https://sankethbk.github.io/blog/tags/page-reclaim/Recent content in Page Reclaim on Sanketh's BlogHugo -- 0.166.0en-usSat, 22 Aug 2026 00:00:00 +0000https://sankethbk.github.io/blog/posts/operating-systems/linux/2026-08-22-page-reclaim/Sat, 22 Aug 2026 00:00:00 +0000https://sankethbk.github.io/blog/posts/operating-systems/linux/2026-08-22-page-reclaim/<h1 id="page-frame-reclaiming">Page Frame Reclaiming</h1> <h2 id="the-page-frame-reclaiming-algorithm">The Page Frame Reclaiming Algorithm</h2> <h3 id="1-unified-page-cache--dual-caching-edge-case">1. Unified Page Cache &amp; Dual-Caching Edge Case</h3> <ul> <li> <p><strong>Unified Page Cache:</strong> Linux no longer separates the &ldquo;disk/buffer cache&rdquo; from the &ldquo;page cache.&rdquo; All file and block I/O goes through a <strong>single unified Page Cache</strong> indexed by <code>(address_space, offset)</code>.</p> </li> <li> <p><strong>The Dual-Caching Exception:</strong></p> </li> <li> <p>Reading via a regular path (e.g., <code>/app/data.db</code>) uses the <strong>file inode&rsquo;s</strong> <code>address_space</code>.</p> </li> <li> <p>Reading via the raw disk node (e.g., <code>/dev/sda1</code>) uses the <strong>block device inode&rsquo;s</strong> <code>address_space</code>.</p>