Anton,
I believe it could be actually some corner case in the logic (inside
WinPcap) that decides when the event needs to be signaled and when it
needs to be reset. If it's an issue like this, it probably shows up
only in a scenario when the traffic is extremely slow or when you
transmit a fixed amount of packets and then nothing else on the wire.
Out of curiosity, why is this delay so critical to you? This delay
usually causes issues only when you try to retransmit the received
packets on a different interface (like a bridge).
Have a nice day
GV
*From:* Anton Tremsin [mailto:[email protected]]
*Sent:* Wednesday, March 23, 2011 7:43 PM
*To:* [email protected]
*Cc:* Gianluca Varenni
*Subject:* Re: [Winpcap-users] Winpcap-users Digest, Vol 72, Issue 8
Hi Gianluca,
You are right again, my comment was not correct. I looked at the
timestamps of all 64 packets in Wireshark to answer your question.
What have looked reliable to me first, now shows that there was some
delay in the data receiving sequence, but I cannot tell precisely
where that delay happens (at which packet number) and how long it is
(changes from one data set to another and it is hard to find where the
gap is). Wireshark shows all 64 packets at once, which looked fine as
if there was no delays.
What is strange in my receiving code now is the following: data comes
with something like 57 - 62 packets, then there is delay equal to the
timeout parameter which I set in the open function. Then the rest of
the data comes. No timeout error shows up in that case. It is only
after the last packet was received the timeout events show up in the
pcap_next_ex() return parameter.
May be it is all in the card driver and I cannot get any help from
Chelsio support, after describing the situation to them. For now I set
the timeout to 1 ms, and handle many timeout events when I do not have
data coming in. That seems to be not an elegant solution, but somewhat
sufficient for the alpha version of our data acquisition. I am trying
to borrow another card (from different manufacturer) to test whether
it is indeed in the card driver. What puzzles me is why setmintocopy()
does not any effect in my code?
Thanks,
Anton
Anton,
You say that "Wireshark grabs data fast and reliably". What do you
mean by that? Do you look at the timestamps in wireshark to say that
it's reliable?
Have a nice day
GV
*From:* [email protected]
<mailto:[email protected]>
[mailto:[email protected]] *On Behalf Of *Anton Tremsin
*Sent:* Wednesday, March 16, 2011 9:24 AM
*To:* [email protected] <mailto:[email protected]>
*Subject:* Re: [Winpcap-users] Winpcap-users Digest, Vol 72, Issue 8
Hi Akif,
I have not tried packet.h, but I see the same behavior: Wireshark
grabs data fast and reliably, while my code using C library has those
delays, which I cannot get rid of. I can receive ~55 packets (8K each)
reliably, with no any delays, but for 64 of them I always have a
delay. setmintocopy does not change anything, as in your case. I also
changed the buffer size , that does not change the delay either, once
it is large enough.
I am stack and really do not know what I can do. I talked to the
manufacturer of my card (Chesio), but they have no idea why that
happens either.
Anton
Hi,
The only thing that amazes me is Wireshark. Why is wireshark able to
capture with such accuracy even if it uses winpcap.
Is it possible to achieve accuracy if i used packet.h functions to
receive and then pcap to send the packets. Anton have you tried that?
BR
> From: [email protected]
<mailto:[email protected]>
> Subject: Winpcap-users Digest, Vol 72, Issue 8
> To: [email protected] <mailto:[email protected]>
> Date: Wed, 16 Mar 2011 02:08:28 -0700
>
> Send Winpcap-users mailing list submissions to
> [email protected] <mailto:[email protected]>
>
> To subscribe or unsubscribe via the World Wide Web, visit
> https://www.winpcap.org/mailman/listinfo/winpcap-users
> or, via email, send a message with subject or body 'help' to
> [email protected]
<mailto:[email protected]>
>
> You can reach the person managing the list at
> [email protected] <mailto:[email protected]>
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Winpcap-users digest..."
>
>
> Today's Topics:
>
> 1. Re: PPP capture (Gianluca Varenni)
> 2. Re: Winpcap-users Digest, Vol 72, Issue 7 (Akif Usman)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Tue, 15 Mar 2011 13:59:25 -0700
> From: Gianluca Varenni <[email protected]>
<mailto:[email protected]>
> To: Anton Tremsin <[email protected]>
<mailto:[email protected]>, "[email protected]"
<mailto:[email protected]>
> <[email protected]> <mailto:[email protected]>
> Subject: Re: [Winpcap-users] PPP capture
> Message-ID:
> <6a8f2e88cff83c43a6aff7fac775b9fc0715174...@mailboxes2.nbttech.com>
<mailto:6a8f2e88cff83c43a6aff7fac775b9fc0715174...@mailboxes2.nbttech.com>
> Content-Type: text/plain; charset="us-ascii"
>
> Anton,
>
> If I remember well, you are capturing from Ethernet, Akif is
capturing from PPP. The code paths for the two types of devices are
completely different (Ethernet goes through the WinPcap kernel driver,
PPP gets captured through Netmon).
>
> Have a nice day
> GV
>
> From: Anton Tremsin [mailto:[email protected]]
> Sent: Monday, March 14, 2011 11:43 PM
> To: [email protected] <mailto:[email protected]>
> Cc: Gianluca Varenni
> Subject: Re: [Winpcap-users] PPP capture
>
> Akif, Gianluca,
>
> As I mentioned in my previous messages, I have exactly the same
problem of delayed packages, with mintocopy set even to 0 (tried other
values as well). I am always sending a set of 64 packets of 8Kbytes
each (that is one image data). The packets are not lost, they always
arrive. However, some of them come with no delay (varied number of
them, sometimes 62, sometimes 57, etc), while the rest of them come
exactly after the delay equal to the setting of the timeout, which I
varied between 1 and 10000 milliseconds. There is no timeout reported
for the packets to arrive with the delay.
>
> I will be very glad if that issue can be solved, which has probably
the same cause as in Akif's application.
>
> Thanks a lot,
>
> Anton
>
> Akif,
>
> This is probably due to the mintocopy and timeout of WinPcap.
WinPcap does not deliver you the packets immediately after they are
received by the driver. Packets are batched in kernel mode and
delivered to the receiving application when
>
>
> There are at least mintocopy bytes in the kernel buffer
>
> After a certain timeout
> (whatever happens first).
>
> In order to reduce the delay, you will need to either reduce the
timeout or the mintocopy.
>
> Have a nice day
> GV
>
> From: [email protected]
<mailto:[email protected]><mailto:[email protected]>
[mailto:[email protected]] On Behalf Of Akif Usman
> Sent: Thursday, March 10, 2011 11:20 AM
> To: [email protected]
<mailto:[email protected]><mailto:[email protected]>
> Subject: [Winpcap-users] PPP capture
>
> HI,
>
> I have installed the winpcap version 3.1 beta and i am capturing
from a PPP interface and it captures perfectly. Now i am trying to
capture from the same PPP interface using my LIBPCAP program and
forward it to another Ethernet interface that connects further to a
second computer (Ethernet NIC) which also has wireshark running on it.
When i capture from the second computer i get a strange offset of 0.5
seconds after every x packets. This is very strange. I dont know why
wireshark is able to capture from PPP interface on the first computer
with proper accuracy and why my LIBPCAP program, which is just
forwarding the packets, is introducing a 0.5s [:-O] delay. Please help
me out as soon as somebody can.
>
> Best Regards
> Fika
>
>
>
>
>
> _______________________________________________
>
> Winpcap-users mailing list
>
> [email protected]
<mailto:[email protected]><mailto:[email protected]>
>
> https://www.winpcap.org/mailman/listinfo/winpcap-users
>
>
>
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
<http://www.winpcap.org/pipermail/winpcap-users/attachments/20110315/7cb51127/attachment-0001.html>
>
> ------------------------------
>
> Message: 2
> Date: Wed, 16 Mar 2011 10:08:25 +0100
> From: Akif Usman <[email protected]>
<mailto:[email protected]>
> To: <[email protected]> <mailto:[email protected]>
> Subject: Re: [Winpcap-users] Winpcap-users Digest, Vol 72, Issue 7
> Message-ID: <[email protected]>
<mailto:[email protected]>
> Content-Type: text/plain; charset="iso-8859-1"
>
>
> HI,
> There are some questiosn i need to ask. Why does wireshark give no
delay upon capture even though it uses Winpcap?
> I am using windows xp for capture and i have checked the capture on
ethernet and there seems to be no problems at all from the capture on
ethernet. I have tried changing mintocopy and the timeout but it gives
me no changes in the performance? Any ideas why?
>
> BR
>
> > From: [email protected]
<mailto:[email protected]>
> > Subject: Winpcap-users Digest, Vol 72, Issue 7
> > To: [email protected] <mailto:[email protected]>
> > Date: Tue, 15 Mar 2011 12:00:02 -0700
> >
> > Send Winpcap-users mailing list submissions to
> > [email protected] <mailto:[email protected]>
> >
> > To subscribe or unsubscribe via the World Wide Web, visit
> > https://www.winpcap.org/mailman/listinfo/winpcap-users
> > or, via email, send a message with subject or body 'help' to
> > [email protected]
<mailto:[email protected]>
> >
> > You can reach the person managing the list at
> > [email protected]
<mailto:[email protected]>
> >
> > When replying, please edit your Subject line so it is more specific
> > than "Re: Contents of Winpcap-users digest..."
> >
> >
> > Today's Topics:
> >
> > 1. Re: PPP capture (Gianluca Varenni)
> > 2. Re: PPP capture (Anton Tremsin)
> >
> >
> > ----------------------------------------------------------------------
> >
> > Message: 1
> > Date: Mon, 14 Mar 2011 18:44:54 -0700
> > From: Gianluca Varenni <[email protected]>
<mailto:[email protected]>
> > To: "[email protected]" <mailto:[email protected]>
<[email protected]> <mailto:[email protected]>
> > Subject: Re: [Winpcap-users] PPP capture
> > Message-ID:
> >
<6a8f2e88cff83c43a6aff7fac775b9fc0715173...@mailboxes2.nbttech.com>
<mailto:6a8f2e88cff83c43a6aff7fac775b9fc0715173...@mailboxes2.nbttech.com>
> > Content-Type: text/plain; charset="us-ascii"
> >
> > Akif,
> >
> > This is probably due to the mintocopy and timeout of WinPcap.
WinPcap does not deliver you the packets immediately after they are
received by the driver. Packets are batched in kernel mode and
delivered to the receiving application when
> >
> >
> > - There are at least mintocopy bytes in the kernel buffer
> >
> > - After a certain timeout
> > (whatever happens first).
> >
> > In order to reduce the delay, you will need to either reduce the
timeout or the mintocopy.
> >
> > Have a nice day
> > GV
> >
> > From: [email protected]
<mailto:[email protected]>
[mailto:[email protected]] On Behalf Of Akif Usman
> > Sent: Thursday, March 10, 2011 11:20 AM
> > To: [email protected] <mailto:[email protected]>
> > Subject: [Winpcap-users] PPP capture
> >
> > HI,
> >
> > I have installed the winpcap version 3.1 beta and i am capturing
from a PPP interface and it captures perfectly. Now i am trying to
capture from the same PPP interface using my LIBPCAP program and
forward it to another Ethernet interface that connects further to a
second computer (Ethernet NIC) which also has wireshark running on it.
When i capture from the second computer i get a strange offset of 0.5
seconds after every x packets. This is very strange. I dont know why
wireshark is able to capture from PPP interface on the first computer
with proper accuracy and why my LIBPCAP program, which is just
forwarding the packets, is introducing a 0.5s [:-O] delay. Please help
me out as soon as somebody can.
> >
> > Best Regards
> > Fika
> > -------------- next part --------------
> > An HTML attachment was scrubbed...
> > URL:
<http://www.winpcap.org/pipermail/winpcap-users/attachments/20110314/fcd4e478/attachment-0001.html>
> >
> > ------------------------------
> >
> > Message: 2
> > Date: Mon, 14 Mar 2011 23:42:30 -0700
> > From: Anton Tremsin <[email protected]>
<mailto:[email protected]>
> > To: [email protected] <mailto:[email protected]>
> > Cc: Gianluca Varenni <[email protected]>
<mailto:[email protected]>
> > Subject: Re: [Winpcap-users] PPP capture
> > Message-ID: <[email protected]>
<mailto:[email protected]>
> > Content-Type: text/plain; charset="iso-8859-1"; Format="flowed"
> >
> > Akif, Gianluca,
> >
> > As I mentioned in my previous messages, I have exactly the same
problem
> > of delayed packages, with mintocopy set even to 0 (tried other
values as
> > well). I am always sending a set of 64 packets of 8Kbytes each
(that is
> > one image data). The packets are not lost, they always arrive.
However,
> > some of them come with no delay (varied number of them, sometimes 62,
> > sometimes 57, etc), while the rest of them come exactly after the
delay
> > equal to the setting of the timeout, which I varied between 1 and
10000
> > milliseconds. There is no timeout reported for the packets to arrive
> > with the delay.
> >
> > I will be very glad if that issue can be solved, which has
probably the
> > same cause as in Akif's application.
> >
> > Thanks a lot,
> >
> > Anton
> > >
> > > Akif,
> > >
> > > This is probably due to the mintocopy and timeout of WinPcap.
WinPcap
> > > does not deliver you the packets immediately after they are
received
> > > by the driver. Packets are batched in kernel mode and delivered
to the
> > > receiving application when
> > >
> > > - There are at least mintocopy bytes in the kernel buffer
> > >
> > > - After a certain timeout
> > >
> > > (whatever happens first).
> > >
> > > In order to reduce the delay, you will need to either reduce the
> > > timeout or the mintocopy.
> > >
> > > Have a nice day
> > >
> > > GV
> > >
> > > *From:* [email protected]
<mailto:[email protected]>
> > > [mailto:[email protected]] *On Behalf Of *Akif Usman
> > > *Sent:* Thursday, March 10, 2011 11:20 AM
> > > *To:* [email protected] <mailto:[email protected]>
> > > *Subject:* [Winpcap-users] PPP capture
> > >
> > > HI,
> > >
> > > I have installed the winpcap version 3.1 beta and i am capturing
from
> > > a PPP interface and it captures perfectly. Now i am trying to
capture
> > > from the same PPP interface using my LIBPCAP program and forward
it to
> > > another Ethernet interface that connects further to a second
computer
> > > (Ethernet NIC) which also has wireshark running on it. When i
capture
> > > from the second computer i get a strange offset of 0.5 seconds
after
> > > every x packets. This is very strange. I dont know why wireshark is
> > > able to capture from PPP interface on the first computer with
proper
> > > accuracy and why my LIBPCAP program, which is just forwarding the
> > > packets, is introducing a 0.5s [:-O] delay. Please help me out
as soon
> > > as somebody can.
> > >
> > > Best Regards
> > >
> > > Fika
> > >
> > >
> > > _______________________________________________
> > > Winpcap-users mailing list
> > > [email protected] <mailto:[email protected]>
> > > https://www.winpcap.org/mailman/listinfo/winpcap-users
> > >
> >
> > -------------- next part --------------
> > An HTML attachment was scrubbed...
> > URL:
<http://www.winpcap.org/pipermail/winpcap-users/attachments/20110314/fde1d594/attachment-0001.html>
> >
> > ------------------------------
> >
> > _______________________________________________
> > Winpcap-users mailing list
> > [email protected] <mailto:[email protected]>
> > https://www.winpcap.org/mailman/listinfo/winpcap-users
> >
> >
> > End of Winpcap-users Digest, Vol 72, Issue 7
> > ********************************************
>
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
<http://www.winpcap.org/pipermail/winpcap-users/attachments/20110316/cd1b5258/attachment.html>
>
> ------------------------------
>
> _______________________________________________
> Winpcap-users mailing list
> [email protected] <mailto:[email protected]>
> https://www.winpcap.org/mailman/listinfo/winpcap-users
>
>
> End of Winpcap-users Digest, Vol 72, Issue 8
> ********************************************
_______________________________________________
Winpcap-users mailing list
[email protected] <mailto:[email protected]>
https://www.winpcap.org/mailman/listinfo/winpcap-users
--
Anton S. Tremsin, Ph.D.
Space Sciences Laboratory
University of California at Berkeley
Berkeley, CA 94720
Tel: (510) 642 4554
_______________________________________________
Winpcap-users mailing list
[email protected] <mailto:[email protected]>
https://www.winpcap.org/mailman/listinfo/winpcap-users
