Troubles with em on FreeBSD 7

Hi!

I'm running a SMP FreeBSD box with mpd5 on it.

# uname -a
FreeBSD xxx.xxxxxxxxx.xxx 7.0-STABLE FreeBSD 7.0-STABLE #0: Sat May 3 12:40:02 EEST 2008 xxxxx@xxxxxxxxxxxxxxxxx:/usr/obj/usr/src/sys/XXXX amd64

# mpd5 -v
Version 5.1 (root@xxxxxxxxxxxxxxxxx 09:53 1-May-2008)

Somehow em0 begins to eat all CPU time of one core.

# top -S
last pid: 55827; load averages: 3.76, 3.42, 3.08 up 0+03:27:38 16:24:20
104 processes: 11 running, 81 sleeping, 12 waiting
CPU states: 1.7% user, 0.0% nice, 21.4% system, 3.0% interrupt, 73.9% idle
Mem: 71M Active, 89M Inact, 340M Wired, 336K Cache, 214M Buf, 7418M Free
Swap: 4096M Total, 4096M Free

PID USERNAME THR PRI NICE SIZE RES STATE C TIME WCPU COMMAND
29 root 1 -68 - 0K 16K CPU5 5 196:41 100.00% em0 taskq
11 root 1 171 ki31 0K 16K CPU7 7 175:41 94.09% idle: cpu7
16 root 1 171 ki31 0K 16K CPU2 2 175:45 91.26% idle: cpu2
15 root 1 171 ki31 0K 16K CPU3 3 180:18 89.45% idle: cpu3
14 root 1 171 ki31 0K 16K CPU4 4 177:13 87.89% idle: cpu4
17 root 1 171 ki31 0K 16K CPU1 1 165:27 86.87% idle: cpu1
12 root 1 171 ki31 0K 16K CPU6 6 176:18 83.25% idle: cpu6
18 root 1 171 ki31 0K 16K RUN 0 157:44 80.66% idle: cpu0
611 root 6 58 0 133M 44320K select 0 0:00 66.26% mpd5
21 root 1 -44 - 0K 16K CPU4 4 48:38 21.39% swi1: net
30 root 1 -68 - 0K 16K - 6 21:41 10.25% em1 taskq

Everything is OK with outbound interface - em1.

Current bandwidth - ~ 80 Mbit/s

There are a lot of input errors on em0 (but no on em1):

# netstat -w 1 -I em0
input (em0) output
packets errs bytes packets errs bytes colls
8012 923 2838565 12504 0 7943345 0
7934 874 2469244 12555 0 7728764 0
7931 976 2712035 12482 0 8006760 0
8015 813 2694716 10669 0 7796656 0
7975 733 2475193 12306 0 8032129 0
7871 825 2548198 12269 0 7789452 0
8072 961 2647014 11924 0 7260788 0
7909 983 2576145 10552 0 7479881 0
^C

And systat -v looks strange with no interrupts on em0:

2 users Load 1.34 1.61 1.62 May 3 14:04

Mem:KB REAL VIRTUAL VN PAGER SWAP PAGER
Tot Share Tot Share Free in out in out
Act 68152 9452 231584 11936 7786368 count
All 108516 10676 4486380 15448 pages
Proc: Interrupts
r p d s w Csw Trp Sys Int Sof Flt 3981 cow 22705 total
47 46k 10k 268k 6697 23k 10k 3973 zfod atkbd0 1
ozfod ata0 irq14
18.3%Sys 2.3%Intr 1.8%User 0.0%Nice 77.6%Idle %ozfod atapci1 19
| | | | | | | | | | | daefr 2001 cpu0: time
=========+> 5699 prcfr 2 em0 irq256
55 dtbuf 12110 totfr 6695 em1 irq257
Namei Name-cache Dir-cache 100000 desvn react 2001 cpu3: time
Calls hits % hits % 4217 numvn pdwak 2001 cpu1: time
12005 12004 100 304 frevn pdpgs 2001 cpu2: time
13 intrn 2001 cpu4: time
Disks ad4 232692 wire 2001 cpu5: time
KB/t 0.00 60640 act 2001 cpu7: time
tps 0 28784 inact 2001 cpu6: time
MB/s 0.00 336 cache
%busy 0 7786032 free
219632 buf

Latency grows up to 400 ms:
# ping 10.0.0.1
PING 10.0.0.1 (10.0.0.1): 56 data bytes
64 bytes from 10.0.0.1: icmp_seq=0 ttl=64 time=17.619 ms
64 bytes from 10.0.0.1: icmp_seq=1 ttl=64 time=27.497 ms
64 bytes from 10.0.0.1: icmp_seq=3 ttl=64 time=16.481 ms
64 bytes from 10.0.0.1: icmp_seq=4 ttl=64 time=24.535 ms
64 bytes from 10.0.0.1: icmp_seq=5 ttl=64 time=13.058 ms
^C
--- 10.0.0.1 ping statistics ---
6 packets transmitted, 5 packets received, 16.7% packet loss
round-trip min/avg/max/stddev = 13.058/19.838/27.497/5.346 ms

# top -S
last pid: 55827; load averages: 3.76, 3.42, 3.08 up 0+03:27:38 16:24:20
104 processes: 11 running, 81 sleeping, 12 waiting
CPU states: 1.7% user, 0.0% nice, 21.4% system, 3.0% interrupt, 73.9% idle
Mem: 71M Active, 89M Inact, 340M Wired, 336K Cache, 214M Buf, 7418M Free
Swap: 4096M Total, 4096M Free

PID USERNAME THR PRI NICE SIZE RES STATE C TIME WCPU COMMAND
29 root 1 -68 - 0K 16K CPU5 5 196:41 100.00% em0 taskq
11 root 1 171 ki31 0K 16K CPU7 7 175:41 94.09% idle: cpu7
16 root 1 171 ki31 0K 16K CPU2 2 175:45 91.26% idle: cpu2
15 root 1 171 ki31 0K 16K CPU3 3 180:18 89.45% idle: cpu3
14 root 1 171 ki31 0K 16K CPU4 4 177:13 87.89% idle: cpu4
17 root 1 171 ki31 0K 16K CPU1 1 165:27 86.87% idle: cpu1
12 root 1 171 ki31 0K 16K CPU6 6 176:18 83.25% idle: cpu6
18 root 1 171 ki31 0K 16K RUN 0 157:44 80.66% idle: cpu0
611 root 6 58 0 133M 44320K select 0 0:00 66.26% mpd5
21 root 1 -44 - 0K 16K CPU4 4 48:38 21.39% swi1: net
30 root 1 -68 - 0K 16K - 6 21:41 10.25% em1 taskq

# sysctl dev.em.0
dev.em.0.%desc: Intel(R) PRO/1000 Network Connection Version - 6.7.3
dev.em.0.%driver: em
dev.em.0.%location: slot=0 function=0
dev.em.0.%pnpinfo: vendor=0x8086 device=0x1096 subvendor=0x15d9 subdevice=0x0000 class=0x020000
dev.em.0.%parent: pci6
dev.em.0.debug: -1
dev.em.0.stats: -1
dev.em.0.rx_int_delay: 0
dev.em.0.tx_int_delay: 66
dev.em.0.rx_abs_int_delay: 66
dev.em.0.tx_abs_int_delay: 66
dev.em.0.rx_processing_limit: -1

I've tried both:
options SCHED_ULE
options SCHED_4BSD

I've added just the following lines in my kernel config:

options IPFIREWALL
options IPFIREWALL_DEFAULT_TO_ACCEPT

options NETGRAPH
options NETGRAPH_PPP
options NETGRAPH_PPTPGRE


My sysctls:
net.inet.ip.forwarding=1
net.inet.ip.fastforwarding=1
net.inet.ip.redirect=0
net.inet.ip.random_id=1
net.inet.ip.ttl=255
net.inet.ip.intr_queue_maxlen=4096

kern.maxfiles=131072
kern.maxfilesperproc=32768
kern.maxprocperuid=32768

kern.ipc.somaxconn=65535
kern.ipc.maxsockets=32768
kern.ipc.maxsockbuf=16777216

net.inet.tcp.rfc1323=1
net.inet.tcp.recvspace=65536
net.inet.tcp.sendspace=32768
net.inet.tcp.sendbuf_max=16777216
net.inet.tcp.recvbuf_max=16777216
net.inet.tcp.sendbuf_auto=1
net.inet.tcp.sendbuf_inc=8192
net.inet.tcp.recvbuf_auto=1
net.inet.tcp.recvbuf_inc=16384
net.inet.tcp.maxtcptw=40960
net.inet.tcp.msl=2500
net.inet.tcp.delayed_ack=0
net.inet.tcp.nolocaltimewait=1

net.inet.udp.checksum=0
net.inet.udp.recvspace=65535
net.inet.udp.maxdgram=57344

net.inet.icmp.icmplim=30

net.inet.tcp.blackhole=2
net.inet.udp.blackhole=1

net.local.stream.recvspace=65535
net.local.stream.sendspace=65535

net.isr.direct=1

kern.timecounter.hardware=TSC

dev.em.0.rx_processing_limit=-1

If I set net.isr.direct to "0", than sw1: net begins to eat 100% of a core, but without errors:
# netstat -w 1 -I em0
input (em0) output
packets errs bytes packets errs bytes colls
6953 0 2860537 8703 0 4882814 0
6785 0 2587635 7683 0 4443958 0
7006 0 2576630 8718 0 4924591 0
6887 0 2652461 8272 0 4548049 0
6854 0 2610157 8689 0 5152459 0
6889 0 2586067 8265 0 5010795 0
6878 0 2586746 8255 0 4734959 0
^C

Moreover, with net.isr.direct=0 I can't create a PPTP tunnel.

Please, help to solve the problem. Thanks!

--
Oleksandr Samoylyk
OVS-RIPE
_______________________________________________
freebsd-net@xxxxxxxxxxx mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-net
To unsubscribe, send any mail to "freebsd-net-unsubscribe@xxxxxxxxxxx"

Relevant Pages

  • Re: Troubles with em on FreeBSD 7
    ... Somehow em0 begins to eat all CPU time of one core. ... PID USERNAME THR PRI NICE SIZE RES STATE C TIME WCPU COMMAND ... packets errs bytes packets errs bytes colls ...
    (freebsd-net)
  • Troubles with em on FreeBSD 7
    ... Somehow em0 begins to eat all CPU time of one core. ... packets errs bytes packets errs bytes colls ...
    (freebsd-net)
  • RE: em driver input errors
    ... As said in my original email, I was getting way more errors with the defaults. ... mailing lists we also added: ... Now em0 is seeing about 1800 interrupts/second, which is way better, but after some time I saw errors again... ... packets errs bytes packets errs bytes colls ...
    (freebsd-net)
  • Troubles with em on FreeBSD 7
    ... Somehow em0 begins to eat all CPU time of one core. ... packets errs bytes packets errs bytes colls ... PING 10.0.0.1: 56 data bytes ...
    (freebsd-net)