nvmet: use try_cmpxchg in nvmet_update_sq_head

Use try_cmpxchg instead of cmpxchg (*ptr, old, new) == old in
nvmet_update_sq_head.  x86 CMPXCHG instruction returns success in ZF flag, so
this change saves a compare after cmpxchg (and related move instruction in
front of cmpxchg).

Also, try_cmpxchg implicitly assigns old *ptr value to "old" when cmpxchg
fails. There is no need to re-read the value in the loop.

Note that the value from *ptr should be read using READ_ONCE to prevent
the compiler from merging, refetching or reordering the read.

No functional change intended.

Signed-off-by: Uros Bizjak <ubizjak@gmail.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>

diff --git a/drivers/nvme/target/core.c b/drivers/nvme/target/core.c
index aecb585..560b6bb 100644 (file)
--- a/drivers/nvme/target/core.c
+++ b/drivers/nvme/target/core.c
@@ -695,11 +695,10 @@ static void nvmet_update_sq_head(struct nvmet_req *req)
        if (req->sq->size) {
                u32 old_sqhd, new_sqhd;
 
+               old_sqhd = READ_ONCE(req->sq->sqhd);
                do {
-                       old_sqhd = req->sq->sqhd;
                        new_sqhd = (old_sqhd + 1) % req->sq->size;
-               } while (cmpxchg(&req->sq->sqhd, old_sqhd, new_sqhd) !=
-                                       old_sqhd);
+               } while (!try_cmpxchg(&req->sq->sqhd, &old_sqhd, new_sqhd));
        }
        req->cqe->sq_head = cpu_to_le16(req->sq->sqhd & 0x0000FFFF);
 }