ocfs2: should call ocfs2_journal_access_di() before ocfs2_delete_entry() in ocfs2_orphan_del()
PATCH 8/11 ocfs2: should call ocfs2_journal_access_di() before ocfs2_delete_entry() in ocfs2_orphan_del()
From: Younger Liu <[email protected]>
Subject: ocfs2: should call ocfs2_journal_access_di() before ocfs2_delete_entry() in ocfs2_orphan_del()
While deleting a file into orphan dir in ocfs2_orphan_del(), it calls ocfs2_delete_entry() before ocfs2_journal_access_di(). If ocfs2_delete_entry() succeeded and ocfs2_journal_access_di() failed, there would be a inconsistency: the file is deleted from orphan dir, but orphan dir dinode is not updated.
So we need to call ocfs2_journal_access_di() before ocfs2_orphan_del().
---
fs/ocfs2/namei.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)
fs/ocfs2/namei.c renamed from fs/ocfs2/namei.c~ocfs2-should-call-ocfs2_journal_access_di-before-ocfs2_delete_entry-in-ocfs2_orphan_del +6 -6
@@ -2207,17 +2207,17 @@ int ocfs2_orphan_del(struct ocfs2_super goto leave; } -/* remove it from the orphan directory */ - status =ocfs2_delete_entry (handle, orphan_dir_inode, &lookup); + status =ocfs2_journal_access_di (handle, +INODE_CACHE (orphan_dir_inode), + orphan_dir_bh, +OCFS2_JOURNAL_ACCESS_WRITE );if (status <0 ) {mlog_errno (status); goto leave; } - status =ocfs2_journal_access_di (handle, -INODE_CACHE (orphan_dir_inode), - orphan_dir_bh, -OCFS2_JOURNAL_ACCESS_WRITE ); +/* remove it from the orphan directory */ + status =ocfs2_delete_entry (handle, orphan_dir_inode, &lookup);if (status <0 ) {mlog_errno (status); goto leave;
_
PATCH 8/11 ocfs2: should call ocfs2_journal_access_di() before ocfs2_delete_entry() in ocfs2_orphan_del()
Quoted 10 lines · akpm at linux-foundation.org
From: Younger Liu <[email protected]>
Subject: ocfs2: should callocfs2_journal_access_di()beforeocfs2_delete_entry()inocfs2_orphan_del()While deleting a file into orphan dir in
ocfs2_orphan_del(), it callsocfs2_delete_entry()beforeocfs2_journal_access_di(). Ifocfs2_delete_entry()succeeded andocfs2_journal_access_di()failed, there would be a inconsistency: the file is deleted from orphan dir, but orphan dir dinode is not updated.So we need to call
ocfs2_journal_access_di()beforeocfs2_orphan_del().
I am confused by this. Re-reading the code in fs/ocfs2/dir.c, __ocfs2_delete_entry() takes great pains to mark the buffer head as part of the journal:
ocfs2_journal_access_func access = ocfs2_journal_access_db; if (OCFS2_I(dir)->ip_dyn_features & OCFS2_INLINE_DATA_FL) access = ocfs2_journal_access_di;
<snip>
status = access(handle, INODE_CACHE(dir), bh, OCFS2_JOURNAL_ACCESS_WRITE);
Am I missing something here?
--Mark
--
Mark Fasheh
PATCH 8/11 ocfs2: should call ocfs2_journal_access_di() before ocfs2_delete_entry() in ocfs2_orphan_del()
Quoted 21 lines · Mark Fasheh
Quoted 10 lines · akpm at linux-foundation.org
From: Younger Liu <[email protected]>
Subject: ocfs2: should callocfs2_journal_access_di()beforeocfs2_delete_entry()inocfs2_orphan_del()While deleting a file into orphan dir in
ocfs2_orphan_del(), it callsocfs2_delete_entry()beforeocfs2_journal_access_di(). Ifocfs2_delete_entry()succeeded andocfs2_journal_access_di()failed, there would be a inconsistency: the file is deleted from orphan dir, but orphan dir dinode is not updated.So we need to call
ocfs2_journal_access_di()beforeocfs2_orphan_del().I am confused by this. Re-reading the code in
fs/ocfs2/dir.c,__ocfs2_delete_entry()takes great pains to mark the buffer head as part of the journal:ocfs2_journal_access_func access = ocfs2_journal_access_db; if (OCFS2_I(dir)->ip_dyn_features & OCFS2_INLINE_DATA_FL) access = ocfs2_journal_access_di;
Quoted 8 lines
<snip>
status = access(handle, INODE_CACHE(dir), bh, OCFS2_JOURNAL_ACCESS_WRITE);Am I missing something here?
--Mark
Hi Mark
I am sorry to reply so late.
I can understand your meaning, But if OCFS2_I(dir) is not inline
data, access = ocfs2_journal_access_db.
Moreover, logically speaking, ocfs2_journal_access_di() should not
been called after ocfs2_delete_entry(). Otherwise, there is a potential risk for inconsistency.
--Younger
Quoted 9 lines
--
Mark Fasheh_______________________________________________
Ocfs2-devel mailing list
Ocfs2-devel at oss.oracle.com
https://oss.oracle.com/mailman/listinfo/ocfs2-devel