From 3aa3b2b2b1edb813dc5342d0108befc39541542d Mon Sep 17 00:00:00 2001 From: Jonathan E Brassow Date: Wed, 7 Mar 2012 19:09:47 +0000 Subject: dm raid: set MD_CHANGE_DEVS when rebuilding The 'rebuild' parameter is used to rebuild individual devices in an array (e.g. resynchronize a RAID1 device or recalculate a parity device in higher RAID). The MD_CHANGE_DEVS flag must be set when this parameter is given in order to write out the superblocks and make the change take immediate effect. The code that handles new devices in super_load already sets MD_CHANGE_DEVS and 'FirstUse'. (The 'FirstUse' flag was being set as a special case for rebuilds in super_init_validation.) Add a condition for rebuilds in super_load to take care of both flags without the special case in 'super_init_validation'. Signed-off-by: Jonathan Brassow Cc: stable@kernel.org Signed-off-by: Alasdair G Kergon --- drivers/md/dm-raid.c | 16 ++++++++++------ 1 file changed, 10 insertions(+), 6 deletions(-) (limited to 'drivers/md/dm-raid.c') diff --git a/drivers/md/dm-raid.c b/drivers/md/dm-raid.c index 86cb7e5d83d5..f53bb389c18f 100644 --- a/drivers/md/dm-raid.c +++ b/drivers/md/dm-raid.c @@ -668,7 +668,14 @@ static int super_load(struct md_rdev *rdev, struct md_rdev *refdev) return ret; sb = page_address(rdev->sb_page); - if (sb->magic != cpu_to_le32(DM_RAID_MAGIC)) { + + /* + * Two cases that we want to write new superblocks and rebuild: + * 1) New device (no matching magic number) + * 2) Device specified for rebuild (!In_sync w/ offset == 0) + */ + if ((sb->magic != cpu_to_le32(DM_RAID_MAGIC)) || + (!test_bit(In_sync, &rdev->flags) && !rdev->recovery_offset)) { super_sync(rdev->mddev, rdev); set_bit(FirstUse, &rdev->flags); @@ -745,11 +752,8 @@ static int super_init_validation(struct mddev *mddev, struct md_rdev *rdev) */ rdev_for_each(r, t, mddev) { if (!test_bit(In_sync, &r->flags)) { - if (!test_bit(FirstUse, &r->flags)) - DMERR("Superblock area of " - "rebuild device %d should have been " - "cleared.", r->raid_disk); - set_bit(FirstUse, &r->flags); + DMINFO("Device %d specified for rebuild: " + "Clearing superblock", r->raid_disk); rebuilds++; } else if (test_bit(FirstUse, &r->flags)) new_devs++; -- cgit v1.2.3 From 0ca93de9b789e0eb05e103f0c04de72df13da73a Mon Sep 17 00:00:00 2001 From: Jonathan E Brassow Date: Wed, 7 Mar 2012 19:09:48 +0000 Subject: dm raid: fix flush support Fix dm-raid flush support. Both md and dm have support for flush, but the dm-raid target forgot to set the flag to indicate that flushes should be passed on. (Important for data integrity e.g. with writeback cache enabled.) Signed-off-by: Jonathan Brassow Acked-by: Mike Snitzer Cc: stable@kernel.org Signed-off-by: Alasdair G Kergon --- drivers/md/dm-raid.c | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/md/dm-raid.c') diff --git a/drivers/md/dm-raid.c b/drivers/md/dm-raid.c index f53bb389c18f..787022c18187 100644 --- a/drivers/md/dm-raid.c +++ b/drivers/md/dm-raid.c @@ -975,6 +975,7 @@ static int raid_ctr(struct dm_target *ti, unsigned argc, char **argv) INIT_WORK(&rs->md.event_work, do_table_event); ti->private = rs; + ti->num_flush_requests = 1; mutex_lock(&rs->md.reconfig_mutex); ret = md_run(&rs->md); -- cgit v1.2.3 From dafb20fa34320a472deb7442f25a0c086e0feb33 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Mon, 19 Mar 2012 12:46:39 +1100 Subject: md: tidy up rdev_for_each usage. md.h has an 'rdev_for_each()' macro for iterating the rdevs in an mddev. However it uses the 'safe' version of list_for_each_entry, and so requires the extra variable, but doesn't include 'safe' in the name, which is useful documentation. Consequently some places use this safe version without needing it, and many use an explicity list_for_each entry. So: - rename rdev_for_each to rdev_for_each_safe - create a new rdev_for_each which uses the plain list_for_each_entry, - use the 'safe' version only where needed, and convert all other list_for_each_entry calls to use rdev_for_each. Signed-off-by: NeilBrown --- drivers/md/dm-raid.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) (limited to 'drivers/md/dm-raid.c') diff --git a/drivers/md/dm-raid.c b/drivers/md/dm-raid.c index 787022c18187..c5a875d7b882 100644 --- a/drivers/md/dm-raid.c +++ b/drivers/md/dm-raid.c @@ -615,14 +615,14 @@ static int read_disk_sb(struct md_rdev *rdev, int size) static void super_sync(struct mddev *mddev, struct md_rdev *rdev) { - struct md_rdev *r, *t; + struct md_rdev *r; uint64_t failed_devices; struct dm_raid_superblock *sb; sb = page_address(rdev->sb_page); failed_devices = le64_to_cpu(sb->failed_devices); - rdev_for_each(r, t, mddev) + rdev_for_each(r, mddev) if ((r->raid_disk >= 0) && test_bit(Faulty, &r->flags)) failed_devices |= (1ULL << r->raid_disk); @@ -707,7 +707,7 @@ static int super_init_validation(struct mddev *mddev, struct md_rdev *rdev) struct dm_raid_superblock *sb; uint32_t new_devs = 0; uint32_t rebuilds = 0; - struct md_rdev *r, *t; + struct md_rdev *r; struct dm_raid_superblock *sb2; sb = page_address(rdev->sb_page); @@ -750,7 +750,7 @@ static int super_init_validation(struct mddev *mddev, struct md_rdev *rdev) * case the In_sync bit will /not/ be set and * recovery_cp must be MaxSector. */ - rdev_for_each(r, t, mddev) { + rdev_for_each(r, mddev) { if (!test_bit(In_sync, &r->flags)) { DMINFO("Device %d specified for rebuild: " "Clearing superblock", r->raid_disk); @@ -782,7 +782,7 @@ static int super_init_validation(struct mddev *mddev, struct md_rdev *rdev) * Now we set the Faulty bit for those devices that are * recorded in the superblock as failed. */ - rdev_for_each(r, t, mddev) { + rdev_for_each(r, mddev) { if (!r->sb_page) continue; sb2 = page_address(r->sb_page); @@ -855,11 +855,11 @@ static int super_validate(struct mddev *mddev, struct md_rdev *rdev) static int analyse_superblocks(struct dm_target *ti, struct raid_set *rs) { int ret; - struct md_rdev *rdev, *freshest, *tmp; + struct md_rdev *rdev, *freshest; struct mddev *mddev = &rs->md; freshest = NULL; - rdev_for_each(rdev, tmp, mddev) { + rdev_for_each(rdev, mddev) { if (!rdev->meta_bdev) continue; @@ -888,7 +888,7 @@ static int analyse_superblocks(struct dm_target *ti, struct raid_set *rs) if (super_validate(mddev, freshest)) return -EINVAL; - rdev_for_each(rdev, tmp, mddev) + rdev_for_each(rdev, mddev) if ((rdev != freshest) && super_validate(mddev, rdev)) return -EINVAL; -- cgit v1.2.3 From 0447568fc51e0268e201f7086d2450cf986e0411 Mon Sep 17 00:00:00 2001 From: Jonathan E Brassow Date: Wed, 28 Mar 2012 18:41:26 +0100 Subject: dm raid: handle failed devices during start up The dm-raid code currently fails to create a RAID array if any of the superblocks cannot be read. This was an oversight as there is already code to handle this case if the values ('- -') were provided for the failed array position. With this patch, if a superblock cannot be read, the array position's fields are initialized as though '- -' was set in the table. That is, the device is failed and the position should not be used, but if there is sufficient redundancy, the array should still be activated. Signed-off-by: Jonathan Brassow Signed-off-by: Alasdair G Kergon --- drivers/md/dm-raid.c | 53 ++++++++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 51 insertions(+), 2 deletions(-) (limited to 'drivers/md/dm-raid.c') diff --git a/drivers/md/dm-raid.c b/drivers/md/dm-raid.c index c5a875d7b882..b0ba52459ed7 100644 --- a/drivers/md/dm-raid.c +++ b/drivers/md/dm-raid.c @@ -604,7 +604,9 @@ static int read_disk_sb(struct md_rdev *rdev, int size) return 0; if (!sync_page_io(rdev, 0, size, rdev->sb_page, READ, 1)) { - DMERR("Failed to read device superblock"); + DMERR("Failed to read superblock of device at position %d", + rdev->raid_disk); + set_bit(Faulty, &rdev->flags); return -EINVAL; } @@ -855,9 +857,25 @@ static int super_validate(struct mddev *mddev, struct md_rdev *rdev) static int analyse_superblocks(struct dm_target *ti, struct raid_set *rs) { int ret; + unsigned redundancy = 0; + struct raid_dev *dev; struct md_rdev *rdev, *freshest; struct mddev *mddev = &rs->md; + switch (rs->raid_type->level) { + case 1: + redundancy = rs->md.raid_disks - 1; + break; + case 4: + case 5: + case 6: + redundancy = rs->raid_type->parity_devs; + break; + default: + ti->error = "Unknown RAID type"; + return -EINVAL; + } + freshest = NULL; rdev_for_each(rdev, mddev) { if (!rdev->meta_bdev) @@ -872,6 +890,37 @@ static int analyse_superblocks(struct dm_target *ti, struct raid_set *rs) case 0: break; default: + dev = container_of(rdev, struct raid_dev, rdev); + if (redundancy--) { + if (dev->meta_dev) + dm_put_device(ti, dev->meta_dev); + + dev->meta_dev = NULL; + rdev->meta_bdev = NULL; + + if (rdev->sb_page) + put_page(rdev->sb_page); + + rdev->sb_page = NULL; + + rdev->sb_loaded = 0; + + /* + * We might be able to salvage the data device + * even though the meta device has failed. For + * now, we behave as though '- -' had been + * set for this device in the table. + */ + if (dev->data_dev) + dm_put_device(ti, dev->data_dev); + + dev->data_dev = NULL; + rdev->bdev = NULL; + + list_del(&rdev->same_set); + + continue; + } ti->error = "Failed to load superblock"; return ret; } @@ -1214,7 +1263,7 @@ static void raid_resume(struct dm_target *ti) static struct target_type raid_target = { .name = "raid", - .version = {1, 1, 0}, + .version = {1, 2, 0}, .module = THIS_MODULE, .ctr = raid_ctr, .dtr = raid_dtr, -- cgit v1.2.3