On 10/25/23 21:04, Bernard Metzler wrote:
-----Original Message----- From: Guoqing Jiang<guoqing.jiang@xxxxxxxxx> Sent: Monday, October 9, 2023 9:18 AM To: Bernard Metzler<BMT@xxxxxxxxxxxxxx>;jgg@xxxxxxxx;leon@xxxxxxxxxx Cc:linux-rdma@xxxxxxxxxxxxxxx Subject: [EXTERNAL] [PATCH 13/19] RDMA/siw: Simplify siw_qp_id2obj Let's set qp and return it. Signed-off-by: Guoqing Jiang<guoqing.jiang@xxxxxxxxx> --- drivers/infiniband/sw/siw/siw.h | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) diff --git a/drivers/infiniband/sw/siw/siw.h b/drivers/infiniband/sw/siw/siw.h index 44684b74550f..e127ef493296 100644 --- a/drivers/infiniband/sw/siw/siw.h +++ b/drivers/infiniband/sw/siw/siw.h @@ -601,12 +601,10 @@ static inline struct siw_qp *siw_qp_id2obj(struct siw_device *sdev, int id) rcu_read_lock(); qp = xa_load(&sdev->qp_xa, id); - if (likely(qp && kref_get_unless_zero(&qp->ref))) { - rcu_read_unlock(); - return qp; - } + if (likely(qp && !kref_get_unless_zero(&qp->ref))) + qp = NULL; rcu_read_unlock(); - return NULL; + return qp; } static inline u32 qp_id(struct siw_qp *qp) -- 2.35.3No let's keep it as is. It openly codes the likely case first. Your code makes the unlikely thing likely.
How about change likely to unlikely? If not, I will drop both 13 and 14. Thanks, Guoqing
