Merge tag 'io_uring-6.0-2022-09-02' of git://git.kernel.dk/linux-block

Pull io_uring fixes from Jens Axboe:

 - A single fix for over-eager retries for networking (Pavel)

 - Revert the notification slot support for zerocopy sends.

   It turns out that even after more than a year or development and
   testing, there's not full agreement on whether just using plain
   ordered notifications is Good Enough to avoid the complexity of using
   the notifications slots. Because of that, we decided that it's best
   left to a future final decision.

   We can always bring back this feature, but we can't really change it
   or remove it once we've released 6.0 with it enabled. The reverts
   leave the usual CQE notifications as the primary interface for
   knowing when data was sent, and when it was acked. (Pavel)

* tag 'io_uring-6.0-2022-09-02' of git://git.kernel.dk/linux-block:
  selftests/net: return back io_uring zc send tests
  io_uring/net: simplify zerocopy send user API
  io_uring/notif: remove notif registration
  Revert "io_uring: rename IORING_OP_FILES_UPDATE"
  Revert "io_uring: add zc notification flush requests"
  selftests/net: temporarily disable io_uring zc test
  io_uring/net: fix overexcessive retries

2 files changed, 41 insertions, 70 deletions

diff --git a/tools/testing/selftests/net/io_uring_zerocopy_tx.c b/tools/testing/selftests/net/io_uring_zerocopy_tx.c
index 9d64c560a2d6..8ce48aca8321 100644
--- a/tools/testing/selftests/net/io_uring_zerocopy_tx.c
+++ b/tools/testing/selftests/net/io_uring_zerocopy_tx.c
@@ -47,7 +47,6 @@ enum {
 	MODE_MIXED	= 3,
 };
 
-static bool cfg_flush		= false;
 static bool cfg_cork		= false;
 static int  cfg_mode		= MODE_ZC_FIXED;
 static int  cfg_nr_reqs		= 8;
@@ -166,21 +165,6 @@ static int io_uring_register_buffers(struct io_uring *ring,
 	return (ret < 0) ? -errno : ret;
 }
 
-static int io_uring_register_notifications(struct io_uring *ring,
-					   unsigned nr,
-					   struct io_uring_notification_slot *slots)
-{
-	int ret;
-	struct io_uring_notification_register r = {
-		.nr_slots = nr,
-		.data = (unsigned long)slots,
-	};
-
-	ret = syscall(__NR_io_uring_register, ring->ring_fd,
-		      IORING_REGISTER_NOTIFIERS, &r, sizeof(r));
-	return (ret < 0) ? -errno : ret;
-}
-
 static int io_uring_mmap(int fd, struct io_uring_params *p,
 			 struct io_uring_sq *sq, struct io_uring_cq *cq)
 {
@@ -297,11 +281,10 @@ static inline void io_uring_prep_send(struct io_uring_sqe *sqe, int sockfd,
 
 static inline void io_uring_prep_sendzc(struct io_uring_sqe *sqe, int sockfd,
 				        const void *buf, size_t len, int flags,
-				        unsigned slot_idx, unsigned zc_flags)
+				        unsigned zc_flags)
 {
 	io_uring_prep_send(sqe, sockfd, buf, len, flags);
-	sqe->opcode = (__u8) IORING_OP_SENDZC_NOTIF;
-	sqe->notification_idx = slot_idx;
+	sqe->opcode = (__u8) IORING_OP_SEND_ZC;
 	sqe->ioprio = zc_flags;
 }
 
@@ -374,7 +357,6 @@ static int do_setup_tx(int domain, int type, int protocol)
 
 static void do_tx(int domain, int type, int protocol)
 {
-	struct io_uring_notification_slot b[1] = {{.tag = NOTIF_TAG}};
 	struct io_uring_sqe *sqe;
 	struct io_uring_cqe *cqe;
 	unsigned long packets = 0, bytes = 0;
@@ -390,10 +372,6 @@ static void do_tx(int domain, int type, int protocol)
 	if (ret)
 		error(1, ret, "io_uring: queue init");
 
-	ret = io_uring_register_notifications(&ring, 1, b);
-	if (ret)
-		error(1, ret, "io_uring: tx ctx registration");
-
 	iov.iov_base = payload;
 	iov.iov_len = cfg_payload_len;
 
@@ -409,9 +387,8 @@ static void do_tx(int domain, int type, int protocol)
 		for (i = 0; i < cfg_nr_reqs; i++) {
 			unsigned zc_flags = 0;
 			unsigned buf_idx = 0;
-			unsigned slot_idx = 0;
 			unsigned mode = cfg_mode;
-			unsigned msg_flags = 0;
+			unsigned msg_flags = MSG_WAITALL;
 
 			if (cfg_mode == MODE_MIXED)
 				mode = rand() % 3;
@@ -423,13 +400,10 @@ static void do_tx(int domain, int type, int protocol)
 						   cfg_payload_len, msg_flags);
 				sqe->user_data = NONZC_TAG;
 			} else {
-				if (cfg_flush) {
-					zc_flags |= IORING_RECVSEND_NOTIF_FLUSH;
-					compl_cqes++;
-				}
+				compl_cqes++;
 				io_uring_prep_sendzc(sqe, fd, payload,
 						     cfg_payload_len,
-						     msg_flags, slot_idx, zc_flags);
+						     msg_flags, zc_flags);
 				if (mode == MODE_ZC_FIXED) {
 					sqe->ioprio |= IORING_RECVSEND_FIXED_BUF;
 					sqe->buf_index = buf_idx;
@@ -442,51 +416,57 @@ static void do_tx(int domain, int type, int protocol)
 		if (ret != cfg_nr_reqs)
 			error(1, ret, "submit");
 
+		if (cfg_cork)
+			do_setsockopt(fd, IPPROTO_UDP, UDP_CORK, 0);
 		for (i = 0; i < cfg_nr_reqs; i++) {
 			ret = io_uring_wait_cqe(&ring, &cqe);
 			if (ret)
 				error(1, ret, "wait cqe");
 
-			if (cqe->user_data == NOTIF_TAG) {
+			if (cqe->user_data != NONZC_TAG &&
+			    cqe->user_data != ZC_TAG)
+				error(1, -EINVAL, "invalid cqe->user_data");
+
+			if (cqe->flags & IORING_CQE_F_NOTIF) {
+				if (cqe->flags & IORING_CQE_F_MORE)
+					error(1, -EINVAL, "invalid notif flags");
 				compl_cqes--;
 				i--;
-			} else if (cqe->user_data != NONZC_TAG &&
-				   cqe->user_data != ZC_TAG) {
-				error(1, cqe->res, "invalid user_data");
-			} else if (cqe->res <= 0 && cqe->res != -EAGAIN) {
+			} else if (cqe->res <= 0) {
+				if (cqe->flags & IORING_CQE_F_MORE)
+					error(1, cqe->res, "more with a failed send");
 				error(1, cqe->res, "send failed");
 			} else {
-				if (cqe->res > 0) {
-					packets++;
-					bytes += cqe->res;
-				}
-				/* failed requests don't flush */
-				if (cfg_flush &&
-				    cqe->res <= 0 &&
-				    cqe->user_data == ZC_TAG)
-					compl_cqes--;
+				if (cqe->user_data == ZC_TAG &&
+				    !(cqe->flags & IORING_CQE_F_MORE))
+					error(1, cqe->res, "missing more flag");
+				packets++;
+				bytes += cqe->res;
 			}
 			io_uring_cqe_seen(&ring);
 		}
-		if (cfg_cork)
-			do_setsockopt(fd, IPPROTO_UDP, UDP_CORK, 0);
 	} while (gettimeofday_ms() < tstop);
 
-	if (close(fd))
-		error(1, errno, "close");
-
-	fprintf(stderr, "tx=%lu (MB=%lu), tx/s=%lu (MB/s=%lu)\n",
-			packets, bytes >> 20,
-			packets / (cfg_runtime_ms / 1000),
-			(bytes >> 20) / (cfg_runtime_ms / 1000));
-
 	while (compl_cqes) {
 		ret = io_uring_wait_cqe(&ring, &cqe);
 		if (ret)
 			error(1, ret, "wait cqe");
+		if (cqe->flags & IORING_CQE_F_MORE)
+			error(1, -EINVAL, "invalid notif flags");
+		if (!(cqe->flags & IORING_CQE_F_NOTIF))
+			error(1, -EINVAL, "missing notif flag");
+
 		io_uring_cqe_seen(&ring);
 		compl_cqes--;
 	}
+
+	fprintf(stderr, "tx=%lu (MB=%lu), tx/s=%lu (MB/s=%lu)\n",
+			packets, bytes >> 20,
+			packets / (cfg_runtime_ms / 1000),
+			(bytes >> 20) / (cfg_runtime_ms / 1000));
+
+	if (close(fd))
+		error(1, errno, "close");
 }
 
 static void do_test(int domain, int type, int protocol)
@@ -500,8 +480,8 @@ static void do_test(int domain, int type, int protocol)
 
 static void usage(const char *filepath)
 {
-	error(1, 0, "Usage: %s [-f] [-n<N>] [-z0] [-s<payload size>] "
-		    "(-4|-6) [-t<time s>] -D<dst_ip> udp", filepath);
+	error(1, 0, "Usage: %s (-4|-6) (udp|tcp) -D<dst_ip> [-s<payload size>] "
+		    "[-t<time s>] [-n<batch>] [-p<port>] [-m<mode>]", filepath);
 }
 
 static void parse_opts(int argc, char **argv)
@@ -519,7 +499,7 @@ static void parse_opts(int argc, char **argv)
 		usage(argv[0]);
 	cfg_payload_len = max_payload_len;
 
-	while ((c = getopt(argc, argv, "46D:p:s:t:n:fc:m:")) != -1) {
+	while ((c = getopt(argc, argv, "46D:p:s:t:n:c:m:")) != -1) {
 		switch (c) {
 		case '4':
 			if (cfg_family != PF_UNSPEC)
@@ -548,9 +528,6 @@ static void parse_opts(int argc, char **argv)
 		case 'n':
 			cfg_nr_reqs = strtoul(optarg, NULL, 0);
 			break;
-		case 'f':
-			cfg_flush = 1;
-			break;
 		case 'c':
 			cfg_cork = strtol(optarg, NULL, 0);
 			break;
@@ -583,8 +560,6 @@ static void parse_opts(int argc, char **argv)
 
 	if (cfg_payload_len > max_payload_len)
 		error(1, 0, "-s: payload exceeds max (%d)", max_payload_len);
-	if (cfg_mode == MODE_NONZC && cfg_flush)
-		error(1, 0, "-f: only zerocopy modes support notifications");
 	if (optind != argc - 1)
 		usage(argv[0]);
 }
diff --git a/tools/testing/selftests/net/io_uring_zerocopy_tx.sh b/tools/testing/selftests/net/io_uring_zerocopy_tx.sh
index 6a65e4437640..32aa6e9dacc2 100755
--- a/tools/testing/selftests/net/io_uring_zerocopy_tx.sh
+++ b/tools/testing/selftests/net/io_uring_zerocopy_tx.sh
@@ -25,15 +25,11 @@ readonly path_sysctl_mem="net.core.optmem_max"
 # No arguments: automated test
 if [[ "$#" -eq "0" ]]; then
 	IPs=( "4" "6" )
-	protocols=( "tcp" "udp" )
 
 	for IP in "${IPs[@]}"; do
-		for proto in "${protocols[@]}"; do
-			for mode in $(seq 1 3); do
-				$0 "$IP" "$proto" -m "$mode" -t 1 -n 32
-				$0 "$IP" "$proto" -m "$mode" -t 1 -n 32 -f
-				$0 "$IP" "$proto" -m "$mode" -t 1 -n 32 -c -f
-			done
+		for mode in $(seq 1 3); do
+			$0 "$IP" udp -m "$mode" -t 1 -n 32
+			$0 "$IP" tcp -m "$mode" -t 1 -n 32
 		done
 	done