_copy_from_iter_flushcache(9)

📦 version 7.3.1, in nixpkgs unstable

💅 _copy_from_iter_flushcache(9) 💅

okay so this is a page from the Kernel API Manual and it's genuinely the moment you've been waiting for, bestie — it's about writing to persistent memory and actually making it stick. let's get into it 🫶

✨ NAME ✨

_copy_from_iter_flushcache — write destination through cpu cache

like yes, the whole vibe is "write it, but flush it through the cache on the way" 💫

🎀 SYNOPSIS 🎀

size_t _copy_from_iter_flushcache (void *addr , size_t bytes , struct
iov_iter *i );

🌸 ARGUMENTS 🌸

  • addr — destination kernel address. this is where the data's going, the main character of the destination situation 👑
  • bytes — total transfer length. how much we're moving, fr fr
  • i — source iterator. the origin story, the place we're reading from 🦋

💖 DESCRIPTION 💖

so the pmem driver arranges for filesystem-dax to use this facility via dax_copy_from_iter, and the whole reason is ensuring that writes to persistent memory are flushed through the CPU cache. no cap — that's the entire point. persistent memory deserves better than "idk it's probably in the cache somewhere" 💀

here's where it gets iconic: it's differentiated from _copy_from_iter_nocache in that it guarantees all data is flushed for all iterator types. all of them. every single one. slay. ✨

the _copy_from_iter_nocache only attempts to bypass the cache for the ITER_IOVEC case — and on some archs it may use instructions that strand dirty-data in the cache 🙈 which is, um, not the vibe we're going for. so if you need the guarantee, this is the function for you. periodt.

🔥 RETURN 🔥

number of bytes copied (may be 0)

literally that's it, no dramatic twist 💜

🧚 SEE ALSO 🧚

Kernel file ./lib/iov_iter.c:

go read them, iykyk 📚💅

🌈 and that's the whole page, bestie! this little gem lives in ./lib and it's dated October 2026 — so go forth, flush that cache, and remember: no dirty-data stranded in the cache on your watch. mwah 💋🎀

📦 where this page comes from