rvt_qp_complete_swqe(9)

πŸ“¦ version 7.3.1, in nixpkgs unstable

πŸ’… rvt_qp_complete_swqe(9) β€” the one where we insert a send completion

Okay bestie, so this is a page straight out of the Kernel API Manual (that's the manual name, bestie, we're in kernel-land now 🧚) and honestly it's a little function with a big job. Its whole thing? Insert send completion. That's the NAME, that's the vibe, that's the moment. Periodt. ✨


✨ SYNOPSIS ✨

u32 rvt_qp_complete_swqe (struct rvt_qp *qp , struct rvt_swqe *wqe ,
enum ib_wc_opcode opcode , enum ib_wc_status status );

Like, it returns a u32 and takes four things. Four! We love a tidy signature. πŸŽ€


🌸 ARGUMENTS 🌸

  • qp β€” the qp. That's it. That's the whole description. The qp is the qp, no cap. πŸ€·β€β™€οΈ
  • wqe β€” the send wqe. Your send work queue entry, the little guy we're finishing up.
  • opcode β€” the wc operation, and heads up, it's driver dependent, so don't come for me if your driver's got opinions. iykyk. πŸ’…
  • status β€” the completion status. How did it go? That's what this is. πŸ¦‹

πŸ”₯ DESCRIPTION πŸ”₯

So here's the lore, bestie. First thing it does is update the s_last information (yes, all of it, that's the assignment), and then it inserts a send completion into the completion queue β€” but only if the qp indicates it should be done. Read that again. Conditional. It's not out here inserting completions willy-nilly, it's checking with the qp first. Respect. πŸ‘‘

And if you want the actual rules on completion control? The page points you at IBTA 10.7.3.1. That's where the info lives. Go read it, it's giving lore drop. πŸ“–

Ngl, this is the kind of function where the description is short but the consequences are real β€” you're doing completion control, so the qp gets the final say on whether that completion goes in. No guessing beyond what the page says. 🫢

⭐ RETURN ⭐

Returns the new last. Yeah. That's the return value, that's what you get back, that's the whole thing. Take it and go slay. πŸ’–


🌈 SEE ALSO 🌈

Buckle up, this list is long and honestly it ate. Everything here is a (9) reference, so they're all kernel API pages too, bestie:

Like, rvt_send_cq(9) and ib_cq_tail(9) / ib_cq_head(9) are literally right there next door and yes that is exactly the energy. Obsessed. πŸ’œ


πŸ’– that's the page bestie, go forth and complete some swqes πŸ’–

Sign-off time, cuties: this lil masterpiece lives in ./include/rdma and it was stamped October 2026. Bookmark it, screenshot it, put it on your fridge. πŸŽ€πŸ¦‹βœ¨

πŸ“¦ where this page comes from