[PATCH] improve reiserfs O_SYNC and fsync speeds

Chris Mason (mason@suse.com)
Tue, 14 Aug 2001 17:00:54 -0400


Hello everyone,

This is an updated version of the reiserfs transaction tracking
patch. It makes O_SYNC and fsyncs faster by only committing the
last transaction a file/dir was invovled in, instead of forcing a
commit on the current transaction. Typical speedup was ~50%.

This has been around for some time and is well tested, it is against
2.4.8-ac4. It might apply cleanly to pure 2.4.8 or 2.4.9-pre
kernels, I'm testing there next.

Alan, please apply.

-chris

# against 2.4.8-ac4
diff -Nru a/fs/reiserfs/dir.c b/fs/reiserfs/dir.c
--- a/fs/reiserfs/dir.c Tue Aug 14 16:56:18 2001
+++ b/fs/reiserfs/dir.c Tue Aug 14 16:56:18 2001
@@ -39,22 +39,10 @@
};

int reiserfs_dir_fsync(struct file *filp, struct dentry *dentry, int datasync) {
- int ret = 0 ;
- int windex ;
- struct reiserfs_transaction_handle th ;
-
lock_kernel();
-
- journal_begin(&th, dentry->d_inode->i_sb, 1) ;
- windex = push_journal_writer("dir_fsync") ;
- reiserfs_prepare_for_journal(th.t_super, SB_BUFFER_WITH_SB(th.t_super), 1) ;
- journal_mark_dirty(&th, dentry->d_inode->i_sb, SB_BUFFER_WITH_SB (dentry->d_inode->i_sb)) ;
- pop_journal_writer(windex) ;
- journal_end_sync(&th, dentry->d_inode->i_sb, 1) ;
-
- unlock_kernel();
-
- return ret ;
+ reiserfs_commit_for_inode(dentry->d_inode) ;
+ unlock_kernel() ;
+ return 0 ;
}


diff -Nru a/fs/reiserfs/file.c b/fs/reiserfs/file.c
--- a/fs/reiserfs/file.c Tue Aug 14 16:56:18 2001
+++ b/fs/reiserfs/file.c Tue Aug 14 16:56:18 2001
@@ -42,6 +42,7 @@
lock_kernel() ;
down (&inode->i_sem);
journal_begin(&th, inode->i_sb, JOURNAL_PER_BALANCE_CNT * 3) ;
+ reiserfs_update_inode_transaction(inode) ;

#ifdef REISERFS_PREALLOCATE
reiserfs_discard_prealloc (&th, inode);
@@ -75,10 +76,7 @@
int datasync
) {
struct inode * p_s_inode = p_s_dentry->d_inode;
- struct reiserfs_transaction_handle th ;
int n_err = 0;
- int windex ;
- int jbegin_count = 1 ;

lock_kernel() ;

@@ -86,14 +84,7 @@
BUG ();

n_err = fsync_inode_buffers(p_s_inode) ;
- /* commit the current transaction to flush any metadata
- ** changes. sys_fsync takes care of flushing the dirty pages for us
- */
- journal_begin(&th, p_s_inode->i_sb, jbegin_count) ;
- windex = push_journal_writer("sync_file") ;
- reiserfs_update_sd(&th, p_s_inode);
- pop_journal_writer(windex) ;
- journal_end_sync(&th, p_s_inode->i_sb,jbegin_count) ;
+ reiserfs_commit_for_inode(p_s_inode) ;
unlock_kernel() ;
return ( n_err < 0 ) ? -EIO : 0;
}
diff -Nru a/fs/reiserfs/inode.c b/fs/reiserfs/inode.c
--- a/fs/reiserfs/inode.c Tue Aug 14 16:56:18 2001
+++ b/fs/reiserfs/inode.c Tue Aug 14 16:56:18 2001
@@ -35,6 +35,7 @@
down (&inode->i_sem);

journal_begin(&th, inode->i_sb, jbegin_count) ;
+ reiserfs_update_inode_transaction(inode) ;
windex = push_journal_writer("delete_inode") ;

reiserfs_delete_object (&th, inode);
@@ -232,6 +233,7 @@
reiserfs_update_sd(th, inode) ;
journal_end(th, s, len) ;
journal_begin(th, s, len) ;
+ reiserfs_update_inode_transaction(inode) ;
}

// it is called by get_block when create == 0. Returns block number
@@ -567,6 +569,7 @@
TYPE_ANY, 3/*key length*/);
if ((new_offset + inode->i_sb->s_blocksize) >= inode->i_size) {
journal_begin(&th, inode->i_sb, jbegin_count) ;
+ reiserfs_update_inode_transaction(inode) ;
transaction_started = 1 ;
}
research:
@@ -591,6 +594,7 @@
if (!transaction_started) {
pathrelse(&path) ;
journal_begin(&th, inode->i_sb, jbegin_count) ;
+ reiserfs_update_inode_transaction(inode) ;
transaction_started = 1 ;
goto research ;
}
@@ -658,6 +662,7 @@
*/
pathrelse(&path) ;
journal_begin(&th, inode->i_sb, jbegin_count) ;
+ reiserfs_update_inode_transaction(inode) ;
transaction_started = 1 ;
goto research;
}
@@ -1277,6 +1282,10 @@
return ;
}
lock_kernel() ;
+
+ /* this is really only used for atime updates, so they don't have
+ ** to be included in O_SYNC or fsync
+ */
journal_begin(&th, inode->i_sb, 1) ;
reiserfs_update_sd (&th, inode);
journal_end(&th, inode->i_sb, 1) ;
@@ -1652,6 +1661,7 @@
** (it will unmap bh if it packs).
*/
journal_begin(&th, p_s_inode->i_sb, JOURNAL_PER_BALANCE_CNT * 2 ) ;
+ reiserfs_update_inode_transaction(p_s_inode) ;
windex = push_journal_writer("reiserfs_vfs_truncate_file") ;
reiserfs_do_truncate (&th, p_s_inode, page, update_timestamps) ;
pop_journal_writer(windex) ;
@@ -1698,6 +1708,7 @@
start_over:
lock_kernel() ;
journal_begin(&th, inode->i_sb, jbegin_count) ;
+ reiserfs_update_inode_transaction(inode) ;

make_cpu_key(&key, inode, byte_offset, TYPE_ANY, 3) ;

@@ -1930,22 +1941,34 @@
static int reiserfs_commit_write(struct file *f, struct page *page,
unsigned from, unsigned to) {
struct inode *inode = page->mapping->host;
+ loff_t pos = ((loff_t)page->index << PAGE_CACHE_SHIFT) + to;
int ret ;

reiserfs_wait_on_write_block(inode->i_sb) ;
+
+ /* generic_commit_write does this for us, but does not update the
+ ** transaction tracking stuff when the size changes. So, we have
+ ** to do the i_size updates here.
+ */
+ if (pos > inode->i_size) {
+ struct reiserfs_transaction_handle th ;
+ lock_kernel() ;
+ journal_begin(&th, inode->i_sb, 1) ;
+ reiserfs_update_inode_transaction(inode) ;
+ inode->i_size = pos ;
+ reiserfs_update_sd(&th, inode) ;
+ journal_end(&th, inode->i_sb, 1) ;
+ unlock_kernel() ;
+ }
+
ret = generic_commit_write(f, page, from, to) ;

/* we test for O_SYNC here so we can commit the transaction
** for any packed tails the file might have had
*/
if (f->f_flags & O_SYNC) {
- struct reiserfs_transaction_handle th ;
lock_kernel() ;
- journal_begin(&th, inode->i_sb, 1) ;
- reiserfs_prepare_for_journal(inode->i_sb,
- SB_BUFFER_WITH_SB(inode->i_sb), 1) ;
- journal_mark_dirty(&th, inode->i_sb, SB_BUFFER_WITH_SB(inode->i_sb)) ;
- journal_end_sync(&th, inode->i_sb, 1) ;
+ reiserfs_commit_for_inode(inode) ;
unlock_kernel() ;
}
return ret ;
diff -Nru a/fs/reiserfs/journal.c b/fs/reiserfs/journal.c
--- a/fs/reiserfs/journal.c Tue Aug 14 16:56:18 2001
+++ b/fs/reiserfs/journal.c Tue Aug 14 16:56:18 2001
@@ -2339,6 +2339,11 @@
** will wait until the current transaction is done/commited before returning
*/
int journal_end_sync(struct reiserfs_transaction_handle *th, struct super_block *p_s_sb, unsigned long nblocks) {
+
+ if (SB_JOURNAL(p_s_sb)->j_len == 0) {
+ reiserfs_prepare_for_journal(p_s_sb, SB_BUFFER_WITH_SB(p_s_sb), 1) ;
+ journal_mark_dirty(th, p_s_sb, SB_BUFFER_WITH_SB(p_s_sb)) ;
+ }
return do_journal_end(th, p_s_sb, nblocks, COMMIT_NOW | WAIT) ;
}

@@ -2627,6 +2632,41 @@
}
}
return 0 ;
+}
+
+void reiserfs_update_inode_transaction(struct inode *inode) {
+
+ inode->u.reiserfs_i.i_trans_index = SB_JOURNAL_LIST_INDEX(inode->i_sb);
+
+ inode->u.reiserfs_i.i_trans_id = SB_JOURNAL(inode->i_sb)->j_trans_id ;
+}
+
+static int reiserfs_inode_in_this_transaction(struct inode *inode) {
+ if (inode->u.reiserfs_i.i_trans_id == SB_JOURNAL(inode->i_sb)->j_trans_id ||
+ inode->u.reiserfs_i.i_trans_id == 0) {
+ return 1;
+ }
+ return 0 ;
+}
+
+void reiserfs_commit_for_inode(struct inode *inode) {
+ struct reiserfs_journal_list *jl ;
+ struct reiserfs_transaction_handle th ;
+ struct super_block *sb = inode->i_sb ;
+
+ jl = SB_JOURNAL_LIST(sb) + inode->u.reiserfs_i.i_trans_index ;
+
+ /* is it from the current transaction, or from an unknown transaction? */
+ if (reiserfs_inode_in_this_transaction(inode)) {
+ journal_join(&th, sb, 1) ;
+ reiserfs_update_inode_transaction(inode) ;
+ journal_end_sync(&th, sb, 1) ;
+ } else if (jl->j_trans_id == inode->u.reiserfs_i.i_trans_id) {
+ flush_commit_list(sb, jl, 1) ;
+ }
+ /* if the transaction id does not match, this list is long since flushed
+ ** and we don't have to do anything here
+ */
}

void reiserfs_restore_prepared_buffer(struct super_block *p_s_sb,
diff -Nru a/fs/reiserfs/namei.c b/fs/reiserfs/namei.c
--- a/fs/reiserfs/namei.c Tue Aug 14 16:56:18 2001
+++ b/fs/reiserfs/namei.c Tue Aug 14 16:56:18 2001
@@ -539,6 +539,8 @@
journal_end(&th, dir->i_sb, jbegin_count) ;
return retval;
}
+ reiserfs_update_inode_transaction(inode) ;
+ reiserfs_update_inode_transaction(dir) ;

inode->i_op = &reiserfs_file_inode_operations;
inode->i_fop = &reiserfs_file_operations;
@@ -598,6 +600,9 @@
//FIXME: needed for block and char devices only
reiserfs_update_sd (&th, inode);

+ reiserfs_update_inode_transaction(inode) ;
+ reiserfs_update_inode_transaction(dir) ;
+
retval = reiserfs_add_entry (&th, dir, dentry->d_name.name, dentry->d_name.len,
inode, 1/*visible*/);
if (retval) {
@@ -653,6 +658,8 @@
journal_end(&th, dir->i_sb, jbegin_count) ;
return retval;
}
+ reiserfs_update_inode_transaction(inode) ;
+ reiserfs_update_inode_transaction(dir) ;

inode->i_op = &reiserfs_dir_inode_operations;
inode->i_fop = &reiserfs_dir_operations;
@@ -721,6 +728,9 @@
}
inode = dentry->d_inode;

+ reiserfs_update_inode_transaction(inode) ;
+ reiserfs_update_inode_transaction(dir) ;
+
if (de.de_objectid != inode->i_ino) {
// FIXME: compare key of an object and a key found in the
// entry
@@ -794,6 +804,9 @@
}
inode = dentry->d_inode;

+ reiserfs_update_inode_transaction(inode) ;
+ reiserfs_update_inode_transaction(dir) ;
+
if (de.de_objectid != inode->i_ino) {
// FIXME: compare key of an object and a key found in the
// entry
@@ -883,6 +896,9 @@
return retval;
}

+ reiserfs_update_inode_transaction(inode) ;
+ reiserfs_update_inode_transaction(dir) ;
+
inode->i_op = &page_symlink_inode_operations;
inode->i_mapping->a_ops = &reiserfs_address_space_operations;

@@ -938,6 +954,10 @@
/* create new entry */
retval = reiserfs_add_entry (&th, dir, dentry->d_name.name, dentry->d_name.len,
inode, 1/*visible*/);
+
+ reiserfs_update_inode_transaction(inode) ;
+ reiserfs_update_inode_transaction(dir) ;
+
if (retval) {
pop_journal_writer(windex) ;
journal_end(&th, dir->i_sb, jbegin_count) ;
@@ -1084,6 +1104,16 @@
return retval;
}

+ reiserfs_update_inode_transaction(old_dir) ;
+ reiserfs_update_inode_transaction(new_dir) ;
+
+ /* this makes it so an fsync on an open fd for the old name will
+ ** commit the rename operation
+ */
+ reiserfs_update_inode_transaction(old_inode) ;
+
+ if (new_inode)
+ reiserfs_update_inode_transaction(new_inode) ;

while (1) {
// look for old name using corresponding entry key (found by reiserfs_find_entry)
diff -Nru a/fs/reiserfs/stree.c b/fs/reiserfs/stree.c
--- a/fs/reiserfs/stree.c Tue Aug 14 16:56:18 2001
+++ b/fs/reiserfs/stree.c Tue Aug 14 16:56:18 2001
@@ -1822,6 +1822,7 @@

journal_end(th, p_s_inode->i_sb, orig_len_alloc) ;
journal_begin(th, p_s_inode->i_sb, orig_len_alloc) ;
+ reiserfs_update_inode_transaction(p_s_inode) ;
}
} while ( n_file_size > ROUND_UP (n_new_file_size) &&
search_for_position_by_key(p_s_inode->i_sb, &s_item_key, &s_search_path) == POSITION_FOUND ) ;
diff -Nru a/include/linux/reiserfs_fs.h b/include/linux/reiserfs_fs.h
--- a/include/linux/reiserfs_fs.h Tue Aug 14 16:56:18 2001
+++ b/include/linux/reiserfs_fs.h Tue Aug 14 16:56:18 2001
@@ -1656,6 +1656,8 @@
*/
#define JOURNAL_BUFFER(j,n) ((j)->j_ap_blocks[((j)->j_start + (n)) % JOURNAL_BLOCK_COUNT])

+void reiserfs_commit_for_inode(struct inode *) ;
+void reiserfs_update_inode_transaction(struct inode *) ;
void reiserfs_wait_on_write_block(struct super_block *s) ;
void reiserfs_block_writes(struct reiserfs_transaction_handle *th) ;
void reiserfs_allow_writes(struct super_block *s) ;
diff -Nru a/include/linux/reiserfs_fs_i.h b/include/linux/reiserfs_fs_i.h
--- a/include/linux/reiserfs_fs_i.h Tue Aug 14 16:56:18 2001
+++ b/include/linux/reiserfs_fs_i.h Tue Aug 14 16:56:18 2001
@@ -40,6 +40,12 @@
is a comment you should make.... -Hans */
//nopack-attribute
int nopack;
+
+ /* we use these for fsync or O_SYNC to decide which transaction needs
+ ** to be committed in order for this inode to be properly flushed
+ */
+ unsigned long i_trans_id ;
+ unsigned long i_trans_index ;
};


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/