]> Pileus Git - ~andy/linux/commitdiff
ocfs2: rework transaction rollback in ocfs2_relink_block_group()
authorJie Liu <jeff.liu@oracle.com>
Wed, 3 Jul 2013 22:00:59 +0000 (15:00 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Wed, 3 Jul 2013 23:07:24 +0000 (16:07 -0700)
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@oracle.com>
Cc: Younger Liu <younger.liu@huawei.com>
Cc: Sunil Mushran <sunil.mushran@gmail.com>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
fs/ocfs2/suballoc.c

index b7e74b580c0f98bdbdf116ab6d43a45d5a3e0ed7..101d16d78162dc9c9ea6d6c089a8072445aafd16 100644 (file)
@@ -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;
 }