From: Achiad Shochat <ach...@mellanox.com>
Under heavy network traffic load we may be in NAPI poll mode for
a long time, thus IRQ affinity changes do not affect NAPI as desired.
Now once in a NAPI poll session we check the channel IRQ affinity,
if it points to a different CPU core than the current, stop polling
and the following device interrupt will fix the NAPI affinity.
Signed-off-by: Achiad Shochat <ach...@mellanox.com>
Signed-off-by: Or Gerlitz <ogerl...@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en.h | 3 +++
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 1 +
drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c | 12 +++++++++++-
3 files changed, 15 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h
b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index 0983a20..3c5c900 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -406,6 +406,9 @@ struct mlx5e_channel {
u8 num_tc;
unsigned long flags;
+ /* data path - accessed per napi poll */
+ struct irq_desc *irq_desc;
+
/* control */
struct mlx5e_priv *priv;
int ix;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 0bab33c..bdbef9f 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -1004,6 +1004,7 @@ static int mlx5e_open_channel(struct mlx5e_priv *priv,
int ix,
goto err_close_sqs;
netif_set_xps_queue(netdev, get_cpu_mask(c->cpu), ix);
+ c->irq_desc = irq_to_desc(c->rq.cq.mcq.irqn);
*cp = c;
return 0;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
index 2c7cb67..99e29c9 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
@@ -30,6 +30,7 @@
* SOFTWARE.
*/
+#include <linux/irq.h>
#include "en.h"
struct mlx5_cqe64 *mlx5e_get_cqe(struct mlx5e_cq *cq)
@@ -49,6 +50,15 @@ struct mlx5_cqe64 *mlx5e_get_cqe(struct mlx5e_cq *cq)
return cqe;
}
+static inline bool mlx5e_no_channel_affinity_change(struct mlx5e_channel *c)
+{
+ int current_cpu = smp_processor_id();
+ struct irq_data *d = irq_desc_get_irq_data(c->irq_desc);
+ struct cpumask *aff = irq_data_get_affinity_mask(d);
+
+ return cpumask_test_cpu(current_cpu, aff);
+}
+
int mlx5e_napi_poll(struct napi_struct *napi, int budget)
{
struct mlx5e_channel *c = container_of(napi, struct mlx5e_channel,
@@ -65,7 +75,7 @@ int mlx5e_napi_poll(struct napi_struct *napi, int budget)
busy |= mlx5e_post_rx_wqes(&c->rq);
- if (busy)
+ if (busy && likely(mlx5e_no_channel_affinity_change(c)))
return budget;
napi_complete(napi);
--
2.3.7
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
