Skip to contentSkip to conversation

cannot metex_unlock without mutex_lock in ocfs2_remove_btree_range

PATCH New Joseph Qi 2 participants 3 messages Jun 10, 2013 OCFS2

lore
Share & tools
lore

Joseph Qi

author
#0 lore raw

PATCH 1/2 cannot metex_unlock without mutex_lock in ocfs2_remove_btree_range

In ocfs2_remove_btree_range, when calling ocfs2_lock_refcount_tree and ocfs2_prepare_refcount_change_for_del failed, it goes to out and then tries to call mutex_unlock without mutex_lock before.

---
 fs/ocfs2/alloc.c |    6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
fs/ocfs2/alloc.c +3 -3
index b8a9d87..941e883 100644
@@ -5655,7 +5655,7 @@ int ocfs2_remove_btree_range(struct inode *inode,
 					       &ref_tree, NULL);
 		if (ret) {
 			mlog_errno(ret);
-			goto out;
+			goto bail;
 		}
 
 		ret = ocfs2_prepare_refcount_change_for_del(inode,
@@ -5666,7 +5666,7 @@ int ocfs2_remove_btree_range(struct inode *inode,
 							    &extra_blocks);
 		if (ret < 0) {
 			mlog_errno(ret);
-			goto out;
+			goto bail;
 		}
 	}
 
@@ -5734,7 +5734,7 @@ out_commit:
 	ocfs2_commit_trans(osb, handle);
 out:
 	mutex_unlock(&tl_inode->i_mutex);
-
+bail:
 	if (meta_ac)
 		ocfs2_free_alloc_context(meta_ac);
 
Signature · 1 line
1.7.9.7
#1 lore raw

PATCH 1/2 cannot metex_unlock without mutex_lock in ocfs2_remove_btree_range

Quoted 3 lines · Joseph Qi

In ocfs2_remove_btree_range, when calling ocfs2_lock_refcount_tree and ocfs2_prepare_refcount_change_for_del failed, it goes to out and then tries to call mutex_unlock without mutex_lock before.

It's better to merge both into one patch because it's apparent that ocfs2_reserve_blocks_for_rec_trunc() need to fix in the same way for the same reason.

Thanks,
-Jeff

Quoted 37 lines
---
 fs/ocfs2/alloc.c |    6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
fs/ocfs2/alloc.c +3 -3
index b8a9d87..941e883 100644
@@ -5655,7 +5655,7 @@ int ocfs2_remove_btree_range(struct inode *inode,
 					       &ref_tree, NULL);
 		if (ret) {
 			mlog_errno(ret);
-			goto out;
+			goto bail;
 		}
 
 		ret = ocfs2_prepare_refcount_change_for_del(inode,
@@ -5666,7 +5666,7 @@ int ocfs2_remove_btree_range(struct inode *inode,
 							    &extra_blocks);
 		if (ret < 0) {
 			mlog_errno(ret);
-			goto out;
+			goto bail;
 		}
 	}
 
@@ -5734,7 +5734,7 @@ out_commit:
 	ocfs2_commit_trans(osb, handle);
 out:
 	mutex_unlock(&tl_inode->i_mutex);
-
+bail:
 	if (meta_ac)
 		ocfs2_free_alloc_context(meta_ac);
 
author
#2 lore raw

PATCH 1/2 cannot metex_unlock without mutex_lock in ocfs2_remove_btree_range

Quoted 41 lines · Jeff Liu
Quoted 3 lines · Joseph Qi

In ocfs2_remove_btree_range, when calling ocfs2_lock_refcount_tree and ocfs2_prepare_refcount_change_for_del failed, it goes to out and then tries to call mutex_unlock without mutex_lock before.

It's better to merge both into one patch because it's apparent that ocfs2_reserve_blocks_for_rec_trunc() need to fix in the same way for the same reason.

Thanks,
-Jeff

Quoted 28 lines
---
 fs/ocfs2/alloc.c |    6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
fs/ocfs2/alloc.c +2 -2
index b8a9d87..941e883 100644
@@ -5655,7 +5655,7 @@ int ocfs2_remove_btree_range(struct inode *inode,
 					       &ref_tree, NULL);
 		if (ret) {
 			mlog_errno(ret);
-			goto out;
+			goto bail;
 		}
 
 		ret = ocfs2_prepare_refcount_change_for_del(inode,
@@ -5666,7 +5666,7 @@ int ocfs2_remove_btree_range(struct inode *inode,
 							    &extra_blocks);
 		if (ret < 0) {
 			mlog_errno(ret);
-			goto out;
+			goto bail;
 		}
 	}
 
Quoted 13 lines

fs/ocfs2/alloc.c out_commit: · line 5740

Quoted 9 lines
fs/ocfs2/alloc.c +1 -1
@@ -5734,7 +5734,7 @@ out_commit:
 	ocfs2_commit_trans(osb, handle);
 out:
 	mutex_unlock(&tl_inode->i_mutex);
-
+bail:
 	if (meta_ac)
 		ocfs2_free_alloc_context(meta_ac);
 

Thanks for your review. I will merge them and then resend the patch.