xe_bo_willneed_put_locked(9)

💅 xe_bo_willneed_put_locked(9) 💅

okay so bestie, we're in the Kernel API Manual today and this is literally the page for xe_bo_willneed_put_locked — yes the whole name, say it with your chest 🫶

---

✨ NAME ✨

xe_bo_willneed_put_locked — the function that releases a WILLNEED holder on a BO. that's it. that's the vibe. 💖

---

🦋 SYNOPSIS 🦋

void xe_bo_willneed_put_locked (struct xe_bo *bo );

no cap, that's the whole signature, keep it exactly like that pls 🙈

---

🎀 ARGUMENTS 🎀

• bo — the buffer object in question. the main character of this whole function ngl 👑

---

🌸 DESCRIPTION 🌸

so here's the tea ☕ — this function decrements willneed_count, and on a 1->0 transition, it marks the BO DONTNEED... but ONLY if the BO still has VMAs (which implies all the active VMAs are already DONTNEED). conditional queen 💅

and if the last VMA is the one being removed? then it preserves the current BO state, so it matches the previous VMA-walk semantics. iykyk 🧚

PURGED is terminal and the BO state is never modified. like, ever. that's a hard no. 💀

⚠️ big one bestie: the caller must hold the BO's dma-resv lock. do not skip this. this is not a drill. 🔥

---

👑 SEE ALSO 👑

• Kernel file ./drivers/gpu/drm/xe/xe_bo.h enum — that's right, a whole enum, go look at it 🍓
• xe_madv_purgeable_state(9)
• xe_bo_is_purged(9)
• xe_bo_madv_is_dontneed(9)
• xe_bo_willneed_get_locked(9)
• xe_bo_vma_count_inc_locked(9)
• xe_bo_vma_count_dec_locked(9)
• xe_bo_size(9)
• xe_bo_has_valid_ccs_bb(9)
• xe_bo_put_deferred(9)
• xe_bo_put_async(9)
• xe_bo_sg_is_contiguous(9)
• struct xe_bo_shrink_flags(9)
• xe_bo_is_mem_type(9)

---

and that's the page!! 🫶 this came from the source at ./drivers/gpu/drm/xe, dated October 2026, and the header/footer kept whispering xe_bo_willneed_put_locked(9) at me the whole time so like. here it is. slay. ⭐🌈💜