Re: [PATCH net-next 1/2] tcp: conditionally call ip_icmp_error() from tcp_v4_err()

[David Ahern <dsahern@kernel.org>]

On 4/18/24 4:15 AM, Eric Dumazet wrote:
> 
> Thanks Paolo
> 
> I found that the following patch is fixing the issue for me.
> 
> diff --git a/tools/testing/selftests/net/nettest.c
> b/tools/testing/selftests/net/nettest.c
> index cd8a580974480212b45d86f35293b77f3d033473..ff25e53024ef6d4101f251c8a8a5e936e44e280f
> 100644
> --- a/tools/testing/selftests/net/nettest.c
> +++ b/tools/testing/selftests/net/nettest.c
> @@ -1744,6 +1744,7 @@ static int connectsock(void *addr, socklen_t
> alen, struct sock_args *args)
>         if (args->bind_test_only)
>                 goto out;
> 
> +       set_recv_attr(sd, args->version);
>         if (connect(sd, addr, alen) < 0) {
>                 if (errno != EINPROGRESS) {
>                         log_err_errno("Failed to connect to remote host");

You have a kernel patch that makes a test fail, and your solution is
changing userspace? The tests are examples of userspace applications and
how they can use APIs, so if the patch breaks a test it is by definition
breaking userspace which is not allowed.