ -7399,23 +7399,21 @@ static int remove_and_add_spares(struct mddev *mddev) } } - if (mddev->degraded) { - list_for_each_entry(rdev, &mddev->disks, same_set) { - if (rdev->raid_disk >= 0 && - !test_bit(In_sync, &rdev->flags) && - !test_bit(Faulty, &rdev->flags)) + list_for_each_entry(rdev, &mddev->disks, same_set) { + if (rdev->raid_disk >= 0 && + !test_bit(In_sync, &rdev->flags) && + !test_bit(Faulty, &rdev->flags)) + spares++; + if (rdev->raid_disk < 0 + && !test_bit(Faulty, &rdev->flags)) { + rdev->recovery_offset = 0; + if (mddev->pers-> + hot_add_disk(mddev, rdev) == 0) { + if (sysfs_link_rdev(mddev, rdev)) + /* failure here is OK */; spares++; - if (rdev->raid_disk < 0 - && !test_bit(Faulty, &rdev->flags)) { - rdev->recovery_offset = 0; - if (mddev->pers-> - hot_add_disk(mddev, rdev) == 0) { - if (sysfs_link_rdev(mddev, rdev)) - /* failure here is OK */; - spares++; - md_new_event(mddev); - set_bit(MD_CHANGE_DEVS, &mddev->flags); - } + md_new_event(mddev); + set_bit(MD_CHANGE_DEVS, &mddev->flags); } } } diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index c80f8c2471cc..b2c83859f75b 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -5376,8 +5376,9 @@ static int raid5_add_disk(struct mddev *mddev, struct md_rdev *rdev) disk = rdev->saved_raid_disk; else disk = first; - for ( ; disk <= last ; disk++) - if ((p=conf->disks + disk)->rdev == NULL) { + for ( ; disk <= last ; disk++) { + p = conf->disks + disk; + if (p->rdev == NULL) { clear_bit(In_sync, &rdev->flags); rdev->raid_disk = disk; err = 0; @@ -5386,6 +5387,17 @@ static int raid5_add_disk(struct mddev *mddev, struct md_rdev *rdev) rcu_assign_pointer(p->rdev, rdev); break; } + if (test_bit(WantReplacement, &p->rdev->flags) && + p->replacement == NULL) { + clear_bit(In_sync, &rdev->flags); + set_bit(Replacement, &rdev->flags); + rdev->raid_disk = disk; + err = 0; + conf->fullsync = 1; + rcu_assign_pointer(p->replacement, rdev); + break; + } + } print_raid5_conf(conf); return err; } -- cgit v1.2.3 From 3a6de2924af602f9c1b5a5154438c37f2d712dfa Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:54 +1100 Subject: md/raid5: Mark device want_replacement when we see a write error. Now that WantReplacement drives are replaced cleanly, mark a drive as WantReplacement when we see a write error. It might get failed soon so the WantReplacement flag is irrelevant, but if the write error is recorded in the bad block log, we still want to activate any spare that might be available. Reviewed-by: Dan Williams Signed-off-by: NeilBrown --- drivers/md/raid5.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'drivers/md') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index b2c83859f75b..360f2b98f62b 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -1796,6 +1796,9 @@ static void raid5_end_write_request(struct bio *bi, int error) if (!uptodate) { set_bit(WriteErrorSeen, &rdev->flags); set_bit(R5_WriteError, &sh->dev[i].flags); + if (!test_and_set_bit(WantReplacement, &rdev->flags)) + set_bit(MD_RECOVERY_NEEDED, + &rdev->mddev->recovery); } else if (is_badblock(rdev, sh->sector, STRIPE_SECTORS, &first_bad, &bad_sectors)) -- cgit v1.2.3 From 69335ef3bc5b766f34db2d688be1d35313138bca Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:54 +1100 Subject: md/raid10: prepare data structures for handling replacement. Allow each slot in the RAID10 to have 2 devices, the want_replacement and the replacement. Also an r10bio to have 2 bios, and for resync/recovery allocate the second bio if there are any replacement devices. Signed-off-by: NeilBrown --- drivers/md/raid10.c | 48 +++++++++++++++++++++++++++++++++++------ drivers/md/raid10.h | 61 ++++++++++++++++++++++++++++++++--------------------- 2 files changed, 78 insertions(+), 31 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index f5088dda4dca..9722065022fa 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c @@ -73,7 +73,8 @@ static void * r10bio_pool_alloc(gfp_t gfp_flags, void *data) struct r10conf *conf = data; int size = offsetof(struct r10bio, devs[conf->copies]); - /* allocate a r10bio with room for raid_disks entries in the bios array */ + /* allocate a r10bio with room for raid_disks entries in the + * bios array */ return kzalloc(size, gfp_flags); } @@ -123,12 +124,19 @@ static void * r10buf_pool_alloc(gfp_t gfp_flags, void *data) if (!bio) goto out_free_bio; r10_bio->devs[j].bio = bio; + if (!conf->have_replacement) + continue; + bio = bio_kmalloc(gfp_flags, RESYNC_PAGES); + if (!bio) + goto out_free_bio; + r10_bio->devs[j].repl_bio = bio; } /* * Allocate RESYNC_PAGES data pages and attach them * where needed. */ for (j = 0 ; j < nalloc; j++) { + struct bio *rbio = r10_bio->devs[j].repl_bio; bio = r10_bio->devs[j].bio; for (i = 0; i < RESYNC_PAGES; i++) { if (j == 1 && !test_bit(MD_RECOVERY_SYNC, @@ -143,6 +151,8 @@ static void * r10buf_pool_alloc(gfp_t gfp_flags, void *data) goto out_free_pages; bio->bi_io_vec[i].bv_page = page; + if (rbio) + rbio->bi_io_vec[i].bv_page = page; } } @@ -156,8 +166,11 @@ out_free_pages: safe_put_page(r10_bio->devs[j].bio->bi_io_vec[i].bv_page); j = -1; out_free_bio: - while ( ++j < nalloc ) + while (++j < nalloc) { bio_put(r10_bio->devs[j].bio); + if (r10_bio->devs[j].repl_bio) + bio_put(r10_bio->devs[j].repl_bio); + } r10bio_pool_free(r10_bio, conf); return NULL; } @@ -178,6 +191,9 @@ static void r10buf_pool_free(void *__r10_bio, void *data) } bio_put(bio); } + bio = r10bio->devs[j].repl_bio; + if (bio) + bio_put(bio); } r10bio_pool_free(r10bio, conf); } @@ -191,6 +207,10 @@ static void put_all_bios(struct r10conf *conf, struct r10bio *r10_bio) if (!BIO_SPECIAL(*bio)) bio_put(*bio); *bio = NULL; + bio = &r10_bio->devs[i].repl_bio; + if (r10_bio->read_slot < 0 && !BIO_SPECIAL(*bio)) + bio_put(*bio); + *bio = NULL; } } @@ -275,19 +295,27 @@ static inline void update_head_pos(int slot, struct r10bio *r10_bio) * Find the disk number which triggered given bio */ static int find_bio_disk(struct r10conf *conf, struct r10bio *r10_bio, - struct bio *bio, int *slotp) + struct bio *bio, int *slotp, int *replp) { int slot; + int repl = 0; - for (slot = 0; slot < conf->copies; slot++) + for (slot = 0; slot < conf->copies; slot++) { if (r10_bio->devs[slot].bio == bio) break; + if (r10_bio->devs[slot].repl_bio == bio) { + repl = 1; + break; + } + } BUG_ON(slot == conf->copies); update_head_pos(slot, r10_bio); if (slotp) *slotp = slot; + if (replp) + *replp = repl; return r10_bio->devs[slot].devnum; } @@ -368,7 +396,7 @@ static void raid10_end_write_request(struct bio *bio, int error) struct r10conf *conf = r10_bio->mddev->private; int slot; - dev = find_bio_disk(conf, r10_bio, bio, &slot); + dev = find_bio_disk(conf, r10_bio, bio, &slot, NULL); /* * this branch is our 'one mirror IO has finished' event handler: @@ -1025,6 +1053,7 @@ read_again: */ plugged = mddev_check_plugged(mddev); + r10_bio->read_slot = -1; /* make sure repl_bio gets freed */ raid10_find_phys(conf, r10_bio); retry_write: blocked_rdev = NULL; @@ -1431,7 +1460,7 @@ static void end_sync_read(struct bio *bio, int error) struct r10conf *conf = r10_bio->mddev->private; int d; - d = find_bio_disk(conf, r10_bio, bio, NULL); + d = find_bio_disk(conf, r10_bio, bio, NULL, NULL); if (test_bit(BIO_UPTODATE, &bio->bi_flags)) set_bit(R10BIO_Uptodate, &r10_bio->state); @@ -1493,7 +1522,7 @@ static void end_sync_write(struct bio *bio, int error) int bad_sectors; int slot; - d = find_bio_disk(conf, r10_bio, bio, &slot); + d = find_bio_disk(conf, r10_bio, bio, &slot, NULL); if (!uptodate) { set_bit(WriteErrorSeen, &conf->mirrors[d].rdev->flags); @@ -2271,9 +2300,14 @@ static void raid10d(struct mddev *mddev) static int init_resync(struct r10conf *conf) { int buffs; + int i; buffs = RESYNC_WINDOW / RESYNC_BLOCK_SIZE; BUG_ON(conf->r10buf_pool); + conf->have_replacement = 0; + for (i = 0; i < conf->raid_disks; i++) + if (conf->mirrors[i].replacement) + conf->have_replacement = 1; conf->r10buf_pool = mempool_create(buffs, r10buf_pool_alloc, r10buf_pool_free, conf); if (!conf->r10buf_pool) return -ENOMEM; diff --git a/drivers/md/raid10.h b/drivers/md/raid10.h index 7facfdf841f4..7c615613c381 100644 --- a/drivers/md/raid10.h +++ b/drivers/md/raid10.h @@ -2,7 +2,7 @@ #define _RAID10_H struct mirror_info { - struct md_rdev *rdev; + struct md_rdev *rdev, *replacement; sector_t head_position; int recovery_disabled; /* matches * mddev->recovery_disabled @@ -18,12 +18,13 @@ struct r10conf { spinlock_t device_lock; /* geometry */ - int near_copies; /* number of copies laid out raid0 style */ + int near_copies; /* number of copies laid out + * raid0 style */ int far_copies; /* number of copies laid out * at large strides across drives */ - int far_offset; /* far_copies are offset by 1 stripe - * instead of many + int far_offset; /* far_copies are offset by 1 + * stripe instead of many */ int copies; /* near_copies * far_copies. * must be <= raid_disks @@ -34,10 +35,11 @@ struct r10conf { * 1 stripe. */ - sector_t dev_sectors; /* temp copy of mddev->dev_sectors */ + sector_t dev_sectors; /* temp copy of + * mddev->dev_sectors */ - int chunk_shift; /* shift from chunks to sectors */ - sector_t chunk_mask; + int chunk_shift; /* shift from chunks to sectors */ + sector_t chunk_mask; struct list_head retry_list; /* queue pending writes and submit them on unplug */ @@ -45,20 +47,22 @@ struct r10conf { int pending_count; spinlock_t resync_lock; - int nr_pending; - int nr_waiting; - int nr_queued; - int barrier; + int nr_pending; + int nr_waiting; + int nr_queued; + int barrier; sector_t next_resync; int fullsync; /* set to 1 if a full sync is needed, * (fresh device added). * Cleared when a sync completes. */ - + int have_replacement; /* There is at least one + * replacement device. + */ wait_queue_head_t wait_barrier; - mempool_t *r10bio_pool; - mempool_t *r10buf_pool; + mempool_t *r10bio_pool; + mempool_t *r10buf_pool; struct page *tmppage; /* When taking over an array from a different personality, we store @@ -98,11 +102,18 @@ struct r10bio { * When resyncing we also use one for each copy. * When reconstructing, we use 2 bios, one for read, one for write. * We choose the number when they are allocated. + * We sometimes need an extra bio to write to the replacement. */ struct { - struct bio *bio; - sector_t addr; - int devnum; + struct bio *bio; + union { + struct bio *repl_bio; /* used for resync and + * writes */ + struct md_rdev *rdev; /* used for reads + * (read_slot >= 0) */ + }; + sector_t addr; + int devnum; } devs[0]; }; @@ -121,17 +132,19 @@ struct r10bio { #define BIO_SPECIAL(bio) ((unsigned long)bio <= 2) /* bits for r10bio.state */ -#define R10BIO_Uptodate 0 -#define R10BIO_IsSync 1 -#define R10BIO_IsRecover 2 -#define R10BIO_Degraded 3 +enum r10bio_state { + R10BIO_Uptodate, + R10BIO_IsSync, + R10BIO_IsRecover, + R10BIO_Degraded, /* Set ReadError on bios that experience a read error * so that raid10d knows what to do with them. */ -#define R10BIO_ReadError 4 + R10BIO_ReadError, /* If a write for this request means we can clear some * known-bad-block records, we set this flag. */ -#define R10BIO_MadeGood 5 -#define R10BIO_WriteError 6 + R10BIO_MadeGood, + R10BIO_WriteError, +}; #endif -- cgit v1.2.3 From 96c3fd1f3802371610c620cff03f9d825707e80e Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:54 +1100 Subject: md/raid10: change read_balance to return an rdev It makes more sense to return an rdev than just an index as read_balance() gets a reference to the rdev and so returning the pointer make this more idiomatic. This will be needed in a future patch when we might return a 'replacement' rdev instead of the main rdev. Signed-off-by: NeilBrown --- drivers/md/raid10.c | 27 +++++++++++++-------------- 1 file changed, 13 insertions(+), 14 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index 9722065022fa..70356c130273 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c @@ -590,7 +590,9 @@ static int raid10_mergeable_bvec(struct request_queue *q, * FIXME: possibly should rethink readbalancing and do it differently * depending on near_copies / far_copies geometry. */ -static int read_balance(struct r10conf *conf, struct r10bio *r10_bio, int *max_sectors) +static struct md_rdev *read_balance(struct r10conf *conf, + struct r10bio *r10_bio, + int *max_sectors) { const sector_t this_sector = r10_bio->sector; int disk, slot; @@ -703,11 +705,11 @@ retry: } r10_bio->read_slot = slot; } else - disk = -1; + rdev = NULL; rcu_read_unlock(); *max_sectors = best_good_sectors; - return disk; + return rdev; } static int raid10_congested(void *data, int bits) @@ -874,7 +876,6 @@ static void unfreeze_array(struct r10conf *conf) static void make_request(struct mddev *mddev, struct bio * bio) { struct r10conf *conf = mddev->private; - struct mirror_info *mirror; struct r10bio *r10_bio; struct bio *read_bio; int i; @@ -973,17 +974,16 @@ static void make_request(struct mddev *mddev, struct bio * bio) /* * read balancing logic: */ - int disk; + struct md_rdev *rdev; int slot; read_again: - disk = read_balance(conf, r10_bio, &max_sectors); - slot = r10_bio->read_slot; - if (disk < 0) { + rdev = read_balance(conf, r10_bio, &max_sectors); + if (!rdev) { raid_end_bio_io(r10_bio); return; } - mirror = conf->mirrors + disk; + slot = r10_bio->read_slot; read_bio = bio_clone_mddev(bio, GFP_NOIO, mddev); md_trim_bio(read_bio, r10_bio->sector - bio->bi_sector, @@ -992,8 +992,8 @@ read_again: r10_bio->devs[slot].bio = read_bio; read_bio->bi_sector = r10_bio->devs[slot].addr + - mirror->rdev->data_offset; - read_bio->bi_bdev = mirror->rdev->bdev; + rdev->data_offset; + read_bio->bi_bdev = rdev->bdev; read_bio->bi_end_io = raid10_end_read_request; read_bio->bi_rw = READ | do_sync; read_bio->bi_private = r10_bio; @@ -2116,8 +2116,8 @@ static void handle_read_error(struct mddev *mddev, struct r10bio *r10_bio) r10_bio->devs[slot].bio = mddev->ro ? IO_BLOCKED : NULL; read_more: - mirror = read_balance(conf, r10_bio, &max_sectors); - if (mirror == -1) { + rdev = read_balance(conf, r10_bio, &max_sectors); + if (rdev == NULL) { printk(KERN_ALERT "md/raid10:%s: %s: unrecoverable I/O" " read error for block %llu\n", mdname(mddev), b, @@ -2131,7 +2131,6 @@ read_more: if (bio) bio_put(bio); slot = r10_bio->read_slot; - rdev = conf->mirrors[mirror].rdev; printk_ratelimited( KERN_ERR "md/raid10:%s: %s: redirecting" -- cgit v1.2.3 From abbf098e6e1e23d5d247b9eaaf325e67f67b0328 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:54 +1100 Subject: md/raid10: preferentially read from replacement device if possible. When reading (for array reads, not for recovery etc) we read from the replacement device if it has recovered far enough. This requires storing the chosen rdev in the 'r10_bio' so we can make sure to drop the ref on the right device when the read finishes. Signed-off-by: NeilBrown --- drivers/md/raid10.c | 36 +++++++++++++++++++++++------------- 1 file changed, 23 insertions(+), 13 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index 70356c130273..5b886218110e 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c @@ -324,11 +324,13 @@ static void raid10_end_read_request(struct bio *bio, int error) int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags); struct r10bio *r10_bio = bio->bi_private; int slot, dev; + struct md_rdev *rdev; struct r10conf *conf = r10_bio->mddev->private; slot = r10_bio->read_slot; dev = r10_bio->devs[slot].devnum; + rdev = r10_bio->devs[slot].rdev; /* * this branch is our 'one mirror IO has finished' event handler: */ @@ -346,7 +348,7 @@ static void raid10_end_read_request(struct bio *bio, int error) */ set_bit(R10BIO_Uptodate, &r10_bio->state); raid_end_bio_io(r10_bio); - rdev_dec_pending(conf->mirrors[dev].rdev, conf->mddev); + rdev_dec_pending(rdev, conf->mddev); } else { /* * oops, read error - keep the refcount on the rdev @@ -355,7 +357,7 @@ static void raid10_end_read_request(struct bio *bio, int error) printk_ratelimited(KERN_ERR "md/raid10:%s: %s: rescheduling sector %llu\n", mdname(conf->mddev), - bdevname(conf->mirrors[dev].rdev->bdev, b), + bdevname(rdev->bdev, b), (unsigned long long)r10_bio->sector); set_bit(R10BIO_ReadError, &r10_bio->state); reschedule_retry(r10_bio); @@ -599,7 +601,7 @@ static struct md_rdev *read_balance(struct r10conf *conf, int sectors = r10_bio->sectors; int best_good_sectors; sector_t new_distance, best_dist; - struct md_rdev *rdev; + struct md_rdev *rdev, *best_rdev; int do_balance; int best_slot; @@ -608,6 +610,7 @@ static struct md_rdev *read_balance(struct r10conf *conf, retry: sectors = r10_bio->sectors; best_slot = -1; + best_rdev = NULL; best_dist = MaxSector; best_good_sectors = 0; do_balance = 1; @@ -629,10 +632,16 @@ retry: if (r10_bio->devs[slot].bio == IO_BLOCKED) continue; disk = r10_bio->devs[slot].devnum; - rdev = rcu_dereference(conf->mirrors[disk].rdev); + rdev = rcu_dereference(conf->mirrors[disk].replacement); + if (rdev == NULL || test_bit(Faulty, &rdev->flags) || + r10_bio->devs[slot].addr + sectors > rdev->recovery_offset) + rdev = rcu_dereference(conf->mirrors[disk].rdev); if (rdev == NULL) continue; - if (!test_bit(In_sync, &rdev->flags)) + if (test_bit(Faulty, &rdev->flags)) + continue; + if (!test_bit(In_sync, &rdev->flags) && + r10_bio->devs[slot].addr + sectors > rdev->recovery_offset) continue; dev_sector = r10_bio->devs[slot].addr; @@ -657,6 +666,7 @@ retry: if (good_sectors > best_good_sectors) { best_good_sectors = good_sectors; best_slot = slot; + best_rdev = rdev; } if (!do_balance) /* Must read from here */ @@ -685,16 +695,15 @@ retry: if (new_distance < best_dist) { best_dist = new_distance; best_slot = slot; + best_rdev = rdev; } } - if (slot == conf->copies) + if (slot >= conf->copies) { slot = best_slot; + rdev = best_rdev; + } if (slot >= 0) { - disk = r10_bio->devs[slot].devnum; - rdev = rcu_dereference(conf->mirrors[disk].rdev); - if (!rdev) - goto retry; atomic_inc(&rdev->nr_pending); if (test_bit(Faulty, &rdev->flags)) { /* Cannot risk returning a device that failed @@ -990,6 +999,7 @@ read_again: max_sectors); r10_bio->devs[slot].bio = read_bio; + r10_bio->devs[slot].rdev = rdev; read_bio->bi_sector = r10_bio->devs[slot].addr + rdev->data_offset; @@ -2088,10 +2098,9 @@ static int narrow_write_error(struct r10bio *r10_bio, int i) static void handle_read_error(struct mddev *mddev, struct r10bio *r10_bio) { int slot = r10_bio->read_slot; - int mirror = r10_bio->devs[slot].devnum; struct bio *bio; struct r10conf *conf = mddev->private; - struct md_rdev *rdev; + struct md_rdev *rdev = r10_bio->devs[slot].rdev; char b[BDEVNAME_SIZE]; unsigned long do_sync; int max_sectors; @@ -2109,7 +2118,7 @@ static void handle_read_error(struct mddev *mddev, struct r10bio *r10_bio) fix_read_error(conf, mddev, r10_bio); unfreeze_array(conf); } - rdev_dec_pending(conf->mirrors[mirror].rdev, mddev); + rdev_dec_pending(rdev, mddev); bio = r10_bio->devs[slot].bio; bdevname(bio->bi_bdev, b); @@ -2144,6 +2153,7 @@ read_more: r10_bio->sector - bio->bi_sector, max_sectors); r10_bio->devs[slot].bio = bio; + r10_bio->devs[slot].rdev = rdev; bio->bi_sector = r10_bio->devs[slot].addr + rdev->data_offset; bio->bi_bdev = rdev->bdev; -- cgit v1.2.3 From c8ab903ea9d7309044910c33dc087418be84f9b5 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:54 +1100 Subject: md/raid10: allow removal of failed replacement devices. Enhance raid10_remove_disk to be able to remove ->replacement as well as ->rdev Signed-off-by: NeilBrown --- drivers/md/raid10.c | 57 ++++++++++++++++++++++++++++++----------------------- 1 file changed, 32 insertions(+), 25 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index 5b886218110e..b2c8998e1397 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c @@ -1429,34 +1429,41 @@ static int raid10_remove_disk(struct mddev *mddev, struct md_rdev *rdev) struct r10conf *conf = mddev->private; int err = 0; int number = rdev->raid_disk; - struct mirror_info *p = conf->mirrors+ number; + struct md_rdev **rdevp; + struct mirror_info *p = conf->mirrors + number; print_conf(conf); - if (rdev == p->rdev) { - if (test_bit(In_sync, &rdev->flags) || - atomic_read(&rdev->nr_pending)) { - err = -EBUSY; - goto abort; - } - /* Only remove faulty devices in recovery - * is not possible. - */ - if (!test_bit(Faulty, &rdev->flags) && - mddev->recovery_disabled != p->recovery_disabled && - enough(conf, -1)) { - err = -EBUSY; - goto abort; - } - p->rdev = NULL; - synchronize_rcu(); - if (atomic_read(&rdev->nr_pending)) { - /* lost the race, try later */ - err = -EBUSY; - p->rdev = rdev; - goto abort; - } - err = md_integrity_register(mddev); + if (rdev == p->rdev) + rdevp = &p->rdev; + else if (rdev == p->replacement) + rdevp = &p->replacement; + else + return 0; + + if (test_bit(In_sync, &rdev->flags) || + atomic_read(&rdev->nr_pending)) { + err = -EBUSY; + goto abort; + } + /* Only remove faulty devices if recovery + * is not possible. + */ + if (!test_bit(Faulty, &rdev->flags) && + mddev->recovery_disabled != p->recovery_disabled && + enough(conf, -1)) { + err = -EBUSY; + goto abort; } + *rdevp = NULL; + synchronize_rcu(); + if (atomic_read(&rdev->nr_pending)) { + /* lost the race, try later */ + err = -EBUSY; + *rdevp = rdev; + goto abort; + } + err = md_integrity_register(mddev); + abort: print_conf(conf); -- cgit v1.2.3 From 475b0321a4df381f64db10ddd750a8b7bb82d88b Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:55 +1100 Subject: md/raid10: writes should get directed to replacement as well as original. When writing, we need to submit two writes, one to the original, and one to the replacements - if there is a replacement. If the write to the replacement results in a write error we just fail the device. We only try to record write errors to the original. This only handles writing new data. Writing for resync/recovery will come later. Signed-off-by: NeilBrown --- drivers/md/raid10.c | 83 +++++++++++++++++++++++++++++++++++++++++++++++------ 1 file changed, 74 insertions(+), 9 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index b2c8998e1397..813f52464f8a 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c @@ -396,17 +396,29 @@ static void raid10_end_write_request(struct bio *bio, int error) int dev; int dec_rdev = 1; struct r10conf *conf = r10_bio->mddev->private; - int slot; + int slot, repl; + struct md_rdev *rdev; - dev = find_bio_disk(conf, r10_bio, bio, &slot, NULL); + dev = find_bio_disk(conf, r10_bio, bio, &slot, &repl); + if (repl) + rdev = conf->mirrors[dev].replacement; + else + rdev = conf->mirrors[dev].rdev; /* * this branch is our 'one mirror IO has finished' event handler: */ if (!uptodate) { - set_bit(WriteErrorSeen, &conf->mirrors[dev].rdev->flags); - set_bit(R10BIO_WriteError, &r10_bio->state); - dec_rdev = 0; + if (repl) + /* Never record new bad blocks to replacement, + * just fail it. + */ + md_error(rdev->mddev, rdev); + else { + set_bit(WriteErrorSeen, &rdev->flags); + set_bit(R10BIO_WriteError, &r10_bio->state); + dec_rdev = 0; + } } else { /* * Set R10BIO_Uptodate in our master bio, so that @@ -423,12 +435,15 @@ static void raid10_end_write_request(struct bio *bio, int error) set_bit(R10BIO_Uptodate, &r10_bio->state); /* Maybe we can clear some bad blocks. */ - if (is_badblock(conf->mirrors[dev].rdev, + if (is_badblock(rdev, r10_bio->devs[slot].addr, r10_bio->sectors, &first_bad, &bad_sectors)) { bio_put(bio); - r10_bio->devs[slot].bio = IO_MADE_GOOD; + if (repl) + r10_bio->devs[slot].repl_bio = IO_MADE_GOOD; + else + r10_bio->devs[slot].bio = IO_MADE_GOOD; dec_rdev = 0; set_bit(R10BIO_MadeGood, &r10_bio->state); } @@ -444,7 +459,6 @@ static void raid10_end_write_request(struct bio *bio, int error) rdev_dec_pending(conf->mirrors[dev].rdev, conf->mddev); } - /* * RAID10 layout manager * As well as the chunksize and raid_disks count, there are two @@ -1073,12 +1087,23 @@ retry_write: for (i = 0; i < conf->copies; i++) { int d = r10_bio->devs[i].devnum; struct md_rdev *rdev = rcu_dereference(conf->mirrors[d].rdev); + struct md_rdev *rrdev = rcu_dereference( + conf->mirrors[d].replacement); if (rdev && unlikely(test_bit(Blocked, &rdev->flags))) { atomic_inc(&rdev->nr_pending); blocked_rdev = rdev; break; } + if (rrdev && unlikely(test_bit(Blocked, &rrdev->flags))) { + atomic_inc(&rrdev->nr_pending); + blocked_rdev = rrdev; + break; + } + if (rrdev && test_bit(Faulty, &rrdev->flags)) + rrdev = NULL; + r10_bio->devs[i].bio = NULL; + r10_bio->devs[i].repl_bio = NULL; if (!rdev || test_bit(Faulty, &rdev->flags)) { set_bit(R10BIO_Degraded, &r10_bio->state); continue; @@ -1127,6 +1152,10 @@ retry_write: } r10_bio->devs[i].bio = bio; atomic_inc(&rdev->nr_pending); + if (rrdev) { + r10_bio->devs[i].repl_bio = bio; + atomic_inc(&rrdev->nr_pending); + } } rcu_read_unlock(); @@ -1135,11 +1164,17 @@ retry_write: int j; int d; - for (j = 0; j < i; j++) + for (j = 0; j < i; j++) { if (r10_bio->devs[j].bio) { d = r10_bio->devs[j].devnum; rdev_dec_pending(conf->mirrors[d].rdev, mddev); } + if (r10_bio->devs[j].repl_bio) { + d = r10_bio->devs[j].devnum; + rdev_dec_pending( + conf->mirrors[d].replacement, mddev); + } + } allow_barrier(conf); md_wait_for_blocked_rdev(blocked_rdev, mddev); wait_barrier(conf); @@ -1186,6 +1221,27 @@ retry_write: bio_list_add(&conf->pending_bio_list, mbio); conf->pending_count++; spin_unlock_irqrestore(&conf->device_lock, flags); + + if (!r10_bio->devs[i].repl_bio) + continue; + + mbio = bio_clone_mddev(bio, GFP_NOIO, mddev); + md_trim_bio(mbio, r10_bio->sector - bio->bi_sector, + max_sectors); + r10_bio->devs[i].repl_bio = mbio; + + mbio->bi_sector = (r10_bio->devs[i].addr+ + conf->mirrors[d].replacement->data_offset); + mbio->bi_bdev = conf->mirrors[d].replacement->bdev; + mbio->bi_end_io = raid10_end_write_request; + mbio->bi_rw = WRITE | do_sync | do_fua; + mbio->bi_private = r10_bio; + + atomic_inc(&r10_bio->remaining); + spin_lock_irqsave(&conf->device_lock, flags); + bio_list_add(&conf->pending_bio_list, mbio); + conf->pending_count++; + spin_unlock_irqrestore(&conf->device_lock, flags); } /* Don't remove the bias on 'remaining' (one_write_done) until @@ -2253,6 +2309,15 @@ static void handle_write_completed(struct r10conf *conf, struct r10bio *r10_bio) } rdev_dec_pending(rdev, conf->mddev); } + bio = r10_bio->devs[m].repl_bio; + rdev = conf->mirrors[dev].replacement; + if (bio == IO_MADE_GOOD) { + rdev_clear_badblocks( + rdev, + r10_bio->devs[m].addr, + r10_bio->sectors); + rdev_dec_pending(rdev, conf->mddev); + } } if (test_bit(R10BIO_WriteError, &r10_bio->state)) -- cgit v1.2.3 From 9ad1aefc8ae8d2e482b4cc4b7199e2354148bbdc Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:55 +1100 Subject: md/raid10: Handle replacement devices during resync. If we need to resync an array which has replacement devices, we always write any block checked to every replacement. If the resync was bitmap-based resync we will then complete the replacement normally. If it was a full resync, we mark the replacements as fully recovered when the resync finishes so no further recovery is needed. Signed-off-by: NeilBrown --- drivers/md/raid10.c | 105 ++++++++++++++++++++++++++++++++++++++++++++++++---- 1 file changed, 98 insertions(+), 7 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index 813f52464f8a..8a6862be9f58 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c @@ -1594,19 +1594,29 @@ static void end_sync_write(struct bio *bio, int error) sector_t first_bad; int bad_sectors; int slot; + int repl; + struct md_rdev *rdev; - d = find_bio_disk(conf, r10_bio, bio, &slot, NULL); + d = find_bio_disk(conf, r10_bio, bio, &slot, &repl); + if (repl) + rdev = conf->mirrors[d].replacement; + else + rdev = conf->mirrors[d].rdev; if (!uptodate) { - set_bit(WriteErrorSeen, &conf->mirrors[d].rdev->flags); - set_bit(R10BIO_WriteError, &r10_bio->state); - } else if (is_badblock(conf->mirrors[d].rdev, + if (repl) + md_error(mddev, rdev); + else { + set_bit(WriteErrorSeen, &rdev->flags); + set_bit(R10BIO_WriteError, &r10_bio->state); + } + } else if (is_badblock(rdev, r10_bio->devs[slot].addr, r10_bio->sectors, &first_bad, &bad_sectors)) set_bit(R10BIO_MadeGood, &r10_bio->state); - rdev_dec_pending(conf->mirrors[d].rdev, mddev); + rdev_dec_pending(rdev, mddev); end_sync_request(r10_bio); } @@ -1710,6 +1720,29 @@ static void sync_request_write(struct mddev *mddev, struct r10bio *r10_bio) generic_make_request(tbio); } + /* Now write out to any replacement devices + * that are active + */ + for (i = 0; i < conf->copies; i++) { + int j, d; + int vcnt = r10_bio->sectors >> (PAGE_SHIFT-9); + + tbio = r10_bio->devs[i].repl_bio; + if (!tbio || !tbio->bi_end_io) + continue; + if (r10_bio->devs[i].bio->bi_end_io != end_sync_write + && r10_bio->devs[i].bio != fbio) + for (j = 0; j < vcnt; j++) + memcpy(page_address(tbio->bi_io_vec[j].bv_page), + page_address(fbio->bi_io_vec[j].bv_page), + PAGE_SIZE); + d = r10_bio->devs[i].devnum; + atomic_inc(&r10_bio->remaining); + md_sync_acct(conf->mirrors[d].replacement->bdev, + tbio->bi_size >> 9); + generic_make_request(tbio); + } + done: if (atomic_dec_and_test(&r10_bio->remaining)) { md_done_sync(mddev, r10_bio->sectors, 1); @@ -2287,6 +2320,22 @@ static void handle_write_completed(struct r10conf *conf, struct r10bio *r10_bio) r10_bio->sectors, 0)) md_error(conf->mddev, rdev); } + rdev = conf->mirrors[dev].replacement; + if (r10_bio->devs[m].repl_bio == NULL) + continue; + if (test_bit(BIO_UPTODATE, + &r10_bio->devs[m].repl_bio->bi_flags)) { + rdev_clear_badblocks( + rdev, + r10_bio->devs[m].addr, + r10_bio->sectors); + } else { + if (!rdev_set_badblocks( + rdev, + r10_bio->devs[m].addr, + r10_bio->sectors, 0)) + md_error(conf->mddev, rdev); + } } put_buf(r10_bio); } else { @@ -2469,9 +2518,22 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, bitmap_end_sync(mddev->bitmap, sect, &sync_blocks, 1); } - } else /* completed sync */ + } else { + /* completed sync */ + if ((!mddev->bitmap || conf->fullsync) + && conf->have_replacement + && test_bit(MD_RECOVERY_SYNC, &mddev->recovery)) { + /* Completed a full sync so the replacements + * are now fully recovered. + */ + for (i = 0; i < conf->raid_disks; i++) + if (conf->mirrors[i].replacement) + conf->mirrors[i].replacement + ->recovery_offset + = MaxSector; + } conf->fullsync = 0; - + } bitmap_close_sync(mddev->bitmap); close_sync(conf); *skipped = 1; @@ -2719,6 +2781,9 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, sector_t first_bad, sector; int bad_sectors; + if (r10_bio->devs[i].repl_bio) + r10_bio->devs[i].repl_bio->bi_end_io = NULL; + bio = r10_bio->devs[i].bio; bio->bi_end_io = NULL; clear_bit(BIO_UPTODATE, &bio->bi_flags); @@ -2749,6 +2814,27 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, conf->mirrors[d].rdev->data_offset; bio->bi_bdev = conf->mirrors[d].rdev->bdev; count++; + + if (conf->mirrors[d].replacement == NULL || + test_bit(Faulty, + &conf->mirrors[d].replacement->flags)) + continue; + + /* Need to set up for writing to the replacement */ + bio = r10_bio->devs[i].repl_bio; + clear_bit(BIO_UPTODATE, &bio->bi_flags); + + sector = r10_bio->devs[i].addr; + atomic_inc(&conf->mirrors[d].rdev->nr_pending); + bio->bi_next = biolist; + biolist = bio; + bio->bi_private = r10_bio; + bio->bi_end_io = end_sync_write; + bio->bi_rw = WRITE; + bio->bi_sector = sector + + conf->mirrors[d].replacement->data_offset; + bio->bi_bdev = conf->mirrors[d].replacement->bdev; + count++; } if (count < 2) { @@ -2757,6 +2843,11 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, if (r10_bio->devs[i].bio->bi_end_io) rdev_dec_pending(conf->mirrors[d].rdev, mddev); + if (r10_bio->devs[i].repl_bio && + r10_bio->devs[i].repl_bio->bi_end_io) + rdev_dec_pending( + conf->mirrors[d].replacement, + mddev); } put_buf(r10_bio); biolist = NULL; -- cgit v1.2.3 From 24afd80d99f80a79d8824d2805114b8b067e9823 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:55 +1100 Subject: md/raid10: handle recovery of replacement devices. If there is a replacement device, then recover to it, reading from any drives - maybe the one being replaced, maybe not. Signed-off-by: NeilBrown --- drivers/md/raid10.c | 110 ++++++++++++++++++++++++++++++++++++++-------------- 1 file changed, 80 insertions(+), 30 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index 8a6862be9f58..403f05ac1f2a 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c @@ -1843,7 +1843,7 @@ static void recovery_request_write(struct mddev *mddev, struct r10bio *r10_bio) { struct r10conf *conf = mddev->private; int d; - struct bio *wbio; + struct bio *wbio, *wbio2; if (!test_bit(R10BIO_Uptodate, &r10_bio->state)) { fix_recovery_read_error(r10_bio); @@ -1855,12 +1855,20 @@ static void recovery_request_write(struct mddev *mddev, struct r10bio *r10_bio) * share the pages with the first bio * and submit the write request */ - wbio = r10_bio->devs[1].bio; d = r10_bio->devs[1].devnum; - - atomic_inc(&conf->mirrors[d].rdev->nr_pending); - md_sync_acct(conf->mirrors[d].rdev->bdev, wbio->bi_size >> 9); - generic_make_request(wbio); + wbio = r10_bio->devs[1].bio; + wbio2 = r10_bio->devs[1].repl_bio; + if (wbio->bi_end_io) { + atomic_inc(&conf->mirrors[d].rdev->nr_pending); + md_sync_acct(conf->mirrors[d].rdev->bdev, wbio->bi_size >> 9); + generic_make_request(wbio); + } + if (wbio2 && wbio2->bi_end_io) { + atomic_inc(&conf->mirrors[d].replacement->nr_pending); + md_sync_acct(conf->mirrors[d].replacement->bdev, + wbio2->bi_size >> 9); + generic_make_request(wbio2); + } } @@ -2590,23 +2598,30 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, sector_t sect; int must_sync; int any_working; - - if (conf->mirrors[i].rdev == NULL || - test_bit(In_sync, &conf->mirrors[i].rdev->flags)) + struct mirror_info *mirror = &conf->mirrors[i]; + + if ((mirror->rdev == NULL || + test_bit(In_sync, &mirror->rdev->flags)) + && + (mirror->replacement == NULL || + test_bit(Faulty, + &mirror->replacement->flags))) continue; still_degraded = 0; /* want to reconstruct this device */ rb2 = r10_bio; sect = raid10_find_virt(conf, sector_nr, i); - /* Unless we are doing a full sync, we only need - * to recover the block if it is set in the bitmap + /* Unless we are doing a full sync, or a replacement + * we only need to recover the block if it is set in + * the bitmap */ must_sync = bitmap_start_sync(mddev->bitmap, sect, &sync_blocks, 1); if (sync_blocks < max_sync) max_sync = sync_blocks; if (!must_sync && + mirror->replacement == NULL && !conf->fullsync) { /* yep, skip the sync_blocks here, but don't assume * that there will never be anything to do here @@ -2676,33 +2691,60 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, bio->bi_end_io = end_sync_read; bio->bi_rw = READ; from_addr = r10_bio->devs[j].addr; - bio->bi_sector = from_addr + - conf->mirrors[d].rdev->data_offset; - bio->bi_bdev = conf->mirrors[d].rdev->bdev; - atomic_inc(&conf->mirrors[d].rdev->nr_pending); - atomic_inc(&r10_bio->remaining); - /* and we write to 'i' */ + bio->bi_sector = from_addr + rdev->data_offset; + bio->bi_bdev = rdev->bdev; + atomic_inc(&rdev->nr_pending); + /* and we write to 'i' (if not in_sync) */ for (k=0; kcopies; k++) if (r10_bio->devs[k].devnum == i) break; BUG_ON(k == conf->copies); - bio = r10_bio->devs[1].bio; - bio->bi_next = biolist; - biolist = bio; - bio->bi_private = r10_bio; - bio->bi_end_io = end_sync_write; - bio->bi_rw = WRITE; to_addr = r10_bio->devs[k].addr; - bio->bi_sector = to_addr + - conf->mirrors[i].rdev->data_offset; - bio->bi_bdev = conf->mirrors[i].rdev->bdev; - r10_bio->devs[0].devnum = d; r10_bio->devs[0].addr = from_addr; r10_bio->devs[1].devnum = i; r10_bio->devs[1].addr = to_addr; + rdev = mirror->rdev; + if (!test_bit(In_sync, &rdev->flags)) { + bio = r10_bio->devs[1].bio; + bio->bi_next = biolist; + biolist = bio; + bio->bi_private = r10_bio; + bio->bi_end_io = end_sync_write; + bio->bi_rw = WRITE; + bio->bi_sector = to_addr + + rdev->data_offset; + bio->bi_bdev = rdev->bdev; + atomic_inc(&r10_bio->remaining); + } else + r10_bio->devs[1].bio->bi_end_io = NULL; + + /* and maybe write to replacement */ + bio = r10_bio->devs[1].repl_bio; + if (bio) + bio->bi_end_io = NULL; + rdev = mirror->replacement; + /* Note: if rdev != NULL, then bio + * cannot be NULL as r10buf_pool_alloc will + * have allocated it. + * So the second test here is pointless. + * But it keeps semantic-checkers happy, and + * this comment keeps human reviewers + * happy. + */ + if (rdev == NULL || bio == NULL || + test_bit(Faulty, &rdev->flags)) + break; + bio->bi_next = biolist; + biolist = bio; + bio->bi_private = r10_bio; + bio->bi_end_io = end_sync_write; + bio->bi_rw = WRITE; + bio->bi_sector = to_addr + rdev->data_offset; + bio->bi_bdev = rdev->bdev; + atomic_inc(&r10_bio->remaining); break; } if (j == conf->copies) { @@ -2720,8 +2762,16 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, for (k = 0; k < conf->copies; k++) if (r10_bio->devs[k].devnum == i) break; - if (!rdev_set_badblocks( - conf->mirrors[i].rdev, + if (!test_bit(In_sync, + &mirror->rdev->flags) + && !rdev_set_badblocks( + mirror->rdev, + r10_bio->devs[k].addr, + max_sync, 0)) + any_working = 0; + if (mirror->replacement && + !rdev_set_badblocks( + mirror->replacement, r10_bio->devs[k].addr, max_sync, 0)) any_working = 0; @@ -2732,7 +2782,7 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, printk(KERN_INFO "md/raid10:%s: insufficient " "working devices for recovery.\n", mdname(mddev)); - conf->mirrors[i].recovery_disabled + mirror->recovery_disabled = mddev->recovery_disabled; } break; -- cgit v1.2.3 From 4ca40c2ce099e4f1ce35445994f49836662596c8 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:55 +1100 Subject: md/raid10: Allow replacement device to be replace old drive. When recovery finish and spare_active is called, check for a replace that might have just become fully synced and mark it as such, marking the original as failed. Then when the original is removed, move the replacement into its position. This means that 'replacement' and spontaneously become NULL in some situations. Make sure we check for those. It also means that 'rdev' and 'replacement' could appear to be identical - check for that too. Signed-off-by: NeilBrown --- drivers/md/raid10.c | 72 +++++++++++++++++++++++++++++++++++++++++++++-------- 1 file changed, 61 insertions(+), 11 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index 403f05ac1f2a..90e951730a23 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c @@ -397,14 +397,17 @@ static void raid10_end_write_request(struct bio *bio, int error) int dec_rdev = 1; struct r10conf *conf = r10_bio->mddev->private; int slot, repl; - struct md_rdev *rdev; + struct md_rdev *rdev = NULL; dev = find_bio_disk(conf, r10_bio, bio, &slot, &repl); if (repl) rdev = conf->mirrors[dev].replacement; - else + if (!rdev) { + smp_rmb(); + repl = 0; rdev = conf->mirrors[dev].rdev; + } /* * this branch is our 'one mirror IO has finished' event handler: */ @@ -1089,6 +1092,8 @@ retry_write: struct md_rdev *rdev = rcu_dereference(conf->mirrors[d].rdev); struct md_rdev *rrdev = rcu_dereference( conf->mirrors[d].replacement); + if (rdev == rrdev) + rrdev = NULL; if (rdev && unlikely(test_bit(Blocked, &rdev->flags))) { atomic_inc(&rdev->nr_pending); blocked_rdev = rdev; @@ -1170,9 +1175,15 @@ retry_write: rdev_dec_pending(conf->mirrors[d].rdev, mddev); } if (r10_bio->devs[j].repl_bio) { + struct md_rdev *rdev; d = r10_bio->devs[j].devnum; - rdev_dec_pending( - conf->mirrors[d].replacement, mddev); + rdev = conf->mirrors[d].replacement; + if (!rdev) { + /* Race with remove_disk */ + smp_mb(); + rdev = conf->mirrors[d].rdev; + } + rdev_dec_pending(rdev, mddev); } } allow_barrier(conf); @@ -1230,6 +1241,10 @@ retry_write: max_sectors); r10_bio->devs[i].repl_bio = mbio; + /* We are actively writing to the original device + * so it cannot disappear, so the replacement cannot + * become NULL here + */ mbio->bi_sector = (r10_bio->devs[i].addr+ conf->mirrors[d].replacement->data_offset); mbio->bi_bdev = conf->mirrors[d].replacement->bdev; @@ -1404,9 +1419,27 @@ static int raid10_spare_active(struct mddev *mddev) */ for (i = 0; i < conf->raid_disks; i++) { tmp = conf->mirrors + i; - if (tmp->rdev - && !test_bit(Faulty, &tmp->rdev->flags) - && !test_and_set_bit(In_sync, &tmp->rdev->flags)) { + if (tmp->replacement + && tmp->replacement->recovery_offset == MaxSector + && !test_bit(Faulty, &tmp->replacement->flags) + && !test_and_set_bit(In_sync, &tmp->replacement->flags)) { + /* Replacement has just become active */ + if (!tmp->rdev + || !test_and_clear_bit(In_sync, &tmp->rdev->flags)) + count++; + if (tmp->rdev) { + /* Replaced device not technically faulty, + * but we need to be sure it gets removed + * and never re-added. + */ + set_bit(Faulty, &tmp->rdev->flags); + sysfs_notify_dirent_safe( + tmp->rdev->sysfs_state); + } + sysfs_notify_dirent_safe(tmp->replacement->sysfs_state); + } else if (tmp->rdev + && !test_bit(Faulty, &tmp->rdev->flags) + && !test_and_set_bit(In_sync, &tmp->rdev->flags)) { count++; sysfs_notify_dirent(tmp->rdev->sysfs_state); } @@ -1506,6 +1539,7 @@ static int raid10_remove_disk(struct mddev *mddev, struct md_rdev *rdev) */ if (!test_bit(Faulty, &rdev->flags) && mddev->recovery_disabled != p->recovery_disabled && + (!p->replacement || p->replacement == rdev) && enough(conf, -1)) { err = -EBUSY; goto abort; @@ -1517,7 +1551,21 @@ static int raid10_remove_disk(struct mddev *mddev, struct md_rdev *rdev) err = -EBUSY; *rdevp = rdev; goto abort; - } + } else if (p->replacement) { + /* We must have just cleared 'rdev' */ + p->rdev = p->replacement; + clear_bit(Replacement, &p->replacement->flags); + smp_mb(); /* Make sure other CPUs may see both as identical + * but will never see neither -- if they are careful. + */ + p->replacement = NULL; + clear_bit(WantReplacement, &rdev->flags); + } else + /* We might have just remove the Replacement as faulty + * Clear the flag just in case + */ + clear_bit(WantReplacement, &rdev->flags); + err = md_integrity_register(mddev); abort: @@ -1595,13 +1643,15 @@ static void end_sync_write(struct bio *bio, int error) int bad_sectors; int slot; int repl; - struct md_rdev *rdev; + struct md_rdev *rdev = NULL; d = find_bio_disk(conf, r10_bio, bio, &slot, &repl); if (repl) rdev = conf->mirrors[d].replacement; - else + if (!rdev) { + smp_mb(); rdev = conf->mirrors[d].rdev; + } if (!uptodate) { if (repl) @@ -2368,7 +2418,7 @@ static void handle_write_completed(struct r10conf *conf, struct r10bio *r10_bio) } bio = r10_bio->devs[m].repl_bio; rdev = conf->mirrors[dev].replacement; - if (bio == IO_MADE_GOOD) { + if (rdev && bio == IO_MADE_GOOD) { rdev_clear_badblocks( rdev, r10_bio->devs[m].addr, -- cgit v1.2.3 From 56a2559bb654ae2555b2ae3b29c837615d0c45c9 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:55 +1100 Subject: md/raid10: recognise replacements when assembling array. If a Replacement is seen, file it as such. If we see two replacements (or two normal devices) for the one slot, abort. Signed-off-by: NeilBrown --- drivers/md/raid10.c | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) (limited to 'drivers/md') diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index 90e951730a23..011827d0df25 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c @@ -3201,6 +3201,16 @@ static int run(struct mddev *mddev) continue; disk = conf->mirrors + disk_idx; + if (test_bit(Replacement, &rdev->flags)) { + if (disk->replacement) + goto out_free_conf; + disk->replacement = rdev; + } else { + if (disk->rdev) + goto out_free_conf; + disk->rdev = rdev; + } + disk->rdev = rdev; disk_stack_limits(mddev->gendisk, rdev->bdev, rdev->data_offset << 9); @@ -3228,6 +3238,13 @@ static int run(struct mddev *mddev) disk = conf->mirrors + i; + if (!disk->rdev && disk->replacement) { + /* The replacement is all we have - use it */ + disk->rdev = disk->replacement; + disk->replacement = NULL; + clear_bit(Replacement, &disk->rdev->flags); + } + if (!disk->rdev || !test_bit(In_sync, &disk->rdev->flags)) { disk->head_position = 0; -- cgit v1.2.3 From b7044d41b5a09ce9082699f74c8f10e0fe59f704 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:56 +1100 Subject: md/raid10: If there is a spare and a want_replacement device, start replacement. When attempting to add a spare to a RAID10 array, also consider adding it as a replacement for a want_replacement device. Signed-off-by: NeilBrown --- drivers/md/raid10.c | 40 ++++++++++++++++++++++++++++++++++++---- 1 file changed, 36 insertions(+), 4 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index 011827d0df25..6e8aa213f0d5 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c @@ -419,6 +419,9 @@ static void raid10_end_write_request(struct bio *bio, int error) md_error(rdev->mddev, rdev); else { set_bit(WriteErrorSeen, &rdev->flags); + if (!test_and_set_bit(WantReplacement, &rdev->flags)) + set_bit(MD_RECOVERY_NEEDED, + &rdev->mddev->recovery); set_bit(R10BIO_WriteError, &r10_bio->state); dec_rdev = 0; } @@ -1481,8 +1484,25 @@ static int raid10_add_disk(struct mddev *mddev, struct md_rdev *rdev) struct mirror_info *p = &conf->mirrors[mirror]; if (p->recovery_disabled == mddev->recovery_disabled) continue; - if (p->rdev) - continue; + if (p->rdev) { + if (!test_bit(WantReplacement, &p->rdev->flags) || + p->replacement != NULL) + continue; + clear_bit(In_sync, &rdev->flags); + set_bit(Replacement, &rdev->flags); + rdev->raid_disk = mirror; + err = 0; + disk_stack_limits(mddev->gendisk, rdev->bdev, + rdev->data_offset << 9); + if (rdev->bdev->bd_disk->queue->merge_bvec_fn) { + blk_queue_max_segments(mddev->queue, 1); + blk_queue_segment_boundary(mddev->queue, + PAGE_CACHE_SIZE - 1); + } + conf->fullsync = 1; + rcu_assign_pointer(p->replacement, rdev); + break; + } disk_stack_limits(mddev->gendisk, rdev->bdev, rdev->data_offset << 9); @@ -1658,6 +1678,9 @@ static void end_sync_write(struct bio *bio, int error) md_error(mddev, rdev); else { set_bit(WriteErrorSeen, &rdev->flags); + if (!test_and_set_bit(WantReplacement, &rdev->flags)) + set_bit(MD_RECOVERY_NEEDED, + &rdev->mddev->recovery); set_bit(R10BIO_WriteError, &r10_bio->state); } } else if (is_badblock(rdev, @@ -1852,8 +1875,13 @@ static void fix_recovery_read_error(struct r10bio *r10_bio) s << 9, bio->bi_io_vec[idx].bv_page, WRITE, false); - if (!ok) + if (!ok) { set_bit(WriteErrorSeen, &rdev->flags); + if (!test_and_set_bit(WantReplacement, + &rdev->flags)) + set_bit(MD_RECOVERY_NEEDED, + &rdev->mddev->recovery); + } } if (!ok) { /* We don't worry if we cannot set a bad block - @@ -1971,8 +1999,12 @@ static int r10_sync_page_io(struct md_rdev *rdev, sector_t sector, if (sync_page_io(rdev, sector, sectors << 9, page, rw, false)) /* success */ return 1; - if (rw == WRITE) + if (rw == WRITE) { set_bit(WriteErrorSeen, &rdev->flags); + if (!test_and_set_bit(WantReplacement, &rdev->flags)) + set_bit(MD_RECOVERY_NEEDED, + &rdev->mddev->recovery); + } /* 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); -- cgit v1.2.3 From 301946364e0aa67c4cfaec82e94c389993c9f7c6 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:56 +1100 Subject: md/raid1: Replace use of mddev->raid_disks with conf->raid_disks. In general mddev->raid_disks can change unexpectedly while conf->raid_disks will only change in a very controlled way. So change some uses of one to the other. The use of mddev->raid_disks will not cause actually problems but this way is more consistent and safer in the long term. Signed-off-by: NeilBrown --- drivers/md/raid1.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c index 4317f8de143a..ab8113cff7c9 100644 --- a/drivers/md/raid1.c +++ b/drivers/md/raid1.c @@ -277,7 +277,8 @@ static inline void update_head_pos(int disk, struct r1bio *r1_bio) static int find_bio_disk(struct r1bio *r1_bio, struct bio *bio) { int mirror; - int raid_disks = r1_bio->mddev->raid_disks; + struct r1conf *conf = r1_bio->mddev->private; + int raid_disks = conf->raid_disks; for (mirror = 0; mirror < raid_disks; mirror++) if (r1_bio->bios[mirror] == bio) @@ -609,7 +610,7 @@ int md_raid1_congested(struct mddev *mddev, int bits) return 1; rcu_read_lock(); - for (i = 0; i < mddev->raid_disks; i++) { + for (i = 0; i < conf->raid_disks; i++) { struct md_rdev *rdev = rcu_dereference(conf->mirrors[i].rdev); if (rdev && !test_bit(Faulty, &rdev->flags)) { struct request_queue *q = bdev_get_queue(rdev->bdev); @@ -1286,7 +1287,7 @@ static int raid1_add_disk(struct mddev *mddev, struct md_rdev *rdev) int mirror = 0; struct mirror_info *p; int first = 0; - int last = mddev->raid_disks - 1; + int last = conf->raid_disks - 1; if (mddev->recovery_disabled == conf->recovery_disabled) return -EBUSY; -- cgit v1.2.3 From 8f19ccb2fd70deb1f278be5e75076074cfddee46 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:56 +1100 Subject: md/raid1: Allocate spare to store replacement devices and their bios. In RAID1, a replacement is much like a normal device, so we just double the size of the relevant arrays and look at all possible devices for reads and writes. This means that the array looks like it is now double the size in some way - we need to be careful about that. In particular, we checking if the array is still degraded while creating a recovery request we need to only consider the first 'half' - i.e. the real (non-replacement) devices. Signed-off-by: NeilBrown --- drivers/md/raid1.c | 64 +++++++++++++++++++++++++++++------------------------- drivers/md/raid1.h | 7 +++++- 2 files changed, 40 insertions(+), 31 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c index ab8113cff7c9..e7768e37f36a 100644 --- a/drivers/md/raid1.c +++ b/drivers/md/raid1.c @@ -135,7 +135,7 @@ out_free_pages: put_page(r1_bio->bios[j]->bi_io_vec[i].bv_page); j = -1; out_free_bio: - while ( ++j < pi->raid_disks ) + while (++j < pi->raid_disks) bio_put(r1_bio->bios[j]); r1bio_pool_free(r1_bio, data); return NULL; @@ -164,7 +164,7 @@ static void put_all_bios(struct r1conf *conf, struct r1bio *r1_bio) { int i; - for (i = 0; i < conf->raid_disks; i++) { + for (i = 0; i < conf->raid_disks * 2; i++) { struct bio **bio = r1_bio->bios + i; if (!BIO_SPECIAL(*bio)) bio_put(*bio); @@ -185,7 +185,7 @@ static void put_buf(struct r1bio *r1_bio) struct r1conf *conf = r1_bio->mddev->private; int i; - for (i=0; iraid_disks; i++) { + for (i = 0; i < conf->raid_disks * 2; i++) { struct bio *bio = r1_bio->bios[i]; if (bio->bi_end_io) rdev_dec_pending(conf->mirrors[i].rdev, r1_bio->mddev); @@ -280,11 +280,11 @@ static int find_bio_disk(struct r1bio *r1_bio, struct bio *bio) struct r1conf *conf = r1_bio->mddev->private; int raid_disks = conf->raid_disks; - for (mirror = 0; mirror < raid_disks; mirror++) + for (mirror = 0; mirror < raid_disks * 2; mirror++) if (r1_bio->bios[mirror] == bio) break; - BUG_ON(mirror == raid_disks); + BUG_ON(mirror == raid_disks * 2); update_head_pos(mirror, r1_bio); return mirror; @@ -506,7 +506,7 @@ static int read_balance(struct r1conf *conf, struct r1bio *r1_bio, int *max_sect start_disk = conf->last_used; } - for (i = 0 ; i < conf->raid_disks ; i++) { + for (i = 0 ; i < conf->raid_disks * 2 ; i++) { sector_t dist; sector_t first_bad; int bad_sectors; @@ -975,7 +975,7 @@ read_again: */ plugged = mddev_check_plugged(mddev); - disks = conf->raid_disks; + disks = conf->raid_disks * 2; retry_write: blocked_rdev = NULL; rcu_read_lock(); @@ -989,7 +989,8 @@ read_again: } r1_bio->bios[i] = NULL; if (!rdev || test_bit(Faulty, &rdev->flags)) { - set_bit(R1BIO_Degraded, &r1_bio->state); + if (i < conf->raid_disks) + set_bit(R1BIO_Degraded, &r1_bio->state); continue; } @@ -1493,7 +1494,7 @@ static int fix_sync_read_error(struct r1bio *r1_bio) } } d++; - if (d == conf->raid_disks) + if (d == conf->raid_disks * 2) d = 0; } while (!success && d != r1_bio->read_disk); @@ -1510,7 +1511,7 @@ static int fix_sync_read_error(struct r1bio *r1_bio) mdname(mddev), bdevname(bio->bi_bdev, b), (unsigned long long)r1_bio->sector); - for (d = 0; d < conf->raid_disks; d++) { + for (d = 0; d < conf->raid_disks * 2; d++) { rdev = conf->mirrors[d].rdev; if (!rdev || test_bit(Faulty, &rdev->flags)) continue; @@ -1536,7 +1537,7 @@ static int fix_sync_read_error(struct r1bio *r1_bio) /* write it back and re-read */ while (d != r1_bio->read_disk) { if (d == 0) - d = conf->raid_disks; + d = conf->raid_disks * 2; d--; if (r1_bio->bios[d]->bi_end_io != end_sync_read) continue; @@ -1551,7 +1552,7 @@ static int fix_sync_read_error(struct r1bio *r1_bio) d = start; while (d != r1_bio->read_disk) { if (d == 0) - d = conf->raid_disks; + d = conf->raid_disks * 2; d--; if (r1_bio->bios[d]->bi_end_io != end_sync_read) continue; @@ -1584,7 +1585,7 @@ static int process_checks(struct r1bio *r1_bio) int primary; int i; - for (primary = 0; primary < conf->raid_disks; primary++) + for (primary = 0; primary < conf->raid_disks * 2; primary++) if (r1_bio->bios[primary]->bi_end_io == end_sync_read && test_bit(BIO_UPTODATE, &r1_bio->bios[primary]->bi_flags)) { r1_bio->bios[primary]->bi_end_io = NULL; @@ -1592,7 +1593,7 @@ static int process_checks(struct r1bio *r1_bio) break; } r1_bio->read_disk = primary; - for (i = 0; i < conf->raid_disks; i++) { + for (i = 0; i < conf->raid_disks * 2; i++) { int j; int vcnt = r1_bio->sectors >> (PAGE_SHIFT- 9); struct bio *pbio = r1_bio->bios[primary]; @@ -1656,7 +1657,7 @@ static void sync_request_write(struct mddev *mddev, struct r1bio *r1_bio) { struct r1conf *conf = mddev->private; int i; - int disks = conf->raid_disks; + int disks = conf->raid_disks * 2; struct bio *bio, *wbio; bio = r1_bio->bios[r1_bio->read_disk]; @@ -1737,7 +1738,7 @@ static void fix_read_error(struct r1conf *conf, int read_disk, success = 1; else { d++; - if (d == conf->raid_disks) + if (d == conf->raid_disks * 2) d = 0; } } while (!success && d != read_disk); @@ -1753,7 +1754,7 @@ static void fix_read_error(struct r1conf *conf, int read_disk, start = d; while (d != read_disk) { if (d==0) - d = conf->raid_disks; + d = conf->raid_disks * 2; d--; rdev = conf->mirrors[d].rdev; if (rdev && @@ -1765,7 +1766,7 @@ static void fix_read_error(struct r1conf *conf, int read_disk, while (d != read_disk) { char b[BDEVNAME_SIZE]; if (d==0) - d = conf->raid_disks; + d = conf->raid_disks * 2; d--; rdev = conf->mirrors[d].rdev; if (rdev && @@ -1887,7 +1888,7 @@ static void handle_sync_write_finished(struct r1conf *conf, struct r1bio *r1_bio { int m; int s = r1_bio->sectors; - for (m = 0; m < conf->raid_disks ; m++) { + for (m = 0; m < conf->raid_disks * 2 ; m++) { struct md_rdev *rdev = conf->mirrors[m].rdev; struct bio *bio = r1_bio->bios[m]; if (bio->bi_end_io == NULL) @@ -1909,7 +1910,7 @@ static void handle_sync_write_finished(struct r1conf *conf, struct r1bio *r1_bio static void handle_write_finished(struct r1conf *conf, struct r1bio *r1_bio) { int m; - for (m = 0; m < conf->raid_disks ; m++) + for (m = 0; m < conf->raid_disks * 2 ; m++) if (r1_bio->bios[m] == IO_MADE_GOOD) { struct md_rdev *rdev = conf->mirrors[m].rdev; rdev_clear_badblocks(rdev, @@ -2184,7 +2185,7 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, int *skipp r1_bio->state = 0; set_bit(R1BIO_IsSync, &r1_bio->state); - for (i=0; i < conf->raid_disks; i++) { + for (i = 0; i < conf->raid_disks * 2; i++) { struct md_rdev *rdev; bio = r1_bio->bios[i]; @@ -2203,7 +2204,8 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, int *skipp rdev = rcu_dereference(conf->mirrors[i].rdev); if (rdev == NULL || test_bit(Faulty, &rdev->flags)) { - still_degraded = 1; + if (i < conf->raid_disks) + still_degraded = 1; } else if (!test_bit(In_sync, &rdev->flags)) { bio->bi_rw = WRITE; bio->bi_end_io = end_sync_write; @@ -2254,7 +2256,7 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, int *skipp * need to mark them bad on all write targets */ int ok = 1; - for (i = 0 ; i < conf->raid_disks ; i++) + for (i = 0 ; i < conf->raid_disks * 2 ; i++) if (r1_bio->bios[i]->bi_end_io == end_sync_write) { struct md_rdev *rdev = rcu_dereference(conf->mirrors[i].rdev); @@ -2323,7 +2325,7 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, int *skipp len = sync_blocks<<9; } - for (i=0 ; i < conf->raid_disks; i++) { + for (i = 0 ; i < conf->raid_disks * 2; i++) { bio = r1_bio->bios[i]; if (bio->bi_end_io) { page = bio->bi_io_vec[bio->bi_vcnt].bv_page; @@ -2356,7 +2358,7 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, int *skipp */ if (test_bit(MD_RECOVERY_REQUESTED, &mddev->recovery)) { atomic_set(&r1_bio->remaining, read_targets); - for (i=0; iraid_disks; i++) { + for (i = 0; i < conf->raid_disks * 2; i++) { bio = r1_bio->bios[i]; if (bio->bi_end_io == end_sync_read) { md_sync_acct(bio->bi_bdev, nr_sectors); @@ -2393,7 +2395,8 @@ static struct r1conf *setup_conf(struct mddev *mddev) if (!conf) goto abort; - conf->mirrors = kzalloc(sizeof(struct mirror_info)*mddev->raid_disks, + conf->mirrors = kzalloc(sizeof(struct mirror_info) + * mddev->raid_disks * 2, GFP_KERNEL); if (!conf->mirrors) goto abort; @@ -2405,7 +2408,7 @@ static struct r1conf *setup_conf(struct mddev *mddev) conf->poolinfo = kzalloc(sizeof(*conf->poolinfo), GFP_KERNEL); if (!conf->poolinfo) goto abort; - conf->poolinfo->raid_disks = mddev->raid_disks; + conf->poolinfo->raid_disks = mddev->raid_disks * 2; conf->r1bio_pool = mempool_create(NR_RAID1_BIOS, r1bio_pool_alloc, r1bio_pool_free, conf->poolinfo); @@ -2438,7 +2441,7 @@ static struct r1conf *setup_conf(struct mddev *mddev) conf->recovery_disabled = mddev->recovery_disabled - 1; conf->last_used = -1; - for (i = 0; i < conf->raid_disks; i++) { + for (i = 0; i < conf->raid_disks * 2; i++) { disk = conf->mirrors + i; @@ -2665,7 +2668,7 @@ static int raid1_reshape(struct mddev *mddev) if (!newpoolinfo) return -ENOMEM; newpoolinfo->mddev = mddev; - newpoolinfo->raid_disks = raid_disks; + newpoolinfo->raid_disks = raid_disks * 2; newpool = mempool_create(NR_RAID1_BIOS, r1bio_pool_alloc, r1bio_pool_free, newpoolinfo); @@ -2673,7 +2676,8 @@ static int raid1_reshape(struct mddev *mddev) kfree(newpoolinfo); return -ENOMEM; } - newmirrors = kzalloc(sizeof(struct mirror_info) * raid_disks, GFP_KERNEL); + newmirrors = kzalloc(sizeof(struct mirror_info) * raid_disks * 2, + GFP_KERNEL); if (!newmirrors) { kfree(newpoolinfo); mempool_destroy(newpool); diff --git a/drivers/md/raid1.h b/drivers/md/raid1.h index c732b6cce935..80ded139314c 100644 --- a/drivers/md/raid1.h +++ b/drivers/md/raid1.h @@ -12,6 +12,9 @@ struct mirror_info { * pool was allocated for, so they know how much to allocate and free. * mddev->raid_disks cannot be used, as it can change while a pool is active * These two datums are stored in a kmalloced struct. + * The 'raid_disks' here is twice the raid_disks in r1conf. + * This allows space for each 'real' device can have a replacement in the + * second half of the array. */ struct pool_info { @@ -21,7 +24,9 @@ struct pool_info { struct r1conf { struct mddev *mddev; - struct mirror_info *mirrors; + struct mirror_info *mirrors; /* twice 'raid_disks' to + * allow for replacements. + */ int raid_disks; /* When choose the best device for a read (read_balance()) -- cgit v1.2.3 From b014f14c81bc5db4e40e2a4db10998fd79c1c64c Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:56 +1100 Subject: md/raid1: Allow a failed replacement device to be removed. Replacement devices are stored at a different offset, so look there too. Signed-off-by: NeilBrown --- drivers/md/raid1.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'drivers/md') diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c index e7768e37f36a..3d9aa0c92821 100644 --- a/drivers/md/raid1.c +++ b/drivers/md/raid1.c @@ -1336,6 +1336,9 @@ static int raid1_remove_disk(struct mddev *mddev, struct md_rdev *rdev) int number = rdev->raid_disk; struct mirror_info *p = conf->mirrors+ number; + if (rdev != p->rdev) + p = conf->mirrors + conf->raid_disks + number; + print_conf(conf); if (rdev == p->rdev) { if (test_bit(In_sync, &rdev->flags) || @@ -1359,6 +1362,9 @@ static int raid1_remove_disk(struct mddev *mddev, struct md_rdev *rdev) err = -EBUSY; p->rdev = rdev; goto abort; + } else { + clear_bit(Replacement, &rdev->flags); + clear_bit(WantReplacement, &rdev->flags); } err = md_integrity_register(mddev); } -- cgit v1.2.3 From 8c7a2c2bcfaf0ee29c74437a4814d1aa780d6a26 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:57 +1100 Subject: md/raid1: handle activation of replacement device when recovery completes. When recovery completes ->spare_active is called. This checks if the replacement is ready and if so it fails the original. Signed-off-by: NeilBrown --- drivers/md/raid1.c | 36 +++++++++++++++++++++++++++++++++--- 1 file changed, 33 insertions(+), 3 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c index 3d9aa0c92821..f68075189df8 100644 --- a/drivers/md/raid1.c +++ b/drivers/md/raid1.c @@ -1265,6 +1265,25 @@ static int raid1_spare_active(struct mddev *mddev) */ for (i = 0; i < conf->raid_disks; i++) { struct md_rdev *rdev = conf->mirrors[i].rdev; + struct md_rdev *repl = conf->mirrors[conf->raid_disks + i].rdev; + if (repl + && repl->recovery_offset == MaxSector + && !test_bit(Faulty, &repl->flags) + && !test_and_set_bit(In_sync, &repl->flags)) { + /* replacement has just become active */ + if (!rdev || + !test_and_clear_bit(In_sync, &rdev->flags)) + count++; + if (rdev) { + /* Replaced device not technically + * faulty, but we need to be sure + * it gets removed and never re-added + */ + set_bit(Faulty, &rdev->flags); + sysfs_notify_dirent_safe( + rdev->sysfs_state); + } + } if (rdev && !test_bit(Faulty, &rdev->flags) && !test_and_set_bit(In_sync, &rdev->flags)) { @@ -1362,10 +1381,21 @@ static int raid1_remove_disk(struct mddev *mddev, struct md_rdev *rdev) err = -EBUSY; p->rdev = rdev; goto abort; - } else { - clear_bit(Replacement, &rdev->flags); + } else if (conf->mirrors[conf->raid_disks + number].rdev) { + /* We just removed a device that is being replaced. + * Move down the replacement. We drain all IO before + * doing this to avoid confusion. + */ + struct md_rdev *repl = + conf->mirrors[conf->raid_disks + number].rdev; + raise_barrier(conf); + clear_bit(Replacement, &repl->flags); + p->rdev = repl; + conf->mirrors[conf->raid_disks + number].rdev = NULL; + lower_barrier(conf); + clear_bit(WantReplacement, &rdev->flags); + } else clear_bit(WantReplacement, &rdev->flags); - } err = md_integrity_register(mddev); } abort: -- cgit v1.2.3 From c19d57980b38a5bb613a898937a1cf85f422fb9b Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:57 +1100 Subject: md/raid1: recognise replacements when assembling arrays. If a Replacement is seen, file it as such. If we see two replacements (or two normal devices) for the one slot, abort. Signed-off-by: NeilBrown --- drivers/md/raid1.c | 25 +++++++++++++++++++++++-- 1 file changed, 23 insertions(+), 2 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c index f68075189df8..6df0c41578af 100644 --- a/drivers/md/raid1.c +++ b/drivers/md/raid1.c @@ -2453,14 +2453,20 @@ static struct r1conf *setup_conf(struct mddev *mddev) conf->poolinfo->mddev = mddev; + err = -EINVAL; spin_lock_init(&conf->device_lock); list_for_each_entry(rdev, &mddev->disks, same_set) { int disk_idx = rdev->raid_disk; if (disk_idx >= mddev->raid_disks || disk_idx < 0) continue; - disk = conf->mirrors + disk_idx; + if (test_bit(Replacement, &rdev->flags)) + disk = conf->mirrors + conf->raid_disks + disk_idx; + else + disk = conf->mirrors + disk_idx; + if (disk->rdev) + goto abort; disk->rdev = rdev; disk->head_position = 0; @@ -2476,11 +2482,27 @@ static struct r1conf *setup_conf(struct mddev *mddev) conf->pending_count = 0; conf->recovery_disabled = mddev->recovery_disabled - 1; + err = -EIO; conf->last_used = -1; for (i = 0; i < conf->raid_disks * 2; i++) { disk = conf->mirrors + i; + if (i < conf->raid_disks && + disk[conf->raid_disks].rdev) { + /* This slot has a replacement. */ + if (!disk->rdev) { + /* No original, just make the replacement + * a recovering spare + */ + disk->rdev = + disk[conf->raid_disks].rdev; + disk[conf->raid_disks].rdev = NULL; + } else if (!test_bit(In_sync, &disk->rdev->flags)) + /* Original is not in_sync - bad */ + goto abort; + } + if (!disk->rdev || !test_bit(In_sync, &disk->rdev->flags)) { disk->head_position = 0; @@ -2494,7 +2516,6 @@ static struct r1conf *setup_conf(struct mddev *mddev) conf->last_used = i; } - err = -EIO; if (conf->last_used < 0) { printk(KERN_ERR "md/raid1:%s: no operational mirrors\n", mdname(mddev)); -- cgit v1.2.3 From 7ef449d1ec3668acbba6af6a500d0c84636b436f Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:57 +1100 Subject: md/raid1: If there is a spare and a want_replacement device, start replacement. When attempting to add a spare to a RAID1 array, also consider adding it as a replacement for a want_replacement device. Signed-off-by: NeilBrown --- drivers/md/raid1.c | 17 +++++++++++++++-- 1 file changed, 15 insertions(+), 2 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c index 6df0c41578af..e3cfca7cde16 100644 --- a/drivers/md/raid1.c +++ b/drivers/md/raid1.c @@ -1315,8 +1315,9 @@ static int raid1_add_disk(struct mddev *mddev, struct md_rdev *rdev) if (rdev->raid_disk >= 0) first = last = rdev->raid_disk; - for (mirror = first; mirror <= last; mirror++) - if ( !(p=conf->mirrors+mirror)->rdev) { + for (mirror = first; mirror <= last; mirror++) { + p = conf->mirrors+mirror; + if (!p->rdev) { disk_stack_limits(mddev->gendisk, rdev->bdev, rdev->data_offset << 9); @@ -1343,6 +1344,18 @@ static int raid1_add_disk(struct mddev *mddev, struct md_rdev *rdev) rcu_assign_pointer(p->rdev, rdev); break; } + if (test_bit(WantReplacement, &p->rdev->flags) && + p[conf->raid_disks].rdev == NULL) { + /* Add this device as a replacement */ + clear_bit(In_sync, &rdev->flags); + set_bit(Replacement, &rdev->flags); + rdev->raid_disk = mirror; + err = 0; + conf->fullsync = 1; + rcu_assign_pointer(p[conf->raid_disks].rdev, rdev); + break; + } + } md_integrity_add_rdev(rdev, mddev); print_conf(conf); return err; -- cgit v1.2.3 From 19d671695e1931ebfd75b2b888778201aefe35ca Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Fri, 23 Dec 2011 10:17:57 +1100 Subject: md/raid1: Mark device want_replacement when we see a write error. Now that WantReplacement drives are replaced cleanly, mark a drive as want_replacement when we see a write error. It might get failed soon so the WantReplacement flag is irrelevant, but if the write error is recorded in the bad block log, we still want to activate any spare that might be available. Signed-off-by: NeilBrown --- drivers/md/raid1.c | 16 +++++++++++++++- 1 file changed, 15 insertions(+), 1 deletion(-) (limited to 'drivers/md') diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c index e3cfca7cde16..cc24f0cb7ee3 100644 --- a/drivers/md/raid1.c +++ b/drivers/md/raid1.c @@ -391,6 +391,11 @@ static void raid1_end_write_request(struct bio *bio, int error) if (!uptodate) { set_bit(WriteErrorSeen, &conf->mirrors[mirror].rdev->flags); + if (!test_and_set_bit(WantReplacement, + &conf->mirrors[mirror].rdev->flags)) + set_bit(MD_RECOVERY_NEEDED, & + conf->mddev->recovery); + set_bit(R1BIO_WriteError, &r1_bio->state); } else { /* @@ -1461,6 +1466,10 @@ static void end_sync_write(struct bio *bio, int error) } while (sectors_to_go > 0); set_bit(WriteErrorSeen, &conf->mirrors[mirror].rdev->flags); + if (!test_and_set_bit(WantReplacement, + &conf->mirrors[mirror].rdev->flags)) + set_bit(MD_RECOVERY_NEEDED, & + mddev->recovery); set_bit(R1BIO_WriteError, &r1_bio->state); } else if (is_badblock(conf->mirrors[mirror].rdev, r1_bio->sector, @@ -1491,8 +1500,13 @@ static int r1_sync_page_io(struct md_rdev *rdev, sector_t sector, if (sync_page_io(rdev, sector, sectors << 9, page, rw, false)) /* success */ return 1; - if (rw == WRITE) + if (rw == WRITE) { set_bit(WriteErrorSeen, &rdev->flags); + if (!test_and_set_bit(WantReplacement, + &rdev->flags)) + set_bit(MD_RECOVERY_NEEDED, & + rdev->mddev->recovery); + } /* 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); -- cgit v1.2.3 From f2a371c5e74dd5685ab47effa4ac7b23b1fdaae5 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Mon, 9 Jan 2012 00:46:41 +1100 Subject: md: notify the 'degraded' sysfs attribute on failure. We currently only 'notify' changes to the 'degraded' attribute when it decreases, not when it increases. Notifying on failure is a little awkward as it happen in interrupt context. So instead, notify when we remove the failed device from the array, which is very soon afterwards. Reported-and-tested-by: Mikhail Balabin Signed-off-by: NeilBrown --- drivers/md/md.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'drivers/md') diff --git a/drivers/md/md.c b/drivers/md/md.c index 1c1c562f63dc..33aa06f12b87 100644 --- a/drivers/md/md.c +++ b/drivers/md/md.c @@ -7383,6 +7383,7 @@ static int remove_and_add_spares(struct mddev *mddev) { struct md_rdev *rdev; int spares = 0; + int removed = 0; mddev->curr_resync_completed = 0; @@ -7396,8 +7397,13 @@ static int remove_and_add_spares(struct mddev *mddev) mddev, rdev) == 0) { sysfs_unlink_rdev(mddev, rdev); rdev->raid_disk = -1; + removed++; } } + if (removed) + sysfs_notify(&mddev->kobj, NULL, + "degraded"); + list_for_each_entry(rdev, &mddev->disks, same_set) { if (rdev->raid_disk >= 0 && -- cgit v1.2.3 From 307729c8bc5b5a41361af8af95906eee7552acb1 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Mon, 9 Jan 2012 01:41:51 +1100 Subject: md/raid1: perform bad-block tests for WriteMostly devices too. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit We normally try to avoid reading from write-mostly devices, but when we do we really have to check for bad blocks and be sure not to try reading them. With the current code, best_good_sectors might not get set and that causes zero-length read requests to be send down which is very confusing. This bug was introduced in commit d2eb35acfdccbe2 and so the patch is suitable for 3.1.x and 3.2.x Reported-and-tested-by: Michał Mirosław Reported-and-tested-by: Art -kwaak- van Breemen Signed-off-by: NeilBrown Cc: stable@vger.kernel.org --- drivers/md/raid1.c | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-) (limited to 'drivers/md') diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c index cc24f0cb7ee3..a368db2431a5 100644 --- a/drivers/md/raid1.c +++ b/drivers/md/raid1.c @@ -531,8 +531,17 @@ static int read_balance(struct r1conf *conf, struct r1bio *r1_bio, int *max_sect if (test_bit(WriteMostly, &rdev->flags)) { /* Don't balance among write-mostly, just * use the first as a last resort */ - if (best_disk < 0) + if (best_disk < 0) { + if (is_badblock(rdev, this_sector, sectors, + &first_bad, &bad_sectors)) { + if (first_bad < this_sector) + /* Cannot use this */ + continue; + best_good_sectors = first_bad - this_sector; + } else + best_good_sectors = sectors; best_disk = disk; + } continue; } /* This is a reasonable device to use. It might -- cgit v1.2.3 From b1bd055d397e09f99dcef9b138ed104ff1812fcb Mon Sep 17 00:00:00 2001 From: "Martin K. Petersen" Date: Wed, 11 Jan 2012 16:27:11 +0100 Subject: block: Introduce blk_set_stacking_limits function Stacking driver queue limits are typically bounded exclusively by the capabilities of the low level devices, not by the stacking driver itself. This patch introduces blk_set_stacking_limits() which has more liberal metrics than the default queue limits function. This allows us to inherit topology parameters from bottom devices without manually tweaking the default limits in each driver prior to calling the stacking function. Since there is now a clear distinction between stacking and low-level devices, blk_set_default_limits() has been modified to carry the more conservative values that we used to manually set in blk_queue_make_request(). Signed-off-by: Martin K. Petersen Acked-by: Mike Snitzer Signed-off-by: Jens Axboe --- block/blk-settings.c | 32 ++++++++++++++++++++++++-------- drivers/md/dm-table.c | 6 +++--- drivers/md/md.c | 1 + include/linux/blkdev.h | 1 + 4 files changed, 29 insertions(+), 11 deletions(-) (limited to 'drivers/md') diff --git a/block/blk-settings.c b/block/blk-settings.c index fa1eb0449a05..d3234fc494ad 100644 --- a/block/blk-settings.c +++ b/block/blk-settings.c @@ -104,9 +104,7 @@ EXPORT_SYMBOL_GPL(blk_queue_lld_busy); * @lim: the queue_limits structure to reset * * Description: - * Returns a queue_limit struct to its default state. Can be used by - * stacking drivers like DM that stage table swaps and reuse an - * existing device queue. + * Returns a queue_limit struct to its default state. */ void blk_set_default_limits(struct queue_limits *lim) { @@ -114,13 +112,12 @@ void blk_set_default_limits(struct queue_limits *lim) lim->max_integrity_segments = 0; lim->seg_boundary_mask = BLK_SEG_BOUNDARY_MASK; lim->max_segment_size = BLK_MAX_SEGMENT_SIZE; - lim->max_sectors = BLK_DEF_MAX_SECTORS; - lim->max_hw_sectors = INT_MAX; + lim->max_sectors = lim->max_hw_sectors = BLK_SAFE_MAX_SECTORS; lim->max_discard_sectors = 0; lim->discard_granularity = 0; lim->discard_alignment = 0; lim->discard_misaligned = 0; - lim->discard_zeroes_data = 1; + lim->discard_zeroes_data = 0; lim->logical_block_size = lim->physical_block_size = lim->io_min = 512; lim->bounce_pfn = (unsigned long)(BLK_BOUNCE_ANY >> PAGE_SHIFT); lim->alignment_offset = 0; @@ -130,6 +127,27 @@ void blk_set_default_limits(struct queue_limits *lim) } EXPORT_SYMBOL(blk_set_default_limits); +/** + * blk_set_stacking_limits - set default limits for stacking devices + * @lim: the queue_limits structure to reset + * + * Description: + * Returns a queue_limit struct to its default state. Should be used + * by stacking drivers like DM that have no internal limits. + */ +void blk_set_stacking_limits(struct queue_limits *lim) +{ + blk_set_default_limits(lim); + + /* Inherit limits from component devices */ + lim->discard_zeroes_data = 1; + lim->max_segments = USHRT_MAX; + lim->max_hw_sectors = UINT_MAX; + + lim->max_sectors = BLK_DEF_MAX_SECTORS; +} +EXPORT_SYMBOL(blk_set_stacking_limits); + /** * blk_queue_make_request - define an alternate make_request function for a device * @q: the request queue for the device to be affected @@ -165,8 +183,6 @@ void blk_queue_make_request(struct request_queue *q, make_request_fn *mfn) q->nr_batching = BLK_BATCH_REQ; blk_set_default_limits(&q->limits); - blk_queue_max_hw_sectors(q, BLK_SAFE_MAX_SECTORS); - q->limits.discard_zeroes_data = 0; /* * by default assume old behaviour and bounce for any highmem page diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c index 8e9132130142..63cc54289aff 100644 --- a/drivers/md/dm-table.c +++ b/drivers/md/dm-table.c @@ -699,7 +699,7 @@ static int validate_hardware_logical_block_alignment(struct dm_table *table, while (i < dm_table_get_num_targets(table)) { ti = dm_table_get_target(table, i++); - blk_set_default_limits(&ti_limits); + blk_set_stacking_limits(&ti_limits); /* combine all target devices' limits */ if (ti->type->iterate_devices) @@ -1221,10 +1221,10 @@ int dm_calculate_queue_limits(struct dm_table *table, struct queue_limits ti_limits; unsigned i = 0; - blk_set_default_limits(limits); + blk_set_stacking_limits(limits); while (i < dm_table_get_num_targets(table)) { - blk_set_default_limits(&ti_limits); + blk_set_stacking_limits(&ti_limits); ti = dm_table_get_target(table, i++); diff --git a/drivers/md/md.c b/drivers/md/md.c index ee981737edfc..114ba155af87 100644 --- a/drivers/md/md.c +++ b/drivers/md/md.c @@ -4622,6 +4622,7 @@ static int md_alloc(dev_t dev, char *name) mddev->queue->queuedata = mddev; blk_queue_make_request(mddev->queue, md_make_request); + blk_set_stacking_limits(&mddev->queue->limits); disk = alloc_disk(1 << shift); if (!disk) { diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index 8bca04873f53..adc34133a56a 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -844,6 +844,7 @@ extern void blk_queue_io_min(struct request_queue *q, unsigned int min); extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt); extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt); extern void blk_set_default_limits(struct queue_limits *lim); +extern void blk_set_stacking_limits(struct queue_limits *lim); extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b, sector_t offset); extern int bdev_stack_limits(struct queue_limits *t, struct block_device *bdev, -- cgit v1.2.3 From ec8013beddd717d1740cfefb1a9b900deef85462 Mon Sep 17 00:00:00 2001 From: Paolo Bonzini Date: Thu, 12 Jan 2012 16:01:29 +0100 Subject: dm: do not forward ioctls from logical volumes to the underlying device A logical volume can map to just part of underlying physical volume. In this case, it must be treated like a partition. Based on a patch from Alasdair G Kergon. Cc: Alasdair G Kergon Cc: dm-devel@redhat.com Signed-off-by: Paolo Bonzini Signed-off-by: Linus Torvalds --- drivers/md/dm-flakey.c | 11 ++++++++++- drivers/md/dm-linear.c | 12 +++++++++++- drivers/md/dm-mpath.c | 6 ++++++ 3 files changed, 27 insertions(+), 2 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/dm-flakey.c b/drivers/md/dm-flakey.c index f84c08029b21..9fb18c147825 100644 --- a/drivers/md/dm-flakey.c +++ b/drivers/md/dm-flakey.c @@ -368,8 +368,17 @@ static int flakey_status(struct dm_target *ti, status_type_t type, static int flakey_ioctl(struct dm_target *ti, unsigned int cmd, unsigned long arg) { struct flakey_c *fc = ti->private; + struct dm_dev *dev = fc->dev; + int r = 0; - return __blkdev_driver_ioctl(fc->dev->bdev, fc->dev->mode, cmd, arg); + /* + * Only pass ioctls through if the device sizes match exactly. + */ + if (fc->start || + ti->len != i_size_read(dev->bdev->bd_inode) >> SECTOR_SHIFT) + r = scsi_verify_blk_ioctl(NULL, cmd); + + return r ? : __blkdev_driver_ioctl(dev->bdev, dev->mode, cmd, arg); } static int flakey_merge(struct dm_target *ti, struct bvec_merge_data *bvm, diff --git a/drivers/md/dm-linear.c b/drivers/md/dm-linear.c index 3921e3bb43c1..9728839f844a 100644 --- a/drivers/md/dm-linear.c +++ b/drivers/md/dm-linear.c @@ -116,7 +116,17 @@ static int linear_ioctl(struct dm_target *ti, unsigned int cmd, unsigned long arg) { struct linear_c *lc = (struct linear_c *) ti->private; - return __blkdev_driver_ioctl(lc->dev->bdev, lc->dev->mode, cmd, arg); + struct dm_dev *dev = lc->dev; + int r = 0; + + /* + * Only pass ioctls through if the device sizes match exactly. + */ + if (lc->start || + ti->len != i_size_read(dev->bdev->bd_inode) >> SECTOR_SHIFT) + r = scsi_verify_blk_ioctl(NULL, cmd); + + return r ? : __blkdev_driver_ioctl(dev->bdev, dev->mode, cmd, arg); } static int linear_merge(struct dm_target *ti, struct bvec_merge_data *bvm, diff --git a/drivers/md/dm-mpath.c b/drivers/md/dm-mpath.c index 5e0090ef4182..801d92d237cf 100644 --- a/drivers/md/dm-mpath.c +++ b/drivers/md/dm-mpath.c @@ -1520,6 +1520,12 @@ static int multipath_ioctl(struct dm_target *ti, unsigned int cmd, spin_unlock_irqrestore(&m->lock, flags); + /* + * Only pass ioctls through if the device sizes match exactly. + */ + if (!r && ti->len != i_size_read(bdev->bd_inode) >> SECTOR_SHIFT) + r = scsi_verify_blk_ioctl(NULL, cmd); + return r ? : __blkdev_driver_ioctl(bdev, mode, cmd, arg); } -- cgit v1.2.3