On Dec 11, 2025, at 7:50 PM, Stephen Hemminger via tcpdump-workers <[email protected]> wrote:
> I have a meta-question about DPDK support in tcpdump/libpcap. > The current model where libpcap is a primary DPDK process is not that useful. > Only really usable as hardware sniffer. > > The better model would be for libpcap to implement the DPDK pdump API > which has tcpdump (or wireshark) running as a secondary process. As this is talking about libpcap, presumably you mean "which would have a program using libpcap, such as tcpdump or Wireshark, running as a seondary process"; they might be the *primary* users of libpcap, but they're not the *only* users (Snort and scapy both use it). Presumably this would still allow it to be used as a sniffer, as that's what tcpdump and Wireshark are, and as that's the primary purpose of libpcap (to allow programs to sniff network traffic) - and presumably it wouldn't require any modifications to programs that use libpcap, i.e., the "secondary process" using libpcap, such as tcpdump or Wireshark, wouldn't be run by a "primary process", it could just be run from the command line or the GUI. > This would allow using tcpdump etc while running any of the many DPDK > routing, switch applications. > > The question is should the new code: > be another capture type: i.e pcap-pdump; or just rewrite existing code in > pcap-dpdk? If it would do a better job of allowing programs to capture network traffic through a DPDK interface than the current code does (even if "better" just means "does an equally good job from the point of view of the program, but is easier to maintain"), then I'm not sure there's a reason to keep the old code around, just replace the existing code (even if the "rewrite" means "throw it away and start from scratch"). > Is there a maintainer for the DPDK code in libpcap? Not really. I don't know to what extent any of us core developers are sufficiently familiar to act as maintainers (I'm not). _______________________________________________ tcpdump-workers mailing list -- [email protected] To unsubscribe send an email to [email protected] %(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s
