启动qnsm-inspect后出错

[peiyuefeng]

启动命令：“./qnsm-inspect -f ./qnsm_inspect.cfg -c . -p 1”
错误信息：

[APP] Initializing CPU core map ...
[APP] CPU core mask = 0x0000000000000000000000000000003f
[APP] Initializing EAL ...
EAL: Detected 7 lcore(s)
EAL: Probing VFIO support...
EAL: PCI device 0000:02:01.0 on NUMA socket -1
EAL: probe driver: 8086:100f net_e1000_em
EAL: PCI device 0000:02:04.0 on NUMA socket -1
EAL: probe driver: 8086:100f net_e1000_em
EAL: PCI device 0000:02:05.0 on NUMA socket -1
EAL: probe driver: 8086:100f net_e1000_em
[APP] Initializing MEMPOOL0 ...
[APP] Initializing MEMPOOL1 ...
[APP] Initializing LINK0 (1) (1 RXQ, 0 TXQ) ...
PANIC in app_init_link():
LINK0 (0): init error (-95)
6: [./qnsm-inspect() [0x40f4fd]]
5: [/lib64/libc.so.6(__libc_start_main+0xf5) [0x7f4756b7bb15]]
4: [./qnsm-inspect() [0x40f3c3]]
3: [./qnsm-inspect() [0x579ab0]]
2: [./qnsm-inspect() [0x40d175]]
1: [./qnsm-inspect() [0x4dc66a]]
Aborted (core dumped)

环境： vmware , 3网卡(intel e1000)，采用NAT模式， 0网卡做管理 ，仅绑定1网卡，做dpdk测试。
qnsm_inspect.cfg如下

[EAL]
log_level = 8
n = 1
socket_mem = 1024
master_lcore = 0
[IDPS]
conf_file = ./suricata.yaml
;mbuf mempool cfg
;add mbuf priavte size para
[MEMPOOL0]
buffer_size = 2304
pool_size = 131072
cache_size = 256
cpu = 0 ;socket_id
private_size = 64 ;sizeof(QNSM_PACKET_INFO)
;for ids
[MEMPOOL1]
buffer_size = 2304
pool_size = 131072
cache_size = 256
cpu = 0
private_size = 64
;link cfg
;[LINK0]
;rss_qs = 0
;rss_proto_ipv4 = TCP UDP
;rss_proto_ipv6 = TCP TCP_EX UDP UDP_EX
;symmetrical_rss = yes
;ip_local_q = 7 reserved for future proto stack app
;arp_q = 8
;rx queue cfg
;http://dpdk.org/doc/guides/nics/ixgbe.html
[RXQ0.0]
size = 2048
burst = 32
[SWQ1]
size = 2048
cpu = 1
mempool = MEMPOOL1
dup = yes
;app cfg
[PIPELINE0]
type = MASTER
core = s0c0
[PIPELINE1]
type = SESSM
core = s0c1
pktq_in = RXQ0.0
pktq_out = SWQ0 SWQ1
timer_period = 10
[PIPELINE2]
type = SIP_IN_AGG
core = s0c2
[PIPELINE3]
type = VIP_AGG
core = s0c3
pktq_in = SWQ0
timer_period = 10
[PIPELINE4]
type = EDGE
core = s0c4
;IPS BEGIN
[PIPELINE5]
type = DETECT
core = s0c5
pktq_in = SWQ1
;IPS END

dpdk16.11.2 中，已经 针对vmware做了patch

sed -i "s/pci_intx_mask_supported(dev)/pci_intx_mask_supported(dev)||true/g" ${RTE_SDK}/lib/librte_eal/linuxapp/igb_uio/igb_uio.c

如果打开qnsm_inspect.cfg中的LINK0的rss配置

[LINK0]
rss_qs = 0
rss_proto_ipv4 = TCP UDP
rss_proto_ipv6 = TCP TCP_EX UDP UDP_EX
symmetrical_rss = yes

出错信息变成了

PANIC in app_link_rss_setup():
LINK0 (0): RSS setup error (null RETA size)
7: [./qnsm-inspect() [0x40f4fd]]
6: [/lib64/libc.so.6(__libc_start_main+0xf5) [0x7f703d1b0b15]]
5: [./qnsm-inspect() [0x40f3c3]]
4: [./qnsm-inspect() [0x5797f8]]
3: [./qnsm-inspect() [0x57729d]]
2: [./qnsm-inspect() [0x40d175]]
1: [./qnsm-inspect() [0x4dc66a]]

[CosmosSun]

@peiyuefeng
您好，qnsm还没有在vmware环境上测试；
从rss的报错看，是dev_info.reta_size为0，确定是e1000网卡吗？

RSS is working as per above but only the E1000, but not for VMXNET3.
具体可以参考：https://communities.vmware.com/thread/540391

另外从dpdk的pmd代码看：
e1000 pmd是支持rss的，eth_igb_infos_get
dev_info->hash_key_size = IGB_HKEY_MAX_INDEX * sizeof(uint32_t); dev_info->reta_size = ETH_RSS_RETA_SIZE_128; dev_info->flow_type_rss_offloads = IGB_RSS_OFFLOAD_ALL;

vmxnet3没有rss，vmxnet3_dev_info_get。

[peiyuefeng]

thx much, i will check that again.

[mflu]

Close it!

[ZHAN-MQ]

Same issue with VirtualBox, and the NIC is e1000