Support for 'netsplice' engine
This IO engine does network transfers, but uses splice() to move data
to/from a socket to a pipe and vmsplice() to move that data to/from
user space.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
diff --git a/engines/net.c b/engines/net.c
index 1a403bd..4936cc2 100644
--- a/engines/net.c
+++ b/engines/net.c
@@ -19,6 +19,8 @@
struct netio_data {
int listenfd;
int send_to_net;
+ int use_splice;
+ int pipes[2];
char host[64];
struct sockaddr_in addr;
};
@@ -50,23 +52,178 @@
return 1;
}
+/*
+ * Receive bytes from a socket and fill them into the internal pipe
+ */
+static int splice_in(struct thread_data *td, struct io_u *io_u)
+{
+ struct netio_data *nd = td->io_ops->data;
+ unsigned int len = io_u->xfer_buflen;
+ struct fio_file *f = io_u->file;
+ int bytes = 0;
+
+ while (len) {
+ int ret = splice(nd->pipes[1], NULL, f->fd, NULL, len, 0);
+
+ if (ret < 0) {
+ if (!bytes)
+ bytes = ret;
+
+ break;
+ } else if (!ret)
+ break;
+
+ bytes += ret;
+ }
+
+ return bytes;
+}
+
+/*
+ * Transmit 'len' bytes from the internal pipe
+ */
+static int splice_out(struct thread_data *td, struct io_u *io_u,
+ unsigned int len)
+{
+ struct netio_data *nd = td->io_ops->data;
+ struct fio_file *f = io_u->file;
+ int bytes = 0;
+
+ while (len) {
+ int ret = splice(nd->pipes[0], NULL, f->fd, NULL, len, 0);
+
+ if (ret < 0) {
+ if (!bytes)
+ bytes = ret;
+
+ break;
+ } else if (!ret)
+ break;
+
+ bytes += ret;
+ len -= ret;
+ }
+
+ return bytes;
+}
+
+/*
+ * vmsplice() pipe to io_u buffer
+ */
+static int vmsplice_io_u_out(struct thread_data *td, struct io_u *io_u,
+ unsigned int len)
+{
+ struct netio_data *nd = td->io_ops->data;
+ struct iovec iov = {
+ .iov_base = io_u->xfer_buf,
+ .iov_len = len,
+ };
+ int bytes = 0;
+
+ while (iov.iov_len) {
+ int ret = vmsplice(nd->pipes[0], &iov, 1, 0);
+
+ if (ret < 0) {
+ if (!bytes)
+ bytes = ret;
+ break;
+ } else if (!ret)
+ break;
+
+ iov.iov_len -= ret;
+ if (iov.iov_len)
+ iov.iov_base += ret;
+ }
+
+ return bytes;
+}
+
+/*
+ * vmsplice() io_u to pipe
+ */
+static int vmsplice_io_u_in(struct thread_data *td, struct io_u *io_u)
+{
+ struct netio_data *nd = td->io_ops->data;
+ struct iovec iov = {
+ .iov_base = io_u->xfer_buf,
+ .iov_len = io_u->xfer_buflen,
+ };
+ unsigned int bytes = 0;
+
+ while (iov.iov_len) {
+ int ret = vmsplice(nd->pipes[1], &iov, 1, 0);
+
+ if (ret < 0)
+ return -1;
+ else if (!ret)
+ return bytes;
+
+ iov.iov_len -= ret;
+ bytes += ret;
+ if (iov.iov_len)
+ iov.iov_base += ret;
+ }
+
+ return bytes;
+}
+
+static int fio_netio_splice_in(struct thread_data *td, struct io_u *io_u)
+{
+ int ret;
+
+ ret = splice_in(td, io_u);
+ if (ret <= 0)
+ return ret;
+
+ return vmsplice_io_u_out(td, io_u, ret);
+}
+
+static int fio_netio_splice_out(struct thread_data *td, struct io_u *io_u)
+{
+ int ret;
+
+ ret = vmsplice_io_u_in(td, io_u);
+ if (ret <= 0)
+ return ret;
+
+ return splice_out(td, io_u, ret);
+}
+
+static int fio_netio_send(struct thread_data *td, struct io_u *io_u)
+{
+ int flags = 0;
+
+ /*
+ * if we are going to write more, set MSG_MORE
+ */
+ if (td->this_io_bytes[DDIR_WRITE] + io_u->xfer_buflen < td->o.size)
+ flags = MSG_MORE;
+
+ return send(io_u->file->fd, io_u->xfer_buf, io_u->xfer_buflen, flags);
+}
+
+static int fio_netio_recv(struct io_u *io_u)
+{
+ int flags = MSG_WAITALL;
+
+ return recv(io_u->file->fd, io_u->xfer_buf, io_u->xfer_buflen, flags);
+}
+
static int fio_netio_queue(struct thread_data *td, struct io_u *io_u)
{
- struct fio_file *f = io_u->file;
- int ret, flags = 0;
+ struct netio_data *nd = td->io_ops->data;
+ int ret;
if (io_u->ddir == DDIR_WRITE) {
- /*
- * if we are going to write more, set MSG_MORE
- */
- if (td->this_io_bytes[DDIR_WRITE] + io_u->xfer_buflen <
- td->o.size)
- flags = MSG_MORE;
-
- ret = send(f->fd, io_u->xfer_buf, io_u->xfer_buflen, flags);
+ if (nd->use_splice)
+ ret = fio_netio_splice_out(td, io_u);
+ else
+ ret = fio_netio_send(td, io_u);
} else if (io_u->ddir == DDIR_READ) {
- flags = MSG_WAITALL;
- ret = recv(f->fd, io_u->xfer_buf, io_u->xfer_buflen, flags);
+ if (nd->use_splice)
+ ret = fio_netio_splice_in(td, io_u);
+ else
+ ret = fio_netio_recv(io_u);
} else
ret = 0; /* must be a SYNC */
@@ -285,7 +442,25 @@
return 0;
}
-static struct ioengine_ops ioengine = {
+static int fio_netio_setup_splice(struct thread_data *td)
+{
+ struct netio_data *nd;
+
+ fio_netio_setup(td);
+
+ nd = td->io_ops->data;
+ if (nd) {
+ if (pipe(nd->pipes) < 0)
+ return 1;
+
+ nd->use_splice = 1;
+ return 0;
+ }
+
+ return 1;
+}
+
+static struct ioengine_ops ioengine_rw = {
.name = "net",
.version = FIO_IOOPS_VERSION,
.prep = fio_netio_prep,
@@ -298,12 +473,27 @@
.flags = FIO_SYNCIO | FIO_DISKLESSIO,
};
+static struct ioengine_ops ioengine_splice = {
+ .name = "netsplice",
+ .version = FIO_IOOPS_VERSION,
+ .prep = fio_netio_prep,
+ .queue = fio_netio_queue,
+ .setup = fio_netio_setup_splice,
+ .init = fio_netio_init,
+ .cleanup = fio_netio_cleanup,
+ .open_file = fio_netio_open_file,
+ .close_file = generic_close_file,
+ .flags = FIO_SYNCIO | FIO_DISKLESSIO,
+};
+
static void fio_init fio_netio_register(void)
{
- register_ioengine(&ioengine);
+ register_ioengine(&ioengine_rw);
+ register_ioengine(&ioengine_splice);
}
static void fio_exit fio_netio_unregister(void)
{
- unregister_ioengine(&ioengine);
+ unregister_ioengine(&ioengine_rw);
+ unregister_ioengine(&ioengine_splice);
}