Bug 1127
Summary: | net/i40e does not preserve RSS redirection table upon port stop/start | ||
---|---|---|---|
Product: | DPDK | Reporter: | Andrew Rybchenko (andrew.rybchenko) |
Component: | ethdev | Assignee: | dev |
Status: | UNCONFIRMED --- | ||
Severity: | normal | CC: | ivan.malov, kaiwenx.deng |
Priority: | Normal | ||
Version: | 22.07 | ||
Target Milestone: | --- | ||
Hardware: | All | ||
OS: | All |
Description
Andrew Rybchenko
2022-11-17 09:09:35 CET
One more point related to the bug in fact. i40e reports invalid RETA table just after port configure (without start) - it pretends to have 64 Rx queues whereas ethdev is already configured to have another number of Rx queues (see [1] line 27): rte_eth_dev_configure(0, 1, 1, { link_speeds=0, rxmode={ mq_mode=RSS, mtu=0, offloads=, flags=HW_STRIP_CRC }, txmode={ mq_mode=NONE, offloads=, pvid=0, flags= }, lbpk_mode=0, rx_conf_adv={ rss_conf={rss_key=, rss_key_len=0, rss_hf=} }, dcb_cap_en=0, intr_conf={ lsc=0, rxq=0 } }) -> 0 rte_eth_dev_rss_reta_query(0, 0x55d83fdddda0, 512) -> 0, reta_conf={ mask=ffffffffffffffff, reta= 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 mask=ffffffffffffffff, reta= 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 mask=ffffffffffffffff, reta= 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 mask=ffffffffffffffff, reta= 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 mask=ffffffffffffffff, reta= 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 mask=ffffffffffffffff, reta= 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 mask=ffffffffffffffff, reta= 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 mask=ffffffffffffffff, reta= 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 } testpmd should be capable to reproduce the problem as well. I think that the problem is related since the bug will be fixed as soon as default RETA is initialized on port configure and not reinitialized on start. [1] https://ts-factory.io/bublik/v2/log/189828?focusId=190422&mode=treeAndinfoAndlog please offer the reproduce steps with standard DPDK API like testpmd, l3fwd and other examples. |