seqlock,lockdep: Fix seqcount_latch_init()
authorPeter Zijlstra <peterz@infradead.org>
Tue, 9 Mar 2021 14:21:18 +0000 (15:21 +0100)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 17 Mar 2021 16:06:34 +0000 (17:06 +0100)
[ Upstream commit 4817a52b306136c8b2b2271d8770401441e4cf79 ]

seqcount_init() must be a macro in order to preserve the static
variable that is used for the lockdep key. Don't then wrap it in an
inline function, which destroys that.

Luckily there aren't many users of this function, but fix it before it
becomes a problem.

Fixes: 80793c3471d9 ("seqlock: Introduce seqcount_latch_t")
Reported-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/YEeFEbNUVkZaXDp4@hirez.programming.kicks-ass.net
Signed-off-by: Sasha Levin <sashal@kernel.org>
include/linux/seqlock.h

index cbfc78b..1ac20d7 100644 (file)
@@ -659,10 +659,7 @@ typedef struct {
  * seqcount_latch_init() - runtime initializer for seqcount_latch_t
  * @s: Pointer to the seqcount_latch_t instance
  */
-static inline void seqcount_latch_init(seqcount_latch_t *s)
-{
-       seqcount_init(&s->seqcount);
-}
+#define seqcount_latch_init(s) seqcount_init(&(s)->seqcount)
 
 /**
  * raw_read_seqcount_latch() - pick even/odd latch data copy