css_next_descendant_pre(9)

📦 version 7.3.1, in nixpkgs unstable

💅 css_next_descendant_pre(9) — find the next descendant for pre-order walk

okay bestie, quick vibe check: this is a page from the Kernel API Manual, and it's genuinely one of those lil functions that runs the whole cgroup family tree walk. not me being dramatic — it's literally the iterator logic behind css_for_each_descendant_pre. let's go 🎀

✨ SYNOPSIS ✨

struct cgroup_subsys_state * css_next_descendant_pre (struct
cgroup_subsys_state *pos , struct cgroup_subsys_state *root );

like, that's the whole thing. she takes two css pointers and hands you back another one 🦋

🌸 ARGUMENTS 🌸

  • pos — where you're standing right now, bestie. pass NULL to kick off the traversal from scratch 🏁
  • root — the css whose descendants we're walking. she's the main character of this whole operation 👑

💖 DESCRIPTION 💖

so this one is meant to be used by css_for_each_descendant_pre — it's the lil engine that finds the next descendant to visit when you're doing a pre-order traversal of root's descendants. and yes, root is included in the iteration — she's literally the first node visited. no cap 👑

now, the locking tea: this function needs cgroup_mutex or RCU read locking. but listen — it does not require the whole traversal to live inside a single critical section. that's the slay part. also, you do not have to hold onto a reference to pos. 💅 she'll still hand you the correct next descendant as long as both pos and root are accessible and pos is a descendant of root. iconic behaviour, fr fr.

about visibility (this is the part where you gotta pay attention, bestie 🫶): if a subsystem synchronizes ->css_online with the start of iteration, then a css that finished ->css_online is guaranteed to be visible in future iterations and will stay visible until the last reference is put. but a css that hasn't finished ->css_online, or one that already finished ->css_offline, may still show up during traversal 😭 — it's genuinely just out there living its life. so it is each subsystem's responsibility to synchronize against on/offlining. that's the assignment, not ours ✨

lowkey important: the on/offlining synchronization is on you (well, on each subsystem). this page is very clear that it won't do it for you 💀

🦋 SEE ALSO 🦋

Kernel file ./kernel/cgroup/cgroup.c, plus a truly unhinged amount of friends:

like… that's a whole group chat right there 😭💀

mwah, that's the page 💜 this lil baddie lives in ./kernel/cgroup and the docs are dated October 2026 — go forth and walk those descendants in pre-order, bestie, you've got this 🫶✨

📦 where this page comes from