sdma_ahg_alloc(9)

💖 sdma_ahg_alloc 💖

okay so bestie, we're in the Kernel API Manual today — yes, the fancy kernel docs, and no, you don't need to be scared, because this one is literally so tiny. It's giving "one job, does it well" energy. Let's get into it ✨

---

🌸 NAME 🌸

sdma_ahg_alloc — allocate an AHG entry 🎀

Like, that's it. That's the whole vibe. It grabs you an AHG entry. Slay.

---

🦋 SYNOPSIS 🦋

int sdma_ahg_alloc (struct sdma_engine *sde );

one function, one argument, one return value. minimalist queen behavior honestly 💅

---

💜 ARGUMENTS 💜

• sde — the engine you're allocating from, bestie. that's the struct sdma_engine * right there. you point at it, it gives you stuff. simple as 🫶

---

⭐ RETURN ⭐

okay so here's the vibe check on what comes back at you:

• 0-31 — success!! 🎉 you got an entry, and the number it hands you back is somewhere in that range. iconic.
• -EOPNOTSUPP — rip 😭 AHG is not enabled, so there's nothing to allocate, and the kernel is politely telling you "nope, not supported."
• -ENOSPC — no cap, there's just no entry available right now. everything's taken. come back later 🌈

like, fr fr, don't assume you always get a happy little number back — check for those negative returns or you WILL have a bad time 💀

---

🔥 SEE ALSO 🔥

the whole friend group, gathered for you 👑

• Kernel file ./drivers/infiniband/hw/hfi1/sdma.c
• sdma_state_name(9)
• sdma_get_descq_cnt(9)
• sdma_engine_get_vl(9)
• sdma_select_engine_vl(9)
• sdma_select_engine_sc(9)
• sdma_seqfile_dump_cpu_list(9)
• sdma_map_init(9)
• sdma_clean(9)
• sdma_init(9)
• sdma_all_running(9)
• sdma_start(9)
• sdma_exit(9)
• __sdma_txclean(9)
• sdma_engine_error(9)
• sdma_seqfile_dump_sde(9)
• sdma_send_txreq(9)
• sdma_send_txlist(9)
• sdma_ahg_free(9)
• _sdma_engine_progress_schedule(9)

---

and that's the whole page, bestie, you ate that 🍓 go forth and allocate responsibly 🧚

this came out of ./drivers/infiniband/hw/hfi1, dated October 2026 — mwah, love u, bye 💜