| 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
| ||
|
Date: Thu, 19 Aug 2021 11:52:20 +0200
From: Magnus Karlsson <magnus.karlsson@...il.com>
To: Maciej Fijalkowski <maciej.fijalkowski@...el.com>
Cc: "Karlsson, Magnus" <magnus.karlsson@...el.com>,
Björn Töpel <bjorn@...nel.org>,
Alexei Starovoitov <ast@...nel.org>,
Daniel Borkmann <daniel@...earbox.net>,
Network Development <netdev@...r.kernel.org>,
Jonathan Lemon <jonathan.lemon@...il.com>,
Ciara Loftus <ciara.loftus@...el.com>,
bpf <bpf@...r.kernel.org>, Yonghong Song <yhs@...com>,
Andrii Nakryiko <andrii@...nel.org>
Subject: Re: [PATCH bpf-next v2 10/16] selftests: xsk: validate tx stats on tx thread
On Thu, Aug 19, 2021 at 11:41 AM Maciej Fijalkowski
<maciej.fijalkowski@...el.com> wrote:
>
> On Tue, Aug 17, 2021 at 11:27:23AM +0200, Magnus Karlsson wrote:
> > From: Magnus Karlsson <magnus.karlsson@...el.com>
> >
> > Validate the tx stats on the Tx thread instead of the Rx
> > tread. Depending on your settings, you might not be allowed to query
> > the statistics of a socket you do not own, so better to do this on the
> > correct thread to start with.
> >
> > Signed-off-by: Magnus Karlsson <magnus.karlsson@...el.com>
> > ---
> > tools/testing/selftests/bpf/xdpxceiver.c | 55 ++++++++++++++++++------
> > 1 file changed, 41 insertions(+), 14 deletions(-)
> >
> > diff --git a/tools/testing/selftests/bpf/xdpxceiver.c b/tools/testing/selftests/bpf/xdpxceiver.c
> > index fe3d281a0575..8ff24472ef1e 100644
> > --- a/tools/testing/selftests/bpf/xdpxceiver.c
> > +++ b/tools/testing/selftests/bpf/xdpxceiver.c
> > @@ -642,23 +642,22 @@ static void tx_only_all(struct ifobject *ifobject)
> > complete_tx_only_all(ifobject);
> > }
> >
> > -static void stats_validate(struct ifobject *ifobject)
> > +static bool rx_stats_are_valid(struct ifobject *ifobject)
> > {
> > + u32 xsk_stat = 0, expected_stat = opt_pkt_count;
> > + struct xsk_socket *xsk = ifobject->xsk->xsk;
> > + int fd = xsk_socket__fd(xsk);
> > struct xdp_statistics stats;
> > socklen_t optlen;
> > int err;
> > - struct xsk_socket *xsk = stat_test_type == STAT_TEST_TX_INVALID ?
> > - ifdict[!ifobject->ifdict_index]->xsk->xsk :
> > - ifobject->xsk->xsk;
> > - int fd = xsk_socket__fd(xsk);
> > - unsigned long xsk_stat = 0, expected_stat = opt_pkt_count;
> > -
> > - sigvar = 0;
> >
> > optlen = sizeof(stats);
> > err = getsockopt(fd, SOL_XDP, XDP_STATISTICS, &stats, &optlen);
> > - if (err)
> > - return;
> > + if (err) {
> > + ksft_test_result_fail("ERROR: [%s] getsockopt(XDP_STATISTICS) error %u %s\n",
> > + __func__, -err, strerror(-err));
> > + return true;
>
> Can we invert the logic or change the name of the func?
> Returning 'true' for error case is a bit confusing given the name of func
> is blah_are_valid, no? If there was an error then I'd return false.
>
> OTOH we're testing faulty socket situations in here, but error from
> getsockopt does not mean that stats were valid.
Yes, this is not that clear. We want the loop on the higher level to
quit, therefore we return true when there is an error. A problem with
the stats tests is that they do not terminate when the stats are
wrong, only when they pass. Once I get the second patch set accepted
(in some form), we should rewrite these stats tests in that new
framework so that they will terminate even when the stats are wrong.
This whole problem will likely disappear at that point. But I will
scratch my head and try to make it better in this patch.
> > + }
> >
> > if (optlen == sizeof(struct xdp_statistics)) {
> > switch (stat_test_type) {
> > @@ -666,8 +665,7 @@ static void stats_validate(struct ifobject *ifobject)
> > xsk_stat = stats.rx_dropped;
> > break;
> > case STAT_TEST_TX_INVALID:
> > - xsk_stat = stats.tx_invalid_descs;
> > - break;
> > + return true;
> > case STAT_TEST_RX_FULL:
> > xsk_stat = stats.rx_ring_full;
> > expected_stat -= RX_FULL_RXQSIZE;
> > @@ -680,8 +678,33 @@ static void stats_validate(struct ifobject *ifobject)
> > }
> >
> > if (xsk_stat == expected_stat)
> > - sigvar = 1;
> > + return true;
> > + }
> > +
> > + return false;
> > +}
> > +
> > +static void tx_stats_validate(struct ifobject *ifobject)
> > +{
> > + struct xsk_socket *xsk = ifobject->xsk->xsk;
> > + int fd = xsk_socket__fd(xsk);
> > + struct xdp_statistics stats;
> > + socklen_t optlen;
> > + int err;
> > +
> > + optlen = sizeof(stats);
> > + err = getsockopt(fd, SOL_XDP, XDP_STATISTICS, &stats, &optlen);
> > + if (err) {
> > + ksft_test_result_fail("ERROR: [%s] getsockopt(XDP_STATISTICS) error %u %s\n",
> > + __func__, -err, strerror(-err));
> > + return;
> > }
> > +
> > + if (stats.tx_invalid_descs == opt_pkt_count)
> > + return;
> > +
> > + ksft_test_result_fail("ERROR: [%s] tx_invalid_descs incorrect. Got [%u] expected [%u]\n",
> > + __func__, stats.tx_invalid_descs, opt_pkt_count);
> > }
> >
> > static void thread_common_ops(struct ifobject *ifobject, void *bufs)
> > @@ -767,6 +790,9 @@ static void *worker_testapp_validate_tx(void *arg)
> > print_verbose("Sending %d packets on interface %s\n", opt_pkt_count, ifobject->ifname);
> > tx_only_all(ifobject);
> >
> > + if (stat_test_type == STAT_TEST_TX_INVALID)
> > + tx_stats_validate(ifobject);
> > +
> > testapp_cleanup_xsk_res(ifobject);
> > pthread_exit(NULL);
> > }
> > @@ -792,7 +818,8 @@ static void *worker_testapp_validate_rx(void *arg)
> > if (test_type != TEST_TYPE_STATS) {
> > rx_pkt(ifobject->xsk, fds);
> > } else {
> > - stats_validate(ifobject);
> > + if (rx_stats_are_valid(ifobject))
> > + break;
> > }
> > if (sigvar)
> > break;
> > --
> > 2.29.0
> >
Powered by blists - more mailing lists