Many OSes, including linux, tie nics to a specific core, so all
interrupts and i/o from that nic will be managed by one single core on
multicore machines.
I.e. all i/o to/from one specific nic will usually go through one
single core while all other cores are sitting idle, which means you
shouldnt get that much reordering from the stack itself.
This is very annoying especially if you have very many cores but not
that very many 10GbE interfaces and you want to drive i/o as hard as
possible.
Very recent linux kernels do have some improvements in this area that
will be interesting to test.
I dont know if BSD and BSD derived systems currently share the
interrupt and i/o load from one nic across multiple cores.
regards
ronnie sahlberg
On Sat, Mar 26, 2011 at 9:33 PM, Martin Visser <martinvisser99@xxxxxxxxx> wrote:
> Hmmm, obviously even on a multi-core machine the network drivers must be
> able to present packets to the OS network stack in the order on the wire.
> What mechanism allows libpcap to do thing in a different order. If what you
> are saying is correct, that has to be a major issue in libpcap?
> Regards, Martin
>
> MartinVisser99@xxxxxxxxx
>
>
> On 26 March 2011 03:46, Guy Harris <guy@xxxxxxxxxxxx> wrote:
>>
>> On Mar 24, 2011, at 9:57 AM, Gurpreet Singh wrote:
>>
>> > It is not a defect but i couldn't find any forum where I can ask this
>> > question.
>>
>> This is probably a good place to ask this question; another place might be
>> ask.wireshark.org.
>>
>> > I am monitoring Eth3 of 2 machines.
>> > 1st machine is sending the RLP(Radio Link Protocol) messages.
>> > 2nd machine receives the RLP Messages.
>> > Data rate would ne 30-40 MBps
>> >
>> > Problem:
>> > Order of the packet on Machine 2 is not same as Machine 1.
>> > Is there any guarantee that wireshark will display the packets in the
>> > same
>> > order as the order of the packets on the wire?
>>
>> Wireshark will, by default, display the packets in the same order that
>> they are delivered to whatever capture mechanism libpcap uses on UN*X and to
>> NDIS (as used by WinPcap) on Windows. That may, or may not, be the same
>> order of the packets on the wire, especially on multi-core machines.
>>
>>
>> ___________________________________________________________________________
>> Sent via: Wireshark-users mailing list <wireshark-users@xxxxxxxxxxxxx>
>> Archives: http://www.wireshark.org/lists/wireshark-users
>> Unsubscribe: https://wireshark.org/mailman/options/wireshark-users
>>
>> mailto:wireshark-users-request@xxxxxxxxxxxxx?subject=unsubscribe
>
>
> ___________________________________________________________________________
> Sent via: Wireshark-users mailing list <wireshark-users@xxxxxxxxxxxxx>
> Archives: http://www.wireshark.org/lists/wireshark-users
> Unsubscribe: https://wireshark.org/mailman/options/wireshark-users
> mailto:wireshark-users-request@xxxxxxxxxxxxx?subject=unsubscribe
>
