defined - Debian bug 171210 says that "sparc" isn't defined by GCC,
which presumably means "with the compiler we're using on Debian" as
there are versions of GCC that *do* define "sparc" on SPARC.
unfortunately, we can't fix "pcap_dump()" and "pcap_dump_close()" to do
that, as any application that tests the return value would fail to work
correctly if linked at runtime with an older libpcap, but we should
perhaps introduce "pcap_dump_ex()" and "pcap_dump_close_ex()" routines
that do return a success-vs-vailure indication.
argument to "pcap_open_live()" a "const" pointer.
Constify some additional device name arguments, and update the man page
to reflect some arguments that were already consts.
Young <dyoung@ojctech.com>, with some minor changes by Jason R. Thorpe
<thorpej@netbsd.org>, and further changes by me to support it on BPF
systems lacking BIOCGDLTLIST and other platforms lacking an equivalent
feature.
Update Jason Thorpe's e-mail address (Zembu is going away, if it hasn't
done so already).
Add APIs to map DLT names to DLT values and vice versa.
can use BPF, you don't need to have the latest DLPI driver or make sure
it's loaded (and maybe we're finally at the point where you can use
BPF).
Reformat a bit.
Note that the DLPI support has problems with DL_PROMISC_MULTI failing.
if, as I suspect is the case, that causes no timeout to be set, and if
that's the same as explicitly clearing the timeout with SBIOCCTIME, that
would appear to mean that it'd wait forever for a full chunk's worth of
packets to arrive.
can have one of two different behaviors, depending on the OS (it means
"don't return from a read until enough data has arrived" on BSD and
Digital/Tru64 UNIX, and means "return immediately" on Solaris, for
example, at least according to the man pages on Digital/Tru64 UNIX and
Solaris and the code in BSD).
"pcap_dispatch()" and "pcap_loop()", give more details on the effect of
the "snaplen" argument to "pcap_open_live()", and suggest 65535 as a
value if you want to capture the entire packet.
particular, point out that it's only used when checking for IPv4
broadcast addresses, and that if you don't care whether those checks are
done correctly, you can supply 0 if the netmask isn't known or isn't
available.
configure script to think it's IRIX, so test for
"/usr/include/linux/socket.h" before testing for
"/usr/include/net/raw.h" (which is done before testing for
"/usr/include/sys/dlpi.h"; hopefully no future IRIX release, or future
release of Solaris/HP-UX/AIX/other OS that supports DLPI, will add
"/usr/include/linux/socket.h" as a "helpful" header file for Linux
compatibility).
of packet headers so that, in all expressions after it, the tests assume
LANE encapsulation of packets. (We also assume the emulated LAN is an
Ethernet LAN, rather than a Token Ring LAN.)
Allow ATM tests to be combined with non-ATM tests in expressions, so
that you can do, for example, "lane and icmp".
guy