diff options
author | Josef Bacik <jbacik@fusionio.com> | 2012-08-17 19:14:17 +0200 |
---|---|---|
committer | Chris Mason <chris.mason@fusionio.com> | 2012-10-01 21:19:03 +0200 |
commit | 5dc562c541e1026df9d43913c2f6b91156e22d32 (patch) | |
tree | a7768100e81b756f2a3edbfcaf99ad77ca7ed605 /fs/btrfs/tree-log.c | |
parent | Btrfs: fix possible corruption when fsyncing written prealloced extents (diff) | |
download | linux-5dc562c541e1026df9d43913c2f6b91156e22d32.tar.xz linux-5dc562c541e1026df9d43913c2f6b91156e22d32.zip |
Btrfs: turbo charge fsync
At least for the vm workload. Currently on fsync we will
1) Truncate all items in the log tree for the given inode if they exist
and
2) Copy all items for a given inode into the log
The problem with this is that for things like VMs you can have lots of
extents from the fragmented writing behavior, and worst yet you may have
only modified a few extents, not the entire thing. This patch fixes this
problem by tracking which transid modified our extent, and then when we do
the tree logging we find all of the extents we've modified in our current
transaction, sort them and commit them. We also only truncate up to the
xattrs of the inode and copy that stuff in normally, and then just drop any
extents in the range we have that exist in the log already. Here are some
numbers of a 50 meg fio job that does random writes and fsync()s after every
write
Original Patched
SATA drive 82KB/s 140KB/s
Fusion drive 431KB/s 2532KB/s
So around 2-6 times faster depending on your hardware. There are a few
corner cases, for example if you truncate at all we have to do it the old
way since there is no way to be sure what is in the log is ok. This
probably could be done smarter, but if you write-fsync-truncate-write-fsync
you deserve what you get. All this work is in RAM of course so if your
inode gets evicted from cache and you read it in and fsync it we'll do it
the slow way if we are still in the same transaction that we last modified
the inode in.
The biggest cool part of this is that it requires no changes to the recovery
code, so if you fsync with this patch and crash and load an old kernel, it
will run the recovery and be a-ok. I have tested this pretty thoroughly
with an fsync tester and everything comes back fine, as well as xfstests.
Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Diffstat (limited to '')
-rw-r--r-- | fs/btrfs/tree-log.c | 220 |
1 files changed, 214 insertions, 6 deletions
diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c index c86670f4f285..f2ff02c55130 100644 --- a/fs/btrfs/tree-log.c +++ b/fs/btrfs/tree-log.c @@ -18,6 +18,7 @@ #include <linux/sched.h> #include <linux/slab.h> +#include <linux/list_sort.h> #include "ctree.h" #include "transaction.h" #include "disk-io.h" @@ -550,7 +551,7 @@ static noinline int replay_one_extent(struct btrfs_trans_handle *trans, saved_nbytes = inode_get_bytes(inode); /* drop any overlapping extents */ - ret = btrfs_drop_extents(trans, inode, start, extent_end, + ret = btrfs_drop_extents(trans, root, inode, start, extent_end, &alloc_hint, 1); BUG_ON(ret); @@ -2803,6 +2804,194 @@ static noinline int copy_items(struct btrfs_trans_handle *trans, return ret; } +static int extent_cmp(void *priv, struct list_head *a, struct list_head *b) +{ + struct extent_map *em1, *em2; + + em1 = list_entry(a, struct extent_map, list); + em2 = list_entry(b, struct extent_map, list); + + if (em1->start < em2->start) + return -1; + else if (em1->start > em2->start) + return 1; + return 0; +} + +struct log_args { + struct extent_buffer *src; + u64 next_offset; + int start_slot; + int nr; +}; + +static int log_one_extent(struct btrfs_trans_handle *trans, + struct inode *inode, struct btrfs_root *root, + struct extent_map *em, struct btrfs_path *path, + struct btrfs_path *dst_path, struct log_args *args) +{ + struct btrfs_root *log = root->log_root; + struct btrfs_file_extent_item *fi; + struct btrfs_key key; + u64 start = em->start; + u64 len = em->len; + u64 num_bytes; + int nritems; + int ret; + + if (BTRFS_I(inode)->logged_trans == trans->transid) { + u64 tmp; + ret = __btrfs_drop_extents(trans, log, inode, dst_path, start, + start + len, &tmp, 0); + if (ret) + return ret; + } + + while (len) { + if (args->nr) + goto next_slot; + key.objectid = btrfs_ino(inode); + key.type = BTRFS_EXTENT_DATA_KEY; + key.offset = start; + + ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); + if (ret < 0) + return ret; + if (ret) { + /* + * This shouldn't happen, but it might so warn and + * return an error. + */ + WARN_ON(1); + return -ENOENT; + } + args->src = path->nodes[0]; +next_slot: + fi = btrfs_item_ptr(args->src, path->slots[0], + struct btrfs_file_extent_item); + if (args->nr && + args->start_slot + args->nr == path->slots[0]) { + args->nr++; + } else if (args->nr) { + ret = copy_items(trans, log, dst_path, args->src, + args->start_slot, args->nr, + LOG_INODE_ALL); + if (ret) + return ret; + args->nr = 1; + args->start_slot = path->slots[0]; + } else if (!args->nr) { + args->nr = 1; + args->start_slot = path->slots[0]; + } + nritems = btrfs_header_nritems(path->nodes[0]); + path->slots[0]++; + num_bytes = btrfs_file_extent_num_bytes(args->src, fi); + if (len < num_bytes) { + /* I _think_ this is ok, envision we write to a + * preallocated space that is adjacent to a previously + * written preallocated space that gets merged when we + * mark this preallocated space written. If we do not + * have the adjacent extent in cache then when we copy + * this extent it could end up being larger than our EM + * thinks it is, which is a-ok, so just set len to 0. + */ + len = 0; + } else { + len -= num_bytes; + } + start += btrfs_file_extent_num_bytes(args->src, fi); + args->next_offset = start; + + if (path->slots[0] < nritems) { + if (len) + goto next_slot; + break; + } + + if (args->nr) { + ret = copy_items(trans, log, dst_path, args->src, + args->start_slot, args->nr, + LOG_INODE_ALL); + if (ret) + return ret; + args->nr = 0; + btrfs_release_path(path); + } + } + + return 0; +} + +static int btrfs_log_changed_extents(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct inode *inode, + struct btrfs_path *path, + struct btrfs_path *dst_path) +{ + struct log_args args; + struct btrfs_root *log = root->log_root; + struct extent_map *em, *n; + struct list_head extents; + struct extent_map_tree *tree = &BTRFS_I(inode)->extent_tree; + u64 test_gen; + int ret = 0; + + INIT_LIST_HEAD(&extents); + + memset(&args, 0, sizeof(args)); + + write_lock(&tree->lock); + test_gen = root->fs_info->last_trans_committed; + + list_for_each_entry_safe(em, n, &tree->modified_extents, list) { + list_del_init(&em->list); + if (em->generation <= test_gen) + continue; + list_add_tail(&em->list, &extents); + } + + list_sort(NULL, &extents, extent_cmp); + + while (!list_empty(&extents)) { + em = list_entry(extents.next, struct extent_map, list); + + list_del_init(&em->list); + + /* + * If we had an error we just need to delete everybody from our + * private list. + */ + if (ret) + continue; + + /* + * If the previous EM and the last extent we left off on aren't + * sequential then we need to copy the items we have and redo + * our search + */ + if (args.nr && em->start != args.next_offset) { + ret = copy_items(trans, log, dst_path, args.src, + args.start_slot, args.nr, + LOG_INODE_ALL); + if (ret) + continue; + btrfs_release_path(path); + args.nr = 0; + } + + ret = log_one_extent(trans, inode, root, em, path, dst_path, &args); + } + + if (!ret && args.nr) + ret = copy_items(trans, log, dst_path, args.src, + args.start_slot, args.nr, LOG_INODE_ALL); + btrfs_release_path(path); + WARN_ON(!list_empty(&extents)); + write_unlock(&tree->lock); + return ret; +} + /* log a single inode in the tree log. * At least one parent directory for this inode must exist in the tree * or be logged already. @@ -2832,6 +3021,7 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans, int nritems; int ins_start_slot = 0; int ins_nr; + bool fast_search = false; u64 ino = btrfs_ino(inode); log = root->log_root; @@ -2851,10 +3041,8 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans, max_key.objectid = ino; - /* today the code can only do partial logging of directories */ - if (!S_ISDIR(inode->i_mode)) - inode_only = LOG_INODE_ALL; + /* today the code can only do partial logging of directories */ if (inode_only == LOG_INODE_EXISTS || S_ISDIR(inode->i_mode)) max_key.type = BTRFS_XATTR_ITEM_KEY; else @@ -2881,7 +3069,16 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans, max_key_type = BTRFS_XATTR_ITEM_KEY; ret = drop_objectid_items(trans, log, path, ino, max_key_type); } else { - ret = btrfs_truncate_inode_items(trans, log, inode, 0, 0); + if (test_and_clear_bit(BTRFS_INODE_NEEDS_FULL_SYNC, + &BTRFS_I(inode)->runtime_flags)) { + ret = btrfs_truncate_inode_items(trans, log, + inode, 0, 0); + } else { + fast_search = true; + max_key.type = BTRFS_XATTR_ITEM_KEY; + ret = drop_objectid_items(trans, log, path, ino, + BTRFS_XATTR_ITEM_KEY); + } } if (ret) { err = ret; @@ -2960,7 +3157,18 @@ next_slot: } ins_nr = 0; } - WARN_ON(ins_nr); + + if (fast_search) { + btrfs_release_path(path); + btrfs_release_path(dst_path); + ret = btrfs_log_changed_extents(trans, root, inode, path, + dst_path); + if (ret) { + err = ret; + goto out_unlock; + } + } + if (inode_only == LOG_INODE_ALL && S_ISDIR(inode->i_mode)) { btrfs_release_path(path); btrfs_release_path(dst_path); |