folio_pte_batch_flags(9)

📦 version 7.3.1, in nixpkgs unstable

💅 folio_pte_batch_flags — a Kernel API Manual moment 💅

okay so bestie, buckle up, because we're diving into folio_pte_batch_flags(9) from the Kernel API Manual and honestly? it's giving "niche but iconic." 🎀

✨ NAME ✨

folio_pte_batch_flags — detect a PTE batch for a large folio 🦋

🌸 SYNOPSIS 🌸

unsigned int folio_pte_batch_flags (struct folio *folio , struct
       vm_area_struct *vma , pte_t *ptep , pte_t *ptentp , unsigned int max_nr
       , fpb_t flags );

like, that's the whole vibe. six args. she's thorough. 💅

💖 ARGUMENTS 💖

  • folio — the large folio we're detecting a PTE batch for. the main character, lowkey. 👑
  • vma — the VMA. heads up bestie: this is only relevant with FPB_MERGE_WRITE — otherwise it can be NULL and nobody will be mad. 🙈
  • ptep — page table pointer for the first entry. ​the starting point, the origin story. ⭐
  • ptentp — pointer to a COPY of the first page table entry, and this function updates its flags based on flags if appropriate. please note the bold COPY, this matters so much. 😭
  • max_nr — the maximum number of table entries to consider. the limit, the boundary, the "bestie no further." 🛑
  • flags — flags that modify the PTE batch semantics. the personality settings, fr fr. 🎀

🦋 DESCRIPTION 🦋

So what does she actually do? Detect a PTE batch: consecutive (present) PTEs that map consecutive pages of the same large folio in a single VMA and a single page table. that's the whole assignment and she ate. 🔥

Every PTE inside a PTE batch has the same PTE bits set — excluding the PFN, the accessed bit, the writable bit, the dirty bit (unless FPB_RESPECT_DIRTY is set) and the soft-dirty bit (unless FPB_RESPECT_SOFT_DIRTY is set). so like, those bits are the exceptions, not the rule. 💀

ptep must map any page of the folio. max_nr must be at least one, and it must be limited by the caller so scanning can't exceed a single VMA and a single page table. no crossing the line, we stay in our lane. ✨

Depending on the FPB_MERGE_* flags, the pte stored at ptentp will be updated — and it is crucial that a pointer to a COPY of the first page table entry, obtained through ptep_get, is provided as ptentp. crucial! capital-C crucial! don't skip this step, i'm literally begging. 🙏

This function will be inlined to optimize based on the input parameters; consider using folio_pte_batch instead if applicable. she's inline-flexible like that. 🧚

🌈 RETURN 🌈

the number of table entries in the batch. that's it, that's the answer. 🫶

⭐ SEE ALSO ⭐

Kernel file ./mm/internal.h and the whole friend group:

ok real talk bestie: get the ptentp COPY right. it's the one thing this page yells at you about, and honestly? she's right. 💅

and that's the page!! source/version is ./mm and the date on this beauty is October 2026 — so yeah, she's from the future and she's thriving. mwah, go read your page tables. 💜🎀✨

📦 where this page comes from