rdma_get_gid_attr(9)

📦 version 7.3.1, in nixpkgs unstable

🌸 rdma_get_gid_attr(9) — the Kernel API Manual, bestie edition 🌸

okay so this is a page from the Kernel API Manual (rdma_get_gid_attr(9)), and honestly? it's a little baddie. let's get into it 💅

✨ NAME ✨

rdma_get_gid_attr — returns GID attributes for a port of a device at a requested gid_index, if a valid GID entry exists. literally that's her whole job 🎀

🦋 SYNOPSIS 🦋

const struct ib_gid_attr * rdma_get_gid_attr (struct ib_device *device ,
u32 port_num , int index );

look at that signature. iconic. no notes 🔥

💖 ARGUMENTS 💖

  • device — the device you're querying. your main character device, if you will 👑
  • port_num — the port number on the device where the GID value is gonna be queried. yes bestie, which port 💫
  • index — index of the GID table entry whose attributes you're querying. it's giving "entry number, please" 🍓

🧚 DESCRIPTION 🧚

so what she actually does: rdma_get_gid_attr acquires a reference count of gid attributes from the cached GID table. that's the whole vibe ✨

⚠️ bestie listen to me: the caller must invoke rdma_put_gid_attr to release the reference to the gid attribute, regardless of link layer. no skipping, no "it'll be fine" — do not leave her hanging, that's so not slay 😭

she returns a pointer to a valid gid attribute — or ERR_PTR for the appropriate error code. so... either you get the goods or you get a little sad pointer-shaped envelope 💌

🌈 SEE ALSO 🌈

Kernel file ./drivers/infiniband/core/cache.c — and also all of these besties, they're in the group chat too:

and that's the page, periodt 🫶 sign-off time: this one lives over in ./drivers/infiniband/core, dated October 2026, and answers to the name rdma_get_gid_attr(9) — mwah 💋💜

📦 where this page comes from