rvt_qp_valid_operation(9)

πŸ“¦ version 7.3.1, in nixpkgs unstable

πŸŽ€ NAME πŸŽ€

rvt_qp_valid_operation β€” validate post send wr request. This page is from the Kernel API Manual, and it’s giving β€œquick little validation moment before the send wr goes anywhere.” ✨

🌈 SYNOPSIS 🌈

int rvt_qp_valid_operation (struct rvt_qp *qp , const struct
       rvt_operation_params *post_parms , const struct ib_send_wr *wr );

🫢 ARGUMENTS 🫢

  • qp β€” the qp. That’s the one, bestie.
  • post_parms β€” the post send table for the driver. Driver’s table, driver’s rules.
  • wr β€” the work request. The wr in question, front and center.

πŸ’… DESCRIPTION πŸ’…

Okay so this routine validates the operation based on the validation table and returns the length of the operation, which can extend beyond ib_send_bw. Operation-dependent flags key the atomic operation validation. πŸ”₯

There is an exception for UD qps: it validates the pd and overrides the length to include the additional UD-specific length. UD representation, we love to see it. 🌸

Returns a negative error or the length of the work request for building the swqe. That’s the return vibe: negative error, or the swqe-building length. Periodt. πŸ‘‘

πŸ”— SEE ALSO πŸ”—

Kernel file ./drivers/infiniband/sw/rdmavt/qp.c is the home base, and the whole squad is:

kisses from source/version ./drivers/infiniband/sw/rdmavt, dated October 2026, and this page is rvt_qp_valid_operation(9) β€” bye bestie πŸ’–βœ¨

πŸ“¦ where this page comes from