Re: libpcap pcap_sendpacket support across platforms.

["Mark Pizzolato" <List-tcpdump-workers () subscriptions pizzolato net>]

The attached revised pcap-bpf.c now addresses the BIOCSHDRCMPLT  issue you
noticed.

- Mark Pizzolato

----- Original Message ----- 
From: "Mark Pizzolato" <List-tcpdump-workers () subscriptions pizzolato net>
To: "Guy Harris" <gharris () sonic net>
Cc: "Mike Schiffman" <mike () infonexus com>; "TCPdump Workers"
<tcpdump-workers () tcpdump org>
Sent: Tuesday, March 23, 2004 4:56 PM
Subject: Re: [tcpdump-workers] libpcap pcap_sendpacket support across
platforms.


> On Tuesday, March 23, 2004 11:22 AM, Guy Harris wrote:
>
> > > Attached is a revised patch, which keeps prior behaviors, while also
> > > providing for pcap_sendpacket when possible.
> >
> > I've merged your changes with a version I'd been working on, based on
> > libnet and some stuff we had at Network Appliance, and checked the
> > result in.  It implements *two* APIs - "pcap_inject()", from OpenBSD,
> > and "pcap_sendpacket()", from WinPcap.
> >
> > Some issues:
> >
> > on BPF, libpcap doesn't do anything with BIOCSHDRCMPLT - libnet
> > does;
>
> Good catch.  I still had that in my application code.
>
> > on DLPI, there's #if'ed-out code to use the appropriate "raw
> > data" primitives on HP-UX - I don't have access to an HP-UX
> > system on which to test them;
>
> I don't have access either.  What I sent does work on Solaris 8 (Both
Intel
> and Sparc).
>
> > on Linux, the version I checked in uses "sendto()", as that's
> > what libnet did;
>
> Hmmmm..  I didn't check with libnet since I had simple working code with a
> very basic send().
>
> What you checked in is broken.  It has an extra right paren on line 687,
and
> after that is fixed, line 725 refers to an undefined sa instead of sa_pkt.
> The attached cleans things up a bit.
>
> I wonder where the sendto() stuff is really necessary.  The simple send()
> worked for us on RH 7.3-Fedora Core1 on Intel.  And RH 6.2 on Sparc, and
> numerous other linux environments.  We've never gotten a complaint....
>
> > on SunOS 4.x with STREAMS NIT, the code I checked in uses
> > "putmsg()" - libnet and your code used "sendto()", but, as I
> > remember, the Network Appliance code used "putmsg()", which is
> > what I'd expect to be the correct call to use on a STREAMS-based
> > mechanism.
>
> Since I didn't have any other reference and don't have a test environment,
I
> worked from what I presumed to be correct in libnet.

Attachment: pcap-bpf.c
Description: