* [PATCH 0/4] patches for bnxt
@ 2017-10-10 14:22 Ajit Khaparde
2017-10-10 14:23 ` [PATCH 1/4] net/bnxt: fix cleanup if a filter allocation fails Ajit Khaparde
` (4 more replies)
0 siblings, 5 replies; 6+ messages in thread
From: Ajit Khaparde @ 2017-10-10 14:22 UTC (permalink / raw)
To: dev; +Cc: ferruh.yigit
Patches against dpdk-next-net.
Please apply.
Ajit Khaparde (4):
net/bnxt: fix cleanup if a filter allocation fails
net/bnxt: prevent programming a duplicate flow
net/bnxt: include fdir support
net/bnxt: fix the association of a MACVLAN per VNIC
drivers/net/bnxt/bnxt_ethdev.c | 384 ++++++++++++++++++++++++++++++++++++++++-
drivers/net/bnxt/bnxt_filter.c | 86 ++++++++-
drivers/net/bnxt/bnxt_filter.h | 2 +
drivers/net/bnxt/bnxt_hwrm.c | 22 +++
drivers/net/bnxt/bnxt_rxq.c | 2 +-
5 files changed, 488 insertions(+), 8 deletions(-)
--
2.13.5 (Apple Git-94)
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH 1/4] net/bnxt: fix cleanup if a filter allocation fails
2017-10-10 14:22 [PATCH 0/4] patches for bnxt Ajit Khaparde
@ 2017-10-10 14:23 ` Ajit Khaparde
2017-10-10 14:23 ` [PATCH 2/4] net/bnxt: prevent programming a duplicate flow Ajit Khaparde
` (3 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: Ajit Khaparde @ 2017-10-10 14:23 UTC (permalink / raw)
To: dev; +Cc: ferruh.yigit
We are not checking if a filter allocation succeeded.
And we end up accessing a null pointer after that.
Also invalidate the fw_l2_filter_id to prevent unnecessary
HW access and hence HWRM command failures during exit.
Signed-off-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
---
drivers/net/bnxt/bnxt_filter.c | 14 ++++++++++++++
1 file changed, 14 insertions(+)
diff --git a/drivers/net/bnxt/bnxt_filter.c b/drivers/net/bnxt/bnxt_filter.c
index 9f171fd99..4c1db559d 100644
--- a/drivers/net/bnxt/bnxt_filter.c
+++ b/drivers/net/bnxt/bnxt_filter.c
@@ -828,12 +828,20 @@ bnxt_validate_and_parse_flow(struct rte_eth_dev *dev,
}
filter->dst_id = vnic->fw_vnic_id;
filter1 = bnxt_get_l2_filter(bp, filter, vnic);
+ if (filter1 == NULL) {
+ rc = -ENOSPC;
+ goto ret;
+ }
filter->fw_l2_filter_id = filter1->fw_l2_filter_id;
RTE_LOG(DEBUG, PMD, "VNIC found\n");
break;
case RTE_FLOW_ACTION_TYPE_DROP:
vnic0 = STAILQ_FIRST(&bp->ff_pool[0]);
filter1 = bnxt_get_l2_filter(bp, filter, vnic0);
+ if (filter1 == NULL) {
+ rc = -ENOSPC;
+ goto ret;
+ }
filter->fw_l2_filter_id = filter1->fw_l2_filter_id;
if (filter->filter_type == HWRM_CFA_EM_FILTER)
filter->flags =
@@ -845,6 +853,10 @@ bnxt_validate_and_parse_flow(struct rte_eth_dev *dev,
case RTE_FLOW_ACTION_TYPE_COUNT:
vnic0 = STAILQ_FIRST(&bp->ff_pool[0]);
filter1 = bnxt_get_l2_filter(bp, filter, vnic0);
+ if (filter1 == NULL) {
+ rc = -ENOSPC;
+ goto ret;
+ }
filter->fw_l2_filter_id = filter1->fw_l2_filter_id;
filter->flags = HWRM_CFA_NTUPLE_FILTER_ALLOC_INPUT_FLAGS_METER;
break;
@@ -892,6 +904,7 @@ bnxt_flow_validate(struct rte_eth_dev *dev,
ret = bnxt_validate_and_parse_flow(dev, pattern, actions, attr,
error, filter);
/* No need to hold on to this filter if we are just validating flow */
+ filter->fw_l2_filter_id = -1;
bnxt_free_filter(bp, filter);
return ret;
@@ -961,6 +974,7 @@ bnxt_flow_create(struct rte_eth_dev *dev,
return flow;
}
free_filter:
+ filter->fw_l2_filter_id = -1;
bnxt_free_filter(bp, filter);
free_flow:
RTE_LOG(ERR, PMD, "Failed to create flow.\n");
--
2.13.5 (Apple Git-94)
^ permalink raw reply related [flat|nested] 6+ messages in thread
* [PATCH 2/4] net/bnxt: prevent programming a duplicate flow
2017-10-10 14:22 [PATCH 0/4] patches for bnxt Ajit Khaparde
2017-10-10 14:23 ` [PATCH 1/4] net/bnxt: fix cleanup if a filter allocation fails Ajit Khaparde
@ 2017-10-10 14:23 ` Ajit Khaparde
2017-10-10 14:23 ` [PATCH 3/4] net/bnxt: include fdir support Ajit Khaparde
` (2 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: Ajit Khaparde @ 2017-10-10 14:23 UTC (permalink / raw)
To: dev; +Cc: ferruh.yigit
Match a flow against existing flows programmed in the HW
and prevent overlapping entries. Also change log level of
some logs to DEBUG.
Signed-off-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
---
drivers/net/bnxt/bnxt_filter.c | 72 +++++++++++++++++++++++++++++++++++++++---
1 file changed, 67 insertions(+), 5 deletions(-)
diff --git a/drivers/net/bnxt/bnxt_filter.c b/drivers/net/bnxt/bnxt_filter.c
index 4c1db559d..029a4eb4b 100644
--- a/drivers/net/bnxt/bnxt_filter.c
+++ b/drivers/net/bnxt/bnxt_filter.c
@@ -325,7 +325,7 @@ bnxt_validate_and_parse_flow_type(const struct rte_flow_item pattern[],
uint32_t en = 0;
use_ntuple = bnxt_filter_type_check(pattern, error);
- RTE_LOG(ERR, PMD, "Use NTUPLE %d\n", use_ntuple);
+ RTE_LOG(DEBUG, PMD, "Use NTUPLE %d\n", use_ntuple);
if (use_ntuple < 0)
return use_ntuple;
@@ -815,7 +815,7 @@ bnxt_validate_and_parse_flow(struct rte_eth_dev *dev,
rc = -rte_errno;
goto ret;
}
- RTE_LOG(ERR, PMD, "Queue index %d\n", act_q->index);
+ RTE_LOG(DEBUG, PMD, "Queue index %d\n", act_q->index);
vnic0 = STAILQ_FIRST(&bp->ff_pool[0]);
vnic = STAILQ_FIRST(&bp->ff_pool[act_q->index]);
@@ -910,6 +910,55 @@ bnxt_flow_validate(struct rte_eth_dev *dev,
return ret;
}
+static int
+bnxt_match_filter(struct bnxt *bp, struct bnxt_filter_info *nf)
+{
+ struct bnxt_filter_info *mf;
+ struct rte_flow *flow;
+ int i;
+
+ for (i = bp->nr_vnics - 1; i >= 0; i--) {
+ struct bnxt_vnic_info *vnic = &bp->vnic_info[i];
+
+ STAILQ_FOREACH(flow, &vnic->flow_list, next) {
+ mf = flow->filter;
+
+ if (mf->filter_type == nf->filter_type &&
+ mf->flags == nf->flags &&
+ mf->src_port == nf->src_port &&
+ mf->src_port_mask == nf->src_port_mask &&
+ mf->dst_port == nf->dst_port &&
+ mf->dst_port_mask == nf->dst_port_mask &&
+ mf->ip_protocol == nf->ip_protocol &&
+ mf->ip_addr_type == nf->ip_addr_type &&
+ mf->ethertype == nf->ethertype &&
+ mf->vni == nf->vni &&
+ mf->tunnel_type == nf->tunnel_type &&
+ mf->l2_ovlan == nf->l2_ovlan &&
+ mf->l2_ovlan_mask == nf->l2_ovlan_mask &&
+ mf->l2_ivlan == nf->l2_ivlan &&
+ mf->l2_ivlan_mask == nf->l2_ivlan_mask &&
+ !memcmp(mf->l2_addr, nf->l2_addr, ETHER_ADDR_LEN) &&
+ !memcmp(mf->l2_addr_mask, nf->l2_addr_mask,
+ ETHER_ADDR_LEN) &&
+ !memcmp(mf->src_macaddr, nf->src_macaddr,
+ ETHER_ADDR_LEN) &&
+ !memcmp(mf->dst_macaddr, nf->dst_macaddr,
+ ETHER_ADDR_LEN) &&
+ !memcmp(mf->src_ipaddr, nf->src_ipaddr,
+ sizeof(nf->src_ipaddr)) &&
+ !memcmp(mf->src_ipaddr_mask, nf->src_ipaddr_mask,
+ sizeof(nf->src_ipaddr_mask)) &&
+ !memcmp(mf->dst_ipaddr, nf->dst_ipaddr,
+ sizeof(nf->dst_ipaddr)) &&
+ !memcmp(mf->dst_ipaddr_mask, nf->dst_ipaddr_mask,
+ sizeof(nf->dst_ipaddr_mask)))
+ return -EEXIST;
+ }
+ }
+ return 0;
+}
+
static struct rte_flow *
bnxt_flow_create(struct rte_eth_dev *dev,
const struct rte_flow_attr *attr,
@@ -949,6 +998,12 @@ bnxt_flow_create(struct rte_eth_dev *dev,
if (ret != 0)
goto free_filter;
+ ret = bnxt_match_filter(bp, filter);
+ if (ret != 0) {
+ RTE_LOG(DEBUG, PMD, "Flow already exists.\n");
+ goto free_filter;
+ }
+
if (filter->filter_type == HWRM_CFA_EM_FILTER) {
filter->enables |=
HWRM_CFA_EM_FLOW_ALLOC_INPUT_ENABLES_L2_FILTER_ID;
@@ -977,9 +1032,13 @@ bnxt_flow_create(struct rte_eth_dev *dev,
filter->fw_l2_filter_id = -1;
bnxt_free_filter(bp, filter);
free_flow:
- RTE_LOG(ERR, PMD, "Failed to create flow.\n");
- rte_flow_error_set(error, -ret,
- RTE_FLOW_ERROR_TYPE_HANDLE, NULL,
+ if (ret == -EEXIST)
+ rte_flow_error_set(error, ret,
+ RTE_FLOW_ERROR_TYPE_HANDLE, NULL,
+ "Matching Flow exists.");
+ else
+ rte_flow_error_set(error, -ret,
+ RTE_FLOW_ERROR_TYPE_HANDLE, NULL,
"Failed to create flow.");
rte_free(flow);
flow = NULL;
@@ -996,6 +1055,9 @@ bnxt_flow_destroy(struct rte_eth_dev *dev,
struct bnxt_vnic_info *vnic = flow->vnic;
int ret = 0;
+ ret = bnxt_match_filter(bp, filter);
+ if (ret == 0)
+ RTE_LOG(ERR, PMD, "Could not find matching flow\n");
if (filter->filter_type == HWRM_CFA_EM_FILTER)
ret = bnxt_hwrm_clear_em_filter(bp, filter);
if (filter->filter_type == HWRM_CFA_NTUPLE_FILTER)
--
2.13.5 (Apple Git-94)
^ permalink raw reply related [flat|nested] 6+ messages in thread
* [PATCH 3/4] net/bnxt: include fdir support
2017-10-10 14:22 [PATCH 0/4] patches for bnxt Ajit Khaparde
2017-10-10 14:23 ` [PATCH 1/4] net/bnxt: fix cleanup if a filter allocation fails Ajit Khaparde
2017-10-10 14:23 ` [PATCH 2/4] net/bnxt: prevent programming a duplicate flow Ajit Khaparde
@ 2017-10-10 14:23 ` Ajit Khaparde
2017-10-10 14:23 ` [PATCH 4/4] net/bnxt: fix the association of a MACVLAN per VNIC Ajit Khaparde
2017-10-11 2:08 ` [PATCH 0/4] patches for bnxt Ferruh Yigit
4 siblings, 0 replies; 6+ messages in thread
From: Ajit Khaparde @ 2017-10-10 14:23 UTC (permalink / raw)
To: dev; +Cc: ferruh.yigit
This patch brings support for Flow Director.
Signed-off-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
---
drivers/net/bnxt/bnxt_ethdev.c | 384 ++++++++++++++++++++++++++++++++++++++++-
drivers/net/bnxt/bnxt_filter.h | 2 +
2 files changed, 384 insertions(+), 2 deletions(-)
diff --git a/drivers/net/bnxt/bnxt_ethdev.c b/drivers/net/bnxt/bnxt_ethdev.c
index 97810f6e5..d3da30189 100644
--- a/drivers/net/bnxt/bnxt_ethdev.c
+++ b/drivers/net/bnxt/bnxt_ethdev.c
@@ -2075,6 +2075,385 @@ bnxt_ntuple_filter(struct rte_eth_dev *dev,
}
return ret;
}
+
+static int
+bnxt_parse_fdir_filter(struct bnxt *bp,
+ struct rte_eth_fdir_filter *fdir,
+ struct bnxt_filter_info *filter)
+{
+ enum rte_fdir_mode fdir_mode =
+ bp->eth_dev->data->dev_conf.fdir_conf.mode;
+ struct bnxt_vnic_info *vnic0, *vnic;
+ struct bnxt_filter_info *filter1;
+ uint32_t en = 0;
+ int i;
+
+ if (fdir_mode == RTE_FDIR_MODE_PERFECT_TUNNEL)
+ return -EINVAL;
+
+ filter->l2_ovlan = fdir->input.flow_ext.vlan_tci;
+ en |= EM_FLOW_ALLOC_INPUT_EN_OVLAN_VID;
+
+ switch (fdir->input.flow_type) {
+ case RTE_ETH_FLOW_IPV4:
+ case RTE_ETH_FLOW_NONFRAG_IPV4_OTHER:
+ /* FALLTHROUGH */
+ filter->src_ipaddr[0] = fdir->input.flow.ip4_flow.src_ip;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_IPADDR;
+ filter->dst_ipaddr[0] = fdir->input.flow.ip4_flow.dst_ip;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_IPADDR;
+ filter->ip_protocol = fdir->input.flow.ip4_flow.proto;
+ en |= NTUPLE_FLTR_ALLOC_IN_EN_IP_PROTO;
+ filter->ip_addr_type =
+ NTUPLE_FLTR_ALLOC_INPUT_IP_ADDR_TYPE_IPV4;
+ filter->src_ipaddr_mask[0] = 0xffffffff;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_IPADDR_MASK;
+ filter->dst_ipaddr_mask[0] = 0xffffffff;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_IPADDR_MASK;
+ filter->ethertype = 0x800;
+ filter->enables |= NTUPLE_FLTR_ALLOC_INPUT_EN_ETHERTYPE;
+ break;
+ case RTE_ETH_FLOW_NONFRAG_IPV4_TCP:
+ filter->src_port = fdir->input.flow.tcp4_flow.src_port;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_PORT;
+ filter->dst_port = fdir->input.flow.tcp4_flow.dst_port;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_PORT;
+ filter->dst_port_mask = 0xffff;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_PORT_MASK;
+ filter->src_port_mask = 0xffff;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_PORT_MASK;
+ filter->src_ipaddr[0] = fdir->input.flow.tcp4_flow.ip.src_ip;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_IPADDR;
+ filter->dst_ipaddr[0] = fdir->input.flow.tcp4_flow.ip.dst_ip;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_IPADDR;
+ filter->ip_protocol = 6;
+ en |= NTUPLE_FLTR_ALLOC_IN_EN_IP_PROTO;
+ filter->ip_addr_type =
+ NTUPLE_FLTR_ALLOC_INPUT_IP_ADDR_TYPE_IPV4;
+ filter->src_ipaddr_mask[0] = 0xffffffff;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_IPADDR_MASK;
+ filter->dst_ipaddr_mask[0] = 0xffffffff;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_IPADDR_MASK;
+ filter->ethertype = 0x800;
+ filter->enables |= NTUPLE_FLTR_ALLOC_INPUT_EN_ETHERTYPE;
+ break;
+ case RTE_ETH_FLOW_NONFRAG_IPV4_UDP:
+ filter->src_port = fdir->input.flow.udp4_flow.src_port;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_PORT;
+ filter->dst_port = fdir->input.flow.udp4_flow.dst_port;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_PORT;
+ filter->dst_port_mask = 0xffff;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_PORT_MASK;
+ filter->src_port_mask = 0xffff;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_PORT_MASK;
+ filter->src_ipaddr[0] = fdir->input.flow.udp4_flow.ip.src_ip;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_IPADDR;
+ filter->dst_ipaddr[0] = fdir->input.flow.udp4_flow.ip.dst_ip;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_IPADDR;
+ filter->ip_protocol = 17;
+ en |= NTUPLE_FLTR_ALLOC_IN_EN_IP_PROTO;
+ filter->ip_addr_type =
+ NTUPLE_FLTR_ALLOC_INPUT_IP_ADDR_TYPE_IPV4;
+ filter->src_ipaddr_mask[0] = 0xffffffff;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_IPADDR_MASK;
+ filter->dst_ipaddr_mask[0] = 0xffffffff;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_IPADDR_MASK;
+ filter->ethertype = 0x800;
+ filter->enables |= NTUPLE_FLTR_ALLOC_INPUT_EN_ETHERTYPE;
+ break;
+ case RTE_ETH_FLOW_IPV6:
+ case RTE_ETH_FLOW_NONFRAG_IPV6_OTHER:
+ /* FALLTHROUGH */
+ filter->ip_addr_type =
+ NTUPLE_FLTR_ALLOC_INPUT_IP_ADDR_TYPE_IPV6;
+ filter->ip_protocol = fdir->input.flow.ipv6_flow.proto;
+ en |= NTUPLE_FLTR_ALLOC_IN_EN_IP_PROTO;
+ rte_memcpy(filter->src_ipaddr,
+ fdir->input.flow.ipv6_flow.src_ip, 16);
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_IPADDR;
+ rte_memcpy(filter->dst_ipaddr,
+ fdir->input.flow.ipv6_flow.dst_ip, 16);
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_IPADDR;
+ memset(filter->dst_ipaddr_mask, 0xff, 16);
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_IPADDR_MASK;
+ memset(filter->src_ipaddr_mask, 0xff, 16);
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_IPADDR_MASK;
+ filter->ethertype = 0x86dd;
+ filter->enables |= NTUPLE_FLTR_ALLOC_INPUT_EN_ETHERTYPE;
+ break;
+ case RTE_ETH_FLOW_NONFRAG_IPV6_TCP:
+ filter->src_port = fdir->input.flow.tcp6_flow.src_port;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_PORT;
+ filter->dst_port = fdir->input.flow.tcp6_flow.dst_port;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_PORT;
+ filter->dst_port_mask = 0xffff;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_PORT_MASK;
+ filter->src_port_mask = 0xffff;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_PORT_MASK;
+ filter->ip_addr_type =
+ NTUPLE_FLTR_ALLOC_INPUT_IP_ADDR_TYPE_IPV6;
+ filter->ip_protocol = fdir->input.flow.tcp6_flow.ip.proto;
+ en |= NTUPLE_FLTR_ALLOC_IN_EN_IP_PROTO;
+ rte_memcpy(filter->src_ipaddr,
+ fdir->input.flow.tcp6_flow.ip.src_ip, 16);
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_IPADDR;
+ rte_memcpy(filter->dst_ipaddr,
+ fdir->input.flow.tcp6_flow.ip.dst_ip, 16);
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_IPADDR;
+ memset(filter->dst_ipaddr_mask, 0xff, 16);
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_IPADDR_MASK;
+ memset(filter->src_ipaddr_mask, 0xff, 16);
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_IPADDR_MASK;
+ filter->ethertype = 0x86dd;
+ filter->enables |= NTUPLE_FLTR_ALLOC_INPUT_EN_ETHERTYPE;
+ break;
+ case RTE_ETH_FLOW_NONFRAG_IPV6_UDP:
+ filter->src_port = fdir->input.flow.udp6_flow.src_port;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_PORT;
+ filter->dst_port = fdir->input.flow.udp6_flow.dst_port;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_PORT;
+ filter->dst_port_mask = 0xffff;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_PORT_MASK;
+ filter->src_port_mask = 0xffff;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_PORT_MASK;
+ filter->ip_addr_type =
+ NTUPLE_FLTR_ALLOC_INPUT_IP_ADDR_TYPE_IPV6;
+ filter->ip_protocol = fdir->input.flow.udp6_flow.ip.proto;
+ en |= NTUPLE_FLTR_ALLOC_IN_EN_IP_PROTO;
+ rte_memcpy(filter->src_ipaddr,
+ fdir->input.flow.udp6_flow.ip.src_ip, 16);
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_IPADDR;
+ rte_memcpy(filter->dst_ipaddr,
+ fdir->input.flow.udp6_flow.ip.dst_ip, 16);
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_IPADDR;
+ memset(filter->dst_ipaddr_mask, 0xff, 16);
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_IPADDR_MASK;
+ memset(filter->src_ipaddr_mask, 0xff, 16);
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_SRC_IPADDR_MASK;
+ filter->ethertype = 0x86dd;
+ filter->enables |= NTUPLE_FLTR_ALLOC_INPUT_EN_ETHERTYPE;
+ break;
+ case RTE_ETH_FLOW_L2_PAYLOAD:
+ filter->ethertype = fdir->input.flow.l2_flow.ether_type;
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_ETHERTYPE;
+ break;
+ case RTE_ETH_FLOW_VXLAN:
+ if (fdir->action.behavior == RTE_ETH_FDIR_REJECT)
+ return -EINVAL;
+ filter->vni = fdir->input.flow.tunnel_flow.tunnel_id;
+ filter->tunnel_type =
+ CFA_NTUPLE_FILTER_ALLOC_REQ_TUNNEL_TYPE_VXLAN;
+ en |= HWRM_CFA_NTUPLE_FILTER_ALLOC_INPUT_ENABLES_TUNNEL_TYPE;
+ break;
+ case RTE_ETH_FLOW_NVGRE:
+ if (fdir->action.behavior == RTE_ETH_FDIR_REJECT)
+ return -EINVAL;
+ filter->vni = fdir->input.flow.tunnel_flow.tunnel_id;
+ filter->tunnel_type =
+ CFA_NTUPLE_FILTER_ALLOC_REQ_TUNNEL_TYPE_NVGRE;
+ en |= HWRM_CFA_NTUPLE_FILTER_ALLOC_INPUT_ENABLES_TUNNEL_TYPE;
+ break;
+ case RTE_ETH_FLOW_UNKNOWN:
+ case RTE_ETH_FLOW_RAW:
+ case RTE_ETH_FLOW_FRAG_IPV4:
+ case RTE_ETH_FLOW_NONFRAG_IPV4_SCTP:
+ case RTE_ETH_FLOW_FRAG_IPV6:
+ case RTE_ETH_FLOW_NONFRAG_IPV6_SCTP:
+ case RTE_ETH_FLOW_IPV6_EX:
+ case RTE_ETH_FLOW_IPV6_TCP_EX:
+ case RTE_ETH_FLOW_IPV6_UDP_EX:
+ case RTE_ETH_FLOW_GENEVE:
+ /* FALLTHROUGH */
+ default:
+ return -EINVAL;
+ }
+
+ vnic0 = STAILQ_FIRST(&bp->ff_pool[0]);
+ vnic = STAILQ_FIRST(&bp->ff_pool[fdir->action.rx_queue]);
+ if (vnic == NULL) {
+ RTE_LOG(ERR, PMD, "Invalid queue %d\n", fdir->action.rx_queue);
+ return -EINVAL;
+ }
+
+
+ if (fdir_mode == RTE_FDIR_MODE_PERFECT_MAC_VLAN) {
+ rte_memcpy(filter->dst_macaddr,
+ fdir->input.flow.mac_vlan_flow.mac_addr.addr_bytes, 6);
+ en |= NTUPLE_FLTR_ALLOC_INPUT_EN_DST_MACADDR;
+ }
+
+ if (fdir->action.behavior == RTE_ETH_FDIR_REJECT) {
+ filter->flags = HWRM_CFA_NTUPLE_FILTER_ALLOC_INPUT_FLAGS_DROP;
+ filter1 = STAILQ_FIRST(&vnic0->filter);
+ //filter1 = bnxt_get_l2_filter(bp, filter, vnic0);
+ } else {
+ filter->dst_id = vnic->fw_vnic_id;
+ for (i = 0; i < ETHER_ADDR_LEN; i++)
+ if (filter->dst_macaddr[i] == 0x00)
+ filter1 = STAILQ_FIRST(&vnic0->filter);
+ else
+ filter1 = bnxt_get_l2_filter(bp, filter, vnic);
+ }
+
+ if (filter1 == NULL)
+ return -EINVAL;
+
+ en |= HWRM_CFA_NTUPLE_FILTER_ALLOC_INPUT_ENABLES_L2_FILTER_ID;
+ filter->fw_l2_filter_id = filter1->fw_l2_filter_id;
+
+ filter->enables = en;
+
+ return 0;
+}
+
+static struct bnxt_filter_info *
+bnxt_match_fdir(struct bnxt *bp, struct bnxt_filter_info *nf)
+{
+ struct bnxt_filter_info *mf = NULL;
+ int i;
+
+ for (i = bp->nr_vnics - 1; i >= 0; i--) {
+ struct bnxt_vnic_info *vnic = &bp->vnic_info[i];
+
+ STAILQ_FOREACH(mf, &vnic->filter, next) {
+ if (mf->filter_type == nf->filter_type &&
+ mf->flags == nf->flags &&
+ mf->src_port == nf->src_port &&
+ mf->src_port_mask == nf->src_port_mask &&
+ mf->dst_port == nf->dst_port &&
+ mf->dst_port_mask == nf->dst_port_mask &&
+ mf->ip_protocol == nf->ip_protocol &&
+ mf->ip_addr_type == nf->ip_addr_type &&
+ mf->ethertype == nf->ethertype &&
+ mf->vni == nf->vni &&
+ mf->tunnel_type == nf->tunnel_type &&
+ mf->l2_ovlan == nf->l2_ovlan &&
+ mf->l2_ovlan_mask == nf->l2_ovlan_mask &&
+ mf->l2_ivlan == nf->l2_ivlan &&
+ mf->l2_ivlan_mask == nf->l2_ivlan_mask &&
+ !memcmp(mf->l2_addr, nf->l2_addr, ETHER_ADDR_LEN) &&
+ !memcmp(mf->l2_addr_mask, nf->l2_addr_mask,
+ ETHER_ADDR_LEN) &&
+ !memcmp(mf->src_macaddr, nf->src_macaddr,
+ ETHER_ADDR_LEN) &&
+ !memcmp(mf->dst_macaddr, nf->dst_macaddr,
+ ETHER_ADDR_LEN) &&
+ !memcmp(mf->src_ipaddr, nf->src_ipaddr,
+ sizeof(nf->src_ipaddr)) &&
+ !memcmp(mf->src_ipaddr_mask, nf->src_ipaddr_mask,
+ sizeof(nf->src_ipaddr_mask)) &&
+ !memcmp(mf->dst_ipaddr, nf->dst_ipaddr,
+ sizeof(nf->dst_ipaddr)) &&
+ !memcmp(mf->dst_ipaddr_mask, nf->dst_ipaddr_mask,
+ sizeof(nf->dst_ipaddr_mask)))
+ return mf;
+ }
+ }
+ return NULL;
+}
+
+static int
+bnxt_fdir_filter(struct rte_eth_dev *dev,
+ enum rte_filter_op filter_op,
+ void *arg)
+{
+ struct bnxt *bp = (struct bnxt *)dev->data->dev_private;
+ struct rte_eth_fdir_filter *fdir = (struct rte_eth_fdir_filter *)arg;
+ struct bnxt_filter_info *filter, *match;
+ struct bnxt_vnic_info *vnic;
+ int ret = 0, i;
+
+ if (filter_op == RTE_ETH_FILTER_NOP)
+ return 0;
+
+ if (arg == NULL && filter_op != RTE_ETH_FILTER_FLUSH)
+ return -EINVAL;
+
+ switch (filter_op) {
+ case RTE_ETH_FILTER_ADD:
+ case RTE_ETH_FILTER_DELETE:
+ /* FALLTHROUGH */
+ filter = bnxt_get_unused_filter(bp);
+ if (filter == NULL) {
+ RTE_LOG(ERR, PMD,
+ "Not enough resources for a new flow.\n");
+ return -ENOMEM;
+ }
+
+ ret = bnxt_parse_fdir_filter(bp, fdir, filter);
+ if (ret != 0)
+ goto free_filter;
+
+ match = bnxt_match_fdir(bp, filter);
+ if (match != NULL && filter_op == RTE_ETH_FILTER_ADD) {
+ RTE_LOG(ERR, PMD, "Flow already exists.\n");
+ ret = -EEXIST;
+ goto free_filter;
+ }
+ if (match == NULL && filter_op == RTE_ETH_FILTER_DELETE) {
+ RTE_LOG(ERR, PMD, "Flow does not exist.\n");
+ ret = -ENOENT;
+ goto free_filter;
+ }
+
+ if (fdir->action.behavior == RTE_ETH_FDIR_REJECT)
+ vnic = STAILQ_FIRST(&bp->ff_pool[0]);
+ else
+ vnic =
+ STAILQ_FIRST(&bp->ff_pool[fdir->action.rx_queue]);
+
+ if (filter_op == RTE_ETH_FILTER_ADD) {
+ filter->filter_type = HWRM_CFA_NTUPLE_FILTER;
+ ret = bnxt_hwrm_set_ntuple_filter(bp,
+ filter->dst_id,
+ filter);
+ if (ret)
+ goto free_filter;
+ STAILQ_INSERT_TAIL(&vnic->filter, filter, next);
+ } else {
+ ret = bnxt_hwrm_clear_ntuple_filter(bp, match);
+ STAILQ_REMOVE(&vnic->filter, match,
+ bnxt_filter_info, next);
+ bnxt_free_filter(bp, match);
+ filter->fw_l2_filter_id = -1;
+ bnxt_free_filter(bp, filter);
+ }
+ break;
+ case RTE_ETH_FILTER_FLUSH:
+ for (i = bp->nr_vnics - 1; i >= 0; i--) {
+ struct bnxt_vnic_info *vnic = &bp->vnic_info[i];
+
+ STAILQ_FOREACH(filter, &vnic->filter, next) {
+ if (filter->filter_type ==
+ HWRM_CFA_NTUPLE_FILTER) {
+ ret =
+ bnxt_hwrm_clear_ntuple_filter(bp,
+ filter);
+ STAILQ_REMOVE(&vnic->filter, filter,
+ bnxt_filter_info, next);
+ }
+ }
+ }
+ return ret;
+ case RTE_ETH_FILTER_UPDATE:
+ case RTE_ETH_FILTER_STATS:
+ case RTE_ETH_FILTER_INFO:
+ /* FALLTHROUGH */
+ RTE_LOG(ERR, PMD, "operation %u not implemented", filter_op);
+ break;
+ default:
+ RTE_LOG(ERR, PMD, "unknown operation %u", filter_op);
+ ret = -EINVAL;
+ break;
+ }
+ return ret;
+
+free_filter:
+ filter->fw_l2_filter_id = -1;
+ bnxt_free_filter(bp, filter);
+ return ret;
+}
+
static int
bnxt_filter_ctrl_op(struct rte_eth_dev *dev __rte_unused,
enum rte_filter_type filter_type,
@@ -2083,12 +2462,13 @@ bnxt_filter_ctrl_op(struct rte_eth_dev *dev __rte_unused,
int ret = 0;
switch (filter_type) {
- case RTE_ETH_FILTER_FDIR:
case RTE_ETH_FILTER_TUNNEL:
- /* FALLTHROUGH */
RTE_LOG(ERR, PMD,
"filter type: %d: To be implemented\n", filter_type);
break;
+ case RTE_ETH_FILTER_FDIR:
+ ret = bnxt_fdir_filter(dev, filter_op, arg);
+ break;
case RTE_ETH_FILTER_NTUPLE:
ret = bnxt_ntuple_filter(dev, filter_op, arg);
break;
diff --git a/drivers/net/bnxt/bnxt_filter.h b/drivers/net/bnxt/bnxt_filter.h
index 8ed9ca124..c47a34193 100644
--- a/drivers/net/bnxt/bnxt_filter.h
+++ b/drivers/net/bnxt/bnxt_filter.h
@@ -158,4 +158,6 @@ struct bnxt_filter_info *bnxt_get_l2_filter(struct bnxt *bp,
HWRM_CFA_NTUPLE_FILTER_ALLOC_INPUT_IP_PROTOCOL_UNKNOWN
#define NTUPLE_FLTR_ALLOC_INPUT_IP_ADDR_TYPE_IPV4 \
HWRM_CFA_NTUPLE_FILTER_ALLOC_INPUT_IP_ADDR_TYPE_IPV4
+#define NTUPLE_FLTR_ALLOC_INPUT_EN_MIRROR_VNIC_ID \
+ HWRM_CFA_NTUPLE_FILTER_ALLOC_INPUT_ENABLES_MIRROR_VNIC_ID
#endif
--
2.13.5 (Apple Git-94)
^ permalink raw reply related [flat|nested] 6+ messages in thread
* [PATCH 4/4] net/bnxt: fix the association of a MACVLAN per VNIC
2017-10-10 14:22 [PATCH 0/4] patches for bnxt Ajit Khaparde
` (2 preceding siblings ...)
2017-10-10 14:23 ` [PATCH 3/4] net/bnxt: include fdir support Ajit Khaparde
@ 2017-10-10 14:23 ` Ajit Khaparde
2017-10-11 2:08 ` [PATCH 0/4] patches for bnxt Ferruh Yigit
4 siblings, 0 replies; 6+ messages in thread
From: Ajit Khaparde @ 2017-10-10 14:23 UTC (permalink / raw)
To: dev; +Cc: ferruh.yigit
We were not associating a MAC+VLAN per VNIC filter correctly.
This patch fixes that. Also set the VLAN type appropriately.
Signed-off-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
---
drivers/net/bnxt/bnxt_hwrm.c | 22 ++++++++++++++++++++++
drivers/net/bnxt/bnxt_rxq.c | 2 +-
2 files changed, 23 insertions(+), 1 deletion(-)
diff --git a/drivers/net/bnxt/bnxt_hwrm.c b/drivers/net/bnxt/bnxt_hwrm.c
index e66994a76..c9b04e248 100644
--- a/drivers/net/bnxt/bnxt_hwrm.c
+++ b/drivers/net/bnxt/bnxt_hwrm.c
@@ -360,7 +360,23 @@ int bnxt_hwrm_set_l2_filter(struct bnxt *bp,
int rc = 0;
struct hwrm_cfa_l2_filter_alloc_input req = {.req_type = 0 };
struct hwrm_cfa_l2_filter_alloc_output *resp = bp->hwrm_cmd_resp_addr;
+ struct rte_eth_conf *dev_conf = &bp->eth_dev->data->dev_conf;
+ const struct rte_eth_vmdq_rx_conf *conf =
+ &dev_conf->rx_adv_conf.vmdq_rx_conf;
uint32_t enables = 0;
+ uint16_t j = dst_id - 1;
+
+ //TODO: Is there a better way to add VLANs to each VNIC in case of VMDQ
+ if (conf->pool_map[j].pools & (1UL << j)) {
+ RTE_LOG(DEBUG, PMD,
+ "Add vlan %u to vmdq pool %u\n",
+ conf->pool_map[j].vlan_id, j);
+
+ filter->l2_ivlan = conf->pool_map[j].vlan_id;
+ filter->enables |=
+ HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_L2_IVLAN |
+ HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_L2_IVLAN_MASK;
+ }
if (filter->fw_l2_filter_id != UINT64_MAX)
bnxt_hwrm_clear_l2_filter(bp, filter);
@@ -385,8 +401,14 @@ int bnxt_hwrm_set_l2_filter(struct bnxt *bp,
HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_L2_OVLAN)
req.l2_ovlan = filter->l2_ovlan;
if (enables &
+ HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_L2_IVLAN)
+ req.l2_ovlan = filter->l2_ivlan;
+ if (enables &
HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_L2_OVLAN_MASK)
req.l2_ovlan_mask = filter->l2_ovlan_mask;
+ if (enables &
+ HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_L2_IVLAN_MASK)
+ req.l2_ovlan_mask = filter->l2_ivlan_mask;
if (enables & HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_SRC_ID)
req.src_id = rte_cpu_to_le_32(filter->src_id);
if (enables & HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_SRC_TYPE)
diff --git a/drivers/net/bnxt/bnxt_rxq.c b/drivers/net/bnxt/bnxt_rxq.c
index bfa5308ad..f2389515d 100644
--- a/drivers/net/bnxt/bnxt_rxq.c
+++ b/drivers/net/bnxt/bnxt_rxq.c
@@ -184,7 +184,7 @@ int bnxt_mq_rx_configure(struct bnxt *bp)
filter->l2_ivlan = conf->pool_map[j].vlan_id;
filter->enables |=
- HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_L2_OVLAN;
+ HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_L2_IVLAN;
}
}
/*
--
2.13.5 (Apple Git-94)
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH 0/4] patches for bnxt
2017-10-10 14:22 [PATCH 0/4] patches for bnxt Ajit Khaparde
` (3 preceding siblings ...)
2017-10-10 14:23 ` [PATCH 4/4] net/bnxt: fix the association of a MACVLAN per VNIC Ajit Khaparde
@ 2017-10-11 2:08 ` Ferruh Yigit
4 siblings, 0 replies; 6+ messages in thread
From: Ferruh Yigit @ 2017-10-11 2:08 UTC (permalink / raw)
To: Ajit Khaparde, dev
On 10/10/2017 3:22 PM, Ajit Khaparde wrote:
> Patches against dpdk-next-net.
> Please apply.
>
> Ajit Khaparde (4):
> net/bnxt: fix cleanup if a filter allocation fails
> net/bnxt: prevent programming a duplicate flow
> net/bnxt: include fdir support
> net/bnxt: fix the association of a MACVLAN per VNIC
Series applied to dpdk-next-net/master, thanks.
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2017-10-11 2:08 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-10-10 14:22 [PATCH 0/4] patches for bnxt Ajit Khaparde
2017-10-10 14:23 ` [PATCH 1/4] net/bnxt: fix cleanup if a filter allocation fails Ajit Khaparde
2017-10-10 14:23 ` [PATCH 2/4] net/bnxt: prevent programming a duplicate flow Ajit Khaparde
2017-10-10 14:23 ` [PATCH 3/4] net/bnxt: include fdir support Ajit Khaparde
2017-10-10 14:23 ` [PATCH 4/4] net/bnxt: fix the association of a MACVLAN per VNIC Ajit Khaparde
2017-10-11 2:08 ` [PATCH 0/4] patches for bnxt Ferruh Yigit
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.