selftests/bpf: add lwt ip encap tests to test_progs
Subsystems (3)
PATCH v2 1/1 selftests/bpf: add lwt ip encap tests to test_progs
Port test_lwt_ip_encap.sh tests onto test_progs.
In addition, this commit adds "egress_md" tests which test a similar flow as egress tests only they use gre devices in collect_md mode for encapsulation and set the tunnel key using bpf_set_tunnel_key().
This introduces minor changes to test_lwt_ip_encap.{sh,c} for consistency with the new tests:
- GRE key must exist as bpf_set_tunnel_key() explicitly sets the TUNNEL_KEY flag
- Source address for GRE traffic is set to IP*_5 instead of IP*_1 since GRE traffic is sent via veth5 so its address is selected when using bpf_set_tunnel_key()
Note: currently these programs use the legacy section name convention as iproute2 lwt configuration does not support providing function names.
----
v2:
- set CONFIG_NET_VRF and CONFIG_NET_L3_MASTER_DEV in reference .config
- add wait after sending large packet to allow it to get delivered as in the .sh test
--- tools/testing/selftests/bpf/config | 2 + .../selftests/bpf/prog_tests/lwt_ip_encap.c | 585 ++++++++++++++++++ .../selftests/bpf/progs/test_lwt_ip_encap.c | 51 +- .../selftests/bpf/test_lwt_ip_encap.sh | 6 +- 4 files changed, 638 insertions(+), 6 deletions(-) create mode 100644 tools/testing/selftests/bpf/prog_tests/lwt_ip_encap.c
tools/testing/selftests/bpf/config +2 -0
@@ -64,3 +64,5 @@ CONFIG_TEST_BPF=m CONFIG_USERFAULTFD=y CONFIG_VXLAN=y CONFIG_XDP_SOCKETS=y +CONFIG_NET_VRF=y +CONFIG_NET_L3_MASTER_DEV=y
tools/testing/selftests/bpf/prog_tests/lwt_ip_encap.c new file +585 -0
585 lines not on this page · raw message
tools/testing/selftests/bpf/progs/test_lwt_ip_encap.c +47 -4
90 lines not on this page · raw message
tools/testing/selftests/bpf/test_lwt_ip_encap.sh +4 -2
@@ -238,7 +238,8 @@ setup() ip -netns ${NS3} -6 route add ${IPv6_6}/128 dev veth8 via ${IPv6_7} # configure IPv4 GRE device in NS3, and a route to it via the "bottom" route - ip -netns ${NS3} tunnel add gre_dev mode gre remote ${IPv4_1} local ${IPv4_GRE} ttl 255 + ip -netns ${NS3} tunnel add gre_dev mode gre remote ${IPv4_5} \ + local ${IPv4_GRE} ttl 255 key 0 ip -netns ${NS3} link set gre_dev up ip -netns ${NS3} addr add ${IPv4_GRE} dev gre_dev ip -netns ${NS1} route add ${IPv4_GRE}/32 dev veth5 via ${IPv4_6} ${VRF}
@@ -246,7 +247,8 @@ setup() # configure IPv6 GRE device in NS3, and a route to it via the "bottom" route - ip -netns ${NS3} -6 tunnel add name gre6_dev mode ip6gre remote ${IPv6_1} local ${IPv6_GRE} ttl 255 + ip -netns ${NS3} -6 tunnel add name gre6_dev mode ip6gre remote ${IPv6_5} \ + local ${IPv6_GRE} ttl 255 key 0 ip -netns ${NS3} link set gre6_dev up ip -netns ${NS3} -6 addr add ${IPv6_GRE} nodad dev gre6_dev ip -netns ${NS1} -6 route add ${IPv6_GRE}/128 dev veth5 via ${IPv6_6} ${VRF}
Signature · 1 line
2.34.1
Daniel Borkmann
maintainerQuoted 12 lines · Eyal Birger
Port
test_lwt_ip_encap.shtests ontotest_progs.In addition, this commit adds "
egress_md" tests which test a similar flow as egress tests only they use gre devices incollect_mdmode for encapsulation and set the tunnel key usingbpf_set_tunnel_key().This introduces minor changes to test_lwt_ip_encap.{sh,c} for consistency with the new tests:
- GRE key must exist as
bpf_set_tunnel_key()explicitly sets theTUNNEL_KEYflag
Quoted 8 lines
- Source address for GRE traffic is set to IP*_5 instead of IP*_1 since GRE traffic is sent via veth5 so its address is selected when using
bpf_set_tunnel_key()Note: currently these programs use the legacy section name convention as iproute2 lwt configuration does not support providing function names.
Signed-off-by Eyal Birger <[email protected]>
[...]
Thanks for following up. Is there now anything that test_lwt_ip_encap.c doesn't cover over test_lwt_ip_encap.sh? If not, I'd vote for removing the latter given the port is then covered in CI via test_progs.
Quoted 25 lines
tools/testing/selftests/bpf/test_lwt_ip_encap.sh +4
-2@@ -238,7 +238,8 @@ setup() ip -netns ${NS3} -6 route add ${IPv6_6}/128 dev veth8 via ${IPv6_7} # configure IPv4 GRE device in NS3, and a route to it via the "bottom" route - ip -netns ${NS3} tunnel add gre_dev mode gre remote ${IPv4_1} local ${IPv4_GRE} ttl 255 + ip -netns ${NS3} tunnel add gre_dev mode gre remote ${IPv4_5} \ + local ${IPv4_GRE} ttl 255 key 0 ip -netns ${NS3} link set gre_dev up ip -netns ${NS3} addr add ${IPv4_GRE} dev gre_dev ip -netns ${NS1} route add ${IPv4_GRE}/32 dev veth5 via ${IPv4_6} ${VRF}@@ -246,7 +247,8 @@ setup() # configure IPv6 GRE device in NS3, and a route to it via the "bottom" route - ip -netns ${NS3} -6 tunnel add name gre6_dev mode ip6gre remote ${IPv6_1} local ${IPv6_GRE} ttl 255 + ip -netns ${NS3} -6 tunnel add name gre6_dev mode ip6gre remote ${IPv6_5} \ + local ${IPv6_GRE} ttl 255 key 0 ip -netns ${NS3} link set gre6_dev up ip -netns ${NS3} -6 addr add ${IPv6_GRE} nodad dev gre6_dev ip -netns ${NS1} -6 route add ${IPv6_GRE}/128 dev veth5 via ${IPv6_6} ${VRF}
Hi Daniel,
Quoted 21 lines · Daniel Borkmann
Quoted 19 lines · Eyal Birger
Port
test_lwt_ip_encap.shtests ontotest_progs.In addition, this commit adds "
egress_md" tests which test a similar flow as egress tests only they use gre devices incollect_mdmode for encapsulation and set the tunnel key usingbpf_set_tunnel_key().This introduces minor changes to test_lwt_ip_encap.{sh,c} for consistency with the new tests:
- GRE key must exist as
bpf_set_tunnel_key()explicitly sets theTUNNEL_KEYflag- Source address for GRE traffic is set to IP*_5 instead of IP*_1 since GRE traffic is sent via veth5 so its address is selected when using
bpf_set_tunnel_key()Note: currently these programs use the legacy section name convention as iproute2 lwt configuration does not support providing function names.
Quoted 6 lines
Quoted 1 line
Signed-off-by Eyal Birger <[email protected]>[...]
Thanks for following up. Is there now anything that
test_lwt_ip_encap.cdoesn't cover overtest_lwt_ip_encap.sh? If not, I'd vote for removing the latter given the port is then covered in CI viatest_progs.
The .c version includes all tests in the .sh and two more. At least when I tried using vmtest the .sh version seems broken.
I can resubmit with an additional patch removing it, or send a follow up patch as you prefer.
Thanks,
Eyal.
Quoted 27 lines
Quoted 26 lines
tools/testing/selftests/bpf/test_lwt_ip_encap.sh +4
-2@@ -238,7 +238,8 @@ setup() ip -netns ${NS3} -6 route add ${IPv6_6}/128 dev veth8 via ${IPv6_7} # configure IPv4 GRE device in NS3, and a route to it via the "bottom" route - ip -netns ${NS3} tunnel add gre_dev mode gre remote ${IPv4_1} local ${IPv4_GRE} ttl 255 + ip -netns ${NS3} tunnel add gre_dev mode gre remote ${IPv4_5} \ + local ${IPv4_GRE} ttl 255 key 0 ip -netns ${NS3} link set gre_dev up ip -netns ${NS3} addr add ${IPv4_GRE} dev gre_dev ip -netns ${NS1} route add ${IPv4_GRE}/32 dev veth5 via ${IPv4_6} ${VRF}@@ -246,7 +247,8 @@ setup() # configure IPv6 GRE device in NS3, and a route to it via the "bottom" route - ip -netns ${NS3} -6 tunnel add name gre6_dev mode ip6gre remote ${IPv6_1} local ${IPv6_GRE} ttl 255 + ip -netns ${NS3} -6 tunnel add name gre6_dev mode ip6gre remote ${IPv6_5} \ + local ${IPv6_GRE} ttl 255 key 0 ip -netns ${NS3} link set gre6_dev up ip -netns ${NS3} -6 addr add ${IPv6_GRE} nodad dev gre6_dev ip -netns ${NS1} -6 route add ${IPv6_GRE}/128 dev veth5 via ${IPv6_6} ${VRF}
Daniel Borkmann
maintainerQuoted 31 lines · Eyal Birger
Hi Daniel,
Quoted 21 lines · Daniel Borkmann
Quoted 19 lines · Eyal Birger
Port
test_lwt_ip_encap.shtests ontotest_progs.In addition, this commit adds "
egress_md" tests which test a similar flow as egress tests only they use gre devices incollect_mdmode for encapsulation and set the tunnel key usingbpf_set_tunnel_key().This introduces minor changes to test_lwt_ip_encap.{sh,c} for consistency with the new tests:
- GRE key must exist as
bpf_set_tunnel_key()explicitly sets theTUNNEL_KEYflag- Source address for GRE traffic is set to IP*_5 instead of IP*_1 since GRE traffic is sent via veth5 so its address is selected when using
bpf_set_tunnel_key()Note: currently these programs use the legacy section name convention as iproute2 lwt configuration does not support providing function names.
Quoted 6 lines
Quoted 1 line
Signed-off-by Eyal Birger <[email protected]>[...]
Thanks for following up. Is there now anything that
test_lwt_ip_encap.cdoesn't cover overtest_lwt_ip_encap.sh? If not, I'd vote for removing the latter given the port is then covered in CI viatest_progs.
Quoted 5 lines
The .c version includes all tests in the .sh and two more. At least when I tried using vmtest the .sh version seems broken.
I can resubmit with an additional patch removing it, or send a follow up patch as you prefer.
+1, could you squash the removal into the single patch and submit as a v3?
Thanks a lot,
Daniel