#include <rte_mbuf.h>
#include <rte_malloc.h>
-#include <rte_ethdev_driver.h>
+#include <ethdev_driver.h>
#include <rte_common.h>
#include <rte_interrupts.h>
#include <rte_debug.h>
#include "mlx5_defs.h"
#include "mlx5.h"
-#include "mlx5_rxtx.h"
+#include "mlx5_tx.h"
+#include "mlx5_rx.h"
#include "mlx5_utils.h"
#include "mlx5_autoconf.h"
(unsigned int)sizeof(rss_hash_default_key),
"wrong RSS default key size.");
-/**
- * Check whether Multi-Packet RQ can be enabled for the device.
- *
- * @param dev
- * Pointer to Ethernet device.
- *
- * @return
- * 1 if supported, negative errno value if not.
- */
-inline int
-mlx5_check_mprq_support(struct rte_eth_dev *dev)
-{
- struct mlx5_priv *priv = dev->data->dev_private;
-
- if (priv->config.mprq.enabled &&
- priv->rxqs_n >= priv->config.mprq.min_rxqs_num)
- return 1;
- return -ENOTSUP;
-}
-
-/**
- * Check whether Multi-Packet RQ is enabled for the Rx queue.
- *
- * @param rxq
- * Pointer to receive queue structure.
- *
- * @return
- * 0 if disabled, otherwise enabled.
- */
-inline int
-mlx5_rxq_mprq_enabled(struct mlx5_rxq_data *rxq)
-{
- return rxq->strd_num_n > 0;
-}
-
-/**
- * Check whether Multi-Packet RQ is enabled for the device.
- *
- * @param dev
- * Pointer to Ethernet device.
- *
- * @return
- * 0 if disabled, otherwise enabled.
- */
-inline int
-mlx5_mprq_enabled(struct rte_eth_dev *dev)
-{
- struct mlx5_priv *priv = dev->data->dev_private;
- uint32_t i;
- uint16_t n = 0;
- uint16_t n_ibv = 0;
-
- if (mlx5_check_mprq_support(dev) < 0)
- return 0;
- /* All the configured queues should be enabled. */
- for (i = 0; i < priv->rxqs_n; ++i) {
- struct mlx5_rxq_data *rxq = (*priv->rxqs)[i];
- struct mlx5_rxq_ctrl *rxq_ctrl = container_of
- (rxq, struct mlx5_rxq_ctrl, rxq);
-
- if (rxq == NULL || rxq_ctrl->type != MLX5_RXQ_TYPE_STANDARD)
- continue;
- n_ibv++;
- if (mlx5_rxq_mprq_enabled(rxq))
- ++n;
- }
- /* Multi-Packet RQ can't be partially configured. */
- MLX5_ASSERT(n == 0 || n == n_ibv);
- return n == n_ibv;
-}
-
/**
* Calculate the number of CQEs in CQ for the Rx queue.
*
/* Iterate on segments. */
for (i = 0; (i != elts_n); ++i) {
+ struct mlx5_eth_rxseg *seg = &rxq_ctrl->rxq.rxseg[i % sges_n];
struct rte_mbuf *buf;
- buf = rte_pktmbuf_alloc(rxq_ctrl->rxq.mp);
+ buf = rte_pktmbuf_alloc(seg->mp);
if (buf == NULL) {
DRV_LOG(ERR, "port %u empty mbuf pool",
PORT_ID(rxq_ctrl->priv));
MLX5_ASSERT(rte_pktmbuf_data_len(buf) == 0);
MLX5_ASSERT(rte_pktmbuf_pkt_len(buf) == 0);
MLX5_ASSERT(!buf->next);
- /* Only the first segment keeps headroom. */
- if (i % sges_n)
- SET_DATA_OFF(buf, 0);
+ SET_DATA_OFF(buf, seg->offset);
PORT(buf) = rxq_ctrl->rxq.port_id;
- DATA_LEN(buf) = rte_pktmbuf_tailroom(buf);
- PKT_LEN(buf) = DATA_LEN(buf);
+ DATA_LEN(buf) = seg->length;
+ PKT_LEN(buf) = seg->length;
NB_SEGS(buf) = 1;
(*rxq_ctrl->rxq.elts)[i] = buf;
}
(1 << rxq->elts_n) * (1 << rxq->strd_num_n) :
(1 << rxq->elts_n);
const uint16_t q_mask = q_n - 1;
- uint16_t used = q_n - (rxq->rq_ci - rxq->rq_pi);
+ uint16_t elts_ci = mlx5_rxq_mprq_enabled(&rxq_ctrl->rxq) ?
+ rxq->elts_ci : rxq->rq_ci;
+ uint16_t used = q_n - (elts_ci - rxq->rq_pi);
uint16_t i;
DRV_LOG(DEBUG, "port %u Rx queue %u freeing %d WRs",
*/
if (mlx5_rxq_check_vec_support(rxq) > 0) {
for (i = 0; i < used; ++i)
- (*rxq->elts)[(rxq->rq_ci + i) & q_mask] = NULL;
- rxq->rq_pi = rxq->rq_ci;
+ (*rxq->elts)[(elts_ci + i) & q_mask] = NULL;
+ rxq->rq_pi = elts_ci;
}
for (i = 0; i != q_n; ++i) {
if ((*rxq->elts)[i] != NULL)
DEV_RX_OFFLOAD_JUMBO_FRAME |
DEV_RX_OFFLOAD_RSS_HASH);
+ if (!config->mprq.enabled)
+ offloads |= RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT;
if (config->hw_fcs_strip)
offloads |= DEV_RX_OFFLOAD_KEEP_CRC;
-
if (config->hw_csum)
offloads |= (DEV_RX_OFFLOAD_IPV4_CKSUM |
DEV_RX_OFFLOAD_UDP_CKSUM |
}
rxq_ctrl = container_of((*priv->rxqs)[idx], struct mlx5_rxq_ctrl, rxq);
return (__atomic_load_n(&rxq_ctrl->refcnt, __ATOMIC_RELAXED) == 1);
-
}
/* Fetches and drops all SW-owned and error CQEs to synchronize CQ. */
rte_io_wmb();
*rxq->cq_db = rte_cpu_to_be_32(rxq->cq_ci);
rte_io_wmb();
- /* Reset RQ consumer before moving queue ro READY state. */
+ /* Reset RQ consumer before moving queue to READY state. */
*rxq->rq_db = rte_cpu_to_be_32(0);
rte_io_wmb();
ret = priv->obj_ops.rxq_obj_modify(rxq_ctrl->obj, MLX5_RXQ_MOD_RST2RDY);
struct mlx5_rxq_data *rxq = (*priv->rxqs)[idx];
struct mlx5_rxq_ctrl *rxq_ctrl =
container_of(rxq, struct mlx5_rxq_ctrl, rxq);
+ struct rte_eth_rxseg_split *rx_seg =
+ (struct rte_eth_rxseg_split *)conf->rx_seg;
+ struct rte_eth_rxseg_split rx_single = {.mp = mp};
+ uint16_t n_seg = conf->rx_nseg;
int res;
+ if (mp) {
+ /*
+ * The parameters should be checked on rte_eth_dev layer.
+ * If mp is specified it means the compatible configuration
+ * without buffer split feature tuning.
+ */
+ rx_seg = &rx_single;
+ n_seg = 1;
+ }
+ if (n_seg > 1) {
+ uint64_t offloads = conf->offloads |
+ dev->data->dev_conf.rxmode.offloads;
+
+ /* The offloads should be checked on rte_eth_dev layer. */
+ MLX5_ASSERT(offloads & DEV_RX_OFFLOAD_SCATTER);
+ if (!(offloads & RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT)) {
+ DRV_LOG(ERR, "port %u queue index %u split "
+ "offload not configured",
+ dev->data->port_id, idx);
+ rte_errno = ENOSPC;
+ return -rte_errno;
+ }
+ MLX5_ASSERT(n_seg < MLX5_MAX_RXQ_NSEG);
+ }
res = mlx5_rx_queue_pre_setup(dev, idx, &desc);
if (res)
return res;
- rxq_ctrl = mlx5_rxq_new(dev, idx, desc, socket, conf, mp);
+ rxq_ctrl = mlx5_rxq_new(dev, idx, desc, socket, conf, rx_seg, n_seg);
if (!rxq_ctrl) {
DRV_LOG(ERR, "port %u unable to allocate queue index %u",
dev->data->port_id, idx);
res = mlx5_rx_queue_pre_setup(dev, idx, &desc);
if (res)
return res;
- if (hairpin_conf->peer_count != 1 ||
- hairpin_conf->peers[0].port != dev->data->port_id ||
- hairpin_conf->peers[0].queue >= priv->txqs_n) {
- DRV_LOG(ERR, "port %u unable to setup hairpin queue index %u "
- " invalid hairpind configuration", dev->data->port_id,
- idx);
+ if (hairpin_conf->peer_count != 1) {
rte_errno = EINVAL;
+ DRV_LOG(ERR, "port %u unable to setup Rx hairpin queue index %u"
+ " peer count is %u", dev->data->port_id,
+ idx, hairpin_conf->peer_count);
return -rte_errno;
}
+ if (hairpin_conf->peers[0].port == dev->data->port_id) {
+ if (hairpin_conf->peers[0].queue >= priv->txqs_n) {
+ rte_errno = EINVAL;
+ DRV_LOG(ERR, "port %u unable to setup Rx hairpin queue"
+ " index %u, Tx %u is larger than %u",
+ dev->data->port_id, idx,
+ hairpin_conf->peers[0].queue, priv->txqs_n);
+ return -rte_errno;
+ }
+ } else {
+ if (hairpin_conf->manual_bind == 0 ||
+ hairpin_conf->tx_explicit == 0) {
+ rte_errno = EINVAL;
+ DRV_LOG(ERR, "port %u unable to setup Rx hairpin queue"
+ " index %u peer port %u with attributes %u %u",
+ dev->data->port_id, idx,
+ hairpin_conf->peers[0].port,
+ hairpin_conf->manual_bind,
+ hairpin_conf->tx_explicit);
+ return -rte_errno;
+ }
+ }
rxq_ctrl = mlx5_rxq_hairpin_new(dev, idx, desc, hairpin_conf);
if (!rxq_ctrl) {
DRV_LOG(ERR, "port %u unable to allocate queue index %u",
unsigned int count = 0;
struct rte_intr_handle *intr_handle = dev->intr_handle;
+ /* Representor shares dev->intr_handle with PF. */
+ if (priv->representor)
+ return 0;
if (!dev->data->dev_conf.intr_conf.rxq)
return 0;
mlx5_rx_intr_vec_disable(dev);
unsigned int rxqs_n = priv->rxqs_n;
unsigned int n = RTE_MIN(rxqs_n, (uint32_t)RTE_MAX_RXTX_INTR_VEC_ID);
+ /* Representor shares dev->intr_handle with PF. */
+ if (priv->representor)
+ return;
if (!dev->data->dev_conf.intr_conf.rxq)
return;
if (!intr_handle->intr_vec)
struct mlx5_rxq_ctrl *
mlx5_rxq_new(struct rte_eth_dev *dev, uint16_t idx, uint16_t desc,
unsigned int socket, const struct rte_eth_rxconf *conf,
- struct rte_mempool *mp)
+ const struct rte_eth_rxseg_split *rx_seg, uint16_t n_seg)
{
struct mlx5_priv *priv = dev->data->dev_private;
struct mlx5_rxq_ctrl *tmpl;
- unsigned int mb_len = rte_pktmbuf_data_room_size(mp);
+ unsigned int mb_len = rte_pktmbuf_data_room_size(rx_seg[0].mp);
struct mlx5_dev_config *config = &priv->config;
uint64_t offloads = conf->offloads |
dev->data->dev_conf.rxmode.offloads;
RTE_PKTMBUF_HEADROOM;
unsigned int max_lro_size = 0;
unsigned int first_mb_free_size = mb_len - RTE_PKTMBUF_HEADROOM;
- const int mprq_en = mlx5_check_mprq_support(dev) > 0;
+ const int mprq_en = mlx5_check_mprq_support(dev) > 0 && n_seg == 1 &&
+ !rx_seg[0].offset && !rx_seg[0].length;
unsigned int mprq_stride_nums = config->mprq.stride_num_n ?
config->mprq.stride_num_n : MLX5_MPRQ_STRIDE_NUM_N;
unsigned int mprq_stride_size = non_scatter_min_mbuf_size <=
* the vector Rx will not be used.
*/
uint16_t desc_n = desc + config->rx_vec_en * MLX5_VPMD_DESCS_PER_LOOP;
+ const struct rte_eth_rxseg_split *qs_seg = rx_seg;
+ unsigned int tail_len;
- if (non_scatter_min_mbuf_size > mb_len && !(offloads &
- DEV_RX_OFFLOAD_SCATTER)) {
- DRV_LOG(ERR, "port %u Rx queue %u: Scatter offload is not"
- " configured and no enough mbuf space(%u) to contain "
- "the maximum RX packet length(%u) with head-room(%u)",
- dev->data->port_id, idx, mb_len, max_rx_pkt_len,
- RTE_PKTMBUF_HEADROOM);
- rte_errno = ENOSPC;
- return NULL;
- }
tmpl = mlx5_malloc(MLX5_MEM_RTE | MLX5_MEM_ZERO,
sizeof(*tmpl) + desc_n * sizeof(struct rte_mbuf *) +
+ (!!mprq_en) *
(desc >> mprq_stride_nums) * sizeof(struct mlx5_mprq_buf *),
0, socket);
-
if (!tmpl) {
rte_errno = ENOMEM;
return NULL;
}
+ MLX5_ASSERT(n_seg && n_seg <= MLX5_MAX_RXQ_NSEG);
+ /*
+ * Build the array of actual buffer offsets and lengths.
+ * Pad with the buffers from the last memory pool if
+ * needed to handle max size packets, replace zero length
+ * with the buffer length from the pool.
+ */
+ tail_len = max_rx_pkt_len;
+ do {
+ struct mlx5_eth_rxseg *hw_seg =
+ &tmpl->rxq.rxseg[tmpl->rxq.rxseg_n];
+ uint32_t buf_len, offset, seg_len;
+
+ /*
+ * For the buffers beyond descriptions offset is zero,
+ * the first buffer contains head room.
+ */
+ buf_len = rte_pktmbuf_data_room_size(qs_seg->mp);
+ offset = (tmpl->rxq.rxseg_n >= n_seg ? 0 : qs_seg->offset) +
+ (tmpl->rxq.rxseg_n ? 0 : RTE_PKTMBUF_HEADROOM);
+ /*
+ * For the buffers beyond descriptions the length is
+ * pool buffer length, zero lengths are replaced with
+ * pool buffer length either.
+ */
+ seg_len = tmpl->rxq.rxseg_n >= n_seg ? buf_len :
+ qs_seg->length ?
+ qs_seg->length :
+ (buf_len - offset);
+ /* Check is done in long int, now overflows. */
+ if (buf_len < seg_len + offset) {
+ DRV_LOG(ERR, "port %u Rx queue %u: Split offset/length "
+ "%u/%u can't be satisfied",
+ dev->data->port_id, idx,
+ qs_seg->length, qs_seg->offset);
+ rte_errno = EINVAL;
+ goto error;
+ }
+ if (seg_len > tail_len)
+ seg_len = buf_len - offset;
+ if (++tmpl->rxq.rxseg_n > MLX5_MAX_RXQ_NSEG) {
+ DRV_LOG(ERR,
+ "port %u too many SGEs (%u) needed to handle"
+ " requested maximum packet size %u, the maximum"
+ " supported are %u", dev->data->port_id,
+ tmpl->rxq.rxseg_n, max_rx_pkt_len,
+ MLX5_MAX_RXQ_NSEG);
+ rte_errno = ENOTSUP;
+ goto error;
+ }
+ /* Build the actual scattering element in the queue object. */
+ hw_seg->mp = qs_seg->mp;
+ MLX5_ASSERT(offset <= UINT16_MAX);
+ MLX5_ASSERT(seg_len <= UINT16_MAX);
+ hw_seg->offset = (uint16_t)offset;
+ hw_seg->length = (uint16_t)seg_len;
+ /*
+ * Advance the segment descriptor, the padding is the based
+ * on the attributes of the last descriptor.
+ */
+ if (tmpl->rxq.rxseg_n < n_seg)
+ qs_seg++;
+ tail_len -= RTE_MIN(tail_len, seg_len);
+ } while (tail_len || !rte_is_power_of_2(tmpl->rxq.rxseg_n));
+ MLX5_ASSERT(tmpl->rxq.rxseg_n &&
+ tmpl->rxq.rxseg_n <= MLX5_MAX_RXQ_NSEG);
+ if (tmpl->rxq.rxseg_n > 1 && !(offloads & DEV_RX_OFFLOAD_SCATTER)) {
+ DRV_LOG(ERR, "port %u Rx queue %u: Scatter offload is not"
+ " configured and no enough mbuf space(%u) to contain "
+ "the maximum RX packet length(%u) with head-room(%u)",
+ dev->data->port_id, idx, mb_len, max_rx_pkt_len,
+ RTE_PKTMBUF_HEADROOM);
+ rte_errno = ENOSPC;
+ goto error;
+ }
tmpl->type = MLX5_RXQ_TYPE_STANDARD;
if (mlx5_mr_btree_init(&tmpl->rxq.mr_ctrl.cache_bh,
MLX5_MR_BTREE_CACHE_N, socket)) {
* - The number of descs is more than the number of strides.
* - max_rx_pkt_len plus overhead is less than the max size
* of a stride or mprq_stride_size is specified by a user.
- * Need to nake sure that there are enough stides to encap
+ * Need to make sure that there are enough strides to encap
* the maximum packet size in case mprq_stride_size is set.
* Otherwise, enable Rx scatter if necessary.
*/
" strd_num_n = %u, strd_sz_n = %u",
dev->data->port_id, idx,
tmpl->rxq.strd_num_n, tmpl->rxq.strd_sz_n);
- } else if (max_rx_pkt_len <= first_mb_free_size) {
+ } else if (tmpl->rxq.rxseg_n == 1) {
+ MLX5_ASSERT(max_rx_pkt_len <= first_mb_free_size);
tmpl->rxq.sges_n = 0;
max_lro_size = max_rx_pkt_len;
} else if (offloads & DEV_RX_OFFLOAD_SCATTER) {
- unsigned int size = non_scatter_min_mbuf_size;
unsigned int sges_n;
if (lro_on_queue && first_mb_free_size <
* Determine the number of SGEs needed for a full packet
* and round it to the next power of two.
*/
- sges_n = log2above((size / mb_len) + !!(size % mb_len));
+ sges_n = log2above(tmpl->rxq.rxseg_n);
if (sges_n > MLX5_MAX_LOG_RQ_SEGS) {
DRV_LOG(ERR,
"port %u too many SGEs (%u) needed to handle"
(!!(dev->data->dev_conf.rxmode.mq_mode & ETH_MQ_RX_RSS));
tmpl->rxq.port_id = dev->data->port_id;
tmpl->priv = priv;
- tmpl->rxq.mp = mp;
+ tmpl->rxq.mp = rx_seg[0].mp;
tmpl->rxq.elts_n = log2above(desc);
tmpl->rxq.rq_repl_thresh =
MLX5_VPMD_RXQ_RPLNSH_THRESH(desc_n);
tmpl->rxq.uar_lock_cq = &priv->sh->uar_lock_cq;
#endif
tmpl->rxq.idx = idx;
- __atomic_add_fetch(&tmpl->refcnt, 1, __ATOMIC_RELAXED);
+ __atomic_fetch_add(&tmpl->refcnt, 1, __ATOMIC_RELAXED);
LIST_INSERT_HEAD(&priv->rxqsctrl, tmpl, next);
return tmpl;
error:
+ mlx5_mr_btree_free(&tmpl->rxq.mr_ctrl.cache_bh);
mlx5_free(tmpl);
return NULL;
}
tmpl->rxq.mr_ctrl.cache_bh = (struct mlx5_mr_btree) { 0 };
tmpl->hairpin_conf = *hairpin_conf;
tmpl->rxq.idx = idx;
- __atomic_add_fetch(&tmpl->refcnt, 1, __ATOMIC_RELAXED);
+ __atomic_fetch_add(&tmpl->refcnt, 1, __ATOMIC_RELAXED);
LIST_INSERT_HEAD(&priv->rxqsctrl, tmpl, next);
return tmpl;
}
if (rxq_data) {
rxq_ctrl = container_of(rxq_data, struct mlx5_rxq_ctrl, rxq);
- __atomic_add_fetch(&rxq_ctrl->refcnt, 1, __ATOMIC_RELAXED);
+ __atomic_fetch_add(&rxq_ctrl->refcnt, 1, __ATOMIC_RELAXED);
}
return rxq_ctrl;
}
mlx5_free(rxq_ctrl->obj);
rxq_ctrl->obj = NULL;
}
- if (rxq_ctrl->type == MLX5_RXQ_TYPE_STANDARD)
+ if (rxq_ctrl->type == MLX5_RXQ_TYPE_STANDARD) {
rxq_free_elts(rxq_ctrl);
+ dev->data->rx_queue_state[idx] = RTE_ETH_QUEUE_STATE_STOPPED;
+ }
if (!__atomic_load_n(&rxq_ctrl->refcnt, __ATOMIC_RELAXED)) {
if (rxq_ctrl->type == MLX5_RXQ_TYPE_STANDARD)
mlx5_mr_btree_free(&rxq_ctrl->rxq.mr_ctrl.cache_bh);
return MLX5_RXQ_TYPE_UNDEFINED;
}
+/*
+ * Get a Rx hairpin queue configuration.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ * @param idx
+ * Rx queue index.
+ *
+ * @return
+ * Pointer to the configuration if a hairpin RX queue, otherwise NULL.
+ */
+const struct rte_eth_hairpin_conf *
+mlx5_rxq_get_hairpin_conf(struct rte_eth_dev *dev, uint16_t idx)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_rxq_ctrl *rxq_ctrl = NULL;
+
+ if (idx < priv->rxqs_n && (*priv->rxqs)[idx]) {
+ rxq_ctrl = container_of((*priv->rxqs)[idx],
+ struct mlx5_rxq_ctrl,
+ rxq);
+ if (rxq_ctrl->type == MLX5_RXQ_TYPE_HAIRPIN)
+ return &rxq_ctrl->hairpin_conf;
+ }
+ return NULL;
+}
+
/**
* Match queues listed in arguments to queues contained in indirection table
* object.
struct mlx5_priv *priv = dev->data->dev_private;
struct mlx5_ind_table_obj *ind_tbl;
+ rte_rwlock_read_lock(&priv->ind_tbls_lock);
LIST_FOREACH(ind_tbl, &priv->ind_tbls, next) {
if ((ind_tbl->queues_n == queues_n) &&
(memcmp(ind_tbl->queues, queues,
ind_tbl->queues_n * sizeof(ind_tbl->queues[0]))
- == 0))
+ == 0)) {
+ __atomic_fetch_add(&ind_tbl->refcnt, 1,
+ __ATOMIC_RELAXED);
break;
+ }
}
- if (ind_tbl) {
- unsigned int i;
-
- rte_atomic32_inc(&ind_tbl->refcnt);
- for (i = 0; i != ind_tbl->queues_n; ++i)
- mlx5_rxq_get(dev, ind_tbl->queues[i]);
- }
+ rte_rwlock_read_unlock(&priv->ind_tbls_lock);
return ind_tbl;
}
* Pointer to Ethernet device.
* @param ind_table
* Indirection table to release.
+ * @param standalone
+ * Indirection table for Standalone queue.
*
* @return
* 1 while a reference on it exists, 0 when freed.
*/
int
mlx5_ind_table_obj_release(struct rte_eth_dev *dev,
- struct mlx5_ind_table_obj *ind_tbl)
+ struct mlx5_ind_table_obj *ind_tbl,
+ bool standalone)
{
struct mlx5_priv *priv = dev->data->dev_private;
- unsigned int i;
+ unsigned int i, ret;
- if (rte_atomic32_dec_and_test(&ind_tbl->refcnt))
- priv->obj_ops.ind_table_destroy(ind_tbl);
+ rte_rwlock_write_lock(&priv->ind_tbls_lock);
+ ret = __atomic_sub_fetch(&ind_tbl->refcnt, 1, __ATOMIC_RELAXED);
+ if (!ret && !standalone)
+ LIST_REMOVE(ind_tbl, next);
+ rte_rwlock_write_unlock(&priv->ind_tbls_lock);
+ if (ret)
+ return 1;
+ priv->obj_ops.ind_table_destroy(ind_tbl);
for (i = 0; i != ind_tbl->queues_n; ++i)
claim_nonzero(mlx5_rxq_release(dev, ind_tbl->queues[i]));
- if (!rte_atomic32_read(&ind_tbl->refcnt)) {
- LIST_REMOVE(ind_tbl, next);
- mlx5_free(ind_tbl);
- return 0;
- }
- return 1;
+ mlx5_free(ind_tbl);
+ return 0;
}
/**
struct mlx5_ind_table_obj *ind_tbl;
int ret = 0;
+ rte_rwlock_read_lock(&priv->ind_tbls_lock);
LIST_FOREACH(ind_tbl, &priv->ind_tbls, next) {
DRV_LOG(DEBUG,
"port %u indirection table obj %p still referenced",
dev->data->port_id, (void *)ind_tbl);
++ret;
}
+ rte_rwlock_read_unlock(&priv->ind_tbls_lock);
+ return ret;
+}
+
+/**
+ * Setup an indirection table structure fields.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ * @param ind_table
+ * Indirection table to modify.
+ *
+ * @return
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
+ */
+int
+mlx5_ind_table_obj_setup(struct rte_eth_dev *dev,
+ struct mlx5_ind_table_obj *ind_tbl)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ uint32_t queues_n = ind_tbl->queues_n;
+ uint16_t *queues = ind_tbl->queues;
+ unsigned int i, j;
+ int ret = 0, err;
+ const unsigned int n = rte_is_power_of_2(queues_n) ?
+ log2above(queues_n) :
+ log2above(priv->config.ind_table_max_size);
+
+ for (i = 0; i != queues_n; ++i) {
+ if (!mlx5_rxq_get(dev, queues[i])) {
+ ret = -rte_errno;
+ goto error;
+ }
+ }
+ ret = priv->obj_ops.ind_table_new(dev, n, ind_tbl);
+ if (ret)
+ goto error;
+ __atomic_fetch_add(&ind_tbl->refcnt, 1, __ATOMIC_RELAXED);
+ return 0;
+error:
+ err = rte_errno;
+ for (j = 0; j < i; j++)
+ mlx5_rxq_release(dev, ind_tbl->queues[j]);
+ rte_errno = err;
+ DRV_LOG(DEBUG, "Port %u cannot setup indirection table.",
+ dev->data->port_id);
return ret;
}
* Queues entering in the indirection table.
* @param queues_n
* Number of queues in the array.
+ * @param standalone
+ * Indirection table for Standalone queue.
*
* @return
* The Verbs/DevX object initialized, NULL otherwise and rte_errno is set.
*/
static struct mlx5_ind_table_obj *
mlx5_ind_table_obj_new(struct rte_eth_dev *dev, const uint16_t *queues,
- uint32_t queues_n)
+ uint32_t queues_n, bool standalone)
{
struct mlx5_priv *priv = dev->data->dev_private;
struct mlx5_ind_table_obj *ind_tbl;
- const unsigned int n = rte_is_power_of_2(queues_n) ?
- log2above(queues_n) :
- log2above(priv->config.ind_table_max_size);
- unsigned int i, j;
int ret;
ind_tbl = mlx5_malloc(MLX5_MEM_ZERO, sizeof(*ind_tbl) +
return NULL;
}
ind_tbl->queues_n = queues_n;
- for (i = 0; i != queues_n; ++i) {
- struct mlx5_rxq_ctrl *rxq = mlx5_rxq_get(dev, queues[i]);
- if (!rxq)
- goto error;
- ind_tbl->queues[i] = queues[i];
+ ind_tbl->queues = (uint16_t *)(ind_tbl + 1);
+ memcpy(ind_tbl->queues, queues, queues_n * sizeof(*queues));
+ ret = mlx5_ind_table_obj_setup(dev, ind_tbl);
+ if (ret < 0) {
+ mlx5_free(ind_tbl);
+ return NULL;
+ }
+ if (!standalone) {
+ rte_rwlock_write_lock(&priv->ind_tbls_lock);
+ LIST_INSERT_HEAD(&priv->ind_tbls, ind_tbl, next);
+ rte_rwlock_write_unlock(&priv->ind_tbls_lock);
}
- ret = priv->obj_ops.ind_table_new(dev, n, ind_tbl);
- if (ret < 0)
- goto error;
- rte_atomic32_inc(&ind_tbl->refcnt);
- LIST_INSERT_HEAD(&priv->ind_tbls, ind_tbl, next);
return ind_tbl;
-error:
- ret = rte_errno;
- for (j = 0; j < i; j++)
- mlx5_rxq_release(dev, ind_tbl->queues[j]);
- rte_errno = ret;
- mlx5_free(ind_tbl);
- DEBUG("Port %u cannot create indirection table.", dev->data->port_id);
- return NULL;
}
/**
- * Get an Rx Hash queue.
+ * Modify an indirection table.
*
* @param dev
* Pointer to Ethernet device.
- * @param rss_conf
- * RSS configuration for the Rx hash queue.
+ * @param ind_table
+ * Indirection table to modify.
* @param queues
- * Queues entering in hash queue. In case of empty hash_fields only the
- * first queue index will be taken for the indirection table.
+ * Queues replacement for the indirection table.
* @param queues_n
- * Number of queues.
+ * Number of queues in the array.
+ * @param standalone
+ * Indirection table for Standalone queue.
*
* @return
- * An hash Rx queue index on success.
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
*/
-uint32_t
-mlx5_hrxq_get(struct rte_eth_dev *dev,
- const uint8_t *rss_key, uint32_t rss_key_len,
- uint64_t hash_fields,
- const uint16_t *queues, uint32_t queues_n)
+int
+mlx5_ind_table_obj_modify(struct rte_eth_dev *dev,
+ struct mlx5_ind_table_obj *ind_tbl,
+ uint16_t *queues, const uint32_t queues_n,
+ bool standalone)
{
struct mlx5_priv *priv = dev->data->dev_private;
- struct mlx5_hrxq *hrxq;
- uint32_t idx;
-
- queues_n = hash_fields ? queues_n : 1;
- ILIST_FOREACH(priv->sh->ipool[MLX5_IPOOL_HRXQ], priv->hrxqs, idx,
- hrxq, next) {
- struct mlx5_ind_table_obj *ind_tbl;
+ unsigned int i, j;
+ int ret = 0, err;
+ const unsigned int n = rte_is_power_of_2(queues_n) ?
+ log2above(queues_n) :
+ log2above(priv->config.ind_table_max_size);
- if (hrxq->shared)
- continue;
- if (hrxq->rss_key_len != rss_key_len)
- continue;
- if (memcmp(hrxq->rss_key, rss_key, rss_key_len))
- continue;
- if (hrxq->hash_fields != hash_fields)
- continue;
- ind_tbl = mlx5_ind_table_obj_get(dev, queues, queues_n);
- if (!ind_tbl)
- continue;
- if (ind_tbl != hrxq->ind_table) {
- mlx5_ind_table_obj_release(dev, ind_tbl);
- continue;
+ MLX5_ASSERT(standalone);
+ RTE_SET_USED(standalone);
+ if (__atomic_load_n(&ind_tbl->refcnt, __ATOMIC_RELAXED) > 1) {
+ /*
+ * Modification of indirection ntables having more than 1
+ * reference unsupported. Intended for standalone indirection
+ * tables only.
+ */
+ DRV_LOG(DEBUG,
+ "Port %u cannot modify indirection table (refcnt> 1).",
+ dev->data->port_id);
+ rte_errno = EINVAL;
+ return -rte_errno;
+ }
+ for (i = 0; i != queues_n; ++i) {
+ if (!mlx5_rxq_get(dev, queues[i])) {
+ ret = -rte_errno;
+ goto error;
}
- rte_atomic32_inc(&hrxq->refcnt);
- return idx;
}
+ MLX5_ASSERT(priv->obj_ops.ind_table_modify);
+ ret = priv->obj_ops.ind_table_modify(dev, n, queues, queues_n, ind_tbl);
+ if (ret)
+ goto error;
+ for (j = 0; j < ind_tbl->queues_n; j++)
+ mlx5_rxq_release(dev, ind_tbl->queues[j]);
+ ind_tbl->queues_n = queues_n;
+ ind_tbl->queues = queues;
return 0;
+error:
+ err = rte_errno;
+ for (j = 0; j < i; j++)
+ mlx5_rxq_release(dev, ind_tbl->queues[j]);
+ rte_errno = err;
+ DRV_LOG(DEBUG, "Port %u cannot setup indirection table.",
+ dev->data->port_id);
+ return ret;
+}
+
+int
+mlx5_hrxq_match_cb(void *tool_ctx __rte_unused, struct mlx5_list_entry *entry,
+ void *cb_ctx)
+{
+ struct mlx5_flow_cb_ctx *ctx = cb_ctx;
+ struct mlx5_flow_rss_desc *rss_desc = ctx->data;
+ struct mlx5_hrxq *hrxq = container_of(entry, typeof(*hrxq), entry);
+
+ return (hrxq->rss_key_len != rss_desc->key_len ||
+ memcmp(hrxq->rss_key, rss_desc->key, rss_desc->key_len) ||
+ hrxq->hash_fields != rss_desc->hash_fields ||
+ hrxq->ind_table->queues_n != rss_desc->queue_num ||
+ memcmp(hrxq->ind_table->queues, rss_desc->queue,
+ rss_desc->queue_num * sizeof(rss_desc->queue[0])));
}
/**
queues, queues_n)) {
ind_tbl = hrxq->ind_table;
} else {
+ if (hrxq->standalone) {
+ /*
+ * Replacement of indirection table unsupported for
+ * stanalone hrxq objects (used by shared RSS).
+ */
+ rte_errno = ENOTSUP;
+ return -rte_errno;
+ }
ind_tbl = mlx5_ind_table_obj_get(dev, queues, queues_n);
if (!ind_tbl)
- ind_tbl = mlx5_ind_table_obj_new(dev, queues, queues_n);
+ ind_tbl = mlx5_ind_table_obj_new(dev, queues, queues_n,
+ hrxq->standalone);
}
if (!ind_tbl) {
rte_errno = ENOMEM;
goto error;
}
if (ind_tbl != hrxq->ind_table) {
- mlx5_ind_table_obj_release(dev, hrxq->ind_table);
+ MLX5_ASSERT(!hrxq->standalone);
+ mlx5_ind_table_obj_release(dev, hrxq->ind_table,
+ hrxq->standalone);
hrxq->ind_table = ind_tbl;
}
hrxq->hash_fields = hash_fields;
return 0;
error:
err = rte_errno;
- if (ind_tbl != hrxq->ind_table)
- mlx5_ind_table_obj_release(dev, ind_tbl);
+ if (ind_tbl != hrxq->ind_table) {
+ MLX5_ASSERT(!hrxq->standalone);
+ mlx5_ind_table_obj_release(dev, ind_tbl, hrxq->standalone);
+ }
rte_errno = err;
return -rte_errno;
}
-/**
- * Release the hash Rx queue.
- *
- * @param dev
- * Pointer to Ethernet device.
- * @param hrxq
- * Index to Hash Rx queue to release.
- *
- * @return
- * 1 while a reference on it exists, 0 when freed.
- */
-int
-mlx5_hrxq_release(struct rte_eth_dev *dev, uint32_t hrxq_idx)
+static void
+__mlx5_hrxq_remove(struct rte_eth_dev *dev, struct mlx5_hrxq *hrxq)
{
struct mlx5_priv *priv = dev->data->dev_private;
- struct mlx5_hrxq *hrxq;
- hrxq = mlx5_ipool_get(priv->sh->ipool[MLX5_IPOOL_HRXQ], hrxq_idx);
- if (!hrxq)
- return 0;
- if (rte_atomic32_dec_and_test(&hrxq->refcnt)) {
#ifdef HAVE_IBV_FLOW_DV_SUPPORT
- mlx5_glue->destroy_flow_action(hrxq->action);
+ mlx5_glue->destroy_flow_action(hrxq->action);
#endif
- priv->obj_ops.hrxq_destroy(hrxq);
- mlx5_ind_table_obj_release(dev, hrxq->ind_table);
- ILIST_REMOVE(priv->sh->ipool[MLX5_IPOOL_HRXQ], &priv->hrxqs,
- hrxq_idx, hrxq, next);
- mlx5_ipool_free(priv->sh->ipool[MLX5_IPOOL_HRXQ], hrxq_idx);
- return 0;
+ priv->obj_ops.hrxq_destroy(hrxq);
+ if (!hrxq->standalone) {
+ mlx5_ind_table_obj_release(dev, hrxq->ind_table,
+ hrxq->standalone);
}
- claim_nonzero(mlx5_ind_table_obj_release(dev, hrxq->ind_table));
- return 1;
+ mlx5_ipool_free(priv->sh->ipool[MLX5_IPOOL_HRXQ], hrxq->idx);
}
/**
- * Create an Rx Hash queue.
+ * Release the hash Rx queue.
*
* @param dev
* Pointer to Ethernet device.
- * @param rss_key
- * RSS key for the Rx hash queue.
- * @param rss_key_len
- * RSS key length.
- * @param hash_fields
- * Verbs protocol hash field to make the RSS on.
- * @param queues
- * Queues entering in hash queue. In case of empty hash_fields only the
- * first queue index will be taken for the indirection table.
- * @param queues_n
- * Number of queues.
- * @param tunnel
- * Tunnel type.
- * @param shared
- * If true new object of Rx Hash queue will be used in shared action.
+ * @param hrxq
+ * Index to Hash Rx queue to release.
*
- * @return
- * The DevX object initialized index, 0 otherwise and rte_errno is set.
+ * @param list
+ * mlx5 list pointer.
+ * @param entry
+ * Hash queue entry pointer.
*/
-uint32_t
-mlx5_hrxq_new(struct rte_eth_dev *dev,
- const uint8_t *rss_key, uint32_t rss_key_len,
- uint64_t hash_fields,
- const uint16_t *queues, uint32_t queues_n,
- int tunnel, bool shared)
+void
+mlx5_hrxq_remove_cb(void *tool_ctx, struct mlx5_list_entry *entry)
+{
+ struct rte_eth_dev *dev = tool_ctx;
+ struct mlx5_hrxq *hrxq = container_of(entry, typeof(*hrxq), entry);
+
+ __mlx5_hrxq_remove(dev, hrxq);
+}
+
+static struct mlx5_hrxq *
+__mlx5_hrxq_create(struct rte_eth_dev *dev,
+ struct mlx5_flow_rss_desc *rss_desc)
{
struct mlx5_priv *priv = dev->data->dev_private;
+ const uint8_t *rss_key = rss_desc->key;
+ uint32_t rss_key_len = rss_desc->key_len;
+ bool standalone = !!rss_desc->shared_rss;
+ const uint16_t *queues =
+ standalone ? rss_desc->const_q : rss_desc->queue;
+ uint32_t queues_n = rss_desc->queue_num;
struct mlx5_hrxq *hrxq = NULL;
uint32_t hrxq_idx = 0;
- struct mlx5_ind_table_obj *ind_tbl;
+ struct mlx5_ind_table_obj *ind_tbl = rss_desc->ind_tbl;
int ret;
- queues_n = hash_fields ? queues_n : 1;
- ind_tbl = mlx5_ind_table_obj_get(dev, queues, queues_n);
+ queues_n = rss_desc->hash_fields ? queues_n : 1;
if (!ind_tbl)
- ind_tbl = mlx5_ind_table_obj_new(dev, queues, queues_n);
- if (!ind_tbl) {
- rte_errno = ENOMEM;
- return 0;
- }
+ ind_tbl = mlx5_ind_table_obj_get(dev, queues, queues_n);
+ if (!ind_tbl)
+ ind_tbl = mlx5_ind_table_obj_new(dev, queues, queues_n,
+ standalone);
+ if (!ind_tbl)
+ return NULL;
hrxq = mlx5_ipool_zmalloc(priv->sh->ipool[MLX5_IPOOL_HRXQ], &hrxq_idx);
if (!hrxq)
goto error;
- hrxq->shared = !!shared;
+ hrxq->standalone = standalone;
+ hrxq->idx = hrxq_idx;
hrxq->ind_table = ind_tbl;
hrxq->rss_key_len = rss_key_len;
- hrxq->hash_fields = hash_fields;
+ hrxq->hash_fields = rss_desc->hash_fields;
memcpy(hrxq->rss_key, rss_key, rss_key_len);
- ret = priv->obj_ops.hrxq_new(dev, hrxq, tunnel);
- if (ret < 0) {
- rte_errno = errno;
+ ret = priv->obj_ops.hrxq_new(dev, hrxq, rss_desc->tunnel);
+ if (ret < 0)
goto error;
- }
- rte_atomic32_inc(&hrxq->refcnt);
- ILIST_INSERT(priv->sh->ipool[MLX5_IPOOL_HRXQ], &priv->hrxqs, hrxq_idx,
- hrxq, next);
- return hrxq_idx;
+ return hrxq;
error:
- ret = rte_errno; /* Save rte_errno before cleanup. */
- mlx5_ind_table_obj_release(dev, ind_tbl);
+ if (!rss_desc->ind_tbl)
+ mlx5_ind_table_obj_release(dev, ind_tbl, standalone);
if (hrxq)
mlx5_ipool_free(priv->sh->ipool[MLX5_IPOOL_HRXQ], hrxq_idx);
- rte_errno = ret; /* Restore rte_errno. */
+ return NULL;
+}
+
+struct mlx5_list_entry *
+mlx5_hrxq_create_cb(void *tool_ctx, void *cb_ctx)
+{
+ struct rte_eth_dev *dev = tool_ctx;
+ struct mlx5_flow_cb_ctx *ctx = cb_ctx;
+ struct mlx5_flow_rss_desc *rss_desc = ctx->data;
+ struct mlx5_hrxq *hrxq;
+
+ hrxq = __mlx5_hrxq_create(dev, rss_desc);
+ return hrxq ? &hrxq->entry : NULL;
+}
+
+struct mlx5_list_entry *
+mlx5_hrxq_clone_cb(void *tool_ctx, struct mlx5_list_entry *entry,
+ void *cb_ctx __rte_unused)
+{
+ struct rte_eth_dev *dev = tool_ctx;
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_hrxq *hrxq;
+ uint32_t hrxq_idx = 0;
+
+ hrxq = mlx5_ipool_zmalloc(priv->sh->ipool[MLX5_IPOOL_HRXQ], &hrxq_idx);
+ if (!hrxq)
+ return NULL;
+ memcpy(hrxq, entry, sizeof(*hrxq) + MLX5_RSS_HASH_KEY_LEN);
+ hrxq->idx = hrxq_idx;
+ return &hrxq->entry;
+}
+
+void
+mlx5_hrxq_clone_free_cb(void *tool_ctx, struct mlx5_list_entry *entry)
+{
+ struct rte_eth_dev *dev = tool_ctx;
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_hrxq *hrxq = container_of(entry, typeof(*hrxq), entry);
+
+ mlx5_ipool_free(priv->sh->ipool[MLX5_IPOOL_HRXQ], hrxq->idx);
+}
+
+/**
+ * Get an Rx Hash queue.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ * @param rss_desc
+ * RSS configuration for the Rx hash queue.
+ *
+ * @return
+ * An hash Rx queue index on success.
+ */
+uint32_t mlx5_hrxq_get(struct rte_eth_dev *dev,
+ struct mlx5_flow_rss_desc *rss_desc)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_hrxq *hrxq;
+ struct mlx5_list_entry *entry;
+ struct mlx5_flow_cb_ctx ctx = {
+ .data = rss_desc,
+ };
+
+ if (rss_desc->shared_rss) {
+ hrxq = __mlx5_hrxq_create(dev, rss_desc);
+ } else {
+ entry = mlx5_list_register(priv->hrxqs, &ctx);
+ if (!entry)
+ return 0;
+ hrxq = container_of(entry, typeof(*hrxq), entry);
+ }
+ if (hrxq)
+ return hrxq->idx;
+ return 0;
+}
+
+/**
+ * Release the hash Rx queue.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ * @param hrxq_idx
+ * Index to Hash Rx queue to release.
+ *
+ * @return
+ * 1 while a reference on it exists, 0 when freed.
+ */
+int mlx5_hrxq_release(struct rte_eth_dev *dev, uint32_t hrxq_idx)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_hrxq *hrxq;
+
+ hrxq = mlx5_ipool_get(priv->sh->ipool[MLX5_IPOOL_HRXQ], hrxq_idx);
+ if (!hrxq)
+ return 0;
+ if (!hrxq->standalone)
+ return mlx5_list_unregister(priv->hrxqs, &hrxq->entry);
+ __mlx5_hrxq_remove(dev, hrxq);
return 0;
}
struct mlx5_hrxq *hrxq = NULL;
int ret;
- if (priv->drop_queue.hrxq) {
- rte_atomic32_inc(&priv->drop_queue.hrxq->refcnt);
+ if (priv->drop_queue.hrxq)
return priv->drop_queue.hrxq;
- }
hrxq = mlx5_malloc(MLX5_MEM_ZERO, sizeof(*hrxq), 0, SOCKET_ID_ANY);
if (!hrxq) {
DRV_LOG(WARNING,
ret = priv->obj_ops.drop_action_create(dev);
if (ret < 0)
goto error;
- rte_atomic32_set(&hrxq->refcnt, 1);
return hrxq;
error:
if (hrxq) {
struct mlx5_priv *priv = dev->data->dev_private;
struct mlx5_hrxq *hrxq = priv->drop_queue.hrxq;
- if (rte_atomic32_dec_and_test(&hrxq->refcnt)) {
- priv->obj_ops.drop_action_destroy(dev);
- mlx5_free(priv->drop_queue.rxq);
- mlx5_free(hrxq->ind_table);
- mlx5_free(hrxq);
- priv->drop_queue.rxq = NULL;
- priv->drop_queue.hrxq = NULL;
- }
+ if (!priv->drop_queue.hrxq)
+ return;
+ priv->obj_ops.drop_action_destroy(dev);
+ mlx5_free(priv->drop_queue.rxq);
+ mlx5_free(hrxq->ind_table);
+ mlx5_free(hrxq);
+ priv->drop_queue.rxq = NULL;
+ priv->drop_queue.hrxq = NULL;
}
/**
* @return
* The number of object not released.
*/
-int
+uint32_t
mlx5_hrxq_verify(struct rte_eth_dev *dev)
{
struct mlx5_priv *priv = dev->data->dev_private;
- struct mlx5_hrxq *hrxq;
- uint32_t idx;
- int ret = 0;
- ILIST_FOREACH(priv->sh->ipool[MLX5_IPOOL_HRXQ], priv->hrxqs, idx,
- hrxq, next) {
- DRV_LOG(DEBUG,
- "port %u hash Rx queue %p still referenced",
- dev->data->port_id, (void *)hrxq);
- ++ret;
- }
- return ret;
+ return mlx5_list_get_entry_num(priv->hrxqs);
}
/**