netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* automatically decrease sender throughput when network traffic is heavy?
@ 2011-04-21  6:50 zhou rui
  0 siblings, 0 replies; only message in thread
From: zhou rui @ 2011-04-21  6:50 UTC (permalink / raw)
  To: netdev

hi
meet a strange situation:

usually throughput between sender(tcpreplay)  ->  receiver(my app) is
about 800~900Mbps(different machine,1G nic)

my kernel module will do a complicated work when it receive a packet,
I register a hook func like this:

 prot_hook.func = my_packet_rcv;
 prot_hook.type = htons(ETH_P_ALL);
 dev_add_pack(&prot_hook);

I suppose the nic driver will drop packets when traffic is heavy,
however, surprised that the sender throught is dreased to 500-600Mbps
why? what happened?

top - 14:00:46 up 5 days, 23:25,  3 users,  load average: 1.20, 0.44, 0.64
Tasks: 126 total,   5 running, 121 sleeping,   0 stopped,   0 zombie
Cpu(s):  1.9%us,  8.9%sy,  0.0%ni, 76.5%id,  0.0%wa,  0.0%hi, 12.6%si,  0.0%st
Mem:     16077M total,     3427M used,    12650M free,       87M buffers
Swap:     2055M total,        0M used,     2055M free,      850M cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
   25 root      20   0     0    0    0 R   98  0.0  10:33.29 ksoftirqd/6


the kernel is 2.6.36.4-0.7-default #5 SMP Thu Apr 14 15:37:53 SGT 2011
x86_64 x86_64 x86_64 GNU/Linux

before that same test was done on kernel 2.6.16 and no this problem seen

thanks
rui

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2011-04-21  6:50 UTC | newest]

Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2011-04-21  6:50 automatically decrease sender throughput when network traffic is heavy? zhou rui

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).