[PATCH net-next v2 0/4] Extend netkit io_uring ZC selftests

[PATCH net-next v2 0/4] Extend netkit io_uring ZC selftests

[Daniel Borkmann]

Small follow-up to the HW net selftests, in particular to add a
selftest showing that also large rx_buf_len for io_uring ZC is
supported with netkit queue leasing.

v1 -> v2:
  - Switch to cmd(..., ns=cfg.netns) (Bobby)
  - Add mp_clear_wait for large rx_buf_len test (Bobby)

Daniel Borkmann (4):
  selftests/net: Move netkit lease hw setup into per-test fixtures
  selftests/net: Use public NetDrvContEnv API in nk_qlease fixtures
  selftests/net: Add netkit io_uring ZC test for large rx_buf_len
  selftests/net: Add hugepage kernel config dependency for zcrx

 tools/testing/selftests/drivers/net/hw/config |   1 +
 .../selftests/drivers/net/hw/nk_qlease.py     | 304 +++++++++++++++---
 .../selftests/drivers/net/lib/py/env.py       |  36 ++-
 3 files changed, 274 insertions(+), 67 deletions(-)

-- 
2.43.0

[PATCH net-next v2 1/4] selftests/net: Move netkit lease hw setup into per-test fixtures

[Daniel Borkmann]

The HW counterpart of nk_qlease.py was carrying its lease setup in main()
and stashing src_queue / nk_queue / nk_*_ifname on cfg, which had drawbacks
called out during the review at [0].

This is the deferred half of the cleanup that landed in commit e254ffb9502c
("selftests/net: Split netdevsim tests from HW tests in nk_qlease") which
was the SW counterpart of nk_qlease.py.

While at it, convert the open-coded "ip netns exec" prefixes in the test
bodies over to the ns= argument of cmd() / bkg().

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/netdev/20260408162238.16709090@kernel.org/ [0]
Reviewed-by: Bobby Eshleman <bobbyeshleman@meta.com>
---
 .../selftests/drivers/net/hw/nk_qlease.py     | 198 +++++++++++++-----
 1 file changed, 149 insertions(+), 49 deletions(-)

diff --git a/tools/testing/selftests/drivers/net/hw/nk_qlease.py b/tools/testing/selftests/drivers/net/hw/nk_qlease.py
index 139a91ebd229..c7cb039b8181 100755
--- a/tools/testing/selftests/drivers/net/hw/nk_qlease.py
+++ b/tools/testing/selftests/drivers/net/hw/nk_qlease.py
@@ -18,8 +18,10 @@ from lib.py import (
     NetNSEnter,
     EthtoolFamily,
     NetdevFamily,
+    RtnlFamily,
 )
 from lib.py import (
+    Netlink,
     bkg,
     cmd,
     defer,
@@ -31,9 +33,117 @@ from lib.py import (
 from lib.py import KsftSkipEx, CmdExitFailure
 
 
-def set_flow_rule(cfg):
+def _create_netkit_pair(cfg, rxqueues=2):
+    if cfg._nk_host_ifname:
+        cmd(f"ip link del dev {cfg._nk_host_ifname}", fail=False)
+        cfg._nk_host_ifname = None
+        cfg.nk_guest_ifname = None
+    if getattr(cfg, "_tc_attached", False):
+        cmd(
+            f"tc filter del dev {cfg.ifname} ingress pref {cfg._bpf_prog_pref}",
+            fail=False,
+        )
+        cfg._tc_attached = False
+
+    all_links = ip("-d link show", json=True)
+    old_idxs = {
+        link["ifindex"]
+        for link in all_links
+        if link.get("linkinfo", {}).get("info_kind") == "netkit"
+    }
+
+    rtnl = RtnlFamily()
+    rtnl.newlink(
+        {
+            "linkinfo": {
+                "kind": "netkit",
+                "data": {
+                    "mode": "l2",
+                    "policy": "forward",
+                    "peer-policy": "forward",
+                },
+            },
+            "num-rx-queues": rxqueues,
+        },
+        flags=[Netlink.NLM_F_CREATE, Netlink.NLM_F_EXCL],
+    )
+
+    all_links = ip("-d link show", json=True)
+    nk_links = [
+        link
+        for link in all_links
+        if link.get("linkinfo", {}).get("info_kind") == "netkit"
+        and link["ifindex"] not in old_idxs
+    ]
+    if len(nk_links) != 2:
+        raise KsftSkipEx("Failed to create netkit pair")
+
+    nk_links.sort(key=lambda x: x["ifindex"])
+    cfg._nk_host_ifname = nk_links[1]["ifname"]
+    cfg.nk_guest_ifname = nk_links[0]["ifname"]
+    cfg.nk_host_ifindex = nk_links[1]["ifindex"]
+    cfg.nk_guest_ifindex = nk_links[0]["ifindex"]
+
+    ip(f"link set dev {cfg.nk_guest_ifname} netns {cfg.netns.name}")
+    ip(f"link set dev {cfg._nk_host_ifname} up")
+    ip(f"-6 addr add fe80::1/64 dev {cfg._nk_host_ifname} nodad")
+    ip(
+        f"-6 route add {cfg.nk_guest_ipv6}/128 via fe80::2 "
+        f"dev {cfg._nk_host_ifname}"
+    )
+    ip(f"link set dev {cfg.nk_guest_ifname} up", ns=cfg.netns)
+    ip(f"-6 addr add fe80::2/64 dev {cfg.nk_guest_ifname}", ns=cfg.netns)
+    ip(
+        f"-6 addr add {cfg.nk_guest_ipv6}/64 dev {cfg.nk_guest_ifname} nodad",
+        ns=cfg.netns,
+    )
+    ip(
+        f"-6 route add default via fe80::1 dev {cfg.nk_guest_ifname}",
+        ns=cfg.netns,
+    )
+
+    cfg._attach_bpf()
+
+
+def _setup_lease(cfg, rxqueues=2):
+    _create_netkit_pair(cfg, rxqueues=rxqueues)
+
+    ethnl = EthtoolFamily()
+    channels = ethnl.channels_get({"header": {"dev-index": cfg.ifindex}})[
+        "combined-count"
+    ]
+    if channels < 2:
+        raise KsftSkipEx(
+            "Test requires NETIF with at least 2 combined channels"
+        )
+    src_queue = channels - 1
+
+    with NetNSEnter(str(cfg.netns)):
+        netdevnl = NetdevFamily()
+        bind_result = netdevnl.queue_create(
+            {
+                "ifindex": cfg.nk_guest_ifindex,
+                "type": "rx",
+                "lease": {
+                    "ifindex": cfg.ifindex,
+                    "queue": {"id": src_queue, "type": "rx"},
+                    "netns-id": 0,
+                },
+            }
+        )
+    return src_queue, bind_result["id"]
+
+
+def _teardown_netkit(cfg):
+    if cfg._nk_host_ifname:
+        cmd(f"ip link del dev {cfg._nk_host_ifname}", fail=False)
+        cfg._nk_host_ifname = None
+        cfg.nk_guest_ifname = None
+
+
+def set_flow_rule(cfg, src_queue):
     output = ethtool(
-        f"-N {cfg.ifname} flow-type tcp6 dst-port {cfg.port} action {cfg.src_queue}"
+        f"-N {cfg.ifname} flow-type tcp6 dst-port {cfg.port} action {src_queue}"
     ).stdout
     values = re.search(r"ID (\d+)", output).group(1)
     return int(values)
@@ -41,6 +151,8 @@ def set_flow_rule(cfg):
 
 def test_iou_zcrx(cfg) -> None:
     cfg.require_ipver("6")
+    src_queue, nk_queue = _setup_lease(cfg)
+    defer(_teardown_netkit, cfg)
     ethnl = EthtoolFamily()
 
     rings = ethnl.rings_get({"header": {"dev-index": cfg.ifindex}})
@@ -65,40 +177,47 @@ def test_iou_zcrx(cfg) -> None:
         },
     )
 
-    ethtool(f"-X {cfg.ifname} equal {cfg.src_queue}")
+    ethtool(f"-X {cfg.ifname} equal {src_queue}")
     defer(ethtool, f"-X {cfg.ifname} default")
 
-    flow_rule_id = set_flow_rule(cfg)
+    flow_rule_id = set_flow_rule(cfg, src_queue)
     defer(ethtool, f"-N {cfg.ifname} delete {flow_rule_id}")
 
-    rx_cmd = f"ip netns exec {cfg.netns.name} {cfg.bin_local} -s -p {cfg.port} -i {cfg.nk_guest_ifname} -q {cfg.nk_queue}"
+    rx_cmd = (
+        f"{cfg.bin_local} -s -p {cfg.port} "
+        f"-i {cfg.nk_guest_ifname} -q {nk_queue}"
+    )
     tx_cmd = f"{cfg.bin_remote} -c -h {cfg.nk_guest_ipv6} -p {cfg.port} -l 12840"
-    with bkg(rx_cmd, exit_wait=True):
+    with bkg(rx_cmd, exit_wait=True, ns=cfg.netns):
         wait_port_listen(cfg.port, proto="tcp", ns=cfg.netns)
         cmd(tx_cmd, host=cfg.remote)
 
 
 def test_attrs(cfg) -> None:
     cfg.require_ipver("6")
+    src_queue, nk_queue = _setup_lease(cfg)
+    defer(_teardown_netkit, cfg)
     netdevnl = NetdevFamily()
     queue_info = netdevnl.queue_get(
-        {"ifindex": cfg.ifindex, "id": cfg.src_queue, "type": "rx"}
+        {"ifindex": cfg.ifindex, "id": src_queue, "type": "rx"}
     )
 
-    ksft_eq(queue_info["id"], cfg.src_queue)
+    ksft_eq(queue_info["id"], src_queue)
     ksft_eq(queue_info["type"], "rx")
     ksft_eq(queue_info["ifindex"], cfg.ifindex)
 
     ksft_in("lease", queue_info)
     lease = queue_info["lease"]
     ksft_eq(lease["ifindex"], cfg.nk_guest_ifindex)
-    ksft_eq(lease["queue"]["id"], cfg.nk_queue)
+    ksft_eq(lease["queue"]["id"], nk_queue)
     ksft_eq(lease["queue"]["type"], "rx")
     ksft_in("netns-id", lease)
 
 
 def test_attach_xdp_with_mp(cfg) -> None:
     cfg.require_ipver("6")
+    src_queue, nk_queue = _setup_lease(cfg)
+    defer(_teardown_netkit, cfg)
     ethnl = EthtoolFamily()
 
     rings = ethnl.rings_get({"header": {"dev-index": cfg.ifindex}})
@@ -123,18 +242,21 @@ def test_attach_xdp_with_mp(cfg) -> None:
         },
     )
 
-    ethtool(f"-X {cfg.ifname} equal {cfg.src_queue}")
+    ethtool(f"-X {cfg.ifname} equal {src_queue}")
     defer(ethtool, f"-X {cfg.ifname} default")
 
     netdevnl = NetdevFamily()
 
-    rx_cmd = f"ip netns exec {cfg.netns.name} {cfg.bin_local} -s -p {cfg.port} -i {cfg.nk_guest_ifname} -q {cfg.nk_queue}"
-    with bkg(rx_cmd):
+    rx_cmd = (
+        f"{cfg.bin_local} -s -p {cfg.port} "
+        f"-i {cfg.nk_guest_ifname} -q {nk_queue}"
+    )
+    with bkg(rx_cmd, ns=cfg.netns):
         wait_port_listen(cfg.port, proto="tcp", ns=cfg.netns)
 
         time.sleep(0.1)
         queue_info = netdevnl.queue_get(
-            {"ifindex": cfg.ifindex, "id": cfg.src_queue, "type": "rx"}
+            {"ifindex": cfg.ifindex, "id": src_queue, "type": "rx"}
         )
         ksft_in("io-uring", queue_info)
 
@@ -144,13 +266,15 @@ def test_attach_xdp_with_mp(cfg) -> None:
 
     time.sleep(0.1)
     queue_info = netdevnl.queue_get(
-        {"ifindex": cfg.ifindex, "id": cfg.src_queue, "type": "rx"}
+        {"ifindex": cfg.ifindex, "id": src_queue, "type": "rx"}
     )
     ksft_not_in("io-uring", queue_info)
 
 
 def test_destroy(cfg) -> None:
     cfg.require_ipver("6")
+    src_queue, nk_queue = _setup_lease(cfg)
+    defer(_teardown_netkit, cfg)
     ethnl = EthtoolFamily()
 
     rings = ethnl.rings_get({"header": {"dev-index": cfg.ifindex}})
@@ -175,16 +299,19 @@ def test_destroy(cfg) -> None:
         },
     )
 
-    ethtool(f"-X {cfg.ifname} equal {cfg.src_queue}")
+    ethtool(f"-X {cfg.ifname} equal {src_queue}")
     defer(ethtool, f"-X {cfg.ifname} default")
 
-    rx_cmd = f"ip netns exec {cfg.netns.name} {cfg.bin_local} -s -p {cfg.port} -i {cfg.nk_guest_ifname} -q {cfg.nk_queue}"
-    rx_proc = cmd(rx_cmd, background=True)
+    rx_cmd = (
+        f"{cfg.bin_local} -s -p {cfg.port} "
+        f"-i {cfg.nk_guest_ifname} -q {nk_queue}"
+    )
+    rx_proc = cmd(rx_cmd, background=True, ns=cfg.netns)
     wait_port_listen(cfg.port, proto="tcp", ns=cfg.netns)
 
     netdevnl = NetdevFamily()
     queue_info = netdevnl.queue_get(
-        {"ifindex": cfg.ifindex, "id": cfg.src_queue, "type": "rx"}
+        {"ifindex": cfg.ifindex, "id": src_queue, "type": "rx"}
     )
     ksft_in("io-uring", queue_info)
 
@@ -199,17 +326,14 @@ def test_destroy(cfg) -> None:
     cfg.nk_guest_ifname = None
 
     queue_info = netdevnl.queue_get(
-        {"ifindex": cfg.ifindex, "id": cfg.src_queue, "type": "rx"}
+        {"ifindex": cfg.ifindex, "id": src_queue, "type": "rx"}
     )
     ksft_not_in("io-uring", queue_info)
 
-    cmd(f"tc filter del dev {cfg.ifname} ingress pref {cfg._bpf_prog_pref}")
-    cfg._tc_attached = False
-
-    flow_rule_id = set_flow_rule(cfg)
+    flow_rule_id = set_flow_rule(cfg, src_queue)
     defer(ethtool, f"-N {cfg.ifname} delete {flow_rule_id}")
 
-    rx_cmd = f"{cfg.bin_local} -s -p {cfg.port} -i {cfg.ifname} -q {cfg.src_queue}"
+    rx_cmd = f"{cfg.bin_local} -s -p {cfg.port} -i {cfg.ifname} -q {src_queue}"
     tx_cmd = f"{cfg.bin_remote} -c -h {cfg.addr_v['6']} -p {cfg.port} -l 12840"
     with bkg(rx_cmd, exit_wait=True):
         wait_port_listen(cfg.port, proto="tcp")
@@ -217,7 +341,7 @@ def test_destroy(cfg) -> None:
     # Short delay since iou cleanup is async and takes a bit of time.
     time.sleep(0.1)
     queue_info = netdevnl.queue_get(
-        {"ifindex": cfg.ifindex, "id": cfg.src_queue, "type": "rx"}
+        {"ifindex": cfg.ifindex, "id": src_queue, "type": "rx"}
     )
     ksft_not_in("io-uring", queue_info)
 
@@ -230,30 +354,6 @@ def main() -> None:
         cfg.bin_remote = cfg.remote.deploy(cfg.bin_local)
         cfg.port = rand_port()
 
-        ethnl = EthtoolFamily()
-        channels = ethnl.channels_get({"header": {"dev-index": cfg.ifindex}})
-        channels = channels["combined-count"]
-        if channels < 2:
-            raise KsftSkipEx("Test requires NETIF with at least 2 combined channels")
-
-        cfg.src_queue = channels - 1
-
-        with NetNSEnter(str(cfg.netns)):
-            netdevnl = NetdevFamily()
-            bind_result = netdevnl.queue_create(
-                {
-                    "ifindex": cfg.nk_guest_ifindex,
-                    "type": "rx",
-                    "lease": {
-                        "ifindex": cfg.ifindex,
-                        "queue": {"id": cfg.src_queue, "type": "rx"},
-                        "netns-id": 0,
-                    },
-                }
-            )
-            cfg.nk_queue = bind_result["id"]
-
-        # test_destroy must be last because it destroys the netkit devices
         ksft_run(
             [test_iou_zcrx, test_attrs, test_attach_xdp_with_mp, test_destroy],
             args=(cfg,),
-- 
2.43.0

[PATCH net-next v2 2/4] selftests/net: Use public NetDrvContEnv API in nk_qlease fixtures

[Daniel Borkmann]

Expose the netkit host ifname as a public attribute nk_host_ifname
(symmetric with the already-public nk_guest_ifname), rename _attach_bpf
to a public attach_bpf, and add a public detach_bpf helper that
encapsulates the tc-filter teardown bookkeeping. Switch the fixtures
to this public API. No functional change and keeps pylint happy.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Bobby Eshleman <bobbyeshleman@meta.com>
---
 .../selftests/drivers/net/hw/nk_qlease.py     | 33 ++++++++---------
 .../selftests/drivers/net/lib/py/env.py       | 36 +++++++++++--------
 2 files changed, 35 insertions(+), 34 deletions(-)

diff --git a/tools/testing/selftests/drivers/net/hw/nk_qlease.py b/tools/testing/selftests/drivers/net/hw/nk_qlease.py
index c7cb039b8181..b97663820ccf 100755
--- a/tools/testing/selftests/drivers/net/hw/nk_qlease.py
+++ b/tools/testing/selftests/drivers/net/hw/nk_qlease.py
@@ -34,16 +34,11 @@ from lib.py import KsftSkipEx, CmdExitFailure
 
 
 def _create_netkit_pair(cfg, rxqueues=2):
-    if cfg._nk_host_ifname:
-        cmd(f"ip link del dev {cfg._nk_host_ifname}", fail=False)
-        cfg._nk_host_ifname = None
+    if cfg.nk_host_ifname:
+        cmd(f"ip link del dev {cfg.nk_host_ifname}", fail=False)
+        cfg.nk_host_ifname = None
         cfg.nk_guest_ifname = None
-    if getattr(cfg, "_tc_attached", False):
-        cmd(
-            f"tc filter del dev {cfg.ifname} ingress pref {cfg._bpf_prog_pref}",
-            fail=False,
-        )
-        cfg._tc_attached = False
+    cfg.detach_bpf()
 
     all_links = ip("-d link show", json=True)
     old_idxs = {
@@ -79,17 +74,17 @@ def _create_netkit_pair(cfg, rxqueues=2):
         raise KsftSkipEx("Failed to create netkit pair")
 
     nk_links.sort(key=lambda x: x["ifindex"])
-    cfg._nk_host_ifname = nk_links[1]["ifname"]
+    cfg.nk_host_ifname = nk_links[1]["ifname"]
     cfg.nk_guest_ifname = nk_links[0]["ifname"]
     cfg.nk_host_ifindex = nk_links[1]["ifindex"]
     cfg.nk_guest_ifindex = nk_links[0]["ifindex"]
 
     ip(f"link set dev {cfg.nk_guest_ifname} netns {cfg.netns.name}")
-    ip(f"link set dev {cfg._nk_host_ifname} up")
-    ip(f"-6 addr add fe80::1/64 dev {cfg._nk_host_ifname} nodad")
+    ip(f"link set dev {cfg.nk_host_ifname} up")
+    ip(f"-6 addr add fe80::1/64 dev {cfg.nk_host_ifname} nodad")
     ip(
         f"-6 route add {cfg.nk_guest_ipv6}/128 via fe80::2 "
-        f"dev {cfg._nk_host_ifname}"
+        f"dev {cfg.nk_host_ifname}"
     )
     ip(f"link set dev {cfg.nk_guest_ifname} up", ns=cfg.netns)
     ip(f"-6 addr add fe80::2/64 dev {cfg.nk_guest_ifname}", ns=cfg.netns)
@@ -102,7 +97,7 @@ def _create_netkit_pair(cfg, rxqueues=2):
         ns=cfg.netns,
     )
 
-    cfg._attach_bpf()
+    cfg.attach_bpf()
 
 
 def _setup_lease(cfg, rxqueues=2):
@@ -135,9 +130,9 @@ def _setup_lease(cfg, rxqueues=2):
 
 
 def _teardown_netkit(cfg):
-    if cfg._nk_host_ifname:
-        cmd(f"ip link del dev {cfg._nk_host_ifname}", fail=False)
-        cfg._nk_host_ifname = None
+    if cfg.nk_host_ifname:
+        cmd(f"ip link del dev {cfg.nk_host_ifname}", fail=False)
+        cfg.nk_host_ifname = None
         cfg.nk_guest_ifname = None
 
 
@@ -320,9 +315,9 @@ def test_destroy(cfg) -> None:
     kill_timer = threading.Timer(1, rx_proc.proc.terminate)
     kill_timer.start()
 
-    ip(f"link del dev {cfg._nk_host_ifname}")
+    ip(f"link del dev {cfg.nk_host_ifname}")
     kill_timer.join()
-    cfg._nk_host_ifname = None
+    cfg.nk_host_ifname = None
     cfg.nk_guest_ifname = None
 
     queue_info = netdevnl.queue_get(
diff --git a/tools/testing/selftests/drivers/net/lib/py/env.py b/tools/testing/selftests/drivers/net/lib/py/env.py
index 2cc78b8a2152..e1038a02760e 100644
--- a/tools/testing/selftests/drivers/net/lib/py/env.py
+++ b/tools/testing/selftests/drivers/net/lib/py/env.py
@@ -341,7 +341,7 @@ class NetDrvContEnv(NetDrvEpEnv):
                  userns=False, **kwargs):
         self.netns = None
         self._userns = userns
-        self._nk_host_ifname = None
+        self.nk_host_ifname = None
         self.nk_guest_ifname = None
         self._tc_clsact_added = False
         self._tc_attached = False
@@ -395,23 +395,23 @@ class NetDrvContEnv(NetDrvEpEnv):
             raise KsftSkipEx("Failed to create netkit pair")
 
         netkit_links.sort(key=lambda x: x['ifindex'])
-        self._nk_host_ifname = netkit_links[1]['ifname']
+        self.nk_host_ifname = netkit_links[1]['ifname']
         self.nk_guest_ifname = netkit_links[0]['ifname']
         self.nk_host_ifindex = netkit_links[1]['ifindex']
         self.nk_guest_ifindex = netkit_links[0]['ifindex']
 
         self._setup_ns()
-        self._attach_bpf()
+        self.attach_bpf()
         if primary_rx_redirect:
             self._attach_primary_rx_redirect_bpf()
 
     def __del__(self):
         if self._primary_rx_redirect_attached:
-            cmd(f"tc filter del dev {self._nk_host_ifname} ingress", fail=False)
+            cmd(f"tc filter del dev {self.nk_host_ifname} ingress", fail=False)
             self._primary_rx_redirect_attached = False
 
         if self._primary_rx_redirect_clsact_added:
-            cmd(f"tc qdisc del dev {self._nk_host_ifname} clsact", fail=False)
+            cmd(f"tc qdisc del dev {self.nk_host_ifname} clsact", fail=False)
             self._primary_rx_redirect_clsact_added = False
 
         if self._tc_attached:
@@ -427,9 +427,9 @@ class NetDrvContEnv(NetDrvEpEnv):
                 host=self.remote, fail=False)
             self._remote_route_added = False
 
-        if self._nk_host_ifname:
-            cmd(f"ip link del dev {self._nk_host_ifname}")
-            self._nk_host_ifname = None
+        if self.nk_host_ifname:
+            cmd(f"ip link del dev {self.nk_host_ifname}")
+            self.nk_host_ifname = None
             self.nk_guest_ifname = None
 
         if self._init_ns_attached:
@@ -470,9 +470,9 @@ class NetDrvContEnv(NetDrvEpEnv):
         self._init_ns_attached = True
         ip("netns set init 0", ns=self.netns)
         ip(f"link set dev {self.nk_guest_ifname} netns {self.netns.name}")
-        ip(f"link set dev {self._nk_host_ifname} up")
-        ip(f"-6 addr add fe80::1/64 dev {self._nk_host_ifname} nodad")
-        ip(f"-6 route add {self.nk_guest_ipv6}/128 via fe80::2 dev {self._nk_host_ifname}")
+        ip(f"link set dev {self.nk_host_ifname} up")
+        ip(f"-6 addr add fe80::1/64 dev {self.nk_host_ifname} nodad")
+        ip(f"-6 route add {self.nk_guest_ipv6}/128 via fe80::2 dev {self.nk_host_ifname}")
 
         ip("link set lo up", ns=self.netns)
         ip(f"link set dev {self.nk_guest_ifname} up", ns=self.netns)
@@ -512,7 +512,13 @@ class NetDrvContEnv(NetDrvEpEnv):
                 return map_id
         raise Exception(f"Failed to find .bss map for prog {prog_id}")
 
-    def _attach_bpf(self):
+    def detach_bpf(self):
+        if self._tc_attached:
+            cmd(f"tc filter del dev {self.ifname} ingress pref "
+                f"{self._bpf_prog_pref}", fail=False)
+            self._tc_attached = False
+
+    def attach_bpf(self):
         bpf_obj = self.test_dir / "nk_forward.bpf.o"
         if not bpf_obj.exists():
             raise KsftSkipEx("BPF prog not found")
@@ -539,9 +545,9 @@ class NetDrvContEnv(NetDrvEpEnv):
         if not bpf_obj.exists():
             raise KsftSkipEx("Primary RX redirect BPF prog not found")
 
-        if self._tc_ensure_clsact(self._nk_host_ifname):
+        if self._tc_ensure_clsact(self.nk_host_ifname):
             self._primary_rx_redirect_clsact_added = True
-        cmd(f"tc filter add dev {self._nk_host_ifname} ingress"
+        cmd(f"tc filter add dev {self.nk_host_ifname} ingress"
             f" bpf obj {bpf_obj} sec tc/ingress direct-action")
         self._primary_rx_redirect_attached = True
 
@@ -550,7 +556,7 @@ class NetDrvContEnv(NetDrvEpEnv):
         self._remote_route_added = True
 
         filters = json.loads(
-            cmd(f"tc -j filter show dev {self._nk_host_ifname} ingress").stdout)
+            cmd(f"tc -j filter show dev {self.nk_host_ifname} ingress").stdout)
         redirect_prog_id = None
         for bpf in filters:
             if 'options' not in bpf:
-- 
2.43.0

[PATCH net-next v2 3/4] selftests/net: Add netkit io_uring ZC test for large rx_buf_len

[Daniel Borkmann]

Add test_iou_zcrx_large_buf, which runs iou-zcrx with rx_buf_len >
page size (-x 2) through a netkit-leased RX queue. The netkit ifindex
is opaque to io_uring, but rx_page_size is honoured by the leased
physical qops via netif_mp_open_rxq()'s lease redirect.

Originally, I also added a BIG TCP variant on top, but dropped it
here as fbnic (and the QEMU fbnic model) has no BIG TCP support
to exercise it as this point.

Tested against the QEMU fbnic emulation. The new test exercises
the > page rx_buf_len path only when the leased NIC advertises
QCFG_RX_PAGE_SIZE; otherwise it skips.

For fbnic, I used Bjorn's patches locally [0]:

  # ./nk_qlease.py
  TAP version 13
  1..5
  ok 1 nk_qlease.test_iou_zcrx
  ok 2 nk_qlease.test_iou_zcrx_large_buf
  ok 3 nk_qlease.test_attrs
  ok 4 nk_qlease.test_attach_xdp_with_mp
  ok 5 nk_qlease.test_destroy
  # Totals: pass:5 fail:0 xfail:0 xpass:0 skip:0 error:0

Without those patches (aka not advertising QCFG_RX_PAGE_SIZE):

  # ./nk_qlease.py
  TAP version 13
  1..5
  ok 1 nk_qlease.test_iou_zcrx
  ok 2 nk_qlease.test_iou_zcrx_large_buf # SKIP Large chunks are not supported -95
  ok 3 nk_qlease.test_attrs
  ok 4 nk_qlease.test_attach_xdp_with_mp
  ok 5 nk_qlease.test_destroy
  # Totals: pass:4 fail:0 xfail:0 xpass:0 skip:1 error:0

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/netdev/20260522113225.241337-1-bjorn@kernel.org/ [0]
---
 .../selftests/drivers/net/hw/nk_qlease.py     | 107 +++++++++++++++++-
 1 file changed, 106 insertions(+), 1 deletion(-)

diff --git a/tools/testing/selftests/drivers/net/hw/nk_qlease.py b/tools/testing/selftests/drivers/net/hw/nk_qlease.py
index b97663820ccf..4f53034c9a50 100755
--- a/tools/testing/selftests/drivers/net/hw/nk_qlease.py
+++ b/tools/testing/selftests/drivers/net/hw/nk_qlease.py
@@ -32,6 +32,31 @@ from lib.py import (
 )
 from lib.py import KsftSkipEx, CmdExitFailure
 
+# iou-zcrx exits with 42 from setup_zcrx() when the NIC does not advertise
+# QCFG_RX_PAGE_SIZE (or otherwise rejects the requested rx_buf_len).
+SKIP_CODE = 42
+
+
+def _restore_hugepages(count):
+    with open("/proc/sys/vm/nr_hugepages", "w", encoding="utf-8") as f:
+        f.write(str(count))
+
+
+def _mp_clear_wait(cfg, src_queue):
+    """Wait for the io_uring memory provider to clear from the leased
+    physical queue; io_uring tears it down asynchronously after the
+    process holding the ifq exits."""
+    netdevnl = NetdevFamily()
+    deadline = time.time() + 5
+    while time.time() < deadline:
+        queue_info = netdevnl.queue_get(
+            {"ifindex": cfg.ifindex, "id": src_queue, "type": "rx"}
+        )
+        if "io-uring" not in queue_info:
+            return
+        time.sleep(0.1)
+    raise TimeoutError("Timed out waiting for memory provider to clear")
+
 
 def _create_netkit_pair(cfg, rxqueues=2):
     if cfg.nk_host_ifname:
@@ -188,6 +213,80 @@ def test_iou_zcrx(cfg) -> None:
         cmd(tx_cmd, host=cfg.remote)
 
 
+def test_iou_zcrx_large_buf(cfg) -> None:
+    """iou-zcrx with rx_buf_len > page size, going through a netkit-leased
+    queue. Exercises the queue rx-buf-len path via netif_mp_open_rxq()'s
+    lease redirect: the netkit ifindex is opaque to io_uring, but
+    rx_page_size is honoured by the *physical* qops because the lease
+    pointer rewrites the request from netkit onto the leased physical
+    rxq before supported_params/validate_qcfg are consulted.
+    """
+    cfg.require_ipver("6")
+    src_queue, nk_queue = _setup_lease(cfg)
+    defer(_teardown_netkit, cfg)
+    ethnl = EthtoolFamily()
+
+    with open("/proc/sys/vm/nr_hugepages", "r+", encoding="utf-8") as f:
+        nr_hugepages = int(f.read().strip())
+        if nr_hugepages < 64:
+            f.seek(0)
+            f.write("64")
+            defer(_restore_hugepages, nr_hugepages)
+
+    rings = ethnl.rings_get({"header": {"dev-index": cfg.ifindex}})
+    rx_rings = rings["rx"]
+    hds_thresh = rings.get("hds-thresh", 0)
+
+    ethnl.rings_set(
+        {
+            "header": {"dev-index": cfg.ifindex},
+            "tcp-data-split": "enabled",
+            "hds-thresh": 0,
+            "rx": 64,
+        }
+    )
+    defer(
+        ethnl.rings_set,
+        {
+            "header": {"dev-index": cfg.ifindex},
+            "tcp-data-split": "unknown",
+            "hds-thresh": hds_thresh,
+            "rx": rx_rings,
+        },
+    )
+
+    ethtool(f"-X {cfg.ifname} equal {src_queue}")
+    defer(ethtool, f"-X {cfg.ifname} default")
+
+    flow_rule_id = set_flow_rule(cfg, src_queue)
+    defer(ethtool, f"-N {cfg.ifname} delete {flow_rule_id}")
+
+    # -x 2 asks iou-zcrx for rx_buf_len = 2 * page_size (8 KiB on x86_64),
+    # backed by a 2 MiB hugepage area so the chunks are physically
+    # contiguous, which is what zcrx requires for non-default rx_buf_len.
+    rx_cmd = (
+        f"{cfg.bin_local} -s -p {cfg.port} "
+        f"-i {cfg.nk_guest_ifname} -q {nk_queue} -x 2"
+    )
+    tx_cmd = f"{cfg.bin_remote} -c -h {cfg.nk_guest_ipv6} -p {cfg.port} -l 12840"
+
+    # Probe via -d (dry run): exits with SKIP_CODE if the leased physical
+    # qops doesn't advertise QCFG_RX_PAGE_SIZE (e.g. older bnxt FW/HW).
+    probe = cmd(rx_cmd + " -d", fail=False, ns=cfg.netns)
+    if probe.ret == SKIP_CODE:
+        msg = probe.stdout.strip() or "rx_buf_len not supported by leased NIC"
+        raise KsftSkipEx(msg)
+
+    # A successful dry run still registered the zcrx ifq on the leased
+    # physical queue; wait for its async teardown before the real server
+    # binds the same queue.
+    _mp_clear_wait(cfg, src_queue)
+
+    with bkg(rx_cmd, exit_wait=True, ns=cfg.netns):
+        wait_port_listen(cfg.port, proto="tcp", ns=cfg.netns)
+        cmd(tx_cmd, host=cfg.remote)
+
+
 def test_attrs(cfg) -> None:
     cfg.require_ipver("6")
     src_queue, nk_queue = _setup_lease(cfg)
@@ -350,7 +449,13 @@ def main() -> None:
         cfg.port = rand_port()
 
         ksft_run(
-            [test_iou_zcrx, test_attrs, test_attach_xdp_with_mp, test_destroy],
+            [
+                test_iou_zcrx,
+                test_iou_zcrx_large_buf,
+                test_attrs,
+                test_attach_xdp_with_mp,
+                test_destroy,
+            ],
             args=(cfg,),
         )
     ksft_exit()
-- 
2.43.0

[PATCH net-next v2 4/4] selftests/net: Add hugepage kernel config dependency for zcrx

[Daniel Borkmann]

test_iou_zcrx_large_buf in drivers/net/hw/nk_qlease.py runs iou-zcrx
with rx_buf_len > page size, backed by a hugepage-mapped area. Thus
add to the Kconfig.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Bobby Eshleman <bobbyeshleman@meta.com>
---
 tools/testing/selftests/drivers/net/hw/config | 1 +
 1 file changed, 1 insertion(+)

diff --git a/tools/testing/selftests/drivers/net/hw/config b/tools/testing/selftests/drivers/net/hw/config
index cd20024218cd..ed8642b68094 100644
--- a/tools/testing/selftests/drivers/net/hw/config
+++ b/tools/testing/selftests/drivers/net/hw/config
@@ -3,6 +3,7 @@ CONFIG_FAIL_FUNCTION=y
 CONFIG_FAULT_INJECTION=y
 CONFIG_FAULT_INJECTION_DEBUG_FS=y
 CONFIG_FUNCTION_ERROR_INJECTION=y
+CONFIG_HUGETLBFS=y
 CONFIG_INET6_ESP=y
 CONFIG_INET6_ESP_OFFLOAD=y
 CONFIG_INET_ESP=y
-- 
2.43.0

Re: [PATCH net-next v2 3/4] selftests/net: Add netkit io_uring ZC test for large rx_buf_len

[Bobby Eshleman]

On Thu, Jun 11, 2026 at 10:25:26AM +0200, Daniel Borkmann wrote:
> Add test_iou_zcrx_large_buf, which runs iou-zcrx with rx_buf_len >
> page size (-x 2) through a netkit-leased RX queue. The netkit ifindex
> is opaque to io_uring, but rx_page_size is honoured by the leased
> physical qops via netif_mp_open_rxq()'s lease redirect.
> 
> Originally, I also added a BIG TCP variant on top, but dropped it
> here as fbnic (and the QEMU fbnic model) has no BIG TCP support
> to exercise it as this point.
> 
> Tested against the QEMU fbnic emulation. The new test exercises
> the > page rx_buf_len path only when the leased NIC advertises
> QCFG_RX_PAGE_SIZE; otherwise it skips.
> 
> For fbnic, I used Bjorn's patches locally [0]:
> 
>   # ./nk_qlease.py
>   TAP version 13
>   1..5
>   ok 1 nk_qlease.test_iou_zcrx
>   ok 2 nk_qlease.test_iou_zcrx_large_buf
>   ok 3 nk_qlease.test_attrs
>   ok 4 nk_qlease.test_attach_xdp_with_mp
>   ok 5 nk_qlease.test_destroy
>   # Totals: pass:5 fail:0 xfail:0 xpass:0 skip:0 error:0
> 
> Without those patches (aka not advertising QCFG_RX_PAGE_SIZE):
> 
>   # ./nk_qlease.py
>   TAP version 13
>   1..5
>   ok 1 nk_qlease.test_iou_zcrx
>   ok 2 nk_qlease.test_iou_zcrx_large_buf # SKIP Large chunks are not supported -95
>   ok 3 nk_qlease.test_attrs
>   ok 4 nk_qlease.test_attach_xdp_with_mp
>   ok 5 nk_qlease.test_destroy
>   # Totals: pass:4 fail:0 xfail:0 xpass:0 skip:1 error:0
> 
> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
> Link: https://lore.kernel.org/netdev/20260522113225.241337-1-bjorn@kernel.org/ [0]
> ---
>  .../selftests/drivers/net/hw/nk_qlease.py     | 107 +++++++++++++++++-
>  1 file changed, 106 insertions(+), 1 deletion(-)
> 
> diff --git a/tools/testing/selftests/drivers/net/hw/nk_qlease.py b/tools/testing/selftests/drivers/net/hw/nk_qlease.py
> index b97663820ccf..4f53034c9a50 100755
> --- a/tools/testing/selftests/drivers/net/hw/nk_qlease.py
> +++ b/tools/testing/selftests/drivers/net/hw/nk_qlease.py
> @@ -32,6 +32,31 @@ from lib.py import (
>  )
>  from lib.py import KsftSkipEx, CmdExitFailure
>  
> +# iou-zcrx exits with 42 from setup_zcrx() when the NIC does not advertise
> +# QCFG_RX_PAGE_SIZE (or otherwise rejects the requested rx_buf_len).
> +SKIP_CODE = 42
> +
> +
> +def _restore_hugepages(count):
> +    with open("/proc/sys/vm/nr_hugepages", "w", encoding="utf-8") as f:
> +        f.write(str(count))
> +
> +
> +def _mp_clear_wait(cfg, src_queue):
> +    """Wait for the io_uring memory provider to clear from the leased
> +    physical queue; io_uring tears it down asynchronously after the
> +    process holding the ifq exits."""
> +    netdevnl = NetdevFamily()
> +    deadline = time.time() + 5
> +    while time.time() < deadline:
> +        queue_info = netdevnl.queue_get(
> +            {"ifindex": cfg.ifindex, "id": src_queue, "type": "rx"}
> +        )
> +        if "io-uring" not in queue_info:
> +            return
> +        time.sleep(0.1)
> +    raise TimeoutError("Timed out waiting for memory provider to clear")
> +
>  
>  def _create_netkit_pair(cfg, rxqueues=2):
>      if cfg.nk_host_ifname:
> @@ -188,6 +213,80 @@ def test_iou_zcrx(cfg) -> None:
>          cmd(tx_cmd, host=cfg.remote)
>  
>  
> +def test_iou_zcrx_large_buf(cfg) -> None:
> +    """iou-zcrx with rx_buf_len > page size, going through a netkit-leased
> +    queue. Exercises the queue rx-buf-len path via netif_mp_open_rxq()'s
> +    lease redirect: the netkit ifindex is opaque to io_uring, but
> +    rx_page_size is honoured by the *physical* qops because the lease
> +    pointer rewrites the request from netkit onto the leased physical
> +    rxq before supported_params/validate_qcfg are consulted.
> +    """
> +    cfg.require_ipver("6")
> +    src_queue, nk_queue = _setup_lease(cfg)
> +    defer(_teardown_netkit, cfg)
> +    ethnl = EthtoolFamily()
> +
> +    with open("/proc/sys/vm/nr_hugepages", "r+", encoding="utf-8") as f:
> +        nr_hugepages = int(f.read().strip())
> +        if nr_hugepages < 64:
> +            f.seek(0)
> +            f.write("64")
> +            defer(_restore_hugepages, nr_hugepages)
> +
> +    rings = ethnl.rings_get({"header": {"dev-index": cfg.ifindex}})
> +    rx_rings = rings["rx"]
> +    hds_thresh = rings.get("hds-thresh", 0)
> +
> +    ethnl.rings_set(
> +        {
> +            "header": {"dev-index": cfg.ifindex},
> +            "tcp-data-split": "enabled",
> +            "hds-thresh": 0,
> +            "rx": 64,
> +        }
> +    )
> +    defer(
> +        ethnl.rings_set,
> +        {
> +            "header": {"dev-index": cfg.ifindex},
> +            "tcp-data-split": "unknown",
> +            "hds-thresh": hds_thresh,
> +            "rx": rx_rings,
> +        },
> +    )
> +
> +    ethtool(f"-X {cfg.ifname} equal {src_queue}")
> +    defer(ethtool, f"-X {cfg.ifname} default")
> +
> +    flow_rule_id = set_flow_rule(cfg, src_queue)
> +    defer(ethtool, f"-N {cfg.ifname} delete {flow_rule_id}")
> +
> +    # -x 2 asks iou-zcrx for rx_buf_len = 2 * page_size (8 KiB on x86_64),
> +    # backed by a 2 MiB hugepage area so the chunks are physically
> +    # contiguous, which is what zcrx requires for non-default rx_buf_len.
> +    rx_cmd = (
> +        f"{cfg.bin_local} -s -p {cfg.port} "
> +        f"-i {cfg.nk_guest_ifname} -q {nk_queue} -x 2"
> +    )
> +    tx_cmd = f"{cfg.bin_remote} -c -h {cfg.nk_guest_ipv6} -p {cfg.port} -l 12840"
> +
> +    # Probe via -d (dry run): exits with SKIP_CODE if the leased physical
> +    # qops doesn't advertise QCFG_RX_PAGE_SIZE (e.g. older bnxt FW/HW).
> +    probe = cmd(rx_cmd + " -d", fail=False, ns=cfg.netns)
> +    if probe.ret == SKIP_CODE:
> +        msg = probe.stdout.strip() or "rx_buf_len not supported by leased NIC"
> +        raise KsftSkipEx(msg)
> +
> +    # A successful dry run still registered the zcrx ifq on the leased
> +    # physical queue; wait for its async teardown before the real server
> +    # binds the same queue.
> +    _mp_clear_wait(cfg, src_queue)
> +
> +    with bkg(rx_cmd, exit_wait=True, ns=cfg.netns):
> +        wait_port_listen(cfg.port, proto="tcp", ns=cfg.netns)
> +        cmd(tx_cmd, host=cfg.remote)
> +
> +
>  def test_attrs(cfg) -> None:
>      cfg.require_ipver("6")
>      src_queue, nk_queue = _setup_lease(cfg)
> @@ -350,7 +449,13 @@ def main() -> None:
>          cfg.port = rand_port()
>  
>          ksft_run(
> -            [test_iou_zcrx, test_attrs, test_attach_xdp_with_mp, test_destroy],
> +            [
> +                test_iou_zcrx,
> +                test_iou_zcrx_large_buf,
> +                test_attrs,
> +                test_attach_xdp_with_mp,
> +                test_destroy,
> +            ],
>              args=(cfg,),
>          )
>      ksft_exit()
> -- 
> 2.43.0
> 

Reviewed-by: Bobby Eshleman <bobbyeshleman@meta.com>