Patchwork [1/2] Make orphan functions no-op in no-journal mode

login
register
mail settings
Submitter Anatol Pomozov
Date Sept. 17, 2012, 5:47 p.m.
Message ID <1347904037-24229-1-git-send-email-anatol.pomozov@gmail.com>
Download mbox | patch
Permalink /patch/184517/
State Superseded
Headers show

Comments

Anatol Pomozov - Sept. 17, 2012, 5:47 p.m.
This avoids using shared mutext and thus improves scalability of
no-journal mode. The goal of this change is similar to 3d287de3b828
but it avoids mutex usage for all ext4_orphan_del(NULL,..) calls.

Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
---
 fs/ext4/namei.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
Anatol Pomozov - Sept. 17, 2012, 5:55 p.m.
Hi

On Mon, Sep 17, 2012 at 10:47 AM, Anatol Pomozov
<anatol.pomozov@gmail.com> wrote:
> This avoids using shared mutext and thus improves scalability of
> no-journal mode. The goal of this change is similar to 3d287de3b828
> but it avoids mutex usage for all ext4_orphan_del(NULL,..) calls.

This change does the same as 3d287de3b828 but for all
ext4_orphan_del() usages. So 3d287de3b828 can be reverted.
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Theodore Ts'o - Sept. 18, 2012, 3:32 a.m.
On Mon, Sep 17, 2012 at 10:47:16AM -0700, Anatol Pomozov wrote:
> This avoids using shared mutext and thus improves scalability of
> no-journal mode. The goal of this change is similar to 3d287de3b828
> but it avoids mutex usage for all ext4_orphan_del(NULL,..) calls.

It doesn't really improve scalability in no-journal except in the
error case.  This is because....

> diff --git a/fs/ext4/namei.c b/fs/ext4/namei.c
> index 2a42cc0..6863cdf 100644
> --- a/fs/ext4/namei.c
> +++ b/fs/ext4/namei.c
> @@ -2362,6 +2362,8 @@ int ext4_orphan_add(handle_t *handle, struct inode *inode)
>  	struct ext4_iloc iloc;
>  	int err = 0, rc;
>  
> +	if (!EXT4_SB(sb)->s_journal)
> +		return 0;
>  	if (!ext4_handle_valid(handle))
>  		return 0;


The two checks above are equivalent.  If (!EXT4_SB(sb)->s_journal),
then ext4_journal_start() will return return an "invalid" handle.  So
this change is purely cosmetic.  I don't object to making the change
for consistency with the change to be made in ext4_orphan_del() below,
but we should remove the !ext4_handle_valid(handle) call in that case.


> @@ -2436,6 +2438,8 @@ int ext4_orphan_del(handle_t *handle, struct inode *inode)
>  	struct ext4_iloc iloc;
>  	int err = 0;
>  
> +	if (!EXT4_SB(inode->i_sb)->s_journal)
> +		return 0;
>  	/* ext4_handle_valid() assumes a valid handle_t pointer */
>  	if (handle && !ext4_handle_valid(handle))
>  		return 0;

We can remove the (handle && !ext4_handle_valid(handle)) conditional
here too.

						- Ted
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/fs/ext4/namei.c b/fs/ext4/namei.c
index 2a42cc0..6863cdf 100644
--- a/fs/ext4/namei.c
+++ b/fs/ext4/namei.c
@@ -2362,6 +2362,8 @@  int ext4_orphan_add(handle_t *handle, struct inode *inode)
 	struct ext4_iloc iloc;
 	int err = 0, rc;
 
+	if (!EXT4_SB(sb)->s_journal)
+		return 0;
 	if (!ext4_handle_valid(handle))
 		return 0;
 
@@ -2436,6 +2438,8 @@  int ext4_orphan_del(handle_t *handle, struct inode *inode)
 	struct ext4_iloc iloc;
 	int err = 0;
 
+	if (!EXT4_SB(inode->i_sb)->s_journal)
+		return 0;
 	/* ext4_handle_valid() assumes a valid handle_t pointer */
 	if (handle && !ext4_handle_valid(handle))
 		return 0;
@@ -2456,7 +2460,7 @@  int ext4_orphan_del(handle_t *handle, struct inode *inode)
 	 * transaction handle with which to update the orphan list on
 	 * disk, but we still need to remove the inode from the linked
 	 * list in memory. */
-	if (sbi->s_journal && !handle)
+	if (!handle)
 		goto out;
 
 	err = ext4_reserve_inode_write(handle, inode, &iloc);