netdev - Re: [PATCH bpf-next 2/2] bpf: Add tests for bpf_sk_storage to bpf_tcp

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20200320053358.kkjljdhd4wh6l5nc@kafai-mbp>
Date:   Thu, 19 Mar 2020 22:33:58 -0700
From:   Martin KaFai Lau <kafai@...com>
To:     Yonghong Song <yhs@...com>
CC:     <bpf@...r.kernel.org>, Alexei Starovoitov <ast@...nel.org>,
        Daniel Borkmann <daniel@...earbox.net>, <kernel-team@...com>,
        <netdev@...r.kernel.org>
Subject: Re: [PATCH bpf-next 2/2] bpf: Add tests for bpf_sk_storage to
 bpf_tcp_ca

On Thu, Mar 19, 2020 at 08:49:22PM -0700, Yonghong Song wrote:
> 
> 
> On 3/19/20 4:50 PM, Martin KaFai Lau wrote:
> > This patch adds test to exercise the bpf_sk_storage_get()
> > and bpf_sk_storage_delete() helper from the bpf_dctcp.c.
> > 
> > Signed-off-by: Martin KaFai Lau <kafai@...com>
> > ---
> >   .../selftests/bpf/prog_tests/bpf_tcp_ca.c     | 28 +++++++++++++++++--
> >   tools/testing/selftests/bpf/progs/bpf_dctcp.c | 16 +++++++++++
> >   2 files changed, 41 insertions(+), 3 deletions(-)
> > 
> > diff --git a/tools/testing/selftests/bpf/prog_tests/bpf_tcp_ca.c b/tools/testing/selftests/bpf/prog_tests/bpf_tcp_ca.c
> > index 8482bbc67eec..9aaecce0bc3c 100644
> > --- a/tools/testing/selftests/bpf/prog_tests/bpf_tcp_ca.c
> > +++ b/tools/testing/selftests/bpf/prog_tests/bpf_tcp_ca.c
> > @@ -11,6 +11,7 @@
> >   static const unsigned int total_bytes = 10 * 1024 * 1024;
> >   static const struct timeval timeo_sec = { .tv_sec = 10 };
> >   static const size_t timeo_optlen = sizeof(timeo_sec);
> > +static int expected_stg = 0xeB9F;
> >   static int stop, duration;
> >   static int settimeo(int fd)
> > @@ -88,7 +89,7 @@ static void *server(void *arg)
> >   	return NULL;
> >   }
> > -static void do_test(const char *tcp_ca)
> > +static void do_test(const char *tcp_ca, const struct bpf_map *sk_stg_map)
> >   {
> >   	struct sockaddr_in6 sa6 = {};
> >   	ssize_t nr_recv = 0, bytes = 0;
> > @@ -110,6 +111,14 @@ static void do_test(const char *tcp_ca)
> >   		return;
> >   	}
> > +	if (sk_stg_map) {
> > +		err = bpf_map_update_elem(bpf_map__fd(sk_stg_map), &fd,
> > +					  &expected_stg, BPF_NOEXIST);
> > +		if (CHECK(err, "bpf_map_update_elem(sk_stg_map)",
> > +			  "err:%d errno:%d\n", err, errno))
> > +			goto done;
> > +	}
> > +
> >   	if (settcpca(lfd, tcp_ca) || settcpca(fd, tcp_ca) ||
> >   	    settimeo(lfd) || settimeo(fd))
> >   		goto done;
> > @@ -149,6 +158,16 @@ static void do_test(const char *tcp_ca)
> >   	CHECK(bytes != total_bytes, "recv", "%zd != %u nr_recv:%zd errno:%d\n",
> >   	      bytes, total_bytes, nr_recv, errno);
> 
> Should the control go to "wait_thread" here if failure?
Thanks for the review!

I did think about that.  I did not bail on this because the
sk_stg_map check below does not depend on the about check.
Hence, I didn't bail here.  I moved the sk_stg_map test
to the very bottom is for this reason also.

Since you asked, I think it makes sense to go back to my first
approach which is to do the below test immediately after
connect() and just bail there.

I will also take this chance to postpone the thread creation
after connect().

> 
> > +	if (sk_stg_map) {
> > +		int tmp_stg;
> > +
> > +		err = bpf_map_lookup_elem(bpf_map__fd(sk_stg_map), &fd,
> > +					  &tmp_stg);
> > +		CHECK(!err || errno != ENOENT,
> > +		      "bpf_map_lookup_elem(sk_stg_map)",
> > +		      "err:%d errno:%d\n", err, errno);
> > +	}
> > +
> >   wait_thread:
> >   	WRITE_ONCE(stop, 1);
> >   	pthread_join(srv_thread, &thread_ret);
> > @@ -175,7 +194,7 @@ static void test_cubic(void)
> >   		return;
> >   	}
> > -	do_test("bpf_cubic");
> > +	do_test("bpf_cubic", NULL);
> >   	bpf_link__destroy(link);
> >   	bpf_cubic__destroy(cubic_skel);
> > @@ -197,7 +216,10 @@ static void test_dctcp(void)
> >   		return;
> >   	}
> > -	do_test("bpf_dctcp");
> > +	do_test("bpf_dctcp", dctcp_skel->maps.sk_stg_map);
> > +	CHECK(dctcp_skel->bss->stg_result != expected_stg,
> > +	      "Unexpected stg_result", "stg_result (%x) != expected_stg (%x)\n",
> > +	      dctcp_skel->bss->stg_result, expected_stg);
> >   	bpf_link__destroy(link);
> >   	bpf_dctcp__destroy(dctcp_skel);
> > diff --git a/tools/testing/selftests/bpf/progs/bpf_dctcp.c b/tools/testing/selftests/bpf/progs/bpf_dctcp.c
> > index 127ea762a062..5c1fc584f3ae 100644
> > --- a/tools/testing/selftests/bpf/progs/bpf_dctcp.c
> > +++ b/tools/testing/selftests/bpf/progs/bpf_dctcp.c
> > @@ -6,6 +6,7 @@
> >    * the kernel BPF logic.
> >    */
> > +#include <stddef.h>
> >   #include <linux/bpf.h>
> >   #include <linux/types.h>
> >   #include <bpf/bpf_helpers.h>
> > @@ -14,6 +15,15 @@
> >   char _license[] SEC("license") = "GPL";
> > +static volatile int stg_result;
> 
> "int stg_result = 0;" should work too.
will use.