cond_synchronize_rcu(9)

📦 version 7.3.1, in nixpkgs unstable

💅 cond_synchronize_rcu(9) 💅

okay bestie, welcome to the Kernel API Manual — this one's cond_synchronize_rcu, and it's literally the "wait for an RCU grace period, but only if you have to" girlie. conditional queen behavior, we stan 🎀

✨ NAME ✨

cond_synchronize_rcu — conditionally wait for an RCU grace period 🦋

🌸 SYNOPSIS 🌸

void cond_synchronize_rcu (unsigned long oldstate );

🎀 ARGUMENTS 🎀

  • oldstate — the value you got back from get_state_synchronize_rcu, start_poll_synchronize_rcu, or start_poll_synchronize_rcu_expedited. like, it's the receipt from your earlier call, keep it safe 💖

💅 DESCRIPTION 💅

so here's the vibe check 👑: if a full RCU grace period has already elapsed since that earlier call to get_state_synchronize_rcu or start_poll_synchronize_rcu, then this function just returns. nothing to do, no waiting, we're done, slay ✨

otherwise? it invokes synchronize_rcu and actually waits for a full grace period. no cap, sometimes you gotta wait 😭

yes, this function does not take counter wrap into account. but bestie — counter wrap is harmless 🙈. if the counter wraps, we've waited for more than 2 billion grace periods (and way more on a 64-bit system!), so waiting for a couple of additional grace periods should be just fine. like genuinely, it's giving "not a problem" 💀

okay and this part is highkey important: this function provides the same memory-ordering guarantees that would be provided by a synchronize_rcu that was invoked at the call to the function that provided oldstate and that returned at the end of this function. periodt 🔥

🌈 SEE ALSO 🌈

Kernel file ./kernel/rcu/tree.c, plus the whole friend group:

and that's the whole page, bestie 🫶 this lil icon lives in ./kernel/rcu and the version/date stamp on it says October 2026 — so go forth, wait conditionally, and remember: the counter wrap cannot hurt you 🧚💜

📦 where this page comes from