iommu_alloc_pages_node_sz(9)

💜 iommu_alloc_pages_node_sz(9) 💜

okay bestie, welcome to the Kernel API Manual, you're gonna love this one ✨

---

🌸 NAME 🌸

iommu_alloc_pages_node_sz — grab yourself a zeroed page of whatever size you want, from a specific NUMA node 🎯 no cap.

---

✨ SYNOPSIS ✨

void * iommu_alloc_pages_node_sz (int nid , gfp_t gfp , size_t size );

literally that's the whole vibe check right there 👑

---

🎀 ARGUMENTS 🎀

• nid — the memory NUMA node id, aka which node you're shopping at 🛍️
• gfp — the buddy allocator flags, the little knobs that say how you want it allocated
• size — the memory size you wanna allocate, and yes it gets rounded up to a power of 2 (more on that below, iykyk)

---

🦋 DESCRIPTION 🦋

So what you get back? The virtual address of the allocated page 💖 that's your prize.

now the important part, bestie, do NOT just leave it lying around:

The page must be freed, either by calling iommu_free_pages or via iommu_put_pages_list 🫶 fr fr, free your pages.

and here's the cute little detail about what you actually got: the returned allocation is round_up_pow_two(size) big 🔥 and it's physically aligned to its size. iconic behavior honestly 💅

---

🌈 SEE ALSO 🌈

peep the kernel file ./drivers/iommu/iommu-pages.c and then go read these besties:

• iommu_free_pages(9)
• iommu_put_pages_list(9)
• iommu_pages_start_incoherent(9)
• iommu_pages_start_incoherent_list(9)
• iommu_pages_stop_incoherent_list(9)
• iommu_pages_free_incoherent(9)

---

and that's the whole page, periodt 💅✨ this cutie lives in ./drivers/iommu, dated October 2026, and it goes by iommu_alloc_pages_node_sz(9) — mwah, go allocate responsibly 🧚💜