io_uring: batch getting pcpu references

percpu_ref_tryget() has its own overhead. Instead getting a reference
for each request, grab a bunch once per io_submit_sqes().

~5% throughput boost for a "submit and wait 128 nops" benchmark.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
__io_req_free_empty() -> __io_req_do_free()

Signed-off-by: Jens Axboe <axboe@kernel.dk>

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 17a199c99c7cca46be5f30246e83ba3a7c7f4c46..9b869fb6c6353dbe7658cc7af03f01777934bf0d 100644 (file)
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -1083,9 +1083,6 @@ static struct io_kiocb *io_get_req(struct io_ring_ctx *ctx,
        gfp_t gfp = GFP_KERNEL | __GFP_NOWARN;
        struct io_kiocb *req;
 
-       if (!percpu_ref_tryget(&ctx->refs))
-               return NULL;
-
        if (!state) {
                req = kmem_cache_alloc(req_cachep, gfp);
                if (unlikely(!req))
@@ -1145,6 +1142,14 @@ static void io_free_req_many(struct io_ring_ctx *ctx, void **reqs, int *nr)
        }
 }
 
+static void __io_req_do_free(struct io_kiocb *req)
+{
+       if (likely(!io_is_fallback_req(req)))
+               kmem_cache_free(req_cachep, req);
+       else
+               clear_bit_unlock(0, (unsigned long *) req->ctx->fallback_req);
+}
+
 static void __io_free_req(struct io_kiocb *req)
 {
        struct io_ring_ctx *ctx = req->ctx;
@@ -1166,11 +1171,9 @@ static void __io_free_req(struct io_kiocb *req)
                        wake_up(&ctx->inflight_wait);
                spin_unlock_irqrestore(&ctx->inflight_lock, flags);
        }
-       percpu_ref_put(&ctx->refs);
-       if (likely(!io_is_fallback_req(req)))
-               kmem_cache_free(req_cachep, req);
-       else
-               clear_bit_unlock(0, (unsigned long *) ctx->fallback_req);
+
+       percpu_ref_put(&req->ctx->refs);
+       __io_req_do_free(req);
 }
 
 static bool io_link_cancel_timeout(struct io_kiocb *req)
@@ -4539,6 +4542,9 @@ static int io_submit_sqes(struct io_ring_ctx *ctx, unsigned int nr,
                        return -EBUSY;
        }
 
+       if (!percpu_ref_tryget_many(&ctx->refs, nr))
+               return -EAGAIN;
+
        if (nr > IO_PLUG_THRESHOLD) {
                io_submit_state_start(&state, nr);
                statep = &state;
@@ -4555,7 +4561,7 @@ static int io_submit_sqes(struct io_ring_ctx *ctx, unsigned int nr,
                        break;
                }
                if (!io_get_sqring(ctx, req, &sqe)) {
-                       __io_free_req(req);
+                       __io_req_do_free(req);
                        break;
                }
 
@@ -4586,6 +4592,8 @@ static int io_submit_sqes(struct io_ring_ctx *ctx, unsigned int nr,
                        break;
        }
 
+       if (submitted != nr)
+               percpu_ref_put_many(&ctx->refs, nr - submitted);
        if (link)
                io_queue_link_head(link);
        if (statep)