Ocfs2 devel - Jun 2013 - [PATCH v2] ocfs2: Rework transaction rollback in ocfs2_relink_block_group()

From: Jie Liu <jeff.liu at oracle.com>

In ocfs2_relink_block_group(), we roll back all those changes if
notify intent to modify buffers for metadata update failed even
if the relevant buffer has not yet been modified/got dirty at that
point, that are not quite right because of:

- None buffer has been modified/dirty if failed to call
  ocfs2_journal_access_gd() against the previous block group buffer
- Only the previous block group buffer has got dirty if failed to
  call ocfs2_journal_access_gd() against the block group buffer
- There is no need to roll back the change for file entry buffer at all

Those problems will not cause anything wrong but unnecessary.
This patch fix them and kill the useless bg_ptr variable as well.

Signed-off-by: Jie Liu <jeff.liu at oracle.com>
Cc: Younger Liu <younger.liu at huawei.com>
---
 fs/ocfs2/suballoc.c |   25 ++++++++++++-------------
 1 file changed, 12 insertions(+), 13 deletions(-)

diff --git a/fs/ocfs2/suballoc.c b/fs/ocfs2/suballoc.c
index b7e74b5..101d16d 100644
--- a/fs/ocfs2/suballoc.c
+++ b/fs/ocfs2/suballoc.c
@@ -1422,7 +1422,7 @@ static int ocfs2_relink_block_group(handle_t *handle,
 	int status;
 	/* there is a really tiny chance the journal calls could fail,
 	 * but we wouldn't want inconsistent blocks in *any* case. */
-	u64 fe_ptr, bg_ptr, prev_bg_ptr;
+	u64 bg_ptr, prev_bg_ptr;
 	struct ocfs2_dinode *fe = (struct ocfs2_dinode *) fe_bh->b_data;
 	struct ocfs2_group_desc *bg = (struct ocfs2_group_desc *) bg_bh->b_data;
 	struct ocfs2_group_desc *prev_bg = (struct ocfs2_group_desc *)
prev_bg_bh->b_data;
@@ -1437,7 +1437,6 @@ static int ocfs2_relink_block_group(handle_t *handle,
 		(unsigned long long)le64_to_cpu(bg->bg_blkno),
 		(unsigned long long)le64_to_cpu(prev_bg->bg_blkno));
 
-	fe_ptr = le64_to_cpu(fe->id2.i_chain.cl_recs[chain].c_blkno);
 	bg_ptr = le64_to_cpu(bg->bg_next_group);
 	prev_bg_ptr = le64_to_cpu(prev_bg->bg_next_group);
 
@@ -1446,7 +1445,7 @@ static int ocfs2_relink_block_group(handle_t *handle,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
-		goto out_rollback;
+		goto out;
 	}
 
 	prev_bg->bg_next_group = bg->bg_next_group;
@@ -1456,7 +1455,7 @@ static int ocfs2_relink_block_group(handle_t *handle,
 					 bg_bh, OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
-		goto out_rollback;
+		goto out_rollback_prev_bg;
 	}
 
 	bg->bg_next_group = fe->id2.i_chain.cl_recs[chain].c_blkno;
@@ -1466,21 +1465,21 @@ static int ocfs2_relink_block_group(handle_t *handle,
 					 fe_bh, OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
-		goto out_rollback;
+		goto out_rollback_bg;
 	}
 
 	fe->id2.i_chain.cl_recs[chain].c_blkno = bg->bg_blkno;
 	ocfs2_journal_dirty(handle, fe_bh);
 
-out_rollback:
-	if (status < 0) {
-		fe->id2.i_chain.cl_recs[chain].c_blkno = cpu_to_le64(fe_ptr);
-		bg->bg_next_group = cpu_to_le64(bg_ptr);
-		prev_bg->bg_next_group = cpu_to_le64(prev_bg_ptr);
-	}
+out:
+	return status;
 
-	if (status)
-		mlog_errno(status);
+out_rollback_bg:
+	bg->bg_next_group = cpu_to_le64(bg_ptr);
+out_rollback_prev_bg:
+	prev_bg->bg_next_group = cpu_to_le64(prev_bg_ptr);
+
+	mlog_errno(status);
 	return status;
 }
 
-- 
1.7.9.5

On 2013/6/20 13:29, Jeff Liu wrote:
> From: Jie Liu <jeff.liu at oracle.com>
> 
> In ocfs2_relink_block_group(), we roll back all those changes if
> notify intent to modify buffers for metadata update failed even
> if the relevant buffer has not yet been modified/got dirty at that
> point, that are not quite right because of:
> 
> - None buffer has been modified/dirty if failed to call
>   ocfs2_journal_access_gd() against the previous block group buffer
> - Only the previous block group buffer has got dirty if failed to
>   call ocfs2_journal_access_gd() against the block group buffer
> - There is no need to roll back the change for file entry buffer at all
> 
> Those problems will not cause anything wrong but unnecessary.
> This patch fix them and kill the useless bg_ptr variable as well.
> 
> Signed-off-by: Jie Liu <jeff.liu at oracle.com>
> Cc: Younger Liu <younger.liu at huawei.com>
Fine.
Reviewed-by: Younger Liu <younger.liu at huawei.com>
> ---
>  fs/ocfs2/suballoc.c |   25 ++++++++++++-------------
>  1 file changed, 12 insertions(+), 13 deletions(-)
> 
> diff --git a/fs/ocfs2/suballoc.c b/fs/ocfs2/suballoc.c
> index b7e74b5..101d16d 100644
> --- a/fs/ocfs2/suballoc.c
> +++ b/fs/ocfs2/suballoc.c
> @@ -1422,7 +1422,7 @@ static int ocfs2_relink_block_group(handle_t *handle,
>  	int status;
>  	/* there is a really tiny chance the journal calls could fail,
>  	 * but we wouldn't want inconsistent blocks in *any* case. */
> -	u64 fe_ptr, bg_ptr, prev_bg_ptr;
> +	u64 bg_ptr, prev_bg_ptr;
>  	struct ocfs2_dinode *fe = (struct ocfs2_dinode *) fe_bh->b_data;
>  	struct ocfs2_group_desc *bg = (struct ocfs2_group_desc *)
bg_bh->b_data;
>  	struct ocfs2_group_desc *prev_bg = (struct ocfs2_group_desc *)
prev_bg_bh->b_data;
> @@ -1437,7 +1437,6 @@ static int ocfs2_relink_block_group(handle_t *handle,
>  		(unsigned long long)le64_to_cpu(bg->bg_blkno),
>  		(unsigned long long)le64_to_cpu(prev_bg->bg_blkno));
>  
> -	fe_ptr = le64_to_cpu(fe->id2.i_chain.cl_recs[chain].c_blkno);
>  	bg_ptr = le64_to_cpu(bg->bg_next_group);
>  	prev_bg_ptr = le64_to_cpu(prev_bg->bg_next_group);
>  
> @@ -1446,7 +1445,7 @@ static int ocfs2_relink_block_group(handle_t *handle,
>  					 OCFS2_JOURNAL_ACCESS_WRITE);
>  	if (status < 0) {
>  		mlog_errno(status);
> -		goto out_rollback;
> +		goto out;
>  	}
>  
>  	prev_bg->bg_next_group = bg->bg_next_group;
> @@ -1456,7 +1455,7 @@ static int ocfs2_relink_block_group(handle_t *handle,
>  					 bg_bh, OCFS2_JOURNAL_ACCESS_WRITE);
>  	if (status < 0) {
>  		mlog_errno(status);
> -		goto out_rollback;
> +		goto out_rollback_prev_bg;
>  	}
>  
>  	bg->bg_next_group = fe->id2.i_chain.cl_recs[chain].c_blkno;
> @@ -1466,21 +1465,21 @@ static int ocfs2_relink_block_group(handle_t
*handle,
>  					 fe_bh, OCFS2_JOURNAL_ACCESS_WRITE);
>  	if (status < 0) {
>  		mlog_errno(status);
> -		goto out_rollback;
> +		goto out_rollback_bg;
>  	}
>  
>  	fe->id2.i_chain.cl_recs[chain].c_blkno = bg->bg_blkno;
>  	ocfs2_journal_dirty(handle, fe_bh);
>  
> -out_rollback:
> -	if (status < 0) {
> -		fe->id2.i_chain.cl_recs[chain].c_blkno = cpu_to_le64(fe_ptr);
> -		bg->bg_next_group = cpu_to_le64(bg_ptr);
> -		prev_bg->bg_next_group = cpu_to_le64(prev_bg_ptr);
> -	}
> +out:
> +	return status;
>  
> -	if (status)
> -		mlog_errno(status);
> +out_rollback_bg:
> +	bg->bg_next_group = cpu_to_le64(bg_ptr);
> +out_rollback_prev_bg:
> +	prev_bg->bg_next_group = cpu_to_le64(prev_bg_ptr);
> +
> +	mlog_errno(status);
>  	return status;
>  }
>  
>

On Thu, 20 Jun 2013 13:29:33 +0800 Jeff Liu <jeff.liu at oracle.com>
wrote:
> From: Jie Liu <jeff.liu at oracle.com>
> 
> In ocfs2_relink_block_group(), we roll back all those changes if
> notify intent to modify buffers for metadata update failed even
> if the relevant buffer has not yet been modified/got dirty at that
> point, that are not quite right because of:
> 
> - None buffer has been modified/dirty if failed to call
>   ocfs2_journal_access_gd() against the previous block group buffer
> - Only the previous block group buffer has got dirty if failed to
>   call ocfs2_journal_access_gd() against the block group buffer
> - There is no need to roll back the change for file entry buffer at all
> 
> Those problems will not cause anything wrong but unnecessary.
> This patch fix them and kill the useless bg_ptr variable as well.
> 
> ...
>
> --- a/fs/ocfs2/suballoc.c
> +++ b/fs/ocfs2/suballoc.c
> @@ -1422,7 +1422,7 @@ static int ocfs2_relink_block_group(handle_t *handle,
>  	int status;
>  	/* there is a really tiny chance the journal calls could fail,
>  	 * but we wouldn't want inconsistent blocks in *any* case. */
> -	u64 fe_ptr, bg_ptr, prev_bg_ptr;
> +	u64 bg_ptr, prev_bg_ptr;
>  	struct ocfs2_dinode *fe = (struct ocfs2_dinode *) fe_bh->b_data;
>  	struct ocfs2_group_desc *bg = (struct ocfs2_group_desc *)
bg_bh->b_data;
>  	struct ocfs2_group_desc *prev_bg = (struct ocfs2_group_desc *)
prev_bg_bh->b_data;
> @@ -1437,7 +1437,6 @@ static int ocfs2_relink_block_group(handle_t *handle,
>  		(unsigned long long)le64_to_cpu(bg->bg_blkno),
>  		(unsigned long long)le64_to_cpu(prev_bg->bg_blkno));
>  
> -	fe_ptr = le64_to_cpu(fe->id2.i_chain.cl_recs[chain].c_blkno);
>  	bg_ptr = le64_to_cpu(bg->bg_next_group);
>  	prev_bg_ptr = le64_to_cpu(prev_bg->bg_next_group);
>  
> @@ -1446,7 +1445,7 @@ static int ocfs2_relink_block_group(handle_t *handle,
>  					 OCFS2_JOURNAL_ACCESS_WRITE);
>  	if (status < 0) {
>  		mlog_errno(status);
> -		goto out_rollback;
> +		goto out;
>  	}
>  
>  	prev_bg->bg_next_group = bg->bg_next_group;
> @@ -1456,7 +1455,7 @@ static int ocfs2_relink_block_group(handle_t *handle,
>  					 bg_bh, OCFS2_JOURNAL_ACCESS_WRITE);
>  	if (status < 0) {
>  		mlog_errno(status);
> -		goto out_rollback;
> +		goto out_rollback_prev_bg;
>  	}
>  
>  	bg->bg_next_group = fe->id2.i_chain.cl_recs[chain].c_blkno;
> @@ -1466,21 +1465,21 @@ static int ocfs2_relink_block_group(handle_t
*handle,
>  					 fe_bh, OCFS2_JOURNAL_ACCESS_WRITE);
>  	if (status < 0) {
>  		mlog_errno(status);
> -		goto out_rollback;
> +		goto out_rollback_bg;
>  	}
>  
>  	fe->id2.i_chain.cl_recs[chain].c_blkno = bg->bg_blkno;
>  	ocfs2_journal_dirty(handle, fe_bh);
>  
> -out_rollback:
> -	if (status < 0) {
> -		fe->id2.i_chain.cl_recs[chain].c_blkno = cpu_to_le64(fe_ptr);
> -		bg->bg_next_group = cpu_to_le64(bg_ptr);
> -		prev_bg->bg_next_group = cpu_to_le64(prev_bg_ptr);
> -	}
> +out:
> +	return status;
>  
> -	if (status)
> -		mlog_errno(status);
> +out_rollback_bg:
> +	bg->bg_next_group = cpu_to_le64(bg_ptr);
> +out_rollback_prev_bg:
> +	prev_bg->bg_next_group = cpu_to_le64(prev_bg_ptr);
> +
> +	mlog_errno(status);
We already called mlog_errno() for this status in all cases.

How does this look?

---
a/fs/ocfs2/suballoc.c~ocfs2-rework-transaction-rollback-in-ocfs2_relink_block_group-fix
+++ a/fs/ocfs2/suballoc.c
@@ -1443,44 +1443,38 @@ static int ocfs2_relink_block_group(hand
 	status = ocfs2_journal_access_gd(handle, INODE_CACHE(alloc_inode),
 					 prev_bg_bh,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
-	if (status < 0) {
-		mlog_errno(status);
+	if (status < 0)
 		goto out;
-	}
 
 	prev_bg->bg_next_group = bg->bg_next_group;
 	ocfs2_journal_dirty(handle, prev_bg_bh);
 
 	status = ocfs2_journal_access_gd(handle, INODE_CACHE(alloc_inode),
 					 bg_bh, OCFS2_JOURNAL_ACCESS_WRITE);
-	if (status < 0) {
-		mlog_errno(status);
+	if (status < 0)
 		goto out_rollback_prev_bg;
-	}
 
 	bg->bg_next_group = fe->id2.i_chain.cl_recs[chain].c_blkno;
 	ocfs2_journal_dirty(handle, bg_bh);
 
 	status = ocfs2_journal_access_di(handle, INODE_CACHE(alloc_inode),
 					 fe_bh, OCFS2_JOURNAL_ACCESS_WRITE);
-	if (status < 0) {
-		mlog_errno(status);
+	if (status < 0)
 		goto out_rollback_bg;
-	}
 
 	fe->id2.i_chain.cl_recs[chain].c_blkno = bg->bg_blkno;
 	ocfs2_journal_dirty(handle, fe_bh);
 
 out:
+	if (status < 0)
+		mlog_errno(status);
 	return status;
 
 out_rollback_bg:
 	bg->bg_next_group = cpu_to_le64(bg_ptr);
 out_rollback_prev_bg:
 	prev_bg->bg_next_group = cpu_to_le64(prev_bg_ptr);
-
-	mlog_errno(status);
-	return status;
+	goto out;
 }


btw, the ocfs2 source and executable cold be made about half the size if
mlog_errno() were to immediately return if status>=0.

it looks good , but it is better that if moving the following two sentence to
before 'ocfs2_journal_dirty(handle, fe_bh);' sentence.
1.ocfs2_journal_dirty(handle, prev_bg_bh);
2.ocfs2_journal_dirty(handle, bg_bh);

because if it fail, it maybe commit by jbd2 journal.

Thanks,
Jensen

On 2013/6/20 13:29, Jeff Liu wrote:
> From: Jie Liu <jeff.liu at oracle.com>
> 
> In ocfs2_relink_block_group(), we roll back all those changes if
> notify intent to modify buffers for metadata update failed even
> if the relevant buffer has not yet been modified/got dirty at that
> point, that are not quite right because of:
> 
> - None buffer has been modified/dirty if failed to call
>   ocfs2_journal_access_gd() against the previous block group buffer
> - Only the previous block group buffer has got dirty if failed to
>   call ocfs2_journal_access_gd() against the block group buffer
> - There is no need to roll back the change for file entry buffer at all
> 
> Those problems will not cause anything wrong but unnecessary.
> This patch fix them and kill the useless bg_ptr variable as well.
> 
> Signed-off-by: Jie Liu <jeff.liu at oracle.com>
> Cc: Younger Liu <younger.liu at huawei.com>
> ---
>  fs/ocfs2/suballoc.c |   25 ++++++++++++-------------
>  1 file changed, 12 insertions(+), 13 deletions(-)
> 
> diff --git a/fs/ocfs2/suballoc.c b/fs/ocfs2/suballoc.c
> index b7e74b5..101d16d 100644
> --- a/fs/ocfs2/suballoc.c
> +++ b/fs/ocfs2/suballoc.c
> @@ -1422,7 +1422,7 @@ static int ocfs2_relink_block_group(handle_t *handle,
>  	int status;
>  	/* there is a really tiny chance the journal calls could fail,
>  	 * but we wouldn't want inconsistent blocks in *any* case. */
> -	u64 fe_ptr, bg_ptr, prev_bg_ptr;
> +	u64 bg_ptr, prev_bg_ptr;
>  	struct ocfs2_dinode *fe = (struct ocfs2_dinode *) fe_bh->b_data;
>  	struct ocfs2_group_desc *bg = (struct ocfs2_group_desc *)
bg_bh->b_data;
>  	struct ocfs2_group_desc *prev_bg = (struct ocfs2_group_desc *)
prev_bg_bh->b_data;
> @@ -1437,7 +1437,6 @@ static int ocfs2_relink_block_group(handle_t *handle,
>  		(unsigned long long)le64_to_cpu(bg->bg_blkno),
>  		(unsigned long long)le64_to_cpu(prev_bg->bg_blkno));
>  
> -	fe_ptr = le64_to_cpu(fe->id2.i_chain.cl_recs[chain].c_blkno);
>  	bg_ptr = le64_to_cpu(bg->bg_next_group);
>  	prev_bg_ptr = le64_to_cpu(prev_bg->bg_next_group);
>  
> @@ -1446,7 +1445,7 @@ static int ocfs2_relink_block_group(handle_t *handle,
>  					 OCFS2_JOURNAL_ACCESS_WRITE);
>  	if (status < 0) {
>  		mlog_errno(status);
> -		goto out_rollback;
> +		goto out;
>  	}
>  
>  	prev_bg->bg_next_group = bg->bg_next_group;
> @@ -1456,7 +1455,7 @@ static int ocfs2_relink_block_group(handle_t *handle,
>  					 bg_bh, OCFS2_JOURNAL_ACCESS_WRITE);
>  	if (status < 0) {
>  		mlog_errno(status);
> -		goto out_rollback;
> +		goto out_rollback_prev_bg;
>  	}
>  
>  	bg->bg_next_group = fe->id2.i_chain.cl_recs[chain].c_blkno;
> @@ -1466,21 +1465,21 @@ static int ocfs2_relink_block_group(handle_t
*handle,
>  					 fe_bh, OCFS2_JOURNAL_ACCESS_WRITE);
>  	if (status < 0) {
>  		mlog_errno(status);
> -		goto out_rollback;
> +		goto out_rollback_bg;
>  	}
>  
>  	fe->id2.i_chain.cl_recs[chain].c_blkno = bg->bg_blkno;
>  	ocfs2_journal_dirty(handle, fe_bh);
>  
> -out_rollback:
> -	if (status < 0) {
> -		fe->id2.i_chain.cl_recs[chain].c_blkno = cpu_to_le64(fe_ptr);
> -		bg->bg_next_group = cpu_to_le64(bg_ptr);
> -		prev_bg->bg_next_group = cpu_to_le64(prev_bg_ptr);
> -	}
> +out:
> +	return status;
>  
> -	if (status)
> -		mlog_errno(status);
> +out_rollback_bg:
> +	bg->bg_next_group = cpu_to_le64(bg_ptr);
> +out_rollback_prev_bg:
> +	prev_bg->bg_next_group = cpu_to_le64(prev_bg_ptr);
> +
> +	mlog_errno(status);
>  	return status;
>  }
>