rdma_rw_ctx_init(9)

📦 version 7.3.1, in nixpkgs unstable

✨ rdma_rw_ctx_init(9) ✨

okay so bestie, this is a page from the Kernel API Manual and it's genuinely about setting up an RDMA READ/WRITE context — like, the whole "get your ducks in a row before you yeet work at the queue" moment 🦋

🌸 NAME 🌸

rdma_rw_ctx_init — initialize a RDMA READ/WRITE context 💅

🎀 SYNOPSIS 🎀

int rdma_rw_ctx_init (struct rdma_rw_ctx *ctx , struct ib_qp *qp , u32
port_num , struct scatterlist *sg , u32 sg_cnt , u32 sg_offset , u64
remote_addr , u32 rkey , enum dma_data_direction dir );

🫶 ARGUMENTS 🫶

like, one by one, no skipping, periodt:

  • ctx — the context you're initializing. this is the glow-up victim, bestie ✨
  • qp — the queue pair we're operating on 👑
  • port_num — the port num that this connection is bound to 🎀
  • sg — the scatterlist we're gonna READ from / WRITE to 🌸
  • sg_cnt — how many entries live in sg, like headcount for the party 🎉
  • sg_offset — the current byte offset into sg. yes, bytes, not vibes 🔥
  • remote_addr — the remote address to read/write, and it's relative to rkey (iykyk) 🌈
  • rkey — the remote key we're operating on 🔑
  • dir — DMA_TO_DEVICE for RDMA WRITE, DMA_FROM_DEVICE for RDMA READ. literally pick the right one or it's giving chaos 😭

💀 DESCRIPTION 💀

okay so here's the tea: it returns the number of WQEs that will be needed on the workqueue if it succeeds 🫶 — or a negative error code if it doesn't. that's it, that's the whole vibe check 💅

👑 SEE ALSO 👑

kernel file ./drivers/infiniband/core/rw.c — and the whole friend group:

mwah 💖 that's the whole page, go be iconic with your queue pairs — shipped out of ./drivers/infiniband/core in October 2026, no cap 🍓✨

📦 where this page comes from