rpc_pipe_msg should be protected by inode->i_lock when purge it.
Subsystems (3)
RFC PATCH 1/1 rpc_pipe_msg should be protected by inode->i_lock when purge it.
Maybe rpc_pipe_msg should be protected by inode->i_lock when purge the free_list.
Signed-off-by Bian Naimeng <[email protected]>
--- net/sunrpc/rpc_pipe.c | 2 ++ 1 files changed, 2 insertions(+), 0 deletions(-)
net/sunrpc/rpc_pipe.c +2 -0
@@ -47,8 +47,10 @@ static void rpc_purge_list(struct rpc_inode *rpci, struct list_head *head,return ;do { msg = list_entry(head->next, struct rpc_pipe_msg, list); +spin_lock (&rpci->vfs_inode->i_lock);list_del (&msg->list); msg->errno = err; +spin_unlock (&rpci->vfs_inode->i_lock);destroy_msg (msg); }while (!list_empty (head));wake_up (&rpci->waitq);
Signature · 7 lines
1.6.5.2 -- Regards Bian Naimeng
Quoted 28 lines · Bian Naimeng
Maybe
rpc_pipe_msgshould be protected byinode->i_lockwhen purge thefree_list.Signed-off-by Bian Naimeng <[email protected]>--- net/sunrpc/rpc_pipe.c | 2 ++ 1 files changed, 2 insertions(+), 0 deletions(-)net/sunrpc/rpc_pipe.c +2
-0@@ -47,8 +47,10 @@ static void rpc_purge_list(struct rpc_inode *rpci, struct list_head *head,return ;do { msg = list_entry(head->next, struct rpc_pipe_msg, list); +spin_lock (&rpci->vfs_inode->i_lock);list_del (&msg->list); msg->errno = err; +spin_unlock (&rpci->vfs_inode->i_lock);destroy_msg (msg); }while (!list_empty (head));wake_up (&rpci->waitq);Signature · 1 line
1.6.5.2
No, we don't need to protect the list 'head'.
Trond