From c3b9a62c8f932f32a733d6b628f61f3f28345727 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Wed, 18 Aug 2010 05:29:12 -0400 Subject: btrfs: replace barriers with explicit flush / FUA usage Switch to the WRITE_FLUSH_FUA flag for log writes, remove the EOPNOTSUPP detection for barriers and stop setting the barrier flag for discards. Signed-off-by: Christoph Hellwig Acked-by: Chris Mason Signed-off-by: Tejun Heo Signed-off-by: Jens Axboe --- fs/btrfs/extent-tree.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 32d094002a57..43dc9ea9aef6 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -1696,7 +1696,7 @@ static void btrfs_issue_discard(struct block_device *bdev, u64 start, u64 len) { blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_KERNEL, - BLKDEV_IFL_WAIT | BLKDEV_IFL_BARRIER); + BLKDEV_IFL_WAIT); } static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr, -- cgit v1.2.2 From dd3932eddf428571762596e17b65f5dc92ca361b Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 16 Sep 2010 20:51:46 +0200 Subject: block: remove BLKDEV_IFL_WAIT All the blkdev_issue_* helpers can only sanely be used for synchronous caller. To issue cache flushes or barriers asynchronously the caller needs to set up a bio by itself with a completion callback to move the asynchronous state machine ahead. So drop the BLKDEV_IFL_WAIT flag that is always specified when calling blkdev_issue_* and also remove the now unused flags argument to blkdev_issue_flush and blkdev_issue_zeroout. For blkdev_issue_discard we need to keep it for the secure discard flag, which gains a more descriptive name and loses the bitops vs flag confusion. Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- fs/btrfs/extent-tree.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 43dc9ea9aef6..0b81ecdb101c 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -1695,8 +1695,7 @@ static int remove_extent_backref(struct btrfs_trans_handle *trans, static void btrfs_issue_discard(struct block_device *bdev, u64 start, u64 len) { - blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_KERNEL, - BLKDEV_IFL_WAIT); + blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_KERNEL, 0); } static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr, -- cgit v1.2.2 From a1f765061e1491d5ec467429d0d6adfd9df2f6d9 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Thu, 16 Sep 2010 14:29:55 -0400 Subject: Btrfs: stop trying to shrink delalloc if there are no inodes to reclaim In very severe ENOSPC cases we can run out of inodes to do delalloc on, which means we'll just keep looping trying to shrink delalloc. Instead, if we fail to shrink delalloc 3 times in a row break out since we're not likely to make any progress. Tested this with a 100mb fs an xfstests test 13. Before the patch it would hang the box, with the patch we get -ENOSPC like we should. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 32d094002a57..c6a5d9095d5f 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3115,6 +3115,7 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans, u64 reserved; u64 max_reclaim; u64 reclaimed = 0; + int no_reclaim = 0; int pause = 1; int ret; @@ -3131,12 +3132,16 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans, while (1) { ret = btrfs_start_one_delalloc_inode(root, trans ? 1 : 0); if (!ret) { + if (no_reclaim > 2) + break; + no_reclaim++; __set_current_state(TASK_INTERRUPTIBLE); schedule_timeout(pause); pause <<= 1; if (pause > HZ / 10) pause = HZ / 10; } else { + no_reclaim = 0; pause = 1; } -- cgit v1.2.2 From 89a55897a2fbbceb94480952784004bf23911d38 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Thu, 14 Oct 2010 14:52:27 -0400 Subject: Btrfs: fix df regression The new ENOSPC stuff breaks out the raid types which breaks the way we were reporting df to the system. This fixes it back so that Available is the total space available to data and used is the actual bytes used by the filesystem. This means that Available is Total - data used - all of the metadata space. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 10 ++++++++++ 1 file changed, 10 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index c6a5d9095d5f..4669c6f8a44d 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -2763,6 +2763,7 @@ static int update_space_info(struct btrfs_fs_info *info, u64 flags, if (found) { spin_lock(&found->lock); found->total_bytes += total_bytes; + found->disk_total += total_bytes * factor; found->bytes_used += bytes_used; found->disk_used += bytes_used * factor; found->full = 0; @@ -2782,6 +2783,7 @@ static int update_space_info(struct btrfs_fs_info *info, u64 flags, BTRFS_BLOCK_GROUP_SYSTEM | BTRFS_BLOCK_GROUP_METADATA); found->total_bytes = total_bytes; + found->disk_total = total_bytes * factor; found->bytes_used = bytes_used; found->disk_used = bytes_used * factor; found->bytes_pinned = 0; @@ -8095,6 +8097,7 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans, struct btrfs_free_cluster *cluster; struct btrfs_key key; int ret; + int factor; root = root->fs_info->extent_root; @@ -8103,6 +8106,12 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans, BUG_ON(!block_group->ro); memcpy(&key, &block_group->key, sizeof(key)); + if (block_group->flags & (BTRFS_BLOCK_GROUP_DUP | + BTRFS_BLOCK_GROUP_RAID1 | + BTRFS_BLOCK_GROUP_RAID10)) + factor = 2; + else + factor = 1; /* make sure this block group isn't part of an allocation cluster */ cluster = &root->fs_info->data_alloc_cluster; @@ -8143,6 +8152,7 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans, spin_lock(&block_group->space_info->lock); block_group->space_info->total_bytes -= block_group->key.offset; block_group->space_info->bytes_readonly -= block_group->key.offset; + block_group->space_info->disk_total -= block_group->key.offset * factor; spin_unlock(&block_group->space_info->lock); btrfs_clear_space_info_full(root->fs_info); -- cgit v1.2.2 From 6d48755d02b150de7f47e7b4753202f2fc9f990f Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 15 Oct 2010 15:13:32 -0400 Subject: Btrfs: fix reservation code for mixed block groups The global reservation stuff tries to add together DATA and METADATA used in order to figure out how much to reserve for everything, but this doesn't work right for mixed block groups. Instead if we have mixed block groups just set data used to 0. Also with mixed block groups we will use bytes_may_use for keeping track of delalloc bytes, so we need to take that into account in our reservation calculations. Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 4669c6f8a44d..0f27f7b48804 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3213,7 +3213,8 @@ static int reserve_metadata_bytes(struct btrfs_block_rsv *block_rsv, spin_lock(&space_info->lock); unused = space_info->bytes_used + space_info->bytes_reserved + - space_info->bytes_pinned + space_info->bytes_readonly; + space_info->bytes_pinned + space_info->bytes_readonly + + space_info->bytes_may_use; if (unused < space_info->total_bytes) unused = space_info->total_bytes - unused; @@ -3507,6 +3508,8 @@ static u64 calc_global_metadata_size(struct btrfs_fs_info *fs_info) sinfo = __find_space_info(fs_info, BTRFS_BLOCK_GROUP_METADATA); spin_lock(&sinfo->lock); + if (sinfo->flags & BTRFS_BLOCK_GROUP_DATA) + data_used = 0; meta_used = sinfo->bytes_used; spin_unlock(&sinfo->lock); @@ -3534,7 +3537,8 @@ static void update_global_block_rsv(struct btrfs_fs_info *fs_info) block_rsv->size = num_bytes; num_bytes = sinfo->bytes_used + sinfo->bytes_pinned + - sinfo->bytes_reserved + sinfo->bytes_readonly; + sinfo->bytes_reserved + sinfo->bytes_readonly + + sinfo->bytes_may_use; if (sinfo->total_bytes > num_bytes) { num_bytes = sinfo->total_bytes - num_bytes; -- cgit v1.2.2 From 0019f10db6f596f3e14a19f9bd7059a1b85b0853 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 15 Oct 2010 15:18:40 -0400 Subject: Btrfs: re-work delalloc flushing Currently we try and flush delalloc, but we only do that in a sort of weak way, which works fine in most cases but if we're under heavy pressure we need to be able to wait for flushing to happen. Also instead of checking the bytes reserved in the block_rsv, check the space info since it is more accurate. The sync option will be used in a future patch. Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 26 ++++++++++++++------------ 1 file changed, 14 insertions(+), 12 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 0f27f7b48804..2846cebc9427 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3111,9 +3111,10 @@ static int maybe_allocate_chunk(struct btrfs_trans_handle *trans, * shrink metadata reservation for delalloc */ static int shrink_delalloc(struct btrfs_trans_handle *trans, - struct btrfs_root *root, u64 to_reclaim) + struct btrfs_root *root, u64 to_reclaim, int sync) { struct btrfs_block_rsv *block_rsv; + struct btrfs_space_info *space_info; u64 reserved; u64 max_reclaim; u64 reclaimed = 0; @@ -3122,9 +3123,10 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans, int ret; block_rsv = &root->fs_info->delalloc_block_rsv; - spin_lock(&block_rsv->lock); - reserved = block_rsv->reserved; - spin_unlock(&block_rsv->lock); + space_info = block_rsv->space_info; + spin_lock(&space_info->lock); + reserved = space_info->bytes_reserved; + spin_unlock(&space_info->lock); if (reserved == 0) return 0; @@ -3132,7 +3134,7 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans, max_reclaim = min(reserved, to_reclaim); while (1) { - ret = btrfs_start_one_delalloc_inode(root, trans ? 1 : 0); + ret = btrfs_start_one_delalloc_inode(root, trans ? 1 : 0, sync); if (!ret) { if (no_reclaim > 2) break; @@ -3147,11 +3149,11 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans, pause = 1; } - spin_lock(&block_rsv->lock); - if (reserved > block_rsv->reserved) - reclaimed = reserved - block_rsv->reserved; - reserved = block_rsv->reserved; - spin_unlock(&block_rsv->lock); + spin_lock(&space_info->lock); + if (reserved > space_info->bytes_reserved) + reclaimed += reserved - space_info->bytes_reserved; + reserved = space_info->bytes_reserved; + spin_unlock(&space_info->lock); if (reserved == 0 || reclaimed >= max_reclaim) break; @@ -3180,7 +3182,7 @@ static int should_retry_reserve(struct btrfs_trans_handle *trans, if (trans && trans->transaction->in_commit) return -ENOSPC; - ret = shrink_delalloc(trans, root, num_bytes); + ret = shrink_delalloc(trans, root, num_bytes, 0); if (ret) return ret; @@ -3729,7 +3731,7 @@ again: block_rsv_add_bytes(block_rsv, to_reserve, 1); if (block_rsv->size > 512 * 1024 * 1024) - shrink_delalloc(NULL, root, to_reserve); + shrink_delalloc(NULL, root, to_reserve, 0); return 0; } -- cgit v1.2.2 From 14ed0ca6e8236f2d264c4a8faec9e3a2b3d04377 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 15 Oct 2010 15:23:48 -0400 Subject: Btrfs: don't allocate chunks as aggressively Because the ENOSPC code over reserves super aggressively we end up allocating chunks way more often than we should. For example with my fs_mark tests on a 2gb fs I can end up reserved 1gb just for metadata, when only 34mb of that is being used. So instead check to see if the amount of space actually used is less than 30% of the total space, and if so don't allocate a chunk, but only if we have at least 256mb of free space to make sure we don't put too much pressure on free space. Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 2846cebc9427..aca3314ef8b9 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3000,8 +3000,7 @@ static void force_metadata_allocation(struct btrfs_fs_info *info) rcu_read_unlock(); } -static int should_alloc_chunk(struct btrfs_space_info *sinfo, - u64 alloc_bytes) +static int should_alloc_chunk(struct btrfs_space_info *sinfo, u64 alloc_bytes) { u64 num_bytes = sinfo->total_bytes - sinfo->bytes_readonly; @@ -3013,6 +3012,10 @@ static int should_alloc_chunk(struct btrfs_space_info *sinfo, alloc_bytes < div_factor(num_bytes, 8)) return 0; + if (num_bytes > 256 * 1024 * 1024 && + sinfo->bytes_used < div_factor(num_bytes, 3)) + return 0; + return 1; } -- cgit v1.2.2 From 8bb8ab2e93f9c3c9453e13be0f37d344a32a3a6d Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 15 Oct 2010 16:52:49 -0400 Subject: Btrfs: rework how we reserve metadata bytes With multi-threaded writes we were getting ENOSPC early because somebody would come in, start flushing delalloc because they couldn't make their reservation, and in the meantime other threads would come in and use the space that was getting freed up, so when the original thread went to check to see if they had space they didn't and they'd return ENOSPC. So instead if we have some free space but not enough for our reservation, take the reservation and then start doing the flushing. The only time we don't take reservations is when we've already overcommitted our space, that way we don't have people who come late to the party way overcommitting ourselves. This also moves all of the retrying and flushing code into reserve_metdata_bytes so it's all uniform. This keeps my fs_mark test from returning -ENOSPC as soon as it starts and actually lets me fill up the disk. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 238 +++++++++++++++++++++++++++---------------------- 1 file changed, 129 insertions(+), 109 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index aca3314ef8b9..180a50146ddf 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3078,38 +3078,6 @@ out: return ret; } -static int maybe_allocate_chunk(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct btrfs_space_info *sinfo, u64 num_bytes) -{ - int ret; - int end_trans = 0; - - if (sinfo->full) - return 0; - - spin_lock(&sinfo->lock); - ret = should_alloc_chunk(sinfo, num_bytes + 2 * 1024 * 1024); - spin_unlock(&sinfo->lock); - if (!ret) - return 0; - - if (!trans) { - trans = btrfs_join_transaction(root, 1); - BUG_ON(IS_ERR(trans)); - end_trans = 1; - } - - ret = do_chunk_alloc(trans, root->fs_info->extent_root, - num_bytes + 2 * 1024 * 1024, - get_alloc_profile(root, sinfo->flags), 0); - - if (end_trans) - btrfs_end_transaction(trans, root); - - return ret == 1 ? 1 : 0; -} - /* * shrink metadata reservation for delalloc */ @@ -3167,79 +3135,138 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans, return reclaimed >= to_reclaim; } -static int should_retry_reserve(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct btrfs_block_rsv *block_rsv, - u64 num_bytes, int *retries) +/* + * Retries tells us how many times we've called reserve_metadata_bytes. The + * idea is if this is the first call (retries == 0) then we will add to our + * reserved count if we can't make the allocation in order to hold our place + * while we go and try and free up space. That way for retries > 1 we don't try + * and add space, we just check to see if the amount of unused space is >= the + * total space, meaning that our reservation is valid. + * + * However if we don't intend to retry this reservation, pass -1 as retries so + * that it short circuits this logic. + */ +static int reserve_metadata_bytes(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_block_rsv *block_rsv, + u64 orig_bytes, int flush) { struct btrfs_space_info *space_info = block_rsv->space_info; - int ret; + u64 unused; + u64 num_bytes = orig_bytes; + int retries = 0; + int ret = 0; + bool reserved = false; - if ((*retries) > 2) - return -ENOSPC; +again: + ret = -ENOSPC; + if (reserved) + num_bytes = 0; - ret = maybe_allocate_chunk(trans, root, space_info, num_bytes); - if (ret) - return 1; + spin_lock(&space_info->lock); + unused = space_info->bytes_used + space_info->bytes_reserved + + space_info->bytes_pinned + space_info->bytes_readonly + + space_info->bytes_may_use; - if (trans && trans->transaction->in_commit) - return -ENOSPC; + /* + * The idea here is that we've not already over-reserved the block group + * then we can go ahead and save our reservation first and then start + * flushing if we need to. Otherwise if we've already overcommitted + * lets start flushing stuff first and then come back and try to make + * our reservation. + */ + if (unused <= space_info->total_bytes) { + unused -= space_info->total_bytes; + if (unused >= num_bytes) { + if (!reserved) + space_info->bytes_reserved += orig_bytes; + ret = 0; + } else { + /* + * Ok set num_bytes to orig_bytes since we aren't + * overocmmitted, this way we only try and reclaim what + * we need. + */ + num_bytes = orig_bytes; + } + } else { + /* + * Ok we're over committed, set num_bytes to the overcommitted + * amount plus the amount of bytes that we need for this + * reservation. + */ + num_bytes = unused - space_info->total_bytes + + (orig_bytes * (retries + 1)); + } - ret = shrink_delalloc(trans, root, num_bytes, 0); - if (ret) - return ret; + /* + * Couldn't make our reservation, save our place so while we're trying + * to reclaim space we can actually use it instead of somebody else + * stealing it from us. + */ + if (ret && !reserved) { + space_info->bytes_reserved += orig_bytes; + reserved = true; + } - spin_lock(&space_info->lock); - if (space_info->bytes_pinned < num_bytes) - ret = 1; spin_unlock(&space_info->lock); - if (ret) - return -ENOSPC; - - (*retries)++; - if (trans) - return -EAGAIN; + if (!ret) + return 0; - trans = btrfs_join_transaction(root, 1); - BUG_ON(IS_ERR(trans)); - ret = btrfs_commit_transaction(trans, root); - BUG_ON(ret); + if (!flush) + goto out; - return 1; -} + /* + * We do synchronous shrinking since we don't actually unreserve + * metadata until after the IO is completed. + */ + ret = shrink_delalloc(trans, root, num_bytes, 1); + if (ret > 0) + return 0; + else if (ret < 0) + goto out; -static int reserve_metadata_bytes(struct btrfs_block_rsv *block_rsv, - u64 num_bytes) -{ - struct btrfs_space_info *space_info = block_rsv->space_info; - u64 unused; - int ret = -ENOSPC; + /* + * So if we were overcommitted it's possible that somebody else flushed + * out enough space and we simply didn't have enough space to reclaim, + * so go back around and try again. + */ + if (retries < 2) { + retries++; + goto again; + } spin_lock(&space_info->lock); - unused = space_info->bytes_used + space_info->bytes_reserved + - space_info->bytes_pinned + space_info->bytes_readonly + - space_info->bytes_may_use; + /* + * Not enough space to be reclaimed, don't bother committing the + * transaction. + */ + if (space_info->bytes_pinned < orig_bytes) + ret = -ENOSPC; + spin_unlock(&space_info->lock); + if (ret) + goto out; - if (unused < space_info->total_bytes) - unused = space_info->total_bytes - unused; - else - unused = 0; + ret = -EAGAIN; + if (trans) + goto out; - if (unused >= num_bytes) { - if (block_rsv->priority >= 10) { - space_info->bytes_reserved += num_bytes; - ret = 0; - } else { - if ((unused + block_rsv->reserved) * - block_rsv->priority >= - (num_bytes + block_rsv->reserved) * 10) { - space_info->bytes_reserved += num_bytes; - ret = 0; - } - } + + ret = -ENOSPC; + trans = btrfs_join_transaction(root, 1); + if (IS_ERR(trans)) + goto out; + ret = btrfs_commit_transaction(trans, root); + if (!ret) + goto again; + +out: + if (reserved) { + spin_lock(&space_info->lock); + space_info->bytes_reserved -= orig_bytes; + spin_unlock(&space_info->lock); } - spin_unlock(&space_info->lock); return ret; } @@ -3383,23 +3410,19 @@ void btrfs_add_durable_block_rsv(struct btrfs_fs_info *fs_info, int btrfs_block_rsv_add(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_block_rsv *block_rsv, - u64 num_bytes, int *retries) + u64 num_bytes) { int ret; if (num_bytes == 0) return 0; -again: - ret = reserve_metadata_bytes(block_rsv, num_bytes); + + ret = reserve_metadata_bytes(trans, root, block_rsv, num_bytes, 1); if (!ret) { block_rsv_add_bytes(block_rsv, num_bytes, 1); return 0; } - ret = should_retry_reserve(trans, root, block_rsv, num_bytes, retries); - if (ret > 0) - goto again; - return ret; } @@ -3434,7 +3457,8 @@ int btrfs_block_rsv_check(struct btrfs_trans_handle *trans, return 0; if (block_rsv->refill_used) { - ret = reserve_metadata_bytes(block_rsv, num_bytes); + ret = reserve_metadata_bytes(trans, root, block_rsv, + num_bytes, 0); if (!ret) { block_rsv_add_bytes(block_rsv, num_bytes, 0); return 0; @@ -3614,7 +3638,7 @@ static u64 calc_trans_metadata_size(struct btrfs_root *root, int num_items) int btrfs_trans_reserve_metadata(struct btrfs_trans_handle *trans, struct btrfs_root *root, - int num_items, int *retries) + int num_items) { u64 num_bytes; int ret; @@ -3624,7 +3648,7 @@ int btrfs_trans_reserve_metadata(struct btrfs_trans_handle *trans, num_bytes = calc_trans_metadata_size(root, num_items); ret = btrfs_block_rsv_add(trans, root, &root->fs_info->trans_block_rsv, - num_bytes, retries); + num_bytes); if (!ret) { trans->bytes_reserved += num_bytes; trans->block_rsv = &root->fs_info->trans_block_rsv; @@ -3698,14 +3722,13 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes) struct btrfs_block_rsv *block_rsv = &root->fs_info->delalloc_block_rsv; u64 to_reserve; int nr_extents; - int retries = 0; int ret; if (btrfs_transaction_in_commit(root->fs_info)) schedule_timeout(1); num_bytes = ALIGN(num_bytes, root->sectorsize); -again: + spin_lock(&BTRFS_I(inode)->accounting_lock); nr_extents = atomic_read(&BTRFS_I(inode)->outstanding_extents) + 1; if (nr_extents > BTRFS_I(inode)->reserved_extents) { @@ -3715,18 +3738,14 @@ again: nr_extents = 0; to_reserve = 0; } + spin_unlock(&BTRFS_I(inode)->accounting_lock); to_reserve += calc_csum_metadata_size(inode, num_bytes); - ret = reserve_metadata_bytes(block_rsv, to_reserve); - if (ret) { - spin_unlock(&BTRFS_I(inode)->accounting_lock); - ret = should_retry_reserve(NULL, root, block_rsv, to_reserve, - &retries); - if (ret > 0) - goto again; + ret = reserve_metadata_bytes(NULL, root, block_rsv, to_reserve, 1); + if (ret) return ret; - } + spin_lock(&BTRFS_I(inode)->accounting_lock); BTRFS_I(inode)->reserved_extents += nr_extents; atomic_inc(&BTRFS_I(inode)->outstanding_extents); spin_unlock(&BTRFS_I(inode)->accounting_lock); @@ -5325,7 +5344,8 @@ use_block_rsv(struct btrfs_trans_handle *trans, block_rsv = get_block_rsv(trans, root); if (block_rsv->size == 0) { - ret = reserve_metadata_bytes(block_rsv, blocksize); + ret = reserve_metadata_bytes(trans, root, block_rsv, + blocksize, 0); if (ret) return ERR_PTR(ret); return block_rsv; -- cgit v1.2.2 From 382279336f428c80f344edfc30d53797e3e76146 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 26 Oct 2010 12:52:53 -0400 Subject: Btrfs: set trans to null in reserve_metadata_bytes if we commit the transaction btrfs_commit_transaction will free our trans, but because we pass trans to shrink_delalloc we could possibly have a use after free situation. So instead if we commit the transaction, set trans to null and set committed to true so we don't keep trying to commit a transaction. This fixes a panic I could reproduce at will. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 9 ++++++--- 1 file changed, 6 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 180a50146ddf..e2dfd4ab3b9b 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3157,6 +3157,7 @@ static int reserve_metadata_bytes(struct btrfs_trans_handle *trans, int retries = 0; int ret = 0; bool reserved = false; + bool committed = false; again: ret = -ENOSPC; @@ -3249,17 +3250,19 @@ again: goto out; ret = -EAGAIN; - if (trans) + if (trans || committed) goto out; - ret = -ENOSPC; trans = btrfs_join_transaction(root, 1); if (IS_ERR(trans)) goto out; ret = btrfs_commit_transaction(trans, root); - if (!ret) + if (!ret) { + trans = NULL; + committed = true; goto again; + } out: if (reserved) { -- cgit v1.2.2 From e9bb7f10d3617304ef94ff7aa8fefbce3078f08b Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 26 Oct 2010 12:55:03 -0400 Subject: Btrfs: remove warn_on from use_block_rsv Because btrfs_dirty_inode does a btrfs_join_transaction, it doesn't actually reserve space. It does this so we can try and dirty the inode quickly without having to deal with the ENOSPC problems. But if it does get back ENOSPC it handles it properly. The problem is use_block_rsv does a WARN_ON whenever this case happens, even tho btrfs_dirty_inode takes it into account and actually expects to get -ENOSPC if things are particularly tight. So instead just remove the warning. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 5 ----- 1 file changed, 5 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index e2dfd4ab3b9b..33785333f293 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5358,11 +5358,6 @@ use_block_rsv(struct btrfs_trans_handle *trans, if (!ret) return block_rsv; - WARN_ON(1); - printk(KERN_INFO"block_rsv size %llu reserved %llu freed %llu %llu\n", - block_rsv->size, block_rsv->reserved, - block_rsv->freed[0], block_rsv->freed[1]); - return ERR_PTR(-ENOSPC); } -- cgit v1.2.2 From 0af3d00bad38d3bb9912a60928ad0669f17bdb76 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Mon, 21 Jun 2010 14:48:16 -0400 Subject: Btrfs: create special free space cache inode In order to save free space cache, we need an inode to hold the data, and we need a special item to point at the right inode for the right block group. So first, create a special item that will point to the right inode, and the number of extent entries we will have and the number of bitmaps we will have. We truncate and pre-allocate space everytime to make sure it's uptodate. This feature will be turned on as soon as you mount with -o space_cache, however it is safe to boot into old kernels, they will just generate the cache the old fashion way. When you boot back into a newer kernel we will notice that we modified and not the cache and automatically discard the cache. Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 231 +++++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 225 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 32d094002a57..aab40fb3faed 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -2688,6 +2688,109 @@ next_block_group(struct btrfs_root *root, return cache; } +static int cache_save_setup(struct btrfs_block_group_cache *block_group, + struct btrfs_trans_handle *trans, + struct btrfs_path *path) +{ + struct btrfs_root *root = block_group->fs_info->tree_root; + struct inode *inode = NULL; + u64 alloc_hint = 0; + int num_pages = 0; + int retries = 0; + int ret = 0; + + /* + * If this block group is smaller than 100 megs don't bother caching the + * block group. + */ + if (block_group->key.offset < (100 * 1024 * 1024)) { + spin_lock(&block_group->lock); + block_group->disk_cache_state = BTRFS_DC_WRITTEN; + spin_unlock(&block_group->lock); + return 0; + } + +again: + inode = lookup_free_space_inode(root, block_group, path); + if (IS_ERR(inode) && PTR_ERR(inode) != -ENOENT) { + ret = PTR_ERR(inode); + btrfs_release_path(root, path); + goto out; + } + + if (IS_ERR(inode)) { + BUG_ON(retries); + retries++; + + if (block_group->ro) + goto out_free; + + ret = create_free_space_inode(root, trans, block_group, path); + if (ret) + goto out_free; + goto again; + } + + /* + * We want to set the generation to 0, that way if anything goes wrong + * from here on out we know not to trust this cache when we load up next + * time. + */ + BTRFS_I(inode)->generation = 0; + ret = btrfs_update_inode(trans, root, inode); + WARN_ON(ret); + + if (i_size_read(inode) > 0) { + ret = btrfs_truncate_free_space_cache(root, trans, path, + inode); + if (ret) + goto out_put; + } + + spin_lock(&block_group->lock); + if (block_group->cached != BTRFS_CACHE_FINISHED) { + spin_unlock(&block_group->lock); + goto out_put; + } + spin_unlock(&block_group->lock); + + num_pages = (int)div64_u64(block_group->key.offset, 1024 * 1024 * 1024); + if (!num_pages) + num_pages = 1; + + /* + * Just to make absolutely sure we have enough space, we're going to + * preallocate 12 pages worth of space for each block group. In + * practice we ought to use at most 8, but we need extra space so we can + * add our header and have a terminator between the extents and the + * bitmaps. + */ + num_pages *= 16; + num_pages *= PAGE_CACHE_SIZE; + + ret = btrfs_check_data_free_space(inode, num_pages); + if (ret) + goto out_put; + + ret = btrfs_prealloc_file_range_trans(inode, trans, 0, 0, num_pages, + num_pages, num_pages, + &alloc_hint); + btrfs_free_reserved_data_space(inode, num_pages); +out_put: + iput(inode); +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; + spin_unlock(&block_group->lock); + + return ret; +} + int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans, struct btrfs_root *root) { @@ -2700,6 +2803,25 @@ int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans, if (!path) return -ENOMEM; +again: + while (1) { + cache = btrfs_lookup_first_block_group(root->fs_info, last); + while (cache) { + if (cache->disk_cache_state == BTRFS_DC_CLEAR) + break; + cache = next_block_group(root, cache); + } + if (!cache) { + if (last == 0) + break; + last = 0; + continue; + } + err = cache_save_setup(cache, trans, path); + last = cache->key.objectid + cache->key.offset; + btrfs_put_block_group(cache); + } + while (1) { if (last == 0) { err = btrfs_run_delayed_refs(trans, root, @@ -2709,6 +2831,11 @@ int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans, cache = btrfs_lookup_first_block_group(root->fs_info, last); while (cache) { + if (cache->disk_cache_state == BTRFS_DC_CLEAR) { + btrfs_put_block_group(cache); + goto again; + } + if (cache->dirty) break; cache = next_block_group(root, cache); @@ -2883,11 +3010,16 @@ int btrfs_check_data_free_space(struct inode *inode, u64 bytes) struct btrfs_space_info *data_sinfo; struct btrfs_root *root = BTRFS_I(inode)->root; u64 used; - int ret = 0, committed = 0; + int ret = 0, committed = 0, alloc_chunk = 1; /* make sure bytes are sectorsize aligned */ bytes = (bytes + root->sectorsize - 1) & ~((u64)root->sectorsize - 1); + if (root == root->fs_info->tree_root) { + alloc_chunk = 0; + committed = 1; + } + data_sinfo = BTRFS_I(inode)->space_info; if (!data_sinfo) goto alloc; @@ -2906,7 +3038,7 @@ again: * if we don't have enough free bytes in this space then we need * to alloc a new chunk. */ - if (!data_sinfo->full) { + if (!data_sinfo->full && alloc_chunk) { u64 alloc_target; data_sinfo->force_alloc = 1; @@ -3777,12 +3909,12 @@ static int update_block_group(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 bytenr, u64 num_bytes, int alloc) { - struct btrfs_block_group_cache *cache; + struct btrfs_block_group_cache *cache = NULL; struct btrfs_fs_info *info = root->fs_info; - int factor; u64 total = num_bytes; u64 old_val; u64 byte_in_group; + int factor; /* block accounting for super block */ spin_lock(&info->delalloc_lock); @@ -3804,11 +3936,17 @@ static int update_block_group(struct btrfs_trans_handle *trans, factor = 2; else factor = 1; + byte_in_group = bytenr - cache->key.objectid; WARN_ON(byte_in_group > cache->key.offset); spin_lock(&cache->space_info->lock); spin_lock(&cache->lock); + + if (btrfs_super_cache_generation(&info->super_copy) != 0 && + cache->disk_cache_state < BTRFS_DC_CLEAR) + cache->disk_cache_state = BTRFS_DC_CLEAR; + cache->dirty = 1; old_val = btrfs_block_group_used(&cache->item); num_bytes = min(total, cache->key.offset - byte_in_group); @@ -7814,6 +7952,40 @@ out: return ret; } +void btrfs_put_block_group_cache(struct btrfs_fs_info *info) +{ + struct btrfs_block_group_cache *block_group; + u64 last = 0; + + while (1) { + struct inode *inode; + + block_group = btrfs_lookup_first_block_group(info, last); + while (block_group) { + spin_lock(&block_group->lock); + if (block_group->iref) + break; + spin_unlock(&block_group->lock); + block_group = next_block_group(info->tree_root, + block_group); + } + if (!block_group) { + if (last == 0) + break; + last = 0; + continue; + } + + inode = block_group->inode; + block_group->iref = 0; + block_group->inode = NULL; + spin_unlock(&block_group->lock); + iput(inode); + last = block_group->key.objectid + block_group->key.offset; + btrfs_put_block_group(block_group); + } +} + int btrfs_free_block_groups(struct btrfs_fs_info *info) { struct btrfs_block_group_cache *block_group; @@ -7897,6 +8069,8 @@ int btrfs_read_block_groups(struct btrfs_root *root) struct btrfs_key key; struct btrfs_key found_key; struct extent_buffer *leaf; + int need_clear = 0; + u64 cache_gen; root = info->extent_root; key.objectid = 0; @@ -7906,6 +8080,11 @@ int btrfs_read_block_groups(struct btrfs_root *root) if (!path) return -ENOMEM; + cache_gen = btrfs_super_cache_generation(&root->fs_info->super_copy); + if (cache_gen != 0 && + btrfs_super_generation(&root->fs_info->super_copy) != cache_gen) + need_clear = 1; + while (1) { ret = find_first_block_group(root, path, &key); if (ret > 0) @@ -7928,6 +8107,9 @@ int btrfs_read_block_groups(struct btrfs_root *root) INIT_LIST_HEAD(&cache->list); INIT_LIST_HEAD(&cache->cluster_list); + if (need_clear) + cache->disk_cache_state = BTRFS_DC_CLEAR; + /* * we only want to have 32k of ram per block group for keeping * track of free space, and if we pass 1/2 of that we want to @@ -8032,6 +8214,7 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans, cache->key.offset = size; cache->key.type = BTRFS_BLOCK_GROUP_ITEM_KEY; cache->sectorsize = root->sectorsize; + cache->fs_info = root->fs_info; /* * we only want to have 32k of ram per block group for keeping track @@ -8088,7 +8271,9 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans, struct btrfs_path *path; struct btrfs_block_group_cache *block_group; struct btrfs_free_cluster *cluster; + struct btrfs_root *tree_root = root->fs_info->tree_root; struct btrfs_key key; + struct inode *inode; int ret; root = root->fs_info->extent_root; @@ -8097,8 +8282,6 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans, BUG_ON(!block_group); BUG_ON(!block_group->ro); - memcpy(&key, &block_group->key, sizeof(key)); - /* make sure this block group isn't part of an allocation cluster */ cluster = &root->fs_info->data_alloc_cluster; spin_lock(&cluster->refill_lock); @@ -8117,6 +8300,40 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans, path = btrfs_alloc_path(); BUG_ON(!path); + inode = lookup_free_space_inode(root, block_group, path); + if (!IS_ERR(inode)) { + btrfs_orphan_add(trans, inode); + clear_nlink(inode); + /* One for the block groups ref */ + spin_lock(&block_group->lock); + if (block_group->iref) { + block_group->iref = 0; + block_group->inode = NULL; + spin_unlock(&block_group->lock); + iput(inode); + } else { + spin_unlock(&block_group->lock); + } + /* One for our lookup ref */ + iput(inode); + } + + key.objectid = BTRFS_FREE_SPACE_OBJECTID; + key.offset = block_group->key.objectid; + key.type = 0; + + ret = btrfs_search_slot(trans, tree_root, &key, path, -1, 1); + if (ret < 0) + goto out; + if (ret > 0) + btrfs_release_path(tree_root, path); + if (ret == 0) { + ret = btrfs_del_item(trans, tree_root, path); + if (ret) + goto out; + btrfs_release_path(tree_root, path); + } + spin_lock(&root->fs_info->block_group_cache_lock); rb_erase(&block_group->cache_node, &root->fs_info->block_group_cache_tree); @@ -8140,6 +8357,8 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans, block_group->space_info->bytes_readonly -= block_group->key.offset; spin_unlock(&block_group->space_info->lock); + memcpy(&key, &block_group->key, sizeof(key)); + btrfs_clear_space_info_full(root->fs_info); btrfs_put_block_group(block_group); -- cgit v1.2.2 From 0cb59c9953171e9adf6da8142a5c85ceb77bb60d Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 2 Jul 2010 12:14:14 -0400 Subject: Btrfs: write out free space cache This is a simple bit, just dump the free space cache out to our preallocated inode when we're writing out dirty block groups. There are a bunch of changes in inode.c in order to account for special cases. Mostly when we're doing the writeout we're holding trans_mutex, so we need to use the nolock transacation functions. Also we can't do asynchronous completions since the async thread could be blocked on already completed IO waiting for the transaction lock. This has been tested with xfstests and btrfs filesystem balance, as well as my ENOSPC tests. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 48 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 48 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index aab40fb3faed..d5455a2bf60b 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -2847,6 +2847,8 @@ again: continue; } + if (cache->disk_cache_state == BTRFS_DC_SETUP) + cache->disk_cache_state = BTRFS_DC_NEED_WRITE; cache->dirty = 0; last = cache->key.objectid + cache->key.offset; @@ -2855,6 +2857,52 @@ again: btrfs_put_block_group(cache); } + while (1) { + /* + * I don't think this is needed since we're just marking our + * preallocated extent as written, but just in case it can't + * hurt. + */ + if (last == 0) { + err = btrfs_run_delayed_refs(trans, root, + (unsigned long)-1); + BUG_ON(err); + } + + cache = btrfs_lookup_first_block_group(root->fs_info, last); + while (cache) { + /* + * Really this shouldn't happen, but it could if we + * couldn't write the entire preallocated extent and + * splitting the extent resulted in a new block. + */ + if (cache->dirty) { + btrfs_put_block_group(cache); + goto again; + } + if (cache->disk_cache_state == BTRFS_DC_NEED_WRITE) + break; + cache = next_block_group(root, cache); + } + if (!cache) { + if (last == 0) + break; + last = 0; + continue; + } + + btrfs_write_out_cache(root, trans, cache, path); + + /* + * If we didn't have an error then the cache state is still + * NEED_WRITE, so we can set it to WRITTEN. + */ + if (cache->disk_cache_state == BTRFS_DC_NEED_WRITE) + cache->disk_cache_state = BTRFS_DC_WRITTEN; + last = cache->key.objectid + cache->key.offset; + btrfs_put_block_group(cache); + } + btrfs_free_path(path); return 0; } -- cgit v1.2.2 From 9d66e233c7042da27ec699453770f41e567a0442 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 25 Aug 2010 16:54:15 -0400 Subject: Btrfs: load free space cache if it exists This patch actually loads the free space cache if it exists. The only thing that really changes here is that we need to cache the block group if we're going to remove an extent from it. Previously we did not do this since the caching kthread would pick it up. With the on disk cache we don't have this luxury so we need to make sure we read the on disk cache in first, and then remove the extent, that way when the extent is unpinned the free space is added to the block group. This has been tested with all sorts of things. Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 50 +++++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 47 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 d5455a2bf60b..9a325e465ad9 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -421,7 +421,9 @@ err: return 0; } -static int cache_block_group(struct btrfs_block_group_cache *cache) +static int cache_block_group(struct btrfs_block_group_cache *cache, + struct btrfs_trans_handle *trans, + int load_cache_only) { struct btrfs_fs_info *fs_info = cache->fs_info; struct btrfs_caching_control *caching_ctl; @@ -432,6 +434,36 @@ static int cache_block_group(struct btrfs_block_group_cache *cache) if (cache->cached != BTRFS_CACHE_NO) return 0; + /* + * We can't do the read from on-disk cache during a commit since we need + * to have the normal tree locking. + */ + if (!trans->transaction->in_commit) { + spin_lock(&cache->lock); + if (cache->cached != BTRFS_CACHE_NO) { + spin_unlock(&cache->lock); + return 0; + } + cache->cached = BTRFS_CACHE_STARTED; + spin_unlock(&cache->lock); + + ret = load_free_space_cache(fs_info, cache); + + spin_lock(&cache->lock); + if (ret == 1) { + cache->cached = BTRFS_CACHE_FINISHED; + cache->last_byte_to_unpin = (u64)-1; + } else { + cache->cached = BTRFS_CACHE_NO; + } + spin_unlock(&cache->lock); + if (ret == 1) + return 0; + } + + if (load_cache_only) + return 0; + caching_ctl = kzalloc(sizeof(*caching_ctl), GFP_KERNEL); BUG_ON(!caching_ctl); @@ -3984,6 +4016,14 @@ static int update_block_group(struct btrfs_trans_handle *trans, factor = 2; else factor = 1; + /* + * If this block group has free space cache written out, we + * need to make sure to load it if we are removing space. This + * is because we need the unpinning stage to actually add the + * space back to the block group, otherwise we will leak space. + */ + if (!alloc && cache->cached == BTRFS_CACHE_NO) + cache_block_group(cache, trans, 1); byte_in_group = bytenr - cache->key.objectid; WARN_ON(byte_in_group > cache->key.offset); @@ -4828,6 +4868,10 @@ have_block_group: if (unlikely(block_group->cached == BTRFS_CACHE_NO)) { u64 free_percent; + ret = cache_block_group(block_group, trans, 1); + if (block_group->cached == BTRFS_CACHE_FINISHED) + goto have_block_group; + free_percent = btrfs_block_group_used(&block_group->item); free_percent *= 100; free_percent = div64_u64(free_percent, @@ -4848,7 +4892,7 @@ 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); + ret = cache_block_group(block_group, trans, 0); BUG_ON(ret); } found_uncached_bg = true; @@ -5405,7 +5449,7 @@ 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); + cache_block_group(block_group, trans, 0); caching_ctl = get_caching_control(block_group); if (!caching_ctl) { -- cgit v1.2.2 From dde5abee12327d59f968bbfc8151e1b04082a2c4 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Thu, 16 Sep 2010 16:17:03 -0400 Subject: Btrfs: check cache->caching_ctl before returning if caching has started With the free space disk caching we can mark the block group as started with the caching, but we don't have a caching ctl. This can race with anybody else who tries to get the caching ctl before we cache (this is very hard to do btw). So instead check to see if cache->caching_ctl is set, and if not return NULL. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 9a325e465ad9..5c9ef3ac25e1 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -242,6 +242,12 @@ get_caching_control(struct btrfs_block_group_cache *cache) return NULL; } + /* We're loading it the fast way, so we don't have a caching_ctl. */ + if (!cache->caching_ctl) { + spin_unlock(&cache->lock); + return NULL; + } + ctl = cache->caching_ctl; atomic_inc(&ctl->count); spin_unlock(&cache->lock); -- cgit v1.2.2 From 67377734fd24c32cbdfeb697c2e2bd7fed519e75 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Thu, 16 Sep 2010 16:19:09 -0400 Subject: Btrfs: add support for mixed data+metadata block groups There are just a few things that need to be fixed in the kernel to support mixed data+metadata block groups. Mostly we just need to make sure that if we are using mixed block groups that we continue to allocate mixed block groups as we need them. Also we need to make sure __find_space_info will find our space info if we search for DATA or METADATA only. Tested this with xfstests and it works nicely. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 22 +++++++++++++++++++--- 1 file changed, 19 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 5c9ef3ac25e1..137833e1fc26 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -547,7 +547,7 @@ static struct btrfs_space_info *__find_space_info(struct btrfs_fs_info *info, rcu_read_lock(); list_for_each_entry_rcu(found, head, list) { - if (found->flags == flags) { + if (found->flags & flags) { rcu_read_unlock(); return found; } @@ -3266,6 +3266,13 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans, } spin_unlock(&space_info->lock); + /* + * If we have mixed data/metadata chunks we want to make sure we keep + * allocating mixed chunks instead of individual chunks. + */ + if (btrfs_mixed_space_info(space_info)) + flags |= (BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA); + /* * if we're doing a data chunk, go ahead and make sure that * we keep a reasonable number of metadata chunks allocated in the @@ -4787,6 +4794,7 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans, bool found_uncached_bg = false; bool failed_cluster_refill = false; bool failed_alloc = false; + bool use_cluster = true; u64 ideal_cache_percent = 0; u64 ideal_cache_offset = 0; @@ -4801,16 +4809,24 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans, return -ENOSPC; } + /* + * If the space info is for both data and metadata it means we have a + * small filesystem and we can't use the clustering stuff. + */ + if (btrfs_mixed_space_info(space_info)) + use_cluster = false; + if (orig_root->ref_cows || empty_size) allowed_chunk_alloc = 1; - if (data & BTRFS_BLOCK_GROUP_METADATA) { + if (data & BTRFS_BLOCK_GROUP_METADATA && use_cluster) { last_ptr = &root->fs_info->meta_alloc_cluster; if (!btrfs_test_opt(root, SSD)) empty_cluster = 64 * 1024; } - if ((data & BTRFS_BLOCK_GROUP_DATA) && btrfs_test_opt(root, SSD)) { + if ((data & BTRFS_BLOCK_GROUP_DATA) && use_cluster && + btrfs_test_opt(root, SSD)) { last_ptr = &root->fs_info->data_alloc_cluster; } -- cgit v1.2.2 From 88c2ba3b069f1e0f4694124d02985fa7620a19f1 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 21 Sep 2010 14:21:34 -0400 Subject: Btrfs: Add a clear_cache mount option If something goes wrong with the free space cache we need a way to make sure it's not loaded on mount and that it's cleared for everybody. When you pass the clear_cache option it will make it so all block groups are setup to be cleared, which keeps them from being loaded and then they will be truncated when the transaction is committed. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 137833e1fc26..1a94ee4c4fbb 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -8198,6 +8198,8 @@ int btrfs_read_block_groups(struct btrfs_root *root) if (cache_gen != 0 && btrfs_super_generation(&root->fs_info->super_copy) != cache_gen) need_clear = 1; + if (btrfs_test_opt(root, CLEAR_CACHE)) + need_clear = 1; while (1) { ret = find_first_block_group(root, path, &key); -- cgit v1.2.2 From 8216ef866df1119fd5a72372b8b29bce49c18590 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Thu, 28 Oct 2010 16:55:47 -0400 Subject: Btrfs: let the user know space caching is enabled If you mount -o space_cache, the option will be persistent across mounts, but to make sure the user knows that they did this, emit a message telling them if they didn't mount with -o space_cache but the feature is still used. Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 1a94ee4c4fbb..d2a7ff53e99a 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -8200,6 +8200,8 @@ int btrfs_read_block_groups(struct btrfs_root *root) need_clear = 1; if (btrfs_test_opt(root, CLEAR_CACHE)) need_clear = 1; + if (!btrfs_test_opt(root, SPACE_CACHE) && cache_gen) + printk(KERN_INFO "btrfs: disk space caching is enabled\n"); while (1) { ret = find_first_block_group(root, path, &key); -- cgit v1.2.2 From e5bc2458293b2af6c0b94435965c68cc70974b56 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Tue, 26 Oct 2010 13:37:56 -0400 Subject: Btrfs: tune the chunk allocation to 5% of the FS as metadata An earlier commit tried to keep us from allocating too many empty metadata chunks. It was somewhat too restrictive and could lead to ENOSPC errors on empty filesystems. This increases the limits to about 5% of the FS size, allowing more metadata chunks to be preallocated. Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 22 ++++++++++++++++++---- 1 file changed, 18 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 372fd224a11d..980d6a3c342c 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -580,6 +580,15 @@ static u64 div_factor(u64 num, int factor) return num; } +static u64 div_factor_fine(u64 num, int factor) +{ + if (factor == 100) + return num; + num *= factor; + do_div(num, 100); + return num; +} + u64 btrfs_find_block_group(struct btrfs_root *root, u64 search_start, u64 search_hint, int owner) { @@ -3218,9 +3227,11 @@ static void force_metadata_allocation(struct btrfs_fs_info *info) rcu_read_unlock(); } -static int should_alloc_chunk(struct btrfs_space_info *sinfo, u64 alloc_bytes) +static int should_alloc_chunk(struct btrfs_root *root, + struct btrfs_space_info *sinfo, u64 alloc_bytes) { u64 num_bytes = sinfo->total_bytes - sinfo->bytes_readonly; + u64 thresh; if (sinfo->bytes_used + sinfo->bytes_reserved + alloc_bytes + 256 * 1024 * 1024 < num_bytes) @@ -3230,8 +3241,10 @@ static int should_alloc_chunk(struct btrfs_space_info *sinfo, u64 alloc_bytes) alloc_bytes < div_factor(num_bytes, 8)) return 0; - if (num_bytes > 256 * 1024 * 1024 && - sinfo->bytes_used < div_factor(num_bytes, 3)) + thresh = btrfs_super_total_bytes(&root->fs_info->super_copy); + thresh = max_t(u64, 256 * 1024 * 1024, div_factor_fine(thresh, 5)); + + if (num_bytes > thresh && sinfo->bytes_used < div_factor(num_bytes, 3)) return 0; return 1; @@ -3265,7 +3278,8 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans, goto out; } - if (!force && !should_alloc_chunk(space_info, alloc_bytes)) { + if (!force && !should_alloc_chunk(extent_root, space_info, + alloc_bytes)) { spin_unlock(&space_info->lock); goto out; } -- cgit v1.2.2 From bf9022e06af553553bc8f4e21ce36147ca6eae68 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Tue, 26 Oct 2010 13:40:45 -0400 Subject: Btrfs: use the flusher threads for delalloc throttling We have a fairly complex set of loops around walking our list of delalloc inodes when we find metadata delalloc space running low. It doesn't work very well, can use large amounts of CPU and doesn't do very efficient writeback. This switches us to kick the bdi flusher threads instead. All dirty data in btrfs is accounted as delalloc data, so this is very similar in terms of what it writes, but we're able to just kick off the IO and wait for progress. Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 33 +++++++++++++++------------------ 1 file changed, 15 insertions(+), 18 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 980d6a3c342c..59c8daaacf0c 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3328,15 +3328,14 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans, u64 reserved; u64 max_reclaim; u64 reclaimed = 0; - int no_reclaim = 0; int pause = 1; - int ret; + int nr_pages = (2 * 1024 * 1024) >> PAGE_CACHE_SHIFT; block_rsv = &root->fs_info->delalloc_block_rsv; space_info = block_rsv->space_info; - spin_lock(&space_info->lock); + + smp_mb(); reserved = space_info->bytes_reserved; - spin_unlock(&space_info->lock); if (reserved == 0) return 0; @@ -3344,20 +3343,11 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans, max_reclaim = min(reserved, to_reclaim); while (1) { - ret = btrfs_start_one_delalloc_inode(root, trans ? 1 : 0, sync); - if (!ret) { - if (no_reclaim > 2) - break; - no_reclaim++; - __set_current_state(TASK_INTERRUPTIBLE); - schedule_timeout(pause); - pause <<= 1; - if (pause > HZ / 10) - pause = HZ / 10; - } else { - no_reclaim = 0; - pause = 1; - } + /* have the flusher threads jump in and do some IO */ + smp_mb(); + nr_pages = min_t(unsigned long, nr_pages, + root->fs_info->delalloc_bytes >> PAGE_CACHE_SHIFT); + writeback_inodes_sb_nr_if_idle(root->fs_info->sb, nr_pages); spin_lock(&space_info->lock); if (reserved > space_info->bytes_reserved) @@ -3370,6 +3360,13 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans, if (trans && trans->transaction->blocked) return -EAGAIN; + + __set_current_state(TASK_INTERRUPTIBLE); + schedule_timeout(pause); + pause <<= 1; + if (pause > HZ / 10) + pause = HZ / 10; + } return reclaimed >= to_reclaim; } -- cgit v1.2.2 From 559af8211433b8c0b20e6c43c61409cb9c9c2996 Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Fri, 29 Oct 2010 15:14:37 -0400 Subject: Btrfs: cleanup warnings from gcc 4.6 (nonbugs) These are all the cases where a variable is set, but not read which are not bugs as far as I can see, but simply leftovers. Still needs more review. Found by gcc 4.6's new warnings Signed-off-by: Andi Kleen Cc: Chris Mason Signed-off-by: Andrew Morton Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 59c8daaacf0c..df754108b952 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5719,7 +5719,6 @@ static noinline void reada_walk_down(struct btrfs_trans_handle *trans, u64 generation; u64 refs; u64 flags; - u64 last = 0; u32 nritems; u32 blocksize; struct btrfs_key key; @@ -5787,7 +5786,6 @@ reada: generation); if (ret) break; - last = bytenr + blocksize; nread++; } wc->reada_slot = slot; -- cgit v1.2.2 From d8e39c457bc1ca2a7304bc086c7b0f0c10854921 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Fri, 29 Oct 2010 15:17:41 -0400 Subject: Btrfs: drop unused variable in block_alloc_rsv The alloc_target variable is not really used. Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 4 ---- 1 file changed, 4 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index df754108b952..a541bc87f04c 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3607,18 +3607,14 @@ struct btrfs_block_rsv *btrfs_alloc_block_rsv(struct btrfs_root *root) { struct btrfs_block_rsv *block_rsv; struct btrfs_fs_info *fs_info = root->fs_info; - u64 alloc_target; block_rsv = kmalloc(sizeof(*block_rsv), GFP_NOFS); if (!block_rsv) return NULL; btrfs_init_block_rsv(block_rsv); - - alloc_target = btrfs_get_alloc_profile(root, 0); block_rsv->space_info = __find_space_info(fs_info, BTRFS_BLOCK_GROUP_METADATA); - return block_rsv; } -- cgit v1.2.2 From 6f33434850ed87dc5e56b60ebbad3d3cf405f296 Mon Sep 17 00:00:00 2001 From: Arne Jansen Date: Fri, 12 Nov 2010 23:17:56 +0000 Subject: btrfs: Fix early enospc because 'unused' calculated with wrong sign. 'unused' calculated with wrong sign in reserve_metadata_bytes(). This might have lead to unwanted over-reservations. Signed-off-by: Arne Jansen Reviewed-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index a541bc87f04c..ddaf6340fe7f 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3413,7 +3413,7 @@ again: * our reservation. */ if (unused <= space_info->total_bytes) { - unused -= space_info->total_bytes; + unused = space_info->total_bytes - unused; if (unused >= num_bytes) { if (!reserved) space_info->bytes_reserved += orig_bytes; -- cgit v1.2.2 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 ddaf6340fe7f..8c56f5b38948 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 v1.2.2 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 8c56f5b38948..cec05e100142 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 v1.2.2 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 cec05e100142..41133b064d72 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 v1.2.2 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 +++++++++++++++-- 1 file changed, 15 insertions(+), 2 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 41133b064d72..4be231e0d2bd 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; -- cgit v1.2.2 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 4be231e0d2bd..7e5162e5c411 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 v1.2.2 From 65e5341b9a0c39767ae1fecc727d70eda0dd6d83 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Fri, 24 Dec 2010 06:41:52 -0500 Subject: Btrfs: fix off by one while setting block groups readonly When we read in block groups, we'll set non-redundant groups readonly if we find a raid1, DUP or raid10 group. But the ro code has an off by one bug in the math around testing to make sure out accounting doesn't go wrong. Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 3 ++- 1 file changed, 2 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 7e5162e5c411..b180efdc8b68 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -7971,13 +7971,14 @@ static int set_block_group_ro(struct btrfs_block_group_cache *cache) if (sinfo->bytes_used + sinfo->bytes_reserved + sinfo->bytes_pinned + sinfo->bytes_may_use + sinfo->bytes_readonly + - cache->reserved_pinned + num_bytes < sinfo->total_bytes) { + cache->reserved_pinned + num_bytes <= sinfo->total_bytes) { sinfo->bytes_readonly += num_bytes; sinfo->bytes_reserved += cache->reserved_pinned; cache->reserved_pinned = 0; cache->ro = 1; ret = 0; } + spin_unlock(&cache->lock); spin_unlock(&sinfo->lock); return ret; -- cgit v1.2.2 From d52a5b5f1fa40804f681cf9868d4a8f90661bdf3 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Wed, 5 Jan 2011 10:07:18 +0000 Subject: btrfs: try to reclaim some space when chunk allocation fails We cannot write data into files when when there is tiny space in the filesystem. Reproduce steps: # mkfs.btrfs /dev/sda1 # mount /dev/sda1 /mnt # dd if=/dev/zero of=/mnt/tmpfile0 bs=4K count=1 # dd if=/dev/zero of=/mnt/tmpfile1 bs=4K count=99999999999999 (fill the filesystem) # umount /mnt # mount /dev/sda1 /mnt # rm -f /mnt/tmpfile0 # dd if=/dev/zero of=/mnt/tmpfile0 bs=4K count=1 (failed with nospec) But if we do the last step again, we can write data successfully. The reason of the problem is that btrfs didn't try to commit the current transaction and reclaim some space when chunk allocation failed. This patch fixes it by committing the current transaction to reclaim some space when chunk allocation fails. Signed-off-by: Miao Xie Reviewed-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index b180efdc8b68..3c71d95111fe 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3162,8 +3162,12 @@ alloc: bytes + 2 * 1024 * 1024, alloc_target, 0); btrfs_end_transaction(trans, root); - if (ret < 0) - return ret; + if (ret < 0) { + if (ret != -ENOSPC) + return ret; + else + goto commit_trans; + } if (!data_sinfo) { btrfs_set_inode_space_info(root, inode); @@ -3174,6 +3178,7 @@ alloc: spin_unlock(&data_sinfo->lock); /* commit the current transaction and try again */ +commit_trans: if (!committed && !root->fs_info->open_ioctl_trans) { committed = 1; trans = btrfs_join_transaction(root, 1); -- cgit v1.2.2 From 7bfc837df935d850fe996dfe92ef48975cd4170a Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Wed, 5 Jan 2011 10:07:26 +0000 Subject: btrfs: restructure find_free_dev_extent() - make it return the start position and length of the max free space when it can not find a suitable free space. - make it more readability Signed-off-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 3c71d95111fe..1e1c9a177626 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -8099,7 +8099,7 @@ int btrfs_can_relocate(struct btrfs_root *root, u64 bytenr) mutex_lock(&root->fs_info->chunk_mutex); list_for_each_entry(device, &fs_devices->alloc_list, dev_alloc_list) { u64 min_free = btrfs_block_group_used(&block_group->item); - u64 dev_offset, max_avail; + u64 dev_offset; /* * check to make sure we can actually find a chunk with enough @@ -8107,7 +8107,7 @@ int btrfs_can_relocate(struct btrfs_root *root, u64 bytenr) */ if (device->total_bytes > device->bytes_used + min_free) { ret = find_free_dev_extent(NULL, device, min_free, - &dev_offset, &max_avail); + &dev_offset, NULL); if (!ret) break; ret = -1; -- cgit v1.2.2 From 6d07bcec969af335d4e35b3921131b7929bd634e Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Wed, 5 Jan 2011 10:07:31 +0000 Subject: btrfs: fix wrong free space information of btrfs When we store data by raid profile in btrfs with two or more different size disks, df command shows there is some free space in the filesystem, but the user can not write any data in fact, df command shows the wrong free space information of btrfs. # mkfs.btrfs -d raid1 /dev/sda9 /dev/sda10 # btrfs-show Label: none uuid: a95cd49e-6e33-45b8-8741-a36153ce4b64 Total devices 2 FS bytes used 28.00KB devid 1 size 5.01GB used 2.03GB path /dev/sda9 devid 2 size 10.00GB used 2.01GB path /dev/sda10 # btrfs device scan /dev/sda9 /dev/sda10 # mount /dev/sda9 /mnt # dd if=/dev/zero of=tmpfile0 bs=4K count=9999999999 (fill the filesystem) # sync # df -TH Filesystem Type Size Used Avail Use% Mounted on /dev/sda9 btrfs 17G 8.6G 5.4G 62% /mnt # btrfs-show Label: none uuid: a95cd49e-6e33-45b8-8741-a36153ce4b64 Total devices 2 FS bytes used 3.99GB devid 1 size 5.01GB used 5.01GB path /dev/sda9 devid 2 size 10.00GB used 4.99GB path /dev/sda10 It is because btrfs cannot allocate chunks when one of the pairing disks has no space, the free space on the other disks can not be used for ever, and should be subtracted from the total space, but btrfs doesn't subtract this space from the total. It is strange to the user. This patch fixes it by calcing the free space that can be used to allocate chunks. Implementation: 1. get all the devices free space, and align them by stripe length. 2. sort the devices by the free space. 3. check the free space of the devices, 3.1. if it is not zero, and then check the number of the devices that has more free space than this device, if the number of the devices is beyond the min stripe number, the free space can be used, and add into total free space. if the number of the devices is below the min stripe number, we can not use the free space, the check ends. 3.2. if the free space is zero, check the next devices, goto 3.1 This implementation is just likely fake chunk allocation. After appling this patch, df can show correct space information: # df -TH Filesystem Type Size Used Avail Use% Mounted on /dev/sda9 btrfs 17G 8.6G 0 100% /mnt Signed-off-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 58 +++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 57 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 1e1c9a177626..04bfc3a2bd9f 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3090,7 +3090,7 @@ static u64 get_alloc_profile(struct btrfs_root *root, u64 flags) return btrfs_reduce_alloc_profile(root, flags); } -static u64 btrfs_get_alloc_profile(struct btrfs_root *root, int data) +u64 btrfs_get_alloc_profile(struct btrfs_root *root, int data) { u64 flags; @@ -8019,6 +8019,62 @@ out: return ret; } +/* + * helper to account the unused space of all the readonly block group in the + * list. takes mirrors into account. + */ +static u64 __btrfs_get_ro_block_group_free_space(struct list_head *groups_list) +{ + struct btrfs_block_group_cache *block_group; + u64 free_bytes = 0; + int factor; + + list_for_each_entry(block_group, groups_list, list) { + spin_lock(&block_group->lock); + + if (!block_group->ro) { + spin_unlock(&block_group->lock); + continue; + } + + if (block_group->flags & (BTRFS_BLOCK_GROUP_RAID1 | + BTRFS_BLOCK_GROUP_RAID10 | + BTRFS_BLOCK_GROUP_DUP)) + factor = 2; + else + factor = 1; + + free_bytes += (block_group->key.offset - + btrfs_block_group_used(&block_group->item)) * + factor; + + spin_unlock(&block_group->lock); + } + + return free_bytes; +} + +/* + * helper to account the unused space of all the readonly block group in the + * space_info. takes mirrors into account. + */ +u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo) +{ + int i; + u64 free_bytes = 0; + + spin_lock(&sinfo->lock); + + for(i = 0; i < BTRFS_NR_RAID_TYPES; i++) + if (!list_empty(&sinfo->block_groups[i])) + free_bytes += __btrfs_get_ro_block_group_free_space( + &sinfo->block_groups[i]); + + spin_unlock(&sinfo->lock); + + return free_bytes; +} + int btrfs_set_block_group_rw(struct btrfs_root *root, struct btrfs_block_group_cache *cache) { -- cgit v1.2.2 From f690efb1aa2a961dd6655529c1797fcac60ad6d9 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 12 Jan 2011 21:04:22 +0000 Subject: Btrfs: don't warn if we get ENOSPC in btrfs_block_rsv_check If we run low on space we could get a bunch of warnings out of btrfs_block_rsv_check, but this is mostly just called via the transaction code to see if we need to end the transaction, it expects to see failures, so let's not WARN and freak everybody out for no reason. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 5 ----- 1 file changed, 5 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 04bfc3a2bd9f..055b837eab19 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3727,11 +3727,6 @@ int btrfs_block_rsv_check(struct btrfs_trans_handle *trans, return 0; } - WARN_ON(1); - printk(KERN_INFO"block_rsv size %llu reserved %llu freed %llu %llu\n", - block_rsv->size, block_rsv->reserved, - block_rsv->freed[0], block_rsv->freed[1]); - return -ENOSPC; } -- cgit v1.2.2 From acce952b0263825da32cf10489413dec78053347 Mon Sep 17 00:00:00 2001 From: liubo Date: Thu, 6 Jan 2011 19:30:25 +0800 Subject: Btrfs: forced readonly mounts on errors This patch comes from "Forced readonly mounts on errors" ideas. As we know, this is the first step in being more fault tolerant of disk corruptions instead of just using BUG() statements. The major content: - add a framework for generating errors that should result in filesystems going readonly. - keep FS state in disk super block. - make sure that all of resource will be freed and released at umount time. - make sure that fter FS is forced readonly on error, there will be no more disk change before FS is corrected. For this, we should stop write operation. After this patch is applied, the conversion from BUG() to such a framework can happen incrementally. Signed-off-by: Liu Bo Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 11 +++++++++++ 1 file changed, 11 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 055b837eab19..bcf303204f7f 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -8642,3 +8642,14 @@ out: btrfs_free_path(path); return ret; } + +int btrfs_error_unpin_extent_range(struct btrfs_root *root, u64 start, u64 end) +{ + return unpin_extent_range(root, start, end); +} + +int btrfs_error_discard_extent(struct btrfs_root *root, u64 bytenr, + u64 num_bytes) +{ + return btrfs_discard_extent(root, bytenr, num_bytes); +} -- cgit v1.2.2 From 3612b49598c303cfb22a4b609427f829828e2427 Mon Sep 17 00:00:00 2001 From: Tsutomu Itoh Date: Tue, 25 Jan 2011 02:51:38 +0000 Subject: btrfs: fix return value check of btrfs_join_transaction() The error check of btrfs_join_transaction()/btrfs_join_transaction_nolock() is added, and the mistake of the error check in several places is corrected. For more stable Btrfs, I think that we should reduce BUG_ON(). But, I think that long time is necessary for this. So, I propose this patch as a short-term solution. With this patch: - To more stable Btrfs, the part that should be corrected is clarified. - The panic isn't done by the NULL pointer reference etc. (even if BUG_ON() is increased temporarily) - The error code is returned in the place where the error can be easily returned. As a long-term plan: - BUG_ON() is reduced by using the forced-readonly framework, etc. Signed-off-by: Tsutomu Itoh Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index bcf303204f7f..98ee139885cc 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -7478,7 +7478,7 @@ int btrfs_drop_dead_reloc_roots(struct btrfs_root *root) BUG_ON(reloc_root->commit_root != NULL); while (1) { trans = btrfs_join_transaction(root, 1); - BUG_ON(!trans); + BUG_ON(IS_ERR(trans)); mutex_lock(&root->fs_info->drop_mutex); ret = btrfs_drop_snapshot(trans, reloc_root); -- cgit v1.2.2 From e9e22899de661af94cb9995885fd04e4c738838b Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Mon, 24 Jan 2011 21:43:19 +0000 Subject: Btrfs: do not release more reserved bytes to the global_block_rsv than we need When we do btrfs_block_rsv_release, if global_block_rsv is not full we will release all the extra bytes to global_block_rsv, even if it's only a little short of the amount of space that we need to reserve. This causes us to starve ourselves of reservable space during the transaction which will force us to shrink delalloc bytes and commit the transaction more often than we should. So instead just add the amount of bytes we need to add to the global reserve so reserved == size, and then add the rest back into the space_info for general use. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 16 ++++++++++++++-- 1 file changed, 14 insertions(+), 2 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 98ee139885cc..7af618dcf2c0 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3589,8 +3589,20 @@ void block_rsv_release_bytes(struct btrfs_block_rsv *block_rsv, if (num_bytes > 0) { if (dest) { - block_rsv_add_bytes(dest, num_bytes, 0); - } else { + spin_lock(&dest->lock); + if (!dest->full) { + u64 bytes_to_add; + + bytes_to_add = dest->size - dest->reserved; + bytes_to_add = min(num_bytes, bytes_to_add); + dest->reserved += bytes_to_add; + if (dest->reserved >= dest->size) + dest->full = 1; + num_bytes -= bytes_to_add; + } + spin_unlock(&dest->lock); + } + if (num_bytes) { spin_lock(&space_info->lock); space_info->bytes_reserved -= num_bytes; spin_unlock(&space_info->lock); -- cgit v1.2.2 From 68a82277b8619e6d0f2738b1d9b160b627e81e92 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Mon, 24 Jan 2011 21:43:20 +0000 Subject: Btrfs: use the global block reserve if we cannot reserve space We call use_block_rsv right before we make an allocation in order to make sure we have enough space. Now normally people have called btrfs_start_transaction() with the appropriate amount of space that we need, so we just use some of that pre-reserved space and move along happily. The problem is where people use btrfs_join_transaction(), which doesn't actually reserve any space. So we try and reserve space here, but we cannot flush delalloc, so this forces us to return -ENOSPC when in reality we have plenty of space. The most common symptom is seeing a bunch of "couldn't dirty inode" messages in syslog. With xfstests 224 we end up falling back to start_transaction and then doing all the flush delalloc stuff which causes to hang for a very long time. So instead steal from the global reserve, which is what this is meant for anyway. With this patch and the other 2 I have sent xfstests 224 now passes successfully. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 28 +++++++++++++++++++++++++++- 1 file changed, 27 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 7af618dcf2c0..ff6bbfd75cf7 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5646,6 +5646,7 @@ use_block_rsv(struct btrfs_trans_handle *trans, struct btrfs_root *root, u32 blocksize) { struct btrfs_block_rsv *block_rsv; + struct btrfs_block_rsv *global_rsv = &root->fs_info->global_block_rsv; int ret; block_rsv = get_block_rsv(trans, root); @@ -5653,14 +5654,39 @@ use_block_rsv(struct btrfs_trans_handle *trans, if (block_rsv->size == 0) { ret = reserve_metadata_bytes(trans, root, block_rsv, blocksize, 0); - if (ret) + /* + * If we couldn't reserve metadata bytes try and use some from + * the global reserve. + */ + if (ret && block_rsv != global_rsv) { + ret = block_rsv_use_bytes(global_rsv, blocksize); + if (!ret) + return global_rsv; + return ERR_PTR(ret); + } else if (ret) { return ERR_PTR(ret); + } return block_rsv; } ret = block_rsv_use_bytes(block_rsv, blocksize); if (!ret) return block_rsv; + if (ret) { + WARN_ON(1); + ret = reserve_metadata_bytes(trans, root, block_rsv, blocksize, + 0); + if (!ret) { + spin_lock(&block_rsv->lock); + block_rsv->size += blocksize; + spin_unlock(&block_rsv->lock); + return block_rsv; + } else if (ret && block_rsv != global_rsv) { + ret = block_rsv_use_bytes(global_rsv, blocksize); + if (!ret) + return global_rsv; + } + } return ERR_PTR(-ENOSPC); } -- cgit v1.2.2 From b1953bcec95c189b1eea690a08e89646d7750bda Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 21 Jan 2011 21:10:01 +0000 Subject: Btrfs: make shrink_delalloc a little friendlier Xfstests 224 will just sit there and spin for ever until eventually we give up flushing delalloc and exit. On my box this took several hours. I could not interrupt this process either, even though we use INTERRUPTIBLE. So do 2 things 1) Keep us from looping over and over again without reclaiming anything 2) If we get interrupted exit the loop I tested this and the test now exits in a reasonable amount of time, and can be interrupted with ctrl+c. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 17 ++++++++++++++--- 1 file changed, 14 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 ff6bbfd75cf7..f96641a93fc9 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3345,8 +3345,10 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans, u64 reserved; u64 max_reclaim; u64 reclaimed = 0; + long time_left; int pause = 1; int nr_pages = (2 * 1024 * 1024) >> PAGE_CACHE_SHIFT; + int loops = 0; block_rsv = &root->fs_info->delalloc_block_rsv; space_info = block_rsv->space_info; @@ -3359,7 +3361,7 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans, max_reclaim = min(reserved, to_reclaim); - while (1) { + while (loops < 1024) { /* have the flusher threads jump in and do some IO */ smp_mb(); nr_pages = min_t(unsigned long, nr_pages, @@ -3367,8 +3369,12 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans, writeback_inodes_sb_nr_if_idle(root->fs_info->sb, nr_pages); spin_lock(&space_info->lock); - if (reserved > space_info->bytes_reserved) + if (reserved > space_info->bytes_reserved) { + loops = 0; reclaimed += reserved - space_info->bytes_reserved; + } else { + loops++; + } reserved = space_info->bytes_reserved; spin_unlock(&space_info->lock); @@ -3379,7 +3385,12 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans, return -EAGAIN; __set_current_state(TASK_INTERRUPTIBLE); - schedule_timeout(pause); + time_left = schedule_timeout(pause); + + /* We were interrupted, exit */ + if (time_left) + break; + pause <<= 1; if (pause > HZ / 10) pause = HZ / 10; -- cgit v1.2.2 From 5df67083488ccbad925f583b698ab38f8629a016 Mon Sep 17 00:00:00 2001 From: Tsutomu Itoh Date: Tue, 1 Feb 2011 09:17:35 +0000 Subject: btrfs: checking NULL or not in some functions Because NULL is returned when the memory allocation fails, it is checked whether it is NULL. Signed-off-by: Tsutomu Itoh Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index f96641a93fc9..9de4ff03882a 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -6496,6 +6496,8 @@ static noinline int relocate_inode_pages(struct inode *inode, u64 start, int ret = 0; ra = kzalloc(sizeof(*ra), GFP_NOFS); + if (!ra) + return -ENOMEM; mutex_lock(&inode->i_mutex); first_index = start >> PAGE_CACHE_SHIFT; -- cgit v1.2.2 From 98d5dc13e7e74b77ca3b4c3cbded9f48d2dbbbb7 Mon Sep 17 00:00:00 2001 From: Tsutomu Itoh Date: Thu, 20 Jan 2011 06:19:37 +0000 Subject: btrfs: fix return value check of btrfs_start_transaction() The error check of btrfs_start_transaction() is added, and the mistake of the error check on several places is corrected. Signed-off-by: Tsutomu Itoh Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 9de4ff03882a..f07ba21cbf06 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -6271,6 +6271,8 @@ int btrfs_drop_snapshot(struct btrfs_root *root, BUG_ON(!wc); trans = btrfs_start_transaction(tree_root, 0); + BUG_ON(IS_ERR(trans)); + if (block_rsv) trans->block_rsv = block_rsv; @@ -6368,6 +6370,7 @@ int btrfs_drop_snapshot(struct btrfs_root *root, btrfs_end_transaction_throttle(trans, tree_root); trans = btrfs_start_transaction(tree_root, 0); + BUG_ON(IS_ERR(trans)); if (block_rsv) trans->block_rsv = block_rsv; } @@ -7587,7 +7590,7 @@ int btrfs_cleanup_reloc_trees(struct btrfs_root *root) if (found) { trans = btrfs_start_transaction(root, 1); - BUG_ON(!trans); + BUG_ON(IS_ERR(trans)); ret = btrfs_commit_transaction(trans, root); BUG_ON(ret); } @@ -7831,7 +7834,7 @@ static noinline int relocate_one_extent(struct btrfs_root *extent_root, trans = btrfs_start_transaction(extent_root, 1); - BUG_ON(!trans); + BUG_ON(IS_ERR(trans)); if (extent_key->objectid == 0) { ret = del_extent_zero(trans, extent_root, path, extent_key); -- cgit v1.2.2 From 3c14874acc71180553fb5aba528e3cf57c5b958b Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 2 Feb 2011 15:53:47 +0000 Subject: Btrfs: exclude super blocks when we read in block groups This has been resulting in a BUT_ON(ret) after btrfs_reserve_extent in btrfs_cow_file_range. The reason is we don't actually calculate the bytes_super for a block group until we go to cache it, which means that the space_info can hand out reservations for space that it doesn't actually have, and we can run out of data space. This is also a problem if you are using space caching since we don't ever calculate bytes_super for the block groups. So instead everytime we read a block group call exclude_super_stripes, which calculates the bytes_super for the block group so it can be left out of the space_info. Then whenever caching completes we just call free_excluded_extents so that the super excluded extents are freed up. Also if we are unmounting and we hit any block groups that haven't been cached we still need to call free_excluded_extents to make sure things are cleaned up properly. Thanks, Reported-by: Arne Jansen Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 26 ++++++++++++++++++-------- 1 file changed, 18 insertions(+), 8 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index f07ba21cbf06..565e22d77b1b 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -320,11 +320,6 @@ static int caching_kthread(void *data) if (!path) return -ENOMEM; - exclude_super_stripes(extent_root, block_group); - spin_lock(&block_group->space_info->lock); - block_group->space_info->bytes_readonly += block_group->bytes_super; - spin_unlock(&block_group->space_info->lock); - last = max_t(u64, block_group->key.objectid, BTRFS_SUPER_INFO_OFFSET); /* @@ -467,8 +462,10 @@ static int cache_block_group(struct btrfs_block_group_cache *cache, cache->cached = BTRFS_CACHE_NO; } spin_unlock(&cache->lock); - if (ret == 1) + if (ret == 1) { + free_excluded_extents(fs_info->extent_root, cache); return 0; + } } if (load_cache_only) @@ -4036,6 +4033,7 @@ void btrfs_delalloc_release_metadata(struct inode *inode, u64 num_bytes) num_bytes = ALIGN(num_bytes, root->sectorsize); atomic_dec(&BTRFS_I(inode)->outstanding_extents); + WARN_ON(atomic_read(&BTRFS_I(inode)->outstanding_extents) < 0); spin_lock(&BTRFS_I(inode)->accounting_lock); nr_extents = atomic_read(&BTRFS_I(inode)->outstanding_extents); @@ -8325,6 +8323,13 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info) if (block_group->cached == BTRFS_CACHE_STARTED) wait_block_group_cache_done(block_group); + /* + * We haven't cached this block group, which means we could + * possibly have excluded extents on this block group. + */ + if (block_group->cached == BTRFS_CACHE_NO) + free_excluded_extents(info->extent_root, block_group); + btrfs_remove_free_space_cache(block_group); btrfs_put_block_group(block_group); @@ -8439,6 +8444,13 @@ int btrfs_read_block_groups(struct btrfs_root *root) cache->flags = btrfs_block_group_flags(&cache->item); cache->sectorsize = root->sectorsize; + /* + * We need to exclude the super stripes now so that the space + * info has super bytes accounted for, otherwise we'll think + * we have more space than we actually do. + */ + exclude_super_stripes(root, cache); + /* * check for two cases, either we are full, and therefore * don't need to bother with the caching work since we won't @@ -8447,12 +8459,10 @@ int btrfs_read_block_groups(struct btrfs_root *root) * time, particularly in the full case. */ if (found_key.offset == btrfs_block_group_used(&cache->item)) { - exclude_super_stripes(root, cache); cache->last_byte_to_unpin = (u64)-1; cache->cached = BTRFS_CACHE_FINISHED; free_excluded_extents(root, cache); } else if (btrfs_block_group_used(&cache->item) == 0) { - exclude_super_stripes(root, cache); cache->last_byte_to_unpin = (u64)-1; cache->cached = BTRFS_CACHE_FINISHED; add_new_free_space(cache, root->fs_info, -- cgit v1.2.2 From c26a920373a983b52223eed5a13b97404d8b4158 Mon Sep 17 00:00:00 2001 From: Tsutomu Itoh Date: Mon, 14 Feb 2011 00:45:29 +0000 Subject: Btrfs: check return value of alloc_extent_map() I add the check on the return value of alloc_extent_map() to several places. In addition, alloc_extent_map() returns only the address or NULL. Therefore, check by IS_ERR() is unnecessary. So, I remove IS_ERR() checking. Signed-off-by: Tsutomu Itoh Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 565e22d77b1b..a7aaa10c5302 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -6584,7 +6584,7 @@ static noinline int relocate_data_extent(struct inode *reloc_inode, u64 end = start + extent_key->offset - 1; em = alloc_extent_map(GFP_NOFS); - BUG_ON(!em || IS_ERR(em)); + BUG_ON(!em); em->start = start; em->len = extent_key->offset; -- cgit v1.2.2 From 91435650c233b93e0da389db74f4b2c11c5ad2d4 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Wed, 16 Feb 2011 13:10:41 -0500 Subject: Btrfs: put ENOSPC debugging under a mount option ENOSPC in btrfs is getting to the point where the extra debugging isn't required. I've put it under mount -o enospc_debug just in case someone is having difficult problems. Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index a7aaa10c5302..d375fc04a065 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5377,7 +5377,7 @@ again: num_bytes, data, 1); goto again; } - if (ret == -ENOSPC) { + if (ret == -ENOSPC && btrfs_test_opt(root, ENOSPC_DEBUG)) { struct btrfs_space_info *sinfo; sinfo = __find_space_info(root->fs_info, data); -- cgit v1.2.2 From c87f08ca44e83b2c8d28f63f9c33f3a270a04bbe Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Wed, 16 Feb 2011 13:57:04 -0500 Subject: Btrfs: allow balance to explicitly allocate chunks as it relocates Btrfs device shrinking and balancing ends up reallocating all the blocks in order to allow COW to move them to new destinations. It is somewhat awkward in terms of ENOSPC because most of the enospc code is built around the idea that some operation on a reference counted tree triggers allocations in the non-reference counted trees. This commit changes the balancing code to deal with enospc by trying to allocate a new chunk. If that allocation succeeds, we go ahead and retry whatever failed due to enospc. Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index d375fc04a065..100e409e9053 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -8066,6 +8066,13 @@ out: return ret; } +int btrfs_force_chunk_alloc(struct btrfs_trans_handle *trans, + struct btrfs_root *root, u64 type) +{ + u64 alloc_flags = get_alloc_profile(root, type); + return do_chunk_alloc(trans, root, 2 * 1024 * 1024, alloc_flags, 1); +} + /* * helper to account the unused space of all the readonly block group in the * list. takes mirrors into account. -- cgit v1.2.2 From 36e39c40b3facc9b489a13f1d301fc53ff6960a3 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Sat, 12 Mar 2011 07:08:42 -0500 Subject: Btrfs: break out of shrink_delalloc earlier Josef had changed shrink_delalloc to exit after three shrink attempts, which wasn't quite enough because new writers could race in and steal free space. But it also fixed deadlocks and stalls as we tried to recover delalloc reservations. The code was tweaked to loop 1024 times, and would reset the counter any time a small amount of progress was made. This was too drastic, and with a lot of writers we can end up stuck in shrink_delalloc forever. The shrink_delalloc loop is fairly complex because the caller is looping too, and the caller will go ahead and force a transaction commit to make sure we reclaim space. This reworks things to exit shrink_delalloc when we've forced some writeback and the delalloc reservations have gone down. This means the writeback has not just started but has also finished at least some of the metadata changes required to reclaim delalloc space. If we've got this wrong, we're returning ENOSPC too early, which is a big improvement over the current behavior of hanging the machine. Test 224 in xfstests hammers on this nicely, and with 1000 writers trying to fill a 1GB drive we get our first ENOSPC at 93% full. The other writers are able to continue until we get 100%. This is a worst case test for btrfs because the 1000 writers are doing small IO, and the small FS size means we don't have a lot of room for metadata chunks. Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 35 +++++++++++++++++++++++------------ 1 file changed, 23 insertions(+), 12 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 100e409e9053..f1db57d4a016 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3343,15 +3343,16 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans, u64 max_reclaim; u64 reclaimed = 0; long time_left; - int pause = 1; int nr_pages = (2 * 1024 * 1024) >> PAGE_CACHE_SHIFT; int loops = 0; + unsigned long progress; block_rsv = &root->fs_info->delalloc_block_rsv; space_info = block_rsv->space_info; smp_mb(); reserved = space_info->bytes_reserved; + progress = space_info->reservation_progress; if (reserved == 0) return 0; @@ -3366,31 +3367,36 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans, writeback_inodes_sb_nr_if_idle(root->fs_info->sb, nr_pages); spin_lock(&space_info->lock); - if (reserved > space_info->bytes_reserved) { - loops = 0; + if (reserved > space_info->bytes_reserved) reclaimed += reserved - space_info->bytes_reserved; - } else { - loops++; - } reserved = space_info->bytes_reserved; spin_unlock(&space_info->lock); + loops++; + if (reserved == 0 || reclaimed >= max_reclaim) break; if (trans && trans->transaction->blocked) return -EAGAIN; - __set_current_state(TASK_INTERRUPTIBLE); - time_left = schedule_timeout(pause); + time_left = schedule_timeout_interruptible(1); /* We were interrupted, exit */ if (time_left) break; - pause <<= 1; - if (pause > HZ / 10) - pause = HZ / 10; + /* we've kicked the IO a few times, if anything has been freed, + * exit. There is no sense in looping here for a long time + * when we really need to commit the transaction, or there are + * just too many writers without enough free space + */ + + if (loops > 3) { + smp_mb(); + if (progress != space_info->reservation_progress) + break; + } } return reclaimed >= to_reclaim; @@ -3613,6 +3619,7 @@ void block_rsv_release_bytes(struct btrfs_block_rsv *block_rsv, if (num_bytes) { spin_lock(&space_info->lock); space_info->bytes_reserved -= num_bytes; + space_info->reservation_progress++; spin_unlock(&space_info->lock); } } @@ -3845,6 +3852,7 @@ static void update_global_block_rsv(struct btrfs_fs_info *fs_info) if (block_rsv->reserved >= block_rsv->size) { num_bytes = block_rsv->reserved - block_rsv->size; sinfo->bytes_reserved -= num_bytes; + sinfo->reservation_progress++; block_rsv->reserved = block_rsv->size; block_rsv->full = 1; } @@ -4006,7 +4014,6 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes) to_reserve = 0; } spin_unlock(&BTRFS_I(inode)->accounting_lock); - to_reserve += calc_csum_metadata_size(inode, num_bytes); ret = reserve_metadata_bytes(NULL, root, block_rsv, to_reserve, 1); if (ret) @@ -4134,6 +4141,7 @@ static int update_block_group(struct btrfs_trans_handle *trans, btrfs_set_block_group_used(&cache->item, old_val); cache->reserved -= num_bytes; cache->space_info->bytes_reserved -= num_bytes; + cache->space_info->reservation_progress++; cache->space_info->bytes_used += num_bytes; cache->space_info->disk_used += num_bytes * factor; spin_unlock(&cache->lock); @@ -4185,6 +4193,7 @@ static int pin_down_extent(struct btrfs_root *root, if (reserved) { cache->reserved -= num_bytes; cache->space_info->bytes_reserved -= num_bytes; + cache->space_info->reservation_progress++; } spin_unlock(&cache->lock); spin_unlock(&cache->space_info->lock); @@ -4235,6 +4244,7 @@ static int update_reserved_bytes(struct btrfs_block_group_cache *cache, space_info->bytes_readonly += num_bytes; cache->reserved -= num_bytes; space_info->bytes_reserved -= num_bytes; + space_info->reservation_progress++; } spin_unlock(&cache->lock); spin_unlock(&space_info->lock); @@ -4713,6 +4723,7 @@ void btrfs_free_tree_block(struct btrfs_trans_handle *trans, if (ret) { spin_lock(&cache->space_info->lock); cache->space_info->bytes_reserved -= buf->len; + cache->space_info->reservation_progress++; spin_unlock(&cache->space_info->lock); } goto out; -- cgit v1.2.2 From 57a45ced94fe48a701361d64230fc16eefa189dd Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 25 Jan 2011 16:30:38 -0500 Subject: Btrfs: change reserved_extents to an atomic_t We track delayed allocation per inodes via 2 counters, one is outstanding_extents and reserved_extents. Outstanding_extents is already an atomic_t, but reserved_extents is not and is protected by a spinlock. So convert this to an atomic_t and instead of using a spinlock, use atomic_cmpxchg when releasing delalloc bytes. This makes our inode 72 bytes smaller, and reduces locking overhead (albiet it was minimal to begin with). Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 42 ++++++++++++++++++++++++++---------------- 1 file changed, 26 insertions(+), 16 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 7b3089b5c2df..27376c97d85f 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3996,6 +3996,7 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes) struct btrfs_block_rsv *block_rsv = &root->fs_info->delalloc_block_rsv; u64 to_reserve; int nr_extents; + int reserved_extents; int ret; if (btrfs_transaction_in_commit(root->fs_info)) @@ -4003,25 +4004,24 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes) num_bytes = ALIGN(num_bytes, root->sectorsize); - spin_lock(&BTRFS_I(inode)->accounting_lock); nr_extents = atomic_read(&BTRFS_I(inode)->outstanding_extents) + 1; - if (nr_extents > BTRFS_I(inode)->reserved_extents) { - nr_extents -= BTRFS_I(inode)->reserved_extents; + reserved_extents = atomic_read(&BTRFS_I(inode)->reserved_extents); + + if (nr_extents > reserved_extents) { + nr_extents -= reserved_extents; to_reserve = calc_trans_metadata_size(root, nr_extents); } else { nr_extents = 0; to_reserve = 0; } - spin_unlock(&BTRFS_I(inode)->accounting_lock); + to_reserve += calc_csum_metadata_size(inode, num_bytes); ret = reserve_metadata_bytes(NULL, root, block_rsv, to_reserve, 1); if (ret) return ret; - spin_lock(&BTRFS_I(inode)->accounting_lock); - BTRFS_I(inode)->reserved_extents += nr_extents; + atomic_add(nr_extents, &BTRFS_I(inode)->reserved_extents); atomic_inc(&BTRFS_I(inode)->outstanding_extents); - spin_unlock(&BTRFS_I(inode)->accounting_lock); block_rsv_add_bytes(block_rsv, to_reserve, 1); @@ -4036,20 +4036,30 @@ void btrfs_delalloc_release_metadata(struct inode *inode, u64 num_bytes) struct btrfs_root *root = BTRFS_I(inode)->root; u64 to_free; int nr_extents; + int reserved_extents; num_bytes = ALIGN(num_bytes, root->sectorsize); atomic_dec(&BTRFS_I(inode)->outstanding_extents); WARN_ON(atomic_read(&BTRFS_I(inode)->outstanding_extents) < 0); - spin_lock(&BTRFS_I(inode)->accounting_lock); - nr_extents = atomic_read(&BTRFS_I(inode)->outstanding_extents); - if (nr_extents < BTRFS_I(inode)->reserved_extents) { - nr_extents = BTRFS_I(inode)->reserved_extents - nr_extents; - BTRFS_I(inode)->reserved_extents -= nr_extents; - } else { - nr_extents = 0; - } - spin_unlock(&BTRFS_I(inode)->accounting_lock); + reserved_extents = atomic_read(&BTRFS_I(inode)->reserved_extents); + do { + int old, new; + + nr_extents = atomic_read(&BTRFS_I(inode)->outstanding_extents); + if (nr_extents >= reserved_extents) { + nr_extents = 0; + break; + } + old = reserved_extents; + nr_extents = reserved_extents - nr_extents; + new = reserved_extents - nr_extents; + old = atomic_cmpxchg(&BTRFS_I(inode)->reserved_extents, + reserved_extents, new); + if (likely(old == reserved_extents)) + break; + reserved_extents = old; + } while (1); to_free = calc_csum_metadata_size(inode, num_bytes); if (nr_extents > 0) -- cgit v1.2.2 From 66b4ffd110f9b48b8d8c1319ee446b53b8d073bf Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Mon, 31 Jan 2011 16:22:42 -0500 Subject: Btrfs: handle errors in btrfs_orphan_cleanup If we cannot truncate an inode for some reason we will never delete the orphan item associated with that inode, which means that we will loop forever in btrfs_orphan_cleanup. Instead of doing this just return error so we fail to mount. It sucks, but hey it's better than hanging. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 3 ++- 1 file changed, 2 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 27376c97d85f..a8f4e8d2ba60 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -7619,7 +7619,8 @@ int btrfs_cleanup_reloc_trees(struct btrfs_root *root) reloc_root = btrfs_read_fs_root_no_name(root->fs_info, &location); BUG_ON(!reloc_root); - btrfs_orphan_cleanup(reloc_root); + ret = btrfs_orphan_cleanup(reloc_root); + BUG_ON(ret); return 0; } -- cgit v1.2.2 From a826d6dcb32d811b4c81df57a5ef1367516586b0 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 16 Mar 2011 13:42:43 -0400 Subject: Btrfs: check items for correctness as we search Currently if we have corrupted items things will blow up in spectacular ways. So as we read in blocks and they are leaves, check the entire leaf to make sure all of the items are correct and point to valid parts in the leaf for the item data the are responsible for. If the item is corrupt we will kick back EIO and not read any of the copies since they are likely to not be correct either. This will catch generic corruptions, it will be up to the individual callers of btrfs_search_slot to make sure their items are right. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index a8f4e8d2ba60..cd794c35a636 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4754,6 +4754,11 @@ pin: } } out: + /* + * Deleting the buffer, clear the corrupt flag since it doesn't matter + * anymore. + */ + clear_bit(EXTENT_BUFFER_CORRUPT, &buf->bflags); btrfs_put_block_group(cache); } -- cgit v1.2.2 From 1abe9b8a138c9988ba8f7bfded6453649a31541f Mon Sep 17 00:00:00 2001 From: liubo Date: Thu, 24 Mar 2011 11:18:59 +0000 Subject: Btrfs: add initial tracepoint support for btrfs Tracepoints can provide insight into why btrfs hits bugs and be greatly helpful for debugging, e.g dd-7822 [000] 2121.641088: btrfs_inode_request: root = 5(FS_TREE), gen = 4, ino = 256, blocks = 8, disk_i_size = 0, last_trans = 8, logged_trans = 0 dd-7822 [000] 2121.641100: btrfs_inode_new: root = 5(FS_TREE), gen = 8, ino = 257, blocks = 0, disk_i_size = 0, last_trans = 0, logged_trans = 0 btrfs-transacti-7804 [001] 2146.935420: btrfs_cow_block: root = 2(EXTENT_TREE), refs = 2, orig_buf = 29368320 (orig_level = 0), cow_buf = 29388800 (cow_level = 0) btrfs-transacti-7804 [001] 2146.935473: btrfs_cow_block: root = 1(ROOT_TREE), refs = 2, orig_buf = 29364224 (orig_level = 0), cow_buf = 29392896 (cow_level = 0) btrfs-transacti-7804 [001] 2146.972221: btrfs_transaction_commit: root = 1(ROOT_TREE), gen = 8 flush-btrfs-2-7821 [001] 2155.824210: btrfs_chunk_alloc: root = 3(CHUNK_TREE), offset = 1103101952, size = 1073741824, num_stripes = 1, sub_stripes = 0, type = DATA flush-btrfs-2-7821 [001] 2155.824241: btrfs_cow_block: root = 2(EXTENT_TREE), refs = 2, orig_buf = 29388800 (orig_level = 0), cow_buf = 29396992 (cow_level = 0) flush-btrfs-2-7821 [001] 2155.824255: btrfs_cow_block: root = 4(DEV_TREE), refs = 2, orig_buf = 29372416 (orig_level = 0), cow_buf = 29401088 (cow_level = 0) flush-btrfs-2-7821 [000] 2155.824329: btrfs_cow_block: root = 3(CHUNK_TREE), refs = 2, orig_buf = 20971520 (orig_level = 0), cow_buf = 20975616 (cow_level = 0) btrfs-endio-wri-7800 [001] 2155.898019: btrfs_cow_block: root = 5(FS_TREE), refs = 2, orig_buf = 29384704 (orig_level = 0), cow_buf = 29405184 (cow_level = 0) btrfs-endio-wri-7800 [001] 2155.898043: btrfs_cow_block: root = 7(CSUM_TREE), refs = 2, orig_buf = 29376512 (orig_level = 0), cow_buf = 29409280 (cow_level = 0) Here is what I have added: 1) ordere_extent: btrfs_ordered_extent_add btrfs_ordered_extent_remove btrfs_ordered_extent_start btrfs_ordered_extent_put These provide critical information to understand how ordered_extents are updated. 2) extent_map: btrfs_get_extent extent_map is used in both read and write cases, and it is useful for tracking how btrfs specific IO is running. 3) writepage: __extent_writepage btrfs_writepage_end_io_hook Pages are cirtical resourses and produce a lot of corner cases during writeback, so it is valuable to know how page is written to disk. 4) inode: btrfs_inode_new btrfs_inode_request btrfs_inode_evict These can show where and when a inode is created, when a inode is evicted. 5) sync: btrfs_sync_file btrfs_sync_fs These show sync arguments. 6) transaction: btrfs_transaction_commit In transaction based filesystem, it will be useful to know the generation and who does commit. 7) back reference and cow: btrfs_delayed_tree_ref btrfs_delayed_data_ref btrfs_delayed_ref_head btrfs_cow_block Btrfs natively supports back references, these tracepoints are helpful on understanding btrfs's COW mechanism. 8) chunk: btrfs_chunk_alloc btrfs_chunk_free Chunk is a link between physical offset and logical offset, and stands for space infomation in btrfs, and these are helpful on tracing space things. 9) reserved_extent: btrfs_reserved_extent_alloc btrfs_reserved_extent_free These can show how btrfs uses its space. Signed-off-by: Liu Bo Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index cd794c35a636..86ea471d3801 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5412,6 +5412,8 @@ again: dump_space_info(sinfo, num_bytes, 1); } + trace_btrfs_reserved_extent_alloc(root, ins->objectid, ins->offset); + return ret; } @@ -5433,6 +5435,8 @@ int btrfs_free_reserved_extent(struct btrfs_root *root, u64 start, u64 len) update_reserved_bytes(cache, len, 0, 1); btrfs_put_block_group(cache); + trace_btrfs_reserved_extent_free(root, start, len); + return ret; } -- cgit v1.2.2 From db5b493ac78e46c7b6bad22cd25d8041564cd8ea Mon Sep 17 00:00:00 2001 From: Tsutomu Itoh Date: Wed, 23 Mar 2011 08:14:16 +0000 Subject: Btrfs: cleanup some BUG_ON() This patch changes some BUG_ON() to the error return. (but, most callers still use BUG_ON()) Signed-off-by: Tsutomu Itoh Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 25 ++++++++++++++++++------- 1 file changed, 18 insertions(+), 7 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 86ea471d3801..a6a8159c5d1e 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5463,7 +5463,8 @@ static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans, size = sizeof(*extent_item) + btrfs_extent_inline_ref_size(type); path = btrfs_alloc_path(); - BUG_ON(!path); + if (!path) + return -ENOMEM; path->leave_spinning = 1; ret = btrfs_insert_empty_item(trans, fs_info->extent_root, path, @@ -6457,10 +6458,14 @@ int btrfs_drop_subtree(struct btrfs_trans_handle *trans, BUG_ON(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID); path = btrfs_alloc_path(); - BUG_ON(!path); + if (!path) + return -ENOMEM; wc = kzalloc(sizeof(*wc), GFP_NOFS); - BUG_ON(!wc); + if (!wc) { + btrfs_free_path(path); + return -ENOMEM; + } btrfs_assert_tree_locked(parent); parent_level = btrfs_header_level(parent); @@ -6918,7 +6923,11 @@ static noinline int get_new_locations(struct inode *reloc_inode, } path = btrfs_alloc_path(); - BUG_ON(!path); + if (!path) { + if (exts != *extents) + kfree(exts); + return -ENOMEM; + } cur_pos = extent_key->objectid - offset; last_byte = extent_key->objectid + extent_key->offset; @@ -7442,7 +7451,8 @@ static noinline int replace_extents_in_leaf(struct btrfs_trans_handle *trans, int ret; new_extent = kmalloc(sizeof(*new_extent), GFP_NOFS); - BUG_ON(!new_extent); + if (!new_extent) + return -ENOMEM; ref = btrfs_lookup_leaf_ref(root, leaf->start); BUG_ON(!ref); @@ -7647,7 +7657,8 @@ static noinline int init_reloc_tree(struct btrfs_trans_handle *trans, return 0; root_item = kmalloc(sizeof(*root_item), GFP_NOFS); - BUG_ON(!root_item); + if (!root_item) + return -ENOMEM; ret = btrfs_copy_root(trans, root, root->commit_root, &eb, BTRFS_TREE_RELOC_OBJECTID); @@ -7673,7 +7684,7 @@ static noinline int init_reloc_tree(struct btrfs_trans_handle *trans, reloc_root = btrfs_read_fs_root_no_radix(root->fs_info->tree_root, &root_key); - BUG_ON(!reloc_root); + BUG_ON(IS_ERR(reloc_root)); reloc_root->last_trans = trans->transid; reloc_root->commit_root = NULL; reloc_root->ref_tree = &root->fs_info->reloc_ref_tree; -- cgit v1.2.2 From 97d9a8a420444eb5b5c071d4b3b9c4100a7ae015 Mon Sep 17 00:00:00 2001 From: Tsutomu Itoh Date: Thu, 24 Mar 2011 06:33:21 +0000 Subject: Btrfs: check return value of read_tree_block() This patch is checking return value of read_tree_block(), and if it is NULL, error processing. Signed-off-by: Tsutomu Itoh Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index a6a8159c5d1e..5bc658a9d85c 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -6067,6 +6067,8 @@ static noinline int do_walk_down(struct btrfs_trans_handle *trans, if (reada && level == 1) reada_walk_down(trans, root, wc, path); next = read_tree_block(root, bytenr, blocksize, generation); + if (!next) + return -EIO; btrfs_tree_lock(next); btrfs_set_lock_blocking(next); } @@ -7937,6 +7939,10 @@ static noinline int relocate_one_extent(struct btrfs_root *extent_root, eb = read_tree_block(found_root, block_start, block_size, 0); + if (!eb) { + ret = -EIO; + goto out; + } btrfs_tree_lock(eb); BUG_ON(level != btrfs_header_level(eb)); -- cgit v1.2.2 From fc0e4a314e361af3b13d9320e92c64118f9a3e61 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Thu, 24 Mar 2011 11:41:21 +0000 Subject: btrfs: use GFP_NOFS instead of GFP_KERNEL In the filesystem context, we must allocate memory by GFP_NOFS, or we may start another filesystem operation and make kswap thread hang up. Signed-off-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 5bc658a9d85c..7922f296420d 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -471,7 +471,7 @@ static int cache_block_group(struct btrfs_block_group_cache *cache, if (load_cache_only) return 0; - caching_ctl = kzalloc(sizeof(*caching_ctl), GFP_KERNEL); + caching_ctl = kzalloc(sizeof(*caching_ctl), GFP_NOFS); BUG_ON(!caching_ctl); INIT_LIST_HEAD(&caching_ctl->list); @@ -1743,7 +1743,7 @@ static int remove_extent_backref(struct btrfs_trans_handle *trans, static void btrfs_issue_discard(struct block_device *bdev, u64 start, u64 len) { - blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_KERNEL, 0); + blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_NOFS, 0); } static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr, -- cgit v1.2.2 From b4d00d569a49fcef02195635dbf8d15904b1fb63 Mon Sep 17 00:00:00 2001 From: Li Dongyang Date: Thu, 24 Mar 2011 10:24:25 +0000 Subject: Btrfs: make update_reserved_bytes() public Make the function public as we should update the reserved extents calculations after taking out an extent for trimming. Signed-off-by: Li Dongyang Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 16 +++++++--------- 1 file changed, 7 insertions(+), 9 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 7922f296420d..0671f5b77eb8 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -36,8 +36,6 @@ static int update_block_group(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 bytenr, u64 num_bytes, int alloc); -static int update_reserved_bytes(struct btrfs_block_group_cache *cache, - u64 num_bytes, int reserve, int sinfo); static int __btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 bytenr, u64 num_bytes, u64 parent, @@ -4233,8 +4231,8 @@ int btrfs_pin_extent(struct btrfs_root *root, * update size of reserved extents. this function may return -EAGAIN * if 'reserve' is true or 'sinfo' is false. */ -static int update_reserved_bytes(struct btrfs_block_group_cache *cache, - u64 num_bytes, int reserve, int sinfo) +int btrfs_update_reserved_bytes(struct btrfs_block_group_cache *cache, + u64 num_bytes, int reserve, int sinfo) { int ret = 0; if (sinfo) { @@ -4714,10 +4712,10 @@ void btrfs_free_tree_block(struct btrfs_trans_handle *trans, WARN_ON(test_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)); btrfs_add_free_space(cache, buf->start, buf->len); - ret = update_reserved_bytes(cache, buf->len, 0, 0); + ret = btrfs_update_reserved_bytes(cache, buf->len, 0, 0); if (ret == -EAGAIN) { /* block group became read-only */ - update_reserved_bytes(cache, buf->len, 0, 1); + btrfs_update_reserved_bytes(cache, buf->len, 0, 1); goto out; } @@ -5206,7 +5204,7 @@ checks: search_start - offset); BUG_ON(offset > search_start); - ret = update_reserved_bytes(block_group, num_bytes, 1, + ret = btrfs_update_reserved_bytes(block_group, num_bytes, 1, (data & BTRFS_BLOCK_GROUP_DATA)); if (ret == -EAGAIN) { btrfs_add_free_space(block_group, offset, num_bytes); @@ -5432,7 +5430,7 @@ int btrfs_free_reserved_extent(struct btrfs_root *root, u64 start, u64 len) ret = btrfs_discard_extent(root, start, len); btrfs_add_free_space(cache, start, len); - update_reserved_bytes(cache, len, 0, 1); + btrfs_update_reserved_bytes(cache, len, 0, 1); btrfs_put_block_group(cache); trace_btrfs_reserved_extent_free(root, start, len); @@ -5634,7 +5632,7 @@ int btrfs_alloc_logged_file_extent(struct btrfs_trans_handle *trans, put_caching_control(caching_ctl); } - ret = update_reserved_bytes(block_group, ins->offset, 1, 1); + ret = btrfs_update_reserved_bytes(block_group, ins->offset, 1, 1); BUG_ON(ret); btrfs_put_block_group(block_group); ret = alloc_reserved_file_extent(trans, root, 0, root_objectid, -- cgit v1.2.2 From 5378e60734f5b7bfe1b43dc191aaf6131c1befe7 Mon Sep 17 00:00:00 2001 From: Li Dongyang Date: Thu, 24 Mar 2011 10:24:27 +0000 Subject: Btrfs: adjust btrfs_discard_extent() return errors and trimmed bytes Callers of btrfs_discard_extent() should check if we are mounted with -o discard, as we want to make fitrim to work even the fs is not mounted with -o discard. Also we should use REQ_DISCARD to map the free extent to get a full mapping, last we only return errors if 1. the error is not a EOPNOTSUPP 2. no device supports discard Signed-off-by: Li Dongyang Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 43 ++++++++++++++++++++++++++----------------- 1 file changed, 26 insertions(+), 17 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 0671f5b77eb8..e990d2d1ba4a 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -1738,39 +1738,45 @@ static int remove_extent_backref(struct btrfs_trans_handle *trans, return ret; } -static void btrfs_issue_discard(struct block_device *bdev, +static int btrfs_issue_discard(struct block_device *bdev, u64 start, u64 len) { - blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_NOFS, 0); + return blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_NOFS, 0); } static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr, - u64 num_bytes) + u64 num_bytes, u64 *actual_bytes) { int ret; - u64 map_length = num_bytes; + u64 discarded_bytes = 0; struct btrfs_multi_bio *multi = NULL; - if (!btrfs_test_opt(root, DISCARD)) - return 0; /* Tell the block device(s) that the sectors can be discarded */ - ret = btrfs_map_block(&root->fs_info->mapping_tree, READ, - bytenr, &map_length, &multi, 0); + ret = btrfs_map_block(&root->fs_info->mapping_tree, REQ_DISCARD, + bytenr, &num_bytes, &multi, 0); if (!ret) { struct btrfs_bio_stripe *stripe = multi->stripes; int i; - if (map_length > num_bytes) - map_length = num_bytes; for (i = 0; i < multi->num_stripes; i++, stripe++) { - btrfs_issue_discard(stripe->dev->bdev, - stripe->physical, - map_length); + ret = btrfs_issue_discard(stripe->dev->bdev, + stripe->physical, + stripe->length); + if (!ret) + discarded_bytes += stripe->length; + else if (ret != -EOPNOTSUPP) + break; } kfree(multi); } + if (discarded_bytes && ret == -EOPNOTSUPP) + ret = 0; + + if (actual_bytes) + *actual_bytes = discarded_bytes; + return ret; } @@ -4371,7 +4377,9 @@ int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans, if (ret) break; - ret = btrfs_discard_extent(root, start, end + 1 - start); + if (btrfs_test_opt(root, DISCARD)) + ret = btrfs_discard_extent(root, start, + end + 1 - start, NULL); clear_extent_dirty(unpin, start, end, GFP_NOFS); unpin_extent_range(root, start, end); @@ -5427,7 +5435,8 @@ int btrfs_free_reserved_extent(struct btrfs_root *root, u64 start, u64 len) return -ENOSPC; } - ret = btrfs_discard_extent(root, start, len); + if (btrfs_test_opt(root, DISCARD)) + ret = btrfs_discard_extent(root, start, len, NULL); btrfs_add_free_space(cache, start, len); btrfs_update_reserved_bytes(cache, len, 0, 1); @@ -8765,7 +8774,7 @@ int btrfs_error_unpin_extent_range(struct btrfs_root *root, u64 start, u64 end) } int btrfs_error_discard_extent(struct btrfs_root *root, u64 bytenr, - u64 num_bytes) + u64 num_bytes, u64 *actual_bytes) { - return btrfs_discard_extent(root, bytenr, num_bytes); + return btrfs_discard_extent(root, bytenr, num_bytes, actual_bytes); } -- cgit v1.2.2 From f7039b1d5c32241f87a513e33120db36bf30264d Mon Sep 17 00:00:00 2001 From: Li Dongyang Date: Thu, 24 Mar 2011 10:24:28 +0000 Subject: Btrfs: add btrfs_trim_fs() to handle FITRIM We take an free extent out from allocator, trim it, then put it back, but before we trim the block group, we should make sure the block group is cached, so plus a little change to make cache_block_group() run without a transaction. Signed-off-by: Li Dongyang Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 50 +++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 49 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 e990d2d1ba4a..1efeda3b2f6f 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -440,7 +440,7 @@ static int cache_block_group(struct btrfs_block_group_cache *cache, * 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 && (!trans->transaction->in_commit) && (root && root != root->fs_info->tree_root)) { spin_lock(&cache->lock); if (cache->cached != BTRFS_CACHE_NO) { @@ -8778,3 +8778,51 @@ int btrfs_error_discard_extent(struct btrfs_root *root, u64 bytenr, { return btrfs_discard_extent(root, bytenr, num_bytes, actual_bytes); } + +int btrfs_trim_fs(struct btrfs_root *root, struct fstrim_range *range) +{ + struct btrfs_fs_info *fs_info = root->fs_info; + struct btrfs_block_group_cache *cache = NULL; + u64 group_trimmed; + u64 start; + u64 end; + u64 trimmed = 0; + int ret = 0; + + cache = btrfs_lookup_block_group(fs_info, range->start); + + while (cache) { + if (cache->key.objectid >= (range->start + range->len)) { + btrfs_put_block_group(cache); + break; + } + + start = max(range->start, cache->key.objectid); + end = min(range->start + range->len, + cache->key.objectid + cache->key.offset); + + if (end - start >= range->minlen) { + if (!block_group_cache_done(cache)) { + ret = cache_block_group(cache, NULL, root, 0); + if (!ret) + wait_block_group_cache_done(cache); + } + ret = btrfs_trim_block_group(cache, + &group_trimmed, + start, + end, + range->minlen); + + trimmed += group_trimmed; + if (ret) { + btrfs_put_block_group(cache); + break; + } + } + + cache = next_block_group(fs_info->tree_root, cache); + } + + range->len = trimmed; + return ret; +} -- cgit v1.2.2 From dac97e516c617f9c797f64b0224050b70aea30c7 Mon Sep 17 00:00:00 2001 From: Yoshinori Sano Date: Tue, 15 Feb 2011 12:01:42 +0000 Subject: Btrfs: fix uncheck memory allocations To make Btrfs code more robust, several return value checks where memory allocation can fail are introduced. I use BUG_ON where I don't know how to handle the error properly, which increases the number of using the notorious BUG_ON, though. Signed-off-by: Yoshinori Sano Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 1efeda3b2f6f..cd0b69f57375 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -6978,6 +6978,10 @@ static noinline int get_new_locations(struct inode *reloc_inode, struct disk_extent *old = exts; max *= 2; exts = kzalloc(sizeof(*exts) * max, GFP_NOFS); + if (!exts) { + ret = -ENOMEM; + goto out; + } memcpy(exts, old, sizeof(*exts) * nr); if (old != *extents) kfree(old); -- cgit v1.2.2 From 9f7c43c96727a53bea45f7f2549d897f0a6117b8 Mon Sep 17 00:00:00 2001 From: liubo Date: Mon, 7 Mar 2011 02:13:33 +0000 Subject: Btrfs: fix memory leak of empty filesystem after balance After Josef's patch(commit 3c14874acc71180553fb5aba528e3cf57c5b958b), btrfs will exclude super bytes when reading block groups(by marking a extent state UPTODATE). However, these bytes do not get freed while balance remove unused block groups, and we won't process those removed ones any more, when we do umount and unload the btrfs module, btrfs hits a memory leak. This patch add the missing free operation. Reproduce steps: $ mkfs.btrfs disk $ mount disk /mnt/btrfs -o loop $ btrfs filesystem balance /mnt/btrfs $ umount /mnt/btrfs $ rmmod btrfs Signed-off-by: Liu Bo Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index cd0b69f57375..a561060f5ffb 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -8669,6 +8669,12 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans, BUG_ON(!block_group); BUG_ON(!block_group->ro); + /* + * Free the reserved super bytes from this block group before + * remove it. + */ + free_excluded_extents(root, block_group); + memcpy(&key, &block_group->key, sizeof(key)); if (block_group->flags & (BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID1 | -- cgit v1.2.2 From c59021f846881a957ac5afe456d0f59d6a517b61 Mon Sep 17 00:00:00 2001 From: liubo Date: Mon, 7 Mar 2011 02:13:14 +0000 Subject: Btrfs: fix OOPS of empty filesystem after balance btrfs will remove unused block groups after balance. When a empty filesystem is balanced, the block group with tag "DATA" may be dropped, and after umount and mount again, it will not find "DATA" space_info and lead to OOPS. So we initial the necessary space_infos(DATA, SYSTEM, METADATA) to avoid OOPS. Reported-by: Daniel J Blueman Signed-off-by: Liu Bo Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 23 +++++++++++++++++++++++ 1 file changed, 23 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index a561060f5ffb..f619c3cb13b7 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -8778,6 +8778,29 @@ out: return ret; } +int btrfs_init_space_info(struct btrfs_fs_info *fs_info) +{ + struct btrfs_space_info *space_info; + int ret; + + ret = update_space_info(fs_info, BTRFS_BLOCK_GROUP_SYSTEM, 0, 0, + &space_info); + if (ret) + return ret; + + ret = update_space_info(fs_info, BTRFS_BLOCK_GROUP_METADATA, 0, 0, + &space_info); + if (ret) + return ret; + + ret = update_space_info(fs_info, BTRFS_BLOCK_GROUP_DATA, 0, 0, + &space_info); + if (ret) + return ret; + + return ret; +} + int btrfs_error_unpin_extent_range(struct btrfs_root *root, u64 start, u64 end) { return unpin_extent_range(root, start, end); -- cgit v1.2.2 From 0e4f8f888845f9dca540ad175884244e5db5eea2 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Fri, 15 Apr 2011 16:05:44 -0400 Subject: Btrfs: don't force chunk allocation in find_free_extent find_free_extent likes to allocate in contiguous clusters, which makes writeback faster, especially on SSD storage. As the FS fragments, these clusters become harder to find and we have to decide between allocating a new chunk to make more clusters or giving up on the cluster to allocate from the free space we have. Right now it creates too many chunks, and you can end up with a whole FS that is mostly empty metadata chunks. This commit changes the allocation code to be more strict and only allocate new chunks when we've made good use of the chunks we already have. Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 95 ++++++++++++++++++++++++++++++++++++++------------ 1 file changed, 73 insertions(+), 22 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index f619c3cb13b7..26479484180d 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -33,6 +33,25 @@ #include "locking.h" #include "free-space-cache.h" +/* control flags for do_chunk_alloc's force field + * CHUNK_ALLOC_NO_FORCE means to only allocate a chunk + * if we really need one. + * + * CHUNK_ALLOC_FORCE means it must try to allocate one + * + * CHUNK_ALLOC_LIMITED means to only try and allocate one + * if we have very few chunks already allocated. This is + * used as part of the clustering code to help make sure + * we have a good pool of storage to cluster in, without + * filling the FS with empty chunks + * + */ +enum { + CHUNK_ALLOC_NO_FORCE = 0, + CHUNK_ALLOC_FORCE = 1, + CHUNK_ALLOC_LIMITED = 2, +}; + static int update_block_group(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 bytenr, u64 num_bytes, int alloc); @@ -3019,7 +3038,7 @@ static int update_space_info(struct btrfs_fs_info *info, u64 flags, found->bytes_readonly = 0; found->bytes_may_use = 0; found->full = 0; - found->force_alloc = 0; + found->force_alloc = CHUNK_ALLOC_NO_FORCE; *space_info = found; list_add_rcu(&found->list, &info->space_info); atomic_set(&found->caching_threads, 0); @@ -3150,7 +3169,7 @@ again: if (!data_sinfo->full && alloc_chunk) { u64 alloc_target; - data_sinfo->force_alloc = 1; + data_sinfo->force_alloc = CHUNK_ALLOC_FORCE; spin_unlock(&data_sinfo->lock); alloc: alloc_target = btrfs_get_alloc_profile(root, 1); @@ -3160,7 +3179,8 @@ alloc: ret = do_chunk_alloc(trans, root->fs_info->extent_root, bytes + 2 * 1024 * 1024, - alloc_target, 0); + alloc_target, + CHUNK_ALLOC_NO_FORCE); btrfs_end_transaction(trans, root); if (ret < 0) { if (ret != -ENOSPC) @@ -3239,31 +3259,56 @@ static void force_metadata_allocation(struct btrfs_fs_info *info) rcu_read_lock(); list_for_each_entry_rcu(found, head, list) { if (found->flags & BTRFS_BLOCK_GROUP_METADATA) - found->force_alloc = 1; + found->force_alloc = CHUNK_ALLOC_FORCE; } rcu_read_unlock(); } static int should_alloc_chunk(struct btrfs_root *root, - struct btrfs_space_info *sinfo, u64 alloc_bytes) + struct btrfs_space_info *sinfo, u64 alloc_bytes, + int force) { u64 num_bytes = sinfo->total_bytes - sinfo->bytes_readonly; + u64 num_allocated = sinfo->bytes_used + sinfo->bytes_reserved; u64 thresh; - if (sinfo->bytes_used + sinfo->bytes_reserved + - alloc_bytes + 256 * 1024 * 1024 < num_bytes) + if (force == CHUNK_ALLOC_FORCE) + return 1; + + /* + * in limited mode, we want to have some free space up to + * about 1% of the FS size. + */ + if (force == CHUNK_ALLOC_LIMITED) { + thresh = btrfs_super_total_bytes(&root->fs_info->super_copy); + thresh = max_t(u64, 64 * 1024 * 1024, + div_factor_fine(thresh, 1)); + + if (num_bytes - num_allocated < thresh) + return 1; + } + + /* + * we have two similar checks here, one based on percentage + * and once based on a hard number of 256MB. The idea + * is that if we have a good amount of free + * room, don't allocate a chunk. A good mount is + * less than 80% utilized of the chunks we have allocated, + * or more than 256MB free + */ + if (num_allocated + alloc_bytes + 256 * 1024 * 1024 < num_bytes) return 0; - if (sinfo->bytes_used + sinfo->bytes_reserved + - alloc_bytes < div_factor(num_bytes, 8)) + if (num_allocated + alloc_bytes < div_factor(num_bytes, 8)) return 0; thresh = btrfs_super_total_bytes(&root->fs_info->super_copy); + + /* 256MB or 5% of the FS */ thresh = max_t(u64, 256 * 1024 * 1024, div_factor_fine(thresh, 5)); if (num_bytes > thresh && sinfo->bytes_used < div_factor(num_bytes, 3)) return 0; - return 1; } @@ -3289,17 +3334,17 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans, spin_lock(&space_info->lock); if (space_info->force_alloc) - force = 1; + force = space_info->force_alloc; if (space_info->full) { spin_unlock(&space_info->lock); goto out; } - if (!force && !should_alloc_chunk(extent_root, space_info, - alloc_bytes)) { + if (!should_alloc_chunk(extent_root, space_info, alloc_bytes, force)) { spin_unlock(&space_info->lock); goto out; } + spin_unlock(&space_info->lock); /* @@ -3327,7 +3372,7 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans, space_info->full = 1; else ret = 1; - space_info->force_alloc = 0; + space_info->force_alloc = CHUNK_ALLOC_NO_FORCE; spin_unlock(&space_info->lock); out: mutex_unlock(&extent_root->fs_info->chunk_mutex); @@ -5303,11 +5348,13 @@ loop: if (allowed_chunk_alloc) { ret = do_chunk_alloc(trans, root, num_bytes + - 2 * 1024 * 1024, data, 1); + 2 * 1024 * 1024, data, + CHUNK_ALLOC_LIMITED); allowed_chunk_alloc = 0; done_chunk_alloc = 1; - } else if (!done_chunk_alloc) { - space_info->force_alloc = 1; + } else if (!done_chunk_alloc && + space_info->force_alloc == CHUNK_ALLOC_NO_FORCE) { + space_info->force_alloc = CHUNK_ALLOC_LIMITED; } if (loop < LOOP_NO_EMPTY_SIZE) { @@ -5393,7 +5440,8 @@ again: */ if (empty_size || root->ref_cows) ret = do_chunk_alloc(trans, root->fs_info->extent_root, - num_bytes + 2 * 1024 * 1024, data, 0); + num_bytes + 2 * 1024 * 1024, data, + CHUNK_ALLOC_NO_FORCE); WARN_ON(num_bytes < root->sectorsize); ret = find_free_extent(trans, root, num_bytes, empty_size, @@ -5405,7 +5453,7 @@ again: num_bytes = num_bytes & ~(root->sectorsize - 1); num_bytes = max(num_bytes, min_alloc_size); do_chunk_alloc(trans, root->fs_info->extent_root, - num_bytes, data, 1); + num_bytes, data, CHUNK_ALLOC_FORCE); goto again; } if (ret == -ENOSPC && btrfs_test_opt(root, ENOSPC_DEBUG)) { @@ -8109,13 +8157,15 @@ int btrfs_set_block_group_ro(struct btrfs_root *root, alloc_flags = update_block_group_flags(root, cache->flags); if (alloc_flags != cache->flags) - do_chunk_alloc(trans, root, 2 * 1024 * 1024, alloc_flags, 1); + do_chunk_alloc(trans, root, 2 * 1024 * 1024, alloc_flags, + CHUNK_ALLOC_FORCE); ret = set_block_group_ro(cache); if (!ret) goto out; alloc_flags = get_alloc_profile(root, cache->space_info->flags); - ret = do_chunk_alloc(trans, root, 2 * 1024 * 1024, alloc_flags, 1); + ret = do_chunk_alloc(trans, root, 2 * 1024 * 1024, alloc_flags, + CHUNK_ALLOC_FORCE); if (ret < 0) goto out; ret = set_block_group_ro(cache); @@ -8128,7 +8178,8 @@ int btrfs_force_chunk_alloc(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 type) { u64 alloc_flags = get_alloc_profile(root, type); - return do_chunk_alloc(trans, root, 2 * 1024 * 1024, alloc_flags, 1); + return do_chunk_alloc(trans, root, 2 * 1024 * 1024, alloc_flags, + CHUNK_ALLOC_FORCE); } /* -- cgit v1.2.2 From 6d74119f1a3efad9dc7f79a16c201242324b731f Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Mon, 11 Apr 2011 20:20:11 -0400 Subject: Btrfs: avoid taking the chunk_mutex in do_chunk_alloc Everytime we try to allocate disk space we try and see if we can pre-emptively allocate a chunk, but in the common case we don't allocate anything, so there is no sense in taking the chunk_mutex at all. So instead if we are allocating a chunk, mark it in the space_info so we don't get two people trying to allocate at the same time. Thanks, Signed-off-by: Josef Bacik Reviewed-by: Liu Bo --- fs/btrfs/extent-tree.c | 30 +++++++++++++++++++++++++----- 1 file changed, 25 insertions(+), 5 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 26479484180d..31f33ba56fe8 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3039,6 +3039,7 @@ static int update_space_info(struct btrfs_fs_info *info, u64 flags, found->bytes_may_use = 0; found->full = 0; found->force_alloc = CHUNK_ALLOC_NO_FORCE; + found->chunk_alloc = 0; *space_info = found; list_add_rcu(&found->list, &info->space_info); atomic_set(&found->caching_threads, 0); @@ -3318,10 +3319,9 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans, { struct btrfs_space_info *space_info; struct btrfs_fs_info *fs_info = extent_root->fs_info; + int wait_for_alloc = 0; int ret = 0; - mutex_lock(&fs_info->chunk_mutex); - flags = btrfs_reduce_alloc_profile(extent_root, flags); space_info = __find_space_info(extent_root->fs_info, flags); @@ -3332,21 +3332,40 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans, } BUG_ON(!space_info); +again: spin_lock(&space_info->lock); if (space_info->force_alloc) force = space_info->force_alloc; if (space_info->full) { spin_unlock(&space_info->lock); - goto out; + return 0; } if (!should_alloc_chunk(extent_root, space_info, alloc_bytes, force)) { spin_unlock(&space_info->lock); - goto out; + return 0; + } else if (space_info->chunk_alloc) { + wait_for_alloc = 1; + } else { + space_info->chunk_alloc = 1; } spin_unlock(&space_info->lock); + mutex_lock(&fs_info->chunk_mutex); + + /* + * The chunk_mutex is held throughout the entirety of a chunk + * allocation, so once we've acquired the chunk_mutex we know that the + * other guy is done and we need to recheck and see if we should + * allocate. + */ + if (wait_for_alloc) { + mutex_unlock(&fs_info->chunk_mutex); + wait_for_alloc = 0; + goto again; + } + /* * If we have mixed data/metadata chunks we want to make sure we keep * allocating mixed chunks instead of individual chunks. @@ -3372,9 +3391,10 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans, space_info->full = 1; else ret = 1; + space_info->force_alloc = CHUNK_ALLOC_NO_FORCE; + space_info->chunk_alloc = 0; spin_unlock(&space_info->lock); -out: mutex_unlock(&extent_root->fs_info->chunk_mutex); return ret; } -- cgit v1.2.2 From 34d52cb6c50b5a43901709998f59fb1c5a43dc4a Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Tue, 29 Mar 2011 13:46:06 +0800 Subject: Btrfs: Make free space cache code generic So we can re-use the code to cache free inode numbers. The change is quite straightforward. Two new structures are introduced. - struct btrfs_free_space_ctl We move those variables that are used for caching free space from struct btrfs_block_group_cache to this new struct. - struct btrfs_free_space_op We do block group specific work (e.g. calculation of extents threshold) through functions registered in this struct. And then we can remove references to struct btrfs_block_group_cache. Signed-off-by: Li Zefan --- fs/btrfs/extent-tree.c | 37 +++++++++++++++++++------------------ 1 file changed, 19 insertions(+), 18 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 31f33ba56fe8..904eae10ec65 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -105,6 +105,7 @@ void btrfs_put_block_group(struct btrfs_block_group_cache *cache) WARN_ON(cache->pinned > 0); WARN_ON(cache->reserved > 0); WARN_ON(cache->reserved_pinned > 0); + kfree(cache->free_space_ctl); kfree(cache); } } @@ -4893,7 +4894,7 @@ wait_block_group_cache_progress(struct btrfs_block_group_cache *cache, return 0; wait_event(caching_ctl->wait, block_group_cache_done(cache) || - (cache->free_space >= num_bytes)); + (cache->free_space_ctl->free_space >= num_bytes)); put_caching_control(caching_ctl); return 0; @@ -8551,10 +8552,16 @@ int btrfs_read_block_groups(struct btrfs_root *root) ret = -ENOMEM; goto error; } + cache->free_space_ctl = kzalloc(sizeof(*cache->free_space_ctl), + GFP_NOFS); + if (!cache->free_space_ctl) { + kfree(cache); + ret = -ENOMEM; + goto error; + } atomic_set(&cache->count, 1); spin_lock_init(&cache->lock); - spin_lock_init(&cache->tree_lock); cache->fs_info = info; INIT_LIST_HEAD(&cache->list); INIT_LIST_HEAD(&cache->cluster_list); @@ -8562,14 +8569,6 @@ int btrfs_read_block_groups(struct btrfs_root *root) if (need_clear) cache->disk_cache_state = BTRFS_DC_CLEAR; - /* - * we only want to have 32k of ram per block group for keeping - * track of free space, and if we pass 1/2 of that we want to - * start converting things over to using bitmaps - */ - cache->extents_thresh = ((1024 * 32) / 2) / - sizeof(struct btrfs_free_space); - read_extent_buffer(leaf, &cache->item, btrfs_item_ptr_offset(leaf, path->slots[0]), sizeof(cache->item)); @@ -8580,6 +8579,8 @@ int btrfs_read_block_groups(struct btrfs_root *root) cache->flags = btrfs_block_group_flags(&cache->item); cache->sectorsize = root->sectorsize; + btrfs_init_free_space_ctl(cache); + /* * We need to exclude the super stripes now so that the space * info has super bytes accounted for, otherwise we'll think @@ -8666,6 +8667,12 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans, cache = kzalloc(sizeof(*cache), GFP_NOFS); if (!cache) return -ENOMEM; + cache->free_space_ctl = kzalloc(sizeof(*cache->free_space_ctl), + GFP_NOFS); + if (!cache->free_space_ctl) { + kfree(cache); + return -ENOMEM; + } cache->key.objectid = chunk_offset; cache->key.offset = size; @@ -8673,19 +8680,13 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans, cache->sectorsize = root->sectorsize; cache->fs_info = root->fs_info; - /* - * we only want to have 32k of ram per block group for keeping track - * of free space, and if we pass 1/2 of that we want to start - * converting things over to using bitmaps - */ - cache->extents_thresh = ((1024 * 32) / 2) / - sizeof(struct btrfs_free_space); atomic_set(&cache->count, 1); spin_lock_init(&cache->lock); - spin_lock_init(&cache->tree_lock); INIT_LIST_HEAD(&cache->list); INIT_LIST_HEAD(&cache->cluster_list); + btrfs_init_free_space_ctl(cache); + btrfs_set_block_group_used(&cache->item, bytes_used); btrfs_set_block_group_chunk_objectid(&cache->item, chunk_objectid); cache->flags = type; -- cgit v1.2.2 From 33345d01522f8152f99dc84a3e7a1a45707f387f Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Wed, 20 Apr 2011 10:31:50 +0800 Subject: Btrfs: Always use 64bit inode number There's a potential problem in 32bit system when we exhaust 32bit inode numbers and start to allocate big inode numbers, because btrfs uses inode->i_ino in many places. So here we always use BTRFS_I(inode)->location.objectid, which is an u64 variable. There are 2 exceptions that BTRFS_I(inode)->location.objectid != inode->i_ino: the btree inode (0 vs 1) and empty subvol dirs (256 vs 2), and inode->i_ino will be used in those cases. Another reason to make this change is I'm going to use a special inode to save free ino cache, and the inode number must be > (u64)-256. Signed-off-by: Li Zefan --- fs/btrfs/extent-tree.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 904eae10ec65..a0e818cb0401 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -7009,8 +7009,8 @@ static noinline int get_new_locations(struct inode *reloc_inode, cur_pos = extent_key->objectid - offset; last_byte = extent_key->objectid + extent_key->offset; - ret = btrfs_lookup_file_extent(NULL, root, path, reloc_inode->i_ino, - cur_pos, 0); + ret = btrfs_lookup_file_extent(NULL, root, path, + btrfs_ino(reloc_inode), cur_pos, 0); if (ret < 0) goto out; if (ret > 0) { @@ -7033,7 +7033,7 @@ static noinline int get_new_locations(struct inode *reloc_inode, btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); if (found_key.offset != cur_pos || found_key.type != BTRFS_EXTENT_DATA_KEY || - found_key.objectid != reloc_inode->i_ino) + found_key.objectid != btrfs_ino(reloc_inode)) break; fi = btrfs_item_ptr(leaf, path->slots[0], @@ -7179,7 +7179,7 @@ next: break; } - if (inode && key.objectid != inode->i_ino) { + if (inode && key.objectid != btrfs_ino(inode)) { BUG_ON(extent_locked); btrfs_release_path(root, path); mutex_unlock(&inode->i_mutex); @@ -7488,7 +7488,7 @@ static noinline int invalidate_extent_cache(struct btrfs_root *root, continue; if (btrfs_file_extent_disk_bytenr(leaf, fi) == 0) continue; - if (!inode || inode->i_ino != key.objectid) { + if (!inode || btrfs_ino(inode) != key.objectid) { iput(inode); inode = btrfs_ilookup(target_root->fs_info->sb, key.objectid, target_root, 1); -- cgit v1.2.2 From 82d5902d9c681be37ffa9d70482907f9f0b7ec1f Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Wed, 20 Apr 2011 10:33:24 +0800 Subject: Btrfs: Support reading/writing on disk free ino cache This is similar to block group caching. We dedicate a special inode in fs tree to save free ino cache. At the very first time we create/delete a file after mount, the free ino cache will be loaded from disk into memory. When the fs tree is commited, the cache will be written back to disk. To keep compatibility, we check the root generation against the generation of the special inode when loading the cache, so the loading will fail if the btrfs filesystem was mounted in an older kernel before. Signed-off-by: Li Zefan --- fs/btrfs/extent-tree.c | 3 ++- 1 file changed, 2 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 a0e818cb0401..95ce8da63b28 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3145,7 +3145,8 @@ int btrfs_check_data_free_space(struct inode *inode, u64 bytes) /* make sure bytes are sectorsize aligned */ bytes = (bytes + root->sectorsize - 1) & ~((u64)root->sectorsize - 1); - if (root == root->fs_info->tree_root) { + if (root == root->fs_info->tree_root || + BTRFS_I(inode)->location.objectid == BTRFS_FREE_INO_OBJECTID) { alloc_chunk = 0; committed = 1; } -- cgit v1.2.2 From 8d413713ca744fa00cf4e05d4054d80727b84789 Mon Sep 17 00:00:00 2001 From: Tsutomu Itoh Date: Mon, 25 Apr 2011 19:43:52 -0400 Subject: Btrfs: check return value of kmalloc() The check on the return value of kmalloc() is added to some places. Signed-off-by: Tsutomu Itoh Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 31f33ba56fe8..cd52f7f556ef 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -8059,6 +8059,10 @@ static noinline int relocate_one_extent(struct btrfs_root *extent_root, u64 group_start = group->key.objectid; new_extents = kmalloc(sizeof(*new_extents), GFP_NOFS); + if (!new_extents) { + ret = -ENOMEM; + goto out; + } nr_extents = 1; ret = get_new_locations(reloc_inode, extent_key, -- cgit v1.2.2 From 62a45b60923a576170a1a0c309c240d9f40d193d Mon Sep 17 00:00:00 2001 From: David Sterba Date: Wed, 20 Apr 2011 15:52:26 +0200 Subject: btrfs: make functions static when possible Signed-off-by: David Sterba --- fs/btrfs/extent-tree.c | 6 +++--- 1 file changed, 3 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 cd52f7f556ef..7cdce82e03e7 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -94,7 +94,7 @@ static int block_group_bits(struct btrfs_block_group_cache *cache, u64 bits) return (cache->flags & bits) == bits; } -void btrfs_get_block_group(struct btrfs_block_group_cache *cache) +static void btrfs_get_block_group(struct btrfs_block_group_cache *cache) { atomic_inc(&cache->count); } @@ -3651,8 +3651,8 @@ static void block_rsv_add_bytes(struct btrfs_block_rsv *block_rsv, spin_unlock(&block_rsv->lock); } -void block_rsv_release_bytes(struct btrfs_block_rsv *block_rsv, - struct btrfs_block_rsv *dest, u64 num_bytes) +static void block_rsv_release_bytes(struct btrfs_block_rsv *block_rsv, + struct btrfs_block_rsv *dest, u64 num_bytes) { struct btrfs_space_info *space_info = block_rsv->space_info; -- cgit v1.2.2 From 172ddd60a662c4d8bf2809462866ddddd6431ea5 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Thu, 21 Apr 2011 00:48:27 +0200 Subject: btrfs: drop gfp parameter from alloc_extent_map pass GFP_NOFS directly to kmem_cache_alloc Signed-off-by: David Sterba --- fs/btrfs/extent-tree.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 7cdce82e03e7..6a3d53783d55 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -6694,7 +6694,7 @@ static noinline int relocate_data_extent(struct inode *reloc_inode, u64 start = extent_key->objectid - offset; u64 end = start + extent_key->offset - 1; - em = alloc_extent_map(GFP_NOFS); + em = alloc_extent_map(); BUG_ON(!em); em->start = start; -- cgit v1.2.2 From b3b4aa74b58bded927f579fff787fb6fa1c0393c Mon Sep 17 00:00:00 2001 From: David Sterba Date: Thu, 21 Apr 2011 01:20:15 +0200 Subject: btrfs: drop unused parameter from btrfs_release_path parameter tree root it's not used since commit 5f39d397dfbe140a14edecd4e73c34ce23c4f9ee ("Btrfs: Create extent_buffer interface for large blocksizes") Signed-off-by: David Sterba --- fs/btrfs/extent-tree.c | 44 ++++++++++++++++++++++---------------------- 1 file changed, 22 insertions(+), 22 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 6a3d53783d55..a160f11465f8 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -379,7 +379,7 @@ again: break; caching_ctl->progress = last; - btrfs_release_path(extent_root, path); + btrfs_release_path(path); up_read(&fs_info->extent_commit_sem); mutex_unlock(&caching_ctl->mutex); if (btrfs_transaction_in_commit(fs_info)) @@ -754,7 +754,7 @@ again: atomic_inc(&head->node.refs); spin_unlock(&delayed_refs->lock); - btrfs_release_path(root->fs_info->extent_root, path); + btrfs_release_path(path); mutex_lock(&head->mutex); mutex_unlock(&head->mutex); @@ -934,7 +934,7 @@ static int convert_extent_item_v0(struct btrfs_trans_handle *trans, break; } } - btrfs_release_path(root, path); + btrfs_release_path(path); if (owner < BTRFS_FIRST_FREE_OBJECTID) new_size += sizeof(*bi); @@ -1042,7 +1042,7 @@ again: return 0; #ifdef BTRFS_COMPAT_EXTENT_TREE_V0 key.type = BTRFS_EXTENT_REF_V0_KEY; - btrfs_release_path(root, path); + btrfs_release_path(path); ret = btrfs_search_slot(trans, root, &key, path, -1, 1); if (ret < 0) { err = ret; @@ -1080,7 +1080,7 @@ again: if (match_extent_data_ref(leaf, ref, root_objectid, owner, offset)) { if (recow) { - btrfs_release_path(root, path); + btrfs_release_path(path); goto again; } err = 0; @@ -1141,7 +1141,7 @@ static noinline int insert_extent_data_ref(struct btrfs_trans_handle *trans, if (match_extent_data_ref(leaf, ref, root_objectid, owner, offset)) break; - btrfs_release_path(root, path); + btrfs_release_path(path); key.offset++; ret = btrfs_insert_empty_item(trans, root, path, &key, size); @@ -1167,7 +1167,7 @@ static noinline int insert_extent_data_ref(struct btrfs_trans_handle *trans, btrfs_mark_buffer_dirty(leaf); ret = 0; fail: - btrfs_release_path(root, path); + btrfs_release_path(path); return ret; } @@ -1293,7 +1293,7 @@ static noinline int lookup_tree_block_ref(struct btrfs_trans_handle *trans, ret = -ENOENT; #ifdef BTRFS_COMPAT_EXTENT_TREE_V0 if (ret == -ENOENT && parent) { - btrfs_release_path(root, path); + btrfs_release_path(path); key.type = BTRFS_EXTENT_REF_V0_KEY; ret = btrfs_search_slot(trans, root, &key, path, -1, 1); if (ret > 0) @@ -1322,7 +1322,7 @@ static noinline int insert_tree_block_ref(struct btrfs_trans_handle *trans, } ret = btrfs_insert_empty_item(trans, root, path, &key, 0); - btrfs_release_path(root, path); + btrfs_release_path(path); return ret; } @@ -1608,7 +1608,7 @@ static int lookup_extent_backref(struct btrfs_trans_handle *trans, if (ret != -ENOENT) return ret; - btrfs_release_path(root, path); + btrfs_release_path(path); *ref_ret = NULL; if (owner < BTRFS_FIRST_FREE_OBJECTID) { @@ -1862,7 +1862,7 @@ static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans, __run_delayed_extent_op(extent_op, leaf, item); btrfs_mark_buffer_dirty(leaf); - btrfs_release_path(root->fs_info->extent_root, path); + btrfs_release_path(path); path->reada = 1; path->leave_spinning = 1; @@ -2361,7 +2361,7 @@ static noinline int check_delayed_ref(struct btrfs_trans_handle *trans, atomic_inc(&head->node.refs); spin_unlock(&delayed_refs->lock); - btrfs_release_path(root->fs_info->extent_root, path); + btrfs_release_path(path); mutex_lock(&head->mutex); mutex_unlock(&head->mutex); @@ -2732,7 +2732,7 @@ static int write_one_cache_group(struct btrfs_trans_handle *trans, bi = btrfs_item_ptr_offset(leaf, path->slots[0]); write_extent_buffer(leaf, &cache->item, bi, sizeof(cache->item)); btrfs_mark_buffer_dirty(leaf); - btrfs_release_path(extent_root, path); + btrfs_release_path(path); fail: if (ret) return ret; @@ -2785,7 +2785,7 @@ again: inode = lookup_free_space_inode(root, block_group, path); if (IS_ERR(inode) && PTR_ERR(inode) != -ENOENT) { ret = PTR_ERR(inode); - btrfs_release_path(root, path); + btrfs_release_path(path); goto out; } @@ -2854,7 +2854,7 @@ again: out_put: iput(inode); out_free: - btrfs_release_path(root, path); + btrfs_release_path(path); out: spin_lock(&block_group->lock); block_group->disk_cache_state = dcs; @@ -4541,7 +4541,7 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans, NULL, refs_to_drop, is_data); BUG_ON(ret); - btrfs_release_path(extent_root, path); + btrfs_release_path(path); path->leave_spinning = 1; key.objectid = bytenr; @@ -4580,7 +4580,7 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans, owner_objectid, 0); BUG_ON(ret < 0); - btrfs_release_path(extent_root, path); + btrfs_release_path(path); path->leave_spinning = 1; key.objectid = bytenr; @@ -4650,7 +4650,7 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans, ret = btrfs_del_items(trans, extent_root, path, path->slots[0], num_to_del); BUG_ON(ret); - btrfs_release_path(extent_root, path); + btrfs_release_path(path); if (is_data) { ret = btrfs_del_csums(trans, root, bytenr, num_bytes); @@ -6480,7 +6480,7 @@ int btrfs_drop_snapshot(struct btrfs_root *root, trans->block_rsv = block_rsv; } } - btrfs_release_path(root, path); + btrfs_release_path(path); BUG_ON(err); ret = btrfs_del_root(trans, tree_root, &root->root_key); @@ -8580,7 +8580,7 @@ int btrfs_read_block_groups(struct btrfs_root *root) memcpy(&cache->key, &found_key, sizeof(found_key)); key.objectid = found_key.objectid + found_key.offset; - btrfs_release_path(root, path); + btrfs_release_path(path); cache->flags = btrfs_block_group_flags(&cache->item); cache->sectorsize = root->sectorsize; @@ -8802,12 +8802,12 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans, if (ret < 0) goto out; if (ret > 0) - btrfs_release_path(tree_root, path); + btrfs_release_path(path); if (ret == 0) { ret = btrfs_del_item(trans, tree_root, path); if (ret) goto out; - btrfs_release_path(tree_root, path); + btrfs_release_path(path); } spin_lock(&root->fs_info->block_group_cache_lock); -- cgit v1.2.2 From 8cc33e5c19bf01c7617608669be8c1b4f663eb2f Mon Sep 17 00:00:00 2001 From: David Sterba Date: Mon, 2 May 2011 15:29:25 +0200 Subject: btrfs: Document a mutex lock/unlock sequence --- fs/btrfs/extent-tree.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index a160f11465f8..fba1348cb2a0 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -756,6 +756,10 @@ again: btrfs_release_path(path); + /* + * Mutex was contended, block until it's released and try + * again + */ mutex_lock(&head->mutex); mutex_unlock(&head->mutex); btrfs_put_delayed_ref(&head->node); @@ -2297,6 +2301,10 @@ again: atomic_inc(&ref->refs); spin_unlock(&delayed_refs->lock); + /* + * Mutex was contended, block until it's + * released and try again + */ mutex_lock(&head->mutex); mutex_unlock(&head->mutex); @@ -2363,6 +2371,10 @@ static noinline int check_delayed_ref(struct btrfs_trans_handle *trans, btrfs_release_path(path); + /* + * Mutex was contended, block until it's released and let + * caller try again + */ mutex_lock(&head->mutex); mutex_unlock(&head->mutex); btrfs_put_delayed_ref(&head->node); -- cgit v1.2.2 From 182608c8294b5fe90d7bbd4b026c82bf0a24b736 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Thu, 5 May 2011 13:13:16 +0200 Subject: btrfs: remove old unused commented out code Remove code which has been #if0-ed out for a very long time and does not seem to be related to current codebase anymore. Signed-off-by: David Sterba --- fs/btrfs/extent-tree.c | 1661 +----------------------------------------------- 1 file changed, 1 insertion(+), 1660 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index fba1348cb2a0..b457f195636e 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -2522,126 +2522,6 @@ out: return ret; } -#if 0 -int btrfs_cache_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root, - struct extent_buffer *buf, u32 nr_extents) -{ - struct btrfs_key key; - struct btrfs_file_extent_item *fi; - u64 root_gen; - u32 nritems; - int i; - int level; - int ret = 0; - int shared = 0; - - if (!root->ref_cows) - return 0; - - if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID) { - shared = 0; - root_gen = root->root_key.offset; - } else { - shared = 1; - root_gen = trans->transid - 1; - } - - level = btrfs_header_level(buf); - nritems = btrfs_header_nritems(buf); - - if (level == 0) { - struct btrfs_leaf_ref *ref; - struct btrfs_extent_info *info; - - ref = btrfs_alloc_leaf_ref(root, nr_extents); - if (!ref) { - ret = -ENOMEM; - goto out; - } - - ref->root_gen = root_gen; - ref->bytenr = buf->start; - ref->owner = btrfs_header_owner(buf); - ref->generation = btrfs_header_generation(buf); - ref->nritems = nr_extents; - info = ref->extents; - - for (i = 0; nr_extents > 0 && i < nritems; i++) { - u64 disk_bytenr; - btrfs_item_key_to_cpu(buf, &key, i); - if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY) - continue; - fi = btrfs_item_ptr(buf, i, - struct btrfs_file_extent_item); - if (btrfs_file_extent_type(buf, fi) == - BTRFS_FILE_EXTENT_INLINE) - continue; - disk_bytenr = btrfs_file_extent_disk_bytenr(buf, fi); - if (disk_bytenr == 0) - continue; - - info->bytenr = disk_bytenr; - info->num_bytes = - btrfs_file_extent_disk_num_bytes(buf, fi); - info->objectid = key.objectid; - info->offset = key.offset; - info++; - } - - ret = btrfs_add_leaf_ref(root, ref, shared); - if (ret == -EEXIST && shared) { - struct btrfs_leaf_ref *old; - old = btrfs_lookup_leaf_ref(root, ref->bytenr); - BUG_ON(!old); - btrfs_remove_leaf_ref(root, old); - btrfs_free_leaf_ref(root, old); - ret = btrfs_add_leaf_ref(root, ref, shared); - } - WARN_ON(ret); - btrfs_free_leaf_ref(root, ref); - } -out: - return ret; -} - -/* when a block goes through cow, we update the reference counts of - * everything that block points to. The internal pointers of the block - * can be in just about any order, and it is likely to have clusters of - * things that are close together and clusters of things that are not. - * - * To help reduce the seeks that come with updating all of these reference - * counts, sort them by byte number before actual updates are done. - * - * struct refsort is used to match byte number to slot in the btree block. - * we sort based on the byte number and then use the slot to actually - * find the item. - * - * struct refsort is smaller than strcut btrfs_item and smaller than - * struct btrfs_key_ptr. Since we're currently limited to the page size - * for a btree block, there's no way for a kmalloc of refsorts for a - * single node to be bigger than a page. - */ -struct refsort { - u64 bytenr; - u32 slot; -}; - -/* - * for passing into sort() - */ -static int refsort_cmp(const void *a_void, const void *b_void) -{ - const struct refsort *a = a_void; - const struct refsort *b = b_void; - - if (a->bytenr < b->bytenr) - return -1; - if (a->bytenr > b->bytenr) - return 1; - return 0; -} -#endif - static int __btrfs_mod_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_buffer *buf, @@ -3223,18 +3103,6 @@ commit_trans: goto again; } -#if 0 /* I hope we never need this code again, just in case */ - printk(KERN_ERR "no space left, need %llu, %llu bytes_used, " - "%llu bytes_reserved, " "%llu bytes_pinned, " - "%llu bytes_readonly, %llu may use %llu total\n", - (unsigned long long)bytes, - (unsigned long long)data_sinfo->bytes_used, - (unsigned long long)data_sinfo->bytes_reserved, - (unsigned long long)data_sinfo->bytes_pinned, - (unsigned long long)data_sinfo->bytes_readonly, - (unsigned long long)data_sinfo->bytes_may_use, - (unsigned long long)data_sinfo->total_bytes); -#endif return -ENOSPC; } data_sinfo->bytes_may_use += bytes; @@ -3867,23 +3735,7 @@ static u64 calc_global_metadata_size(struct btrfs_fs_info *fs_info) u64 meta_used; u64 data_used; int csum_size = btrfs_super_csum_size(&fs_info->super_copy); -#if 0 - /* - * per tree used space accounting can be inaccuracy, so we - * can't rely on it. - */ - spin_lock(&fs_info->extent_root->accounting_lock); - num_bytes = btrfs_root_used(&fs_info->extent_root->root_item); - spin_unlock(&fs_info->extent_root->accounting_lock); - spin_lock(&fs_info->csum_root->accounting_lock); - num_bytes += btrfs_root_used(&fs_info->csum_root->root_item); - spin_unlock(&fs_info->csum_root->accounting_lock); - - spin_lock(&fs_info->tree_root->accounting_lock); - num_bytes += btrfs_root_used(&fs_info->tree_root->root_item); - spin_unlock(&fs_info->tree_root->accounting_lock); -#endif sinfo = __find_space_info(fs_info, BTRFS_BLOCK_GROUP_DATA); spin_lock(&sinfo->lock); data_used = sinfo->bytes_used; @@ -3936,10 +3788,7 @@ static void update_global_block_rsv(struct btrfs_fs_info *fs_info) block_rsv->reserved = block_rsv->size; block_rsv->full = 1; } -#if 0 - printk(KERN_INFO"global block rsv size %llu reserved %llu\n", - block_rsv->size, block_rsv->reserved); -#endif + spin_unlock(&sinfo->lock); spin_unlock(&block_rsv->lock); } @@ -6596,1514 +6445,6 @@ int btrfs_drop_subtree(struct btrfs_trans_handle *trans, return ret; } -#if 0 -static unsigned long calc_ra(unsigned long start, unsigned long last, - unsigned long nr) -{ - return min(last, start + nr - 1); -} - -static noinline int relocate_inode_pages(struct inode *inode, u64 start, - u64 len) -{ - u64 page_start; - u64 page_end; - unsigned long first_index; - unsigned long last_index; - unsigned long i; - struct page *page; - struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; - struct file_ra_state *ra; - struct btrfs_ordered_extent *ordered; - unsigned int total_read = 0; - unsigned int total_dirty = 0; - int ret = 0; - - ra = kzalloc(sizeof(*ra), GFP_NOFS); - if (!ra) - return -ENOMEM; - - mutex_lock(&inode->i_mutex); - first_index = start >> PAGE_CACHE_SHIFT; - last_index = (start + len - 1) >> PAGE_CACHE_SHIFT; - - /* make sure the dirty trick played by the caller work */ - ret = invalidate_inode_pages2_range(inode->i_mapping, - first_index, last_index); - if (ret) - goto out_unlock; - - file_ra_state_init(ra, inode->i_mapping); - - for (i = first_index ; i <= last_index; i++) { - if (total_read % ra->ra_pages == 0) { - btrfs_force_ra(inode->i_mapping, ra, NULL, i, - calc_ra(i, last_index, ra->ra_pages)); - } - total_read++; -again: - if (((u64)i << PAGE_CACHE_SHIFT) > i_size_read(inode)) - BUG_ON(1); - page = grab_cache_page(inode->i_mapping, i); - if (!page) { - ret = -ENOMEM; - goto out_unlock; - } - if (!PageUptodate(page)) { - btrfs_readpage(NULL, page); - lock_page(page); - if (!PageUptodate(page)) { - unlock_page(page); - page_cache_release(page); - ret = -EIO; - goto out_unlock; - } - } - wait_on_page_writeback(page); - - page_start = (u64)page->index << PAGE_CACHE_SHIFT; - page_end = page_start + PAGE_CACHE_SIZE - 1; - lock_extent(io_tree, page_start, page_end, GFP_NOFS); - - ordered = btrfs_lookup_ordered_extent(inode, page_start); - if (ordered) { - unlock_extent(io_tree, page_start, page_end, GFP_NOFS); - unlock_page(page); - page_cache_release(page); - btrfs_start_ordered_extent(inode, ordered, 1); - btrfs_put_ordered_extent(ordered); - goto again; - } - set_page_extent_mapped(page); - - if (i == first_index) - set_extent_bits(io_tree, page_start, page_end, - EXTENT_BOUNDARY, GFP_NOFS); - btrfs_set_extent_delalloc(inode, page_start, page_end); - - set_page_dirty(page); - total_dirty++; - - unlock_extent(io_tree, page_start, page_end, GFP_NOFS); - unlock_page(page); - page_cache_release(page); - } - -out_unlock: - kfree(ra); - mutex_unlock(&inode->i_mutex); - balance_dirty_pages_ratelimited_nr(inode->i_mapping, total_dirty); - return ret; -} - -static noinline int relocate_data_extent(struct inode *reloc_inode, - struct btrfs_key *extent_key, - u64 offset) -{ - struct btrfs_root *root = BTRFS_I(reloc_inode)->root; - struct extent_map_tree *em_tree = &BTRFS_I(reloc_inode)->extent_tree; - struct extent_map *em; - u64 start = extent_key->objectid - offset; - u64 end = start + extent_key->offset - 1; - - em = alloc_extent_map(); - BUG_ON(!em); - - em->start = start; - em->len = extent_key->offset; - em->block_len = extent_key->offset; - em->block_start = extent_key->objectid; - em->bdev = root->fs_info->fs_devices->latest_bdev; - set_bit(EXTENT_FLAG_PINNED, &em->flags); - - /* setup extent map to cheat btrfs_readpage */ - lock_extent(&BTRFS_I(reloc_inode)->io_tree, start, end, GFP_NOFS); - while (1) { - int ret; - write_lock(&em_tree->lock); - ret = add_extent_mapping(em_tree, em); - write_unlock(&em_tree->lock); - if (ret != -EEXIST) { - free_extent_map(em); - break; - } - btrfs_drop_extent_cache(reloc_inode, start, end, 0); - } - unlock_extent(&BTRFS_I(reloc_inode)->io_tree, start, end, GFP_NOFS); - - return relocate_inode_pages(reloc_inode, start, extent_key->offset); -} - -struct btrfs_ref_path { - u64 extent_start; - u64 nodes[BTRFS_MAX_LEVEL]; - u64 root_objectid; - u64 root_generation; - u64 owner_objectid; - u32 num_refs; - int lowest_level; - int current_level; - int shared_level; - - struct btrfs_key node_keys[BTRFS_MAX_LEVEL]; - u64 new_nodes[BTRFS_MAX_LEVEL]; -}; - -struct disk_extent { - u64 ram_bytes; - u64 disk_bytenr; - u64 disk_num_bytes; - u64 offset; - u64 num_bytes; - u8 compression; - u8 encryption; - u16 other_encoding; -}; - -static int is_cowonly_root(u64 root_objectid) -{ - if (root_objectid == BTRFS_ROOT_TREE_OBJECTID || - root_objectid == BTRFS_EXTENT_TREE_OBJECTID || - root_objectid == BTRFS_CHUNK_TREE_OBJECTID || - root_objectid == BTRFS_DEV_TREE_OBJECTID || - root_objectid == BTRFS_TREE_LOG_OBJECTID || - root_objectid == BTRFS_CSUM_TREE_OBJECTID) - return 1; - return 0; -} - -static noinline int __next_ref_path(struct btrfs_trans_handle *trans, - struct btrfs_root *extent_root, - struct btrfs_ref_path *ref_path, - int first_time) -{ - struct extent_buffer *leaf; - struct btrfs_path *path; - struct btrfs_extent_ref *ref; - struct btrfs_key key; - struct btrfs_key found_key; - u64 bytenr; - u32 nritems; - int level; - int ret = 1; - - path = btrfs_alloc_path(); - if (!path) - return -ENOMEM; - - if (first_time) { - ref_path->lowest_level = -1; - ref_path->current_level = -1; - ref_path->shared_level = -1; - goto walk_up; - } -walk_down: - level = ref_path->current_level - 1; - while (level >= -1) { - u64 parent; - if (level < ref_path->lowest_level) - break; - - if (level >= 0) - bytenr = ref_path->nodes[level]; - else - bytenr = ref_path->extent_start; - BUG_ON(bytenr == 0); - - parent = ref_path->nodes[level + 1]; - ref_path->nodes[level + 1] = 0; - ref_path->current_level = level; - BUG_ON(parent == 0); - - key.objectid = bytenr; - key.offset = parent + 1; - key.type = BTRFS_EXTENT_REF_KEY; - - ret = btrfs_search_slot(trans, extent_root, &key, path, 0, 0); - if (ret < 0) - goto out; - BUG_ON(ret == 0); - - leaf = path->nodes[0]; - nritems = btrfs_header_nritems(leaf); - if (path->slots[0] >= nritems) { - ret = btrfs_next_leaf(extent_root, path); - if (ret < 0) - goto out; - if (ret > 0) - goto next; - leaf = path->nodes[0]; - } - - btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); - if (found_key.objectid == bytenr && - found_key.type == BTRFS_EXTENT_REF_KEY) { - if (level < ref_path->shared_level) - ref_path->shared_level = level; - goto found; - } -next: - level--; - btrfs_release_path(extent_root, path); - cond_resched(); - } - /* reached lowest level */ - ret = 1; - goto out; -walk_up: - level = ref_path->current_level; - while (level < BTRFS_MAX_LEVEL - 1) { - u64 ref_objectid; - - if (level >= 0) - bytenr = ref_path->nodes[level]; - else - bytenr = ref_path->extent_start; - - BUG_ON(bytenr == 0); - - key.objectid = bytenr; - key.offset = 0; - key.type = BTRFS_EXTENT_REF_KEY; - - ret = btrfs_search_slot(trans, extent_root, &key, path, 0, 0); - if (ret < 0) - goto out; - - leaf = path->nodes[0]; - nritems = btrfs_header_nritems(leaf); - if (path->slots[0] >= nritems) { - ret = btrfs_next_leaf(extent_root, path); - if (ret < 0) - goto out; - if (ret > 0) { - /* the extent was freed by someone */ - if (ref_path->lowest_level == level) - goto out; - btrfs_release_path(extent_root, path); - goto walk_down; - } - leaf = path->nodes[0]; - } - - btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); - if (found_key.objectid != bytenr || - found_key.type != BTRFS_EXTENT_REF_KEY) { - /* the extent was freed by someone */ - if (ref_path->lowest_level == level) { - ret = 1; - goto out; - } - btrfs_release_path(extent_root, path); - goto walk_down; - } -found: - ref = btrfs_item_ptr(leaf, path->slots[0], - struct btrfs_extent_ref); - ref_objectid = btrfs_ref_objectid(leaf, ref); - if (ref_objectid < BTRFS_FIRST_FREE_OBJECTID) { - if (first_time) { - level = (int)ref_objectid; - BUG_ON(level >= BTRFS_MAX_LEVEL); - ref_path->lowest_level = level; - ref_path->current_level = level; - ref_path->nodes[level] = bytenr; - } else { - WARN_ON(ref_objectid != level); - } - } else { - WARN_ON(level != -1); - } - first_time = 0; - - if (ref_path->lowest_level == level) { - ref_path->owner_objectid = ref_objectid; - ref_path->num_refs = btrfs_ref_num_refs(leaf, ref); - } - - /* - * the block is tree root or the block isn't in reference - * counted tree. - */ - if (found_key.objectid == found_key.offset || - is_cowonly_root(btrfs_ref_root(leaf, ref))) { - ref_path->root_objectid = btrfs_ref_root(leaf, ref); - ref_path->root_generation = - btrfs_ref_generation(leaf, ref); - if (level < 0) { - /* special reference from the tree log */ - ref_path->nodes[0] = found_key.offset; - ref_path->current_level = 0; - } - ret = 0; - goto out; - } - - level++; - BUG_ON(ref_path->nodes[level] != 0); - ref_path->nodes[level] = found_key.offset; - ref_path->current_level = level; - - /* - * the reference was created in the running transaction, - * no need to continue walking up. - */ - if (btrfs_ref_generation(leaf, ref) == trans->transid) { - ref_path->root_objectid = btrfs_ref_root(leaf, ref); - ref_path->root_generation = - btrfs_ref_generation(leaf, ref); - ret = 0; - goto out; - } - - btrfs_release_path(extent_root, path); - cond_resched(); - } - /* reached max tree level, but no tree root found. */ - BUG(); -out: - btrfs_free_path(path); - return ret; -} - -static int btrfs_first_ref_path(struct btrfs_trans_handle *trans, - struct btrfs_root *extent_root, - struct btrfs_ref_path *ref_path, - u64 extent_start) -{ - memset(ref_path, 0, sizeof(*ref_path)); - ref_path->extent_start = extent_start; - - return __next_ref_path(trans, extent_root, ref_path, 1); -} - -static int btrfs_next_ref_path(struct btrfs_trans_handle *trans, - struct btrfs_root *extent_root, - struct btrfs_ref_path *ref_path) -{ - return __next_ref_path(trans, extent_root, ref_path, 0); -} - -static noinline int get_new_locations(struct inode *reloc_inode, - struct btrfs_key *extent_key, - u64 offset, int no_fragment, - struct disk_extent **extents, - int *nr_extents) -{ - struct btrfs_root *root = BTRFS_I(reloc_inode)->root; - struct btrfs_path *path; - struct btrfs_file_extent_item *fi; - struct extent_buffer *leaf; - struct disk_extent *exts = *extents; - struct btrfs_key found_key; - u64 cur_pos; - u64 last_byte; - u32 nritems; - int nr = 0; - int max = *nr_extents; - int ret; - - WARN_ON(!no_fragment && *extents); - if (!exts) { - max = 1; - exts = kmalloc(sizeof(*exts) * max, GFP_NOFS); - if (!exts) - return -ENOMEM; - } - - path = btrfs_alloc_path(); - if (!path) { - if (exts != *extents) - kfree(exts); - return -ENOMEM; - } - - cur_pos = extent_key->objectid - offset; - last_byte = extent_key->objectid + extent_key->offset; - ret = btrfs_lookup_file_extent(NULL, root, path, reloc_inode->i_ino, - cur_pos, 0); - if (ret < 0) - goto out; - if (ret > 0) { - ret = -ENOENT; - goto out; - } - - while (1) { - leaf = path->nodes[0]; - nritems = btrfs_header_nritems(leaf); - if (path->slots[0] >= nritems) { - ret = btrfs_next_leaf(root, path); - if (ret < 0) - goto out; - if (ret > 0) - break; - leaf = path->nodes[0]; - } - - btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); - if (found_key.offset != cur_pos || - found_key.type != BTRFS_EXTENT_DATA_KEY || - found_key.objectid != reloc_inode->i_ino) - break; - - fi = btrfs_item_ptr(leaf, path->slots[0], - struct btrfs_file_extent_item); - if (btrfs_file_extent_type(leaf, fi) != - BTRFS_FILE_EXTENT_REG || - btrfs_file_extent_disk_bytenr(leaf, fi) == 0) - break; - - if (nr == max) { - struct disk_extent *old = exts; - max *= 2; - exts = kzalloc(sizeof(*exts) * max, GFP_NOFS); - if (!exts) { - ret = -ENOMEM; - goto out; - } - memcpy(exts, old, sizeof(*exts) * nr); - if (old != *extents) - kfree(old); - } - - exts[nr].disk_bytenr = - btrfs_file_extent_disk_bytenr(leaf, fi); - exts[nr].disk_num_bytes = - btrfs_file_extent_disk_num_bytes(leaf, fi); - exts[nr].offset = btrfs_file_extent_offset(leaf, fi); - exts[nr].num_bytes = btrfs_file_extent_num_bytes(leaf, fi); - exts[nr].ram_bytes = btrfs_file_extent_ram_bytes(leaf, fi); - exts[nr].compression = btrfs_file_extent_compression(leaf, fi); - exts[nr].encryption = btrfs_file_extent_encryption(leaf, fi); - exts[nr].other_encoding = btrfs_file_extent_other_encoding(leaf, - fi); - BUG_ON(exts[nr].offset > 0); - BUG_ON(exts[nr].compression || exts[nr].encryption); - BUG_ON(exts[nr].num_bytes != exts[nr].disk_num_bytes); - - cur_pos += exts[nr].num_bytes; - nr++; - - if (cur_pos + offset >= last_byte) - break; - - if (no_fragment) { - ret = 1; - goto out; - } - path->slots[0]++; - } - - BUG_ON(cur_pos + offset > last_byte); - if (cur_pos + offset < last_byte) { - ret = -ENOENT; - goto out; - } - ret = 0; -out: - btrfs_free_path(path); - if (ret) { - if (exts != *extents) - kfree(exts); - } else { - *extents = exts; - *nr_extents = nr; - } - return ret; -} - -static noinline int replace_one_extent(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct btrfs_path *path, - struct btrfs_key *extent_key, - struct btrfs_key *leaf_key, - struct btrfs_ref_path *ref_path, - struct disk_extent *new_extents, - int nr_extents) -{ - struct extent_buffer *leaf; - struct btrfs_file_extent_item *fi; - struct inode *inode = NULL; - struct btrfs_key key; - u64 lock_start = 0; - u64 lock_end = 0; - u64 num_bytes; - u64 ext_offset; - u64 search_end = (u64)-1; - u32 nritems; - int nr_scaned = 0; - int extent_locked = 0; - int extent_type; - int ret; - - memcpy(&key, leaf_key, sizeof(key)); - if (ref_path->owner_objectid != BTRFS_MULTIPLE_OBJECTIDS) { - if (key.objectid < ref_path->owner_objectid || - (key.objectid == ref_path->owner_objectid && - key.type < BTRFS_EXTENT_DATA_KEY)) { - key.objectid = ref_path->owner_objectid; - key.type = BTRFS_EXTENT_DATA_KEY; - key.offset = 0; - } - } - - while (1) { - ret = btrfs_search_slot(trans, root, &key, path, 0, 1); - if (ret < 0) - goto out; - - leaf = path->nodes[0]; - nritems = btrfs_header_nritems(leaf); -next: - if (extent_locked && ret > 0) { - /* - * the file extent item was modified by someone - * before the extent got locked. - */ - unlock_extent(&BTRFS_I(inode)->io_tree, lock_start, - lock_end, GFP_NOFS); - extent_locked = 0; - } - - if (path->slots[0] >= nritems) { - if (++nr_scaned > 2) - break; - - BUG_ON(extent_locked); - ret = btrfs_next_leaf(root, path); - if (ret < 0) - goto out; - if (ret > 0) - break; - leaf = path->nodes[0]; - nritems = btrfs_header_nritems(leaf); - } - - btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); - - if (ref_path->owner_objectid != BTRFS_MULTIPLE_OBJECTIDS) { - if ((key.objectid > ref_path->owner_objectid) || - (key.objectid == ref_path->owner_objectid && - key.type > BTRFS_EXTENT_DATA_KEY) || - key.offset >= search_end) - break; - } - - if (inode && key.objectid != inode->i_ino) { - BUG_ON(extent_locked); - btrfs_release_path(root, path); - mutex_unlock(&inode->i_mutex); - iput(inode); - inode = NULL; - continue; - } - - if (key.type != BTRFS_EXTENT_DATA_KEY) { - path->slots[0]++; - ret = 1; - goto next; - } - fi = btrfs_item_ptr(leaf, path->slots[0], - struct btrfs_file_extent_item); - extent_type = btrfs_file_extent_type(leaf, fi); - if ((extent_type != BTRFS_FILE_EXTENT_REG && - extent_type != BTRFS_FILE_EXTENT_PREALLOC) || - (btrfs_file_extent_disk_bytenr(leaf, fi) != - extent_key->objectid)) { - path->slots[0]++; - ret = 1; - goto next; - } - - num_bytes = btrfs_file_extent_num_bytes(leaf, fi); - ext_offset = btrfs_file_extent_offset(leaf, fi); - - if (search_end == (u64)-1) { - search_end = key.offset - ext_offset + - btrfs_file_extent_ram_bytes(leaf, fi); - } - - if (!extent_locked) { - lock_start = key.offset; - lock_end = lock_start + num_bytes - 1; - } else { - if (lock_start > key.offset || - lock_end + 1 < key.offset + num_bytes) { - unlock_extent(&BTRFS_I(inode)->io_tree, - lock_start, lock_end, GFP_NOFS); - extent_locked = 0; - } - } - - if (!inode) { - btrfs_release_path(root, path); - - inode = btrfs_iget_locked(root->fs_info->sb, - key.objectid, root); - if (inode->i_state & I_NEW) { - BTRFS_I(inode)->root = root; - BTRFS_I(inode)->location.objectid = - key.objectid; - BTRFS_I(inode)->location.type = - BTRFS_INODE_ITEM_KEY; - BTRFS_I(inode)->location.offset = 0; - btrfs_read_locked_inode(inode); - unlock_new_inode(inode); - } - /* - * some code call btrfs_commit_transaction while - * holding the i_mutex, so we can't use mutex_lock - * here. - */ - if (is_bad_inode(inode) || - !mutex_trylock(&inode->i_mutex)) { - iput(inode); - inode = NULL; - key.offset = (u64)-1; - goto skip; - } - } - - if (!extent_locked) { - struct btrfs_ordered_extent *ordered; - - btrfs_release_path(root, path); - - lock_extent(&BTRFS_I(inode)->io_tree, lock_start, - lock_end, GFP_NOFS); - ordered = btrfs_lookup_first_ordered_extent(inode, - lock_end); - if (ordered && - ordered->file_offset <= lock_end && - ordered->file_offset + ordered->len > lock_start) { - unlock_extent(&BTRFS_I(inode)->io_tree, - lock_start, lock_end, GFP_NOFS); - btrfs_start_ordered_extent(inode, ordered, 1); - btrfs_put_ordered_extent(ordered); - key.offset += num_bytes; - goto skip; - } - if (ordered) - btrfs_put_ordered_extent(ordered); - - extent_locked = 1; - continue; - } - - if (nr_extents == 1) { - /* update extent pointer in place */ - btrfs_set_file_extent_disk_bytenr(leaf, fi, - new_extents[0].disk_bytenr); - btrfs_set_file_extent_disk_num_bytes(leaf, fi, - new_extents[0].disk_num_bytes); - btrfs_mark_buffer_dirty(leaf); - - btrfs_drop_extent_cache(inode, key.offset, - key.offset + num_bytes - 1, 0); - - ret = btrfs_inc_extent_ref(trans, root, - new_extents[0].disk_bytenr, - new_extents[0].disk_num_bytes, - leaf->start, - root->root_key.objectid, - trans->transid, - key.objectid); - BUG_ON(ret); - - ret = btrfs_free_extent(trans, root, - extent_key->objectid, - extent_key->offset, - leaf->start, - btrfs_header_owner(leaf), - btrfs_header_generation(leaf), - key.objectid, 0); - BUG_ON(ret); - - btrfs_release_path(root, path); - key.offset += num_bytes; - } else { - BUG_ON(1); -#if 0 - u64 alloc_hint; - u64 extent_len; - int i; - /* - * drop old extent pointer at first, then insert the - * new pointers one bye one - */ - btrfs_release_path(root, path); - ret = btrfs_drop_extents(trans, root, inode, key.offset, - key.offset + num_bytes, - key.offset, &alloc_hint); - BUG_ON(ret); - - for (i = 0; i < nr_extents; i++) { - if (ext_offset >= new_extents[i].num_bytes) { - ext_offset -= new_extents[i].num_bytes; - continue; - } - extent_len = min(new_extents[i].num_bytes - - ext_offset, num_bytes); - - ret = btrfs_insert_empty_item(trans, root, - path, &key, - sizeof(*fi)); - BUG_ON(ret); - - leaf = path->nodes[0]; - fi = btrfs_item_ptr(leaf, path->slots[0], - struct btrfs_file_extent_item); - btrfs_set_file_extent_generation(leaf, fi, - trans->transid); - btrfs_set_file_extent_type(leaf, fi, - BTRFS_FILE_EXTENT_REG); - btrfs_set_file_extent_disk_bytenr(leaf, fi, - new_extents[i].disk_bytenr); - btrfs_set_file_extent_disk_num_bytes(leaf, fi, - new_extents[i].disk_num_bytes); - btrfs_set_file_extent_ram_bytes(leaf, fi, - new_extents[i].ram_bytes); - - btrfs_set_file_extent_compression(leaf, fi, - new_extents[i].compression); - btrfs_set_file_extent_encryption(leaf, fi, - new_extents[i].encryption); - btrfs_set_file_extent_other_encoding(leaf, fi, - new_extents[i].other_encoding); - - btrfs_set_file_extent_num_bytes(leaf, fi, - extent_len); - ext_offset += new_extents[i].offset; - btrfs_set_file_extent_offset(leaf, fi, - ext_offset); - btrfs_mark_buffer_dirty(leaf); - - btrfs_drop_extent_cache(inode, key.offset, - key.offset + extent_len - 1, 0); - - ret = btrfs_inc_extent_ref(trans, root, - new_extents[i].disk_bytenr, - new_extents[i].disk_num_bytes, - leaf->start, - root->root_key.objectid, - trans->transid, key.objectid); - BUG_ON(ret); - btrfs_release_path(root, path); - - inode_add_bytes(inode, extent_len); - - ext_offset = 0; - num_bytes -= extent_len; - key.offset += extent_len; - - if (num_bytes == 0) - break; - } - BUG_ON(i >= nr_extents); -#endif - } - - if (extent_locked) { - unlock_extent(&BTRFS_I(inode)->io_tree, lock_start, - lock_end, GFP_NOFS); - extent_locked = 0; - } -skip: - if (ref_path->owner_objectid != BTRFS_MULTIPLE_OBJECTIDS && - key.offset >= search_end) - break; - - cond_resched(); - } - ret = 0; -out: - btrfs_release_path(root, path); - if (inode) { - mutex_unlock(&inode->i_mutex); - if (extent_locked) { - unlock_extent(&BTRFS_I(inode)->io_tree, lock_start, - lock_end, GFP_NOFS); - } - iput(inode); - } - return ret; -} - -int btrfs_reloc_tree_cache_ref(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct extent_buffer *buf, u64 orig_start) -{ - int level; - int ret; - - BUG_ON(btrfs_header_generation(buf) != trans->transid); - BUG_ON(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID); - - level = btrfs_header_level(buf); - if (level == 0) { - struct btrfs_leaf_ref *ref; - struct btrfs_leaf_ref *orig_ref; - - orig_ref = btrfs_lookup_leaf_ref(root, orig_start); - if (!orig_ref) - return -ENOENT; - - ref = btrfs_alloc_leaf_ref(root, orig_ref->nritems); - if (!ref) { - btrfs_free_leaf_ref(root, orig_ref); - return -ENOMEM; - } - - ref->nritems = orig_ref->nritems; - memcpy(ref->extents, orig_ref->extents, - sizeof(ref->extents[0]) * ref->nritems); - - btrfs_free_leaf_ref(root, orig_ref); - - ref->root_gen = trans->transid; - ref->bytenr = buf->start; - ref->owner = btrfs_header_owner(buf); - ref->generation = btrfs_header_generation(buf); - - ret = btrfs_add_leaf_ref(root, ref, 0); - WARN_ON(ret); - btrfs_free_leaf_ref(root, ref); - } - return 0; -} - -static noinline int invalidate_extent_cache(struct btrfs_root *root, - struct extent_buffer *leaf, - struct btrfs_block_group_cache *group, - struct btrfs_root *target_root) -{ - struct btrfs_key key; - struct inode *inode = NULL; - struct btrfs_file_extent_item *fi; - struct extent_state *cached_state = NULL; - u64 num_bytes; - u64 skip_objectid = 0; - u32 nritems; - u32 i; - - nritems = btrfs_header_nritems(leaf); - for (i = 0; i < nritems; i++) { - btrfs_item_key_to_cpu(leaf, &key, i); - if (key.objectid == skip_objectid || - key.type != BTRFS_EXTENT_DATA_KEY) - continue; - fi = btrfs_item_ptr(leaf, i, struct btrfs_file_extent_item); - if (btrfs_file_extent_type(leaf, fi) == - BTRFS_FILE_EXTENT_INLINE) - continue; - if (btrfs_file_extent_disk_bytenr(leaf, fi) == 0) - continue; - if (!inode || inode->i_ino != key.objectid) { - iput(inode); - inode = btrfs_ilookup(target_root->fs_info->sb, - key.objectid, target_root, 1); - } - if (!inode) { - skip_objectid = key.objectid; - continue; - } - num_bytes = btrfs_file_extent_num_bytes(leaf, fi); - - lock_extent_bits(&BTRFS_I(inode)->io_tree, key.offset, - key.offset + num_bytes - 1, 0, &cached_state, - GFP_NOFS); - btrfs_drop_extent_cache(inode, key.offset, - key.offset + num_bytes - 1, 1); - unlock_extent_cached(&BTRFS_I(inode)->io_tree, key.offset, - key.offset + num_bytes - 1, &cached_state, - GFP_NOFS); - cond_resched(); - } - iput(inode); - return 0; -} - -static noinline int replace_extents_in_leaf(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct extent_buffer *leaf, - struct btrfs_block_group_cache *group, - struct inode *reloc_inode) -{ - struct btrfs_key key; - struct btrfs_key extent_key; - struct btrfs_file_extent_item *fi; - struct btrfs_leaf_ref *ref; - struct disk_extent *new_extent; - u64 bytenr; - u64 num_bytes; - u32 nritems; - u32 i; - int ext_index; - int nr_extent; - int ret; - - new_extent = kmalloc(sizeof(*new_extent), GFP_NOFS); - if (!new_extent) - return -ENOMEM; - - ref = btrfs_lookup_leaf_ref(root, leaf->start); - BUG_ON(!ref); - - ext_index = -1; - nritems = btrfs_header_nritems(leaf); - for (i = 0; i < nritems; i++) { - btrfs_item_key_to_cpu(leaf, &key, i); - if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY) - continue; - fi = btrfs_item_ptr(leaf, i, struct btrfs_file_extent_item); - if (btrfs_file_extent_type(leaf, fi) == - BTRFS_FILE_EXTENT_INLINE) - continue; - bytenr = btrfs_file_extent_disk_bytenr(leaf, fi); - num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi); - if (bytenr == 0) - continue; - - ext_index++; - if (bytenr >= group->key.objectid + group->key.offset || - bytenr + num_bytes <= group->key.objectid) - continue; - - extent_key.objectid = bytenr; - extent_key.offset = num_bytes; - extent_key.type = BTRFS_EXTENT_ITEM_KEY; - nr_extent = 1; - ret = get_new_locations(reloc_inode, &extent_key, - group->key.objectid, 1, - &new_extent, &nr_extent); - if (ret > 0) - continue; - BUG_ON(ret < 0); - - BUG_ON(ref->extents[ext_index].bytenr != bytenr); - BUG_ON(ref->extents[ext_index].num_bytes != num_bytes); - ref->extents[ext_index].bytenr = new_extent->disk_bytenr; - ref->extents[ext_index].num_bytes = new_extent->disk_num_bytes; - - btrfs_set_file_extent_disk_bytenr(leaf, fi, - new_extent->disk_bytenr); - btrfs_set_file_extent_disk_num_bytes(leaf, fi, - new_extent->disk_num_bytes); - btrfs_mark_buffer_dirty(leaf); - - ret = btrfs_inc_extent_ref(trans, root, - new_extent->disk_bytenr, - new_extent->disk_num_bytes, - leaf->start, - root->root_key.objectid, - trans->transid, key.objectid); - BUG_ON(ret); - - ret = btrfs_free_extent(trans, root, - bytenr, num_bytes, leaf->start, - btrfs_header_owner(leaf), - btrfs_header_generation(leaf), - key.objectid, 0); - BUG_ON(ret); - cond_resched(); - } - kfree(new_extent); - BUG_ON(ext_index + 1 != ref->nritems); - btrfs_free_leaf_ref(root, ref); - return 0; -} - -int btrfs_free_reloc_root(struct btrfs_trans_handle *trans, - struct btrfs_root *root) -{ - struct btrfs_root *reloc_root; - int ret; - - if (root->reloc_root) { - reloc_root = root->reloc_root; - root->reloc_root = NULL; - list_add(&reloc_root->dead_list, - &root->fs_info->dead_reloc_roots); - - btrfs_set_root_bytenr(&reloc_root->root_item, - reloc_root->node->start); - btrfs_set_root_level(&root->root_item, - btrfs_header_level(reloc_root->node)); - memset(&reloc_root->root_item.drop_progress, 0, - sizeof(struct btrfs_disk_key)); - reloc_root->root_item.drop_level = 0; - - ret = btrfs_update_root(trans, root->fs_info->tree_root, - &reloc_root->root_key, - &reloc_root->root_item); - BUG_ON(ret); - } - return 0; -} - -int btrfs_drop_dead_reloc_roots(struct btrfs_root *root) -{ - struct btrfs_trans_handle *trans; - struct btrfs_root *reloc_root; - struct btrfs_root *prev_root = NULL; - struct list_head dead_roots; - int ret; - unsigned long nr; - - INIT_LIST_HEAD(&dead_roots); - list_splice_init(&root->fs_info->dead_reloc_roots, &dead_roots); - - while (!list_empty(&dead_roots)) { - reloc_root = list_entry(dead_roots.prev, - struct btrfs_root, dead_list); - list_del_init(&reloc_root->dead_list); - - BUG_ON(reloc_root->commit_root != NULL); - while (1) { - trans = btrfs_join_transaction(root, 1); - BUG_ON(IS_ERR(trans)); - - mutex_lock(&root->fs_info->drop_mutex); - ret = btrfs_drop_snapshot(trans, reloc_root); - if (ret != -EAGAIN) - break; - mutex_unlock(&root->fs_info->drop_mutex); - - nr = trans->blocks_used; - ret = btrfs_end_transaction(trans, root); - BUG_ON(ret); - btrfs_btree_balance_dirty(root, nr); - } - - free_extent_buffer(reloc_root->node); - - ret = btrfs_del_root(trans, root->fs_info->tree_root, - &reloc_root->root_key); - BUG_ON(ret); - mutex_unlock(&root->fs_info->drop_mutex); - - nr = trans->blocks_used; - ret = btrfs_end_transaction(trans, root); - BUG_ON(ret); - btrfs_btree_balance_dirty(root, nr); - - kfree(prev_root); - prev_root = reloc_root; - } - if (prev_root) { - btrfs_remove_leaf_refs(prev_root, (u64)-1, 0); - kfree(prev_root); - } - return 0; -} - -int btrfs_add_dead_reloc_root(struct btrfs_root *root) -{ - list_add(&root->dead_list, &root->fs_info->dead_reloc_roots); - return 0; -} - -int btrfs_cleanup_reloc_trees(struct btrfs_root *root) -{ - struct btrfs_root *reloc_root; - struct btrfs_trans_handle *trans; - struct btrfs_key location; - int found; - int ret; - - mutex_lock(&root->fs_info->tree_reloc_mutex); - ret = btrfs_find_dead_roots(root, BTRFS_TREE_RELOC_OBJECTID, NULL); - BUG_ON(ret); - found = !list_empty(&root->fs_info->dead_reloc_roots); - mutex_unlock(&root->fs_info->tree_reloc_mutex); - - if (found) { - trans = btrfs_start_transaction(root, 1); - BUG_ON(IS_ERR(trans)); - ret = btrfs_commit_transaction(trans, root); - BUG_ON(ret); - } - - location.objectid = BTRFS_DATA_RELOC_TREE_OBJECTID; - location.offset = (u64)-1; - location.type = BTRFS_ROOT_ITEM_KEY; - - reloc_root = btrfs_read_fs_root_no_name(root->fs_info, &location); - BUG_ON(!reloc_root); - ret = btrfs_orphan_cleanup(reloc_root); - BUG_ON(ret); - return 0; -} - -static noinline int init_reloc_tree(struct btrfs_trans_handle *trans, - struct btrfs_root *root) -{ - struct btrfs_root *reloc_root; - struct extent_buffer *eb; - struct btrfs_root_item *root_item; - struct btrfs_key root_key; - int ret; - - BUG_ON(!root->ref_cows); - if (root->reloc_root) - return 0; - - root_item = kmalloc(sizeof(*root_item), GFP_NOFS); - if (!root_item) - return -ENOMEM; - - ret = btrfs_copy_root(trans, root, root->commit_root, - &eb, BTRFS_TREE_RELOC_OBJECTID); - BUG_ON(ret); - - root_key.objectid = BTRFS_TREE_RELOC_OBJECTID; - root_key.offset = root->root_key.objectid; - root_key.type = BTRFS_ROOT_ITEM_KEY; - - memcpy(root_item, &root->root_item, sizeof(root_item)); - btrfs_set_root_refs(root_item, 0); - btrfs_set_root_bytenr(root_item, eb->start); - btrfs_set_root_level(root_item, btrfs_header_level(eb)); - btrfs_set_root_generation(root_item, trans->transid); - - btrfs_tree_unlock(eb); - free_extent_buffer(eb); - - ret = btrfs_insert_root(trans, root->fs_info->tree_root, - &root_key, root_item); - BUG_ON(ret); - kfree(root_item); - - reloc_root = btrfs_read_fs_root_no_radix(root->fs_info->tree_root, - &root_key); - BUG_ON(IS_ERR(reloc_root)); - reloc_root->last_trans = trans->transid; - reloc_root->commit_root = NULL; - reloc_root->ref_tree = &root->fs_info->reloc_ref_tree; - - root->reloc_root = reloc_root; - return 0; -} - -/* - * Core function of space balance. - * - * The idea is using reloc trees to relocate tree blocks in reference - * counted roots. There is one reloc tree for each subvol, and all - * reloc trees share same root key objectid. Reloc trees are snapshots - * of the latest committed roots of subvols (root->commit_root). - * - * To relocate a tree block referenced by a subvol, there are two steps. - * COW the block through subvol's reloc tree, then update block pointer - * in the subvol to point to the new block. Since all reloc trees share - * same root key objectid, doing special handing for tree blocks owned - * by them is easy. Once a tree block has been COWed in one reloc tree, - * we can use the resulting new block directly when the same block is - * required to COW again through other reloc trees. By this way, relocated - * tree blocks are shared between reloc trees, so they are also shared - * between subvols. - */ -static noinline int relocate_one_path(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct btrfs_path *path, - struct btrfs_key *first_key, - struct btrfs_ref_path *ref_path, - struct btrfs_block_group_cache *group, - struct inode *reloc_inode) -{ - struct btrfs_root *reloc_root; - struct extent_buffer *eb = NULL; - struct btrfs_key *keys; - u64 *nodes; - int level; - int shared_level; - int lowest_level = 0; - int ret; - - if (ref_path->owner_objectid < BTRFS_FIRST_FREE_OBJECTID) - lowest_level = ref_path->owner_objectid; - - if (!root->ref_cows) { - path->lowest_level = lowest_level; - ret = btrfs_search_slot(trans, root, first_key, path, 0, 1); - BUG_ON(ret < 0); - path->lowest_level = 0; - btrfs_release_path(root, path); - return 0; - } - - mutex_lock(&root->fs_info->tree_reloc_mutex); - ret = init_reloc_tree(trans, root); - BUG_ON(ret); - reloc_root = root->reloc_root; - - shared_level = ref_path->shared_level; - ref_path->shared_level = BTRFS_MAX_LEVEL - 1; - - keys = ref_path->node_keys; - nodes = ref_path->new_nodes; - memset(&keys[shared_level + 1], 0, - sizeof(*keys) * (BTRFS_MAX_LEVEL - shared_level - 1)); - memset(&nodes[shared_level + 1], 0, - sizeof(*nodes) * (BTRFS_MAX_LEVEL - shared_level - 1)); - - if (nodes[lowest_level] == 0) { - path->lowest_level = lowest_level; - ret = btrfs_search_slot(trans, reloc_root, first_key, path, - 0, 1); - BUG_ON(ret); - for (level = lowest_level; level < BTRFS_MAX_LEVEL; level++) { - eb = path->nodes[level]; - if (!eb || eb == reloc_root->node) - break; - nodes[level] = eb->start; - if (level == 0) - btrfs_item_key_to_cpu(eb, &keys[level], 0); - else - btrfs_node_key_to_cpu(eb, &keys[level], 0); - } - if (nodes[0] && - ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) { - eb = path->nodes[0]; - ret = replace_extents_in_leaf(trans, reloc_root, eb, - group, reloc_inode); - BUG_ON(ret); - } - btrfs_release_path(reloc_root, path); - } else { - ret = btrfs_merge_path(trans, reloc_root, keys, nodes, - lowest_level); - BUG_ON(ret); - } - - /* - * replace tree blocks in the fs tree with tree blocks in - * the reloc tree. - */ - ret = btrfs_merge_path(trans, root, keys, nodes, lowest_level); - BUG_ON(ret < 0); - - if (ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) { - ret = btrfs_search_slot(trans, reloc_root, first_key, path, - 0, 0); - BUG_ON(ret); - extent_buffer_get(path->nodes[0]); - eb = path->nodes[0]; - btrfs_release_path(reloc_root, path); - ret = invalidate_extent_cache(reloc_root, eb, group, root); - BUG_ON(ret); - free_extent_buffer(eb); - } - - mutex_unlock(&root->fs_info->tree_reloc_mutex); - path->lowest_level = 0; - return 0; -} - -static noinline int relocate_tree_block(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct btrfs_path *path, - struct btrfs_key *first_key, - struct btrfs_ref_path *ref_path) -{ - int ret; - - ret = relocate_one_path(trans, root, path, first_key, - ref_path, NULL, NULL); - BUG_ON(ret); - - return 0; -} - -static noinline int del_extent_zero(struct btrfs_trans_handle *trans, - struct btrfs_root *extent_root, - struct btrfs_path *path, - struct btrfs_key *extent_key) -{ - int ret; - - ret = btrfs_search_slot(trans, extent_root, extent_key, path, -1, 1); - if (ret) - goto out; - ret = btrfs_del_item(trans, extent_root, path); -out: - btrfs_release_path(extent_root, path); - return ret; -} - -static noinline struct btrfs_root *read_ref_root(struct btrfs_fs_info *fs_info, - struct btrfs_ref_path *ref_path) -{ - struct btrfs_key root_key; - - root_key.objectid = ref_path->root_objectid; - root_key.type = BTRFS_ROOT_ITEM_KEY; - if (is_cowonly_root(ref_path->root_objectid)) - root_key.offset = 0; - else - root_key.offset = (u64)-1; - - return btrfs_read_fs_root_no_name(fs_info, &root_key); -} - -static noinline int relocate_one_extent(struct btrfs_root *extent_root, - struct btrfs_path *path, - struct btrfs_key *extent_key, - struct btrfs_block_group_cache *group, - struct inode *reloc_inode, int pass) -{ - struct btrfs_trans_handle *trans; - struct btrfs_root *found_root; - struct btrfs_ref_path *ref_path = NULL; - struct disk_extent *new_extents = NULL; - int nr_extents = 0; - int loops; - int ret; - int level; - struct btrfs_key first_key; - u64 prev_block = 0; - - - trans = btrfs_start_transaction(extent_root, 1); - BUG_ON(IS_ERR(trans)); - - if (extent_key->objectid == 0) { - ret = del_extent_zero(trans, extent_root, path, extent_key); - goto out; - } - - ref_path = kmalloc(sizeof(*ref_path), GFP_NOFS); - if (!ref_path) { - ret = -ENOMEM; - goto out; - } - - for (loops = 0; ; loops++) { - if (loops == 0) { - ret = btrfs_first_ref_path(trans, extent_root, ref_path, - extent_key->objectid); - } else { - ret = btrfs_next_ref_path(trans, extent_root, ref_path); - } - if (ret < 0) - goto out; - if (ret > 0) - break; - - if (ref_path->root_objectid == BTRFS_TREE_LOG_OBJECTID || - ref_path->root_objectid == BTRFS_TREE_RELOC_OBJECTID) - continue; - - found_root = read_ref_root(extent_root->fs_info, ref_path); - BUG_ON(!found_root); - /* - * for reference counted tree, only process reference paths - * rooted at the latest committed root. - */ - if (found_root->ref_cows && - ref_path->root_generation != found_root->root_key.offset) - continue; - - if (ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) { - if (pass == 0) { - /* - * copy data extents to new locations - */ - u64 group_start = group->key.objectid; - ret = relocate_data_extent(reloc_inode, - extent_key, - group_start); - if (ret < 0) - goto out; - break; - } - level = 0; - } else { - level = ref_path->owner_objectid; - } - - if (prev_block != ref_path->nodes[level]) { - struct extent_buffer *eb; - u64 block_start = ref_path->nodes[level]; - u64 block_size = btrfs_level_size(found_root, level); - - eb = read_tree_block(found_root, block_start, - block_size, 0); - if (!eb) { - ret = -EIO; - goto out; - } - btrfs_tree_lock(eb); - BUG_ON(level != btrfs_header_level(eb)); - - if (level == 0) - btrfs_item_key_to_cpu(eb, &first_key, 0); - else - btrfs_node_key_to_cpu(eb, &first_key, 0); - - btrfs_tree_unlock(eb); - free_extent_buffer(eb); - prev_block = block_start; - } - - mutex_lock(&extent_root->fs_info->trans_mutex); - btrfs_record_root_in_trans(found_root); - mutex_unlock(&extent_root->fs_info->trans_mutex); - if (ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) { - /* - * try to update data extent references while - * keeping metadata shared between snapshots. - */ - if (pass == 1) { - ret = relocate_one_path(trans, found_root, - path, &first_key, ref_path, - group, reloc_inode); - if (ret < 0) - goto out; - continue; - } - /* - * use fallback method to process the remaining - * references. - */ - if (!new_extents) { - u64 group_start = group->key.objectid; - new_extents = kmalloc(sizeof(*new_extents), - GFP_NOFS); - if (!new_extents) { - ret = -ENOMEM; - goto out; - } - nr_extents = 1; - ret = get_new_locations(reloc_inode, - extent_key, - group_start, 1, - &new_extents, - &nr_extents); - if (ret) - goto out; - } - ret = replace_one_extent(trans, found_root, - path, extent_key, - &first_key, ref_path, - new_extents, nr_extents); - } else { - ret = relocate_tree_block(trans, found_root, path, - &first_key, ref_path); - } - if (ret < 0) - goto out; - } - ret = 0; -out: - btrfs_end_transaction(trans, extent_root); - kfree(new_extents); - kfree(ref_path); - return ret; -} -#endif - static u64 update_block_group_flags(struct btrfs_root *root, u64 flags) { u64 num_devices; -- cgit v1.2.2 From 1aba86d67f340a8001d67183ec32e8a62e3ec658 Mon Sep 17 00:00:00 2001 From: liubo Date: Fri, 8 Apr 2011 08:44:37 +0000 Subject: Btrfs: fix easily get into ENOSPC in mixed case When a btrfs disk is created by mixed data & metadata option, it will have no pure data or pure metadata space info. In btrfs's for-linus branch, commit 78b1ea13838039cd88afdd62519b40b344d6c920 (Btrfs: fix OOPS of empty filesystem after balance) initializes space infos at the very beginning. The problem is this initialization does not take the mixed case into account, which will cause btrfs will easily get into ENOSPC in mixed case. Signed-off-by: Liu Bo Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 37 ++++++++++++++++++++++++++----------- 1 file changed, 26 insertions(+), 11 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index cd52f7f556ef..9ee6bd55e16c 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -8856,23 +8856,38 @@ out: int btrfs_init_space_info(struct btrfs_fs_info *fs_info) { struct btrfs_space_info *space_info; + struct btrfs_super_block *disk_super; + u64 features; + u64 flags; + int mixed = 0; int ret; - ret = update_space_info(fs_info, BTRFS_BLOCK_GROUP_SYSTEM, 0, 0, - &space_info); - if (ret) - return ret; + disk_super = &fs_info->super_copy; + if (!btrfs_super_root(disk_super)) + return 1; - ret = update_space_info(fs_info, BTRFS_BLOCK_GROUP_METADATA, 0, 0, - &space_info); - if (ret) - return ret; + features = btrfs_super_incompat_flags(disk_super); + if (features & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS) + mixed = 1; - ret = update_space_info(fs_info, BTRFS_BLOCK_GROUP_DATA, 0, 0, - &space_info); + flags = BTRFS_BLOCK_GROUP_SYSTEM; + ret = update_space_info(fs_info, flags, 0, 0, &space_info); if (ret) - return ret; + goto out; + if (mixed) { + flags = BTRFS_BLOCK_GROUP_METADATA | BTRFS_BLOCK_GROUP_DATA; + ret = update_space_info(fs_info, flags, 0, 0, &space_info); + } else { + flags = BTRFS_BLOCK_GROUP_METADATA; + ret = update_space_info(fs_info, flags, 0, 0, &space_info); + if (ret) + goto out; + + flags = BTRFS_BLOCK_GROUP_DATA; + ret = update_space_info(fs_info, flags, 0, 0, &space_info); + } +out: return ret; } -- cgit v1.2.2 From 16cdcec736cd214350cdb591bf1091f8beedefa0 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Fri, 22 Apr 2011 18:12:22 +0800 Subject: btrfs: implement delayed inode items operation Changelog V5 -> V6: - Fix oom when the memory load is high, by storing the delayed nodes into the root's radix tree, and letting btrfs inodes go. Changelog V4 -> V5: - Fix the race on adding the delayed node to the inode, which is spotted by Chris Mason. - Merge Chris Mason's incremental patch into this patch. - Fix deadlock between readdir() and memory fault, which is reported by Itaru Kitayama. Changelog V3 -> V4: - Fix nested lock, which is reported by Itaru Kitayama, by updating space cache inode in time. Changelog V2 -> V3: - Fix the race between the delayed worker and the task which does delayed items balance, which is reported by Tsutomu Itoh. - Modify the patch address David Sterba's comment. - Fix the bug of the cpu recursion spinlock, reported by Chris Mason Changelog V1 -> V2: - break up the global rb-tree, use a list to manage the delayed nodes, which is created for every directory and file, and used to manage the delayed directory name index items and the delayed inode item. - introduce a worker to deal with the delayed nodes. Compare with Ext3/4, the performance of file creation and deletion on btrfs is very poor. the reason is that btrfs must do a lot of b+ tree insertions, such as inode item, directory name item, directory name index and so on. If we can do some delayed b+ tree insertion or deletion, we can improve the performance, so we made this patch which implemented delayed directory name index insertion/deletion and delayed inode update. Implementation: - introduce a delayed root object into the filesystem, that use two lists to manage the delayed nodes which are created for every file/directory. One is used to manage all the delayed nodes that have delayed items. And the other is used to manage the delayed nodes which is waiting to be dealt with by the work thread. - Every delayed node has two rb-tree, one is used to manage the directory name index which is going to be inserted into b+ tree, and the other is used to manage the directory name index which is going to be deleted from b+ tree. - introduce a worker to deal with the delayed operation. This worker is used to deal with the works of the delayed directory name index items insertion and deletion and the delayed inode update. When the delayed items is beyond the lower limit, we create works for some delayed nodes and insert them into the work queue of the worker, and then go back. When the delayed items is beyond the upper bound, we create works for all the delayed nodes that haven't been dealt with, and insert them into the work queue of the worker, and then wait for that the untreated items is below some threshold value. - When we want to insert a directory name index into b+ tree, we just add the information into the delayed inserting rb-tree. And then we check the number of the delayed items and do delayed items balance. (The balance policy is above.) - When we want to delete a directory name index from the b+ tree, we search it in the inserting rb-tree at first. If we look it up, just drop it. If not, add the key of it into the delayed deleting rb-tree. Similar to the delayed inserting rb-tree, we also check the number of the delayed items and do delayed items balance. (The same to inserting manipulation) - When we want to update the metadata of some inode, we cached the data of the inode into the delayed node. the worker will flush it into the b+ tree after dealing with the delayed insertion and deletion. - We will move the delayed node to the tail of the list after we access the delayed node, By this way, we can cache more delayed items and merge more inode updates. - If we want to commit transaction, we will deal with all the delayed node. - the delayed node will be freed when we free the btrfs inode. - Before we log the inode items, we commit all the directory name index items and the delayed inode update. I did a quick test by the benchmark tool[1] and found we can improve the performance of file creation by ~15%, and file deletion by ~20%. Before applying this patch: Create files: Total files: 50000 Total time: 1.096108 Average time: 0.000022 Delete files: Total files: 50000 Total time: 1.510403 Average time: 0.000030 After applying this patch: Create files: Total files: 50000 Total time: 0.932899 Average time: 0.000019 Delete files: Total files: 50000 Total time: 1.215732 Average time: 0.000024 [1] http://marc.info/?l=linux-btrfs&m=128212635122920&q=p3 Many thanks for Kitayama-san's help! Signed-off-by: Miao Xie Reviewed-by: David Sterba Tested-by: Tsutomu Itoh Tested-by: Itaru Kitayama Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 18 ++++++------------ 1 file changed, 6 insertions(+), 12 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 9ee6bd55e16c..7b0433866f36 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3973,12 +3973,6 @@ static void release_global_block_rsv(struct btrfs_fs_info *fs_info) WARN_ON(fs_info->chunk_block_rsv.reserved > 0); } -static u64 calc_trans_metadata_size(struct btrfs_root *root, int num_items) -{ - return (root->leafsize + root->nodesize * (BTRFS_MAX_LEVEL - 1)) * - 3 * num_items; -} - int btrfs_trans_reserve_metadata(struct btrfs_trans_handle *trans, struct btrfs_root *root, int num_items) @@ -3989,7 +3983,7 @@ int btrfs_trans_reserve_metadata(struct btrfs_trans_handle *trans, if (num_items == 0 || root->fs_info->chunk_root == root) return 0; - num_bytes = calc_trans_metadata_size(root, num_items); + num_bytes = btrfs_calc_trans_metadata_size(root, num_items); ret = btrfs_block_rsv_add(trans, root, &root->fs_info->trans_block_rsv, num_bytes); if (!ret) { @@ -4028,14 +4022,14 @@ int btrfs_orphan_reserve_metadata(struct btrfs_trans_handle *trans, * If all of the metadata space is used, we can commit * transaction and use space it freed. */ - u64 num_bytes = calc_trans_metadata_size(root, 4); + u64 num_bytes = btrfs_calc_trans_metadata_size(root, 4); return block_rsv_migrate_bytes(src_rsv, dst_rsv, num_bytes); } void btrfs_orphan_release_metadata(struct inode *inode) { struct btrfs_root *root = BTRFS_I(inode)->root; - u64 num_bytes = calc_trans_metadata_size(root, 4); + u64 num_bytes = btrfs_calc_trans_metadata_size(root, 4); btrfs_block_rsv_release(root, root->orphan_block_rsv, num_bytes); } @@ -4049,7 +4043,7 @@ int btrfs_snap_reserve_metadata(struct btrfs_trans_handle *trans, * two for root back/forward refs, two for directory entries * and one for root of the snapshot. */ - u64 num_bytes = calc_trans_metadata_size(root, 5); + u64 num_bytes = btrfs_calc_trans_metadata_size(root, 5); dst_rsv->space_info = src_rsv->space_info; return block_rsv_migrate_bytes(src_rsv, dst_rsv, num_bytes); } @@ -4078,7 +4072,7 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes) if (nr_extents > reserved_extents) { nr_extents -= reserved_extents; - to_reserve = calc_trans_metadata_size(root, nr_extents); + to_reserve = btrfs_calc_trans_metadata_size(root, nr_extents); } else { nr_extents = 0; to_reserve = 0; @@ -4132,7 +4126,7 @@ void btrfs_delalloc_release_metadata(struct inode *inode, u64 num_bytes) to_free = calc_csum_metadata_size(inode, num_bytes); if (nr_extents > 0) - to_free += calc_trans_metadata_size(root, nr_extents); + to_free += btrfs_calc_trans_metadata_size(root, nr_extents); btrfs_block_rsv_release(root, &root->fs_info->delalloc_block_rsv, to_free); -- cgit v1.2.2 From 7a7eaa40a39bde4eefc91aadeb1ce3dc4e6a1252 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 13 Apr 2011 12:54:33 -0400 Subject: Btrfs: take away the num_items argument from btrfs_join_transaction I keep forgetting that btrfs_join_transaction() just ignores the num_items argument, which leads me to sending pointless patches and looking stupid :). So just kill the num_items argument from btrfs_join_transaction and btrfs_start_ioctl_transaction, since neither of them use it. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 12 ++++++------ 1 file changed, 6 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 9ee6bd55e16c..941b28e78931 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3174,7 +3174,7 @@ again: spin_unlock(&data_sinfo->lock); alloc: alloc_target = btrfs_get_alloc_profile(root, 1); - trans = btrfs_join_transaction(root, 1); + trans = btrfs_join_transaction(root); if (IS_ERR(trans)) return PTR_ERR(trans); @@ -3202,7 +3202,7 @@ alloc: commit_trans: if (!committed && !root->fs_info->open_ioctl_trans) { committed = 1; - trans = btrfs_join_transaction(root, 1); + trans = btrfs_join_transaction(root); if (IS_ERR(trans)) return PTR_ERR(trans); ret = btrfs_commit_transaction(trans, root); @@ -3589,7 +3589,7 @@ again: goto out; ret = -ENOSPC; - trans = btrfs_join_transaction(root, 1); + trans = btrfs_join_transaction(root); if (IS_ERR(trans)) goto out; ret = btrfs_commit_transaction(trans, root); @@ -3816,7 +3816,7 @@ int btrfs_block_rsv_check(struct btrfs_trans_handle *trans, if (trans) return -EAGAIN; - trans = btrfs_join_transaction(root, 1); + trans = btrfs_join_transaction(root); BUG_ON(IS_ERR(trans)); ret = btrfs_commit_transaction(trans, root); return 0; @@ -7649,7 +7649,7 @@ int btrfs_drop_dead_reloc_roots(struct btrfs_root *root) BUG_ON(reloc_root->commit_root != NULL); while (1) { - trans = btrfs_join_transaction(root, 1); + trans = btrfs_join_transaction(root); BUG_ON(IS_ERR(trans)); mutex_lock(&root->fs_info->drop_mutex); @@ -8176,7 +8176,7 @@ int btrfs_set_block_group_ro(struct btrfs_root *root, BUG_ON(cache->ro); - trans = btrfs_join_transaction(root, 1); + trans = btrfs_join_transaction(root); BUG_ON(IS_ERR(trans)); alloc_flags = update_block_group_flags(root, cache->flags); -- cgit v1.2.2 From a4abeea41adfa3c143c289045f4625dfaeba2212 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Mon, 11 Apr 2011 17:25:13 -0400 Subject: Btrfs: kill trans_mutex We use trans_mutex for lots of things, here's a basic list 1) To serialize trans_handles joining the currently running transaction 2) To make sure that no new trans handles are started while we are committing 3) To protect the dead_roots list and the transaction lists Really the serializing trans_handles joining is not too hard, and can really get bogged down in acquiring a reference to the transaction. So replace the trans_mutex with a trans_lock spinlock and use it to do the following 1) Protect fs_info->running_transaction. All trans handles have to do is check this, and then take a reference of the transaction and keep on going. 2) Protect the fs_info->trans_list. This doesn't get used too much, basically it just holds the current transactions, which will usually just be the currently committing transaction and the currently running transaction at most. 3) Protect the dead roots list. This is only ever processed by splicing the list so this is relatively simple. 4) Protect the fs_info->reloc_ctl stuff. This is very lightweight and was using the trans_mutex before, so this is a pretty straightforward change. 5) Protect fs_info->no_trans_join. Because we don't hold the trans_lock over the entirety of the commit we need to have a way to block new people from creating a new transaction while we're doing our work. So we set no_trans_join and in join_transaction we test to see if that is set, and if it is we do a wait_on_commit. 6) Make the transaction use count atomic so we don't need to take locks to modify it when we're dropping references. 7) Add a commit_lock to the transaction to make sure multiple people trying to commit the same transaction don't race and commit at the same time. 8) Make open_ioctl_trans an atomic so we don't have to take any locks for ioctl trans. I have tested this with xfstests, but obviously it is a pretty hairy change so lots of testing is greatly appreciated. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 3 ++- 1 file changed, 2 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 941b28e78931..ca599654ce19 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3200,7 +3200,8 @@ alloc: /* commit the current transaction and try again */ commit_trans: - if (!committed && !root->fs_info->open_ioctl_trans) { + if (!committed && + !atomic_read(&root->fs_info->open_ioctl_trans)) { committed = 1; trans = btrfs_join_transaction(root); if (IS_ERR(trans)) -- cgit v1.2.2 From fcb80c2affd63237cff5b34cba5756be7c976a5a Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 3 May 2011 10:40:22 -0400 Subject: Btrfs: fix how we do space reservation for truncate The ceph guys keep running into problems where we have space reserved in our orphan block rsv when freeing it up. This is because they tend to do snapshots alot, so their truncates tend to use a bunch of space, so when we go to do things like update the inode we have to steal reservation space in order to make the reservation happen. This happens because truncate can use as much space as it freaking feels like, but we still have to hold space for removing the orphan item and updating the inode, which will definitely always happen. So in order to fix this we need to split all of the reservation stuf up. So with this patch we have 1) The orphan block reserve which only holds the space for deleting our orphan item when everything is over. 2) The truncate block reserve which gets allocated and used specifically for the space that the truncate will use on a per truncate basis. 3) The transaction will always have 1 item's worth of data reserved so we can update the inode normally. Hopefully this will make the ceph problem go away. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 46 ++++++++++++++++++++++++++++++++++++---------- 1 file changed, 36 insertions(+), 10 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index ca599654ce19..a2ca561c70f0 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3980,6 +3980,37 @@ static u64 calc_trans_metadata_size(struct btrfs_root *root, int num_items) 3 * num_items; } +int btrfs_truncate_reserve_metadata(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_block_rsv *rsv) +{ + struct btrfs_block_rsv *trans_rsv = &root->fs_info->trans_block_rsv; + u64 num_bytes; + int ret; + + /* + * Truncate should be freeing data, but give us 2 items just in case it + * needs to use some space. We may want to be smarter about this in the + * future. + */ + num_bytes = calc_trans_metadata_size(root, 2); + + /* We already have enough bytes, just return */ + if (rsv->reserved >= num_bytes) + return 0; + + num_bytes -= rsv->reserved; + + /* + * You should have reserved enough space before hand to do this, so this + * should not fail. + */ + ret = block_rsv_migrate_bytes(trans_rsv, rsv, num_bytes); + BUG_ON(ret); + + return 0; +} + int btrfs_trans_reserve_metadata(struct btrfs_trans_handle *trans, struct btrfs_root *root, int num_items) @@ -4020,23 +4051,18 @@ int btrfs_orphan_reserve_metadata(struct btrfs_trans_handle *trans, struct btrfs_block_rsv *dst_rsv = root->orphan_block_rsv; /* - * one for deleting orphan item, one for updating inode and - * two for calling btrfs_truncate_inode_items. - * - * btrfs_truncate_inode_items is a delete operation, it frees - * more space than it uses in most cases. So two units of - * metadata space should be enough for calling it many times. - * If all of the metadata space is used, we can commit - * transaction and use space it freed. + * We need to hold space in order to delete our orphan item once we've + * added it, so this takes the reservation so we can release it later + * when we are truly done with the orphan item. */ - u64 num_bytes = calc_trans_metadata_size(root, 4); + u64 num_bytes = calc_trans_metadata_size(root, 1); return block_rsv_migrate_bytes(src_rsv, dst_rsv, num_bytes); } void btrfs_orphan_release_metadata(struct inode *inode) { struct btrfs_root *root = BTRFS_I(inode)->root; - u64 num_bytes = calc_trans_metadata_size(root, 4); + u64 num_bytes = calc_trans_metadata_size(root, 1); btrfs_block_rsv_release(root, root->orphan_block_rsv, num_bytes); } -- cgit v1.2.2 From d82a6f1d7e8b61ed5996334d0db66651bb43641d Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 11 May 2011 15:26:06 -0400 Subject: Btrfs: kill BTRFS_I(inode)->block_group Originally this was going to be used as a way to give hints to the allocator, but frankly we can get much better hints elsewhere and it's not even used at all for anything usefull. In addition to be completely useless, when we initialize an inode we try and find a freeish block group to set as the inodes block group, and with a completely full 40gb fs this takes _forever_, so I imagine with say 1tb fs this is just unbearable. So just axe the thing altoghether, we don't need it and it saves us 8 bytes in the inode and saves us 500 microseconds per inode lookup in my testcase. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 10 ++-------- 1 file changed, 2 insertions(+), 8 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index a2ca561c70f0..9f0a4e3bd8a9 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5319,6 +5319,7 @@ checks: btrfs_add_free_space(block_group, offset, search_start - offset); BUG_ON(offset > search_start); + btrfs_put_block_group(block_group); break; loop: failed_cluster_refill = false; @@ -5411,14 +5412,7 @@ loop: ret = -ENOSPC; } else if (!ins->objectid) { ret = -ENOSPC; - } - - /* we found what we needed */ - if (ins->objectid) { - if (!(data & BTRFS_BLOCK_GROUP_DATA)) - trans->block_group = block_group->key.objectid; - - btrfs_put_block_group(block_group); + } else if (ins->objectid) { ret = 0; } -- cgit v1.2.2 From 589d8ade83f07c0f11c8191c0ca309f34d7a2c14 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 11 May 2011 17:30:53 -0400 Subject: Btrfs: try not to sleep as much when doing slow caching When the fs is super full and we unmount the fs, we could get stuck in this thing where unmount is waiting for the caching kthread to make progress and the caching kthread keeps scheduling because we're in the middle of a commit. So instead just let the caching kthread keep going and only yeild if need_resched(). This makes my horrible umount case go from taking up to 10 minutes to taking less than 20 seconds. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 19 +++++++++++-------- 1 file changed, 11 insertions(+), 8 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 9f0a4e3bd8a9..96be62450318 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -378,15 +378,18 @@ again: if (ret) break; - caching_ctl->progress = last; - btrfs_release_path(extent_root, path); - up_read(&fs_info->extent_commit_sem); - mutex_unlock(&caching_ctl->mutex); - if (btrfs_transaction_in_commit(fs_info)) - schedule_timeout(1); - else + if (need_resched() || + btrfs_next_leaf(extent_root, path)) { + caching_ctl->progress = last; + btrfs_release_path(extent_root, path); + up_read(&fs_info->extent_commit_sem); + mutex_unlock(&caching_ctl->mutex); cond_resched(); - goto again; + goto again; + } + leaf = path->nodes[0]; + nritems = btrfs_header_nritems(leaf); + continue; } if (key.objectid < block_group->key.objectid) { -- cgit v1.2.2 From 026fd317828500524cdc7e5ff9e8e7923abb2868 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 13 May 2011 10:32:11 -0400 Subject: Btrfs: don't always do readahead Our readahead is sort of sloppy, and really isn't always needed. For example if ls is doing a stating ls (which is the default) it's going to stat in non-disk order, so if say you have a directory with a stupid amount of files, readahead is going to do nothing but waste time in the case of doing the stat. Taking the unconditional readahead out made my test go from 57 minutes to 36 minutes. This means that everywhere we do loop through the tree we want to make sure we do set path->reada properly, so I went through and found all of the places where we loop through the path and set reada to 1. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 3 ++- 1 file changed, 2 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 96be62450318..1ba2cc58eab5 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -347,7 +347,7 @@ static int caching_kthread(void *data) */ path->skip_locking = 1; path->search_commit_root = 1; - path->reada = 2; + path->reada = 1; key.objectid = last; key.offset = 0; @@ -8556,6 +8556,7 @@ int btrfs_read_block_groups(struct btrfs_root *root) path = btrfs_alloc_path(); if (!path) return -ENOMEM; + path->reada = 1; cache_gen = btrfs_super_cache_generation(&root->fs_info->super_copy); if (cache_gen != 0 && -- cgit v1.2.2 From cca1c81f43e26ab60c0d1090fb90992358d69bdf Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 13 May 2011 11:07:12 -0400 Subject: Btrfs: don't try to allocate from a block group that doesn't have enough space If we have a very large filesystem, we can spend a lot of time in find_free_extent just trying to allocate from empty block groups. So instead check to see if the block group even has enough space for the allocation, and if not go on to the next block group. Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 1ba2cc58eab5..c8c318494dee 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5159,6 +5159,14 @@ have_block_group: if (unlikely(block_group->ro)) goto loop; + spin_lock(&block_group->tree_lock); + if (cached && + block_group->free_space < num_bytes + empty_size) { + spin_unlock(&block_group->tree_lock); + goto loop; + } + spin_unlock(&block_group->tree_lock); + /* * Ok we want to try and use the cluster allocator, so lets look * there, unless we are on LOOP_NO_EMPTY_SIZE, since we will -- cgit v1.2.2 From c4f675cd40d955d539180506c09515c90169b15b Mon Sep 17 00:00:00 2001 From: Sergei Trofimovich Date: Fri, 20 May 2011 20:20:30 +0000 Subject: btrfs: don't spin in shrink_delalloc if there is nothing to free Observed as a large delay when --mixed filesystem is filled up. Test example: 1. create tiny --mixed FS: $ dd if=/dev/zero of=2G.img seek=$((2048 * 1024 * 1024 - 1)) count=1 bs=1 $ mkfs.btrfs --mixed 2G.img $ mount -oloop 2G.img /mnt/ut/ 2. Try to fill it up: $ dd if=/dev/urandom of=10M.file bs=10240 count=1024 $ seq 1 256 | while read file_no; do echo $file_no; time cp 10M.file ${file_no}.copy; done Up to '200.copy' it goes fast, but when disk fills-up each -ENOSPC message takes 3 seconds to pop-up _every_ ENOSPC (and in usermode linux it's even more: 30-60 seconds!). (Maybe, time depends on kernel's timer resolution). No IO, no CPU load, just rescheduling. Some debugging revealed busy spinning in shrink_delalloc. Signed-off-by: Sergei Trofimovich Reviewed-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 9ee6bd55e16c..9f5fdd37451d 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3425,6 +3425,10 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans, if (reserved == 0) return 0; + /* nothing to shrink - nothing to reclaim */ + if (root->fs_info->delalloc_bytes == 0) + return 0; + max_reclaim = min(reserved, to_reclaim); while (loops < 1024) { -- cgit v1.2.2 From 1cd307990d6e2b4965620e339a92e0d7ae853e13 Mon Sep 17 00:00:00 2001 From: Tsutomu Itoh Date: Thu, 19 May 2011 05:19:08 +0000 Subject: Btrfs: BUG_ON is deleted from the caller of btrfs_truncate_item & btrfs_extend_item Currently, btrfs_truncate_item and btrfs_extend_item returns only 0. So, the check by BUG_ON in the caller is unnecessary. Signed-off-by: Tsutomu Itoh Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 3 --- 1 file changed, 3 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 9f5fdd37451d..103e141afeb3 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -947,7 +947,6 @@ static int convert_extent_item_v0(struct btrfs_trans_handle *trans, BUG_ON(ret); ret = btrfs_extend_item(trans, root, path, new_size); - BUG_ON(ret); leaf = path->nodes[0]; item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item); @@ -1555,7 +1554,6 @@ int setup_inline_extent_backref(struct btrfs_trans_handle *trans, size = btrfs_extent_inline_ref_size(type); ret = btrfs_extend_item(trans, root, path, size); - BUG_ON(ret); ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item); refs = btrfs_extent_refs(leaf, ei); @@ -1684,7 +1682,6 @@ int update_inline_extent_backref(struct btrfs_trans_handle *trans, end - ptr - size); item_size -= size; ret = btrfs_truncate_item(trans, root, path, item_size, 1); - BUG_ON(ret); } btrfs_mark_buffer_dirty(leaf); return 0; -- cgit v1.2.2 From 7841cb2898f66a73062c64d0ef5733dde7279e46 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Tue, 31 May 2011 18:07:27 +0200 Subject: btrfs: add helper for fs_info->closing wrap checking of filesystem 'closing' flag and fix a few missing memory barriers. Signed-off-by: David Sterba --- fs/btrfs/extent-tree.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index c9173a7827b0..5b9b6b6df242 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -366,8 +366,7 @@ again: nritems = btrfs_header_nritems(leaf); while (1) { - smp_mb(); - if (fs_info->closing > 1) { + if (btrfs_fs_closing(fs_info) > 1) { last = (u64)-1; break; } -- cgit v1.2.2 From f2bb8f5cfb3bce595b2de251ed7638047fc4e530 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 25 May 2011 13:10:16 -0400 Subject: Btrfs: don't commit the transaction if we dont have enough pinned bytes I noticed when running an enospc test that we would get stuck committing the transaction in check_data_space even though we truly didn't have enough space. So check to see if bytes_pinned is bigger than num_bytes, if it's not don't commit the transaction. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 5b9b6b6df242..0d0a3fe77bb7 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3089,6 +3089,13 @@ alloc: } goto again; } + + /* + * If we have less pinned bytes than we want to allocate then + * don't bother committing the transaction, it won't help us. + */ + if (data_sinfo->bytes_pinned < bytes) + committed = 1; spin_unlock(&data_sinfo->lock); /* commit the current transaction and try again */ -- cgit v1.2.2 From 723bda2083d44edbd6be0f0b09f902120dc07442 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 27 May 2011 16:11:38 -0400 Subject: Btrfs: fix the allocator loop logic I was testing with empty_cluster = 0 to try and reproduce a problem and kept hitting early enospc panics. This was because our loop logic was a little confused. So this is what I did 1) Make the loop variable the ultimate decider on wether we should loop again isntead of checking to see if we had an uncached bg, empty size or empty cluster. 2) Increment loop before checking to see what we are on to make the loop definitions make more sense. 3) If we are on the chunk alloc loop don't set empty_size/empty_cluster to 0 unless we didn't actually allocate a chunk. If we did allocate a chunk we should be able to easily setup a new cluster so clearing empty_size/empty_cluster makes us less efficient. This kept me from hitting panics while trying to reproduce the other problem. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 48 +++++++++++++++++++++++++----------------------- 1 file changed, 25 insertions(+), 23 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 0d0a3fe77bb7..b42efc2ded51 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5218,9 +5218,7 @@ loop: * LOOP_NO_EMPTY_SIZE, set empty_size and empty_cluster to 0 and try * again */ - if (!ins->objectid && loop < LOOP_NO_EMPTY_SIZE && - (found_uncached_bg || empty_size || empty_cluster || - allowed_chunk_alloc)) { + if (!ins->objectid && loop < LOOP_NO_EMPTY_SIZE) { index = 0; if (loop == LOOP_FIND_IDEAL && found_uncached_bg) { found_uncached_bg = false; @@ -5260,32 +5258,36 @@ loop: goto search; } - if (loop < LOOP_CACHING_WAIT) { - loop++; - goto search; - } + loop++; if (loop == LOOP_ALLOC_CHUNK) { - empty_size = 0; - empty_cluster = 0; - } + if (allowed_chunk_alloc) { + ret = do_chunk_alloc(trans, root, num_bytes + + 2 * 1024 * 1024, data, + CHUNK_ALLOC_LIMITED); + allowed_chunk_alloc = 0; + if (ret == 1) + done_chunk_alloc = 1; + } else if (!done_chunk_alloc && + space_info->force_alloc == + CHUNK_ALLOC_NO_FORCE) { + space_info->force_alloc = CHUNK_ALLOC_LIMITED; + } - if (allowed_chunk_alloc) { - ret = do_chunk_alloc(trans, root, num_bytes + - 2 * 1024 * 1024, data, - CHUNK_ALLOC_LIMITED); - allowed_chunk_alloc = 0; - done_chunk_alloc = 1; - } else if (!done_chunk_alloc && - space_info->force_alloc == CHUNK_ALLOC_NO_FORCE) { - space_info->force_alloc = CHUNK_ALLOC_LIMITED; + /* + * We didn't allocate a chunk, go ahead and drop the + * empty size and loop again. + */ + if (!done_chunk_alloc) + loop = LOOP_NO_EMPTY_SIZE; } - if (loop < LOOP_NO_EMPTY_SIZE) { - loop++; - goto search; + if (loop == LOOP_NO_EMPTY_SIZE) { + empty_size = 0; + empty_cluster = 0; } - ret = -ENOSPC; + + goto search; } else if (!ins->objectid) { ret = -ENOSPC; } else if (ins->objectid) { -- cgit v1.2.2 From f4c44016218a6fce357715b9bbabbbbe1f69853c Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Mon, 13 Jun 2011 11:30:47 -0400 Subject: Btrfs: drop the delalloc_bytes check in shrink_delalloc Even when delalloc_bytes is zero, we may need to sleep while waiting for delalloc space. Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 4 ---- 1 file changed, 4 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index b42efc2ded51..1f61bf5b4960 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3314,10 +3314,6 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans, if (reserved == 0) return 0; - /* nothing to shrink - nothing to reclaim */ - if (root->fs_info->delalloc_bytes == 0) - return 0; - max_reclaim = min(reserved, to_reclaim); while (loops < 1024) { -- cgit v1.2.2 From e0f5406727f1dfdc47b8ba4a0ff6eae4b0b5ed4c Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Sat, 18 Jun 2011 20:26:38 +0000 Subject: Btrfs: fix type mismatch in find_free_extent() data parameter should be u64 because a full-sized chunk flags field is passed instead of 0/1 for distinguishing data from metadata. All underlying functions expect u64. Signed-off-by: Ilya Dryomov Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 1f61bf5b4960..71cd456fdb60 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4842,7 +4842,7 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans, u64 num_bytes, u64 empty_size, u64 search_start, u64 search_end, u64 hint_byte, struct btrfs_key *ins, - int data) + u64 data) { int ret = 0; struct btrfs_root *root = orig_root->fs_info->extent_root; @@ -4869,7 +4869,7 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans, space_info = __find_space_info(root->fs_info, data); if (!space_info) { - printk(KERN_ERR "No space info for %d\n", data); + printk(KERN_ERR "No space info for %llu\n", data); return -ENOSPC; } -- cgit v1.2.2