On Mon, Feb 05, 2024 at 06:55:44AM -0500, Jeff Layton wrote:

On Mon, 2024-02-05 at 12:36 +0100, Christian Brauner wrote:

quoted

quoted

diff --git a/include/linux/filelock.h b/include/linux/filelock.h
index 085ff6ba0653..a814664b1053 100644
--- a/include/linux/filelock.h
+++ b/include/linux/filelock.h

@@ -147,6 +147,29 @@ int fcntl_setlk64(unsigned int, struct file *, unsigned int,

 int fcntl_setlease(unsigned int fd, struct file *filp, int arg);
 int fcntl_getlease(struct file *filp);
 







+static inline bool lock_is_unlock(struct file_lock *fl)
+{
+	return fl->fl_type == F_UNLCK;
+}
+
+static inline bool lock_is_read(struct file_lock *fl)
+{
+	return fl->fl_type == F_RDLCK;
+}
+
+static inline bool lock_is_write(struct file_lock *fl)
+{
+	return fl->fl_type == F_WRLCK;
+}
+
+static inline void locks_wake_up(struct file_lock *fl)
+{
+	wake_up(&fl->fl_wait);
+}
+
+/* for walking lists of file_locks linked by fl_list */
+#define for_each_file_lock(_fl, _head)	list_for_each_entry(_fl, _head, fl_list)
+

This causes a build warning for fs/ceph/ and fs/afs when
!CONFIG_FILE_LOCKING. I'm about to fold the following diff into this
patch. The diff looks a bit wonky but essentially I've moved
lock_is_unlock(), lock_is_{read,write}(), locks_wake_up() and
for_each_file_lock() out of the ifdef CONFIG_FILE_LOCKING:

I sent a patch for this problem yesterday. Did you not get it?

Whoops, probably missed it on the trip back from fosdem.
I'll double check now.