From: Andrii Nakryiko <andrii.nakryiko@gmail.com>
To: Stanislav Fomichev <sdf@fomichev.me>
Cc: Andrii Nakryiko <andriin@fb.com>, bpf <bpf@vger.kernel.org>,
Networking <netdev@vger.kernel.org>,
Alexei Starovoitov <ast@fb.com>,
Daniel Borkmann <daniel@iogearbox.net>,
Stanislav Fomichev <sdf@google.com>,
Kernel Team <kernel-team@fb.com>
Subject: Re: [PATCH bpf-next] selftests/bpf: make tcp_rtt test more robust to failures
Date: Wed, 11 Mar 2020 14:01:38 -0700 [thread overview]
Message-ID: <CAEf4BzZpL83aAhDWTyNoXtJp5W8S4Q_=+2_0UNeY=eb14hS8aQ@mail.gmail.com> (raw)
In-Reply-To: <20200311204106.GA2125642@mini-arch.hsd1.ca.comcast.net>
On Wed, Mar 11, 2020 at 1:41 PM Stanislav Fomichev <sdf@fomichev.me> wrote:
>
> On 03/11, Andrii Nakryiko wrote:
> > Switch to non-blocking accept and wait for server thread to exit before
> > proceeding. I noticed that sometimes tcp_rtt server thread failure would
> > "spill over" into other tests (that would run after tcp_rtt), probably just
> > because server thread exits much later and tcp_rtt doesn't wait for it.
> >
> > Fixes: 8a03222f508b ("selftests/bpf: test_progs: fix client/server race in tcp_rtt")
> > Signed-off-by: Andrii Nakryiko <andriin@fb.com>
> > ---
> > .../selftests/bpf/prog_tests/tcp_rtt.c | 30 +++++++++++--------
> > 1 file changed, 18 insertions(+), 12 deletions(-)
> >
> > diff --git a/tools/testing/selftests/bpf/prog_tests/tcp_rtt.c b/tools/testing/selftests/bpf/prog_tests/tcp_rtt.c
> > index f4cd60d6fba2..d235eea0de27 100644
> > --- a/tools/testing/selftests/bpf/prog_tests/tcp_rtt.c
> > +++ b/tools/testing/selftests/bpf/prog_tests/tcp_rtt.c
> > @@ -188,7 +188,7 @@ static int start_server(void)
> > };
> > int fd;
> >
> > - fd = socket(AF_INET, SOCK_STREAM, 0);
> > + fd = socket(AF_INET, SOCK_STREAM | SOCK_NONBLOCK, 0);
> > if (fd < 0) {
> > log_err("Failed to create server socket");
> > return -1;
> > @@ -205,6 +205,7 @@ static int start_server(void)
> >
> > static pthread_mutex_t server_started_mtx = PTHREAD_MUTEX_INITIALIZER;
> > static pthread_cond_t server_started = PTHREAD_COND_INITIALIZER;
> > +static volatile bool server_done = false;
> >
> > static void *server_thread(void *arg)
> > {
> > @@ -222,23 +223,22 @@ static void *server_thread(void *arg)
> >
> > if (CHECK_FAIL(err < 0)) {
> > perror("Failed to listed on socket");
> > - return NULL;
> > + return ERR_PTR(err);
> > }
> >
> > - client_fd = accept(fd, (struct sockaddr *)&addr, &len);
> > + while (!server_done) {
> > + client_fd = accept(fd, (struct sockaddr *)&addr, &len);
> > + if (client_fd == -1 && errno == EAGAIN)
> > + continue;
> > + break;
> > + }
> > if (CHECK_FAIL(client_fd < 0)) {
> > perror("Failed to accept client");
> > - return NULL;
> > + return ERR_PTR(err);
> > }
> >
> > - /* Wait for the next connection (that never arrives)
> > - * to keep this thread alive to prevent calling
> > - * close() on client_fd.
> > - */
> > - if (CHECK_FAIL(accept(fd, (struct sockaddr *)&addr, &len) >= 0)) {
> > - perror("Unexpected success in second accept");
> > - return NULL;
> > - }
> > + while (!server_done)
> > + usleep(50);
> >
> > close(client_fd);
> >
> > @@ -249,6 +249,7 @@ void test_tcp_rtt(void)
> > {
> > int server_fd, cgroup_fd;
> > pthread_t tid;
> > + void *server_res;
> >
> > cgroup_fd = test__join_cgroup("/tcp_rtt");
> > if (CHECK_FAIL(cgroup_fd < 0))
> > @@ -267,6 +268,11 @@ void test_tcp_rtt(void)
> > pthread_mutex_unlock(&server_started_mtx);
> >
> > CHECK_FAIL(run_test(cgroup_fd, server_fd));
> > +
> > + server_done = true;
>
> [..]
> > + pthread_join(tid, &server_res);
> > + CHECK_FAIL(IS_ERR(server_res));
>
> I wonder if we add (move) close(server_fd) before pthread_join(), can we
> fix this issue without using non-blocking socket? The accept() should
> return as soon as server_fd is closed so it's essentially your
> 'server_done'.
That was my first attempt. Amazingly, closing listening socket FD
doesn't unblock accept()...
>
> > +
> > close_server_fd:
> > close(server_fd);
> > close_cgroup_fd:
> > --
> > 2.17.1
> >
next prev parent reply other threads:[~2020-03-11 21:01 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-03-11 19:15 [PATCH bpf-next] selftests/bpf: make tcp_rtt test more robust to failures Andrii Nakryiko
2020-03-11 20:41 ` Stanislav Fomichev
2020-03-11 21:01 ` Andrii Nakryiko [this message]
2020-03-11 22:14 ` Stanislav Fomichev
2020-03-11 22:24 ` Andrii Nakryiko
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='CAEf4BzZpL83aAhDWTyNoXtJp5W8S4Q_=+2_0UNeY=eb14hS8aQ@mail.gmail.com' \
--to=andrii.nakryiko@gmail.com \
--cc=andriin@fb.com \
--cc=ast@fb.com \
--cc=bpf@vger.kernel.org \
--cc=daniel@iogearbox.net \
--cc=kernel-team@fb.com \
--cc=netdev@vger.kernel.org \
--cc=sdf@fomichev.me \
--cc=sdf@google.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).