From 2c810cddc44d6f95cef75df3f07fc0850ff92417 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Mon, 21 May 2012 09:27:00 +1000 Subject: md: allow a reshape operation to be reversed. Currently a reshape operation always progresses from the start of the array to the end unless the number of devices is being reduced, in which case it progressed in the opposite direction. To reverse a partial reshape which changes the number of devices you can stop the array and re-assemble with the raid-disks numbers reversed and it will undo. However for a reshape that does not change the number of devices it is not possible to reverse the reshape in the middle - you have to wait until it completes. So add a 'reshape_direction' attribute with is either 'forwards' or 'backwards' and can be explicitly set when delta_disks is zero. This will become more important when we allow the data_offset to change in a reshape. Then the explicit statement of what direction is being used will be more useful. This can be enabled in raid5 trivially as it already supports reverse reshape and just needs to use a different trigger to request it. Signed-off-by: NeilBrown --- drivers/md/raid5.c | 23 ++++++++++++----------- 1 file changed, 12 insertions(+), 11 deletions(-) (limited to 'drivers/md/raid5.c') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index f351422938e..0abbd3447cf 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -3970,13 +3970,13 @@ static void make_request(struct mddev *mddev, struct bio * bi) * to check again. */ spin_lock_irq(&conf->device_lock); - if (mddev->delta_disks < 0 + if (mddev->reshape_backwards ? logical_sector < conf->reshape_progress : logical_sector >= conf->reshape_progress) { disks = conf->previous_raid_disks; previous = 1; } else { - if (mddev->delta_disks < 0 + if (mddev->reshape_backwards ? logical_sector < conf->reshape_safe : logical_sector >= conf->reshape_safe) { spin_unlock_irq(&conf->device_lock); @@ -4009,7 +4009,7 @@ static void make_request(struct mddev *mddev, struct bio * bi) */ int must_retry = 0; spin_lock_irq(&conf->device_lock); - if (mddev->delta_disks < 0 + if (mddev->reshape_backwards ? logical_sector >= conf->reshape_progress : logical_sector < conf->reshape_progress) /* mismatch, need to try again */ @@ -4108,11 +4108,11 @@ static sector_t reshape_request(struct mddev *mddev, sector_t sector_nr, int *sk if (sector_nr == 0) { /* If restarting in the middle, skip the initial sectors */ - if (mddev->delta_disks < 0 && + if (mddev->reshape_backwards && conf->reshape_progress < raid5_size(mddev, 0, 0)) { sector_nr = raid5_size(mddev, 0, 0) - conf->reshape_progress; - } else if (mddev->delta_disks >= 0 && + } else if (!mddev->reshape_backwards && conf->reshape_progress > 0) sector_nr = conf->reshape_progress; sector_div(sector_nr, new_data_disks); @@ -4147,7 +4147,7 @@ static sector_t reshape_request(struct mddev *mddev, sector_t sector_nr, int *sk sector_div(readpos, data_disks); safepos = conf->reshape_safe; sector_div(safepos, data_disks); - if (mddev->delta_disks < 0) { + if (mddev->reshape_backwards) { writepos -= min_t(sector_t, reshape_sectors, writepos); readpos += reshape_sectors; safepos += reshape_sectors; @@ -4174,7 +4174,7 @@ static sector_t reshape_request(struct mddev *mddev, sector_t sector_nr, int *sk * Maybe that number should be configurable, but I'm not sure it is * worth it.... maybe it could be a multiple of safemode_delay??? */ - if ((mddev->delta_disks < 0 + if ((mddev->reshape_backwards ? (safepos > writepos && readpos < writepos) : (safepos < writepos && readpos > writepos)) || time_after(jiffies, conf->reshape_checkpoint + 10*HZ)) { @@ -4195,7 +4195,7 @@ static sector_t reshape_request(struct mddev *mddev, sector_t sector_nr, int *sk sysfs_notify(&mddev->kobj, NULL, "sync_completed"); } - if (mddev->delta_disks < 0) { + if (mddev->reshape_backwards) { BUG_ON(conf->reshape_progress == 0); stripe_addr = writepos; BUG_ON((mddev->dev_sectors & @@ -4239,7 +4239,7 @@ static sector_t reshape_request(struct mddev *mddev, sector_t sector_nr, int *sk list_add(&sh->lru, &stripes); } spin_lock_irq(&conf->device_lock); - if (mddev->delta_disks < 0) + if (mddev->reshape_backwards) conf->reshape_progress -= reshape_sectors * new_data_disks; else conf->reshape_progress += reshape_sectors * new_data_disks; @@ -5008,7 +5008,7 @@ static int run(struct mddev *mddev) mdname(mddev)); return -EINVAL; } - } else if (mddev->delta_disks < 0 + } else if (mddev->reshape_backwards ? (here_new * mddev->new_chunk_sectors <= here_old * mddev->chunk_sectors) : (here_new * mddev->new_chunk_sectors >= @@ -5535,7 +5535,7 @@ static int raid5_start_reshape(struct mddev *mddev) conf->chunk_sectors = mddev->new_chunk_sectors; conf->prev_algo = conf->algorithm; conf->algorithm = mddev->new_layout; - if (mddev->delta_disks < 0) + if (mddev->reshape_backwards) conf->reshape_progress = raid5_size(mddev, 0, 0); else conf->reshape_progress = 0; @@ -5663,6 +5663,7 @@ static void raid5_finish_reshape(struct mddev *mddev) mddev->chunk_sectors = conf->chunk_sectors; mddev->reshape_position = MaxSector; mddev->delta_disks = 0; + mddev->reshape_backwards = 0; } } -- cgit v1.2.3-70-g09d2 From c6563a8c38fde3c1c7fc925a10bde3ca20799301 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Mon, 21 May 2012 09:27:00 +1000 Subject: md: add possibility to change data-offset for devices. When reshaping we can avoid costly intermediate backup by changing the 'start' address of the array on the device (if there is enough room). So as a first step, allow such a change to be requested through sysfs, and recorded in v1.x metadata. (As we didn't previous check that all 'pad' fields were zero, we need a new FEATURE flag for this. A (belatedly) check that all remaining 'pad' fields are zero to avoid a repeat of this) The new data offset must be requested separately for each device. This allows each to have a different change in the data offset. This is not likely to be used often but as data_offset can be set per-device, new_data_offset should be too. This patch also removes the 'acknowledged' arg to rdev_set_badblocks as it is never used and never will be. At the same time we add a new arg ('in_new') which is currently always zero but will be used more soon. When a reshape finishes we will need to update the data_offset and rdev->sectors. So provide an exported function to do that. Signed-off-by: NeilBrown --- drivers/md/md.c | 217 +++++++++++++++++++++++++++++++++++++++++----- drivers/md/md.h | 7 +- drivers/md/raid1.c | 4 +- drivers/md/raid10.c | 8 +- drivers/md/raid5.c | 10 ++- include/linux/raid/md_p.h | 10 ++- 6 files changed, 222 insertions(+), 34 deletions(-) (limited to 'drivers/md/raid5.c') diff --git a/drivers/md/md.c b/drivers/md/md.c index 44bb1d52dd4..9fa98fc74b0 100644 --- a/drivers/md/md.c +++ b/drivers/md/md.c @@ -1035,12 +1035,17 @@ static unsigned int calc_sb_csum(mdp_super_t * sb) struct super_type { char *name; struct module *owner; - int (*load_super)(struct md_rdev *rdev, struct md_rdev *refdev, + int (*load_super)(struct md_rdev *rdev, + struct md_rdev *refdev, int minor_version); - int (*validate_super)(struct mddev *mddev, struct md_rdev *rdev); - void (*sync_super)(struct mddev *mddev, struct md_rdev *rdev); + int (*validate_super)(struct mddev *mddev, + struct md_rdev *rdev); + void (*sync_super)(struct mddev *mddev, + struct md_rdev *rdev); unsigned long long (*rdev_size_change)(struct md_rdev *rdev, sector_t num_sectors); + int (*allow_new_offset)(struct md_rdev *rdev, + unsigned long long new_offset); }; /* @@ -1112,6 +1117,7 @@ static int super_90_load(struct md_rdev *rdev, struct md_rdev *refdev, int minor rdev->preferred_minor = sb->md_minor; rdev->data_offset = 0; + rdev->new_data_offset = 0; rdev->sb_size = MD_SB_BYTES; rdev->badblocks.shift = -1; @@ -1438,6 +1444,12 @@ super_90_rdev_size_change(struct md_rdev *rdev, sector_t num_sectors) return num_sectors; } +static int +super_90_allow_new_offset(struct md_rdev *rdev, unsigned long long new_offset) +{ + /* non-zero offset changes not possible with v0.90 */ + return new_offset == 0; +} /* * version 1 superblock @@ -1473,6 +1485,7 @@ static int super_1_load(struct md_rdev *rdev, struct md_rdev *refdev, int minor_ struct mdp_superblock_1 *sb; int ret; sector_t sb_start; + sector_t sectors; char b[BDEVNAME_SIZE], b2[BDEVNAME_SIZE]; int bmask; @@ -1527,9 +1540,18 @@ static int super_1_load(struct md_rdev *rdev, struct md_rdev *refdev, int minor_ bdevname(rdev->bdev,b)); return -EINVAL; } + if (sb->pad0 || + sb->pad3[0] || + memcmp(sb->pad3, sb->pad3+1, sizeof(sb->pad3) - sizeof(sb->pad3[1]))) + /* Some padding is non-zero, might be a new feature */ + return -EINVAL; rdev->preferred_minor = 0xffff; rdev->data_offset = le64_to_cpu(sb->data_offset); + rdev->new_data_offset = rdev->data_offset; + if ((le32_to_cpu(sb->feature_map) & MD_FEATURE_RESHAPE_ACTIVE) && + (le32_to_cpu(sb->feature_map) & MD_FEATURE_NEW_OFFSET)) + rdev->new_data_offset += (s32)le32_to_cpu(sb->new_offset); atomic_set(&rdev->corrected_errors, le32_to_cpu(sb->cnt_corrected_read)); rdev->sb_size = le32_to_cpu(sb->max_dev) * 2 + 256; @@ -1540,6 +1562,9 @@ static int super_1_load(struct md_rdev *rdev, struct md_rdev *refdev, int minor_ if (minor_version && rdev->data_offset < sb_start + (rdev->sb_size/512)) return -EINVAL; + if (minor_version + && rdev->new_data_offset < sb_start + (rdev->sb_size/512)) + return -EINVAL; if (sb->level == cpu_to_le32(LEVEL_MULTIPATH)) rdev->desc_nr = -1; @@ -1611,16 +1636,14 @@ static int super_1_load(struct md_rdev *rdev, struct md_rdev *refdev, int minor_ else ret = 0; } - if (minor_version) - rdev->sectors = (i_size_read(rdev->bdev->bd_inode) >> 9) - - le64_to_cpu(sb->data_offset); - else - rdev->sectors = rdev->sb_start; - if (rdev->sectors < le64_to_cpu(sb->data_size)) + if (minor_version) { + sectors = (i_size_read(rdev->bdev->bd_inode) >> 9); + sectors -= rdev->data_offset; + } else + sectors = rdev->sb_start; + if (sectors < le64_to_cpu(sb->data_size)) return -EINVAL; rdev->sectors = le64_to_cpu(sb->data_size); - if (le64_to_cpu(sb->size) > rdev->sectors) - return -EINVAL; return ret; } @@ -1745,7 +1768,6 @@ static void super_1_sync(struct mddev *mddev, struct md_rdev *rdev) sb->feature_map = 0; sb->pad0 = 0; sb->recovery_offset = cpu_to_le64(0); - memset(sb->pad1, 0, sizeof(sb->pad1)); memset(sb->pad3, 0, sizeof(sb->pad3)); sb->utime = cpu_to_le64((__u64)mddev->utime); @@ -1767,6 +1789,8 @@ static void super_1_sync(struct mddev *mddev, struct md_rdev *rdev) sb->devflags |= WriteMostly1; else sb->devflags &= ~WriteMostly1; + sb->data_offset = cpu_to_le64(rdev->data_offset); + sb->data_size = cpu_to_le64(rdev->sectors); if (mddev->bitmap && mddev->bitmap_info.file == NULL) { sb->bitmap_offset = cpu_to_le32((__u32)mddev->bitmap_info.offset); @@ -1795,6 +1819,12 @@ static void super_1_sync(struct mddev *mddev, struct md_rdev *rdev) mddev->reshape_backwards) sb->feature_map |= cpu_to_le32(MD_FEATURE_RESHAPE_BACKWARDS); + if (rdev->new_data_offset != rdev->data_offset) { + sb->feature_map + |= cpu_to_le32(MD_FEATURE_NEW_OFFSET); + sb->new_offset = cpu_to_le32((__u32)(rdev->new_data_offset + - rdev->data_offset)); + } } if (rdev->badblocks.count == 0) @@ -1871,6 +1901,8 @@ super_1_rdev_size_change(struct md_rdev *rdev, sector_t num_sectors) sector_t max_sectors; if (num_sectors && num_sectors < rdev->mddev->dev_sectors) return 0; /* component must fit device */ + if (rdev->data_offset != rdev->new_data_offset) + return 0; /* too confusing */ if (rdev->sb_start < rdev->data_offset) { /* minor versions 1 and 2; superblock before data */ max_sectors = i_size_read(rdev->bdev->bd_inode) >> 9; @@ -1898,6 +1930,40 @@ super_1_rdev_size_change(struct md_rdev *rdev, sector_t num_sectors) rdev->sb_page); md_super_wait(rdev->mddev); return num_sectors; + +} + +static int +super_1_allow_new_offset(struct md_rdev *rdev, + unsigned long long new_offset) +{ + /* All necessary checks on new >= old have been done */ + struct bitmap *bitmap; + if (new_offset >= rdev->data_offset) + return 1; + + /* with 1.0 metadata, there is no metadata to tread on + * so we can always move back */ + if (rdev->mddev->minor_version == 0) + return 1; + + /* otherwise we must be sure not to step on + * any metadata, so stay: + * 36K beyond start of superblock + * beyond end of badblocks + * beyond write-intent bitmap + */ + if (rdev->sb_start + (32+4)*2 > new_offset) + return 0; + bitmap = rdev->mddev->bitmap; + if (bitmap && !rdev->mddev->bitmap_info.file && + rdev->sb_start + rdev->mddev->bitmap_info.offset + + bitmap->file_pages * (PAGE_SIZE>>9) > new_offset) + return 0; + if (rdev->badblocks.sector + rdev->badblocks.size > new_offset) + return 0; + + return 1; } static struct super_type super_types[] = { @@ -1908,6 +1974,7 @@ static struct super_type super_types[] = { .validate_super = super_90_validate, .sync_super = super_90_sync, .rdev_size_change = super_90_rdev_size_change, + .allow_new_offset = super_90_allow_new_offset, }, [1] = { .name = "md-1", @@ -1916,6 +1983,7 @@ static struct super_type super_types[] = { .validate_super = super_1_validate, .sync_super = super_1_sync, .rdev_size_change = super_1_rdev_size_change, + .allow_new_offset = super_1_allow_new_offset, }, }; @@ -2823,9 +2891,8 @@ offset_show(struct md_rdev *rdev, char *page) static ssize_t offset_store(struct md_rdev *rdev, const char *buf, size_t len) { - char *e; - unsigned long long offset = simple_strtoull(buf, &e, 10); - if (e==buf || (*e && *e != '\n')) + unsigned long long offset; + if (strict_strtoull(buf, 10, &offset) < 0) return -EINVAL; if (rdev->mddev->pers && rdev->raid_disk >= 0) return -EBUSY; @@ -2840,6 +2907,63 @@ offset_store(struct md_rdev *rdev, const char *buf, size_t len) static struct rdev_sysfs_entry rdev_offset = __ATTR(offset, S_IRUGO|S_IWUSR, offset_show, offset_store); +static ssize_t new_offset_show(struct md_rdev *rdev, char *page) +{ + return sprintf(page, "%llu\n", + (unsigned long long)rdev->new_data_offset); +} + +static ssize_t new_offset_store(struct md_rdev *rdev, + const char *buf, size_t len) +{ + unsigned long long new_offset; + struct mddev *mddev = rdev->mddev; + + if (strict_strtoull(buf, 10, &new_offset) < 0) + return -EINVAL; + + if (mddev->sync_thread) + return -EBUSY; + if (new_offset == rdev->data_offset) + /* reset is always permitted */ + ; + else if (new_offset > rdev->data_offset) { + /* must not push array size beyond rdev_sectors */ + if (new_offset - rdev->data_offset + + mddev->dev_sectors > rdev->sectors) + return -E2BIG; + } + /* Metadata worries about other space details. */ + + /* decreasing the offset is inconsistent with a backwards + * reshape. + */ + if (new_offset < rdev->data_offset && + mddev->reshape_backwards) + return -EINVAL; + /* Increasing offset is inconsistent with forwards + * reshape. reshape_direction should be set to + * 'backwards' first. + */ + if (new_offset > rdev->data_offset && + !mddev->reshape_backwards) + return -EINVAL; + + if (mddev->pers && mddev->persistent && + !super_types[mddev->major_version] + .allow_new_offset(rdev, new_offset)) + return -E2BIG; + rdev->new_data_offset = new_offset; + if (new_offset > rdev->data_offset) + mddev->reshape_backwards = 1; + else if (new_offset < rdev->data_offset) + mddev->reshape_backwards = 0; + + return len; +} +static struct rdev_sysfs_entry rdev_new_offset = +__ATTR(new_offset, S_IRUGO|S_IWUSR, new_offset_show, new_offset_store); + static ssize_t rdev_size_show(struct md_rdev *rdev, char *page) { @@ -2884,6 +3008,8 @@ rdev_size_store(struct md_rdev *rdev, const char *buf, size_t len) if (strict_blocks_to_sectors(buf, §ors) < 0) return -EINVAL; + if (rdev->data_offset != rdev->new_data_offset) + return -EINVAL; /* too confusing */ if (my_mddev->pers && rdev->raid_disk >= 0) { if (my_mddev->persistent) { sectors = super_types[my_mddev->major_version]. @@ -3020,6 +3146,7 @@ static struct attribute *rdev_default_attrs[] = { &rdev_errors.attr, &rdev_slot.attr, &rdev_offset.attr, + &rdev_new_offset.attr, &rdev_size.attr, &rdev_recovery_start.attr, &rdev_bad_blocks.attr, @@ -3094,6 +3221,7 @@ int md_rdev_init(struct md_rdev *rdev) rdev->raid_disk = -1; rdev->flags = 0; rdev->data_offset = 0; + rdev->new_data_offset = 0; rdev->sb_events = 0; rdev->last_read_error.tv_sec = 0; rdev->last_read_error.tv_nsec = 0; @@ -3598,7 +3726,17 @@ raid_disks_store(struct mddev *mddev, const char *buf, size_t len) if (mddev->pers) rv = update_raid_disks(mddev, n); else if (mddev->reshape_position != MaxSector) { + struct md_rdev *rdev; int olddisks = mddev->raid_disks - mddev->delta_disks; + + rdev_for_each(rdev, mddev) { + if (olddisks < n && + rdev->data_offset < rdev->new_data_offset) + return -EINVAL; + if (olddisks > n && + rdev->data_offset > rdev->new_data_offset) + return -EINVAL; + } mddev->delta_disks = n - olddisks; mddev->raid_disks = n; mddev->reshape_backwards = (mddev->delta_disks < 0); @@ -4445,6 +4583,7 @@ reshape_position_show(struct mddev *mddev, char *page) static ssize_t reshape_position_store(struct mddev *mddev, const char *buf, size_t len) { + struct md_rdev *rdev; char *e; unsigned long long new = simple_strtoull(buf, &e, 10); if (mddev->pers) @@ -4457,6 +4596,8 @@ reshape_position_store(struct mddev *mddev, const char *buf, size_t len) mddev->new_level = mddev->level; mddev->new_layout = mddev->layout; mddev->new_chunk_sectors = mddev->chunk_sectors; + rdev_for_each(rdev, mddev) + rdev->new_data_offset = rdev->data_offset; return len; } @@ -6001,6 +6142,7 @@ static int update_size(struct mddev *mddev, sector_t num_sectors) static int update_raid_disks(struct mddev *mddev, int raid_disks) { int rv; + struct md_rdev *rdev; /* change the number of raid disks */ if (mddev->pers->check_reshape == NULL) return -EINVAL; @@ -6009,6 +6151,16 @@ static int update_raid_disks(struct mddev *mddev, int raid_disks) return -EINVAL; if (mddev->sync_thread || mddev->reshape_position != MaxSector) return -EBUSY; + + rdev_for_each(rdev, mddev) { + if (mddev->raid_disks < raid_disks && + rdev->data_offset < rdev->new_data_offset) + return -EINVAL; + if (mddev->raid_disks > raid_disks && + rdev->data_offset > rdev->new_data_offset) + return -EINVAL; + } + mddev->delta_disks = raid_disks - mddev->raid_disks; if (mddev->delta_disks < 0) mddev->reshape_backwards = 1; @@ -7709,6 +7861,20 @@ void md_wait_for_blocked_rdev(struct md_rdev *rdev, struct mddev *mddev) } EXPORT_SYMBOL(md_wait_for_blocked_rdev); +void md_finish_reshape(struct mddev *mddev) +{ + /* called be personality module when reshape completes. */ + struct md_rdev *rdev; + + rdev_for_each(rdev, mddev) { + if (rdev->data_offset > rdev->new_data_offset) + rdev->sectors += rdev->data_offset - rdev->new_data_offset; + else + rdev->sectors -= rdev->new_data_offset - rdev->data_offset; + rdev->data_offset = rdev->new_data_offset; + } +} +EXPORT_SYMBOL(md_finish_reshape); /* Bad block management. * We can record which blocks on each device are 'bad' and so just @@ -7957,10 +8123,15 @@ static int md_set_badblocks(struct badblocks *bb, sector_t s, int sectors, } int rdev_set_badblocks(struct md_rdev *rdev, sector_t s, int sectors, - int acknowledged) + int is_new) { - int rv = md_set_badblocks(&rdev->badblocks, - s + rdev->data_offset, sectors, acknowledged); + int rv; + if (is_new) + s += rdev->new_data_offset; + else + s += rdev->data_offset; + rv = md_set_badblocks(&rdev->badblocks, + s, sectors, 0); if (rv) { /* Make sure they get written out promptly */ sysfs_notify_dirent_safe(rdev->sysfs_state); @@ -8066,11 +8237,15 @@ out: return rv; } -int rdev_clear_badblocks(struct md_rdev *rdev, sector_t s, int sectors) +int rdev_clear_badblocks(struct md_rdev *rdev, sector_t s, int sectors, + int is_new) { + if (is_new) + s += rdev->new_data_offset; + else + s += rdev->data_offset; return md_clear_badblocks(&rdev->badblocks, - s + rdev->data_offset, - sectors); + s, sectors); } EXPORT_SYMBOL_GPL(rdev_clear_badblocks); diff --git a/drivers/md/md.h b/drivers/md/md.h index d51c0ca3777..98913e8dac1 100644 --- a/drivers/md/md.h +++ b/drivers/md/md.h @@ -55,6 +55,7 @@ struct md_rdev { int sb_loaded; __u64 sb_events; sector_t data_offset; /* start of data in array */ + sector_t new_data_offset;/* only relevant while reshaping */ sector_t sb_start; /* offset of the super block (in 512byte sectors) */ int sb_size; /* bytes in the superblock */ int preferred_minor; /* autorun support */ @@ -193,8 +194,9 @@ static inline int is_badblock(struct md_rdev *rdev, sector_t s, int sectors, return 0; } extern int rdev_set_badblocks(struct md_rdev *rdev, sector_t s, int sectors, - int acknowledged); -extern int rdev_clear_badblocks(struct md_rdev *rdev, sector_t s, int sectors); + int is_new); +extern int rdev_clear_badblocks(struct md_rdev *rdev, sector_t s, int sectors, + int is_new); extern void md_ack_all_badblocks(struct badblocks *bb); struct mddev { @@ -592,6 +594,7 @@ extern void md_write_start(struct mddev *mddev, struct bio *bi); extern void md_write_end(struct mddev *mddev); extern void md_done_sync(struct mddev *mddev, int blocks, int ok); extern void md_error(struct mddev *mddev, struct md_rdev *rdev); +extern void md_finish_reshape(struct mddev *mddev); extern int mddev_congested(struct mddev *mddev, int bits); extern void md_flush_request(struct mddev *mddev, struct bio *bio); diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c index 15dd59b84e9..71a7dc038a8 100644 --- a/drivers/md/raid1.c +++ b/drivers/md/raid1.c @@ -2024,7 +2024,7 @@ static void handle_sync_write_finished(struct r1conf *conf, struct r1bio *r1_bio continue; if (test_bit(BIO_UPTODATE, &bio->bi_flags) && test_bit(R1BIO_MadeGood, &r1_bio->state)) { - rdev_clear_badblocks(rdev, r1_bio->sector, s); + rdev_clear_badblocks(rdev, r1_bio->sector, s, 0); } if (!test_bit(BIO_UPTODATE, &bio->bi_flags) && test_bit(R1BIO_WriteError, &r1_bio->state)) { @@ -2044,7 +2044,7 @@ static void handle_write_finished(struct r1conf *conf, struct r1bio *r1_bio) struct md_rdev *rdev = conf->mirrors[m].rdev; rdev_clear_badblocks(rdev, r1_bio->sector, - r1_bio->sectors); + r1_bio->sectors, 0); rdev_dec_pending(rdev, conf->mddev); } else if (r1_bio->bios[m] != NULL) { /* This drive got a write error. We need to diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index 3f91c2e1dfe..832fb4d5665 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c @@ -2480,7 +2480,7 @@ static void handle_write_completed(struct r10conf *conf, struct r10bio *r10_bio) rdev_clear_badblocks( rdev, r10_bio->devs[m].addr, - r10_bio->sectors); + r10_bio->sectors, 0); } else { if (!rdev_set_badblocks( rdev, @@ -2496,7 +2496,7 @@ static void handle_write_completed(struct r10conf *conf, struct r10bio *r10_bio) rdev_clear_badblocks( rdev, r10_bio->devs[m].addr, - r10_bio->sectors); + r10_bio->sectors, 0); } else { if (!rdev_set_badblocks( rdev, @@ -2515,7 +2515,7 @@ static void handle_write_completed(struct r10conf *conf, struct r10bio *r10_bio) rdev_clear_badblocks( rdev, r10_bio->devs[m].addr, - r10_bio->sectors); + r10_bio->sectors, 0); rdev_dec_pending(rdev, conf->mddev); } else if (bio != NULL && !test_bit(BIO_UPTODATE, &bio->bi_flags)) { @@ -2532,7 +2532,7 @@ static void handle_write_completed(struct r10conf *conf, struct r10bio *r10_bio) rdev_clear_badblocks( rdev, r10_bio->devs[m].addr, - r10_bio->sectors); + r10_bio->sectors, 0); rdev_dec_pending(rdev, conf->mddev); } } diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 0abbd3447cf..3705585d756 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -3561,7 +3561,7 @@ finish: if (test_and_clear_bit(R5_MadeGood, &dev->flags)) { rdev = conf->disks[i].rdev; rdev_clear_badblocks(rdev, sh->sector, - STRIPE_SECTORS); + STRIPE_SECTORS, 0); rdev_dec_pending(rdev, conf->mddev); } if (test_and_clear_bit(R5_MadeGoodRepl, &dev->flags)) { @@ -3570,7 +3570,7 @@ finish: /* rdev have been moved down */ rdev = conf->disks[i].rdev; rdev_clear_badblocks(rdev, sh->sector, - STRIPE_SECTORS); + STRIPE_SECTORS, 0); rdev_dec_pending(rdev, conf->mddev); } } @@ -5505,10 +5505,14 @@ static int raid5_start_reshape(struct mddev *mddev) if (!check_stripe_cache(mddev)) return -ENOSPC; - rdev_for_each(rdev, mddev) + rdev_for_each(rdev, mddev) { + /* Don't support changing data_offset yet */ + if (rdev->new_data_offset != rdev->data_offset) + return -EINVAL; if (!test_bit(In_sync, &rdev->flags) && !test_bit(Faulty, &rdev->flags)) spares++; + } if (spares - mddev->degraded < mddev->delta_disks - conf->max_degraded) /* Not enough devices even to make a degraded array diff --git a/include/linux/raid/md_p.h b/include/linux/raid/md_p.h index 07e05f92d05..ee753536ab7 100644 --- a/include/linux/raid/md_p.h +++ b/include/linux/raid/md_p.h @@ -233,7 +233,10 @@ struct mdp_superblock_1 { __le32 delta_disks; /* change in number of raid_disks */ __le32 new_layout; /* new layout */ __le32 new_chunk; /* new chunk size (512byte sectors) */ - __u8 pad1[128-124]; /* set to 0 when written */ + __le32 new_offset; /* signed number to add to data_offset in new + * layout. 0 == no-change. This can be + * different on each device in the array. + */ /* constant this-device information - 64 bytes */ __le64 data_offset; /* sector start of data, often 0 */ @@ -285,11 +288,14 @@ struct mdp_superblock_1 { * of devices, but is going * backwards anyway. */ +#define MD_FEATURE_NEW_OFFSET 64 /* new_offset must be honoured */ #define MD_FEATURE_ALL (MD_FEATURE_BITMAP_OFFSET \ |MD_FEATURE_RECOVERY_OFFSET \ |MD_FEATURE_RESHAPE_ACTIVE \ |MD_FEATURE_BAD_BLOCKS \ |MD_FEATURE_REPLACEMENT \ - |MD_FEATURE_RESHAPE_BACKWARDS) + |MD_FEATURE_RESHAPE_BACKWARDS \ + |MD_FEATURE_NEW_OFFSET \ + ) #endif -- cgit v1.2.3-70-g09d2 From 05616be5e11f66888b66554957dbecdd90658a84 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Mon, 21 May 2012 09:27:00 +1000 Subject: md/raid5: Use correct data_offset for all IO. As there can now be two different data_offsets - an 'old' and a 'new' - we need to carefully choose between them. Signed-off-by: NeilBrown --- drivers/md/raid5.c | 72 ++++++++++++++++++++++++++++++++++++++++++++---------- 1 file changed, 59 insertions(+), 13 deletions(-) (limited to 'drivers/md/raid5.c') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 3705585d756..71d1de909ba 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -488,6 +488,27 @@ get_active_stripe(struct r5conf *conf, sector_t sector, return sh; } +/* Determine if 'data_offset' or 'new_data_offset' should be used + * in this stripe_head. + */ +static int use_new_offset(struct r5conf *conf, struct stripe_head *sh) +{ + sector_t progress = conf->reshape_progress; + /* Need a memory barrier to make sure we see the value + * of conf->generation, or ->data_offset that was set before + * reshape_progress was updated. + */ + smp_rmb(); + if (progress == MaxSector) + return 0; + if (sh->generation == conf->generation - 1) + return 0; + /* We are in a reshape, and this is a new-generation stripe, + * so use new_data_offset. + */ + return 1; +} + static void raid5_end_read_request(struct bio *bi, int error); static void @@ -603,7 +624,12 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s) __func__, (unsigned long long)sh->sector, bi->bi_rw, i); atomic_inc(&sh->count); - bi->bi_sector = sh->sector + rdev->data_offset; + if (use_new_offset(conf, sh)) + bi->bi_sector = (sh->sector + + rdev->new_data_offset); + else + bi->bi_sector = (sh->sector + + rdev->data_offset); bi->bi_flags = 1 << BIO_UPTODATE; bi->bi_idx = 0; bi->bi_io_vec[0].bv_len = STRIPE_SIZE; @@ -627,7 +653,12 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s) __func__, (unsigned long long)sh->sector, rbi->bi_rw, i); atomic_inc(&sh->count); - rbi->bi_sector = sh->sector + rrdev->data_offset; + if (use_new_offset(conf, sh)) + rbi->bi_sector = (sh->sector + + rrdev->new_data_offset); + else + 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; @@ -1648,7 +1679,7 @@ static void raid5_end_read_request(struct bio * bi, int error) int uptodate = test_bit(BIO_UPTODATE, &bi->bi_flags); char b[BDEVNAME_SIZE]; struct md_rdev *rdev = NULL; - + sector_t s; for (i=0 ; idev[i].req) @@ -1671,6 +1702,10 @@ static void raid5_end_read_request(struct bio * bi, int error) if (!rdev) rdev = conf->disks[i].rdev; + if (use_new_offset(conf, sh)) + s = sh->sector + rdev->new_data_offset; + else + s = sh->sector + rdev->data_offset; if (uptodate) { set_bit(R5_UPTODATE, &sh->dev[i].flags); if (test_bit(R5_ReadError, &sh->dev[i].flags)) { @@ -1683,8 +1718,7 @@ static void raid5_end_read_request(struct bio * bi, int error) "md/raid:%s: read error corrected" " (%lu sectors at %llu on %s)\n", mdname(conf->mddev), STRIPE_SECTORS, - (unsigned long long)(sh->sector - + rdev->data_offset), + (unsigned long long)s, bdevname(rdev->bdev, b)); atomic_add(STRIPE_SECTORS, &rdev->corrected_errors); clear_bit(R5_ReadError, &sh->dev[i].flags); @@ -1704,8 +1738,7 @@ static void raid5_end_read_request(struct bio * bi, int error) "md/raid:%s: read error on replacement device " "(sector %llu on %s).\n", mdname(conf->mddev), - (unsigned long long)(sh->sector - + rdev->data_offset), + (unsigned long long)s, bdn); else if (conf->mddev->degraded >= conf->max_degraded) printk_ratelimited( @@ -1713,8 +1746,7 @@ static void raid5_end_read_request(struct bio * bi, int error) "md/raid:%s: read error not correctable " "(sector %llu on %s).\n", mdname(conf->mddev), - (unsigned long long)(sh->sector - + rdev->data_offset), + (unsigned long long)s, bdn); else if (test_bit(R5_ReWrite, &sh->dev[i].flags)) /* Oh, no!!! */ @@ -1723,8 +1755,7 @@ static void raid5_end_read_request(struct bio * bi, int error) "md/raid:%s: read error NOT corrected!! " "(sector %llu on %s).\n", mdname(conf->mddev), - (unsigned long long)(sh->sector - + rdev->data_offset), + (unsigned long long)s, bdn); else if (atomic_read(&rdev->read_errors) > conf->max_nr_stripes) @@ -3842,6 +3873,7 @@ static int chunk_aligned_read(struct mddev *mddev, struct bio * raid_bio) raid_bio->bi_next = (void*)rdev; align_bi->bi_bdev = rdev->bdev; align_bi->bi_flags &= ~(1 << BIO_SEG_VALID); + /* No reshape active, so we can trust rdev->data_offset */ align_bi->bi_sector += rdev->data_offset; if (!bio_fits_rdev(align_bi) || @@ -5182,9 +5214,12 @@ static int run(struct mddev *mddev) blk_queue_io_opt(mddev->queue, chunk_size * (conf->raid_disks - conf->max_degraded)); - rdev_for_each(rdev, mddev) + rdev_for_each(rdev, mddev) { disk_stack_limits(mddev->gendisk, rdev->bdev, rdev->data_offset << 9); + disk_stack_limits(mddev->gendisk, rdev->bdev, + rdev->new_data_offset << 9); + } } return 0; @@ -5539,12 +5574,16 @@ static int raid5_start_reshape(struct mddev *mddev) conf->chunk_sectors = mddev->new_chunk_sectors; conf->prev_algo = conf->algorithm; conf->algorithm = mddev->new_layout; + conf->generation++; + /* Code that selects data_offset needs to see the generation update + * if reshape_progress has been set - so a memory barrier needed. + */ + smp_mb(); if (mddev->reshape_backwards) conf->reshape_progress = raid5_size(mddev, 0, 0); else conf->reshape_progress = 0; conf->reshape_safe = conf->reshape_progress; - conf->generation++; spin_unlock_irq(&conf->device_lock); /* Add some new drives, as many as will fit. @@ -5596,6 +5635,9 @@ static int raid5_start_reshape(struct mddev *mddev) mddev->recovery = 0; spin_lock_irq(&conf->device_lock); mddev->raid_disks = conf->raid_disks = conf->previous_raid_disks; + rdev_for_each(rdev, mddev) + rdev->new_data_offset = rdev->data_offset; + smp_wmb(); conf->reshape_progress = MaxSector; mddev->reshape_position = MaxSector; spin_unlock_irq(&conf->device_lock); @@ -5614,9 +5656,13 @@ static void end_reshape(struct r5conf *conf) { if (!test_bit(MD_RECOVERY_INTR, &conf->mddev->recovery)) { + struct md_rdev *rdev; spin_lock_irq(&conf->device_lock); conf->previous_raid_disks = conf->raid_disks; + rdev_for_each(rdev, conf->mddev) + rdev->data_offset = rdev->new_data_offset; + smp_wmb(); conf->reshape_progress = MaxSector; spin_unlock_irq(&conf->device_lock); wake_up(&conf->wait_for_overlap); -- cgit v1.2.3-70-g09d2 From b5254dd5fdd9abcacadb5101beb35df9ae8cc564 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Mon, 21 May 2012 09:27:01 +1000 Subject: md/raid5: allow for change in data_offset while managing a reshape. The important issue here is incorporating the different in data_offset into calculations concerning when we might need to over-write data that is still thought to be valid. To this end we find the minimum offset difference across all devices and add that where appropriate. Signed-off-by: NeilBrown --- drivers/md/raid5.c | 109 +++++++++++++++++++++++++++++++++++++---------------- drivers/md/raid5.h | 6 +++ 2 files changed, 82 insertions(+), 33 deletions(-) (limited to 'drivers/md/raid5.c') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 71d1de909ba..0172bdd37b4 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -4165,13 +4165,11 @@ static sector_t reshape_request(struct mddev *mddev, sector_t sector_nr, int *sk else reshape_sectors = mddev->chunk_sectors; - /* we update the metadata when there is more than 3Meg - * in the block range (that is rather arbitrary, should - * probably be time based) or when the data about to be - * copied would over-write the source of the data at - * the front of the range. - * i.e. one new_stripe along from reshape_progress new_maps - * to after where reshape_safe old_maps to + /* We update the metadata at least every 10 seconds, or when + * the data about to be copied would over-write the source of + * the data at the front of the range. i.e. one new_stripe + * along from reshape_progress new_maps to after where + * reshape_safe old_maps to */ writepos = conf->reshape_progress; sector_div(writepos, new_data_disks); @@ -4189,11 +4187,29 @@ static sector_t reshape_request(struct mddev *mddev, sector_t sector_nr, int *sk safepos -= min_t(sector_t, reshape_sectors, safepos); } + /* Having calculated the 'writepos' possibly use it + * to set 'stripe_addr' which is where we will write to. + */ + if (mddev->reshape_backwards) { + BUG_ON(conf->reshape_progress == 0); + stripe_addr = writepos; + BUG_ON((mddev->dev_sectors & + ~((sector_t)reshape_sectors - 1)) + - reshape_sectors - stripe_addr + != sector_nr); + } else { + BUG_ON(writepos != sector_nr + reshape_sectors); + stripe_addr = sector_nr; + } + /* 'writepos' is the most advanced device address we might write. * 'readpos' is the least advanced device address we might read. * 'safepos' is the least address recorded in the metadata as having * been reshaped. - * If 'readpos' is behind 'writepos', then there is no way that we can + * If there is a min_offset_diff, these are adjusted either by + * increasing the safepos/readpos if diff is negative, or + * increasing writepos if diff is positive. + * If 'readpos' is then behind 'writepos', there is no way that we can * ensure safety in the face of a crash - that must be done by userspace * making a backup of the data. So in that case there is no particular * rush to update metadata. @@ -4206,6 +4222,12 @@ static sector_t reshape_request(struct mddev *mddev, sector_t sector_nr, int *sk * Maybe that number should be configurable, but I'm not sure it is * worth it.... maybe it could be a multiple of safemode_delay??? */ + if (conf->min_offset_diff < 0) { + safepos += -conf->min_offset_diff; + readpos += -conf->min_offset_diff; + } else + writepos += conf->min_offset_diff; + if ((mddev->reshape_backwards ? (safepos > writepos && readpos < writepos) : (safepos < writepos && readpos > writepos)) || @@ -4227,17 +4249,6 @@ static sector_t reshape_request(struct mddev *mddev, sector_t sector_nr, int *sk sysfs_notify(&mddev->kobj, NULL, "sync_completed"); } - if (mddev->reshape_backwards) { - BUG_ON(conf->reshape_progress == 0); - stripe_addr = writepos; - BUG_ON((mddev->dev_sectors & - ~((sector_t)reshape_sectors - 1)) - - reshape_sectors - stripe_addr - != sector_nr); - } else { - BUG_ON(writepos != sector_nr + reshape_sectors); - stripe_addr = sector_nr; - } INIT_LIST_HEAD(&stripes); for (i = 0; i < reshape_sectors; i += STRIPE_SECTORS) { int j; @@ -4984,16 +4995,42 @@ static int run(struct mddev *mddev) struct md_rdev *rdev; sector_t reshape_offset = 0; int i; + long long min_offset_diff = 0; + int first = 1; if (mddev->recovery_cp != MaxSector) printk(KERN_NOTICE "md/raid:%s: not clean" " -- starting background reconstruction\n", mdname(mddev)); + + rdev_for_each(rdev, mddev) { + long long diff; + if (rdev->raid_disk < 0) + continue; + diff = (rdev->new_data_offset - rdev->data_offset); + if (first) { + min_offset_diff = diff; + first = 0; + } else if (mddev->reshape_backwards && + diff < min_offset_diff) + min_offset_diff = diff; + else if (!mddev->reshape_backwards && + diff > min_offset_diff) + min_offset_diff = diff; + } + if (mddev->reshape_position != MaxSector) { /* Check that we can continue the reshape. - * Currently only disks can change, it must - * increase, and we must be past the point where - * a stripe over-writes itself + * Difficulties arise if the stripe we would write to + * next is at or after the stripe we would read from next. + * For a reshape that changes the number of devices, this + * is only possible for a very short time, and mdadm makes + * sure that time appears to have past before assembling + * the array. So we fail if that time hasn't passed. + * For a reshape that keeps the number of devices the same + * mdadm must be monitoring the reshape can keeping the + * critical areas read-only and backed up. It will start + * the array in read-only mode, so we check for that. */ sector_t here_new, here_old; int old_disks; @@ -5025,26 +5062,34 @@ static int run(struct mddev *mddev) /* here_old is the first stripe that we might need to read * from */ if (mddev->delta_disks == 0) { + if ((here_new * mddev->new_chunk_sectors != + here_old * mddev->chunk_sectors)) { + printk(KERN_ERR "md/raid:%s: reshape position is" + " confused - aborting\n", mdname(mddev)); + return -EINVAL; + } /* We cannot be sure it is safe to start an in-place - * reshape. It is only safe if user-space if monitoring + * reshape. It is only safe if user-space is monitoring * and taking constant backups. * mdadm always starts a situation like this in * readonly mode so it can take control before * allowing any writes. So just check for that. */ - if ((here_new * mddev->new_chunk_sectors != - here_old * mddev->chunk_sectors) || - mddev->ro == 0) { - printk(KERN_ERR "md/raid:%s: in-place reshape must be started" - " in read-only mode - aborting\n", + if (abs(min_offset_diff) >= mddev->chunk_sectors && + abs(min_offset_diff) >= mddev->new_chunk_sectors) + /* not really in-place - so OK */; + else if (mddev->ro == 0) { + printk(KERN_ERR "md/raid:%s: in-place reshape " + "must be started in read-only mode " + "- aborting\n", mdname(mddev)); return -EINVAL; } } else if (mddev->reshape_backwards - ? (here_new * mddev->new_chunk_sectors <= + ? (here_new * mddev->new_chunk_sectors + min_offset_diff <= here_old * mddev->chunk_sectors) : (here_new * mddev->new_chunk_sectors >= - here_old * mddev->chunk_sectors)) { + here_old * mddev->chunk_sectors + (-min_offset_diff))) { /* Reading from the same stripe as writing to - bad */ printk(KERN_ERR "md/raid:%s: reshape_position too early for " "auto-recovery - aborting.\n", @@ -5069,6 +5114,7 @@ static int run(struct mddev *mddev) if (IS_ERR(conf)) return PTR_ERR(conf); + conf->min_offset_diff = min_offset_diff; mddev->thread = conf->thread; conf->thread = NULL; mddev->private = conf; @@ -5541,9 +5587,6 @@ static int raid5_start_reshape(struct mddev *mddev) return -ENOSPC; rdev_for_each(rdev, mddev) { - /* Don't support changing data_offset yet */ - if (rdev->new_data_offset != rdev->data_offset) - return -EINVAL; if (!test_bit(In_sync, &rdev->flags) && !test_bit(Faulty, &rdev->flags)) spares++; diff --git a/drivers/md/raid5.h b/drivers/md/raid5.h index 8d8e13934a4..c6bdfa01d98 100644 --- a/drivers/md/raid5.h +++ b/drivers/md/raid5.h @@ -385,6 +385,12 @@ struct r5conf { short generation; /* increments with every reshape */ unsigned long reshape_checkpoint; /* Time we last updated * metadata */ + long long min_offset_diff; /* minimum difference between + * data_offset and + * new_data_offset across all + * devices. May be negative, + * but is closest to zero. + */ struct list_head handle_list; /* stripes needing handling */ struct list_head hold_list; /* preread ready stripes */ -- cgit v1.2.3-70-g09d2 From cceeca43b5ad96766098144a3fd757e03de9f6f8 Mon Sep 17 00:00:00 2001 From: Shaohua Li Date: Tue, 22 May 2012 13:55:04 +1000 Subject: raid5: remove unused variables The two variables are useless. Signed-off-by: Shaohua Li Signed-off-by: NeilBrown --- drivers/md/raid5.c | 4 ---- 1 file changed, 4 deletions(-) (limited to 'drivers/md/raid5.c') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 0172bdd37b4..962c8f40908 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -3985,12 +3985,10 @@ static void make_request(struct mddev *mddev, struct bio * bi) plugged = mddev_check_plugged(mddev); for (;logical_sector < last_sector; logical_sector += STRIPE_SECTORS) { DEFINE_WAIT(w); - int disks, data_disks; int previous; retry: previous = 0; - disks = conf->raid_disks; prepare_to_wait(&conf->wait_for_overlap, &w, TASK_UNINTERRUPTIBLE); if (unlikely(conf->reshape_progress != MaxSector)) { /* spinlock is needed as reshape_progress may be @@ -4005,7 +4003,6 @@ static void make_request(struct mddev *mddev, struct bio * bi) if (mddev->reshape_backwards ? logical_sector < conf->reshape_progress : logical_sector >= conf->reshape_progress) { - disks = conf->previous_raid_disks; previous = 1; } else { if (mddev->reshape_backwards @@ -4018,7 +4015,6 @@ static void make_request(struct mddev *mddev, struct bio * bi) } spin_unlock_irq(&conf->device_lock); } - data_disks = disks - conf->max_degraded; new_sector = raid5_compute_sector(conf, logical_sector, previous, -- cgit v1.2.3-70-g09d2 From bc0934f0477d0a2350a478004799d9c064923b7b Mon Sep 17 00:00:00 2001 From: Shaohua Li Date: Tue, 22 May 2012 13:55:05 +1000 Subject: raid5: support sync request REQ_SYNC is ignored in current raid5 code. Block layer does use it to do policy, for example ioscheduler. This patch adds it. Signed-off-by: Shaohua Li Signed-off-by: NeilBrown --- drivers/md/raid5.c | 12 ++++++++++-- drivers/md/raid5.h | 1 + 2 files changed, 11 insertions(+), 2 deletions(-) (limited to 'drivers/md/raid5.c') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 962c8f40908..7bfd59b313d 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -539,6 +539,8 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s) replace_only = 1; } else continue; + if (test_and_clear_bit(R5_SyncIO, &sh->dev[i].flags)) + rw |= REQ_SYNC; bi = &sh->dev[i].req; rbi = &sh->dev[i].rreq; /* For writing to replacement */ @@ -1145,6 +1147,8 @@ ops_run_biodrain(struct stripe_head *sh, struct dma_async_tx_descriptor *tx) dev->sector + STRIPE_SECTORS) { if (wbi->bi_rw & REQ_FUA) set_bit(R5_WantFUA, &dev->flags); + if (wbi->bi_rw & REQ_SYNC) + set_bit(R5_SyncIO, &dev->flags); tx = async_copy_data(1, wbi, dev->page, dev->sector, tx); wbi = r5_next_bio(wbi, dev->sector); @@ -1162,13 +1166,15 @@ static void ops_complete_reconstruct(void *stripe_head_ref) int pd_idx = sh->pd_idx; int qd_idx = sh->qd_idx; int i; - bool fua = false; + bool fua = false, sync = false; pr_debug("%s: stripe %llu\n", __func__, (unsigned long long)sh->sector); - for (i = disks; i--; ) + for (i = disks; i--; ) { fua |= test_bit(R5_WantFUA, &sh->dev[i].flags); + sync |= test_bit(R5_SyncIO, &sh->dev[i].flags); + } for (i = disks; i--; ) { struct r5dev *dev = &sh->dev[i]; @@ -1177,6 +1183,8 @@ static void ops_complete_reconstruct(void *stripe_head_ref) set_bit(R5_UPTODATE, &dev->flags); if (fua) set_bit(R5_WantFUA, &dev->flags); + if (sync) + set_bit(R5_SyncIO, &dev->flags); } } diff --git a/drivers/md/raid5.h b/drivers/md/raid5.h index c6bdfa01d98..2164021f3b5 100644 --- a/drivers/md/raid5.h +++ b/drivers/md/raid5.h @@ -285,6 +285,7 @@ enum r5dev_flags { */ R5_Wantdrain, /* dev->towrite needs to be drained */ R5_WantFUA, /* Write should be FUA */ + R5_SyncIO, /* The IO is sync */ 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 */ -- cgit v1.2.3-70-g09d2 From a4a6125a074e1b08ee8ae34f700c5bca19eb9d18 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Tue, 22 May 2012 13:55:27 +1000 Subject: md: allow array to be resized while bitmap is present. Now that bitmaps can be resized, we can allow an array to be resized while the bitmap is present. This only covers resizing that involves changing the effective size of member devices, not resizing that changes the number of devices. Signed-off-by: NeilBrown --- drivers/md/md.c | 6 +----- drivers/md/raid1.c | 11 +++++++++-- drivers/md/raid10.c | 10 ++++++++-- drivers/md/raid5.c | 14 ++++++++++---- 4 files changed, 28 insertions(+), 13 deletions(-) (limited to 'drivers/md/raid5.c') diff --git a/drivers/md/md.c b/drivers/md/md.c index 9e2336fbbd3..86adf4ac46c 100644 --- a/drivers/md/md.c +++ b/drivers/md/md.c @@ -6153,11 +6153,7 @@ static int update_size(struct mddev *mddev, sector_t num_sectors) */ if (mddev->sync_thread) return -EBUSY; - if (mddev->bitmap) - /* Sorry, cannot grow a bitmap yet, just remove it, - * grow, and re-add. - */ - return -EBUSY; + rdev_for_each(rdev, mddev) { sector_t avail = rdev->sectors; diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c index 22cfc6660b1..8e717bd518e 100644 --- a/drivers/md/raid1.c +++ b/drivers/md/raid1.c @@ -2752,9 +2752,16 @@ static int raid1_resize(struct mddev *mddev, sector_t sectors) * any io in the removed space completes, but it hardly seems * worth it. */ - md_set_array_sectors(mddev, raid1_size(mddev, sectors, 0)); - if (mddev->array_sectors > raid1_size(mddev, sectors, 0)) + sector_t newsize = raid1_size(mddev, sectors, 0); + if (mddev->external_size && + mddev->array_sectors > newsize) return -EINVAL; + if (mddev->bitmap) { + int ret = bitmap_resize(mddev->bitmap, newsize, 0, 0); + if (ret) + return ret; + } + md_set_array_sectors(mddev, newsize); set_capacity(mddev->gendisk, mddev->array_sectors); revalidate_disk(mddev->gendisk); if (sectors > mddev->dev_sectors && diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index fb9062b5022..8fe3aa46998 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c @@ -3678,9 +3678,15 @@ static int raid10_resize(struct mddev *mddev, sector_t sectors) oldsize = raid10_size(mddev, 0, 0); size = raid10_size(mddev, sectors, 0); - md_set_array_sectors(mddev, size); - if (mddev->array_sectors > size) + if (mddev->external_size && + mddev->array_sectors > size) return -EINVAL; + if (mddev->bitmap) { + int ret = bitmap_resize(mddev->bitmap, size, 0, 0); + if (ret) + return ret; + } + md_set_array_sectors(mddev, size); set_capacity(mddev->gendisk, mddev->array_sectors); revalidate_disk(mddev->gendisk); if (sectors > mddev->dev_sectors && diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 7bfd59b313d..eab6168bb7f 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -5503,12 +5503,18 @@ static int raid5_resize(struct mddev *mddev, sector_t sectors) * any io in the removed space completes, but it hardly seems * worth it. */ + sector_t newsize; sectors &= ~((sector_t)mddev->chunk_sectors - 1); - md_set_array_sectors(mddev, raid5_size(mddev, sectors, - mddev->raid_disks)); - if (mddev->array_sectors > - raid5_size(mddev, sectors, mddev->raid_disks)) + newsize = raid5_size(mddev, sectors, mddev->raid_disks); + if (mddev->external_size && + mddev->array_sectors > newsize) return -EINVAL; + if (mddev->bitmap) { + int ret = bitmap_resize(mddev->bitmap, sectors, 0, 0); + if (ret) + return ret; + } + md_set_array_sectors(mddev, newsize); set_capacity(mddev->gendisk, mddev->array_sectors); revalidate_disk(mddev->gendisk); if (sectors > mddev->dev_sectors && -- cgit v1.2.3-70-g09d2 From 30b67645faadcdc32072f999796b99bd0a20f2d5 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Tue, 22 May 2012 13:55:28 +1000 Subject: md/raid5: Allow reshape while a bitmap is present. We always should have allowed this. A raid5 reshape doesn't change the size of the bitmap, so not need to restrict it. Also add a test to make sure we don't try to start a reshape on a failed array. Signed-off-by: NeilBrown --- drivers/md/raid5.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'drivers/md/raid5.c') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index eab6168bb7f..8fba57e8f78 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -5559,9 +5559,6 @@ static int check_reshape(struct mddev *mddev) mddev->new_layout == mddev->layout && mddev->new_chunk_sectors == mddev->chunk_sectors) return 0; /* nothing to do */ - if (mddev->bitmap) - /* Cannot grow a bitmap yet */ - return -EBUSY; if (has_failed(conf)) return -EINVAL; if (mddev->delta_disks < 0) { @@ -5596,6 +5593,9 @@ static int raid5_start_reshape(struct mddev *mddev) if (!check_stripe_cache(mddev)) return -ENOSPC; + if (has_failed(conf)) + return -EINVAL; + rdev_for_each(rdev, mddev) { if (!test_bit(In_sync, &rdev->flags) && !test_bit(Faulty, &rdev->flags)) -- cgit v1.2.3-70-g09d2 From da7613b8b042ab1d3ea55459ed97301e4a2097bb Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Tue, 22 May 2012 13:55:33 +1000 Subject: md/raid5: improve removal of extra devices after reshape. After a reshape which reduced the number of devices we need to disconnect the extra devices. The code for this doesn't currently handle 'replacement' devices. It is very unlikely that such devices will be present, but it is safest to handle them anyway. So simplify the handling. Just clear In_sync and leave it to remove_and_add_spaces (which will be called soon) to do the real works. Signed-off-by: NeilBrown --- drivers/md/raid5.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'drivers/md/raid5.c') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 8fba57e8f78..d26767246d2 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -5755,11 +5755,11 @@ 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, rdev) == 0) { - sysfs_unlink_rdev(mddev, rdev); - rdev->raid_disk = -1; - } + if (rdev) + clear_bit(In_sync, &rdev->flags); + rdev = conf->disks[d].replacement; + if (rdev) + clear_bit(In_sync, &rdev->flags); } } mddev->layout = conf->algorithm; -- cgit v1.2.3-70-g09d2