[PATCH v2 2/3] io_uring: batch getting pcpu references

[Pavel Begunkov]

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

basic benchmark with submit and wait 128 non-linked nops showed ~5%
performance gain. (7044 KIOPS vs 7423)

Signed-off-by: Pavel Begunkov <asml.silence@xxxxxxxxx>
---
 fs/io_uring.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 513f1922ce6a..5392134f042f 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -1045,9 +1045,6 @@ 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))
@@ -4391,6 +4388,7 @@ static int io_submit_sqes(struct io_ring_ctx *ctx, unsigned int nr,
 	struct io_submit_state state, *statep = NULL;
 	struct io_kiocb *link = NULL;
 	int i, submitted = 0;
+	unsigned int extra_refs;
 	bool mm_fault = false;
 
 	/* if we have a backlog and couldn't flush it all, return BUSY */
@@ -4400,6 +4398,10 @@ 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;
+	extra_refs = nr;
+
 	if (nr > IO_PLUG_THRESHOLD) {
 		io_submit_state_start(&state, nr);
 		statep = &state;
@@ -4415,6 +4417,7 @@ static int io_submit_sqes(struct io_ring_ctx *ctx, unsigned int nr,
 				submitted = -EAGAIN;
 			break;
 		}
+		--extra_refs;
 		if (!io_get_sqring(ctx, req, &sqe)) {
 			__io_free_req(req);
 			break;
@@ -4451,6 +4454,8 @@ static int io_submit_sqes(struct io_ring_ctx *ctx, unsigned int nr,
 		io_queue_link_head(link);
 	if (statep)
 		io_submit_state_end(&state);
+	if (extra_refs)
+		percpu_ref_put_many(&ctx->refs, extra_refs);
 
 	 /* Commit SQ ring head once we've consumed and submitted all SQEs */
 	io_commit_sqring(ctx);
-- 
2.24.0