[dpdk-dev] [PATCH] net/enic: move link checking init to probe time
The enic DMAs link status information to the host and this requires a
little setup. This setup was being done as a result of calling
rte_eth_dev_start(). But applications expect to be able to check link
status before calling rte_eth_dev_start().
This patch moves he link status setup to enic_init() which is called
at device probe time so that link status can be checked anytime.
Fixes: fefed3d1e62c ("enic: new driver")
Signed-off-by: Nelson Escobar
Reviewed-by: John Daley
---
drivers/net/enic/enic_main.c | 8 ++--
1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/drivers/net/enic/enic_main.c b/drivers/net/enic/enic_main.c
index b4ca371..eb32ac1 100644
--- a/drivers/net/enic/enic_main.c
+++ b/drivers/net/enic/enic_main.c
@@ -430,7 +430,6 @@ int enic_enable(struct enic *enic)
eth_dev->data->dev_link.link_speed = vnic_dev_port_speed(enic->vdev);
eth_dev->data->dev_link.link_duplex = ETH_LINK_FULL_DUPLEX;
- vnic_dev_notify_set(enic->vdev, -1); /* No Intr for notify */
if (enic_clsf_init(enic))
dev_warning(enic, "Init of hash table for clsf failed."\
@@ -820,7 +819,6 @@ int enic_disable(struct enic *enic)
}
vnic_dev_set_reset_flag(enic->vdev, 1);
- vnic_dev_notify_unset(enic->vdev);
for (i = 0; i < enic->wq_count; i++)
vnic_wq_clean(>wq[i], enic_free_wq_buf);
@@ -1022,6 +1020,9 @@ static void enic_dev_deinit(struct enic *enic)
{
struct rte_eth_dev *eth_dev = enic->rte_dev;
+ /* stop link status checking */
+ vnic_dev_notify_unset(enic->vdev);
+
rte_free(eth_dev->data->mac_addrs);
}
@@ -1137,6 +1138,9 @@ static int enic_dev_init(struct enic *enic)
vnic_dev_set_reset_flag(enic->vdev, 0);
+ /* set up link status checking */
+ vnic_dev_notify_set(enic->vdev, -1); /* No Intr for notify */
+
return 0;
}
--
2.7.0
[dpdk-dev] [PATCH] net/enic: fix segfault after receiving error interrupt
When enic's interrupt handler is called indicating an error, it scans
through the receive queues (RQs) on the adapter looking for errors.
But since the inclusion of rx scatter, some of the RQs may not be in
use, and you shouldn't check them for errors.
Fixes: 856d7ba7ed22 ("net/enic: support scattered Rx")
Signed-off-by: Nelson Escobar
Reviewed-by: John Daley
---
drivers/net/enic/enic_main.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/net/enic/enic_main.c b/drivers/net/enic/enic_main.c
index 9ec2a2d..fbd4089 100644
--- a/drivers/net/enic/enic_main.c
+++ b/drivers/net/enic/enic_main.c
@@ -123,6 +123,8 @@ static void enic_log_q_error(struct enic *enic)
}
for (i = 0; i < enic_vnic_rq_count(enic); i++) {
+ if (!enic->rq[i].in_use)
+ continue;
error_status = vnic_rq_error_status(>rq[i]);
if (error_status)
dev_err(enic, "RQ[%d] error_status %d\n", i,
--
2.7.0
[dpdk-dev] [PATCH] net/enic: fix calculation of truncated packets
The calculation of truncated packets didn't take into account packet
errors due to the adapter not having buffers, causing both the
ipackets, and imissed counts to be wrong if such errors occurred. In
order to properly calculate the number of packets truncated, we need
to subtract the count of errors due to no buffers.
Fixes: c44d9f01adf3 ("net/enic: count truncated packets")
Signed-off-by: Nelson Escobar
---
drivers/net/enic/enic_main.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/net/enic/enic_main.c b/drivers/net/enic/enic_main.c
index d8669cc..9ec2a2d 100644
--- a/drivers/net/enic/enic_main.c
+++ b/drivers/net/enic/enic_main.c
@@ -172,7 +172,8 @@ void enic_dev_stats_get(struct enic *enic, struct
rte_eth_stats *r_stats)
* which can make ibytes be slightly higher than it should be.
*/
rx_packet_errors = rte_atomic64_read(_stats->rx_packet_errors);
- rx_truncated = rx_packet_errors - stats->rx.rx_errors;
+ rx_truncated = rx_packet_errors - stats->rx.rx_errors -
+ stats->rx.rx_no_bufs;
r_stats->ipackets = stats->rx.rx_frames_ok - rx_truncated;
r_stats->opackets = stats->tx.tx_frames_ok;
--
2.7.0
[dpdk-dev] [PATCH 1/2] net/enic: fix setting MAC address when a port is restarted
enic_disable() removed the MAC address when a port was shut down but
enic_enable() didn't add the MAC address back when the port was
started again. Move where we set the MAC address for the adapter from
enic_setup_finish() to a enic_enable() so that port restarting works
properly.
Fixes: fefed3d1e62c ("enic: new driver")
Signed-off-by: Nelson Escobar
Reviewed-by: John Daley
---
drivers/net/enic/enic_main.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/enic/enic_main.c b/drivers/net/enic/enic_main.c
index 8eacf2d..1f3d14e 100644
--- a/drivers/net/enic/enic_main.c
+++ b/drivers/net/enic/enic_main.c
@@ -455,6 +455,8 @@ int enic_enable(struct enic *enic)
for (index = 0; index < enic->rq_count; index++)
enic_start_rq(enic, index);
+ vnic_dev_add_addr(enic->vdev, enic->mac_addr);
+
vnic_dev_enable_wait(enic->vdev);
/* Register and enable error interrupt */
@@ -971,8 +973,6 @@ int enic_setup_finish(struct enic *enic)
return -1;
}
- vnic_dev_add_addr(enic->vdev, enic->mac_addr);
-
/* Default conf */
vnic_dev_packet_filter(enic->vdev,
1 /* directed */,
--
2.7.0
[dpdk-dev] [PATCH] net/enic: fix checking for sufficient resources
The enic PMD was using the same variables in the enic structure to
track two different things. Initially rq_count, wq_count, cq_count,
and intr_count were set to the values obtained from the VIC adapters
as the maximum resources allocated on the VIC, then in
enic_set_vnic_res(), they were set to the counts of resources actually
used, discarding the initial values. The checks in enic_set_vnic_res()
were technically incorrect if it is called more than once on a port,
which happens when using bonding, but were harmless in practice as the
checks couldn't fail on the second call.
The enic rx-scatter patch misunderstood the subtleties of
enic_set_vnic_res(), and naively added a multiply by two to the
rq_count check. This resulted in the rq_count check failing when
enic_set_vnic_res() was called a second time, ie when using bonding.
This patch adds new variables to the enic structure to track the
maximum resources the VIC is configured to provide so that the
information isn't later lost and calls to enic_set_vnic_res() do
the expected thing.
Fixes: 856d7ba7ed22 ("net/enic: support scattered Rx")
Signed-off-by: Nelson Escobar
---
drivers/net/enic/enic.h| 6 ++
drivers/net/enic/enic_ethdev.c | 5 +++--
drivers/net/enic/enic_main.c | 14 --
drivers/net/enic/enic_res.c| 12 ++--
4 files changed, 23 insertions(+), 14 deletions(-)
diff --git a/drivers/net/enic/enic.h b/drivers/net/enic/enic.h
index 53fed0b..33057d4 100644
--- a/drivers/net/enic/enic.h
+++ b/drivers/net/enic/enic.h
@@ -152,6 +152,12 @@ struct enic {
/* software counters */
struct enic_soft_stats soft_stats;
+ /* configured resources on vic */
+ unsigned int conf_rq_count;
+ unsigned int conf_wq_count;
+ unsigned int conf_cq_count;
+ unsigned int conf_intr_count;
+
/* linked list storing memory allocations */
LIST_HEAD(enic_memzone_list, enic_memzone_entry) memzone_list;
rte_spinlock_t memzone_list_lock;
diff --git a/drivers/net/enic/enic_ethdev.c b/drivers/net/enic/enic_ethdev.c
index a7ce064..66b81a2 100644
--- a/drivers/net/enic/enic_ethdev.c
+++ b/drivers/net/enic/enic_ethdev.c
@@ -436,8 +436,9 @@ static void enicpmd_dev_info_get(struct rte_eth_dev
*eth_dev,
struct enic *enic = pmd_priv(eth_dev);
ENICPMD_FUNC_TRACE();
- device_info->max_rx_queues = enic->rq_count;
- device_info->max_tx_queues = enic->wq_count;
+ /* Scattered Rx uses two receive queues per rx queue exposed to dpdk */
+ device_info->max_rx_queues = enic->conf_rq_count / 2;
+ device_info->max_tx_queues = enic->conf_wq_count;
device_info->min_rx_bufsize = ENIC_MIN_MTU;
device_info->max_rx_pktlen = enic->rte_dev->data->mtu
+ ETHER_HDR_LEN + 4;
diff --git a/drivers/net/enic/enic_main.c b/drivers/net/enic/enic_main.c
index 0475cc1..d4e43b5 100644
--- a/drivers/net/enic/enic_main.c
+++ b/drivers/net/enic/enic_main.c
@@ -1016,21 +1016,23 @@ int enic_set_vnic_res(struct enic *enic)
/* With Rx scatter support, two RQs are now used per RQ used by
* the application.
*/
- if (enic->rq_count < (eth_dev->data->nb_rx_queues * 2)) {
+ if (enic->conf_rq_count < eth_dev->data->nb_rx_queues) {
dev_err(dev, "Not enough Receive queues. Requested:%u which
uses %d RQs on VIC, Configured:%u\n",
eth_dev->data->nb_rx_queues,
- eth_dev->data->nb_rx_queues * 2, enic->rq_count);
+ eth_dev->data->nb_rx_queues * 2, enic->conf_rq_count);
rc = -EINVAL;
}
- if (enic->wq_count < eth_dev->data->nb_tx_queues) {
+ if (enic->conf_wq_count < eth_dev->data->nb_tx_queues) {
dev_err(dev, "Not enough Transmit queues. Requested:%u,
Configured:%u\n",
- eth_dev->data->nb_tx_queues, enic->wq_count);
+ eth_dev->data->nb_tx_queues, enic->conf_wq_count);
rc = -EINVAL;
}
- if (enic->cq_count < (enic->rq_count + enic->wq_count)) {
+ if (enic->conf_cq_count < (eth_dev->data->nb_rx_queues +
+ eth_dev->data->nb_tx_queues)) {
dev_err(dev, "Not enough Completion queues. Required:%u,
Configured:%u\n",
- enic->rq_count + enic->wq_count, enic->cq_count);
+ (eth_dev->data->nb_rx_queues +
+eth_dev->data->nb_tx_queues), enic->conf_cq_count);
rc = -EINVAL;
}
diff --git a/drivers/net/enic/enic_res.c b/drivers/net/enic/enic_res.c
index b271d34..84c5d33 100644
--- a/drivers/net/enic/enic_res.c
+++ b/drivers/net/enic/enic
[dpdk-dev] [PATCH v2] enic: fix issues when using Rx scatter with multiple RQs
The Rx scatter patch failed to make a few changes and resulted in
problems when using multiple receive queues in dpdk (ie RSS) since the
wrong adapter resources were being used.
- get and use the correct completion queue index associated with a
receive queue.
- set the correct receive queue index when using RSS
Fixes: 14a261bf0520 ("enic: add scattered Rx support")
Signed-off-by: Nelson Escobar
Reviewed-by: John Daley
---
drivers/net/enic/enic.h | 6 +-
drivers/net/enic/enic_main.c | 10 ++
2 files changed, 11 insertions(+), 5 deletions(-)
diff --git a/drivers/net/enic/enic.h b/drivers/net/enic/enic.h
index ed5f18d..1dd8aee 100644
--- a/drivers/net/enic/enic.h
+++ b/drivers/net/enic/enic.h
@@ -174,7 +174,11 @@ static inline unsigned int enic_vnic_rq_count(struct enic
*enic)
static inline unsigned int enic_cq_rq(__rte_unused struct enic *enic, unsigned
int rq)
{
- return rq;
+ /* Scatter rx uses two receive queues together with one
+* completion queue, so the completion queue number is no
+* longer the same as the rq number.
+*/
+ return rq / 2;
}
static inline unsigned int enic_cq_wq(struct enic *enic, unsigned int wq)
diff --git a/drivers/net/enic/enic_main.c b/drivers/net/enic/enic_main.c
index 15389e5..68dbe40 100644
--- a/drivers/net/enic/enic_main.c
+++ b/drivers/net/enic/enic_main.c
@@ -238,19 +238,20 @@ void enic_init_vnic_resources(struct enic *enic)
struct vnic_rq *data_rq;
for (index = 0; index < enic->rq_count; index++) {
+ cq_idx = enic_cq_rq(enic, enic_sop_rq(index));
+
vnic_rq_init(>rq[enic_sop_rq(index)],
- enic_cq_rq(enic, index),
+ cq_idx,
error_interrupt_enable,
error_interrupt_offset);
data_rq = >rq[enic_data_rq(index)];
if (data_rq->in_use)
vnic_rq_init(data_rq,
-enic_cq_rq(enic, index),
+cq_idx,
error_interrupt_enable,
error_interrupt_offset);
- cq_idx = enic_cq_rq(enic, index);
vnic_cq_init(>cq[cq_idx],
0 /* flow_control_enable */,
1 /* color_enable */,
@@ -899,7 +900,8 @@ static int enic_set_rsscpu(struct enic *enic, u8
rss_hash_bits)
return -ENOMEM;
for (i = 0; i < (1 << rss_hash_bits); i++)
- (*rss_cpu_buf_va).cpu[i/4].b[i%4] = i % enic->rq_count;
+ (*rss_cpu_buf_va).cpu[i / 4].b[i % 4] =
+ enic_sop_rq(i % enic->rq_count);
err = enic_set_rss_cpu(enic,
rss_cpu_buf_pa,
--
2.7.0
[dpdk-dev] [PATCH] enic: fix issues when using Rx scatter with multiple RQs
The Rx scatter patch failed to make a few changes and resulted
in problems when using multiple RQs since the wrong RQ or CQ
was being used.
Fixes: 14a261bf0520 ("enic: add scattered Rx support")
Signed-off-by: Nelson Escobar
Reviewed-by: John Daley
---
drivers/net/enic/enic.h | 2 +-
drivers/net/enic/enic_main.c | 10 ++
2 files changed, 7 insertions(+), 5 deletions(-)
diff --git a/drivers/net/enic/enic.h b/drivers/net/enic/enic.h
index ed5f18d..15b1d45 100644
--- a/drivers/net/enic/enic.h
+++ b/drivers/net/enic/enic.h
@@ -174,7 +174,7 @@ static inline unsigned int enic_vnic_rq_count(struct enic
*enic)
static inline unsigned int enic_cq_rq(__rte_unused struct enic *enic, unsigned
int rq)
{
- return rq;
+ return rq / 2;
}
static inline unsigned int enic_cq_wq(struct enic *enic, unsigned int wq)
diff --git a/drivers/net/enic/enic_main.c b/drivers/net/enic/enic_main.c
index 15389e5..68dbe40 100644
--- a/drivers/net/enic/enic_main.c
+++ b/drivers/net/enic/enic_main.c
@@ -238,19 +238,20 @@ void enic_init_vnic_resources(struct enic *enic)
struct vnic_rq *data_rq;
for (index = 0; index < enic->rq_count; index++) {
+ cq_idx = enic_cq_rq(enic, enic_sop_rq(index));
+
vnic_rq_init(>rq[enic_sop_rq(index)],
- enic_cq_rq(enic, index),
+ cq_idx,
error_interrupt_enable,
error_interrupt_offset);
data_rq = >rq[enic_data_rq(index)];
if (data_rq->in_use)
vnic_rq_init(data_rq,
-enic_cq_rq(enic, index),
+cq_idx,
error_interrupt_enable,
error_interrupt_offset);
- cq_idx = enic_cq_rq(enic, index);
vnic_cq_init(>cq[cq_idx],
0 /* flow_control_enable */,
1 /* color_enable */,
@@ -899,7 +900,8 @@ static int enic_set_rsscpu(struct enic *enic, u8
rss_hash_bits)
return -ENOMEM;
for (i = 0; i < (1 << rss_hash_bits); i++)
- (*rss_cpu_buf_va).cpu[i/4].b[i%4] = i % enic->rq_count;
+ (*rss_cpu_buf_va).cpu[i / 4].b[i % 4] =
+ enic_sop_rq(i % enic->rq_count);
err = enic_set_rss_cpu(enic,
rss_cpu_buf_pa,
--
2.7.0
[dpdk-dev] [PATCH v2] enic: fix free function to actually free memory
enic_alloc_consistent() allocated memory, but enic_free_consistent()
was an empty function, so allocated memory was never freed.
This commit adds a list and lock to the enic structure to keep track
of the memzones allocated in enic_alloc_consistent(), and
enic_free_consistent() uses that information to properly free memory.
Fixes: fefed3d1e62c ("enic: new driver")
Signed-off-by: Nelson Escobar
Reviewed-by: John Daley
---
v2:
- updated commit message to mention memzone tracking
drivers/net/enic/base/vnic_dev.c | 14 +-
drivers/net/enic/base/vnic_dev.h | 2 +-
drivers/net/enic/enic.h | 12 +
drivers/net/enic/enic_main.c | 56 +---
4 files changed, 67 insertions(+), 17 deletions(-)
diff --git a/drivers/net/enic/base/vnic_dev.c b/drivers/net/enic/base/vnic_dev.c
index e8a5028..fc2e4cc 100644
--- a/drivers/net/enic/base/vnic_dev.c
+++ b/drivers/net/enic/base/vnic_dev.c
@@ -83,7 +83,7 @@ struct vnic_dev {
struct vnic_intr_coal_timer_info intr_coal_timer_info;
void *(*alloc_consistent)(void *priv, size_t size,
dma_addr_t *dma_handle, u8 *name);
- void (*free_consistent)(struct rte_pci_device *hwdev,
+ void (*free_consistent)(void *priv,
size_t size, void *vaddr,
dma_addr_t dma_handle);
};
@@ -101,7 +101,7 @@ void *vnic_dev_priv(struct vnic_dev *vdev)
void vnic_register_cbacks(struct vnic_dev *vdev,
void *(*alloc_consistent)(void *priv, size_t size,
dma_addr_t *dma_handle, u8 *name),
- void (*free_consistent)(struct rte_pci_device *hwdev,
+ void (*free_consistent)(void *priv,
size_t size, void *vaddr,
dma_addr_t dma_handle))
{
@@ -807,7 +807,7 @@ int vnic_dev_notify_unsetcmd(struct vnic_dev *vdev)
int vnic_dev_notify_unset(struct vnic_dev *vdev)
{
if (vdev->notify && !vnic_dev_in_reset(vdev)) {
- vdev->free_consistent(vdev->pdev,
+ vdev->free_consistent(vdev->priv,
sizeof(struct vnic_devcmd_notify),
vdev->notify,
vdev->notify_pa);
@@ -924,16 +924,16 @@ void vnic_dev_unregister(struct vnic_dev *vdev)
{
if (vdev) {
if (vdev->notify)
- vdev->free_consistent(vdev->pdev,
+ vdev->free_consistent(vdev->priv,
sizeof(struct vnic_devcmd_notify),
vdev->notify,
vdev->notify_pa);
if (vdev->stats)
- vdev->free_consistent(vdev->pdev,
+ vdev->free_consistent(vdev->priv,
sizeof(struct vnic_stats),
vdev->stats, vdev->stats_pa);
if (vdev->fw_info)
- vdev->free_consistent(vdev->pdev,
+ vdev->free_consistent(vdev->priv,
sizeof(struct vnic_devcmd_fw_info),
vdev->fw_info, vdev->fw_info_pa);
kfree(vdev);
@@ -1041,7 +1041,7 @@ int vnic_dev_classifier(struct vnic_dev *vdev, u8 cmd,
u16 *entry,
ret = vnic_dev_cmd(vdev, CMD_ADD_FILTER, , , wait);
*entry = (u16)a0;
- vdev->free_consistent(vdev->pdev, tlv_size, tlv_va, tlv_pa);
+ vdev->free_consistent(vdev->priv, tlv_size, tlv_va, tlv_pa);
} else if (cmd == CLSF_DEL) {
a0 = *entry;
ret = vnic_dev_cmd(vdev, CMD_DEL_FILTER, , , wait);
diff --git a/drivers/net/enic/base/vnic_dev.h b/drivers/net/enic/base/vnic_dev.h
index 113d6ac..689442f 100644
--- a/drivers/net/enic/base/vnic_dev.h
+++ b/drivers/net/enic/base/vnic_dev.h
@@ -102,7 +102,7 @@ unsigned int vnic_dev_get_res_count(struct vnic_dev *vdev,
void vnic_register_cbacks(struct vnic_dev *vdev,
void *(*alloc_consistent)(void *priv, size_t size,
dma_addr_t *dma_handle, u8 *name),
- void (*free_consistent)(struct rte_pci_device *hwdev,
+ void (*free_consistent)(void *priv,
size_t size, void *vaddr,
dma_addr_t dma_handle));
void __iomem *vnic_dev_get_res(struct vnic_dev *vdev, enum vnic_res_type type,
diff --git a/drivers/net/enic/enic.h b/drivers/net/enic/enic.h
index 1e6914e..9f94afb 100644
--- a/drivers/net/enic/enic.h
+++ b/drivers/net/enic/enic.h
@@ -46,6 +46,8 @@
#include "vnic_rss.h"
#include "enic_res.h"
#include "cq_enet_desc.h"
+#include
+#include
#define DRV_NAME "enic_pmd"
#define DRV_DESCRIPTION"Cisco VIC Ethernet NIC Poll-mode
Driver"
@@ -95,6 +97,11 @@ struct enic_soft_stats {
rte_atomic64_t rx_packet_errors;
[dpdk-dev] [PATCH v2] enic: fix name of hash table used for enic classifiers
The enic_clsf_init() function is called once per enic instance, but it
used a static name to create the hash table. Consequently when using
more than one enic instance, there was a name collision which caused
errors:
EAL: memzone_reserve_aligned_thread_unsafe():
memzone already exists
RING: Cannot reserve memory
HASH: memory allocation failed
PMD: rte_enic_pmd: Init of hash table for clsf failed.
Flow director feature will not work
This patch changes the name to be unique per enic instance.
Fixes: fefed3d1e62c ("enic: new driver")
Signed-off-by: Nelson Escobar
Reviewed-by: John Daley
---
v2:
- use snprintf instead of sprintf
drivers/net/enic/enic_clsf.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/drivers/net/enic/enic_clsf.c b/drivers/net/enic/enic_clsf.c
index edb56e1..2050818 100644
--- a/drivers/net/enic/enic_clsf.c
+++ b/drivers/net/enic/enic_clsf.c
@@ -239,15 +239,16 @@ void enic_clsf_destroy(struct enic *enic)
int enic_clsf_init(struct enic *enic)
{
+ char clsf_name[RTE_HASH_NAMESIZE];
struct rte_hash_parameters hash_params = {
- .name = "enicpmd_clsf_hash",
+ .name = clsf_name,
.entries = ENICPMD_CLSF_HASH_ENTRIES,
.key_len = sizeof(struct rte_eth_fdir_filter),
.hash_func = DEFAULT_HASH_FUNC,
.hash_func_init_val = 0,
.socket_id = SOCKET_ID_ANY,
};
-
+ snprintf(clsf_name, RTE_HASH_NAMESIZE, "enic_clsf_%s", enic->bdf_name);
enic->fdir.hash = rte_hash_create(_params);
memset(>fdir.stats, 0, sizeof(enic->fdir.stats));
enic->fdir.stats.free = ENICPMD_FDIR_MAX;
--
2.7.0
[dpdk-dev] [PATCH v2] enic: scattered Rx
For performance reasons, this patch uses 2 VIC RQs per RQ presented to
DPDK.
The VIC requires that each descriptor be marked as either a start of
packet (SOP) descriptor or a non-SOP descriptor. A one RQ solution
requires skipping descriptors when receiving small packets and results
in bad performance when receiving many small packets.
The 2 RQ solution makes use of the VIC feature that allows a receive
on primary queue to 'spill over' into another queue if the receive is
too large to fit in the buffer assigned to the descriptor on the
primary queue. This means that there is no skipping of descriptors
when receiving small packets and results in much better performance.
Signed-off-by: Nelson Escobar
Reviewed-by: John Daley
---
v2:
- fixes upstream checkpatch complaint
- fixes bug where packet type and flags were set on last mbuf
instead of first mbuf of scattered receive
- adds ethernet hdr length to mtu when calculating the number of
mbufs it would take to receive maximum sized packet
doc/guides/nics/overview.rst | 2 +-
drivers/net/enic/base/rq_enet_desc.h | 2 +-
drivers/net/enic/base/vnic_rq.c | 8 +-
drivers/net/enic/base/vnic_rq.h | 18 ++-
drivers/net/enic/enic.h | 22 ++-
drivers/net/enic/enic_ethdev.c | 10 +-
drivers/net/enic/enic_main.c | 277 +++
drivers/net/enic/enic_res.c | 5 +-
drivers/net/enic/enic_rxtx.c | 140 --
9 files changed, 361 insertions(+), 123 deletions(-)
diff --git a/doc/guides/nics/overview.rst b/doc/guides/nics/overview.rst
index 2200171..d0ae847 100644
--- a/doc/guides/nics/overview.rst
+++ b/doc/guides/nics/overview.rst
@@ -94,7 +94,7 @@ Most of these differences are summarized below.
Queue start/stop Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y Y Y
MTU update Y Y Y Y Y Y Y Y Y Y
Jumbo frame Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y Y Y
- Scattered Rx Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y Y
+ Scattered Rx Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y Y
LRO Y Y Y Y
TSO Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Promiscuous mode Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y Y Y Y Y
diff --git a/drivers/net/enic/base/rq_enet_desc.h
b/drivers/net/enic/base/rq_enet_desc.h
index 7292d9d..13e24b4 100644
--- a/drivers/net/enic/base/rq_enet_desc.h
+++ b/drivers/net/enic/base/rq_enet_desc.h
@@ -55,7 +55,7 @@ enum rq_enet_type_types {
#define RQ_ENET_TYPE_BITS 2
#define RQ_ENET_TYPE_MASK ((1 << RQ_ENET_TYPE_BITS) - 1)
-static inline void rq_enet_desc_enc(struct rq_enet_desc *desc,
+static inline void rq_enet_desc_enc(volatile struct rq_enet_desc *desc,
u64 address, u8 type, u16 length)
{
desc->address = cpu_to_le64(address);
diff --git a/drivers/net/enic/base/vnic_rq.c b/drivers/net/enic/base/vnic_rq.c
index cb62c5e..0e700a1 100644
--- a/drivers/net/enic/base/vnic_rq.c
+++ b/drivers/net/enic/base/vnic_rq.c
@@ -84,11 +84,12 @@ void vnic_rq_init_start(struct vnic_rq *rq, unsigned int
cq_index,
iowrite32(cq_index, >ctrl->cq_index);
iowrite32(error_interrupt_enable, >ctrl->error_interrupt_enable);
iowrite32(error_interrupt_offset, >ctrl->error_interrupt_offset);
- iowrite32(0, >ctrl->dropped_packet_count);
iowrite32(0, >ctrl->error_status);
iowrite32(fetch_index, >ctrl->fetch_index);
iowrite32(posted_index, >ctrl->posted_index);
-
+ if (rq->is_sop)
+ iowrite32(((rq->is_sop << 10) | rq->data_queue_idx),
+ >ctrl->data_ring);
}
void vnic_rq_init(struct vnic_rq *rq, unsigned int cq_index,
@@ -96,6 +97,7 @@ void vnic_rq_init(struct vnic_rq *rq, unsigned int cq_index,
unsigned int error_interrupt_offset)
{
u32 fetch_index = 0;
+
/* Use current fetch_index as the ring starting point */
fetch_index = ioread32(>ctrl->fetch_index);
@@ -110,6 +112,8 @@ void vnic_rq_init(struct vnic_rq *rq, unsigned int cq_index,
error_interrupt_offset);
rq->rxst_idx = 0;
rq->tot_pkts = 0;
+ rq->pkt_first_seg = NULL;
+ rq->pkt_last_seg = NULL;
}
void vnic_rq_error_out(struct vnic_rq *rq, unsigned int error)
diff --git a/drivers/net/enic/base/vnic_rq.h b/drivers/net/enic/base/vnic_rq.h
index e083ccc..fd9e170 100644
--- a/drivers/net/enic/base/vnic_rq.h
+++ b/drivers/net/enic/base/vnic_rq.h
@@ -60,10 +60,18 @@ struct vnic_rq_ctrl {
u32 pad7;
u32 error_status; /* 0x48 */
u32 pad8;
- u32 dropped_packet_count; /* 0x50 */
+ u32 tcp_sn;
[dpdk-dev] [PATCH] enic: scattered Rx
For performance reasons, this patch uses 2 VIC RQs per RQ presented to
DPDK.
The VIC requires that each descriptor be marked as either a start of
packet (SOP) descriptor or a non-SOP descriptor. A one RQ solution
requires skipping descriptors when receiving small packets and results
in bad performance when receiving many small packets.
The 2 RQ solution makes use of the VIC feature that allows a receive
on primary queue to 'spill over' into another queue if the receive is
too large to fit in the buffer assigned to the descriptor on the
primary queue. This means that there is no skipping of descriptors
when receiving small packets and results in much better performance.
Signed-off-by: Nelson Escobar
Reviewed-by: John Daley
---
doc/guides/nics/overview.rst | 2 +-
drivers/net/enic/base/rq_enet_desc.h | 2 +-
drivers/net/enic/base/vnic_rq.c | 8 +-
drivers/net/enic/base/vnic_rq.h | 18 ++-
drivers/net/enic/enic.h | 22 ++-
drivers/net/enic/enic_ethdev.c | 10 +-
drivers/net/enic/enic_main.c | 277 +++
drivers/net/enic/enic_res.c | 5 +-
drivers/net/enic/enic_rxtx.c | 137 +++--
9 files changed, 358 insertions(+), 123 deletions(-)
diff --git a/doc/guides/nics/overview.rst b/doc/guides/nics/overview.rst
index 2200171..d0ae847 100644
--- a/doc/guides/nics/overview.rst
+++ b/doc/guides/nics/overview.rst
@@ -94,7 +94,7 @@ Most of these differences are summarized below.
Queue start/stop Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y Y Y
MTU update Y Y Y Y Y Y Y Y Y Y
Jumbo frame Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y Y Y
- Scattered Rx Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y Y
+ Scattered Rx Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y Y
LRO Y Y Y Y
TSO Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Promiscuous mode Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y Y Y Y Y
diff --git a/drivers/net/enic/base/rq_enet_desc.h
b/drivers/net/enic/base/rq_enet_desc.h
index 7292d9d..13e24b4 100644
--- a/drivers/net/enic/base/rq_enet_desc.h
+++ b/drivers/net/enic/base/rq_enet_desc.h
@@ -55,7 +55,7 @@ enum rq_enet_type_types {
#define RQ_ENET_TYPE_BITS 2
#define RQ_ENET_TYPE_MASK ((1 << RQ_ENET_TYPE_BITS) - 1)
-static inline void rq_enet_desc_enc(struct rq_enet_desc *desc,
+static inline void rq_enet_desc_enc(volatile struct rq_enet_desc *desc,
u64 address, u8 type, u16 length)
{
desc->address = cpu_to_le64(address);
diff --git a/drivers/net/enic/base/vnic_rq.c b/drivers/net/enic/base/vnic_rq.c
index cb62c5e..0e700a1 100644
--- a/drivers/net/enic/base/vnic_rq.c
+++ b/drivers/net/enic/base/vnic_rq.c
@@ -84,11 +84,12 @@ void vnic_rq_init_start(struct vnic_rq *rq, unsigned int
cq_index,
iowrite32(cq_index, >ctrl->cq_index);
iowrite32(error_interrupt_enable, >ctrl->error_interrupt_enable);
iowrite32(error_interrupt_offset, >ctrl->error_interrupt_offset);
- iowrite32(0, >ctrl->dropped_packet_count);
iowrite32(0, >ctrl->error_status);
iowrite32(fetch_index, >ctrl->fetch_index);
iowrite32(posted_index, >ctrl->posted_index);
-
+ if (rq->is_sop)
+ iowrite32(((rq->is_sop << 10) | rq->data_queue_idx),
+ >ctrl->data_ring);
}
void vnic_rq_init(struct vnic_rq *rq, unsigned int cq_index,
@@ -96,6 +97,7 @@ void vnic_rq_init(struct vnic_rq *rq, unsigned int cq_index,
unsigned int error_interrupt_offset)
{
u32 fetch_index = 0;
+
/* Use current fetch_index as the ring starting point */
fetch_index = ioread32(>ctrl->fetch_index);
@@ -110,6 +112,8 @@ void vnic_rq_init(struct vnic_rq *rq, unsigned int cq_index,
error_interrupt_offset);
rq->rxst_idx = 0;
rq->tot_pkts = 0;
+ rq->pkt_first_seg = NULL;
+ rq->pkt_last_seg = NULL;
}
void vnic_rq_error_out(struct vnic_rq *rq, unsigned int error)
diff --git a/drivers/net/enic/base/vnic_rq.h b/drivers/net/enic/base/vnic_rq.h
index e083ccc..fd9e170 100644
--- a/drivers/net/enic/base/vnic_rq.h
+++ b/drivers/net/enic/base/vnic_rq.h
@@ -60,10 +60,18 @@ struct vnic_rq_ctrl {
u32 pad7;
u32 error_status; /* 0x48 */
u32 pad8;
- u32 dropped_packet_count; /* 0x50 */
+ u32 tcp_sn; /* 0x50 */
u32 pad9;
- u32 dropped_packet_count_rc;/* 0x58 */
+ u32 unused; /* 0x58 */
u32 pad10;
+ u32 dca_select; /* 0x60 */
+ u32 pad11;
+ u32 dca_value; /* 0x68
[dpdk-dev] [PATCH] enic: fix releasing of mbufs when tearing down rq
When trying to release the mbufs, the function was incorrectly
iterating over the max size configured instead of the actual size
of the ring.
Fixes: cbb44dddcade ("enic: receive path performance improvements")
Signed-off-by: Nelson Escobar
Reviewed-by: John Daley
---
drivers/net/enic/enic_main.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/enic/enic_main.c b/drivers/net/enic/enic_main.c
index 0576a6e..9b6fe36 100644
--- a/drivers/net/enic/enic_main.c
+++ b/drivers/net/enic/enic_main.c
@@ -80,7 +80,7 @@ static int is_eth_addr_valid(uint8_t *addr)
}
static void
-enic_rxmbuf_queue_release(struct enic *enic, struct vnic_rq *rq)
+enic_rxmbuf_queue_release(__rte_unused struct enic *enic, struct vnic_rq *rq)
{
uint16_t i;
@@ -89,7 +89,7 @@ enic_rxmbuf_queue_release(struct enic *enic, struct vnic_rq
*rq)
return;
}
- for (i = 0; i < enic->config.rq_desc_count; i++) {
+ for (i = 0; i < rq->ring.desc_count; i++) {
if (rq->mbuf_ring[i]) {
rte_pktmbuf_free_seg(rq->mbuf_ring[i]);
rq->mbuf_ring[i] = NULL;
--
2.7.0
[dpdk-dev] [PATCH] enic: fix free function to actually free memory
enic_alloc_consistent() allocated memory, but enic_free_consistent()
was an empty function, so allocated memory was never freed.
Fixes: fefed3d1e62c ("enic: new driver")
Signed-off-by: Nelson Escobar
Reviewed-by: John Daley
---
drivers/net/enic/base/vnic_dev.c | 14 +-
drivers/net/enic/base/vnic_dev.h | 2 +-
drivers/net/enic/enic.h | 12 +
drivers/net/enic/enic_main.c | 56 +---
4 files changed, 67 insertions(+), 17 deletions(-)
diff --git a/drivers/net/enic/base/vnic_dev.c b/drivers/net/enic/base/vnic_dev.c
index e8a5028..fc2e4cc 100644
--- a/drivers/net/enic/base/vnic_dev.c
+++ b/drivers/net/enic/base/vnic_dev.c
@@ -83,7 +83,7 @@ struct vnic_dev {
struct vnic_intr_coal_timer_info intr_coal_timer_info;
void *(*alloc_consistent)(void *priv, size_t size,
dma_addr_t *dma_handle, u8 *name);
- void (*free_consistent)(struct rte_pci_device *hwdev,
+ void (*free_consistent)(void *priv,
size_t size, void *vaddr,
dma_addr_t dma_handle);
};
@@ -101,7 +101,7 @@ void *vnic_dev_priv(struct vnic_dev *vdev)
void vnic_register_cbacks(struct vnic_dev *vdev,
void *(*alloc_consistent)(void *priv, size_t size,
dma_addr_t *dma_handle, u8 *name),
- void (*free_consistent)(struct rte_pci_device *hwdev,
+ void (*free_consistent)(void *priv,
size_t size, void *vaddr,
dma_addr_t dma_handle))
{
@@ -807,7 +807,7 @@ int vnic_dev_notify_unsetcmd(struct vnic_dev *vdev)
int vnic_dev_notify_unset(struct vnic_dev *vdev)
{
if (vdev->notify && !vnic_dev_in_reset(vdev)) {
- vdev->free_consistent(vdev->pdev,
+ vdev->free_consistent(vdev->priv,
sizeof(struct vnic_devcmd_notify),
vdev->notify,
vdev->notify_pa);
@@ -924,16 +924,16 @@ void vnic_dev_unregister(struct vnic_dev *vdev)
{
if (vdev) {
if (vdev->notify)
- vdev->free_consistent(vdev->pdev,
+ vdev->free_consistent(vdev->priv,
sizeof(struct vnic_devcmd_notify),
vdev->notify,
vdev->notify_pa);
if (vdev->stats)
- vdev->free_consistent(vdev->pdev,
+ vdev->free_consistent(vdev->priv,
sizeof(struct vnic_stats),
vdev->stats, vdev->stats_pa);
if (vdev->fw_info)
- vdev->free_consistent(vdev->pdev,
+ vdev->free_consistent(vdev->priv,
sizeof(struct vnic_devcmd_fw_info),
vdev->fw_info, vdev->fw_info_pa);
kfree(vdev);
@@ -1041,7 +1041,7 @@ int vnic_dev_classifier(struct vnic_dev *vdev, u8 cmd,
u16 *entry,
ret = vnic_dev_cmd(vdev, CMD_ADD_FILTER, , , wait);
*entry = (u16)a0;
- vdev->free_consistent(vdev->pdev, tlv_size, tlv_va, tlv_pa);
+ vdev->free_consistent(vdev->priv, tlv_size, tlv_va, tlv_pa);
} else if (cmd == CLSF_DEL) {
a0 = *entry;
ret = vnic_dev_cmd(vdev, CMD_DEL_FILTER, , , wait);
diff --git a/drivers/net/enic/base/vnic_dev.h b/drivers/net/enic/base/vnic_dev.h
index 113d6ac..689442f 100644
--- a/drivers/net/enic/base/vnic_dev.h
+++ b/drivers/net/enic/base/vnic_dev.h
@@ -102,7 +102,7 @@ unsigned int vnic_dev_get_res_count(struct vnic_dev *vdev,
void vnic_register_cbacks(struct vnic_dev *vdev,
void *(*alloc_consistent)(void *priv, size_t size,
dma_addr_t *dma_handle, u8 *name),
- void (*free_consistent)(struct rte_pci_device *hwdev,
+ void (*free_consistent)(void *priv,
size_t size, void *vaddr,
dma_addr_t dma_handle));
void __iomem *vnic_dev_get_res(struct vnic_dev *vdev, enum vnic_res_type type,
diff --git a/drivers/net/enic/enic.h b/drivers/net/enic/enic.h
index 1e6914e..9f94afb 100644
--- a/drivers/net/enic/enic.h
+++ b/drivers/net/enic/enic.h
@@ -46,6 +46,8 @@
#include "vnic_rss.h"
#include "enic_res.h"
#include "cq_enet_desc.h"
+#include
+#include
#define DRV_NAME "enic_pmd"
#define DRV_DESCRIPTION"Cisco VIC Ethernet NIC Poll-mode
Driver"
@@ -95,6 +97,11 @@ struct enic_soft_stats {
rte_atomic64_t rx_packet_errors;
};
+struct enic_memzone_entry {
+ const struct rte_memzone *rz;
+ LIST_ENTRY(enic_memzone_entry) entries;
+};
+
/* Per-instance private data structure */
struct enic {
struct enic *next;
@@ -140,6 +147,11 @@ struct enic {
/* software c
[dpdk-dev] [PATCH] enic: improve ptype handling
- add l4 ptypes to the ones we report as supporting
- report/use RTE_PTYPE_L3_IPV4_EXT_UNKNOWN and
RTE_PTYPE_L3_IPV6_EXT_UNKNOWN instead of RTE_PTYPE_L3_IPV4 and
RTE_PTYPE_L3_IPV6 as vic can't distinguish between packets with
extentions and those without extentions.
- correctly set the ptype bits set on packets that are both tcp/udp
and a frag
- set RTE_PTYPE_L4_NONFRAG on ip packets we know are not udp, tcp,
or fragments.
Fixes: 947d860c821f ("enic: improve Rx performance")
Signed-off-by: Nelson Escobar
Reviewed-by: John Daley
---
drivers/net/enic/enic_ethdev.c | 8 ++--
drivers/net/enic/enic_rxtx.c | 39 +--
2 files changed, 31 insertions(+), 16 deletions(-)
diff --git a/drivers/net/enic/enic_ethdev.c b/drivers/net/enic/enic_ethdev.c
index 697ff82..003dec0 100644
--- a/drivers/net/enic/enic_ethdev.c
+++ b/drivers/net/enic/enic_ethdev.c
@@ -455,8 +455,12 @@ static void enicpmd_dev_info_get(struct rte_eth_dev
*eth_dev,
static const uint32_t *enicpmd_dev_supported_ptypes_get(struct rte_eth_dev
*dev)
{
static const uint32_t ptypes[] = {
- RTE_PTYPE_L3_IPV4,
- RTE_PTYPE_L3_IPV6,
+ RTE_PTYPE_L3_IPV4_EXT_UNKNOWN,
+ RTE_PTYPE_L3_IPV6_EXT_UNKNOWN,
+ RTE_PTYPE_L4_TCP,
+ RTE_PTYPE_L4_UDP,
+ RTE_PTYPE_L4_FRAG,
+ RTE_PTYPE_L4_NONFRAG,
RTE_PTYPE_UNKNOWN
};
diff --git a/drivers/net/enic/enic_rxtx.c b/drivers/net/enic/enic_rxtx.c
index 8fe334f..972eae2 100644
--- a/drivers/net/enic/enic_rxtx.c
+++ b/drivers/net/enic/enic_rxtx.c
@@ -148,20 +148,31 @@ enic_cq_rx_flags_to_pkt_type(struct cq_desc *cqd)
struct cq_enet_rq_desc *cqrd = (struct cq_enet_rq_desc *)cqd;
uint8_t cqrd_flags = cqrd->flags;
static const uint32_t cq_type_table[128] __rte_cache_aligned = {
- [32] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4,
- [34] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4
- | RTE_PTYPE_L4_UDP,
- [36] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4
- | RTE_PTYPE_L4_TCP,
- [96] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4
- | RTE_PTYPE_L4_FRAG,
- [16] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6,
- [18] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6
- | RTE_PTYPE_L4_UDP,
- [20] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6
- | RTE_PTYPE_L4_TCP,
- [80] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6
- | RTE_PTYPE_L4_FRAG,
+ [0x00] = RTE_PTYPE_UNKNOWN,
+ [0x20] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN
+ | RTE_PTYPE_L4_NONFRAG,
+ [0x22] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN
+ | RTE_PTYPE_L4_UDP,
+ [0x24] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN
+ | RTE_PTYPE_L4_TCP,
+ [0x60] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN
+ | RTE_PTYPE_L4_FRAG,
+ [0x62] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN
+ | RTE_PTYPE_L4_UDP,
+ [0x64] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN
+ | RTE_PTYPE_L4_TCP,
+ [0x10] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN
+ | RTE_PTYPE_L4_NONFRAG,
+ [0x12] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN
+ | RTE_PTYPE_L4_UDP,
+ [0x14] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN
+ | RTE_PTYPE_L4_TCP,
+ [0x50] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN
+ | RTE_PTYPE_L4_FRAG,
+ [0x52] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN
+ | RTE_PTYPE_L4_UDP,
+ [0x54] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN
+ | RTE_PTYPE_L4_TCP,
/* All others reserved */
};
cqrd_flags &= CQ_ENET_RQ_DESC_FLAGS_IPV4_FRAGMENT
--
2.7.0
[dpdk-dev] [PATCH] enic: fix name of hash table used for enic classifiers
The enic_clsf_init() function is called once per enic instance, but it
used a static name to create the hash table. Consequently when using
more than one enic instance, there was a name collision which caused
errors:
EAL: memzone_reserve_aligned_thread_unsafe():
memzone already exists
RING: Cannot reserve memory
HASH: memory allocation failed
PMD: rte_enic_pmd: Init of hash table for clsf failed.
Flow director feature will not work
This patch changes the name to be unique per enic instance.
Fixes: fefed3d1e62c ("enic: new driver")
Signed-off-by: Nelson Escobar
Reviewed-by: John Daley
---
drivers/net/enic/enic_clsf.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/drivers/net/enic/enic_clsf.c b/drivers/net/enic/enic_clsf.c
index edb56e1..2ec77f5 100644
--- a/drivers/net/enic/enic_clsf.c
+++ b/drivers/net/enic/enic_clsf.c
@@ -239,15 +239,16 @@ void enic_clsf_destroy(struct enic *enic)
int enic_clsf_init(struct enic *enic)
{
+ char clsf_name[RTE_HASH_NAMESIZE];
struct rte_hash_parameters hash_params = {
- .name = "enicpmd_clsf_hash",
+ .name = clsf_name,
.entries = ENICPMD_CLSF_HASH_ENTRIES,
.key_len = sizeof(struct rte_eth_fdir_filter),
.hash_func = DEFAULT_HASH_FUNC,
.hash_func_init_val = 0,
.socket_id = SOCKET_ID_ANY,
};
-
+ sprintf(clsf_name, "enic_clsf_%s", enic->bdf_name);
enic->fdir.hash = rte_hash_create(_params);
memset(>fdir.stats, 0, sizeof(enic->fdir.stats));
enic->fdir.stats.free = ENICPMD_FDIR_MAX;
--
2.7.0
[dpdk-dev] [PATCH] bonding: fix bond link detect in non-interrupt mode
Thomas,
Sorry, I should have included the following fixes line:
Fixes: a45b288ef21a("bond: support link status polling")
Please add it to the commit message when applying if there are no other
problems with the patch.
Nelson.
On 3/31/2016 6:40 AM, Thomas Monjalon wrote:
> 2016-03-25 17:44, John Daley:
>> From: Nelson Escobar
>>
>> Stopping then re-starting a bond interface containing slaves that
>> used polling for link detection caused the bond to think all slave
>> links were down and inactive.
>>
>> Move the start of the polling for link from slave_add() to
>> bond_ethdev_start() and in bond_ethdev_stop() make sure we clear
>> the last_link_status of the slaves.
>>
>> Signed-off-by: Nelson Escobar
>> Signed-off-by: John Daley
>
> A "Fixes:" line would be appreciated to know the origin of the bug.
> Thanks
>
