aboutsummaryrefslogtreecommitdiffstats
path: root/vnet
diff options
context:
space:
mode:
authorShesha Sreenivasamurthy <shesha@cisco.com>2016-02-03 09:38:36 -0800
committerGerrit Code Review <gerrit@fd.io>2016-02-10 13:14:17 +0000
commitb8f45b37292fcb3b19c4ffcdb91947a9eec03500 (patch)
tree1ee2b5a9b78907025c9d14633bf3474ec94cd71a /vnet
parent1b1695316dab02fdbb626ca73ad61a92c31e417e (diff)
Vhost multi-q support
Remove multi-q support patch for DPDK-2.1.0 Change-Id: I685f3886b3883803210221682fa80353670a6970 Signed-off-by: Shesha Sreenivasamurthy <shesha@cisco.com>
Diffstat (limited to 'vnet')
-rw-r--r--vnet/vnet/devices/dpdk/device.c32
-rw-r--r--vnet/vnet/devices/dpdk/dpdk.h7
-rw-r--r--vnet/vnet/devices/dpdk/dpdk_priv.h25
-rw-r--r--vnet/vnet/devices/dpdk/vhost_user.c448
-rw-r--r--vnet/vnet/devices/virtio/vhost-user.h18
5 files changed, 415 insertions, 115 deletions
diff --git a/vnet/vnet/devices/dpdk/device.c b/vnet/vnet/devices/dpdk/device.c
index 92f8dcb7870..832853134bb 100644
--- a/vnet/vnet/devices/dpdk/device.c
+++ b/vnet/vnet/devices/dpdk/device.c
@@ -16,6 +16,7 @@
#include <vppinfra/vec.h>
#include <vppinfra/format.h>
#include <vlib/unix/cj.h>
+#include <assert.h>
#include <vnet/ethernet/ethernet.h>
#include <vnet/devices/dpdk/dpdk.h>
@@ -261,22 +262,37 @@ u32 tx_burst_vector_internal (vlib_main_t * vm,
}
else if (xd->dev_type == VNET_DPDK_DEV_VHOST_USER)
{
+ u32 offset = 0;
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ if (PREDICT_TRUE(xd->lockp == NULL)) {
+ dpdk_device_and_queue_t * dq;
+ vec_foreach (dq, dm->devices_by_cpu[vm->cpu_index])
+ {
+ if (xd->device_index == dq->device)
+ break;
+ }
+ assert (dq);
+ offset = dq->queue_id * VIRTIO_QNUM;
+ } else {
+ offset = queue_id * VIRTIO_QNUM;
+ }
+#endif
if (PREDICT_TRUE(tx_head > tx_tail))
{
/* no wrap, transmit in one burst */
- rv = rte_vhost_enqueue_burst(&xd->vu_vhost_dev, VIRTIO_RXQ,
+ rv = rte_vhost_enqueue_burst(&xd->vu_vhost_dev, offset + VIRTIO_RXQ,
&tx_vector[tx_tail],
(uint16_t) (tx_head-tx_tail));
if (PREDICT_TRUE(rv > 0))
{
- if (dpdk_vhost_user_want_interrupt(xd, VIRTIO_RXQ)) {
- dpdk_vu_vring *vring = &(xd->vu_intf->vrings[VIRTIO_RXQ]);
+ if (dpdk_vhost_user_want_interrupt(xd, offset + VIRTIO_RXQ)) {
+ dpdk_vu_vring *vring = &(xd->vu_intf->vrings[offset + VIRTIO_RXQ]);
vring->n_since_last_int += rv;
f64 now = vlib_time_now (vm);
if (vring->int_deadline < now ||
vring->n_since_last_int > dm->vhost_coalesce_frames)
- dpdk_vhost_user_send_interrupt(vm, xd, VIRTIO_RXQ);
+ dpdk_vhost_user_send_interrupt(vm, xd, offset + VIRTIO_RXQ);
}
int c = rv;
@@ -291,20 +307,20 @@ u32 tx_burst_vector_internal (vlib_main_t * vm,
* so we can try to transmit the rest. If we didn't transmit
* everything, stop now.
*/
- rv = rte_vhost_enqueue_burst(&xd->vu_vhost_dev, VIRTIO_RXQ,
+ rv = rte_vhost_enqueue_burst(&xd->vu_vhost_dev, offset + VIRTIO_RXQ,
&tx_vector[tx_tail],
(uint16_t) (DPDK_TX_RING_SIZE - tx_tail));
if (PREDICT_TRUE(rv > 0))
{
- if (dpdk_vhost_user_want_interrupt(xd, VIRTIO_RXQ)) {
- dpdk_vu_vring *vring = &(xd->vu_intf->vrings[VIRTIO_RXQ]);
+ if (dpdk_vhost_user_want_interrupt(xd, offset + VIRTIO_RXQ)) {
+ dpdk_vu_vring *vring = &(xd->vu_intf->vrings[offset + VIRTIO_RXQ]);
vring->n_since_last_int += rv;
f64 now = vlib_time_now (vm);
if (vring->int_deadline < now ||
vring->n_since_last_int > dm->vhost_coalesce_frames)
- dpdk_vhost_user_send_interrupt(vm, xd, VIRTIO_RXQ);
+ dpdk_vhost_user_send_interrupt(vm, xd, offset + VIRTIO_RXQ);
}
int c = rv;
diff --git a/vnet/vnet/devices/dpdk/dpdk.h b/vnet/vnet/devices/dpdk/dpdk.h
index 82456a0661c..73ba0f40334 100644
--- a/vnet/vnet/devices/dpdk/dpdk.h
+++ b/vnet/vnet/devices/dpdk/dpdk.h
@@ -137,6 +137,9 @@ typedef struct {
int callfd;
int kickfd;
int errfd;
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ int enabled;
+#endif
u32 callfd_idx;
u32 n_since_last_int;
f64 int_deadline;
@@ -154,7 +157,11 @@ typedef struct {
u64 feature_mask;
u32 num_vrings;
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ dpdk_vu_vring vrings[VHOST_MAX_QUEUE_PAIRS * 2];
+#else
dpdk_vu_vring vrings[2];
+#endif
u64 region_addr[VHOST_MEMORY_MAX_NREGIONS];
u32 region_fd[VHOST_MEMORY_MAX_NREGIONS];
} dpdk_vu_intf_t;
diff --git a/vnet/vnet/devices/dpdk/dpdk_priv.h b/vnet/vnet/devices/dpdk/dpdk_priv.h
index e499e2aec0b..bb26c3fcc15 100644
--- a/vnet/vnet/devices/dpdk/dpdk_priv.h
+++ b/vnet/vnet/devices/dpdk/dpdk_priv.h
@@ -342,30 +342,41 @@ dpdk_rx_burst ( dpdk_main_t * dm, dpdk_device_t * xd, u16 queue_id)
vlib_main_t * vm = vlib_get_main();
vlib_buffer_main_t * bm = vm->buffer_main;
unsigned socket_id = rte_socket_id();
+ u32 offset = 0;
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ offset = queue_id * VIRTIO_QNUM;
+
+ struct vhost_virtqueue *vq =
+ xd->vu_vhost_dev.virtqueue[offset + VIRTIO_TXQ];
+
+ if (PREDICT_FALSE(!vq->enabled))
+ return 0;
+#else
if (PREDICT_FALSE(!xd->vu_is_running))
return 0;
+#endif
- n_buffers = rte_vhost_dequeue_burst(&xd->vu_vhost_dev, VIRTIO_TXQ,
+ n_buffers = rte_vhost_dequeue_burst(&xd->vu_vhost_dev, offset + VIRTIO_TXQ,
bm->pktmbuf_pools[socket_id],
xd->rx_vectors[queue_id], VLIB_FRAME_SIZE);
f64 now = vlib_time_now (vm);
/* send pending interrupts if needed */
- if (dpdk_vhost_user_want_interrupt(xd, VIRTIO_TXQ)) {
- dpdk_vu_vring *vring = &(xd->vu_intf->vrings[VIRTIO_TXQ]);
+ if (dpdk_vhost_user_want_interrupt(xd, offset + VIRTIO_TXQ)) {
+ dpdk_vu_vring *vring = &(xd->vu_intf->vrings[offset + VIRTIO_TXQ]);
vring->n_since_last_int += n_buffers;
if ((vring->n_since_last_int && (vring->int_deadline < now))
|| (vring->n_since_last_int > dm->vhost_coalesce_frames))
- dpdk_vhost_user_send_interrupt(vm, xd, VIRTIO_TXQ);
+ dpdk_vhost_user_send_interrupt(vm, xd, offset + VIRTIO_TXQ);
}
- if (dpdk_vhost_user_want_interrupt(xd, VIRTIO_RXQ)) {
- dpdk_vu_vring *vring = &(xd->vu_intf->vrings[VIRTIO_RXQ]);
+ if (dpdk_vhost_user_want_interrupt(xd, offset + VIRTIO_RXQ)) {
+ dpdk_vu_vring *vring = &(xd->vu_intf->vrings[offset + VIRTIO_RXQ]);
if (vring->n_since_last_int && (vring->int_deadline < now))
- dpdk_vhost_user_send_interrupt(vm, xd, VIRTIO_RXQ);
+ dpdk_vhost_user_send_interrupt(vm, xd, offset + VIRTIO_RXQ);
}
}
diff --git a/vnet/vnet/devices/dpdk/vhost_user.c b/vnet/vnet/devices/dpdk/vhost_user.c
index 5ab4c22ed3e..1500879f1c5 100644
--- a/vnet/vnet/devices/dpdk/vhost_user.c
+++ b/vnet/vnet/devices/dpdk/vhost_user.c
@@ -12,6 +12,7 @@
* See the License for the specific language governing permissions and
* limitations under the License.
*/
+#include <assert.h>
#include <sys/socket.h>
#include <sys/un.h>
#include <sys/stat.h>
@@ -38,6 +39,30 @@
#define DBG_SOCK(args...)
#endif
+static const char *vhost_message_str[] = {
+ [VHOST_USER_NONE] = "VHOST_USER_NONE",
+ [VHOST_USER_GET_FEATURES] = "VHOST_USER_GET_FEATURES",
+ [VHOST_USER_SET_FEATURES] = "VHOST_USER_SET_FEATURES",
+ [VHOST_USER_SET_OWNER] = "VHOST_USER_SET_OWNER",
+ [VHOST_USER_RESET_OWNER] = "VHOST_USER_RESET_OWNER",
+ [VHOST_USER_SET_MEM_TABLE] = "VHOST_USER_SET_MEM_TABLE",
+ [VHOST_USER_SET_LOG_BASE] = "VHOST_USER_SET_LOG_BASE",
+ [VHOST_USER_SET_LOG_FD] = "VHOST_USER_SET_LOG_FD",
+ [VHOST_USER_SET_VRING_NUM] = "VHOST_USER_SET_VRING_NUM",
+ [VHOST_USER_SET_VRING_ADDR] = "VHOST_USER_SET_VRING_ADDR",
+ [VHOST_USER_SET_VRING_BASE] = "VHOST_USER_SET_VRING_BASE",
+ [VHOST_USER_GET_VRING_BASE] = "VHOST_USER_GET_VRING_BASE",
+ [VHOST_USER_SET_VRING_KICK] = "VHOST_USER_SET_VRING_KICK",
+ [VHOST_USER_SET_VRING_CALL] = "VHOST_USER_SET_VRING_CALL",
+ [VHOST_USER_SET_VRING_ERR] = "VHOST_USER_SET_VRING_ERR",
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ [VHOST_USER_GET_PROTOCOL_FEATURES] = "VHOST_USER_GET_PROTOCOL_FEATURES",
+ [VHOST_USER_SET_PROTOCOL_FEATURES] = "VHOST_USER_SET_PROTOCOL_FEATURES",
+ [VHOST_USER_GET_QUEUE_NUM] = "VHOST_USER_GET_QUEUE_NUM",
+ [VHOST_USER_SET_VRING_ENABLE] = "VHOST_USER_SET_VRING_ENABLE",
+#endif
+};
+
/*
* DPDK vhost-user functions
*/
@@ -122,6 +147,30 @@ dpdk_vhost_user_device_from_sw_if_index(u32 sw_if_index)
return dpdk_vhost_user_device_from_hw_if_index(sw->hw_if_index);
}
+static void stop_processing_packets(u32 hw_if_index, u8 idx)
+{
+ dpdk_device_t *xd =
+ dpdk_vhost_user_device_from_hw_if_index(hw_if_index);
+ assert(xd);
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ xd->vu_vhost_dev.virtqueue[idx]->enabled = 0;
+#else
+ xd->vu_is_running = 0;
+#endif
+}
+
+static void disable_interface(dpdk_device_t * xd)
+{
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ u8 idx;
+ int numqs = xd->vu_vhost_dev.virt_qp_nb * VIRTIO_QNUM;
+ for (idx = 0; idx < numqs; idx++)
+ xd->vu_vhost_dev.virtqueue[idx]->enabled = 0;
+#endif
+
+ xd->vu_is_running = 0;
+}
+
static inline void * map_guest_mem(dpdk_device_t * xd, u64 addr)
{
dpdk_vu_intf_t * vui = xd->vu_intf;
@@ -133,7 +182,7 @@ static inline void * map_guest_mem(dpdk_device_t * xd, u64 addr)
return (void *) (vui->region_addr[i] + addr - mem->regions[i].guest_phys_address);
}
}
- DBG_SOCK("failed to map guest mem addr %llx", addr);
+ DBG_SOCK("failed to map guest mem addr %lx", addr);
return 0;
}
@@ -146,6 +195,11 @@ dpdk_create_vhost_user_if_internal (u32 * hw_if_index, u32 if_id)
vnet_sw_interface_t * sw;
clib_error_t * error;
dpdk_device_and_queue_t * dq;
+ int num_qpairs = 1;
+
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ num_qpairs = dm->use_rss < 1 ? 1 : dm->use_rss;
+#endif
dpdk_device_t * xd = NULL;
u8 addr[6];
@@ -179,12 +233,16 @@ dpdk_create_vhost_user_if_internal (u32 * hw_if_index, u32 if_id)
xd->vu_if_id = if_id;
// reset virtqueues
- for (j = 0; j < VIRTIO_QNUM; j++)
- {
+ for (j = 0; j < num_qpairs * VIRTIO_QNUM; j++) {
memset(xd->vu_vhost_dev.virtqueue[j], 0, sizeof(struct vhost_virtqueue));
- }
+ xd->vu_vhost_dev.virtqueue[j]->kickfd = -1;
+ xd->vu_vhost_dev.virtqueue[j]->callfd = -1;
+ xd->vu_vhost_dev.virtqueue[j]->backend = -1;
+ }
+
// reset lockp
- memset ((void *) xd->lockp, 0, CLIB_CACHE_LINE_BYTES);
+ if (xd->lockp)
+ memset ((void *) xd->lockp, 0, CLIB_CACHE_LINE_BYTES);
// reset tx vectors
for (j = 0; j < tm->n_vlib_mains; j++)
@@ -205,7 +263,12 @@ dpdk_create_vhost_user_if_internal (u32 * hw_if_index, u32 if_id)
// vui was not retrieved from inactive ifaces - create new
vec_add2_aligned (dm->devices, xd, 1, CLIB_CACHE_LINE_BYTES);
xd->dev_type = VNET_DPDK_DEV_VHOST_USER;
- xd->rx_q_used = 1;
+ xd->rx_q_used = num_qpairs;
+ xd->tx_q_used = num_qpairs;
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ xd->vu_vhost_dev.virt_qp_nb = num_qpairs;
+#endif
+
vec_validate_aligned (xd->rx_vectors, xd->rx_q_used, CLIB_CACHE_LINE_BYTES);
if (if_id == (u32)~0)
@@ -221,15 +284,30 @@ dpdk_create_vhost_user_if_internal (u32 * hw_if_index, u32 if_id)
VHOST_MEMORY_MAX_NREGIONS *
sizeof(struct virtio_memory_regions));
- for (j = 0; j < VIRTIO_QNUM; j++)
- {
+ /* Will be set when guest sends VHOST_USER_SET_MEM_TABLE cmd */
+ xd->vu_vhost_dev.mem->nregions = 0;
+
+ /*
+ * New virtqueue structure is an array of VHOST_MAX_QUEUE_PAIRS * 2
+ * We need to allocate numq pairs.
+ */
+ for (j = 0; j < num_qpairs * VIRTIO_QNUM; j++) {
xd->vu_vhost_dev.virtqueue[j] = clib_mem_alloc (sizeof(struct vhost_virtqueue));
memset(xd->vu_vhost_dev.virtqueue[j], 0, sizeof(struct vhost_virtqueue));
- }
+ xd->vu_vhost_dev.virtqueue[j]->kickfd = -1;
+ xd->vu_vhost_dev.virtqueue[j]->callfd = -1;
+ xd->vu_vhost_dev.virtqueue[j]->backend = -1;
+ }
- xd->lockp = clib_mem_alloc_aligned (CLIB_CACHE_LINE_BYTES,
- CLIB_CACHE_LINE_BYTES);
- memset ((void *) xd->lockp, 0, CLIB_CACHE_LINE_BYTES);
+ xd->lockp = NULL;
+ if (xd->tx_q_used < dm->input_cpu_count) {
+ xd->lockp = clib_mem_alloc_aligned (CLIB_CACHE_LINE_BYTES,
+ CLIB_CACHE_LINE_BYTES);
+ memset ((void *) xd->lockp, 0, CLIB_CACHE_LINE_BYTES);
+ }
+
+ DBG_SOCK("tm->n_vlib_mains: %d. TX %d, RX: %d, num_qpairs: %d, Lock: %p",
+ tm->n_vlib_mains, xd->tx_q_used, xd->rx_q_used, num_qpairs, xd->lockp);
vec_validate_aligned (xd->tx_vectors, tm->n_vlib_mains,
CLIB_CACHE_LINE_BYTES);
@@ -253,7 +331,9 @@ dpdk_create_vhost_user_if_internal (u32 * hw_if_index, u32 if_id)
CLIB_CACHE_LINE_BYTES);
}
- {
+ /*
+ * Generate random MAC address for the interface
+ */
f64 now = vlib_time_now(vm);
u32 rnd;
rnd = (u32) (now * 1e6);
@@ -262,7 +342,6 @@ dpdk_create_vhost_user_if_internal (u32 * hw_if_index, u32 if_id)
memcpy (addr+2, &rnd, sizeof(rnd));
addr[0] = 2;
addr[1] = 0xfe;
- }
error = ethernet_register_interface
(dm->vnet_main,
@@ -283,34 +362,59 @@ dpdk_create_vhost_user_if_internal (u32 * hw_if_index, u32 if_id)
*hw_if_index = xd->vlib_hw_if_index;
- int cpu = (xd->device_index % dm->input_cpu_count) +
- dm->input_cpu_first_index;
+ DBG_SOCK("xd->device_index: %d, dm->input_cpu_count: %d, "
+ "dm->input_cpu_first_index: %d\n", xd->device_index,
+ dm->input_cpu_count, dm->input_cpu_first_index);
+
+ int q, next_cpu = 0;
+ for (q = 0; q < num_qpairs; q++) {
+ int cpu = dm->input_cpu_first_index +
+ (next_cpu % dm->input_cpu_count);
- vec_add2(dm->devices_by_cpu[cpu], dq, 1);
- dq->device = xd->device_index;
- dq->queue_id = 0;
+ unsigned lcore = vlib_worker_threads[cpu].dpdk_lcore_id;
+ vec_validate(xd->cpu_socket_id_by_queue, q);
+ xd->cpu_socket_id_by_queue[q] = rte_lcore_to_socket_id(lcore);
- // start polling if it was not started yet (because of no phys ifaces)
- if (tm->n_vlib_mains == 1 && dpdk_input_node.state != VLIB_NODE_STATE_POLLING)
- vlib_node_set_state (vm, dpdk_input_node.index, VLIB_NODE_STATE_POLLING);
+ vec_add2(dm->devices_by_cpu[cpu], dq, 1);
+ dq->device = xd->device_index;
+ dq->queue_id = q;
+ DBG_SOCK("CPU for %d = %d. QID: %d", *hw_if_index, cpu, dq->queue_id);
- if (tm->n_vlib_mains > 1 && tm->main_thread_is_io_node)
- vlib_node_set_state (vm, dpdk_io_input_node.index, VLIB_NODE_STATE_POLLING);
+ // start polling if it was not started yet (because of no phys ifaces)
+ if (tm->n_vlib_mains == 1 && dpdk_input_node.state != VLIB_NODE_STATE_POLLING)
+ vlib_node_set_state (vm, dpdk_input_node.index, VLIB_NODE_STATE_POLLING);
- if (tm->n_vlib_mains > 1 && !tm->main_thread_is_io_node)
- vlib_node_set_state (vlib_mains[cpu], dpdk_input_node.index,
- VLIB_NODE_STATE_POLLING);
+ if (tm->n_vlib_mains > 1 && tm->main_thread_is_io_node)
+ vlib_node_set_state (vm, dpdk_io_input_node.index, VLIB_NODE_STATE_POLLING);
+
+ if (tm->n_vlib_mains > 1 && !tm->main_thread_is_io_node)
+ vlib_node_set_state (vlib_mains[cpu], dpdk_input_node.index,
+ VLIB_NODE_STATE_POLLING);
+ next_cpu++;
+ }
vlib_worker_thread_barrier_release (vm);
return 0;
}
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+static clib_error_t *
+dpdk_vhost_user_set_protocol_features(u32 hw_if_index, u64 prot_features)
+{
+ dpdk_device_t * xd;
+ xd = dpdk_vhost_user_device_from_hw_if_index(hw_if_index);
+ assert(xd);
+ xd->vu_vhost_dev.protocol_features = prot_features;
+ return 0;
+}
+#endif
+
static clib_error_t *
dpdk_vhost_user_get_features(u32 hw_if_index, u64 * features)
{
*features = rte_vhost_feature_get();
- DBG_SOCK("supported features: 0x%x", *features);
+ DBG_SOCK("supported features: 0x%lx", *features);
return 0;
}
@@ -331,10 +435,13 @@ dpdk_vhost_user_set_features(u32 hw_if_index, u64 features)
if (xd->vu_vhost_dev.features & (1 << VIRTIO_NET_F_MRG_RXBUF))
hdr_len = sizeof(struct virtio_net_hdr_mrg_rxbuf);
- xd->vu_vhost_dev.virtqueue[VIRTIO_RXQ]->vhost_hlen = hdr_len;
- xd->vu_vhost_dev.virtqueue[VIRTIO_TXQ]->vhost_hlen = hdr_len;
-
- xd->vu_is_running = 0;
+ int numqs = VIRTIO_QNUM;
+ u8 idx;
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ numqs = xd->vu_vhost_dev.virt_qp_nb * VIRTIO_QNUM;
+#endif
+ for (idx = 0; idx < numqs; idx++)
+ xd->vu_vhost_dev.virtqueue[idx]->vhost_hlen = hdr_len;
return 0;
}
@@ -386,9 +493,7 @@ dpdk_vhost_user_set_mem_table(u32 hw_if_index, vhost_user_memory_t * vum, int fd
}
}
- xd->vu_is_running = 0;
-
- DBG_SOCK("done");
+ disable_interface(xd);
return 0;
}
@@ -407,7 +512,7 @@ dpdk_vhost_user_set_vring_num(u32 hw_if_index, u8 idx, u32 num)
vq = xd->vu_vhost_dev.virtqueue[idx];
vq->size = num;
- xd->vu_is_running = 0;
+ stop_processing_packets(hw_if_index, idx);
return 0;
}
@@ -418,7 +523,8 @@ dpdk_vhost_user_set_vring_addr(u32 hw_if_index, u8 idx, u64 desc, u64 used, u64
dpdk_device_t * xd;
struct vhost_virtqueue *vq;
- DBG_SOCK("idx %u desc 0x%x used 0x%x avail 0x%x", idx, desc, used, avail);
+ DBG_SOCK("idx %u desc 0x%lx used 0x%lx avail 0x%lx",
+ idx, desc, used, avail);
if (!(xd = dpdk_vhost_user_device_from_hw_if_index(hw_if_index))) {
clib_warning("not a vhost-user interface");
@@ -434,7 +540,7 @@ dpdk_vhost_user_set_vring_addr(u32 hw_if_index, u8 idx, u64 desc, u64 used, u64
clib_warning("falied to set vring addr");
}
- xd->vu_is_running = 0;
+ stop_processing_packets(hw_if_index, idx);
return 0;
}
@@ -453,7 +559,42 @@ dpdk_vhost_user_get_vring_base(u32 hw_if_index, u8 idx, u32 * num)
vq = xd->vu_vhost_dev.virtqueue[idx];
*num = vq->last_used_idx;
- DBG_SOCK("idx %u num %u", idx, *num);
+/*
+ * From spec:
+ * Client must start ring upon receiving a kick
+ * (that is, detecting that file descriptor is readable)
+ * on the descriptor specified by VHOST_USER_SET_VRING_KICK,
+ * and stop ring upon receiving VHOST_USER_GET_VRING_BASE.
+ */
+ dpdk_vu_intf_t *vui = xd->vu_intf;
+ DBG_SOCK("Stopping vring Q %u of device %d", idx, hw_if_index);
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ vui->vrings[idx].enabled = 0; /* Reset local copy */
+ vui->vrings[idx].callfd = -1; /* Reset FD */
+ vq->enabled = 0;
+ vq->desc = NULL;
+ vq->used = NULL;
+ vq->avail = NULL;
+
+ /* Check if all Qs are disabled */
+ int numqs = xd->vu_vhost_dev.virt_qp_nb * VIRTIO_QNUM;
+ for (idx = 0; idx < numqs; idx++) {
+ if (xd->vu_vhost_dev.virtqueue[idx]->enabled)
+ break;
+ }
+
+ /* If all vrings are disabed then disable device */
+ if (idx == numqs) {
+ DBG_SOCK("Device %d disabled", hw_if_index);
+ xd->vu_is_running = 0;
+ }
+#else
+ vq->desc = NULL;
+ vq->used = NULL;
+ vq->avail = NULL;
+ xd->vu_is_running = 0;
+#endif
+
return 0;
}
@@ -474,7 +615,7 @@ dpdk_vhost_user_set_vring_base(u32 hw_if_index, u8 idx, u32 num)
vq->last_used_idx = num;
vq->last_used_idx_res = num;
- xd->vu_is_running = 0;
+ stop_processing_packets(hw_if_index, idx);
return 0;
}
@@ -484,50 +625,126 @@ dpdk_vhost_user_set_vring_kick(u32 hw_if_index, u8 idx, int fd)
{
dpdk_main_t * dm = &dpdk_main;
dpdk_device_t * xd;
- struct vhost_virtqueue *vq, *vq0, *vq1;
-
- DBG_SOCK("idx %u fd %d", idx, fd);
+ dpdk_vu_vring *vring;
+ struct vhost_virtqueue *vq0, *vq1, *vq;
+ int index, vu_is_running = 0;
if (!(xd = dpdk_vhost_user_device_from_hw_if_index(hw_if_index))) {
clib_warning("not a vhost-user interface");
return 0;
}
+ vring = &xd->vu_intf->vrings[idx];
vq = xd->vu_vhost_dev.virtqueue[idx];
vq->kickfd = fd;
- vq0 = xd->vu_vhost_dev.virtqueue[0];
- vq1 = xd->vu_vhost_dev.virtqueue[1];
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ vq->enabled = (vq->desc && vq->avail && vq->used && vring->enabled) ? 1 : 0;
+#endif
+
+ /*
+ * Set xd->vu_is_running if at least one pair of
+ * RX/TX queues are enabled.
+ */
+ int numqs = VIRTIO_QNUM;
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ numqs = xd->vu_vhost_dev.virt_qp_nb * VIRTIO_QNUM;
+#endif
+
+ for (index = 0; index < numqs; index += 2) {
+ vq0 = xd->vu_vhost_dev.virtqueue[index]; /* RX */
+ vq1 = xd->vu_vhost_dev.virtqueue[index + 1]; /* TX */
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ if (vq0->enabled && vq1->enabled)
+#else
+ if (vq0->desc && vq0->avail && vq0->used &&
+ vq1->desc && vq1->avail && vq1->used)
+#endif
+ {
+ vu_is_running = 1;
+ break;
+ }
+ }
+ DBG_SOCK("SET_VRING_KICK - idx %d, running %d, fd: %d",
+ idx, vu_is_running, fd);
+
+ xd->vu_is_running = vu_is_running;
+ if (xd->vu_is_running && xd->admin_up) {
+ vnet_hw_interface_set_flags (dm->vnet_main,
+ xd->vlib_hw_if_index, VNET_HW_INTERFACE_FLAG_LINK_UP |
+ ETH_LINK_FULL_DUPLEX );
+ }
+
+ return 0;
+}
+
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+dpdk_vhost_user_set_vring_enable(u32 hw_if_index, u8 idx, int enable)
+{
+ dpdk_device_t * xd;
+ struct vhost_virtqueue *vq;
+ dpdk_vu_intf_t *vui;
- if (vq0->desc && vq0->avail && vq0->used &&
- vq1->desc && vq1->avail && vq1->used) {
- xd->vu_is_running = 1;
- if (xd->admin_up)
- vnet_hw_interface_set_flags (dm->vnet_main, xd->vlib_hw_if_index,
- VNET_HW_INTERFACE_FLAG_LINK_UP |
- ETH_LINK_FULL_DUPLEX );
+ if (!(xd = dpdk_vhost_user_device_from_hw_if_index(hw_if_index))) {
+ clib_warning("not a vhost-user interface");
+ return 0;
}
+ vui = xd->vu_intf;
+ /*
+ * Guest vhost driver wrongly enables queue before
+ * setting the vring address. Therefore, save a
+ * local copy. Reflect it in vq structure if addresses
+ * are set. If not, vq will be enabled when vring
+ * is kicked.
+ */
+ vui->vrings[idx].enabled = enable; /* Save local copy */
+
+ vq = xd->vu_vhost_dev.virtqueue[idx];
+ if (vq->desc && vq->avail && vq->used)
+ xd->vu_vhost_dev.virtqueue[idx]->enabled = enable;
+
return 0;
}
+#endif
+static clib_error_t * dpdk_vhost_user_callfd_read_ready (unix_file_t * uf)
+{
+ __attribute__((unused)) int n;
+ u8 buff[8];
+ n = read(uf->file_descriptor, ((char*)&buff), 8);
+ return 0;
+}
static clib_error_t *
dpdk_vhost_user_set_vring_call(u32 hw_if_index, u8 idx, int fd)
{
dpdk_device_t * xd;
struct vhost_virtqueue *vq;
+ unix_file_t template = {0};
- DBG_SOCK("idx %u fd %d", idx, fd);
+ DBG_SOCK("SET_VRING_CALL - idx %d, fd %d", idx, fd);
if (!(xd = dpdk_vhost_user_device_from_hw_if_index(hw_if_index))) {
clib_warning("not a vhost-user interface");
return 0;
}
+ dpdk_vu_intf_t *vui = xd->vu_intf;
+
+ /* if there is old fd, delete it */
+ if (vui->vrings[idx].callfd > 0) {
+ unix_file_t * uf = pool_elt_at_index (unix_main.file_pool,
+ vui->vrings[idx].callfd_idx);
+ unix_file_del (&unix_main, uf);
+ }
+ vui->vrings[idx].callfd = fd;
+ template.read_function = dpdk_vhost_user_callfd_read_ready;
+ template.file_descriptor = fd;
+ vui->vrings[idx].callfd_idx = unix_file_add (&unix_main, &template);
+
vq = xd->vu_vhost_dev.virtqueue[idx];
- /* reset callfd to force no interrupts */
- vq->callfd = -1;
+ vq->callfd = -1; /* We use locally saved vring->callfd; */
return 0;
}
@@ -563,10 +780,7 @@ dpdk_vhost_user_send_interrupt(vlib_main_t * vm, dpdk_device_t * xd, int idx)
/* if vm is interested in interrupts */
if((vring->callfd > 0) && !(vq->avail->flags & VRING_AVAIL_F_NO_INTERRUPT)) {
- u64 x = 1;
- int rv __attribute__((unused));
- /* $$$$ pay attention to rv */
- rv = write(vring->callfd, &x, sizeof(x));
+ eventfd_write(vring->callfd, (eventfd_t)1);
vring->n_since_last_int = 0;
vring->int_deadline = vlib_time_now(vm) + dm->vhost_coalesce_time;
}
@@ -588,7 +802,12 @@ dpdk_vhost_user_vui_init(vnet_main_t * vnm,
memset(vui, 0, sizeof(*vui));
vui->unix_fd = sockfd;
- vui->num_vrings = 2;
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ vui->num_vrings = xd->vu_vhost_dev.virt_qp_nb * VIRTIO_QNUM;
+#else
+ vui->num_vrings = VIRTIO_QNUM;
+#endif
+ DBG_SOCK("dpdk_vhost_user_vui_init VRINGS: %d", vui->num_vrings);
vui->sock_is_server = is_server;
strncpy(vui->sock_filename, sock_filename, ARRAY_LEN(vui->sock_filename)-1);
vui->sock_errno = 0;
@@ -638,14 +857,6 @@ dpdk_vhost_user_if_disconnect(dpdk_device_t * xd)
DBG_SOCK("interface ifindex %d disconnected", xd->vlib_sw_if_index);
}
-static clib_error_t * dpdk_vhost_user_callfd_read_ready (unix_file_t * uf)
-{
- __attribute__((unused)) int n;
- u8 buff[8];
- n = read(uf->file_descriptor, ((char*)&buff), 8);
- return 0;
-}
-
static clib_error_t * dpdk_vhost_user_socket_read (unix_file_t * uf)
{
int n;
@@ -660,7 +871,6 @@ static clib_error_t * dpdk_vhost_user_socket_read (unix_file_t * uf)
struct cmsghdr *cmsg;
uword * p;
u8 q;
- unix_file_t template = {0};
vnet_main_t * vnm = vnet_get_main();
p = hash_get (dm->vu_sw_if_index_by_sock_fd, uf->file_descriptor);
@@ -719,12 +929,13 @@ static clib_error_t * dpdk_vhost_user_socket_read (unix_file_t * uf)
rv = read(uf->file_descriptor, ((char*)&msg) + n, msg.size);
}
+ DBG_SOCK("VPP VHOST message %s", vhost_message_str[msg.request]);
switch (msg.request) {
case VHOST_USER_GET_FEATURES:
DBG_SOCK("if %d msg VHOST_USER_GET_FEATURES",
xd->vlib_hw_if_index);
- msg.flags |= 4;
+ msg.flags |= VHOST_USER_REPLY_MASK;
dpdk_vhost_user_get_features(xd->vlib_hw_if_index, &msg.u64);
msg.u64 &= vui->feature_mask;
@@ -732,7 +943,7 @@ static clib_error_t * dpdk_vhost_user_socket_read (unix_file_t * uf)
break;
case VHOST_USER_SET_FEATURES:
- DBG_SOCK("if %d msg VHOST_USER_SET_FEATURES features 0x%016llx",
+ DBG_SOCK("if %d msg VHOST_USER_SET_FEATURES features 0x%016lx",
xd->vlib_hw_if_index, msg.u64);
dpdk_vhost_user_set_features(xd->vlib_hw_if_index, msg.u64);
@@ -792,39 +1003,30 @@ static clib_error_t * dpdk_vhost_user_socket_read (unix_file_t * uf)
break;
case VHOST_USER_SET_VRING_CALL:
- DBG_SOCK("if %d msg VHOST_USER_SET_VRING_CALL u64 %d",
- xd->vlib_hw_if_index, msg.u64);
-
q = (u8) (msg.u64 & 0xFF);
+ DBG_SOCK("if %d msg VHOST_USER_SET_VRING_CALL u64 %lx, idx: %d",
+ xd->vlib_hw_if_index, msg.u64, q);
+
if (!(msg.u64 & 0x100))
{
if (number_of_fds != 1)
goto close_socket;
-
- /* if there is old fd, delete it */
- if (vui->vrings[q].callfd) {
- unix_file_t * uf = pool_elt_at_index (unix_main.file_pool,
- vui->vrings[q].callfd_idx);
- unix_file_del (&unix_main, uf);
- }
- vui->vrings[q].callfd = fds[0];
- template.read_function = dpdk_vhost_user_callfd_read_ready;
- template.file_descriptor = fds[0];
- vui->vrings[q].callfd_idx = unix_file_add (&unix_main, &template);
+ fd = fds[0];
+ } else {
+ fd = -1;
}
- else
- vui->vrings[q].callfd = -1;
+ dpdk_vhost_user_set_vring_call(xd->vlib_hw_if_index, q, fd);
- dpdk_vhost_user_set_vring_call(xd->vlib_hw_if_index, q, vui->vrings[q].callfd);
break;
case VHOST_USER_SET_VRING_KICK:
- DBG_SOCK("if %d msg VHOST_USER_SET_VRING_KICK u64 %d",
- xd->vlib_hw_if_index, msg.u64);
q = (u8) (msg.u64 & 0xFF);
+ DBG_SOCK("if %d msg VHOST_USER_SET_VRING_KICK u64 %lx, idx: %d",
+ xd->vlib_hw_if_index, msg.u64, q);
+
if (!(msg.u64 & 0x100))
{
if (number_of_fds != 1)
@@ -839,11 +1041,12 @@ static clib_error_t * dpdk_vhost_user_socket_read (unix_file_t * uf)
break;
case VHOST_USER_SET_VRING_ERR:
- DBG_SOCK("if %d msg VHOST_USER_SET_VRING_ERR u64 %d",
- xd->vlib_hw_if_index, msg.u64);
q = (u8) (msg.u64 & 0xFF);
+ DBG_SOCK("if %d msg VHOST_USER_SET_VRING_ERR u64 %lx, idx: %d",
+ xd->vlib_hw_if_index, msg.u64, q);
+
if (!(msg.u64 & 0x100))
{
if (number_of_fds != 1)
@@ -868,7 +1071,7 @@ static clib_error_t * dpdk_vhost_user_socket_read (unix_file_t * uf)
DBG_SOCK("if %d msg VHOST_USER_GET_VRING_BASE idx %d num %d",
xd->vlib_hw_if_index, msg.state.index, msg.state.num);
- msg.flags |= 4;
+ msg.flags |= VHOST_USER_REPLY_MASK;
msg.size = sizeof(msg.state);
dpdk_vhost_user_get_vring_base(xd->vlib_hw_if_index, msg.state.index, &msg.state.num);
@@ -889,6 +1092,44 @@ static clib_error_t * dpdk_vhost_user_socket_read (unix_file_t * uf)
xd->vlib_hw_if_index);
break;
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ case VHOST_USER_GET_PROTOCOL_FEATURES:
+ DBG_SOCK("if %d msg VHOST_USER_GET_PROTOCOL_FEATURES",
+ xd->vlib_hw_if_index);
+
+ msg.flags |= VHOST_USER_REPLY_MASK;
+ msg.u64 = VHOST_USER_PROTOCOL_FEATURES;
+ DBG_SOCK("VHOST_USER_PROTOCOL_FEATURES: %llx", VHOST_USER_PROTOCOL_FEATURES);
+ msg.size = sizeof(msg.u64);
+ break;
+
+ case VHOST_USER_SET_PROTOCOL_FEATURES:
+ DBG_SOCK("if %d msg VHOST_USER_SET_PROTOCOL_FEATURES",
+ xd->vlib_hw_if_index);
+
+ DBG_SOCK("VHOST_USER_SET_PROTOCOL_FEATURES: 0x%lx",
+ msg.u64);
+ dpdk_vhost_user_set_protocol_features(xd->vlib_hw_if_index,
+ msg.u64);
+ break;
+
+ case VHOST_USER_SET_VRING_ENABLE:
+ DBG_SOCK("%d VPP VHOST_USER_SET_VRING_ENABLE IDX: %d, Enable: %d",
+ xd->vlib_hw_if_index, msg.state.index, msg.state.num);
+ dpdk_vhost_user_set_vring_enable
+ (xd->vlib_hw_if_index, msg.state.index, msg.state.num);
+ break;
+
+ case VHOST_USER_GET_QUEUE_NUM:
+ DBG_SOCK("if %d msg VHOST_USER_GET_QUEUE_NUM:",
+ xd->vlib_hw_if_index);
+
+ msg.flags |= VHOST_USER_REPLY_MASK;
+ msg.u64 = xd->vu_vhost_dev.virt_qp_nb;
+ msg.size = sizeof(msg.u64);
+ break;
+#endif
+
default:
DBG_SOCK("unknown vhost-user message %d received. closing socket",
msg.request);
@@ -904,10 +1145,11 @@ static clib_error_t * dpdk_vhost_user_socket_read (unix_file_t * uf)
vnet_hw_interface_set_flags (vnm, xd->vlib_hw_if_index, VNET_HW_INTERFACE_FLAG_LINK_UP);
vui->is_up = 1;
+ xd->admin_up = 1;
}
/* if we need to reply */
- if (msg.flags & 4)
+ if (msg.flags & VHOST_USER_REPLY_MASK)
{
n = send(uf->file_descriptor, &msg, VHOST_USER_MSG_HDR_SZ + msg.size, 0);
if (n != (msg.size + VHOST_USER_MSG_HDR_SZ))
@@ -1470,7 +1712,7 @@ show_dpdk_vhost_user_command_fn (vlib_main_t * vm,
u32 virtio_net_hdr_sz = (vui->num_vrings > 0 ?
vhost_dev->virtqueue[0]->vhost_hlen : 0);
- vlib_cli_output (vm, "Interface: %s (ifindex %d)",
+ vlib_cli_output (vm, "Interface: %v (ifindex %d)",
hi->name, hw_if_indices[i]);
vlib_cli_output (vm, "virtio_net_hdr_sz %d\n features (0x%llx): \n",
@@ -1506,8 +1748,9 @@ show_dpdk_vhost_user_command_fn (vlib_main_t * vm,
}
for (q = 0; q < vui->num_vrings; q++) {
struct vhost_virtqueue *vq = vhost_dev->virtqueue[q];
+ const char *qtype = (q & 1) ? "TX" : "RX";
- vlib_cli_output(vm, "\n Virtqueue %d\n", q);
+ vlib_cli_output(vm, "\n Virtqueue %d (%s)\n", q/2, qtype);
vlib_cli_output(vm, " qsz %d last_used_idx %d last_used_idx_res %d\n",
vq->size, vq->last_used_idx, vq->last_used_idx_res);
@@ -1516,12 +1759,17 @@ show_dpdk_vhost_user_command_fn (vlib_main_t * vm,
vlib_cli_output(vm, " avail.flags %x avail.idx %d used.flags %x used.idx %d\n",
vq->avail->flags, vq->avail->idx, vq->used->flags, vq->used->idx);
- vlib_cli_output(vm, " kickfd %d callfd %d errfd %d\n",
- vui->vrings[q].kickfd,
- vui->vrings[q].callfd,
- vui->vrings[q].errfd);
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ vlib_cli_output(vm, " kickfd %d callfd %d errfd %d enabled %d\n",
+ vq->kickfd, vq->callfd, vui->vrings[q].errfd, vq->enabled);
+
+ if (show_descr && vq->enabled) {
+#else
+ vlib_cli_output(vm, " kickfd %d callfd %d errfd\n",
+ vq->kickfd, vq->callfd, vui->vrings[q].errfd);
if (show_descr) {
+#endif
vlib_cli_output(vm, "\n descriptor table:\n");
vlib_cli_output(vm, " id addr len flags next user_addr\n");
vlib_cli_output(vm, " ===== ================== ===== ====== ===== ==================\n");
diff --git a/vnet/vnet/devices/virtio/vhost-user.h b/vnet/vnet/devices/virtio/vhost-user.h
index 3b57bcbfc16..1fbc0f0bdd7 100644
--- a/vnet/vnet/devices/virtio/vhost-user.h
+++ b/vnet/vnet/devices/virtio/vhost-user.h
@@ -24,6 +24,18 @@
#define VHOST_NET_VRING_NUM 2
#define VIRTQ_DESC_F_NEXT 1
+#define VHOST_USER_REPLY_MASK (0x1 << 2)
+
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+#define VHOST_USER_PROTOCOL_F_MQ 0
+#define VHOST_USER_PROTOCOL_FEATURES (1ULL << VHOST_USER_PROTOCOL_F_MQ)
+
+/* If multiqueue is provided by host, then we suppport it. */
+#define VIRTIO_NET_CTRL_MQ 4
+#define VIRTIO_NET_CTRL_MQ_VQ_PAIRS_SET 0
+#define VIRTIO_NET_CTRL_MQ_VQ_PAIRS_MIN 1
+#define VIRTIO_NET_CTRL_MQ_VQ_PAIRS_MAX 0x8000
+#endif
#define foreach_virtio_net_feature \
_ (VIRTIO_NET_F_MRG_RXBUF, 15) \
@@ -83,6 +95,12 @@ typedef enum vhost_user_req {
VHOST_USER_SET_VRING_KICK = 12,
VHOST_USER_SET_VRING_CALL = 13,
VHOST_USER_SET_VRING_ERR = 14,
+#if RTE_VERSION >= RTE_VERSION_NUM(2, 2, 0, 0)
+ VHOST_USER_GET_PROTOCOL_FEATURES = 15,
+ VHOST_USER_SET_PROTOCOL_FEATURES = 16,
+ VHOST_USER_GET_QUEUE_NUM = 17,
+ VHOST_USER_SET_VRING_ENABLE = 18,
+#endif
VHOST_USER_MAX
} vhost_user_req_t;