Re: [PATCH bpf-next v5 3/4] selftests/bpf: add bpf_icmp_send kfunc IPv6 tests

[Mahe Tardy <mahe.tardy@gmail.com>]

On Fri, May 15, 2026 at 08:44:44PM +0000, sashiko-bot@kernel.org wrote:
> Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
> - [Medium] Buffer truncation and type confusion when calling getsockname() for IPv6 sockets.
> --
> 
> commit 3f16e831f177058fc11380fe48e08f94594b83bf
> Author: Mahe Tardy <mahe.tardy@gmail.com>
> 
> selftests/bpf: add bpf_icmp_send kfunc IPv6 tests
> 
> This commit extends the existing bpf_icmp_send kfunc tests with IPv6
> support. It updates the test suite to handle IPv6 connections and
> ICMPv6 error queues.
> 
> > diff --git a/tools/testing/selftests/bpf/prog_tests/icmp_send_kfunc.c b/tools/testing/selftests/bpf/prog_tests/icmp_send_kfunc.c
> > index b98c0312adadc..d9badfc6e620c 100644
> > --- a/tools/testing/selftests/bpf/prog_tests/icmp_send_kfunc.c
> > +++ b/tools/testing/selftests/bpf/prog_tests/icmp_send_kfunc.c
> 
> [ ... ]
> 
> > @@ -78,14 +92,13 @@ static void read_icmp_errqueue(int sockfd, int expected_code)
> >  }
> >  
> >  static void trigger_prog_read_icmp_errqueue(struct icmp_send *skel,
> > -					    int code)
> > +					    int code, int af, const char *ip)
> >  {
> >  	int srv_fd = -1, client_fd = -1;
> >  	struct sockaddr_in addr;
> >  	socklen_t len = sizeof(addr);
> >  
> > -	srv_fd = start_server(AF_INET, SOCK_STREAM, "127.0.0.1", 0,
> > -			      TIMEOUT_MS);
> > +	srv_fd = start_server(af, SOCK_STREAM, ip, 0, TIMEOUT_MS);
> >  	if (!ASSERT_GE(srv_fd, 0, "start_server"))
> >  		return;
> >  
> > @@ -94,6 +107,8 @@ static void trigger_prog_read_icmp_errqueue(struct icmp_send *skel,
> >  		return;
> >  	}
> >  	skel->bss->server_port = ntohs(addr.sin_port);
> 
> When af is AF_INET6, start_server() creates an IPv6 socket, but addr
> is hardcoded as struct sockaddr_in which is 16 bytes.
> 
> Does this cause getsockname() to silently truncate the 28-byte
> sockaddr_in6 structure because len is initialized to sizeof(addr)?
> 
> While reading addr.sin_port happens to work because sin_port and sin6_port
> share the same offset, would it be better to use struct sockaddr_storage 
> here (like in connect_to_fd_nonblock() earlier in the file) and extract
> the port conditionally based on the address family to avoid type confusion?

I don't think we really care about this since we are just reading the
sin_port.

> 
> -- 
> Sashiko AI review · https://sashiko.dev/#/patchset/20260515194746.50920-1-mahe.tardy@gmail.com?part=3