[PATCH] fs: reduce work in fdget_pos()

[Mateusz Guzik]

1. predict the file was found
2. explicitly compare the ref to "one", ignoring the dead zone

The latter arguably improves the behavior to begin with. Suppose the
count turned bad -- the previously used ref routine is going to check
for it and return 0, indicating the count does not necessitate taking
->f_pos_lock. But there very well may be several users.

i.e. not paying for special-casing the dead zone improves semantics.

Sizes are as follows (in bytes; gcc 13, x86-64):
stock:		316
likely(): 	296
likely()+ref:	278

Signed-off-by: Mateusz Guzik <mjguzik@xxxxxxxxx>
---
 fs/file.c                |  5 +++--
 include/linux/file_ref.h | 14 ++++++++++++++
 2 files changed, 17 insertions(+), 2 deletions(-)

diff --git a/fs/file.c b/fs/file.c
index ddefb5c80398..190463e61934 100644
--- a/fs/file.c
+++ b/fs/file.c
@@ -1193,7 +1193,8 @@ struct fd fdget_raw(unsigned int fd)
 static inline bool file_needs_f_pos_lock(struct file *file)
 {
 	return (file->f_mode & FMODE_ATOMIC_POS) &&
-		(file_count(file) > 1 || file->f_op->iterate_shared);
+		(__file_ref_read_raw(&file->f_ref) != FILE_REF_ONEREF ||
+		file->f_op->iterate_shared);
 }
 
 bool file_seek_cur_needs_f_lock(struct file *file)
@@ -1211,7 +1212,7 @@ struct fd fdget_pos(unsigned int fd)
 	struct fd f = fdget(fd);
 	struct file *file = fd_file(f);
 
-	if (file && file_needs_f_pos_lock(file)) {
+	if (likely(file) && file_needs_f_pos_lock(file)) {
 		f.word |= FDPUT_POS_UNLOCK;
 		mutex_lock(&file->f_pos_lock);
 	}
diff --git a/include/linux/file_ref.h b/include/linux/file_ref.h
index 6ef92d765a66..7db62fbc0500 100644
--- a/include/linux/file_ref.h
+++ b/include/linux/file_ref.h
@@ -208,4 +208,18 @@ static inline unsigned long file_ref_read(file_ref_t *ref)
 	return c >= FILE_REF_RELEASED ? 0 : c + 1;
 }
 
+/*
+ * __file_ref_read_raw - Return the value stored in ref->refcnt
+ * @ref: Pointer to the reference count
+ *
+ * Return: The raw value found in the counter
+ *
+ * A hack for file_needs_f_pos_lock(), you probably want to use
+ * file_ref_read() instead.
+ */
+static inline unsigned long __file_ref_read_raw(file_ref_t *ref)
+{
+	return atomic_long_read(&ref->refcnt);
+}
+
 #endif
-- 
2.43.0