free_pgtables(9)

📦 version 7.3.1, in nixpkgs unstable

💅 free_pgtables(9)

okay so this is one of those kernel API manual pages — section 9, bestie, we're in the deep internals now 🫶 — and it's literally about freeing a range of page tables. yes, the actual page tables. the scaffolding under the scaffolding. we love a cleanup arc ✨

🌸 NAME 🌸

free_pgtables — frees a range of page tables, no cap, that's the whole job 💅

⭐ SYNOPSIS ⭐

void free_pgtables (struct mmu_gather *tlb , struct unmap_desc *unmap );

like, that's it. that's the whole signature. void return, two pointers in, page tables out (of existence) 🔥

🎀 ARGUMENTS 🎀

  • tlb — the mmu gather. yes, that mmu gather. the lil collection bag of things-to-be-flushed 🧺
  • unmap — the unmap_desc. it's giving "the descriptor of what we're unmapping" and honestly? iconic 👑

🦋 NOTE 🦋

okay so here's the tea, and you actually need this part ☕

pg_start and pg_end are provided to indicate the absolute range of the page tables that should be removed. and listen — this range can differ from the vma mappings on some archs, specifically archs that may have mappings that need to be removed outside the vmas 🙈 so don't just assume the vmas tell the whole story, bestie.

also, prev->vm_end and next->vm_start are often used. not always, but often. iykyk 💅

and then there's this one: vma_end differs from pg_end when a dup_mmap failed and the tree has unrelated data to the mm_struct being torn down 😭 so if you're sitting there like "why don't these match", that's why. it's not a bug, it's the failure path, and she is messy 💀

🌈 SEE ALSO 🌈

kernel file ./mm/memory.c, plus this absolutely stacked cast of characters — genuinely a group chat worth joining 🫶

and that's the page, bestie 💖 this one lives in ./mm, dated October 2026, and it's filed as free_pgtables(9) — go forth and free those page tables responsibly, mwah 🎀✨

📦 where this page comes from