AuxXxilium/linux_dsm_epyc7002
1
0
Fork 0
mirror of https://github.com/AuxXxilium/linux_dsm_epyc7002.git synced 2024-12-17 10:52:26 +07:00
Code Issues Actions Packages Projects Releases Wiki Activity

mlx4: avoid unnecessary dirtying of critical fields

Browse Source 
While stressing a 40Gbit mlx4 NIC with busy polling, I found false
sharing in mlx4 driver that can be easily avoided.

This patch brings an additional 7 % performance improvement in UDP_RR
workload.

1) If we received no frame during one mlx4_en_process_rx_cq()
   invocation, no need to call mlx4_cq_set_ci() and/or dirty ring->cons

2) Do not refill rx buffers if we have plenty of them.
   This avoids false sharing and allows some bulk/batch optimizations.
   Page allocator and its locks will thank us.

Finally, mlx4_en_poll_rx_cq() should not return 0 if it determined
cpu handling NIC IRQ should be changed. We should return budget-1
instead, to not fool net_rx_action() and its netdev_budget.

v2: keep AVG_PERF_COUNTER(... polled) even if polled is 0

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Tariq Toukan <tariqt@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit is contained in:
Eric Dumazet 2016-11-20 09:24:36 -08:00 committed by David S. Miller
parent b668534c1d
commit dad42c3038
1 changed files with 30 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

47
drivers/net/ethernet/mellanox/mlx4/en_rx.c
View File

@ -688,18 +688,23 @@ static void validate_loopback(struct mlx4_en_priv *priv, struct sk_buff *skb)
dev_kfree_skb_any(skb); dev_kfree_skb_any(skb);
} }
static void mlx4_en_refill_rx_buffers(struct mlx4_en_priv *priv, static bool mlx4_en_refill_rx_buffers(struct mlx4_en_priv *priv,
struct mlx4_en_rx_ring *ring) struct mlx4_en_rx_ring *ring)
{ {
int index = ring->prod & ring->size_mask; u32 missing = ring->actual_size - (ring->prod - ring->cons);
while ((u32) (ring->prod - ring->cons) < ring->actual_size) { /* Try to batch allocations, but not too much. */
if (mlx4_en_prepare_rx_desc(priv, ring, index, if (missing < 8)
return false;
do {
if (mlx4_en_prepare_rx_desc(priv, ring,
ring->prod & ring->size_mask,
GFP_ATOMIC | __GFP_COLD)) GFP_ATOMIC | __GFP_COLD))
break; break;
ring->prod++; ring->prod++;
index = ring->prod & ring->size_mask; } while (--missing);
}
return true;
} }
/* When hardware doesn't strip the vlan, we need to calculate the checksum /* When hardware doesn't strip the vlan, we need to calculate the checksum
@ -1081,15 +1086,20 @@ int mlx4_en_process_rx_cq(struct net_device *dev, struct mlx4_en_cq *cq, int bud
out: out:
rcu_read_unlock(); rcu_read_unlock();
if (doorbell_pending)
mlx4_en_xmit_doorbell(priv->tx_ring[TX_XDP][cq->ring]);
if (polled) {
if (doorbell_pending)
mlx4_en_xmit_doorbell(priv->tx_ring[TX_XDP][cq->ring]);
mlx4_cq_set_ci(&cq->mcq);
wmb(); /* ensure HW sees CQ consumer before we post new buffers */
ring->cons = cq->mcq.cons_index;
}
AVG_PERF_COUNTER(priv->pstats.rx_coal_avg, polled); AVG_PERF_COUNTER(priv->pstats.rx_coal_avg, polled);
mlx4_cq_set_ci(&cq->mcq);
wmb(); /* ensure HW sees CQ consumer before we post new buffers */ if (mlx4_en_refill_rx_buffers(priv, ring))
ring->cons = cq->mcq.cons_index; mlx4_en_update_rx_prod_db(ring);
mlx4_en_refill_rx_buffers(priv, ring);
mlx4_en_update_rx_prod_db(ring);
return polled; return polled;
} }
@ -1131,10 +1141,13 @@ int mlx4_en_poll_rx_cq(struct napi_struct *napi, int budget)
return budget; return budget;
/* Current cpu is not according to smp_irq_affinity - /* Current cpu is not according to smp_irq_affinity -
* probably affinity changed. need to stop this NAPI * probably affinity changed. Need to stop this NAPI
* poll, and restart it on the right CPU * poll, and restart it on the right CPU.
* Try to avoid returning a too small value (like 0),
* to not fool net_rx_action() and its netdev_budget
*/ */
done = 0; if (done)
done--;
} }
/* Done for now */ /* Done for now */
if (napi_complete_done(napi, done)) if (napi_complete_done(napi, done))