d/freebsd-nq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Remove RSS support in ENA

Browse Source 
Receive Side Scaling is optional feature that could be enabled in kernel
configuration by defining flag RSS.

Kernel uses hash to store and find protocol control block which is
stored in hash tables.
Kernel and NIC hash functions must be consistent. Otherwise case lookup
fails.

To achieve this kernel provides API to set proper hash key to NIC.
As it is not possible to change key for virtual ENA NIC, this driver
cannot support RSS function.

ENA is designed to work in virtual environments so supporting hardware
version of this card is unnecessary.

Submitted by:  Rafal Kozik <rk@semihalf.com>
Obtained from: Semihalf
Sponsored by:  Amazon, Inc.
This commit is contained in:
Marcin Wojtas 2019-05-30 13:12:14 +00:00
parent 40621d71fd
commit 277f11c401
1 changed files with 3 additions and 53 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

56
sys/dev/ena/ena.c
View File

@ -59,11 +59,9 @@ __FBSDID("$FreeBSD$");
#include <net/if_arp.h>
#include <net/if_dl.h>
#include <net/if_media.h>
#include <net/rss_config.h>
#include <net/if_types.h>
#include <net/if_vlan_var.h>
#include <netinet/in_rss.h>
#include <netinet/in_systm.h>
#include <netinet/in.h>
#include <netinet/if_ether.h>
@ -585,9 +583,6 @@ ena_setup_tx_resources(struct ena_adapter *adapter, int qid)
struct ena_que *que = &adapter->que[qid];
struct ena_ring *tx_ring = que->tx_ring;
int size, i, err;
#ifdef RSS
cpuset_t cpu_mask;
#endif
size = sizeof(struct ena_tx_buffer) * tx_ring->ring_size;
@ -638,16 +633,8 @@ ena_setup_tx_resources(struct ena_adapter *adapter, int qid)
goto err_buf_info_unmap;
}
/* RSS set cpu for thread */
#ifdef RSS
CPU_SETOF(que->cpu, &cpu_mask);
taskqueue_start_threads_cpuset(&tx_ring->enqueue_tq, 1, PI_NET,
&cpu_mask, "%s tx_ring enq (bucket %d)",
device_get_nameunit(adapter->pdev), que->cpu);
#else /* RSS */
taskqueue_start_threads(&tx_ring->enqueue_tq, 1, PI_NET,
"%s txeq %d", device_get_nameunit(adapter->pdev), que->cpu);
#endif /* RSS */
return (0);
@ -780,9 +767,6 @@ ena_setup_rx_resources(struct ena_adapter *adapter, unsigned int qid)
struct ena_que *que = &adapter->que[qid];
struct ena_ring *rx_ring = que->rx_ring;
int size, err, i;
#ifdef RSS
cpuset_t cpu_mask;
#endif
size = sizeof(struct ena_rx_buffer) * rx_ring->ring_size;
@ -836,16 +820,8 @@ ena_setup_rx_resources(struct ena_adapter *adapter, unsigned int qid)
rx_ring->cmpl_tq = taskqueue_create_fast("ena RX completion", M_WAITOK,
taskqueue_thread_enqueue, &rx_ring->cmpl_tq);
/* RSS set cpu for thread */
#ifdef RSS
CPU_SETOF(que->cpu, &cpu_mask);
taskqueue_start_threads_cpuset(&rx_ring->cmpl_tq, 1, PI_NET, &cpu_mask,
"%s rx_ring cmpl (bucket %d)",
device_get_nameunit(adapter->pdev), que->cpu);
#else
taskqueue_start_threads(&rx_ring->cmpl_tq, 1, PI_NET,
"%s rx_ring cmpl %d", device_get_nameunit(adapter->pdev), que->cpu);
#endif
return (0);
@ -1908,12 +1884,9 @@ ena_setup_io_intr(struct ena_adapter *adapter)
adapter->msix_entries[irq_idx].vector;
ena_trace(ENA_INFO | ENA_IOQ, "ena_setup_io_intr vector: %d\n",
adapter->msix_entries[irq_idx].vector);
#ifdef RSS
adapter->que[i].cpu = adapter->irq_tbl[irq_idx].cpu =
rss_getcpu(i % rss_getnumbuckets());
#else
/*
* We still want to bind rings to the corresponding cpu
* We want to bind rings to the corresponding cpu
* using something similar to the RSS round-robin technique.
*/
if (unlikely(last_bind_cpu < 0))
@ -1921,7 +1894,6 @@ ena_setup_io_intr(struct ena_adapter *adapter)
adapter->que[i].cpu = adapter->irq_tbl[irq_idx].cpu =
last_bind_cpu;
last_bind_cpu = CPU_NEXT(last_bind_cpu);
#endif
}
}
@ -2010,13 +1982,8 @@ ena_request_io_irq(struct ena_adapter *adapter)
}
irq->requested = true;
#ifdef RSS
ena_trace(ENA_INFO, "queue %d - RSS bucket %d\n",
i - ENA_IO_IRQ_FIRST_IDX, irq->cpu);
#else
ena_trace(ENA_INFO, "queue %d - cpu %d\n",
i - ENA_IO_IRQ_FIRST_IDX, irq->cpu);
#endif
}
return (rc);
@ -2952,16 +2919,7 @@ ena_mq_start(if_t ifp, struct mbuf *m)
* It should improve performance.
*/
if (M_HASHTYPE_GET(m) != M_HASHTYPE_NONE) {
#ifdef RSS
if (rss_hash2bucket(m->m_pkthdr.flowid,
M_HASHTYPE_GET(m), &i) == 0) {
i = i % adapter->num_queues;
} else
#endif
{
i = m->m_pkthdr.flowid % adapter->num_queues;
}
i = m->m_pkthdr.flowid % adapter->num_queues;
} else {
i = curcpu % adapter->num_queues;
}
@ -3017,9 +2975,6 @@ ena_calc_io_queue_num(struct ena_adapter *adapter,
/* 1 IRQ for for mgmnt and 1 IRQ for each TX/RX pair */
io_queue_num = min_t(int, io_queue_num,
pci_msix_count(adapter->pdev) - 1);
#ifdef RSS
io_queue_num = min_t(int, io_queue_num, rss_getnumbuckets());
#endif
return (io_queue_num);
}
@ -3077,12 +3032,7 @@ ena_rss_init_default(struct ena_adapter *adapter)
}
for (i = 0; i < ENA_RX_RSS_TABLE_SIZE; i++) {
#ifdef RSS
qid = rss_get_indirection_to_bucket(i);
qid = qid % adapter->num_queues;
#else
qid = i % adapter->num_queues;
#endif
rc = ena_com_indirect_table_fill_entry(ena_dev, i,
ENA_IO_RXQ_IDX(qid));
if (unlikely((rc != 0) && (rc != EOPNOTSUPP))) {