Date: Thu, 11 Nov 2010 11:56:37 -0800 (PST) From: Kirill Yelizarov <ykirill@yahoo.com> To: freebsd-stable@freebsd.org Subject: Re: icmp packets on em larger than 1472 [SEC=UNCLASSIFIED] Message-ID: <985402.19076.qm@web120519.mail.ne1.yahoo.com> In-Reply-To: <4CDC3D7B.7000108@sentex.net>
next in thread | previous in thread | raw e-mail | index | archive | help
=0A=0A--- On Thu, 11/11/10, Mike Tancsa <mike@sentex.net> wrote:=0A=0A> Fro= m: Mike Tancsa <mike@sentex.net>=0A> Subject: Re: icmp packets on em larger= than 1472 [SEC=3DUNCLASSIFIED]=0A> To: freebsd-stable@freebsd.org=0A> Date= : Thursday, November 11, 2010, 10:01 PM=0A> On 11/11/2010 12:26 AM, Kevin O= berman=0A> wrote:=0A> >> Date: Thu, 11 Nov 2010 13:01:26 +0800=0A> >> From:= "Wilkinson, Alex" <alex.wilkinson@dsto.defence.gov.au>=0A> >> Sender: owne= r-freebsd-stable@freebsd.org=0A> >>=0A> >>=0A> >>=A0 =A0=A0=A00n Wed, Nov 1= 0, 2010 at=0A> 04:21:12AM -0800, Kirill Yelizarov wrote: =0A> >>=0A> >>=A0 = =A0=A0=A0>All my em cards=0A> running 8.1 stable don't reply to icmp echo r= equests packets=0A> larger than 1472 bytes.=0A> >>=A0 =A0=A0=A0>=0A> >>=A0 = =A0=A0=A0>On stable 7.2 the same=0A> hardware works as expected:=0A> >>=A0 = =A0=A0=A0># ping -s 1500=0A> 192.168.64.99=0A> >>=A0 =A0=A0=A0>PING 192.168= .64.99=0A> (192.168.64.99): 1500 data bytes=0A> >>=A0 =A0=A0=A0>1508 bytes = from=0A> 192.168.64.99: icmp_seq=3D0 ttl=3D63 time=3D1.249 ms=0A> >>=A0 =A0= =A0=A0>1508 bytes from=0A> 192.168.64.99: icmp_seq=3D1 ttl=3D63 time=3D1.15= 8 ms=0A> >>=A0 =A0=A0=A0>=0A> >>=A0 =A0=A0=A0>Here is the dump on em=0A> in= terface=0A> >>=A0 =A0=A0=A0>15:06:31.452043 IP=0A> 192.168.66.65 > *****: I= CMP echo request, id 28729, seq=0A> 5, length 1480=0A> >>=A0 =A0=A0=A0>15:0= 6:31.452047 IP=0A> 192.168.66.65 > ****: icmp=0A> >>=A0 =A0=A0=A0>15:06:31.= 452069 IP=0A> **** > 192.168.66.65: ICMP echo reply, id 28729, seq 5,=0A> l= ength 1480=0A> >>=A0 =A0=A0=A0>15:06:31.452071 IP ***=0A> > 192.168.66.65: = icmp=0A> >>=A0 =A0=A0=A0> =0A> >>=A0 =A0=A0=A0>Same ping from same=0A> sour= ce (it's a 8.1 stable with fxp interface) to em card=0A> running 8.1 stable= =0A> >>=A0 =A0=A0=A0>#pciconf -lv=0A> >>=A0=0A> =A0=A0=A0>em0@pci0:3:4:0:= =A0=A0=A0=0A> class=3D0x020000 card=3D0x10798086 chip=3D0x10798086 rev=3D0x= 03=0A> hdr=3D0x00=0A> >>=A0 =A0=A0=A0>=A0 =A0=0A> vendor=A0 =A0=A0=A0=3D 'I= ntel Corporation'=0A> >>=A0 =A0=A0=A0>=A0 =A0=0A> device=A0 =A0=A0=A0=3D 'D= ual Port Gigabit Ethernet=0A> Controller (82546EB)'=0A> >>=A0 =A0=A0=A0>=A0= =A0=0A> class=A0 =A0 =A0 =3D network=0A> >>=A0 =A0=A0=A0>=A0 =A0=0A> subcl= ass=A0=A0=A0=3D ethernet=0A> >>=A0 =A0=A0=A0>=0A> >>=A0 =A0=A0=A0># ping -s= 1472=0A> 192.168.64.200=0A> >>=A0 =A0=A0=A0>PING 192.168.64.200=0A> (192.1= 68.64.200): 1472 data bytes=0A> >>=A0 =A0=A0=A0>1480 bytes from=0A> 192.168= .64.200: icmp_seq=3D0 ttl=3D63 time=3D0.848 ms=0A> >>=A0 =A0=A0=A0>^C=0A> >= >=A0 =A0=A0=A0>=0A> >>=A0 =A0=A0=A0># ping -s 1473=0A> 192.168.64.200=0A> >= >=A0 =A0=A0=A0>PING 192.168.64.200=0A> (192.168.64.200): 1473 data bytes=0A= > >>=A0 =A0=A0=A0>^C=0A> >>=A0 =A0=A0=A0>--- 192.168.64.200=0A> ping statis= tics ---=0A> >>=A0 =A0=A0=A0>4 packets transmitted,=0A> 0 packets received,= 100.0% packet loss=0A> >>=0A> >> works fine for me:=0A> >>=0A> >> FreeBSD = 8.1-STABLE #0 r213395=0A> >>=0A> >> em0@pci0:0:25:0:class=3D0x020000 card= =3D0x3035103c=0A> chip=3D0x10de8086 rev=3D0x02 hdr=3D0x00=0A> >>=A0 =A0=A0= =A0vendor=A0=0A> =A0=A0=A0=3D 'Intel Corporation'=0A> >>=A0 =A0=A0=A0device= =A0=0A> =A0=A0=A0=3D 'Intel Gigabit network connection=0A> (82567LM-3 )'=0A= > >>=A0 =A0=A0=A0class=A0 =A0 =A0=0A> =3D network=0A> >>=A0=0A> =A0=A0=A0su= bclass=A0=A0=A0=3D ethernet=0A> >>=0A> >> #ping -s 1473 host=0A> >> PING ho= st(192.168.1.1): 1473 data bytes=0A> >> 1481 bytes from 192.168.1.1: icmp_s= eq=3D0 ttl=3D253=0A> time=3D31.506 ms=0A> >> 1481 bytes from 192.168.1.1: i= cmp_seq=3D1 ttl=3D253=0A> time=3D31.493 ms=0A> >> 1481 bytes from 192.168.1= .1: icmp_seq=3D2 ttl=3D253=0A> time=3D31.550 ms=0A> >> ^C=0A> > =0A> > The = reason the '-s 1500' worked was that the packets=0A> were fragmented. If=0A= > > I add the '-D' option, '-s 1473' fails on v7 and v8.=0A> Are the V8 sys= tems=0A> > where you see if failing without the '-D' on the same=0A> networ= k segment?=0A> > If not, it is likely that an intervening device is=0A> ref= using to fragment=0A> > the packet. (Some routers deliberately don't fragme= nt=0A> ICMP Echos Request=0A> > packets.) =0A> =0A> =0A> I am not sure I fo= llow. If you do a=0A> ping -s 1473 -D=0A> on an interface that has the defa= ult MTU of 1500, it wont=0A> work, as the=0A> entire packet is going to be = 1501 (note the data bytes)=0A> =0A> eg.=0A> # ping=A0 -q -s 1472 -c 1 192.1= 68.43.219=0A> PING 192.168.43.219 (192.168.43.219): 1472 data bytes=0A> =0A= > --- 192.168.43.219 ping statistics ---=0A> 1 packets transmitted, 1 packe= ts received, 0.0% packet=0A> loss=0A> round-trip min/avg/max/stddev =3D 1.7= 14/1.714/1.714/0.000 ms=0A> on 192.168.43.219, I see=0A> =0A> and on .43.21= 9, I see=0A> =0A> 0(ich10)# tcpdump -vvvni em2 icmp=0A> tcpdump: listening = on em2, link-type EN10MB (Ethernet),=0A> capture size 96=0A> bytes=0A> 13:4= 9:17.564482 IP (tos 0x0, ttl 63, id 53656, offset 0,=0A> flags [none],=0A> = proto ICMP (1), length 1500)=0A> =A0 =A0 192.168.42.11 > 192.168.43.219: IC= MP echo=0A> request, id 23315, seq 0,=0A> length 1480=0A> 13:49:17.564499 I= P (tos 0x0, ttl 64, id 14346, offset 0,=0A> flags [none],=0A> proto ICMP (1= ), length 1500)=0A> =A0 =A0 192.168.43.219 > 192.168.42.11: ICMP echo=0A> r= eply, id 23315, seq 0,=0A> length 1480=0A> =0A> =0A> Note the length is 150= 0 of the packet.=0A> =0A> That being said, if its failing on the em nic whe= re you=0A> dont specify the=0A> -D flag on the ping, then there is a bug so= mewhere.=A0=0A> On certain em nics,=0A> I found doing=0A> ifconfig em0 -rxc= sum=0A> ifconfig em0 -txcsum=0A> ifconfig em0 -tso=0A> =0A> works around a = number of bugs=0A> =0A> =A0=A0=A0 ---Mike=0A> =0AYes, i know it. This was t= he first thing i tried. Sorry, i didn't mention it. =0A=0AKirill=0A> =0A> = =0A> _______________________________________________=0A> freebsd-stable@fre= ebsd.org=0A> mailing list=0A> http://lists.freebsd.org/mailman/listinfo/fre= ebsd-stable=0A> To unsubscribe, send any mail to "freebsd-stable-unsubscrib= e@freebsd.org"=0A> =0A=0A
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?985402.19076.qm>