[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <4cbabdf1-af2c-490a-a41a-b40c1539c1cb@linux.dev>
Date: Mon, 3 Nov 2025 13:31:51 -0800
From: Martin KaFai Lau <martin.lau@...ux.dev>
To: Alexis Lothoré (eBPF Foundation)
<alexis.lothore@...tlin.com>
Cc: Alexei Starovoitov <ast@...nel.org>,
Daniel Borkmann <daniel@...earbox.net>, Andrii Nakryiko <andrii@...nel.org>,
Eduard Zingerman <eddyz87@...il.com>, Song Liu <song@...nel.org>,
Yonghong Song <yonghong.song@...ux.dev>,
John Fastabend <john.fastabend@...il.com>, KP Singh <kpsingh@...nel.org>,
Stanislav Fomichev <sdf@...ichev.me>, Hao Luo <haoluo@...gle.com>,
Jiri Olsa <jolsa@...nel.org>, Shuah Khan <shuah@...nel.org>,
ebpf@...uxfoundation.org, Thomas Petazzoni <thomas.petazzoni@...tlin.com>,
bpf@...r.kernel.org, linux-kselftest@...r.kernel.org,
linux-kernel@...r.kernel.org,
Bastien Curutchet <bastien.curutchet@...tlin.com>
Subject: Re: [PATCH bpf-next 3/3] selftests/bpf: use start_server_str rather
than start_reuseport_server in tc_tunnel
On 10/31/25 2:01 AM, Alexis Lothoré (eBPF Foundation) wrote:
> test_tc_tunnel currently uses start_reuseport_server because it needs to
> frequently start and stop the server, so we need SO_REUSEPORT to avoid
> getting errors on server restart due to the socket being in TIME_WAIT
> state. But the test is only using one server at a time, so it is a bit
> confusing to use this API.
>
> Replace start_reuseport with start_sever_str, and provided the relevant
> callback to set SO_REUSEPORT.
>
> Signed-off-by: Alexis Lothoré (eBPF Foundation) <alexis.lothore@...tlin.com>
> ---
> .../selftests/bpf/prog_tests/test_tc_tunnel.c | 24 +++++++++++++++-------
> 1 file changed, 17 insertions(+), 7 deletions(-)
>
> diff --git a/tools/testing/selftests/bpf/prog_tests/test_tc_tunnel.c b/tools/testing/selftests/bpf/prog_tests/test_tc_tunnel.c
> index deea90aaefad..8e3fe6dc6221 100644
> --- a/tools/testing/selftests/bpf/prog_tests/test_tc_tunnel.c
> +++ b/tools/testing/selftests/bpf/prog_tests/test_tc_tunnel.c
> @@ -69,7 +69,7 @@ struct subtest_cfg {
> int client_egress_prog_fd;
> int server_ingress_prog_fd;
> char extra_decap_mod_args[TUNNEL_ARGS_MAX_LEN];
> - int *server_fd;
> + int server_fd;
> };
>
> struct connection {
> @@ -131,20 +131,30 @@ static void set_subtest_addresses(struct subtest_cfg *cfg)
> }
> }
>
> +static int reuseport_cb(int fd, void *opts)
> +{
> + int one = 1;
> +
> + return setsockopt(fd, SOL_SOCKET, SO_REUSEPORT, &one, sizeof(one));
> +}
> +
> static int run_server(struct subtest_cfg *cfg)
> {
> int family = cfg->ipproto == 6 ? AF_INET6 : AF_INET;
> + struct network_helper_opts opts = {
> + .timeout_ms = TIMEOUT_MS,
> + .post_socket_cb = reuseport_cb,
> + };
> struct nstoken *nstoken;
>
> nstoken = open_netns(SERVER_NS);
> if (!ASSERT_OK_PTR(nstoken, "open server ns"))
> return -1;
>
> - cfg->server_fd = start_reuseport_server(family, SOCK_STREAM,
> - cfg->server_addr, TEST_PORT,
> - TIMEOUT_MS, 1);
> + cfg->server_fd = start_server_str(family, SOCK_STREAM, cfg->server_addr,
> + TEST_PORT, &opts);
I meant to directly enable SO_REUSE"ADDR" in the start_server_addr()
instead of each individual test using SO_REUSEPORT. I think all tcp server
in test_progs should have it enabled. Something like this:
diff --git a/tools/testing/selftests/bpf/network_helpers.c b/tools/testing/selftests/bpf/network_helpers.c
index 8bb09167399a..e8b33d902bbc 100644
--- a/tools/testing/selftests/bpf/network_helpers.c
+++ b/tools/testing/selftests/bpf/network_helpers.c
@@ -97,7 +97,7 @@ int settimeo(int fd, int timeout_ms)
int start_server_addr(int type, const struct sockaddr_storage *addr, socklen_t addrlen,
const struct network_helper_opts *opts)
{
- int fd;
+ int fd, on = 1;
if (!opts)
opts = &default_opts;
@@ -111,6 +111,12 @@ int start_server_addr(int type, const struct sockaddr_storage *addr, socklen_t a
if (settimeo(fd, opts->timeout_ms))
goto error_close;
+ if (type == SOCK_STREAM &&
+ setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, &on, sizeof(on))) {
+ log_err("Failed to enable SO_REUSEADDR");
+ goto error_close;
+ }
+
if (opts->post_socket_cb &&
opts->post_socket_cb(fd, opts->cb_opts)) {
log_err("Failed to call post_socket_cb");
I have applied patch 1 and 2.
Powered by blists - more mailing lists