On Sat, Jun 20, 2026 at 04:09:21PM -0400, Jon Maloy wrote:

Since commit b4dace8 ("fwd: Direct inbound spliced forwards to the guest's external address"), inbound splice connections target the guest's external address (addr_seen) rather than loopback. This requires the namespace tap interface to be configured with that address.

setup_pasta() starts pasta without --config-net, so the tap interface in the namespace is never brought up and has no addresses or routes. Splice connections to addr_seen fail because the address is unreachable in the namespace, breaking all pasta TCP and UDP tests.

Add --config-net to the pasta command line in setup_pasta(), matching what setup_passt_in_ns() and setup_two_guests() already do.

Signed-off-by: Jon Maloy

No. We use setup_pasta() to test NDP and DHCP behaviour with pasta, so we need it not to preconfigure the network.

--- test/lib/setup | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/test/lib/setup b/test/lib/setup index 59945987..dbf8079d 100755 --- a/test/lib/setup +++ b/test/lib/setup @@ -107,7 +107,7 @@ setup_pasta() { [ ${DEBUG} -eq 1 ] && __opts="${__opts} -d" [ ${TRACE} -eq 1 ] && __opts="${__opts} --trace"

- context_run_bg passt "./pasta ${__opts} -f -t 10002 -T 10003 -u 10002 -U 10003 -P ${STATESETUP}/passt.pid $(${NSTOOL} info -pw ${STATESETUP}/ns.hold)" + context_run_bg passt "./pasta ${__opts} -f -t 10002 -T 10003 -u 10002 -U 10003 --config-net -P ${STATESETUP}/passt.pid $(${NSTOOL} info -pw ${STATESETUP}/ns.hold)"

# pidfile isn't created until pasta is ready wait_for [ -f "${STATESETUP}/passt.pid" ] -- 2.52.0

-- David Gibson (he or they) | I'll have my music baroque, and my code david AT gibson.dropbear.id.au | minimalist, thank you, not the other way | around. http://www.ozlabs.org/~dgibson