https://bugs.dpdk.org/show_bug.cgi?id=1393
Bug ID: 1393
Summary: examples/l3fwd: in eventmode RX queue setup doesn't
obey numa
Product: DPDK
Version: 23.11
Hardware: All
OS: All
Status: UNCONFIRMED
Severity: normal
Priority: Normal
Component: examples
Assignee: dev@dpdk.org
Reporter: konstantin.v.anan...@yandex.ru
Target Milestone: ---
Reproducible with latest DPDK main branch (24.03.rc1)
Starting l3fwd in event mode with all cores/NICs on numa-socket #1 will fail:
dpdk-l3fwd --lcores=49,51,53,57 -n 6 \
-a ca:00.0 -a ca:00.1 -a cb:00.0 -a cb:00.1 -s 0x8000000000000 --vdev event_sw0
-- \
-L -P -p f --rx-queue-size 1024 --tx-queue-size 1024 --mode eventdev \
--eventq-sched=ordered --rule_ipv4=test/l3fwd_lpm_v4_u1.cfg \
--rule_ipv6=test/l3fwd_lpm_v6_u1.cfg
....
...
ETHDEV: Ambiguous Rx mempools configuration
EAL: Error - exiting with code: 1
Cause: rte_eth_rx_queue_setup: err=-22, port=0, eth_qid: 0
Reason for that:
all lcores I specified are from numa-socket #1, so there is no mempool for
numa-socket #0, while in event mode l3fwd_eth_dev_port_setup() *always* tries
to use mempool for numa-socket 0.
examples/l3fwd/l3fwd_event.c:
static void
l3fwd_eth_dev_port_setup(struct rte_eth_conf *port_conf)
{
...
for (eth_qid = 0; eth_qid < evt_rsrc->eth_rx_queues;
eth_qid++) {
if (!evt_rsrc->per_port_pool)
ret = rte_eth_rx_queue_setup(port_id, eth_qid,
nb_rxd, 0, &rxconf,
evt_rsrc->pkt_pool[0][0]);
else
ret = rte_eth_rx_queue_setup(port_id, eth_qid,
nb_rxd, 0, &rxconf,
evt_rsrc->pkt_pool[port_id][0]);
For proper numa support, I expect it to be something like:
int socket_id = l3fwd_event_deduce_rx_service_socket_id(...);
ret = rte_eth_rx_queue_setup(..., evt_rsrc->pkt_pool[port_id][socket_id]);
To overcome the problem, use '--no-numa' l3fwd cmd-line flag.
--
You are receiving this mail because:
You are the assignee for the bug.
