__rxe_init_mcg(9)

📦 version 7.3.1, in nixpkgs unstable

💫 __rxe_init_mcg(9) 💫

okay so bestie, welcome to the Kernel API Manual ✨ — this is the page for __rxe_init_mcg, and no cap, it's a little cutie. Let's get into it 🎀

🌸 NAME 🌸

__rxe_init_mcg — initialize a new mcg 💅

✨ SYNOPSIS ✨

void __rxe_init_mcg (struct rxe_dev *rxe , union ib_gid *mgid , struct
rxe_mcg *mcg );

like, that's the whole signature. she's not returning anything, she's void, she just does her thing 🫶

🦋 ARGUMENTS 🦋

  • rxe — the rxe device, bestie. this is the device we're working with 💖
  • mgid — the multicast address, but as a gid 👑
  • mcg — the brand new mcg object. freshly made. iconic 🌟

🔥 DESCRIPTION 🔥

so what she actually does is initialize the mcg fields. that's the whole vibe 💅

here's the part that matters, though — the mcg is private at this point and it is not yet visible in mcg_tree. because of that, this call may totally run without rxe->mcg_lock, no stress, no drama 🧚

then later, __rxe_publish_mcg is the one that makes it visible — and that happens under the lock, once the mcg is actually ready. so like, init first, publish later, lock only when publishing. periodt ✨

lowkey the whole point: don't go looking for it in mcg_tree yet, she hasn't been introduced to the tree 💀

🌈 SEE ALSO 🌈

kernel file ./drivers/infiniband/sw/rxe/rxe_mcast.c — and honestly the whole friend group 💜:

and that's the page!! 🍓 this one lives in ./drivers/infiniband/sw/rxe, dated October 2026, and it's __rxe_init_mcg(9) all the way down 💖 slay, go initialize some mcg 🎀🫶💅

📦 where this page comes from