xe_page_reclaim_skip(9)

📦 version 7.3.1, in nixpkgs unstable

💅 xe_page_reclaim_skip(9) — the kickstart kernel API manual, bestie edition 🎀

Okay so real talk: this one lives in section 9 of the kernel API manual, and it's a lil boolean gatekeeper that decides whether PRL should be skipped for a VMA. Lowkey iconic behaviour, honestly. ✨

🌸 NAME 🌸

xe_page_reclaim_skip — decide whether PRL should be skipped for a VMA. That's it, that's the job, and she does it flawlessly. 👑

🦋 SYNOPSIS 🦋

bool xe_page_reclaim_skip (struct xe_tile *tile , struct xe_vma *vma );

Look at her. Two args, one bool return. No notes. 🫶

🎀 ARGUMENTS 🎀

  • tile — the struct xe_tile * that owns the VMA. Like, whose VMA is this? This tile's. Periodt.
  • vma — the struct xe_vma * currently under consideration. She's on the vibe check list. 💅

✨ DESCRIPTION ✨

So here's the lore: PPC flushing may be handled by HW for specific PAT encodings. Yes, the hardware can just do it for you sometimes, slay. 💖

And because of that, we skip PPC flushing/Page Reclaim for the scenarios listed below — the flushes would be redundant, and redundant flushes are not the moment. 🙈

  • pat_index is transient display (1). That's the whole bullet. She's a short queen but she matters. 👑

And for cases of NULL VMA, there should be no corresponding PRL entry, so skip over. Not me checking and finding nothing to reclaim 🫶 — if there's no VMA, there's no PRL entry to worry about, so we just move on.

🔥 RETURN 🔥

  • true when page reclamation is unnecessary 🌈
  • false otherwise 💀

⭐ SEE ALSO ⭐

Kernel file ./drivers/gpu/drm/xe/xe_page_reclaim.c

Go read the whole friend group, they're all linked and they all ate. 👑

And that's the page, bestie!! 🧚 Big thank-you to ./drivers/gpu/drm/xe for hosting this whole situation, October 2026 edition — mwah 💜, and this has been xe_page_reclaim_skip(9), signing off. 🍓

📦 where this page comes from