skwq_has_sleeper(9)

🌸 skwq_has_sleeper — a kernel API manual, bestie edition 🌸

okay so this is from the Kernel API Manual, page skwq_has_sleeper(9), and honestly? it's a little baddie. let's get into it 💅

---

✨ NAME ✨

skwq_has_sleeper — check if there are any waiting processes 👀

like... is anybody in the queue? hello?? 🫶

---

🦋 SYNOPSIS 🦋

bool skwq_has_sleeper (struct socket_wq *wq );

that's it. that's the whole thing. short and iconic 💖

---

🎀 ARGUMENTS 🎀

• wq — a struct socket_wq. that's the whole description, she's mysterious like that 🙈

---

⭐ RETURN ⭐

returns true if the socket_wq has waiting processes. no cap, that's the vibe check 🔥

---

💀 DESCRIPTION 💀

okay so here's the tea, bestie. the whole point of skwq_has_sleeper and sock_poll_wait is to wrap the memory barrier call. they exist because of a race that got found hanging out in the tcp code 😭 like, actual bug energy.

consider the following tcp code paths:

CPU1                CPU2
sys_select          receive packet
...                 ...
__add_wait_queue    update tp->rcv_nxt
...                 ...
tp->rcv_nxt check   sock_def_readable
...                 {
schedule               rcu_read_lock;
                       wq = rcu_dereference(sk->sk_wq);
                       if (wq && waitqueue_active(wq->wait))
                           wake_up_interruptible(wq->wait)
                       ...
                    }

so the race for tcp fires when the __add_wait_queue changes that CPU1 did just... stay chilling in its cache. and the same goes for the tp->rcv_nxt update on the CPU2 side. 💀 so then CPU1 could end up calling schedule and sleep forever if there's no more data on the socket. forever!! not me having a whole existential crisis about a sleep that never wakes up 😭

like genuinely, that's the danger here — a process stuck sleeping with no wakeup coming. that's why the memory barrier wrapping exists. slay for correctness, not for vibes 🫶

---

👑 SEE ALSO 👑

all the siblings and cousins, in the kernel file ./include/net/sock.h:

• struct sock_common(9)
• struct sock(9)
• sk_user_data_is_nocopy(9)
• __locked_read_sk_user_data_with_flags(9)
• __rcu_dereference_sk_user_data_with_flags(9)
• sk_for_each_entry_offset_rcu(9)
• SOCK_CONNECT_BIND(9)
• lock_sock_fast(9)
• unlock_sock_fast(9)
• sk_wmem_alloc_get(9)
• sk_rmem_alloc_get(9)
• sk_has_allocations(9)
• sock_poll_wait(9)
• sk_page_frag(9)
• _sock_tx_timestamp(9)
• sk_eat_skb(9)

---

🌈 and that's the page!! it lives in ./include/net and the date on this one is October 2026 — so yeah, future-coded manual page, we love to see it. go read sock_poll_wait(9) next, iykyk 💜🎀✨