Re: [net-next,v9,06/15] quic: add stream management

This is an AI-generated review of your patch. The human sending this
email has considered the AI review valid, or at least plausible.

For local reproduction steps see https://netdev-ai.bots.linux.dev/ai-local.html
---
quic: add stream management

This patch introduces stream management infrastructure for the QUIC
protocol, including a hash table for tracking active streams, functions
for acquiring and releasing streams, and logic for detecting stream
ID limits.

> diff --git a/net/quic/socket.c b/net/quic/socket.c
> index b86989080184..d6f25669c693 100644
> --- a/net/quic/socket.c
> +++ b/net/quic/socket.c
> @@ -41,6 +41,9 @@ static int quic_init_sock(struct sock *sk)
>  	sk->sk_write_space = quic_write_space;
>  	sock_set_flag(sk, SOCK_USE_WRITE_QUEUE);
>
> +	if (quic_stream_init(quic_streams(sk)))
> +		return -ENOMEM;

If quic_stream_init() fails here, the function returns before calling
sk_sockets_allocated_inc() and sock_prot_inuse_add() below. However,
when initialization fails, inet_create() calls sk_common_release() which
invokes quic_destroy_sock():

quic_init_sock()
    quic_stream_init() fails
    return -ENOMEM
        -> sk_common_release()
            -> quic_destroy_sock()
                -> sk_sockets_allocated_dec()  // underflow
                -> sock_prot_inuse_add(..., -1)  // underflow

Can this cause counter underflow since the increments never executed but
the decrements in quic_destroy_sock() are unconditional?

> +
>  	WRITE_ONCE(sk->sk_sndbuf, READ_ONCE(sysctl_quic_wmem[1]));
>  	WRITE_ONCE(sk->sk_rcvbuf, READ_ONCE(sysctl_quic_rmem[1]));
>
> @@ -52,6 +55,8 @@ static int quic_init_sock(struct sock *sk)
>
>  static void quic_destroy_sock(struct sock *sk)
>  {
> +	quic_stream_free(quic_streams(sk));
> +
>  	quic_data_free(quic_ticket(sk));

[ ... ]