svc_rdma_listen_handler(9)

💅 svc_rdma_listen_handler 💅

okay so bestie, welcome to the svc_rdma_listen_handler(9) glow-up — it's a page from the Kernel API Manual and it's lowkey iconic. it's all about catching the CM events that show up on a listening endpoint, and honestly? she's doing the most with barely any lines. let's get into it ✨

---

🌸 SYNOPSIS 🌸

int svc_rdma_listen_handler (struct rdma_cm_id *cma_id , struct rdma_cm_event *event );

that's the whole vibe. two args, returns an int, no cap. 🫶

---

🎀 ARGUMENTS 🎀

• cma_id — the server's listener rdma_cm_id, aka the main character of this function. it's the id doing the listening, periodt.
• event — literally the details of the event. that's it, that's the tea. 🍓

---

✨ DESCRIPTION ✨

so here's how she reports back — the return values are doing a whole "stay or go" thing:

• 0 — do not destroy cma_id. she's staying, leave her alone 💖
• 1 — destroy cma_id. yeah bestie, it's over, time to clean up 💀

NB: there is never a DEVICE_REMOVAL event for INADDR_ANY listeners. like, never ever. don't come looking for one, she's not showing up 🙈

---

🦋 SEE ALSO 🦋

• Kernel file ./net/sunrpc/xprtrdma/svc_rdma_transport.c — go read the source, iykyk 👑
• svc_rdma_xprt_deferred_close(9)
• svc_rdma_cma_handler(9)
• svc_rdma_free(9)

---

and that's the whole page, bestie — go off and handle those CM events 💅✨

mwah 💋 this one came to you from ./net/sunrpc/xprtrdma, dated October 2026, and it's svc_rdma_listen_handler(9) forever and always 🎀