Btrfs: avoid tree log commit when there are no changes

rpm has a habit of running fdatasync when the file hasn't
changed.  We already detect if a file hasn't been changed
in the current transaction but it might have been sent to
the tree-log in this transaction and not changed since
the last call to fsync.

In this case, we want to avoid a tree log sync, which includes
a number of synchronous writes and barriers.  This commit
extends the existing tracking of the last transaction to change
a file to also track the last sub-transaction.

The end result is that rpm -ivh and -Uvh are roughly twice as fast,
and on par with ext3.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
This commit is contained in:
Chris Mason 2009-10-13 13:21:08 -04:00
parent 4722607db6
commit 257c62e1bc
8 changed files with 71 additions and 16 deletions

View File

@ -86,6 +86,12 @@ struct btrfs_inode {
* transid of the trans_handle that last modified this inode
*/
u64 last_trans;
/*
* log transid when this inode was last modified
*/
u64 last_sub_trans;
/*
* transid that last logged this inode
*/

View File

@ -1009,6 +1009,7 @@ struct btrfs_root {
atomic_t log_writers;
atomic_t log_commit[2];
unsigned long log_transid;
unsigned long last_log_commit;
unsigned long log_batch;
pid_t log_start_pid;
bool log_multiple_pids;

View File

@ -919,6 +919,7 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
atomic_set(&root->log_writers, 0);
root->log_batch = 0;
root->log_transid = 0;
root->last_log_commit = 0;
extent_io_tree_init(&root->dirty_log_pages,
fs_info->btree_inode->i_mapping, GFP_NOFS);
@ -1089,6 +1090,7 @@ int btrfs_add_log_tree(struct btrfs_trans_handle *trans,
WARN_ON(root->log_root);
root->log_root = log_root;
root->log_transid = 0;
root->last_log_commit = 0;
return 0;
}

View File

@ -1087,8 +1087,10 @@ out_nolock:
btrfs_end_transaction(trans, root);
else
btrfs_commit_transaction(trans, root);
} else {
} else if (ret != BTRFS_NO_LOG_SYNC) {
btrfs_commit_transaction(trans, root);
} else {
btrfs_end_transaction(trans, root);
}
}
if (file->f_flags & O_DIRECT) {
@ -1138,6 +1140,13 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync)
int ret = 0;
struct btrfs_trans_handle *trans;
/* we wait first, since the writeback may change the inode */
root->log_batch++;
/* the VFS called filemap_fdatawrite for us */
btrfs_wait_ordered_range(inode, 0, (u64)-1);
root->log_batch++;
/*
* check the transaction that last modified this inode
* and see if its already been committed
@ -1145,6 +1154,11 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync)
if (!BTRFS_I(inode)->last_trans)
goto out;
/*
* if the last transaction that changed this file was before
* the current transaction, we can bail out now without any
* syncing
*/
mutex_lock(&root->fs_info->trans_mutex);
if (BTRFS_I(inode)->last_trans <=
root->fs_info->last_trans_committed) {
@ -1154,13 +1168,6 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync)
}
mutex_unlock(&root->fs_info->trans_mutex);
root->log_batch++;
filemap_fdatawrite(inode->i_mapping);
btrfs_wait_ordered_range(inode, 0, (u64)-1);
root->log_batch++;
if (datasync && !(inode->i_state & I_DIRTY_PAGES))
goto out;
/*
* ok we haven't committed the transaction yet, lets do a commit
*/
@ -1189,6 +1196,7 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync)
*/
mutex_unlock(&dentry->d_inode->i_mutex);
if (ret != BTRFS_NO_LOG_SYNC) {
if (ret > 0) {
ret = btrfs_commit_transaction(trans, root);
} else {
@ -1198,6 +1206,9 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync)
else
ret = btrfs_commit_transaction(trans, root);
}
} else {
ret = btrfs_end_transaction(trans, root);
}
mutex_lock(&dentry->d_inode->i_mutex);
out:
return ret > 0 ? EIO : ret;

View File

@ -3480,6 +3480,7 @@ static noinline void init_btrfs_i(struct inode *inode)
bi->generation = 0;
bi->sequence = 0;
bi->last_trans = 0;
bi->last_sub_trans = 0;
bi->logged_trans = 0;
bi->delalloc_bytes = 0;
bi->reserved_bytes = 0;
@ -4980,7 +4981,9 @@ again:
set_page_dirty(page);
SetPageUptodate(page);
BTRFS_I(inode)->last_trans = root->fs_info->generation + 1;
BTRFS_I(inode)->last_trans = root->fs_info->generation;
BTRFS_I(inode)->last_sub_trans = BTRFS_I(inode)->root->log_transid;
unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
out_unlock:
@ -5100,6 +5103,7 @@ struct inode *btrfs_alloc_inode(struct super_block *sb)
if (!ei)
return NULL;
ei->last_trans = 0;
ei->last_sub_trans = 0;
ei->logged_trans = 0;
ei->outstanding_extents = 0;
ei->reserved_extents = 0;

View File

@ -79,6 +79,7 @@ static inline void btrfs_set_inode_last_trans(struct btrfs_trans_handle *trans,
struct inode *inode)
{
BTRFS_I(inode)->last_trans = trans->transaction->transid;
BTRFS_I(inode)->last_sub_trans = BTRFS_I(inode)->root->log_transid;
}
int btrfs_end_transaction(struct btrfs_trans_handle *trans,

View File

@ -1980,6 +1980,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
int ret;
struct btrfs_root *log = root->log_root;
struct btrfs_root *log_root_tree = root->fs_info->log_root_tree;
u64 log_transid = 0;
mutex_lock(&root->log_mutex);
index1 = root->log_transid % 2;
@ -2018,6 +2019,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
btrfs_set_root_node(&log->root_item, log->node);
root->log_batch = 0;
log_transid = root->log_transid;
root->log_transid++;
log->log_transid = root->log_transid;
root->log_start_pid = 0;
@ -2095,6 +2097,11 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
write_ctree_super(trans, root->fs_info->tree_root, 1);
ret = 0;
mutex_lock(&root->log_mutex);
if (root->last_log_commit < log_transid)
root->last_log_commit = log_transid;
mutex_unlock(&root->log_mutex);
out_wake_log_root:
atomic_set(&log_root_tree->log_commit[index2], 0);
smp_mb();
@ -2862,6 +2869,21 @@ out:
return ret;
}
static int inode_in_log(struct btrfs_trans_handle *trans,
struct inode *inode)
{
struct btrfs_root *root = BTRFS_I(inode)->root;
int ret = 0;
mutex_lock(&root->log_mutex);
if (BTRFS_I(inode)->logged_trans == trans->transid &&
BTRFS_I(inode)->last_sub_trans <= root->last_log_commit)
ret = 1;
mutex_unlock(&root->log_mutex);
return ret;
}
/*
* helper function around btrfs_log_inode to make sure newly created
* parent directories also end up in the log. A minimal inode and backref
@ -2901,6 +2923,11 @@ int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
if (ret)
goto end_no_trans;
if (inode_in_log(trans, inode)) {
ret = BTRFS_NO_LOG_SYNC;
goto end_no_trans;
}
start_log_trans(trans, root);
ret = btrfs_log_inode(trans, root, inode, inode_only);

View File

@ -19,6 +19,9 @@
#ifndef __TREE_LOG_
#define __TREE_LOG_
/* return value for btrfs_log_dentry_safe that means we don't need to log it at all */
#define BTRFS_NO_LOG_SYNC 256
int btrfs_sync_log(struct btrfs_trans_handle *trans,
struct btrfs_root *root);
int btrfs_free_log(struct btrfs_trans_handle *trans, struct btrfs_root *root);