diff --git a/lib_acl/src/private/private_vstream.c b/lib_acl/src/private/private_vstream.c
index 9d201dd82..f56f0a57f 100644
--- a/lib_acl/src/private/private_vstream.c
+++ b/lib_acl/src/private/private_vstream.c
@@ -416,7 +416,7 @@ TAG_AGAIN:
 	return (n);
 }
 
-static int __loop_writen(ACL_VSTREAM *stream, const void *vptr, size_t dlen)
+static int __loop_writen(ACL_VSTREAM *stream, const void *vptr, ssize_t dlen)
 {
 	const unsigned char *ptr;
 	int   n;
@@ -433,6 +433,8 @@ static int __loop_writen(ACL_VSTREAM *stream, const void *vptr, size_t dlen)
 			return (ACL_VSTREAM_EOF);
 		}
 
+		assert(n <= dlen);
+
 		dlen  -= n;
 		ptr   += n;
 	}
@@ -448,7 +450,7 @@ int private_vstream_writen(ACL_VSTREAM *stream, const void *vptr, size_t dlen)
 		if (private_vstream_fflush(stream) == ACL_VSTREAM_EOF)
 			return (ACL_VSTREAM_EOF);
 	}
-	return (__loop_writen(stream, vptr, dlen));
+	return (__loop_writen(stream, vptr, (ssize_t) dlen));
 }
 
 int private_vstream_write(ACL_VSTREAM *stream, const void *vptr, size_t dlen)
diff --git a/lib_fiber/c/src/event.c b/lib_fiber/c/src/event.c
index 842c77a7f..eccc149a7 100644
--- a/lib_fiber/c/src/event.c
+++ b/lib_fiber/c/src/event.c
@@ -166,6 +166,12 @@ int event_checkfd(EVENT *ev, FILE_EVENT *fe)
 			acl_fiber_set_error(0);
 			return 0;
 		}
+#if defined(HAS_IO_URING)
+	} else if (EVENT_IS_IO_URING(ev)) {
+		fe->type = TYPE_FILE | TYPE_EVENTABLE;
+		acl_fiber_set_error(0);
+		return 1;
+#endif
 	} else {
 		fe->type = TYPE_FILE;
 		acl_fiber_set_error(0);
@@ -250,20 +256,24 @@ int event_add_read(EVENT *ev, FILE_EVENT *fe, event_proc *proc)
 		return 0;
 	}
 
+	fe->r_proc = proc;
+
 	if (fe->oper & EVENT_DEL_READ) {
 		fe->oper &= ~EVENT_DEL_READ;
 	}
 
 	if (!(fe->mask & EVENT_READ)) {
+		if (EVENT_IS_IO_URING(ev)) {
+			ev->add_read(ev, fe);
+		}
 		// we should check the fd's type for the first time.
-		if (fe->me.parent == &fe->me) {
+		else if (fe->me.parent == &fe->me) {
 			ring_prepend(&ev->events, &fe->me);
 		}
 
 		fe->oper |= EVENT_ADD_READ;
 	}
 
-	fe->r_proc = proc;
 	return 1;
 }
 
@@ -293,19 +303,22 @@ int event_add_write(EVENT *ev, FILE_EVENT *fe, event_proc *proc)
 		return 0;
 	}
 
+	fe->w_proc = proc;
+
 	if (fe->oper & EVENT_DEL_WRITE) {
 		fe->oper &= ~EVENT_DEL_WRITE;
 	}
 
 	if (!(fe->mask & EVENT_WRITE)) {
-		if (fe->me.parent == &fe->me) {
+		if (EVENT_IS_IO_URING(ev)) {
+			ev->add_write(ev, fe);
+		} else if (fe->me.parent == &fe->me) {
 			ring_prepend(&ev->events, &fe->me);
 		}
 
 		fe->oper |= EVENT_ADD_WRITE;
 	}
 
-	fe->w_proc = proc;
 	return 1;
 }
 
diff --git a/lib_fiber/c/src/event.h b/lib_fiber/c/src/event.h
index fc54b80f7..e9106e33e 100644
--- a/lib_fiber/c/src/event.h
+++ b/lib_fiber/c/src/event.h
@@ -60,6 +60,15 @@ typedef void epoll_proc(EVENT *ev, EPOLL_EVENT *ee);
 typedef struct IOCP_EVENT IOCP_EVENT;
 #endif
 
+#ifdef HAS_IO_URING
+typedef struct IO_URING_CTX {
+	FILE_EVENT *fe;
+	int res;
+	int cnt;
+	unsigned mask;
+} IO_URING_CTX;
+#endif
+
 /**
  * for each connection fd
  */
@@ -238,8 +247,6 @@ struct FILE_EVENT {
 		} sendmsg_ctx;
 	} out;
 
-	int res;
-
 	union {
 		struct {
 			struct sockaddr_in addr;
@@ -250,6 +257,8 @@ struct FILE_EVENT {
 		char  *path;
 	} var;
 
+	struct IO_URING_CTX reader_ctx;
+	struct IO_URING_CTX writer_ctx;
 	struct __kernel_timespec rts;
 	struct __kernel_timespec wts;
 	int           r_timeout;
diff --git a/lib_fiber/c/src/event/event_io_uring.c b/lib_fiber/c/src/event/event_io_uring.c
index 24e3e4a2a..8018e450b 100644
--- a/lib_fiber/c/src/event/event_io_uring.c
+++ b/lib_fiber/c/src/event/event_io_uring.c
@@ -13,6 +13,7 @@ typedef struct EVENT_URING {
 	struct io_uring ring;
 	size_t sqe_size;
 	size_t appending;
+	size_t loop_count;
 } EVENT_URING;
 
 static void event_uring_free(EVENT *ev)
@@ -41,7 +42,8 @@ static void add_read_wait(EVENT_URING *ep, FILE_EVENT *fe, int tmo_ms)
 
 	sqe = io_uring_get_sqe(&ep->ring);
 	io_uring_prep_poll_add(sqe, fe->fd, POLLIN | POLLHUP | POLLERR);
-	io_uring_sqe_set_data(sqe, fe);
+	fe->reader_ctx.fe = fe;
+	io_uring_sqe_set_data(sqe, &fe->reader_ctx);
 	sqe->flags = IOSQE_IO_LINK;
 
 	TRY_SUBMMIT(ep);
@@ -62,6 +64,7 @@ static int event_uring_add_read(EVENT_URING *ep, FILE_EVENT *fe)
 	}
 
 	fe->mask |= EVENT_READ;
+	fe->reader_ctx.mask = EVENT_READ;
 
 	if (LIKELY(!(fe->mask & (EVENT_POLLIN | EVENT_ACCEPT)))) {
 		struct io_uring_sqe *sqe = io_uring_get_sqe(&ep->ring);
@@ -95,7 +98,9 @@ static int event_uring_add_read(EVENT_URING *ep, FILE_EVENT *fe)
 				fe->in.read_ctx.len,
 				fe->in.read_ctx.off);
 		}
-		io_uring_sqe_set_data(sqe, fe);
+
+		fe->reader_ctx.fe = fe;
+		io_uring_sqe_set_data(sqe, &fe->reader_ctx);
 
 		TRY_SUBMMIT(ep);
 	} else if (fe->mask & EVENT_POLLIN) {
@@ -106,7 +111,9 @@ static int event_uring_add_read(EVENT_URING *ep, FILE_EVENT *fe)
 		io_uring_prep_accept(sqe, fe->fd,
 			(struct sockaddr*) &fe->var.peer.addr,
 			(socklen_t*) &fe->var.peer.len, 0);
-		io_uring_sqe_set_data(sqe, fe);
+
+		fe->reader_ctx.fe = fe;
+		io_uring_sqe_set_data(sqe, &fe->reader_ctx);
 
 		TRY_SUBMMIT(ep);
 	}
@@ -120,7 +127,8 @@ static void add_write_wait(EVENT_URING *ep, FILE_EVENT *fe, int tmo_ms)
 
 	sqe = io_uring_get_sqe(&ep->ring);
 	io_uring_prep_poll_add(sqe, fe->fd, POLLOUT | POLLHUP | POLLERR);
-	io_uring_sqe_set_data(sqe, fe);
+	fe->writer_ctx.fe = fe;
+	io_uring_sqe_set_data(sqe, &fe->writer_ctx);
 	sqe->flags = IOSQE_IO_LINK;
 
 	TRY_SUBMMIT(ep);
@@ -141,6 +149,8 @@ static int event_uring_add_write(EVENT_URING *ep, FILE_EVENT *fe)
 	}
 
 	fe->mask |= EVENT_WRITE;
+	fe->writer_ctx.mask = EVENT_WRITE;
+	fe->writer_ctx.cnt++;
 
 	if (LIKELY(!(fe->mask & (EVENT_POLLOUT | EVENT_CONNECT)))) {
 		struct io_uring_sqe *sqe = io_uring_get_sqe(&ep->ring);
@@ -174,7 +184,9 @@ static int event_uring_add_write(EVENT_URING *ep, FILE_EVENT *fe)
 				fe->out.write_ctx.len,
 				fe->out.write_ctx.off);
 		}
-		io_uring_sqe_set_data(sqe, fe);
+
+		fe->writer_ctx.fe = fe;
+		io_uring_sqe_set_data(sqe, &fe->writer_ctx);
 
 		TRY_SUBMMIT(ep);
 	} else if (fe->mask & EVENT_POLLOUT) {
@@ -185,7 +197,9 @@ static int event_uring_add_write(EVENT_URING *ep, FILE_EVENT *fe)
 		io_uring_prep_connect(sqe, fe->fd,
 			(struct sockaddr*) &fe->var.peer.addr,
 			(socklen_t) fe->var.peer.len);
-		io_uring_sqe_set_data(sqe, fe);
+
+		fe->writer_ctx.fe = fe;
+		io_uring_sqe_set_data(sqe, &fe->writer_ctx);
 
 		TRY_SUBMMIT(ep);
 	}
@@ -199,7 +213,8 @@ void event_uring_file_close(EVENT *ev, FILE_EVENT *fe)
 	struct io_uring_sqe *sqe = io_uring_get_sqe(&ep->ring);
 
 	io_uring_prep_close(sqe, fe->fd);
-	io_uring_sqe_set_data(sqe, fe);
+	fe->reader_ctx.fe = fe;
+	io_uring_sqe_set_data(sqe, &fe->reader_ctx);
 	TRY_SUBMMIT(ep);
 }
 
@@ -210,7 +225,8 @@ void event_uring_file_openat(EVENT *ev, FILE_EVENT *fe, int dirfd,
 	struct io_uring_sqe *sqe = io_uring_get_sqe(&ep->ring);
 
 	io_uring_prep_openat(sqe, dirfd, pathname, flags, mode);
-	io_uring_sqe_set_data(sqe, fe);
+	fe->reader_ctx.fe = fe;
+	io_uring_sqe_set_data(sqe, &fe->reader_ctx);
 	TRY_SUBMMIT(ep);
 }
 
@@ -220,7 +236,8 @@ void event_uring_file_unlink(EVENT *ev, FILE_EVENT *fe, const char *pathname)
 	struct io_uring_sqe *sqe = io_uring_get_sqe(&ep->ring);
 
 	io_uring_prep_unlink(sqe, pathname, 0);
-	io_uring_sqe_set_data(sqe, fe);
+	fe->reader_ctx.fe = fe;
+	io_uring_sqe_set_data(sqe, &fe->reader_ctx);
 	TRY_SUBMMIT(ep);
 }
 
@@ -232,7 +249,8 @@ void event_uring_file_statx(EVENT *ev, FILE_EVENT *fe, int dirfd,
 	struct io_uring_sqe *sqe = io_uring_get_sqe(&ep->ring);
 
 	io_uring_prep_statx(sqe, dirfd, pathname, flags, mask, statxbuf);
-	io_uring_sqe_set_data(sqe, fe);
+	fe->reader_ctx.fe = fe;
+	io_uring_sqe_set_data(sqe, &fe->reader_ctx);
 	TRY_SUBMMIT(ep);
 }
 
@@ -243,7 +261,8 @@ void event_uring_file_renameat2(EVENT *ev, FILE_EVENT *fe, int olddirfd,
 	struct io_uring_sqe *sqe = io_uring_get_sqe(&ep->ring);
 
 	io_uring_prep_renameat(sqe, olddirfd, oldpath, newdirfd, newpath, flags);
-	io_uring_sqe_set_data(sqe, fe);
+	fe->reader_ctx.fe = fe;
+	io_uring_sqe_set_data(sqe, &fe->reader_ctx);
 	TRY_SUBMMIT(ep);
 }
 
@@ -254,7 +273,8 @@ void event_uring_mkdirat(EVENT *ev, FILE_EVENT *fe, int dirfd,
 	struct io_uring_sqe *sqe = io_uring_get_sqe(&ep->ring);
 
 	io_uring_prep_mkdirat(sqe, dirfd, pathname, mode);
-	io_uring_sqe_set_data(sqe, fe);
+	fe->reader_ctx.fe = fe;
+	io_uring_sqe_set_data(sqe, &fe->reader_ctx);
 	TRY_SUBMMIT(ep);
 }
 
@@ -267,7 +287,8 @@ void event_uring_splice(EVENT *ev, FILE_EVENT *fe, int fd_in, loff_t off_in,
 
 	io_uring_prep_splice(sqe, fd_in, off_in, fd_out, off_out,
 		len, splice_flags);
-	io_uring_sqe_set_data(sqe, fe);
+	fe->reader_ctx.fe = fe;
+	io_uring_sqe_set_data(sqe, &fe->reader_ctx);
 	sqe->flags |= sqe_flags;
 	sqe->opcode = opcode;
 	TRY_SUBMMIT(ep);
@@ -284,7 +305,8 @@ void event_uring_sendfile(EVENT *ev, FILE_EVENT *fe, int out, int in,
 	unsigned flags = SPLICE_F_MOVE | SPLICE_F_MORE; // | SPLICE_F_NONBLOCK;
 
 	io_uring_prep_splice(sqe, in, off, fe->var.pipefd[1], -1, cnt, flags);
-	io_uring_sqe_set_data(sqe, fe);
+	fe->reader_ctx.fe = fe;
+	io_uring_sqe_set_data(sqe, &fe->reader_ctx);
 	sqe->flags |= IOSQE_IO_LINK | SPLICE_F_FD_IN_FIXED | IOSQE_ASYNC;
 	sqe->opcode = IORING_OP_SPLICE;
 
@@ -293,7 +315,8 @@ void event_uring_sendfile(EVENT *ev, FILE_EVENT *fe, int out, int in,
 	flags = 0;
 	sqe = io_uring_get_sqe(&ep->ring);
 	io_uring_prep_splice(sqe, fe->var.pipefd[0], -1, out, -1, cnt, flags);
-	io_uring_sqe_set_data(sqe, fe);
+	fe->reader_ctx.fe = fe;
+	io_uring_sqe_set_data(sqe, &fe->reader_ctx);
 	sqe->opcode = IORING_OP_SPLICE;
 
 	TRY_SUBMMIT(ep);
@@ -324,7 +347,7 @@ static int event_uring_del_write(EVENT_URING *ep UNUSED, FILE_EVENT *fe)
 
 static void handle_read(EVENT *ev, FILE_EVENT *fe, int res)
 {
-	fe->res = res;
+	fe->reader_ctx.res = res;
 
 	if (LIKELY(!(fe->mask & (EVENT_ACCEPT | EVENT_POLLIN)))) {
 		if ((fe->type & TYPE_FILE) && res > 0) {
@@ -335,7 +358,7 @@ static void handle_read(EVENT *ev, FILE_EVENT *fe, int res)
 			}
 		}
 	} else if (fe->mask & EVENT_ACCEPT) {
-		// fe->res = res;
+		// Do nothing!
 	} else if (fe->mask & EVENT_POLLIN) {
 		if (res & (POLLIN | ERR)) {
 			if (res & POLLERR) {
@@ -363,7 +386,7 @@ static void handle_read(EVENT *ev, FILE_EVENT *fe, int res)
 
 static void handle_write(EVENT *ev, FILE_EVENT *fe, int res)
 {
-	fe->res = res;
+	fe->writer_ctx.res = res;
 
 	if (LIKELY(!(fe->mask & (EVENT_CONNECT | EVENT_POLLOUT)))) {
 		if ((fe->type & TYPE_FILE) && res > 0) {
@@ -374,7 +397,7 @@ static void handle_write(EVENT *ev, FILE_EVENT *fe, int res)
 			}
 		}
 	} else if (fe->mask & EVENT_CONNECT) {
-		//fe->res = res;
+		// Do nothing!
 	} else if (fe->mask & EVENT_POLLOUT) {
 		if (res & (POLLOUT | ERR)) {
 			if (res & POLLERR) {
@@ -402,12 +425,21 @@ static void handle_write(EVENT *ev, FILE_EVENT *fe, int res)
 	}
 }
 
-static void handle_one(EVENT *ev, FILE_EVENT *fe, int res)
+static void handle_one(EVENT *ev, IO_URING_CTX *ctx, int res)
 {
-	if ((fe->mask & EVENT_READ) && fe->r_proc) {
+	FILE_EVENT *fe = ctx->fe;
+
+	if (ctx == &fe->reader_ctx && ctx->mask == EVENT_READ && fe->r_proc) {
+		fe->mask &= ~EVENT_READ;
 		handle_read(ev, fe, res);
+		ctx->mask = 0;
 		return;
-	} else if ((fe->mask & EVENT_WRITE) && fe->w_proc) {
+	}
+
+	if (ctx == &fe->writer_ctx && ctx->mask == EVENT_WRITE && fe->w_proc) {
+		fe->writer_ctx.cnt--;
+		fe->mask &= ~EVENT_WRITE;
+		ctx->mask = 0;
 		handle_write(ev, fe, res);
 		return;
 	}
@@ -424,7 +456,7 @@ static void handle_one(EVENT *ev, FILE_EVENT *fe, int res)
 		| EVENT_DIR_MKDIRAT \
 		| EVENT_SPLICE)
 
-	fe->res = res;
+	fe->reader_ctx.res = res;
 
 	if (fe->mask & FLAGS) {
 		fe->r_proc(ev, fe);
@@ -436,31 +468,27 @@ static void handle_one(EVENT *ev, FILE_EVENT *fe, int res)
 
 static int peek_more(EVENT_URING *ep)
 {
-#define	PEEK_FOREACH
-//#define	PEEK_BATCH
-
-#if	defined(PEEK_FOREACH)
-
 	struct io_uring_cqe *cqe;
 	unsigned head, count = 0;
-	FILE_EVENT *fe;
+	IO_URING_CTX *ctx;
 	int ret;
 
 	io_uring_for_each_cqe(&ep->ring, head, cqe) {
 		count++;
-		fe = (FILE_EVENT*) io_uring_cqe_get_data(cqe);
+		ctx = (IO_URING_CTX*) io_uring_cqe_get_data(cqe);
 		ret = cqe->res;
+		//io_uring_cqe_seen(&ep->ring, cqe);
 
 		if (ret == -ENOBUFS) {
 			msg_error("%s(%d): ENOBUFS error", __FUNCTION__, __LINE__);
 			return -1;
 		}
 
-		if (ret == -ETIME || ret == -ECANCELED || fe == NULL) {
+		if (ret == -ETIME || ret == -ECANCELED || ctx == NULL) {
 			continue;
 		}
 
-		handle_one((EVENT*) ep, fe, ret);
+		handle_one((EVENT*) ep, ctx, ret);
 	}
 
 	if (count > 0) {
@@ -468,75 +496,13 @@ static int peek_more(EVENT_URING *ep)
 	}
 
 	return count;
-
-#elif	defined(PEEK_BATCH)
-
-#define	PEEK_MAX	100
-
-	struct io_uring_cqe *cqes[PEEK_MAX + 1];
-	FILE_EVENT *fe;
-	unsigned n, i;
-	int ret, cnt = 0;
-
-	while ((n = io_uring_peek_batch_cqe(&ep->ring, cqes, PEEK_MAX)) > 0) {
-		for (i = 0; i < n; i++) {
-			ret = cqes[i]->res;
-			fe = (FILE_EVENT*) io_uring_cqe_get_data(cqes[i]);
-
-			if (ret == -ENOBUFS) {
-				msg_error("%s(%d): ENOBUFS error",
-					__FUNCTION__, __LINE__);
-				return -1;
-			}
-
-			if (ret == -ETIME || ret == -ECANCELED || fe == NULL) {
-				continue;
-			}
-
-			handle_one((EVENT*) ep, fe, ret);
-		}
-
-		io_uring_cq_advance(&ep->ring, n);
-		cnt += n;
-	}
-
-	return cnt;
-
-#else
-
-	int cnt = 0, ret;
-	struct io_uring_cqe *cqe;
-	FILE_EVENT *fe;
-
-	while (1) {
-		ret = io_uring_peek_cqe(&ep->ring, &cqe);
-		if (ret == -EAGAIN) {
-			break;
-		}
-
-		ret = cqe->res;
-		fe = (FILE_EVENT*) io_uring_cqe_get_data(cqe);
-		io_uring_cqe_seen(&ep->ring, cqe);
-
-		if (ret == -ETIME || ret == -ECANCELED || fe == NULL) {
-			continue;
-		} else if (ret < 0) {
-			return -1;
-		}
-
-		handle_one((EVENT*) ep, fe, ret);
-		cnt++;
-	}
-	return cnt;
-
-#endif
 }
 
 static int submit_and_wait(EVENT_URING *ep, int timeout)
 {
 	struct __kernel_timespec ts, *tp;
 	struct io_uring_cqe *cqe;
-	FILE_EVENT *fe;
+	IO_URING_CTX *ctx;
 	int ret;
 
 	if (timeout >= 0) {
@@ -551,7 +517,8 @@ static int submit_and_wait(EVENT_URING *ep, int timeout)
 
 	if (ep->appending > 0) {
 		ep->appending = 0;  \
-		ret = io_uring_submit_and_wait_timeout(&ep->ring, &cqe, 1, tp, NULL);
+		ret = io_uring_submit_and_wait_timeout(&ep->ring, &cqe,
+				1, tp, NULL);
 	} else {
 		ret = io_uring_wait_cqes(&ep->ring, &cqe, 1, tp, NULL);
 	}
@@ -569,7 +536,7 @@ static int submit_and_wait(EVENT_URING *ep, int timeout)
 	}
 
 	ret = cqe->res;
-	fe = (FILE_EVENT*) io_uring_cqe_get_data(cqe);
+	ctx = (IO_URING_CTX*) io_uring_cqe_get_data(cqe);
 
 	io_uring_cqe_seen(&ep->ring, cqe);
 
@@ -578,11 +545,11 @@ static int submit_and_wait(EVENT_URING *ep, int timeout)
 		return -1;
 	}
 
-	if (ret == -ETIME || ret == -ECANCELED || fe == NULL) {
+	if (ret == -ETIME || ret == -ECANCELED || ctx == NULL) {
 		return 1;
 	}
 
-	handle_one((EVENT*) ep, fe, ret);
+	handle_one((EVENT*) ep, ctx, ret);
 	return 1;
 }
 
@@ -591,6 +558,7 @@ static int event_uring_wait(EVENT *ev, int timeout)
 	EVENT_URING *ep = (EVENT_URING*) ev;
 	int ret, count = 0;
 
+	ep->loop_count++;
 	ret = submit_and_wait(ep, timeout);
 	if (ret == 0) {
 		return 0;
diff --git a/lib_fiber/c/src/fbase_event.c b/lib_fiber/c/src/fbase_event.c
index dcfa71523..2ffa1658e 100644
--- a/lib_fiber/c/src/fbase_event.c
+++ b/lib_fiber/c/src/fbase_event.c
@@ -21,6 +21,8 @@
 #include "common/iostuff.h"
 #include "fiber.h"
 
+//#undef	HAS_EVENTFD
+
 void fbase_event_open(FIBER_BASE *fbase)
 {
 #if defined(HAS_EVENTFD)
@@ -55,8 +57,10 @@ void fbase_event_open(FIBER_BASE *fbase)
 			__FILE__, __LINE__, __FUNCTION__, (int) fbase->event_in);
 	}
 
-	non_blocking(fbase->event_in, 1);
-	non_blocking(fbase->event_out, 1);
+	//if (!EVENT_IS_IO_URING(fiber_io_event())) {
+		non_blocking(fbase->event_in, 1);
+		non_blocking(fbase->event_out, 1);
+	//}
 }
 
 void fbase_event_close(FIBER_BASE *fbase)
@@ -76,20 +80,27 @@ int fbase_event_wait(FIBER_BASE *fbase)
 {
 	long long n;
 	int  ret, interrupt = 0, err;
+	FILE_EVENT *fe;
 
 	if (fbase->event_in == INVALID_SOCKET) {
 		msg_fatal("%s(%d), %s: invalid event_in=%d",
 			__FILE__, __LINE__, __FUNCTION__, (int) fbase->event_in);
 	}
 
+	fe = fiber_file_open_read(fbase->event_in);
+	fe->type = TYPE_SPIPE | TYPE_EVENTABLE;
+
 	while (1) {
 		//if (acl_fiber_scheduled() && read_wait(fbase->event_in, -1) == -1) {
+#if 1
 		if (read_wait(fbase->event_in, -1) == -1) {
 			msg_error("%s(%d), %s: read_wait error=%s, fd=%d",
 				__FILE__, __LINE__, __FUNCTION__,
 				last_serror(), fbase->event_in);
 			return -1;
 		}
+#endif
+
 #ifdef SYS_WIN
 		ret = (int) acl_fiber_recv(fbase->event_in, (char*) &n, sizeof(n), 0);
 #else
@@ -141,6 +152,7 @@ int fbase_event_wakeup(FIBER_BASE *fbase)
 {
 	long long n = 1;
 	int  ret, interrupt = 0;
+	FILE_EVENT *fe;
 
 	/**
 	 * if (LIKELY(atomic_int64_cas(fbase->atomic, 0, 1) != 0)) {
@@ -154,6 +166,9 @@ int fbase_event_wakeup(FIBER_BASE *fbase)
 			fbase, (int) fbase->event_out);
 	}
 
+	fe = fiber_file_open_write(fbase->event_out);
+	fe->type = TYPE_SPIPE | TYPE_EVENTABLE;
+
 	while (1) {
 #ifdef SYS_WIN
 		ret = (int) acl_fiber_send(fbase->event_out, (char*) &n, sizeof(n), 0);
@@ -166,9 +181,11 @@ int fbase_event_wakeup(FIBER_BASE *fbase)
 		}
 
 		if (ret >= 0) {
-			msg_fatal("%s(%d), %s: write ret=%d invalid length, "
-				"interrupt=%d", __FILE__, __LINE__,
-				__FUNCTION__, ret, interrupt);
+			msg_fatal("%s(%d), %s: fiber=%d, write ret=%d invalid"
+				" length, interrupt=%d, fd=%d",
+				__FILE__, __LINE__, __FUNCTION__,
+				acl_fiber_self(), ret, interrupt,
+				(int) fbase->event_out);
 		}
 
 		if (acl_fiber_last_error() == EINTR) {
diff --git a/lib_fiber/c/src/file_event.c b/lib_fiber/c/src/file_event.c
index dde66767f..79088d410 100644
--- a/lib_fiber/c/src/file_event.c
+++ b/lib_fiber/c/src/file_event.c
@@ -26,6 +26,8 @@ void file_event_init(FILE_EVENT *fe, socket_t fd)
 	memset(&fe->in, 0, sizeof(fe->in));
 	memset(&fe->out, 0, sizeof(fe->out));
 	memset(&fe->var, 0, sizeof(fe->var));
+	memset(&fe->reader_ctx, 0, sizeof(fe->reader_ctx));
+	memset(&fe->writer_ctx, 0, sizeof(fe->writer_ctx));
 	fe->r_timeout = -1;
 	fe->w_timeout = -1;
 #endif
@@ -56,6 +58,7 @@ FILE_EVENT *file_event_alloc(socket_t fd)
 
 static void file_event_free(FILE_EVENT *fe)
 {
+	memset(fe, 0, sizeof(*fe));
 	mem_free(fe);
 }
 
diff --git a/lib_fiber/c/src/hook/file.c b/lib_fiber/c/src/hook/file.c
index a474b1ed4..4ba8887c0 100644
--- a/lib_fiber/c/src/hook/file.c
+++ b/lib_fiber/c/src/hook/file.c
@@ -66,10 +66,10 @@ int file_close(EVENT *ev, FILE_EVENT *fe)
 
 	fe->mask &= ~EVENT_FILE_CLOSE;
 
-	if (fe->res == 0) {
+	if (fe->reader_ctx.res == 0) {
 		return 0;
 	} else {
-		acl_fiber_set_error(-fe->res);
+		acl_fiber_set_error(-fe->reader_ctx.res);
 		return -1;
 	}
 }
@@ -109,14 +109,14 @@ int openat(int dirfd, const char *pathname, int flags, ...)
 	free(fe->var.path);
 	fe->var.path = NULL;
 
-	if (fe->res >= 0) {
-		fe->fd   = fe->res;
+	if (fe->reader_ctx.res >= 0) {
+		fe->fd   = fe->reader_ctx.res;
 		fe->type = TYPE_FILE | TYPE_EVENTABLE;
 		fiber_file_set(fe);  // Save the fe for the future using.
 		return fe->fd;
 	}
 
-	acl_fiber_set_error(-fe->res);
+	acl_fiber_set_error(-fe->reader_ctx.res);
 	file_event_unrefer(fe);
 	return -1;
 }
@@ -162,11 +162,11 @@ int unlink(const char *pathname)
 	free(fe->var.path);
 	fe->var.path = NULL;
 
-	if (fe->res == 0) {
+	if (fe->reader_ctx.res == 0) {
 		file_event_unrefer(fe);
 		return 0;
 	} else {
-		acl_fiber_set_error(-fe->res);
+		acl_fiber_set_error(-fe->reader_ctx.res);
 		file_event_unrefer(fe);
 		return -1;
 	}
@@ -204,11 +204,11 @@ int renameat2(int olddirfd, const char *oldpath,
 	free(fe->in.read_ctx.buf);
 	free(fe->var.path);
 
-	if (fe->res == 0) {
+	if (fe->reader_ctx.res == 0) {
 		file_event_unrefer(fe);
 		return 0;
 	} else {
-		acl_fiber_set_error(-fe->res);
+		acl_fiber_set_error(-fe->reader_ctx.res);
 		file_event_unrefer(fe);
 		return -1;
 	}
@@ -257,13 +257,13 @@ int statx(int dirfd, const char *pathname, int flags, unsigned int mask,
 	free(fe->in.read_ctx.buf);
 	fe->in.read_ctx.buf = NULL;
 
-	if (fe->res == 0) {
+	if (fe->reader_ctx.res == 0) {
 		memcpy(statxbuf, fe->var.statxbuf, sizeof(struct statx));
 		free(fe->var.statxbuf);
 		file_event_unrefer(fe);
 		return 0;
 	} else {
-		acl_fiber_set_error(-fe->res);
+		acl_fiber_set_error(-fe->reader_ctx.res);
 		free(fe->var.statxbuf);
 		file_event_unrefer(fe);
 		return -1;
@@ -324,11 +324,11 @@ int mkdirat(int dirfd, const char *pathname, mode_t mode)
 	fe->mask &= ~EVENT_DIR_MKDIRAT;
 	free(fe->var.path);
 
-	if (fe->res == 0) {
+	if (fe->reader_ctx.res == 0) {
 		file_event_unrefer(fe);
 		return 0;
 	} else {
-		acl_fiber_set_error(-fe->res);
+		acl_fiber_set_error(-fe->reader_ctx.res);
 		file_event_unrefer(fe);
 		return -1;
 	}
@@ -443,21 +443,21 @@ ssize_t splice(int fd_in, loff_t *poff_in, int fd_out,
 
 	fe->mask &= ~EVENT_SPLICE;
 
-	if (fe->res < 0) {
-		acl_fiber_set_error(-fe->res);
+	if (fe->reader_ctx.res < 0) {
+		acl_fiber_set_error(-fe->reader_ctx.res);
 		file_event_unrefer(fe);
 		return -1;
 	}
 	
 	if (off_in != -1 && poff_in) {
-		*poff_in += fe->res;
+		*poff_in += fe->reader_ctx.res;
 	}
 
 	if (off_out != -1 && poff_out) {
-		*poff_out += fe->res;
+		*poff_out += fe->reader_ctx.res;
 	}
 
-	ret = fe->res;
+	ret = fe->reader_ctx.res;
 	file_event_unrefer(fe);
 	return ret;
 }
diff --git a/lib_fiber/c/src/hook/io.c b/lib_fiber/c/src/hook/io.c
index 0c21e3028..dfc13b684 100644
--- a/lib_fiber/c/src/hook/io.c
+++ b/lib_fiber/c/src/hook/io.c
@@ -127,7 +127,7 @@ static int iocp_wait_read(FILE_EVENT *fe)
 		// Must clear the EVENT_READ flags in order to set IO event
 		// for each IO process.
 		fe->mask &= ~EVENT_READ;
-		fe->res   = 0;
+		fe->reader_ctx.res = 0;
 
 		if (fiber_wait_read(fe) < 0) {
 			return -1;
@@ -144,8 +144,8 @@ static int iocp_wait_read(FILE_EVENT *fe)
 			return -1;
 		}
 
-		if (fe->res >= 0) {
-			return fe->res;
+		if (fe->reader_ctx.res >= 0) {
+			return fe->reader_ctx.res;
 		}
 
 		err = acl_fiber_last_error();
@@ -168,14 +168,14 @@ int fiber_iocp_read(FILE_EVENT *fe, char *buf, int len)
 	 * which maybe used in iocp_add_read() and set for polling read status.
 	 */
 	if (fe->sock_type == SOCK_DGRAM && fe->rbuf == fe->packet
-		&& fe->res > 0) {
+		&& fe->reader_ctx.res > 0) {
 
-		if (fe->res < len) {
-			len = fe->res;
+		if (fe->reader_ctx.res < len) {
+			len = fe->reader_ctx.res;
 		}
 		memcpy(buf, fe->packet, len);
 		fe->rbuf = NULL;
-		fe->res  = 0;
+		fe->reader_ctx.res = 0;
 		return len;
 	}
 
@@ -530,14 +530,14 @@ static int iocp_wait_write(FILE_EVENT *fe)
 		int err;
 
 		fe->mask &= ~EVENT_WRITE;
-		fe->res = 0;
+		fe->writer_ctx.res = -1;
 
 		if (wait_write(fe) == -1) {
 			return -1;
 		}
 
-		if (fe->res >= 0) {
-			return fe->res;
+		if (fe->writer_ctx.res >= 0) {
+			return fe->writer_ctx.res;
 		}
 
 		err = acl_fiber_last_error();
diff --git a/lib_fiber/c/src/hook/socket.c b/lib_fiber/c/src/hook/socket.c
index 5bc7c0bf7..6d0467bc6 100644
--- a/lib_fiber/c/src/hook/socket.c
+++ b/lib_fiber/c/src/hook/socket.c
@@ -87,14 +87,14 @@ static socket_t fiber_iocp_accept(FILE_EVENT *fe)
 {
 	fe->mask &= ~EVENT_READ;
 	fe->mask |= EVENT_ACCEPT;
-	fe->res   = INVALID_SOCKET;
+	fe->reader_ctx.res = INVALID_SOCKET;
 
 	if (fiber_wait_read(fe) < 0) {
 		msg_error("%s(%d): fiber_wait_read error=%s, fd=%d",
 			__FUNCTION__, __LINE__, last_serror(), (int) fe->fd);
 		return INVALID_SOCKET;
 	}
-	return fe->res;
+	return fe->reader_ctx.res;
 }
 #endif
 
@@ -292,8 +292,8 @@ static socket_t fiber_iocp_connect(FILE_EVENT *fe)
 	}
 
 	fe->mask &= ~EVENT_CONNECT;
-	if (fe->res < 0) {
-		acl_fiber_set_error(-fe->res);
+	if (fe->writer_ctx.res < 0) {
+		acl_fiber_set_error(-fe->writer_ctx.res);
 		return -1;
 	}
 	return 0;