Re: [Patch bpf] skmsg: check sk_rcvbuf limit before queuing to ingress_skb
flat view
From: Jakub Sitnicki <jakub@cloudflare.com>
Date: 2021-06-30 09:21:31
Also in:
bpf
On Tue, Jun 29, 2021 at 08:20 AM CEST, Cong Wang wrote: [...]
quoted hunk ↗ jump to hunk
@@ -854,7 +854,8 @@ static int sk_psock_skb_redirect(struct sk_psock *from, struct sk_buff *skb) return -EIO; } spin_lock_bh(&psock_other->ingress_lock); - if (!sk_psock_test_state(psock_other, SK_PSOCK_TX_ENABLED)) { + if (!sk_psock_test_state(psock_other, SK_PSOCK_TX_ENABLED) || + atomic_read(&sk_other->sk_rmem_alloc) > sk_other->sk_rcvbuf) { spin_unlock_bh(&psock_other->ingress_lock); skb_bpf_redirect_clear(skb); sock_drop(from->sk, skb);@@ -930,7 +931,8 @@ static int sk_psock_verdict_apply(struct sk_psock *psock, struct sk_buff *skb, } if (err < 0) { spin_lock_bh(&psock->ingress_lock); - if (sk_psock_test_state(psock, SK_PSOCK_TX_ENABLED)) { + if (sk_psock_test_state(psock, SK_PSOCK_TX_ENABLED) && + atomic_read(&sk_other->sk_rmem_alloc) <= sk_other->sk_rcvbuf) { skb_queue_tail(&psock->ingress_skb, skb); schedule_work(&psock->work); err = 0;
I belive access to sk_rcvbuf should be annotated with READ_ONCE (for
KCSAN's sake) as we don't lock the egress socket. See 8265792bf887 [1]
("net: silence KCSAN warnings around sk_add_backlog() calls") for
guidance.
[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=8265792bf8871acc2d00fd03883d830e2249d395