Date: Sun, 10 Oct 2010 10:40:14 +0000 From: four.harrisons@googlemail.com To: kes-kes@yandex.ru Cc: freebsd-questions@freebsd.org Subject: Re[2]: How to obtain which interrupts cause system to hang? Message-ID: <4cb2ce15.4808cc0a.019a.13fb@mx.google.com>
next in thread | raw e-mail | index | archive | help
On Sun, 10 Oct 2010, ??????? ??????? wrote:=0D=0A > >> #systat -v=0D=0A =
> >> 1 users Load 0.74 0.71 0.55 Oct 9 19:53=
=0D=0A > IS> [..]=0D=0A > >> Proc: =
Interrupts=0D=0A > >> r p d s w Csw T=
rp Sys Int Sof Flt 24 cow 2008 total=0D=0A > >> 2 3 =
39 23k 67 563 9 1710 47 15 zfod 9 ata0 irq14=0D=0A=
> >> ozfod =
nfe0 irq23=0D=0A > >> 23.1%Sys 50.8%Intr 1.3%User 0.0%Nice 24.8%I=
dle %ozfod 1999 cpu0: time=0D=0A > >> | | | | | |=
| | | | | daefr=0D=0A > >> =3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D+++++++++++++++++++++++++> 6 prcfr=0D=0A >=20=
=0D=0A > IS> Yes, system and esp. interrupt time is heavy .. 23k context =
switches!?=0D=0A >=20=0D=0A > IS> In addition to b. f.'s good advice .. a=
s you later said, 2000 Hz slicing=0D=0A > IS> _should_ be ok, unless a sl=
ow CPU? Or perhaps a fast CPU throttled back=0D=0A > IS> too far .. powe=
rd? Check sysctl dev.cpu.0.freq while this is happening.=0D=0A >=20=0D=0A=
> IS> Disable p4tcc if it's a modern CPU; that usually hurts more than h=
elps.=0D=0A > IS> Disable polling if you're using that .. you haven't pro=
vided much info,=0D=0A > IS> like is this with any network load, despite =
nfe0 showing no interrupts?=0D=0A=0D=0A > Polling is ON. Traffice is abou=
t 60Mbit/s routed from nfe0 to vlan4 on rl0=0D=0A > when interrupts are h=
appen traffic slow down to 25-30Mbit/s.=0D=0A=0D=0AOut of my depth. If i=
t's a net problem - maybe not - you may do better=20=0D=0Ain freebsd-net@=
if you provide enough information (dmesg plus ifconfig,=20=0D=0Avmstat -=
i etc, normally and while this problem is happening).=0D=0A=0D=0A > There=
is no p4tcc option in KERNEL config file.=0D=0A=0D=0ANo, it can be enabl=
ed by cpufreq(4). See dmesg for acpi_throttle or=20=0D=0Ap4tcc, but it l=
ooks like you might not have device cpufreq in your=20=0D=0Akernel or loa=
ded, or dev.cpu.0.freq and more would have shown below.=0D=0A=0D=0A > dis=
able/enable polling does not help. situation still same.=0D=0A=0D=0A > sy=
sctl -a | grep freq=0D=0A > kern.acct_chkfreq: 15=0D=0A > kern.timecounte=
r.tc.i8254.frequency: 1193182=0D=0A > kern.timecounter.tc.ACPI-fast.frequ=
ency: 3579545=0D=0A > kern.timecounter.tc.TSC.frequency: 1809280975=0D=0A=
> net.inet.sctp.sack_freq: 2=0D=0A > debug.cpufreq.verbose: 0=0D=0A > de=
bug.cpufreq.lowest: 0=0D=0A > machdep.acpi_timer_freq: 3579545=0D=0A > ma=
chdep.tsc_freq: 1809280975=0D=0A > machdep.i8254_freq: 1193182=0D=0A=0D=0A=
Only useful for what it doesn't show :)=0D=0A=0D=0A > >> How to obtain w=
hat nasty happen, which process take 36-50% of CPU=0D=0A > >> resource?=0D=
=0A >=20=0D=0A > IS> Try 'top -S'. It's almost certainly system process[e=
s], not shown above.=0D=0A=0D=0ADoes that not show anything? Also, somet=
hing like 'ps auxww | less'=20=0D=0Ashould show you what's using all that=
CPU. I'm out of wild clues.=0D=0A=0D=0Acheers, Ian=0D=0A_______________=
________________________________=0D=0Afreebsd-questions@freebsd.org maili=
ng list=0D=0Ahttp://lists.freebsd.org/mailman/listinfo/freebsd-questions=0D=
=0ATo unsubscribe, send any mail to "freebsd-questions-unsubscribe@freebs=
d.org"=0D=0A
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4cb2ce15.4808cc0a.019a.13fb>