The GRO self-test, gro.c, currently constructs IPv6 packets containing a
Hop-by-Hop Options header (IPPROTO_HOPOPTS) to ensure the GRO path
correctly handles IPv6 extension headers.
However, network elements may be configured to drop packets with the
Hop-by-Hop Options header (HBH). This causes the self-test to fail
in environments where such network elements are present.
To improve the robustness and reliability of this test in diverse
network environments, switch from using IPPROTO_HOPOPTS to
IPPROTO_DSTOPTS (Destination Options).
The Destination Options header is less likely to be dropped by
intermediate routers and still serves the core purpose of the test:
validating GRO's handling of an IPv6 extension header. This change
ensures the test can execute successfully without being incorrectly
failed by network policies outside the kernel's control.
Fixes: 7d1575014a63 ("selftests/net: GRO coalesce test")
Reviewed-by: Willem de Bruijn <willemb(a)google.com>
Signed-off-by: Anubhav Singh <anubhavsinggh(a)google.com>
---
tools/testing/selftests/net/gro.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/tools/testing/selftests/net/gro.c b/tools/testing/selftests/net/gro.c
index 2b1d9f2b3e9e..d8c29fe39c1d 100644
--- a/tools/testing/selftests/net/gro.c
+++ b/tools/testing/selftests/net/gro.c
@@ -754,11 +754,11 @@ static void send_ipv6_exthdr(int fd, struct sockaddr_ll *daddr, char *ext_data1,
static char exthdr_pck[sizeof(buf) + MIN_EXTHDR_SIZE];
create_packet(buf, 0, 0, PAYLOAD_LEN, 0);
- add_ipv6_exthdr(buf, exthdr_pck, IPPROTO_HOPOPTS, ext_data1);
+ add_ipv6_exthdr(buf, exthdr_pck, IPPROTO_DSTOPTS, ext_data1);
write_packet(fd, exthdr_pck, total_hdr_len + PAYLOAD_LEN + MIN_EXTHDR_SIZE, daddr);
create_packet(buf, PAYLOAD_LEN * 1, 0, PAYLOAD_LEN, 0);
- add_ipv6_exthdr(buf, exthdr_pck, IPPROTO_HOPOPTS, ext_data2);
+ add_ipv6_exthdr(buf, exthdr_pck, IPPROTO_DSTOPTS, ext_data2);
write_packet(fd, exthdr_pck, total_hdr_len + PAYLOAD_LEN + MIN_EXTHDR_SIZE, daddr);
}
--
2.51.1.851.g4ebd6896fd-goog
Due to the gro_sender sending data packets and FIN packets
in very quick succession, these are received almost simultaneously
by the gro_receiver. FIN packets are sometimes processed before the
data packets leading to intermittent (~1/100) test failures.
This change adds a delay of 100ms before sending FIN packets
in gro:tcp test to avoid the out-of-order delivery. The same
mitigation already exists for the gro:ip test.
Fixes: 7d1575014a63 ("selftests/net: GRO coalesce test")
Reviewed-by: Willem de Bruijn <willemb(a)google.com>
Signed-off-by: Anubhav Singh <anubhavsinggh(a)google.com>
---
tools/testing/selftests/net/gro.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/tools/testing/selftests/net/gro.c b/tools/testing/selftests/net/gro.c
index 2b1d9f2b3e9e..3fa63bd85dea 100644
--- a/tools/testing/selftests/net/gro.c
+++ b/tools/testing/selftests/net/gro.c
@@ -989,6 +989,7 @@ static void check_recv_pkts(int fd, int *correct_payload,
static void gro_sender(void)
{
+ const int fin_delay_us = 100 * 1000;
static char fin_pkt[MAX_HDR_LEN];
struct sockaddr_ll daddr = {};
int txfd = -1;
@@ -1032,15 +1033,22 @@ static void gro_sender(void)
write_packet(txfd, fin_pkt, total_hdr_len, &daddr);
} else if (strcmp(testname, "tcp") == 0) {
send_changed_checksum(txfd, &daddr);
+ /* Adding sleep before sending FIN so that it is not
+ * received prior to other packets.
+ */
+ usleep(fin_delay_us);
write_packet(txfd, fin_pkt, total_hdr_len, &daddr);
send_changed_seq(txfd, &daddr);
+ usleep(fin_delay_us);
write_packet(txfd, fin_pkt, total_hdr_len, &daddr);
send_changed_ts(txfd, &daddr);
+ usleep(fin_delay_us);
write_packet(txfd, fin_pkt, total_hdr_len, &daddr);
send_diff_opt(txfd, &daddr);
+ usleep(fin_delay_us);
write_packet(txfd, fin_pkt, total_hdr_len, &daddr);
} else if (strcmp(testname, "ip") == 0) {
send_changed_ECN(txfd, &daddr);
--
2.51.1.851.g4ebd6896fd-goog