diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index a336282..2a65991 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -30,6 +30,7 @@ stages:
   - dnf --enablerepo=powertools install -y clang
   - dnf --enablerepo=powertools install -y llvm
   - dnf --enablerepo=powertools install -y libbpf-devel
+  - dnf --enablerepo=powertools install -y liburing-devel
   - chmod +x ./ci/travis.sh
   script:
   - yum makecache
diff --git a/CMakeLists.txt b/CMakeLists.txt
index 82277fa..fe75298 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -28,6 +28,14 @@ elseif(ASAN_OPTION MATCHES "THREAD")
 endif()
 # end of for ASAN
 
+find_package(LIBURING)
+if (SUPPORT_LIBURING)
+	add_definitions(-DSUPPORT_LIBURING=1)
+	message(STATUS "Support liburing")
+else()
+	message(STATUS "Not support liburing")
+endif()
+
 find_package(LIBBPF)
 if (SUPPORT_BPF)
 	add_definitions(-DSUPPORT_BPF=1)
diff --git a/cmake/FindLIBURING.cmake b/cmake/FindLIBURING.cmake
new file mode 100644
index 0000000..c1bfdec
--- /dev/null
+++ b/cmake/FindLIBURING.cmake
@@ -0,0 +1,8 @@
+include(FindPkgConfig)
+pkg_check_modules(LIBURING liburing)
+
+if(LIBURING_FOUND)
+        set(SUPPORT_LIBURING true)
+else()
+        set(SUPPORT_LIBURING false)
+endif()
\ No newline at end of file
diff --git a/conf/kni/kni.conf b/conf/kni/kni.conf
index 577bedf..78b926f 100644
--- a/conf/kni/kni.conf
+++ b/conf/kni/kni.conf
@@ -22,6 +22,23 @@ bpf_debug_log=0
 # 4: BPF 使用四元组分流
 bpf_hash_mode=2
 
+[io_uring]
+enable_iouring=1
+enable_debuglog=0
+ring_size=1024
+buff_size=2048
+# io_uring_setup() flags
+# IORING_SETUP_IOPOLL	(1U << 0)	/* io_context is polled */
+# IORING_SETUP_SQPOLL	(1U << 1)	/* SQ poll thread */
+# IORING_SETUP_SQ_AFF	(1U << 2)	/* sq_thread_cpu is valid */
+# IORING_SETUP_CQSIZE	(1U << 3)	/* app defines CQ size */
+# IORING_SETUP_CLAMP	(1U << 4)	/* clamp SQ/CQ ring sizes */
+# IORING_SETUP_ATTACH_WQ	(1U << 5)	/* attach to existing wq */
+# IORING_SETUP_R_DISABLED	(1U << 6)	/* start with ring disabled */
+# IORING_SETUP_SUBMIT_ALL	(1U << 7)	/* continue submit on error */
+flags=0
+sq_thread_idle=0
+
 [tfe0]
 enabled = 1
 dev_eth_symbol = ens1f5
diff --git a/entry/CMakeLists.txt b/entry/CMakeLists.txt
index f2883da..76e98d7 100644
--- a/entry/CMakeLists.txt
+++ b/entry/CMakeLists.txt
@@ -1,11 +1,15 @@
 set(CMAKE_INSTALL_PREFIX /home/mesasoft/sapp_run)
-add_library(kni SHARED src/kni_entry.cpp src/tfe_mgr.cpp src/kni_tap_rss.cpp src/kni_pxy_tcp_option.cpp src/kni_dynamic_bypass.cpp)
+add_library(kni SHARED src/kni_entry.cpp src/tfe_mgr.cpp src/kni_tap_rss.cpp src/kni_iouring.cpp src/kni_pxy_tcp_option.cpp src/kni_dynamic_bypass.cpp)
 target_include_directories(kni PUBLIC ${CMAKE_CURRENT_LIST_DIR}/include)
 
+target_link_libraries(kni common MESA_prof_load MESA_htable MESA_field_stat maatframe mrzcpd dabloom)
+
 if (SUPPORT_BPF)
-    target_link_libraries(kni common MESA_prof_load MESA_htable MESA_field_stat maatframe mrzcpd dabloom bpf)
-else()
-    target_link_libraries(kni common MESA_prof_load MESA_htable MESA_field_stat maatframe mrzcpd dabloom)
+    target_link_libraries(kni bpf)
+endif()
+
+if (SUPPORT_LIBURING)
+    target_link_libraries(kni uring)
 endif()
 
 install(TARGETS kni LIBRARY DESTINATION ${CMAKE_INSTALL_PREFIX}/plug/business/kni COMPONENT LIBRARIES)
diff --git a/entry/include/kni_entry.h b/entry/include/kni_entry.h
index b5f9670..11be0c6 100644
--- a/entry/include/kni_entry.h
+++ b/entry/include/kni_entry.h
@@ -7,6 +7,7 @@
 #include <tsg/tsg_statistic.h>
 #include "tfe_mgr.h"
 #include <tsg/tsg_label.h>
+#include "kni_iouring.h"
 
 #define BURST_MAX   1
 #define CALLER_SAPP 0
@@ -196,6 +197,9 @@ struct per_thread_handle{
 	MESA_htable_handle tuple2stream_htable;
 	MESA_htable_handle traceid2sslinfo_htable;
 	struct expiry_dablooms_handle *dabloom_handle;
+#if (SUPPORT_LIBURING)
+	struct io_uring_handle *iouring_handle;
+#endif
 };
 
 struct tuple2stream_htable_value{
@@ -211,6 +215,7 @@ struct security_policy_shunt_tsg_diagnose{
 };
 
 struct kni_handle{
+	struct io_uring_conf iouring_conf;
 	struct kni_marsio_handle *marsio_handle;
 	struct bpf_ctx *tap_bpf_ctx;
 	struct kni_maat_handle *maat_handle;
diff --git a/entry/include/kni_iouring.h b/entry/include/kni_iouring.h
new file mode 100644
index 0000000..35c5ec0
--- /dev/null
+++ b/entry/include/kni_iouring.h
@@ -0,0 +1,69 @@
+#ifndef _KNI_IOURING_H_
+#define _KNI_IOURING_H_
+
+#include <stdio.h>
+#include <string.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <arpa/inet.h>
+
+#if (SUPPORT_LIBURING)
+#include <liburing.h>
+#endif
+
+#ifdef __cplusplus
+extern "C"
+{
+#endif
+
+#define MAX_BATCH_CQE_NUM 128
+
+    struct io_uring_conf
+    {
+        int enable_iouring;
+        int enable_debuglog;
+
+        int ring_size;
+        int buff_size;
+
+        int flags;
+        int sq_thread_idle; // milliseconds
+    };
+
+#if (SUPPORT_LIBURING)
+    enum evtype
+    {
+        EVTYPE_UNKNOWN = 0,
+        EVTYPE_READ = 1,
+        EVTYPE_WRITE = 2,
+    };
+
+    struct user_data
+    {
+        int sockfd;
+        enum evtype type;
+        struct iovec vec;
+    };
+
+    struct io_uring_handle
+    {
+        struct io_uring ring;
+        struct io_uring_params params;
+
+        int ring_size;
+        int buff_size;
+    };
+
+    struct user_data *io_uring_user_data_create(int sockfd, enum evtype type, int buff_size);
+    void io_uring_user_data_destory(struct user_data *conn);
+
+    struct io_uring_handle *io_uring_handle_create(int ring_size, int buff_size, int flags, int sq_thread_idle);
+    void io_uring_handle_destory(struct io_uring_handle *handle);
+
+#endif
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
\ No newline at end of file
diff --git a/entry/include/kni_tap_rss.h b/entry/include/kni_tap_rss.h
index 1421562..b82ef38 100644
--- a/entry/include/kni_tap_rss.h
+++ b/entry/include/kni_tap_rss.h
@@ -6,7 +6,7 @@ extern "C"
 {
 #endif
 
-#define TAP_RSS_LOG_TAG "TAP_RSS :"
+#define TAP_RSS_LOG_TAG "TAP_RSS: "
 
 struct bpf_ctx;
 
diff --git a/entry/src/kni_entry.cpp b/entry/src/kni_entry.cpp
index b0d581b..db81b59 100644
--- a/entry/src/kni_entry.cpp
+++ b/entry/src/kni_entry.cpp
@@ -64,6 +64,9 @@ struct kni_handle *g_kni_handle = NULL;
 struct kni_field_stat_handle *g_kni_fs_handle = NULL;
 int *arr_last_tfe_dispatch_index = NULL;
 
+static int io_uring_add_write(int thread_seq, char *raw_data, uint16_t raw_len, addr_type_t addr_type);
+static int io_uring_add_read(int thread_seq);
+static int io_uring_polling(int thread_seq);
 
 static char* stream_errmsg_session_record(enum intercept_error _errno){
 	switch(_errno){
@@ -808,14 +811,23 @@ static int send_to_tfe_normal_mode(char *raw_data, uint16_t raw_len, int thread_
 
 static int send_to_tfe_tap_mode(char *raw_data, uint16_t raw_len, int thread_seq, addr_type_t addr_type, void *logger)
 {
-	int tap_fd = g_kni_handle->threads_handle[thread_seq].tap_fd;
-	char *send_buff = g_kni_handle->threads_handle[thread_seq].buff;
-
-	add_ether_header(send_buff, raw_data, raw_len, addr_type);
-	int ret = kni_tap_write_per_thread(tap_fd, send_buff, raw_len + sizeof(struct ethhdr), logger);
-	if (ret < 0)
+	if (g_kni_handle->iouring_conf.enable_iouring)
 	{
-		return -1;
+		if (io_uring_add_write(thread_seq, raw_data, raw_len, addr_type) == -1)
+		{
+			return -1;
+		}
+	}
+	else
+	{
+		int tap_fd = g_kni_handle->threads_handle[thread_seq].tap_fd;
+		char *send_buff = g_kni_handle->threads_handle[thread_seq].buff;
+		add_ether_header(send_buff, raw_data, raw_len, addr_type);
+		int ret = kni_tap_write_per_thread(tap_fd, send_buff, raw_len + sizeof(struct ethhdr), logger);
+		if (ret < 0)
+		{
+			return -1;
+		}
 	}
 
 	return 0;
@@ -1978,21 +1990,35 @@ extern "C" char kni_polling_all_entry(const struct streaminfo *stream, void** pm
 	else if (g_kni_handle->deploy_mode == KNI_DEPLOY_MODE_TAP)
 	{
 		int tap_fd = g_kni_handle->threads_handle[thread_seq].tap_fd;
-		char *recv_buff = g_kni_handle->threads_handle[thread_seq].buff;
-		int recv_buff_size = g_kni_handle->threads_handle[thread_seq].buff_size;
-
-		// 只支持一个 TFE, 每次最多收 100 个包
-		for (int j = 0; j < 100; j++)
+		if (g_kni_handle->iouring_conf.enable_iouring)
 		{
-			if (kni_tap_read_per_thread(tap_fd, recv_buff, recv_buff_size, logger) > 0)
+			if (io_uring_polling(thread_seq) == -1)
 			{
-				struct ethhdr *ether_hdr = (struct ethhdr *)recv_buff;
-				tuple2stream_htable_search(tuple2stream_htable, ether_hdr, thread_seq);
-				flag = POLLING_STATE_WORK;
+				flag = POLLING_STATE_IDLE;
 			}
 			else
 			{
-				break;
+				flag = POLLING_STATE_WORK;
+			}
+		}
+		else
+		{
+			char *recv_buff = g_kni_handle->threads_handle[thread_seq].buff;
+			int recv_buff_size = g_kni_handle->threads_handle[thread_seq].buff_size;
+
+			// 只支持一个 TFE, 每次最多收 100 个包
+			for (int j = 0; j < 100; j++)
+			{
+				if (kni_tap_read_per_thread(tap_fd, recv_buff, recv_buff_size, logger) > 0)
+				{
+					struct ethhdr *ether_hdr = (struct ethhdr *)recv_buff;
+					tuple2stream_htable_search(tuple2stream_htable, ether_hdr, thread_seq);
+					flag = POLLING_STATE_WORK;
+				}
+				else
+				{
+					break;
+				}
 			}
 		}
 	}
@@ -2669,6 +2695,13 @@ extern "C" int kni_init(){
 			goto error_out;
 		}
 
+		MESA_load_profile_int_nodef(profile, "io_uring", "enable_iouring", (int *)&g_kni_handle->iouring_conf.enable_iouring);
+		MESA_load_profile_int_nodef(profile, "io_uring", "enable_debuglog", (int *)&g_kni_handle->iouring_conf.enable_debuglog);
+		MESA_load_profile_int_nodef(profile, "io_uring", "ring_size", (int *)&g_kni_handle->iouring_conf.ring_size);
+		MESA_load_profile_int_nodef(profile, "io_uring", "buff_size", (int *)&g_kni_handle->iouring_conf.buff_size);
+		MESA_load_profile_int_nodef(profile, "io_uring", "flags", (int *)&g_kni_handle->iouring_conf.flags);
+		MESA_load_profile_int_nodef(profile, "io_uring", "sq_thread_idle", (int *)&g_kni_handle->iouring_conf.sq_thread_idle);
+
 		int tap_allow_mutilthread = 0;
 		MESA_load_profile_int_nodef(profile, "tap", "tap_allow_mutilthread", &tap_allow_mutilthread);
 		if (tap_allow_mutilthread)
@@ -2677,6 +2710,7 @@ extern "C" int kni_init(){
 			uint32_t bpf_hash_mode = 2;
 			uint32_t bpf_queue_num = get_thread_count();
 			char bpf_obj[1024] = {0};
+
 			MESA_load_profile_int_nodef(profile, "tap", "bpf_debug_log", (int *)&bpf_debug_log);
 			MESA_load_profile_int_nodef(profile, "tap", "bpf_hash_mode", (int *)&bpf_hash_mode);
 			ret = MESA_load_profile_string_nodef(profile, "tap", "bpf_obj", bpf_obj, sizeof(bpf_obj));
@@ -2740,8 +2774,29 @@ extern "C" int kni_init(){
 			{
 				goto error_out;
 			}
-			g_kni_handle->threads_handle[i].buff_size = KNI_MTU;
-			g_kni_handle->threads_handle[i].buff = ALLOC(char, g_kni_handle->threads_handle[i].buff_size);
+			if (g_kni_handle->iouring_conf.enable_iouring)
+			{
+#if (SUPPORT_LIBURING)
+				g_kni_handle->threads_handle[i].iouring_handle = io_uring_handle_create(
+					g_kni_handle->iouring_conf.ring_size, g_kni_handle->iouring_conf.buff_size,
+					g_kni_handle->iouring_conf.flags, g_kni_handle->iouring_conf.sq_thread_idle);
+				if (g_kni_handle->threads_handle[i].iouring_handle == NULL)
+				{
+					KNI_LOG_ERROR(local_logger, TAP_RSS_LOG_TAG "Failed at kni create iouring");
+					goto error_out;
+				}
+
+				io_uring_add_read(i);
+#else
+				KNI_LOG_ERROR(local_logger, TAP_RSS_LOG_TAG "Unsupport io_uring !!!");
+				exit(-1);
+#endif
+			}
+			else
+			{
+				g_kni_handle->threads_handle[i].buff_size = KNI_MTU;
+				g_kni_handle->threads_handle[i].buff = ALLOC(char, g_kni_handle->threads_handle[i].buff_size);
+			}
 		}
 	}
 	//init ssl dynamic bypass htable
@@ -2819,3 +2874,159 @@ error_out:
 	kni_destroy();
 	exit(0);
 }
+
+#if (SUPPORT_LIBURING)
+
+static int io_uring_add_write(int thread_seq, char *raw_data, uint16_t raw_len, addr_type_t addr_type)
+{
+	int tap_fd = g_kni_handle->threads_handle[thread_seq].tap_fd;
+	struct io_uring_handle *handle = g_kni_handle->threads_handle[thread_seq].iouring_handle;
+	struct io_uring *ring = &handle->ring;
+
+	struct io_uring_sqe *sqe = io_uring_get_sqe(ring);
+	if (sqe == NULL)
+	{
+		KNI_LOG_ERROR(g_kni_handle->local_logger, TAP_RSS_LOG_TAG "kni the io_uring submission queue is full, drop write packet");
+		return -1;
+	}
+
+	struct user_data *conn = io_uring_user_data_create(tap_fd, EVTYPE_WRITE, handle->buff_size);
+	conn->vec.iov_len = sizeof(struct ethhdr) + raw_len;
+	add_ether_header(conn->vec.iov_base, raw_data, raw_len, addr_type);
+
+	io_uring_prep_writev(sqe, conn->sockfd, &conn->vec, 1, 0);
+	io_uring_sqe_set_data(sqe, conn);
+
+	if (g_kni_handle->iouring_conf.enable_debuglog)
+	{
+		KNI_LOG_DEBUG(g_kni_handle->local_logger, TAP_RSS_LOG_TAG "Add event: write, thread: %02d, sockfd: %02d, sqe: %p, user_data: %p, iov_base: %p, iovec_len: %04d",
+					  thread_seq, conn->sockfd, sqe, sqe->user_data, conn->vec.iov_base, conn->vec.iov_len);
+	}
+	io_uring_submit(ring);
+
+	return 0;
+}
+
+static int io_uring_add_read(int thread_seq)
+{
+	int tap_fd = g_kni_handle->threads_handle[thread_seq].tap_fd;
+	struct io_uring_handle *handle = g_kni_handle->threads_handle[thread_seq].iouring_handle;
+	struct io_uring *ring = &handle->ring;
+
+	struct io_uring_sqe *sqe = io_uring_get_sqe(ring);
+	if (sqe == NULL)
+	{
+		KNI_LOG_ERROR(g_kni_handle->local_logger, TAP_RSS_LOG_TAG "kni the io_uring submission queue is full, drop read packet");
+		return -1;
+	}
+
+	struct user_data *conn = io_uring_user_data_create(tap_fd, EVTYPE_READ, handle->buff_size);
+	io_uring_prep_readv(sqe, conn->sockfd, &conn->vec, 1, 0);
+	io_uring_sqe_set_data(sqe, conn);
+
+	if (g_kni_handle->iouring_conf.enable_debuglog)
+	{
+		KNI_LOG_DEBUG(g_kni_handle->local_logger, TAP_RSS_LOG_TAG "Add event: read, thread: %02d, sockfd: %02d, sqe: %p, user_data: %p, iov_base: %p, iovec_len: %04d",
+					  thread_seq, conn->sockfd, sqe, sqe->user_data, conn->vec.iov_base, conn->vec.iov_len);
+	}
+	io_uring_submit(ring);
+
+	return 0;
+}
+
+static int io_uring_polling(int thread_seq)
+{
+	int tap_fd = g_kni_handle->threads_handle[thread_seq].tap_fd;
+	MESA_htable_handle tuple2stream_htable = g_kni_handle->threads_handle[thread_seq].tuple2stream_htable;
+	struct io_uring_handle *handle = g_kni_handle->threads_handle[thread_seq].iouring_handle;
+	struct io_uring *ring = &handle->ring;
+
+	struct io_uring_cqe *cqes[MAX_BATCH_CQE_NUM];
+	int ret = io_uring_peek_batch_cqe(ring, cqes, MAX_BATCH_CQE_NUM);
+	if (ret <= 0)
+	{
+		return -1;
+	}
+
+	for (int i = 0; i < ret; i++)
+	{
+		struct io_uring_cqe *cqe = cqes[i];
+		if (cqe == NULL)
+		{
+			KNI_LOG_ERROR(g_kni_handle->local_logger, TAP_RSS_LOG_TAG "kni iouring cqe return NULL");
+			continue;
+		}
+		if (cqe->user_data == NULL)
+		{
+			KNI_LOG_ERROR(g_kni_handle->local_logger, TAP_RSS_LOG_TAG "kni iouring cqe->user_data return NULL");
+			continue;
+		}
+
+		struct user_data *conn = (struct user_data *)cqe->user_data;
+		if (g_kni_handle->iouring_conf.enable_debuglog)
+		{
+			KNI_LOG_DEBUG(g_kni_handle->local_logger, TAP_RSS_LOG_TAG "Handle event: %s, thread: %02d, sockfd: %02d, cqe: %p, user_data: %p, iov_base: %p, iovec_len: %04d, cqe->res: %04d",
+						  (conn->type == EVTYPE_READ ? "read" : (conn->type == EVTYPE_WRITE ? "write" : "unknown")),
+						  thread_seq, conn->sockfd, cqe, cqe->user_data, conn->vec.iov_base, conn->vec.iov_len, cqe->res);
+		}
+		switch (conn->type)
+		{
+		case EVTYPE_READ:
+			if (cqe->res > 0)
+			{
+				struct ethhdr *ether_hdr = (struct ethhdr *)conn->vec.iov_base;
+				tuple2stream_htable_search(tuple2stream_htable, ether_hdr, thread_seq);
+			}
+			else
+			{
+				KNI_LOG_ERROR(g_kni_handle->local_logger, TAP_RSS_LOG_TAG "kni iouring tapfd[%d] read error, (%d, %s)", tap_fd, -cqe->res, strerror(-cqe->res));
+			}
+			io_uring_user_data_destory(conn);
+			io_uring_add_read(thread_seq);
+			break;
+		case EVTYPE_WRITE:
+			if (cqe->res < 0)
+			{
+				KNI_LOG_ERROR(g_kni_handle->local_logger, TAP_RSS_LOG_TAG "kni iouring tapfd[%d] write error, (%d, %s)", tap_fd, -cqe->res, strerror(-cqe->res));
+			}
+			io_uring_user_data_destory(conn);
+			break;
+		case EVTYPE_UNKNOWN:
+			assert(0);
+			break;
+		default:
+			assert(0);
+			break;
+		}
+		io_uring_cqe_seen(&handle->ring, cqe);
+	}
+	return 0;
+}
+
+#else
+
+static int io_uring_add_write(int thread_seq, char *raw_data, uint16_t raw_len, addr_type_t addr_type)
+{
+	KNI_LOG_ERROR(g_kni_handle->local_logger, TAP_RSS_LOG_TAG "Unsupport io_uring !!!");
+	exit(-1);
+
+	return -1;
+}
+
+static int io_uring_add_read(int thread_seq)
+{
+	KNI_LOG_ERROR(g_kni_handle->local_logger, TAP_RSS_LOG_TAG "Unsupport io_uring !!!");
+	exit(-1);
+
+	return -1;
+}
+
+static int io_uring_polling(int thread_seq)
+{
+	KNI_LOG_ERROR(g_kni_handle->local_logger, TAP_RSS_LOG_TAG "Unsupport io_uring !!!");
+	exit(-1);
+
+	return -1;
+}
+
+#endif
\ No newline at end of file
diff --git a/entry/src/kni_iouring.cpp b/entry/src/kni_iouring.cpp
new file mode 100644
index 0000000..d9845cc
--- /dev/null
+++ b/entry/src/kni_iouring.cpp
@@ -0,0 +1,79 @@
+#include "kni_iouring.h"
+
+#if (SUPPORT_LIBURING)
+
+struct user_data *io_uring_user_data_create(int sockfd, enum evtype type, int buff_size)
+{
+    struct user_data *conn = (struct user_data *)calloc(1, sizeof(struct user_data));
+    conn->sockfd = sockfd;
+    conn->type = type;
+    conn->vec.iov_base = (void *)calloc(buff_size, sizeof(char));
+    conn->vec.iov_len = buff_size;
+
+    return conn;
+}
+
+void io_uring_user_data_destory(struct user_data *conn)
+{
+    if (conn)
+    {
+        if (conn->vec.iov_base)
+        {
+            free(conn->vec.iov_base);
+            conn->vec.iov_base = NULL;
+        }
+
+        free(conn);
+        conn = NULL;
+    }
+}
+
+void io_uring_handle_destory(struct io_uring_handle *handle)
+{
+    if (handle)
+    {
+        io_uring_queue_exit(&handle->ring);
+
+        free(handle);
+        handle = NULL;
+    }
+}
+
+struct io_uring_handle *io_uring_handle_create(int ring_size, int buff_size, int flags, int sq_thread_idle)
+{
+    struct io_uring_handle *handle = (struct io_uring_handle *)calloc(1, sizeof(struct io_uring_handle));
+    if (handle == NULL)
+    {
+        return NULL;
+    }
+
+    handle->ring_size = ring_size;
+    handle->buff_size = buff_size;
+
+    /*
+     * 参考资料：https://unixism.net/loti/tutorial/sq_poll.html#sq-poll
+     *
+     * 执行下面命令进行验证 IORING_SETUP_SQPOLL:
+     * sudo bpftrace -e 'tracepoint:io_uring:io_uring_submit_sqe {printf("%s(%d)\n", comm, pid);}'
+     */
+    if (flags)
+    {
+        handle->params.flags |= flags;
+    }
+    if (sq_thread_idle)
+    {
+        handle->params.sq_thread_idle = sq_thread_idle; // milliseconds
+    }
+
+    int ret = io_uring_queue_init_params(ring_size, &handle->ring, &handle->params);
+    if (ret)
+    {
+        fprintf(stderr, "Failed at io_uring_queue_init_params(), %s\n", strerror(-ret));
+        io_uring_handle_destory(handle);
+        return NULL;
+    }
+
+    return handle;
+}
+
+#endif
\ No newline at end of file
diff --git a/entry/src/kni_tap_rss.cpp b/entry/src/kni_tap_rss.cpp
index 667049c..3fa0c0c 100644
--- a/entry/src/kni_tap_rss.cpp
+++ b/entry/src/kni_tap_rss.cpp
@@ -95,7 +95,7 @@ struct bpf_ctx *kni_tap_global_load_rss_bpf(const char *bpf_obj_file, uint32_t b
 
     if (bpf_prog_load(ctx->bpf_file, BPF_PROG_TYPE_SOCKET_FILTER, &ctx->bpf_obj, &ctx->bpf_prog_fd) < 0)
     {
-        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to load bpf object %s, aborting: %s\n", ctx->bpf_file, strerror(errno));
+        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to load bpf object %s, aborting: %s", ctx->bpf_file, strerror(errno));
         goto error;
     }
 
@@ -114,7 +114,7 @@ error:
 #else
 struct bpf_ctx *kni_tap_global_load_rss_bpf(const char *bpf_obj_file, uint32_t bpf_queue_num, uint32_t bpf_hash_mode, uint32_t bpf_debug_log, void *logger)
 {
-    KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "not support bpf\n");
+    KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "not support bpf");
     return NULL;
 }
 #endif
@@ -129,7 +129,7 @@ int kni_tap_open_per_thread(const char *tap_dev, int tap_flags, int bpf_prog_fd,
     tap_fd = open(TUN_PATH, O_RDWR);
     if (tap_fd == -1)
     {
-        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to open " TUN_PATH ", aborting: %s\n", strerror(errno));
+        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to open " TUN_PATH ", aborting: %s", strerror(errno));
         return -1;
     }
 
@@ -138,7 +138,7 @@ int kni_tap_open_per_thread(const char *tap_dev, int tap_flags, int bpf_prog_fd,
     strcpy(ifr.ifr_name, tap_dev);
     if (ioctl(tap_fd, TUNSETIFF, &ifr) == -1)
     {
-        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to attach %s, aborting: %s\n", tap_dev, strerror(errno));
+        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to attach %s, aborting: %s", tap_dev, strerror(errno));
         goto error;
     }
 
@@ -148,7 +148,7 @@ int kni_tap_open_per_thread(const char *tap_dev, int tap_flags, int bpf_prog_fd,
      */
     if (ioctl(tap_fd, TUNSETPERSIST, 1) == -1)
     {
-        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to set persist on %s, aborting: %s\n", tap_dev, strerror(errno));
+        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to set persist on %s, aborting: %s", tap_dev, strerror(errno));
         goto error;
     }
 
@@ -158,7 +158,7 @@ int kni_tap_open_per_thread(const char *tap_dev, int tap_flags, int bpf_prog_fd,
         // Set bpf
         if (ioctl(tap_fd, TUNSETSTEERINGEBPF, (void *)&bpf_prog_fd) == -1)
         {
-            KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to set bpf on %s, aborting: %s\n", tap_dev, strerror(errno));
+            KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to set bpf on %s, aborting: %s", tap_dev, strerror(errno));
             goto error;
         }
     }
@@ -168,13 +168,13 @@ int kni_tap_open_per_thread(const char *tap_dev, int tap_flags, int bpf_prog_fd,
     nonblock_flags = fcntl(tap_fd, F_GETFL);
     if (nonblock_flags == -1)
     {
-        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to get nonblock flags on %s fd, aborting: %s\n", tap_dev, strerror(errno));
+        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to get nonblock flags on %s fd, aborting: %s", tap_dev, strerror(errno));
         goto error;
     }
     nonblock_flags |= O_NONBLOCK;
     if (fcntl(tap_fd, F_SETFL, nonblock_flags) == -1)
     {
-        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to set nonblock flags on %s fd, aborting: %s\n", tap_dev, strerror(errno));
+        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to set nonblock flags on %s fd, aborting: %s", tap_dev, strerror(errno));
         goto error;
     }
 
@@ -182,7 +182,7 @@ int kni_tap_open_per_thread(const char *tap_dev, int tap_flags, int bpf_prog_fd,
     fd = socket(PF_INET, SOCK_DGRAM, 0);
     if (fd == -1)
     {
-        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to create socket, aborting: %s\n", strerror(errno));
+        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to create socket, aborting: %s", strerror(errno));
         goto error;
     }
 
@@ -190,14 +190,14 @@ int kni_tap_open_per_thread(const char *tap_dev, int tap_flags, int bpf_prog_fd,
     strcpy(ifr.ifr_name, tap_dev);
     if (ioctl(fd, SIOCGIFMTU, &ifr) < 0)
     {
-        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to get MTU on %s, aborting: %s\n", tap_dev, strerror(errno));
+        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to get MTU on %s, aborting: %s", tap_dev, strerror(errno));
         goto error;
     }
 
     // Set eth up
     if (ioctl(fd, SIOCGIFFLAGS, &ifr) == -1)
     {
-        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to get link status on %s, aborting: %s\n", tap_dev, strerror(errno));
+        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to get link status on %s, aborting: %s", tap_dev, strerror(errno));
         goto error;
     }
 
@@ -206,12 +206,12 @@ int kni_tap_open_per_thread(const char *tap_dev, int tap_flags, int bpf_prog_fd,
         ifr.ifr_flags |= IFF_UP;
         if (ioctl(fd, SIOCSIFFLAGS, &ifr) < 0)
         {
-            KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to set link status on %s, aborting: %s\n", tap_dev, strerror(errno));
+            KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to set link status on %s, aborting: %s", tap_dev, strerror(errno));
             goto error;
         }
     }
 
-    KNI_LOG_INFO(logger, TAP_RSS_LOG_TAG "using tap device %s with MTU %d\n", tap_dev, ifr.ifr_mtu);
+    KNI_LOG_INFO(logger, TAP_RSS_LOG_TAG "using tap device %s with MTU %d", tap_dev, ifr.ifr_mtu);
     close(fd);
 
     return tap_fd;
@@ -248,7 +248,7 @@ int kni_tap_read_per_thread(int tap_fd, char *buff, int buff_size, void *logger)
     {
         if (errno != EWOULDBLOCK && errno != EAGAIN)
         {
-            KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to read data from tapfd %d, aborting: %s\n", tap_fd, strerror(errno));
+            KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "unable to read data from tapfd %d, aborting: %s", tap_fd, strerror(errno));
         }
     }
 
@@ -260,7 +260,7 @@ int kni_tap_write_per_thread(int tap_fd, const char *data, int data_len, void *l
     int ret = write(tap_fd, data, data_len);
     if (ret != data_len)
     {
-        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "need send %dB, only send %dB, aborting: %s\n", data_len, ret, strerror(errno));
+        KNI_LOG_ERROR(logger, TAP_RSS_LOG_TAG "need send %dB, only send %dB, aborting: %s", data_len, ret, strerror(errno));
     }
 
     return ret;