我已经在运行CentOS 3.10的大学提供的节点上成功编译并安装了netmap。
我试图使用示例应用程序pkt-gen来生成和接收数据包，以测量产生的系统开销，即在线路和应用程序之间移动数据包的时间。我第一次运行pkt-gen它运行正常。从那时起，当它运行时，pkt-gen首先发送一些数据包然后不发送任何导致后续打印：
[root@netzero4 pkt-gen]# ./pkt-gen -i enp4s0f0 -f tx
097.483435 main [2736] interface is enp4s0f0
097.483523 main [2861] using default burst size: 512
097.483911 main [2869] running on 1 cpus (have 8)
097.484431 extract_ip_range [464] range is 10.0.0.1:1234 to 10.0.0.1:1234
097.484455 extract_ip_range [464] range is 10.1.0.1:1234 to 10.1.0.1:1234
097.484862 nm_open [855] overriding ARG1 0
097.484880 nm_open [859] overriding ARG2 0
097.484890 nm_open [863] overriding ARG3 0
097.484900 nm_open [867] overriding RING_CFG
097.484910 nm_open [876] overriding ifname enp4s0f0 ringid 0x0 flags 0x8001
097.699352 main [2968] mapped 334980KB at 0x7fe8c80ad000
Sending on netmap:enp4s0f0: 8 queues, 1 threads and 1 cpus.
10.0.0.1 -> 10.1.0.1 (00:00:00:00:00:00 -> ff:ff:ff:ff:ff:ff)
097.699424 main [3065] Sending 512 packets every 0.000000000 s
097.699456 start_threads [2408] Wait 2 secs for phy reset
099.699618 start_threads [2410] Ready...
099.699784 sender_body [1519] start, fd 3 main_fd 3
100.700333 main_thread [2498] 4.085 Kpps (4.088 Kpkts 1.962 Mbps in 1000631 usec) 511.00 avg_batch 0 min_space
101.701306 main_thread [2498] 0.000 pps (0.000 pkts 0.000 bps in 1000973 usec) 0.00 avg_batch 99999 min_space
101.705783 sender_body [1588] poll error/timeout on queue 0: Success
102.702387 main_thread [2498] 0.000 pps (0.000 pkts 0.000 bps in 1001081 usec) 0.00 avg_batch 99999 min_space
103.703483 main_thread [2498] 0.000 pps (0.000 pkts 0.000 bps in 1001096 usec) 0.00 avg_batch 99999 min_space
103.707902 sender_body [1588] poll error/timeout on queue 0: Success
104.704579 main_thread [2498] 0.000 pps (0.000 pkts 0.000 bps in 1001095 usec) 0.00 avg_batch 99999 min_space
105.705686 main_thread [2498] 0.000 pps (0.000 pkts 0.000 bps in 1001108 usec) 0.00 avg_batch 99999 min_space
105.710016 sender_body [1588] poll error/timeout on queue 0: Success
106.706773 main_thread [2498] 0.000 pps (0.000 pkts 0.000 bps in 1001087 usec) 0.00 avg_batch 99999 min_space
107.707874 main_thread [2498] 0.000 pps (0.000 pkts 0.000 bps in 1001101 usec) 0.00 avg_batch 99999 min_space
107.712132 sender_body [1588] poll error/timeout on queue 0: Success
108.708973 main_thread [2498] 0.000 pps (0.000 pkts 0.000 bps in 1001099 usec) 0.00 avg_batch 99999 min_space
^C109.028581 sigint_h [519] received control-C on thread 0x7fe8dd4be740
109.028633 main_thread [2498] 0.000 pps (0.000 pkts 0.000 bps in 319662 usec) 0.00 avg_batch 99999 min_space
109.714223 sender_body [1645] flush tail 511 head 511 on thread 0x7fe8c80ac700
110.029723 main_thread [2498] 0.000 pps (0.000 pkts 0.000 bps in 1001088 usec) 0.00 avg_batch 99999 min_space
Sent 4088 packets 245280 bytes 8 events 60 bytes each in 10.01 seconds.
Speed: 408.208 pps Bandwidth: 195.940 Kbps (raw 274.316 Kbps). Average batch: 511.00 pkts
我也尝试过为netmap提供全新安装，但结果是一样的。我做了一些研究，并且显然这个问题是因为戒指缓冲区出现了问题。但是根据this帖子它已经修复了。我不太熟悉使用Linux，但可能是因为NIC存在某种问题吗？