[PATCH liburing 0/5] zcrx example and other changes

[PATCH liburing 0/5] zcrx example and other changes

[Pavel Begunkov]

We need a simple example for zcrx to show case how the api works
and how to use features. Patch 5 is a brushed up version of the
zcrx selftest.

Apart from that update headers and use the kernel return zcrx id.

Pavel Begunkov (5):
  Update io_uring.h for zcrx
  tests/zcrx: rename a test
  tests/zcrx: use returned right zcrx id
  examples: add extra helpers
  examples: add a zcrx example

 examples/Makefile               |   1 +
 examples/helpers.c              |  15 ++
 examples/helpers.h              |   6 +
 examples/zcrx.c                 | 335 ++++++++++++++++++++++++++++++++
 src/include/liburing/io_uring.h |   4 +-
 test/zcrx.c                     |  21 +-
 6 files changed, 372 insertions(+), 10 deletions(-)
 create mode 100644 examples/zcrx.c

-- 
2.48.1

[PATCH liburing 1/5] Update io_uring.h for zcrx

[Pavel Begunkov]

Pull in an updated version if struct io_uring_zcrx_ifq_reg

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 src/include/liburing/io_uring.h | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/src/include/liburing/io_uring.h b/src/include/liburing/io_uring.h
index cb9f82fa..572ff59f 100644
--- a/src/include/liburing/io_uring.h
+++ b/src/include/liburing/io_uring.h
@@ -977,7 +977,9 @@ struct io_uring_zcrx_ifq_reg {
 	__u64	region_ptr; /* struct io_uring_region_desc * */
 
 	struct io_uring_zcrx_offsets offsets;
-	__u64	__resv[4];
+	__u32	zcrx_id;
+	__u32	__resv2;
+	__u64	__resv[3];
 };
 
 #ifdef __cplusplus
-- 
2.48.1

[PATCH liburing 2/5] tests/zcrx: rename a test

[Pavel Begunkov]

There should've been some mishap for a test to be called
"test_invalid_invalid_request", rename it.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 test/zcrx.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/test/zcrx.c b/test/zcrx.c
index 95230c65..0fa8f2bd 100644
--- a/test/zcrx.c
+++ b/test/zcrx.c
@@ -401,7 +401,7 @@ static struct io_uring_cqe *submit_and_wait_one(struct io_uring *ring)
 	return cqe;
 }
 
-static int test_invalid_invalid_request(void *area)
+static int test_invalid_zcrx_request(void *area)
 {
 	struct io_uring_cqe *cqe;
 	struct io_uring_sqe *sqe;
@@ -901,7 +901,7 @@ int main(int argc, char *argv[])
 		return ret;
 	}
 
-	ret = test_invalid_invalid_request(area);
+	ret = test_invalid_zcrx_request(area);
 	if (ret) {
 		fprintf(stderr, "test_invalid_ifq_collision failed\n");
 		return ret;
-- 
2.48.1

[PATCH liburing 3/5] tests/zcrx: use returned right zcrx id

[Pavel Begunkov]

Now we return the zcrx id back to the user, use it.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 test/zcrx.c | 17 ++++++++++-------
 1 file changed, 10 insertions(+), 7 deletions(-)

diff --git a/test/zcrx.c b/test/zcrx.c
index 0fa8f2bd..b60462a6 100644
--- a/test/zcrx.c
+++ b/test/zcrx.c
@@ -344,7 +344,7 @@ static int test_area_access(void)
 	return T_EXIT_PASS;
 }
 
-static int create_ring_with_ifq(struct io_uring *ring, void *area)
+static int create_ring_with_ifq(struct io_uring *ring, void *area, __u32 *id)
 {
 	struct io_uring_zcrx_area_reg area_reg = {
 		.addr = (__u64)(unsigned long)area,
@@ -371,6 +371,7 @@ static int create_ring_with_ifq(struct io_uring *ring, void *area)
 		fprintf(stderr, "ifq register failed %d\n", ret);
 		return T_EXIT_FAIL;
 	}
+	*id = reg.zcrx_id;
 	return 0;
 }
 
@@ -406,9 +407,10 @@ static int test_invalid_zcrx_request(void *area)
 	struct io_uring_cqe *cqe;
 	struct io_uring_sqe *sqe;
 	struct io_uring ring;
+	__u32 zcrx_id;
 	int ret, fds[2];
 
-	ret = create_ring_with_ifq(&ring, area);
+	ret = create_ring_with_ifq(&ring, area, &zcrx_id);
 	if (ret != T_SETUP_OK) {
 		fprintf(stderr, "ifq-ring create failed: %d\n", ret);
 		return T_EXIT_FAIL;
@@ -422,7 +424,7 @@ static int test_invalid_zcrx_request(void *area)
 
 	/* invalid file */
 	sqe = io_uring_get_sqe(&ring);
-	test_io_uring_prep_zcrx(sqe, ring.ring_fd, 0);
+	test_io_uring_prep_zcrx(sqe, ring.ring_fd, zcrx_id);
 
 	cqe = submit_and_wait_one(&ring);
 	if (!cqe) {
@@ -441,7 +443,7 @@ static int test_invalid_zcrx_request(void *area)
 
 	/* invalid ifq idx */
 	sqe = io_uring_get_sqe(&ring);
-	test_io_uring_prep_zcrx(sqe, fds[0], 1);
+	test_io_uring_prep_zcrx(sqe, fds[0], zcrx_id + 1);
 
 	cqe = submit_and_wait_one(&ring);
 	if (!cqe) {
@@ -478,7 +480,8 @@ struct recv_data {
 	struct io_uring_zcrx_rq rq_ring;
 };
 
-static int recv_prep(struct io_uring *ring, struct recv_data *rd, int *sock)
+static int recv_prep(struct io_uring *ring, struct recv_data *rd, int *sock,
+		     __u32 zcrx_id)
 {
 	struct sockaddr_in saddr;
 	struct io_uring_sqe *sqe;
@@ -526,7 +529,7 @@ static int recv_prep(struct io_uring *ring, struct recv_data *rd, int *sock)
 
 	sqe = io_uring_get_sqe(ring);
 	io_uring_prep_rw(IORING_OP_RECV_ZC, sqe, use_fd, NULL, 0, 0);
-	sqe->zcrx_ifq_idx = 0;
+	sqe->zcrx_ifq_idx = zcrx_id;
 	sqe->ioprio |= IORING_RECV_MULTISHOT;
 	sqe->user_data = 2;
 
@@ -693,7 +696,7 @@ static void *recv_fn(void *data)
 	rd->rq_ring.rq_tail = 0;
 	rd->rq_ring.ring_entries = reg.rq_entries;
 
-	ret = recv_prep(&ring, rd, &sock);
+	ret = recv_prep(&ring, rd, &sock, reg.zcrx_id);
 	if (ret) {
 		fprintf(stderr, "recv_prep failed: %d\n", ret);
 		goto err;
-- 
2.48.1

[PATCH liburing 4/5] examples: add extra helpers

[Pavel Begunkov]

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 examples/helpers.c | 15 +++++++++++++++
 examples/helpers.h |  6 ++++++
 2 files changed, 21 insertions(+)

diff --git a/examples/helpers.c b/examples/helpers.c
index 59b31ecb..0c9e0160 100644
--- a/examples/helpers.c
+++ b/examples/helpers.c
@@ -9,6 +9,7 @@
 #include <sys/socket.h>
 #include <sys/time.h>
 #include <unistd.h>
+#include <stdarg.h>
 
 #include "helpers.h"
 
@@ -70,3 +71,17 @@ void *aligned_alloc(size_t alignment, size_t size)
 
 	return ret;
 }
+
+void t_error(int status, int errnum, const char *format, ...)
+{
+	va_list args;
+	va_start(args, format);
+
+	vfprintf(stderr, format, args);
+	if (errnum)
+		fprintf(stderr, ": %s", strerror(errnum));
+
+	fprintf(stderr, "\n");
+	va_end(args);
+	exit(status);
+}
diff --git a/examples/helpers.h b/examples/helpers.h
index d73ee4a5..c8bf7332 100644
--- a/examples/helpers.h
+++ b/examples/helpers.h
@@ -2,6 +2,10 @@
 #ifndef LIBURING_EX_HELPERS_H
 #define LIBURING_EX_HELPERS_H
 
+#include <stddef.h>
+
+#define T_ALIGN_UP(v, align) (((v) + (align) - 1) & ~((align) - 1))
+
 int setup_listening_socket(int port, int ipv6);
 
 /*
@@ -11,4 +15,6 @@ int setup_listening_socket(int port, int ipv6);
  */
 void *aligned_alloc(size_t alignment, size_t size);
 
+void t_error(int status, int errnum, const char *format, ...);
+
 #endif
-- 
2.48.1

[PATCH liburing 5/5] examples: add a zcrx example

[Pavel Begunkov]

Copy-pasted from selftests/.../iou_zcrx.c by David and brushed up on
top.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 examples/Makefile |   1 +
 examples/zcrx.c   | 335 ++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 336 insertions(+)
 create mode 100644 examples/zcrx.c

diff --git a/examples/Makefile b/examples/Makefile
index 7b740ca9..47bdfbf3 100644
--- a/examples/Makefile
+++ b/examples/Makefile
@@ -32,6 +32,7 @@ example_srcs := \
 	send-zerocopy.c \
 	rsrc-update-bench.c \
 	proxy.c \
+	zcrx.c \
 	kdigest.c
 
 all_targets :=
diff --git a/examples/zcrx.c b/examples/zcrx.c
new file mode 100644
index 00000000..8393cfe3
--- /dev/null
+++ b/examples/zcrx.c
@@ -0,0 +1,335 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <assert.h>
+#include <errno.h>
+#include <error.h>
+#include <fcntl.h>
+#include <limits.h>
+#include <stdbool.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <stdarg.h>
+
+#include <arpa/inet.h>
+#include <linux/errqueue.h>
+#include <linux/if_packet.h>
+#include <linux/ipv6.h>
+#include <linux/socket.h>
+#include <linux/sockios.h>
+#include <net/ethernet.h>
+#include <net/if.h>
+#include <netinet/in.h>
+#include <netinet/ip.h>
+#include <netinet/ip6.h>
+#include <netinet/tcp.h>
+#include <netinet/udp.h>
+#include <sys/epoll.h>
+#include <sys/ioctl.h>
+#include <sys/mman.h>
+#include <sys/resource.h>
+#include <sys/socket.h>
+#include <sys/stat.h>
+#include <sys/time.h>
+#include <sys/types.h>
+#include <sys/un.h>
+#include <sys/wait.h>
+
+#include "liburing.h"
+#include "helpers.h"
+
+#define PAGE_SIZE (4096)
+#define AREA_SIZE (8192 * PAGE_SIZE)
+#define SEND_SIZE (512 * 4096)
+
+static int cfg_port = 8000;
+static const char *cfg_ifname;
+static int cfg_queue_id = -1;
+static bool cfg_oneshot;
+static int cfg_oneshot_recvs;
+static bool cfg_verify_data = false;
+static struct sockaddr_in6 cfg_addr;
+
+static void *area_ptr;
+static void *ring_ptr;
+static size_t ring_size;
+static struct io_uring_zcrx_rq rq_ring;
+static unsigned long area_token;
+static int connfd;
+static bool stop;
+static size_t received;
+static __u32 zcrx_id;
+
+static inline size_t get_refill_ring_size(unsigned int rq_entries)
+{
+	ring_size = rq_entries * sizeof(struct io_uring_zcrx_rqe);
+	/* add space for the header (head/tail/etc.) */
+	ring_size += PAGE_SIZE;
+	return T_ALIGN_UP(ring_size, 4096);
+}
+
+static void setup_zcrx(struct io_uring *ring)
+{
+	unsigned int ifindex;
+	unsigned int rq_entries = 4096;
+	int ret;
+
+	ifindex = if_nametoindex(cfg_ifname);
+	if (!ifindex)
+		t_error(1, 0, "bad interface name: %s", cfg_ifname);
+
+	area_ptr = mmap(NULL,
+			AREA_SIZE,
+			PROT_READ | PROT_WRITE,
+			MAP_ANONYMOUS | MAP_PRIVATE,
+			0,
+			0);
+	if (area_ptr == MAP_FAILED)
+		t_error(1, 0, "mmap(): zero copy area");
+
+	ring_size = get_refill_ring_size(rq_entries);
+	ring_ptr = mmap(NULL,
+			ring_size,
+			PROT_READ | PROT_WRITE,
+			MAP_ANONYMOUS | MAP_PRIVATE,
+			0,
+			0);
+	if (ring_ptr == MAP_FAILED)
+		t_error(1, 0, "mmap(): refill ring");
+
+	struct io_uring_region_desc region_reg = {
+		.size = ring_size,
+		.user_addr = (__u64)(unsigned long)ring_ptr,
+		.flags = IORING_MEM_REGION_TYPE_USER,
+	};
+
+	struct io_uring_zcrx_area_reg area_reg = {
+		.addr = (__u64)(unsigned long)area_ptr,
+		.len = AREA_SIZE,
+		.flags = 0,
+	};
+
+	struct io_uring_zcrx_ifq_reg reg = {
+		.if_idx = ifindex,
+		.if_rxq = cfg_queue_id,
+		.rq_entries = rq_entries,
+		.area_ptr = (__u64)(unsigned long)&area_reg,
+		.region_ptr = (__u64)(unsigned long)&region_reg,
+	};
+
+	ret = io_uring_register_ifq(ring, &reg);
+	if (ret)
+		t_error(1, 0, "io_uring_register_ifq(): %d", ret);
+
+	rq_ring.khead = (unsigned int *)((char *)ring_ptr + reg.offsets.head);
+	rq_ring.ktail = (unsigned int *)((char *)ring_ptr + reg.offsets.tail);
+	rq_ring.rqes = (struct io_uring_zcrx_rqe *)((char *)ring_ptr + reg.offsets.rqes);
+	rq_ring.rq_tail = 0;
+	rq_ring.ring_entries = reg.rq_entries;
+
+	zcrx_id = reg.zcrx_id;
+	area_token = area_reg.rq_area_token;
+}
+
+static void add_accept(struct io_uring *ring, int sockfd)
+{
+	struct io_uring_sqe *sqe = io_uring_get_sqe(ring);
+
+	io_uring_prep_accept(sqe, sockfd, NULL, NULL, 0);
+	sqe->user_data = 1;
+}
+
+static void add_recvzc(struct io_uring *ring, int sockfd)
+{
+	struct io_uring_sqe *sqe = io_uring_get_sqe(ring);
+
+	io_uring_prep_rw(IORING_OP_RECV_ZC, sqe, sockfd, NULL, 0, 0);
+	sqe->ioprio |= IORING_RECV_MULTISHOT;
+	sqe->zcrx_ifq_idx = zcrx_id;
+	sqe->user_data = 2;
+}
+
+static void add_recvzc_oneshot(struct io_uring *ring, int sockfd, size_t len)
+{
+	struct io_uring_sqe *sqe = io_uring_get_sqe(ring);
+
+	io_uring_prep_rw(IORING_OP_RECV_ZC, sqe, sockfd, NULL, len, 0);
+	sqe->ioprio |= IORING_RECV_MULTISHOT;
+	sqe->zcrx_ifq_idx = zcrx_id;
+	sqe->user_data = 2;
+}
+
+static void process_accept(struct io_uring *ring, struct io_uring_cqe *cqe)
+{
+	if (cqe->res < 0)
+		t_error(1, 0, "accept()");
+	if (connfd)
+		t_error(1, 0, "Unexpected second connection");
+
+	connfd = cqe->res;
+	if (cfg_oneshot)
+		add_recvzc_oneshot(ring, connfd, PAGE_SIZE);
+	else
+		add_recvzc(ring, connfd);
+}
+
+static void verify_data(char *data, size_t size, unsigned long seq)
+{
+	int i;
+
+	if (!cfg_verify_data)
+		return;
+
+	for (i = 0; i < size; i++) {
+		char expected = 'a' + (seq + i) % 26;
+
+		if (data[i] != expected)
+			t_error(1, 0, "payload mismatch at %i", i);
+	}
+}
+
+static void process_recvzc(struct io_uring *ring, struct io_uring_cqe *cqe)
+{
+	unsigned rq_mask = rq_ring.ring_entries - 1;
+	struct io_uring_zcrx_cqe *rcqe;
+	struct io_uring_zcrx_rqe *rqe;
+	uint64_t mask;
+	char *data;
+
+	if (cqe->res < 0)
+		t_error(1, 0, "recvzc(): %d", cqe->res);
+
+	if (cqe->res == 0 && cqe->flags == 0 && cfg_oneshot_recvs == 0) {
+		stop = true;
+		return;
+	}
+
+	if (cfg_oneshot) {
+		if (cqe->res == 0 && cqe->flags == 0 && cfg_oneshot_recvs) {
+			add_recvzc_oneshot(ring, connfd, PAGE_SIZE);
+			cfg_oneshot_recvs--;
+		}
+	} else if (!(cqe->flags & IORING_CQE_F_MORE)) {
+		add_recvzc(ring, connfd);
+	}
+
+	rcqe = (struct io_uring_zcrx_cqe *)(cqe + 1);
+	mask = (1ULL << IORING_ZCRX_AREA_SHIFT) - 1;
+	data = (char *)area_ptr + (rcqe->off & mask);
+
+	verify_data(data, cqe->res, received);
+	received += cqe->res;
+
+	/* processed, return back to the kernel */
+	rqe = &rq_ring.rqes[rq_ring.rq_tail & rq_mask];
+	rqe->off = (rcqe->off & ~IORING_ZCRX_AREA_MASK) | area_token;
+	rqe->len = cqe->res;
+	io_uring_smp_store_release(rq_ring.ktail, ++rq_ring.rq_tail);
+}
+
+static void server_loop(struct io_uring *ring)
+{
+	struct io_uring_cqe *cqe;
+	unsigned int head, count = 0;
+
+	io_uring_submit_and_wait(ring, 1);
+
+	io_uring_for_each_cqe(ring, head, cqe) {
+		if (cqe->user_data == 1)
+			process_accept(ring, cqe);
+		else if (cqe->user_data == 2)
+			process_recvzc(ring, cqe);
+		else
+			t_error(1, 0, "unknown cqe");
+		count++;
+	}
+	io_uring_cq_advance(ring, count);
+}
+
+static void run_server(void)
+{
+	unsigned int flags = 0;
+	struct io_uring ring;
+	int fd, enable, ret;
+
+	fd = socket(AF_INET6, SOCK_STREAM, 0);
+	if (fd == -1)
+		t_error(1, 0, "socket()");
+
+	enable = 1;
+	ret = setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, &enable, sizeof(int));
+	if (ret < 0)
+		t_error(1, 0, "setsockopt(SO_REUSEADDR)");
+
+	ret = bind(fd, (struct sockaddr *)&cfg_addr, sizeof(cfg_addr));
+	if (ret < 0)
+		t_error(1, 0, "bind()");
+
+	if (listen(fd, 1024) < 0)
+		t_error(1, 0, "listen()");
+
+	flags |= IORING_SETUP_COOP_TASKRUN;
+	flags |= IORING_SETUP_SINGLE_ISSUER;
+	flags |= IORING_SETUP_DEFER_TASKRUN;
+	flags |= IORING_SETUP_SUBMIT_ALL;
+	flags |= IORING_SETUP_CQE32;
+
+	ret = io_uring_queue_init(512, &ring, flags);
+	if (ret)
+		t_error(1, ret, "ring init failed");
+
+	setup_zcrx(&ring);
+	add_accept(&ring, fd);
+
+	while (!stop)
+		server_loop(&ring);
+}
+
+static void usage(const char *filepath)
+{
+	t_error(1, 0, "Usage: %s (-4|-6) -p<port> -i<ifname> -q<rxq_id>", filepath);
+}
+
+static void parse_opts(int argc, char **argv)
+{
+	struct sockaddr_in6 *addr6 = (void *) &cfg_addr;
+	int c;
+
+	if (argc <= 1)
+		usage(argv[0]);
+
+	while ((c = getopt(argc, argv, "vp:i:q:o:")) != -1) {
+		switch (c) {
+		case 'p':
+			cfg_port = strtoul(optarg, NULL, 0);
+			break;
+		case 'i':
+			cfg_ifname = optarg;
+			break;
+		case 'o': {
+			cfg_oneshot = true;
+			cfg_oneshot_recvs = strtoul(optarg, NULL, 0);
+			break;
+		}
+		case 'q':
+			cfg_queue_id = strtoul(optarg, NULL, 0);
+			break;
+		case 'v':
+			cfg_verify_data = true;
+			break;
+		}
+	}
+
+	memset(addr6, 0, sizeof(*addr6));
+	addr6->sin6_family = AF_INET6;
+	addr6->sin6_port = htons(cfg_port);
+	addr6->sin6_addr = in6addr_any;
+}
+
+int main(int argc, char **argv)
+{
+	parse_opts(argc, argv);
+	run_server();
+	return 0;
+}
-- 
2.48.1

Re: [PATCH liburing 0/5] zcrx example and other changes

[Jens Axboe]

On Wed, 16 Apr 2025 10:01:12 +0100, Pavel Begunkov wrote:
> We need a simple example for zcrx to show case how the api works
> and how to use features. Patch 5 is a brushed up version of the
> zcrx selftest.
> 
> Apart from that update headers and use the kernel return zcrx id.
> 
> Pavel Begunkov (5):
>   Update io_uring.h for zcrx
>   tests/zcrx: rename a test
>   tests/zcrx: use returned right zcrx id
>   examples: add extra helpers
>   examples: add a zcrx example
> 
> [...]

Applied, thanks!

[1/5] Update io_uring.h for zcrx
      (no commit info)
[2/5] tests/zcrx: rename a test
      (no commit info)
[3/5] tests/zcrx: use returned right zcrx id
      (no commit info)
[4/5] examples: add extra helpers
      (no commit info)
[5/5] examples: add a zcrx example
      (no commit info)

Best regards,
-- 
Jens Axboe