From 2b20982e3154266106573beac2a4d4ba57a2789a Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 3 Dec 2010 13:17:53 -0500 Subject: Btrfs: deal with space cache errors better Currently if the space cache inode generation number doesn't match the generation number in the space cache header we will just fail to load the space cache, but we won't mark the space cache as an error, so we'll keep getting that error each time somebody tries to cache that block group until we actually clear the thing. Fix this by marking the space cache as having an error so we only get the message once. This patch also makes it so that we don't try and setup space cache for a block group that isn't cached, since we won't be able to write it out anyway. None of these problems are actual problems, they are just annoying and sub-optimal. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index ddaf6340fe7..8c56f5b3894 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -2742,6 +2742,7 @@ static int cache_save_setup(struct btrfs_block_group_cache *block_group, struct btrfs_root *root = block_group->fs_info->tree_root; struct inode *inode = NULL; u64 alloc_hint = 0; + int dcs = BTRFS_DC_ERROR; int num_pages = 0; int retries = 0; int ret = 0; @@ -2796,6 +2797,8 @@ again: spin_lock(&block_group->lock); if (block_group->cached != BTRFS_CACHE_FINISHED) { + /* We're not cached, don't bother trying to write stuff out */ + dcs = BTRFS_DC_WRITTEN; spin_unlock(&block_group->lock); goto out_put; } @@ -2822,6 +2825,8 @@ again: ret = btrfs_prealloc_file_range_trans(inode, trans, 0, 0, num_pages, num_pages, num_pages, &alloc_hint); + if (!ret) + dcs = BTRFS_DC_SETUP; btrfs_free_reserved_data_space(inode, num_pages); out_put: iput(inode); @@ -2829,10 +2834,7 @@ out_free: btrfs_release_path(root, path); out: spin_lock(&block_group->lock); - if (ret) - block_group->disk_cache_state = BTRFS_DC_ERROR; - else - block_group->disk_cache_state = BTRFS_DC_SETUP; + block_group->disk_cache_state = dcs; spin_unlock(&block_group->lock); return ret; -- cgit v1.2.3-70-g09d2 From b8399dee478db7939cd0d6fda8ecacddf2facd03 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 8 Dec 2010 09:15:11 -0500 Subject: Btrfs: do not do fast caching if we are allocating blocks for tree_root Since the fast caching uses normal tree locking, we can possibly deadlock if we get to the caching via a btrfs_search_slot() on the tree_root. So just check to see if the root we are on is the tree root, and just don't do the fast caching. Reported-by: Sage Weil Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 18 ++++++++++++------ 1 file changed, 12 insertions(+), 6 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 8c56f5b3894..cec05e10014 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -429,6 +429,7 @@ err: static int cache_block_group(struct btrfs_block_group_cache *cache, struct btrfs_trans_handle *trans, + struct btrfs_root *root, int load_cache_only) { struct btrfs_fs_info *fs_info = cache->fs_info; @@ -442,9 +443,12 @@ static int cache_block_group(struct btrfs_block_group_cache *cache, /* * We can't do the read from on-disk cache during a commit since we need - * to have the normal tree locking. + * to have the normal tree locking. Also if we are currently trying to + * allocate blocks for the tree root we can't do the fast caching since + * we likely hold important locks. */ - if (!trans->transaction->in_commit) { + if (!trans->transaction->in_commit && + (root && root != root->fs_info->tree_root)) { spin_lock(&cache->lock); if (cache->cached != BTRFS_CACHE_NO) { spin_unlock(&cache->lock); @@ -4083,7 +4087,7 @@ static int update_block_group(struct btrfs_trans_handle *trans, * space back to the block group, otherwise we will leak space. */ if (!alloc && cache->cached == BTRFS_CACHE_NO) - cache_block_group(cache, trans, 1); + cache_block_group(cache, trans, NULL, 1); byte_in_group = bytenr - cache->key.objectid; WARN_ON(byte_in_group > cache->key.offset); @@ -4937,7 +4941,8 @@ have_block_group: if (unlikely(block_group->cached == BTRFS_CACHE_NO)) { u64 free_percent; - ret = cache_block_group(block_group, trans, 1); + ret = cache_block_group(block_group, trans, + orig_root, 1); if (block_group->cached == BTRFS_CACHE_FINISHED) goto have_block_group; @@ -4961,7 +4966,8 @@ have_block_group: if (loop > LOOP_CACHING_NOWAIT || (loop > LOOP_FIND_IDEAL && atomic_read(&space_info->caching_threads) < 2)) { - ret = cache_block_group(block_group, trans, 0); + ret = cache_block_group(block_group, trans, + orig_root, 0); BUG_ON(ret); } found_uncached_bg = true; @@ -5518,7 +5524,7 @@ int btrfs_alloc_logged_file_extent(struct btrfs_trans_handle *trans, u64 num_bytes = ins->offset; block_group = btrfs_lookup_block_group(root->fs_info, ins->objectid); - cache_block_group(block_group, trans, 0); + cache_block_group(block_group, trans, NULL, 0); caching_ctl = get_caching_control(block_group); if (!caching_ctl) { -- cgit v1.2.3-70-g09d2 From 84cd948cb11041f205242de457e680b9bb872a36 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 8 Dec 2010 12:24:01 -0500 Subject: Btrfs: do not BUG if we fail to remove the orphan item for dead snapshots Not being able to delete an orphan item isn't a horrible thing. The worst that happens is the next time around we try and do the orphan cleanup and we can't find the referenced object and just delete the item and move on. Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index cec05e10014..41133b064d7 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -6309,9 +6309,13 @@ int btrfs_drop_snapshot(struct btrfs_root *root, NULL, NULL); BUG_ON(ret < 0); if (ret > 0) { - ret = btrfs_del_orphan_item(trans, tree_root, - root->root_key.objectid); - BUG_ON(ret); + /* if we fail to delete the orphan item this time + * around, it'll get picked up the next time. + * + * The most common failure here is just -ENOENT. + */ + btrfs_del_orphan_item(trans, tree_root, + root->root_key.objectid); } } -- cgit v1.2.3-70-g09d2 From cd02dca56442e1504fd6bc5b96f7f1870162b266 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Mon, 13 Dec 2010 14:56:23 -0500 Subject: Btrfs: account for missing devices in RAID allocation profiles When we mount in RAID degraded mode without adding a new device to replace the failed one, we can end up using the wrong RAID flags for allocations. This results in strange combinations of block groups (raid1 in a raid10 filesystem) and corruptions when we try to allocate blocks from single spindle chunks on drives that are actually missing. The first device has two small 4MB chunks in it that mkfs creates and these are usually unused in a raid1 or raid10 setup. But, in -o degraded, the allocator will fall back to these because the mask of desired raid groups isn't correct. The fix here is to count the missing devices as we build up the list of devices in the system. This count is used when picking the raid level to make sure we continue using the same levels that were in place before we lost a drive. Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 17 +++++++++++++++-- fs/btrfs/volumes.c | 20 +++++++++++++++++++- fs/btrfs/volumes.h | 2 ++ 3 files changed, 36 insertions(+), 3 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 41133b064d7..4be231e0d2b 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3044,7 +3044,13 @@ static void set_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags) u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags) { - u64 num_devices = root->fs_info->fs_devices->rw_devices; + /* + * we add in the count of missing devices because we want + * to make sure that any RAID levels on a degraded FS + * continue to be honored. + */ + u64 num_devices = root->fs_info->fs_devices->rw_devices + + root->fs_info->fs_devices->missing_devices; if (num_devices == 1) flags &= ~(BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID0); @@ -7891,7 +7897,14 @@ static u64 update_block_group_flags(struct btrfs_root *root, u64 flags) u64 stripped = BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10; - num_devices = root->fs_info->fs_devices->rw_devices; + /* + * we add in the count of missing devices because we want + * to make sure that any RAID levels on a degraded FS + * continue to be honored. + */ + num_devices = root->fs_info->fs_devices->rw_devices + + root->fs_info->fs_devices->missing_devices; + if (num_devices == 1) { stripped |= BTRFS_BLOCK_GROUP_DUP; stripped = flags & ~stripped; diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 91851b555e2..177b7317959 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -413,12 +413,16 @@ static noinline int device_list_add(const char *path, device->fs_devices = fs_devices; fs_devices->num_devices++; - } else if (strcmp(device->name, path)) { + } else if (!device->name || strcmp(device->name, path)) { name = kstrdup(path, GFP_NOFS); if (!name) return -ENOMEM; kfree(device->name); device->name = name; + if (device->missing) { + fs_devices->missing_devices--; + device->missing = 0; + } } if (found_transid > fs_devices->latest_trans) { @@ -1238,6 +1242,9 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path) device->fs_devices->num_devices--; + if (device->missing) + root->fs_info->fs_devices->missing_devices--; + next_device = list_entry(root->fs_info->fs_devices->devices.next, struct btrfs_device, dev_list); if (device->bdev == root->fs_info->sb->s_bdev) @@ -3084,7 +3091,9 @@ static struct btrfs_device *add_missing_dev(struct btrfs_root *root, device->devid = devid; device->work.func = pending_bios_fn; device->fs_devices = fs_devices; + device->missing = 1; fs_devices->num_devices++; + fs_devices->missing_devices++; spin_lock_init(&device->io_lock); INIT_LIST_HEAD(&device->dev_alloc_list); memcpy(device->uuid, dev_uuid, BTRFS_UUID_SIZE); @@ -3282,6 +3291,15 @@ static int read_one_dev(struct btrfs_root *root, device = add_missing_dev(root, devid, dev_uuid); if (!device) return -ENOMEM; + } else if (!device->missing) { + /* + * this happens when a device that was properly setup + * in the device info lists suddenly goes bad. + * device->bdev is NULL, and so we have to set + * device->missing to one here + */ + root->fs_info->fs_devices->missing_devices++; + device->missing = 1; } } diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h index 31b0fabdd2e..a668c011698 100644 --- a/fs/btrfs/volumes.h +++ b/fs/btrfs/volumes.h @@ -45,6 +45,7 @@ struct btrfs_device { int barriers; int writeable; int in_fs_metadata; + int missing; spinlock_t io_lock; @@ -94,6 +95,7 @@ struct btrfs_fs_devices { u64 num_devices; u64 open_devices; u64 rw_devices; + u64 missing_devices; u64 total_rw_bytes; struct block_device *latest_bdev; -- cgit v1.2.3-70-g09d2 From 83a50de97fe96aca82389e061862ed760ece2283 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Mon, 13 Dec 2010 15:06:46 -0500 Subject: Btrfs: prevent RAID level downgrades when space is low The extent allocator has code that allows us to fill allocations from any available block group, even if it doesn't match the raid level we've requested. This was put in because adding a new drive to a filesystem made with the default mkfs options actually upgrades the metadata from single spindle dup to full RAID1. But, the code also allows us to allocate from a raid0 chunk when we really want a raid1 or raid10 chunk. This can cause big trouble because mkfs creates a small (4MB) raid0 chunk for data and metadata which then goes unused for raid1/raid10 installs. The allocator will happily wander in and allocate from that chunk when things get tight, which is not correct. The fix here is to make sure that we provide duplication when the caller has asked for it. It does all the dups to be any raid level, which preserves the dup->raid1 upgrade abilities. Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 20 +++++++++++++++++++- 1 file changed, 19 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 4be231e0d2b..7e5162e5c41 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4943,6 +4943,25 @@ search: btrfs_get_block_group(block_group); search_start = block_group->key.objectid; + /* + * this can happen if we end up cycling through all the + * raid types, but we want to make sure we only allocate + * for the proper type. + */ + if (!block_group_bits(block_group, data)) { + u64 extra = BTRFS_BLOCK_GROUP_DUP | + BTRFS_BLOCK_GROUP_RAID1 | + BTRFS_BLOCK_GROUP_RAID10; + + /* + * if they asked for extra copies and this block group + * doesn't provide them, bail. This does allow us to + * fill raid0 from raid1. + */ + if ((data & extra) && !(block_group->flags & extra)) + goto loop; + } + have_block_group: if (unlikely(block_group->cached == BTRFS_CACHE_NO)) { u64 free_percent; @@ -8273,7 +8292,6 @@ int btrfs_read_block_groups(struct btrfs_root *root) break; if (ret != 0) goto error; - leaf = path->nodes[0]; btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); cache = kzalloc(sizeof(*cache), GFP_NOFS); -- cgit v1.2.3-70-g09d2