fscrypt_encr...cache_blocks(9)

💅 fscrypt_encrypt_pagecache_blocks(9) — like, the whole kernel API manual vibe, bestie ✨

okay so real talk, this page lives in the Kernel API Manual and it's the whole fscrypt_encr...cache_blocks(9) situation. yes the title got truncated, no that's not my fault, the terminal is just built like that 🙈 anyway. sit down, grab a strawberry drink 🍓, we're learning how pagecache folios get encrypted.

---

✨ NAME ✨

fscrypt_encrypt_pagecache_blocks — encrypt data from a pagecache folio 🎀

like... that's literally her job. that's the whole thing.

---

🌸 SYNOPSIS 🌸

struct page * fscrypt_encrypt_pagecache_blocks (struct folio *folio ,
size_t len , size_t offs , gfp_t gfp_flags );

look at that signature. iconic. no notes. 🔥

---

🦋 ARGUMENTS 🦋

• folio — the locked pagecache folio holding the data you wanna encrypt. key word being locked, bestie, do not come in here unlocked 🔒
• len — size of the data to encrypt, in bytes 📏
• offs — offset within the page where the data to encrypt starts, in bytes 📍
• gfp_flags — memory allocation flags; the deets are down below, keep reading 👀

---

💖 DESCRIPTION 💖

so what actually happens? this thingy allocates a brand new bounce page and encrypts the given data into it. that's the move. 🫶

the length and the offset of the data must be aligned to the file's crypto data unit size, no cap. and here's the cute part: aligning to the filesystem block size totally fulfills this requirement, because the filesystem block size is always a multiple of the data unit size. math said slay. 🧚

in the bounce page, the ciphertext data ends up at the same offset where the plaintext data was sitting in the source page. matching. coordinated. outfits on point. 👑

⚠️ okay but any other parts of that bounce page? left uninitialized. do not go reading them, do not go trusting them, they are not your friends. 💀

this one is for use by the ->writepages method of non-block-based filesystems. iykyk.

now the allocation tea ☕ — the bounce page allocation is mempool-backed, so it will always succeed when gfp_flags includes __GFP_DIRECT_REclaim... wait no, __GFP_DIRECT_RECLAIM, e.g. when it's GFP_NOFS. slay, guaranteed. 💅

however (and this is the big however, the main character plot twist): only the first page of each bio can be allocated this way. to prevent deadlocks, for any additional pages you must use a mask like GFP_NOWAIT. periodt. don't skip this part or you're gonna have a bad time 😭

---

🌈 RETURN 🌈

• ✅ the new encrypted bounce page on success
• 💀 an ERR_PTR on failure

that's it. that's the whole return. she's simple, she's effective.

---

⭐ SEE ALSO ⭐

go touch some other pages, bestie:

• Kernel file ./fs/crypto/crypto.c
• fscrypt_free_bounce_page(9)
• fscrypt_encrypt_block_inplace(9)
• fscrypt_decrypt_block_inplace(9)
• fscrypt_initialize(9)

---

okay that's the whole page!! mwah 💋 this little baddie comes to you straight out of ./fs/crypto, dated October 2026, and yes she's still doing the fscrypt_encr...cache_blocks(9) thing in the corner. go encrypt responsibly, ily, bye 🎀✨🦋