iommufd_hwpt_paging_alloc(9)

💅 iommufd_hwpt_paging_alloc(9) — from the Kernel API Manual, bestie

okay so real talk, this one is a function, and it lives in section 9 of the kernel API manual. yes, the fancy one. get comfy 🎀

---

✨ NAME ✨

iommufd_hwpt_paging_alloc — bestie, this is how you get a PAGING iommu_domain for a device. like, that's its whole job. iconic of it honestly 👑

---

🌸 SYNOPSIS 🌸

struct iommufd_hwpt_paging * iommufd_hwpt_paging_alloc (struct
iommufd_ctx *ictx , struct iommufd_ioas *ioas , struct iommufd_device
*idev , ioasid_t pasid , u32 flags , bool immediate_attach , const
struct iommu_user_data *user_data );

look at all those parameters. she has range 💜

---

🦋 ARGUMENTS 🦋

• ictx — the iommufd context. the mothership, if you will 🛸
• ioas — the IOAS you're linking this domain up with. it's giving "we're official now" 🫶
• idev — the device you're grabbing an iommu_domain for. main character device energy 🔥
• pasid — the PASID you're grabbing an iommu_domain for. yes, per-PASID, no cap
• flags — flags coming straight from userspace. whatever they handed you 📥
• immediate_attach — true if idev should get attached to the hwpt. like, right away. no waiting around ⚡
• user_data — the user-provided, driver-specific data describing the domain you're about to create. it's the blueprint, basically 📝

---

🧚 DESCRIPTION 🧚

So you want a new iommu_domain? Say less. This function allocates one and hands it back to you as a hw_pagetable. The HWPT gets linked to the ioas you passed in, and by the time you get your grubby little hands back on it, the underlying iommu_domain is fully populated — like, entirely. done. finished. slay 💅

⚠️ okay this part is important, bestie, don't skip it: the caller must hold the ioas->mutex until after iommufd_object_abort_and_destroy or iommufd_object_finalize has been called on the returned hwpt. that's iommufd_object_abort_and_destroy or iommufd_object_finalize — pick whichever one fits, but the mutex stays held through it. no early letting go. hold it. 🔒

---

🌈 SEE ALSO 🌈

• Kernel file ./drivers/iommu/iommufd/hw_pagetable.c — go read the source, it's right there 🍓
• iommufd_hwpt_nested_alloc(9) — the nested sibling 👯
• iommufd_viommu_alloc_hwpt_nested(9) — and the viommu nested one too 🦋

---

kk that's the whole page!! 💖 this came out of ./drivers/iommu/iommufd, dated October 2026, and the page itself is iommufd_hwpt_paging_alloc(9) — remember the mutex thing, and drink some water. mwah 🎀✨