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