OSDN Git Service

io_uring: optimise hot path restricted checks
authorPavel Begunkov <asml.silence@gmail.com>
Sat, 26 Jun 2021 20:40:47 +0000 (21:40 +0100)
committerJens Axboe <axboe@kernel.dk>
Wed, 30 Jun 2021 20:15:40 +0000 (14:15 -0600)
Move likely/unlikely from io_check_restriction() to specifically
ctx->restricted check, because doesn't do what it supposed to and make
the common path take an extra jump.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/22bf70d0a543dfc935d7276bdc73081784e30698.1624739600.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
fs/io_uring.c

index ce88ad5..a5df65f 100644 (file)
@@ -6506,7 +6506,7 @@ static inline bool io_check_restriction(struct io_ring_ctx *ctx,
                                        struct io_kiocb *req,
                                        unsigned int sqe_flags)
 {
-       if (!ctx->restricted)
+       if (likely(!ctx->restricted))
                return true;
 
        if (!test_bit(req->opcode, ctx->restrictions.sqe_op))
@@ -6549,7 +6549,7 @@ static int io_init_req(struct io_ring_ctx *ctx, struct io_kiocb *req,
                return -EINVAL;
        if (unlikely(req->opcode >= IORING_OP_LAST))
                return -EINVAL;
-       if (unlikely(!io_check_restriction(ctx, req, sqe_flags)))
+       if (!io_check_restriction(ctx, req, sqe_flags))
                return -EACCES;
 
        if ((sqe_flags & IOSQE_BUFFER_SELECT) &&