IGMP(4)

💖 IGMP(4) — the Kernel Interfaces Manual glow-up 💖

okay bestie, so this is the Kernel Interfaces Manual, and today's main character is igmp — the Internet Group Management Protocol 💅✨

🦋 NAME 🦋

igmp — Internet Group Management Protocol, no cap 💅

✨ SYNOPSIS ✨

#include <sys/types.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#include <netinet/igmp.h>

int
socket(AF_INET, SOCK_RAW, IPPROTO_IGMP);

🌸 DESCRIPTION 🌸

okay so IGMP is literally the control plane protocol that IPv4 hosts and routers use to spread the word about multicast group membership info 🫶. Normally you're not touching this directly — it's the kernel itself doing the work, responding to multicast membership requests from user apps. Routing protocols can open a raw socket to talk to igmp directly though, iykyk.

And fr fr, as of FreeBSD 8.0, IGMP version 3 is implemented 👑. This is the era that adds support for Source-Specific Multicast (SSM) — meaning apps can tell upstream multicast routers "bestie, I only wanna receive multicast streams from these particular sources," and that's iconic.

🧚 SYSCTL VARIABLES 🧚

• net.inet.igmp.stats — this opaque read-only variable exposes the stack-wide IGMPv3 protocol statistics to netstat(1). basically the receipts 💅
• net.inet.igmp.ifinfo — this opaque read-only variable exposes the per-link IGMPv3 status to ifmcstat(8). per-link vibes ✨
• net.inet.igmp.gsrdelay — this one specifies the time threshold, in seconds, for processing Group-and-Source Specific Queries (GSR). and here's the tea: since GSR query processing requires maintaining state on the host, it can cause memory to be allocated, so it's lowkey a potential attack point for Denial-of-Service (DoS) 😭. If more than one GSR query comes in within this threshold, it gets dropped — all to mitigate the potential for DoS. protect your bag bestie 🫶
• net.inet.igmp.default_version — controls the default version of IGMP used on all links. this sysctl is normally set to 3 by default, periodt.
• net.inet.igmp.legacysupp — if this is non-zero, then IGMP v1 and v2 membership reports received on a link are allowed to suppress the IGMP v3 state-change reports that this host would otherwise issue. this sysctl is normally enabled by default 💖
• net.inet.igmp.v2enable — if non-zero, then IGMP v2 membership queries get processed by this host, and backwards compatibility is enabled until the v2 'Old Querier Present' timer expires. normally enabled by default ✨
• net.inet.igmp.v1enable — if non-zero, then IGMP v1 membership queries get processed by this host, and backwards compatibility is enabled until the v1 'Old Querier Present' timer expires. normally enabled by default 💅
• net.inet.igmp.sendlocal — if non-zero, then IGMP state-changes for groups in the 224.0.0.0/24 link-scope prefix will be issued. this behaviour is recommended if you're deploying FreeBSD in a network environment with layer 2 devices that snoop IGMP traffic to mitigate multicast propagation throughout the network 👀. normally enabled by default.
• net.inet.igmp.sendra — if non-zero, then IGMP v2 and v3 reports will contain the IP Router Alert option. normally enabled by default 🎀
• net.inet.igmp.recvifkludge — if non-zero, then received IGMP reports containing 0.0.0.0 as their source get rewritten to contain the subnet address. useful when there are hosts on-link that haven't been configured with a primary IPv4 address yet — a little glow-up for the unconfigured girlies 💖. normally enabled by default.

🔥 SEE ALSO 🔥

netstat(1), sourcefilter(3), inet(4), multicast(4), ifmcstat(8) — go read these next, bestie, they're all connected 🌈

💜 HISTORY 💜

The igmp manual page re-appeared in FreeBSD 8.0. a comeback era, honestly 👑

and that's the whole page, bestie 🫶 this was the Debian version from March 9, 2009 — yeah, she's a little vintage but she ate 💅✨