Merge branch 'netkit-support-for-io_uring-zero-copy-and-af_xdp'

Daniel Borkmann says:

====================
netkit: Support for io_uring zero-copy and AF_XDP

Containers use virtual netdevs to route traffic from a physical netdev
in the host namespace. They do not have access to the physical netdev
in the host and thus can't use memory providers or AF_XDP that require
reconfiguring/restarting queues in the physical netdev.

This patchset adds the concept of queue leasing to virtual netdevs that
allow containers to use memory providers and AF_XDP at native speed.
Leased queues are bound to a real queue in a physical netdev and act
as a proxy.

Memory providers and AF_XDP operations take an ifindex and queue id,
so containers would pass in an ifindex for a virtual netdev and a queue
id of a leased queue, which then gets proxied to the underlying real
queue.

We have implemented support for this concept in netkit and tested the
latter against Nvidia ConnectX-6 (mlx5) as well as Broadcom BCM957504
(bnxt_en) 100G NICs. For more details see the individual patches.
====================

Link: https://patch.msgid.link/20260402231031.447597-1-daniel@iogearbox.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

4 files changed, 1421 insertions, 2 deletions

diff --git a/tools/include/uapi/linux/netdev.h b/tools/include/uapi/linux/netdev.h
index e0b579a1df4f..7df1056a35fd 100644
--- a/tools/include/uapi/linux/netdev.h
+++ b/tools/include/uapi/linux/netdev.h
@@ -160,6 +160,7 @@ enum {
 	NETDEV_A_QUEUE_DMABUF,
 	NETDEV_A_QUEUE_IO_URING,
 	NETDEV_A_QUEUE_XSK,
+	NETDEV_A_QUEUE_LEASE,
 
 	__NETDEV_A_QUEUE_MAX,
 	NETDEV_A_QUEUE_MAX = (__NETDEV_A_QUEUE_MAX - 1)
@@ -203,6 +204,15 @@ enum {
 };
 
 enum {
+	NETDEV_A_LEASE_IFINDEX = 1,
+	NETDEV_A_LEASE_QUEUE,
+	NETDEV_A_LEASE_NETNS_ID,
+
+	__NETDEV_A_LEASE_MAX,
+	NETDEV_A_LEASE_MAX = (__NETDEV_A_LEASE_MAX - 1)
+};
+
+enum {
 	NETDEV_A_DMABUF_IFINDEX = 1,
 	NETDEV_A_DMABUF_QUEUES,
 	NETDEV_A_DMABUF_FD,
@@ -228,6 +238,7 @@ enum {
 	NETDEV_CMD_BIND_RX,
 	NETDEV_CMD_NAPI_SET,
 	NETDEV_CMD_BIND_TX,
+	NETDEV_CMD_QUEUE_CREATE,
 
 	__NETDEV_CMD_MAX,
 	NETDEV_CMD_MAX = (__NETDEV_CMD_MAX - 1)
diff --git a/tools/testing/selftests/drivers/net/hw/Makefile b/tools/testing/selftests/drivers/net/hw/Makefile
index deeca3f8d080..28d245e11bc4 100644
--- a/tools/testing/selftests/drivers/net/hw/Makefile
+++ b/tools/testing/selftests/drivers/net/hw/Makefile
@@ -35,6 +35,7 @@ TEST_PROGS = \
 	loopback.sh \
 	nic_timestamp.py \
 	nk_netns.py \
+	nk_qlease.py \
 	pp_alloc_fail.py \
 	rss_api.py \
 	rss_ctx.py \
diff --git a/tools/testing/selftests/drivers/net/hw/lib/py/__init__.py b/tools/testing/selftests/drivers/net/hw/lib/py/__init__.py
index df4da5078c48..84a4dab6c649 100644
--- a/tools/testing/selftests/drivers/net/hw/lib/py/__init__.py
+++ b/tools/testing/selftests/drivers/net/hw/lib/py/__init__.py
@@ -20,7 +20,7 @@ try:
     # Import one by one to avoid pylint false positives
     from net.lib.py import NetNS, NetNSEnter, NetdevSimDev
     from net.lib.py import EthtoolFamily, NetdevFamily, NetshaperFamily, \
-        NlError, RtnlFamily, DevlinkFamily, PSPFamily
+        NlError, RtnlFamily, DevlinkFamily, PSPFamily, Netlink
     from net.lib.py import CmdExitFailure
     from net.lib.py import bkg, cmd, bpftool, bpftrace, defer, ethtool, \
         fd_read_timeout, ip, rand_port, rand_ports, wait_port_listen, \
@@ -36,7 +36,7 @@ try:
 
     __all__ = ["NetNS", "NetNSEnter", "NetdevSimDev",
                "EthtoolFamily", "NetdevFamily", "NetshaperFamily",
-               "NlError", "RtnlFamily", "DevlinkFamily", "PSPFamily",
+               "NlError", "RtnlFamily", "DevlinkFamily", "PSPFamily", "Netlink",
                "CmdExitFailure",
                "bkg", "cmd", "bpftool", "bpftrace", "defer", "ethtool",
                "fd_read_timeout", "ip", "rand_port", "rand_ports",
diff --git a/tools/testing/selftests/drivers/net/hw/nk_qlease.py b/tools/testing/selftests/drivers/net/hw/nk_qlease.py
new file mode 100755
index 000000000000..2bc5ffe96c7d
--- /dev/null
+++ b/tools/testing/selftests/drivers/net/hw/nk_qlease.py
@@ -0,0 +1,1407 @@
+#!/usr/bin/env python3
+# SPDX-License-Identifier: GPL-2.0
+
+import errno
+import re
+import time
+import threading
+from os import path
+from lib.py import (
+    ksft_run,
+    ksft_exit,
+    ksft_eq,
+    ksft_ne,
+    ksft_in,
+    ksft_not_in,
+    ksft_raises,
+)
+from lib.py import (
+    NetDrvContEnv,
+    NetNS,
+    NetNSEnter,
+    EthtoolFamily,
+    NetdevFamily,
+    RtnlFamily,
+    NetdevSimDev,
+)
+from lib.py import (
+    NlError,
+    Netlink,
+    bkg,
+    cmd,
+    defer,
+    ethtool,
+    ip,
+    rand_port,
+    wait_port_listen,
+)
+from lib.py import KsftSkipEx, CmdExitFailure
+
+
+def set_flow_rule(cfg):
+    output = ethtool(
+        f"-N {cfg.ifname} flow-type tcp6 dst-port {cfg.port} action {cfg.src_queue}"
+    ).stdout
+    values = re.search(r"ID (\d+)", output).group(1)
+    return int(values)
+
+
+def create_netkit(rxqueues):
+    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
+    ]
+    nk_links.sort(key=lambda x: x["ifindex"])
+    return (
+        nk_links[1]["ifname"],
+        nk_links[1]["ifindex"],
+        nk_links[0]["ifname"],
+        nk_links[0]["ifindex"],
+    )
+
+
+def create_netkit_single(rxqueues):
+    rtnl = RtnlFamily()
+    rtnl.newlink(
+        {
+            "linkinfo": {
+                "kind": "netkit",
+                "data": {
+                    "mode": "l2",
+                    "pairing": "single",
+                },
+            },
+            "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 "UP" not in link.get("flags", [])
+    ]
+    return nk_links[0]["ifname"], nk_links[0]["ifindex"]
+
+
+def test_remove_phys(netns) -> None:
+    nsimdev = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev.remove)
+    nsim = nsimdev.nsims[0]
+    ip(f"link set dev {nsim.ifname} up")
+
+    nk_host, _, nk_guest, nk_guest_idx = create_netkit(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_host}", fail=False)
+
+    ip(f"link set dev {nk_guest} netns {netns.name}")
+    ip(f"link set dev {nk_host} up")
+    ip(f"link set dev {nk_guest} up", ns=netns)
+
+    src_queue = 1
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        result = netdevnl.queue_create(
+            {
+                "ifindex": nk_guest_idx,
+                "type": "rx",
+                "lease": {
+                    "ifindex": nsim.ifindex,
+                    "queue": {"id": src_queue, "type": "rx"},
+                    "netns-id": 0,
+                },
+            }
+        )
+        nk_queue_id = result["id"]
+
+    netdevnl = NetdevFamily()
+    queue_info = netdevnl.queue_get(
+        {"ifindex": nsim.ifindex, "id": src_queue, "type": "rx"}
+    )
+    ksft_in("lease", queue_info)
+    ksft_eq(queue_info["lease"]["ifindex"], nk_guest_idx)
+    ksft_eq(queue_info["lease"]["queue"]["id"], nk_queue_id)
+
+    nsimdev.remove()
+    time.sleep(0.1)
+    ret = cmd(f"ip link show dev {nk_host}", fail=False)
+    ksft_ne(ret.ret, 0)
+
+
+def test_double_lease(netns) -> None:
+    nsimdev = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev.remove)
+    nsim = nsimdev.nsims[0]
+    ip(f"link set dev {nsim.ifname} up")
+
+    nk_host, _, nk_guest, nk_guest_idx = create_netkit(rxqueues=3)
+    defer(cmd, f"ip link del dev {nk_host}")
+
+    ip(f"link set dev {nk_guest} netns {netns.name}")
+    ip(f"link set dev {nk_host} up")
+    ip(f"link set dev {nk_guest} up", ns=netns)
+
+    src_queue = 1
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        result = netdevnl.queue_create(
+            {
+                "ifindex": nk_guest_idx,
+                "type": "rx",
+                "lease": {
+                    "ifindex": nsim.ifindex,
+                    "queue": {"id": src_queue, "type": "rx"},
+                    "netns-id": 0,
+                },
+            }
+        )
+        ksft_eq(result["id"], 1)
+
+        with ksft_raises(NlError) as e:
+            netdevnl.queue_create(
+                {
+                    "ifindex": nk_guest_idx,
+                    "type": "rx",
+                    "lease": {
+                        "ifindex": nsim.ifindex,
+                        "queue": {"id": src_queue, "type": "rx"},
+                        "netns-id": 0,
+                    },
+                }
+            )
+        ksft_eq(e.exception.nl_msg.error, -errno.EBUSY)
+
+
+def test_virtual_lessor(netns) -> None:
+    nk_host_a, _, nk_guest_a, nk_guest_a_idx = create_netkit(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_host_a}")
+    ip(f"link set dev {nk_host_a} up")
+    ip(f"link set dev {nk_guest_a} up")
+
+    nk_host_b, _, nk_guest_b, nk_guest_b_idx = create_netkit(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_host_b}")
+
+    ip(f"link set dev {nk_guest_b} netns {netns.name}")
+    ip(f"link set dev {nk_host_b} up")
+    ip(f"link set dev {nk_guest_b} up", ns=netns)
+
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        with ksft_raises(NlError) as e:
+            netdevnl.queue_create(
+                {
+                    "ifindex": nk_guest_b_idx,
+                    "type": "rx",
+                    "lease": {
+                        "ifindex": nk_guest_a_idx,
+                        "queue": {"id": 0, "type": "rx"},
+                        "netns-id": 0,
+                    },
+                }
+            )
+        ksft_eq(e.exception.nl_msg.error, -errno.EINVAL)
+
+
+def test_phys_lessee(_netns) -> None:
+    nsimdev_a = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev_a.remove)
+    nsim_a = nsimdev_a.nsims[0]
+    ip(f"link set dev {nsim_a.ifname} up")
+
+    nsimdev_b = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev_b.remove)
+    nsim_b = nsimdev_b.nsims[0]
+    ip(f"link set dev {nsim_b.ifname} up")
+
+    netdevnl = NetdevFamily()
+    with ksft_raises(NlError) as e:
+        netdevnl.queue_create(
+            {
+                "ifindex": nsim_a.ifindex,
+                "type": "rx",
+                "lease": {
+                    "ifindex": nsim_b.ifindex,
+                    "queue": {"id": 0, "type": "rx"},
+                },
+            }
+        )
+    ksft_eq(e.exception.nl_msg.error, -errno.EINVAL)
+
+
+def test_different_lessors(netns) -> None:
+    nsimdev_a = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev_a.remove)
+    nsim_a = nsimdev_a.nsims[0]
+    ip(f"link set dev {nsim_a.ifname} up")
+
+    nsimdev_b = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev_b.remove)
+    nsim_b = nsimdev_b.nsims[0]
+    ip(f"link set dev {nsim_b.ifname} up")
+
+    nk_host, _, nk_guest, nk_guest_idx = create_netkit(rxqueues=3)
+    defer(cmd, f"ip link del dev {nk_host}", fail=False)
+
+    ip(f"link set dev {nk_guest} netns {netns.name}")
+    ip(f"link set dev {nk_host} up")
+    ip(f"link set dev {nk_guest} up", ns=netns)
+
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        netdevnl.queue_create(
+            {
+                "ifindex": nk_guest_idx,
+                "type": "rx",
+                "lease": {
+                    "ifindex": nsim_a.ifindex,
+                    "queue": {"id": 1, "type": "rx"},
+                    "netns-id": 0,
+                },
+            }
+        )
+
+        with ksft_raises(NlError) as e:
+            netdevnl.queue_create(
+                {
+                    "ifindex": nk_guest_idx,
+                    "type": "rx",
+                    "lease": {
+                        "ifindex": nsim_b.ifindex,
+                        "queue": {"id": 1, "type": "rx"},
+                        "netns-id": 0,
+                    },
+                }
+            )
+        ksft_eq(e.exception.nl_msg.error, -errno.EOPNOTSUPP)
+
+
+def test_queue_out_of_range(netns) -> None:
+    nsimdev = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev.remove)
+    nsim = nsimdev.nsims[0]
+    ip(f"link set dev {nsim.ifname} up")
+
+    nk_host, _, nk_guest, nk_guest_idx = create_netkit(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_host}", fail=False)
+
+    ip(f"link set dev {nk_guest} netns {netns.name}")
+    ip(f"link set dev {nk_host} up")
+    ip(f"link set dev {nk_guest} up", ns=netns)
+
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        with ksft_raises(NlError) as e:
+            netdevnl.queue_create(
+                {
+                    "ifindex": nk_guest_idx,
+                    "type": "rx",
+                    "lease": {
+                        "ifindex": nsim.ifindex,
+                        "queue": {"id": 2, "type": "rx"},
+                        "netns-id": 0,
+                    },
+                }
+            )
+        ksft_eq(e.exception.nl_msg.error, -errno.ERANGE)
+
+
+def test_resize_leased(netns) -> None:
+    nsimdev = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev.remove)
+    nsim = nsimdev.nsims[0]
+    ip(f"link set dev {nsim.ifname} up")
+
+    nk_host, _, nk_guest, nk_guest_idx = create_netkit(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_host}", fail=False)
+
+    ip(f"link set dev {nk_guest} netns {netns.name}")
+    ip(f"link set dev {nk_host} up")
+    ip(f"link set dev {nk_guest} up", ns=netns)
+
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        netdevnl.queue_create(
+            {
+                "ifindex": nk_guest_idx,
+                "type": "rx",
+                "lease": {
+                    "ifindex": nsim.ifindex,
+                    "queue": {"id": 1, "type": "rx"},
+                    "netns-id": 0,
+                },
+            }
+        )
+
+    ethnl = EthtoolFamily()
+    with ksft_raises(NlError) as e:
+        ethnl.channels_set({"header": {"dev-index": nsim.ifindex}, "combined-count": 1})
+    ksft_eq(e.exception.nl_msg.error, -errno.EINVAL)
+
+
+def test_self_lease(_netns) -> None:
+    nk_host, _, _, nk_guest_idx = create_netkit(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_host}", fail=False)
+
+    netdevnl = NetdevFamily()
+    with ksft_raises(NlError) as e:
+        netdevnl.queue_create(
+            {
+                "ifindex": nk_guest_idx,
+                "type": "rx",
+                "lease": {
+                    "ifindex": nk_guest_idx,
+                    "queue": {"id": 0, "type": "rx"},
+                },
+            }
+        )
+    ksft_eq(e.exception.nl_msg.error, -errno.EINVAL)
+
+
+def test_veth_queue_create(netns) -> None:
+    nsimdev = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev.remove)
+    nsim = nsimdev.nsims[0]
+    ip(f"link set dev {nsim.ifname} up")
+
+    ip("link add veth0 type veth peer name veth1")
+    defer(cmd, "ip link del dev veth0", fail=False)
+
+    all_links = ip("-d link show", json=True)
+    veth_peer = [
+        link
+        for link in all_links
+        if link.get("ifname") == "veth1"
+    ]
+    veth_peer_idx = veth_peer[0]["ifindex"]
+
+    ip(f"link set dev veth1 netns {netns.name}")
+    ip("link set dev veth0 up")
+    ip("link set dev veth1 up", ns=netns)
+
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        with ksft_raises(NlError) as e:
+            netdevnl.queue_create(
+                {
+                    "ifindex": veth_peer_idx,
+                    "type": "rx",
+                    "lease": {
+                        "ifindex": nsim.ifindex,
+                        "queue": {"id": 1, "type": "rx"},
+                        "netns-id": 0,
+                    },
+                }
+            )
+        ksft_eq(e.exception.nl_msg.error, -errno.EINVAL)
+
+
+def test_create_tx_type(netns) -> None:
+    nsimdev = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev.remove)
+    nsim = nsimdev.nsims[0]
+    ip(f"link set dev {nsim.ifname} up")
+
+    nk_host, _, nk_guest, nk_guest_idx = create_netkit(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_host}", fail=False)
+
+    ip(f"link set dev {nk_guest} netns {netns.name}")
+    ip(f"link set dev {nk_host} up")
+    ip(f"link set dev {nk_guest} up", ns=netns)
+
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        with ksft_raises(NlError) as e:
+            netdevnl.queue_create(
+                {
+                    "ifindex": nk_guest_idx,
+                    "type": "tx",
+                    "lease": {
+                        "ifindex": nsim.ifindex,
+                        "queue": {"id": 1, "type": "rx"},
+                        "netns-id": 0,
+                    },
+                }
+            )
+        ksft_eq(e.exception.nl_msg.error, -errno.EINVAL)
+
+
+def test_create_primary(_netns) -> None:
+    nsimdev = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev.remove)
+    nsim = nsimdev.nsims[0]
+    ip(f"link set dev {nsim.ifname} up")
+
+    nk_host, nk_host_idx, _, _ = create_netkit(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_host}", fail=False)
+
+    ip(f"link set dev {nk_host} up")
+
+    netdevnl = NetdevFamily()
+    with ksft_raises(NlError) as e:
+        netdevnl.queue_create(
+            {
+                "ifindex": nk_host_idx,
+                "type": "rx",
+                "lease": {
+                    "ifindex": nsim.ifindex,
+                    "queue": {"id": 1, "type": "rx"},
+                },
+            }
+        )
+    ksft_eq(e.exception.nl_msg.error, -errno.EOPNOTSUPP)
+
+
+def test_create_limit(netns) -> None:
+    nsimdev = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev.remove)
+    nsim = nsimdev.nsims[0]
+    ip(f"link set dev {nsim.ifname} up")
+
+    nk_host, _, nk_guest, nk_guest_idx = create_netkit(rxqueues=1)
+    defer(cmd, f"ip link del dev {nk_host}", fail=False)
+
+    ip(f"link set dev {nk_guest} netns {netns.name}")
+    ip(f"link set dev {nk_host} up")
+    ip(f"link set dev {nk_guest} up", ns=netns)
+
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        with ksft_raises(NlError) as e:
+            netdevnl.queue_create(
+                {
+                    "ifindex": nk_guest_idx,
+                    "type": "rx",
+                    "lease": {
+                        "ifindex": nsim.ifindex,
+                        "queue": {"id": 1, "type": "rx"},
+                        "netns-id": 0,
+                    },
+                }
+            )
+        ksft_eq(e.exception.nl_msg.error, -errno.EINVAL)
+
+
+def test_link_flap_phys(netns) -> None:
+    nsimdev = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev.remove)
+    nsim = nsimdev.nsims[0]
+    ip(f"link set dev {nsim.ifname} up")
+
+    nk_host, _, nk_guest, nk_guest_idx = create_netkit(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_host}")
+
+    ip(f"link set dev {nk_guest} netns {netns.name}")
+    ip(f"link set dev {nk_host} up")
+    ip(f"link set dev {nk_guest} up", ns=netns)
+
+    src_queue = 1
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        result = netdevnl.queue_create(
+            {
+                "ifindex": nk_guest_idx,
+                "type": "rx",
+                "lease": {
+                    "ifindex": nsim.ifindex,
+                    "queue": {"id": src_queue, "type": "rx"},
+                    "netns-id": 0,
+                },
+            }
+        )
+        nk_queue_id = result["id"]
+
+    netdevnl = NetdevFamily()
+    queue_info = netdevnl.queue_get(
+        {"ifindex": nsim.ifindex, "id": src_queue, "type": "rx"}
+    )
+    ksft_in("lease", queue_info)
+    ksft_eq(queue_info["lease"]["queue"]["id"], nk_queue_id)
+
+    # Link flap the physical device
+    ip(f"link set dev {nsim.ifname} down")
+    ip(f"link set dev {nsim.ifname} up")
+
+    # Verify lease survives the flap
+    queue_info = netdevnl.queue_get(
+        {"ifindex": nsim.ifindex, "id": src_queue, "type": "rx"}
+    )
+    ksft_in("lease", queue_info)
+    ksft_eq(queue_info["lease"]["queue"]["id"], nk_queue_id)
+
+
+def test_queue_get_virtual(netns) -> None:
+    nsimdev = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev.remove)
+    nsim = nsimdev.nsims[0]
+    ip(f"link set dev {nsim.ifname} up")
+
+    nk_host, _, nk_guest, nk_guest_idx = create_netkit(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_host}")
+
+    ip(f"link set dev {nk_guest} netns {netns.name}")
+    ip(f"link set dev {nk_host} up")
+    ip(f"link set dev {nk_guest} up", ns=netns)
+
+    src_queue = 1
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        result = netdevnl.queue_create(
+            {
+                "ifindex": nk_guest_idx,
+                "type": "rx",
+                "lease": {
+                    "ifindex": nsim.ifindex,
+                    "queue": {"id": src_queue, "type": "rx"},
+                    "netns-id": 0,
+                },
+            }
+        )
+        nk_queue_id = result["id"]
+
+        # queue-get on virtual device's leased queue should not show lease
+        # info (lease info is only shown from the physical device's side)
+        queue_info = netdevnl.queue_get(
+            {"ifindex": nk_guest_idx, "id": nk_queue_id, "type": "rx"}
+        )
+        ksft_eq(queue_info["id"], nk_queue_id)
+        ksft_eq(queue_info["ifindex"], nk_guest_idx)
+        ksft_not_in("lease", queue_info)
+
+        # Default queue (not leased) also has no lease info
+        queue_info = netdevnl.queue_get(
+            {"ifindex": nk_guest_idx, "id": 0, "type": "rx"}
+        )
+        ksft_not_in("lease", queue_info)
+
+
+def test_remove_virt_first(netns) -> None:
+    nsimdev = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev.remove)
+    nsim = nsimdev.nsims[0]
+    ip(f"link set dev {nsim.ifname} up")
+
+    nk_host, _, nk_guest, nk_guest_idx = create_netkit(rxqueues=2)
+
+    ip(f"link set dev {nk_guest} netns {netns.name}")
+    ip(f"link set dev {nk_host} up")
+    ip(f"link set dev {nk_guest} up", ns=netns)
+
+    src_queue = 1
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        result = netdevnl.queue_create(
+            {
+                "ifindex": nk_guest_idx,
+                "type": "rx",
+                "lease": {
+                    "ifindex": nsim.ifindex,
+                    "queue": {"id": src_queue, "type": "rx"},
+                    "netns-id": 0,
+                },
+            }
+        )
+        ksft_eq(result["id"], 1)
+
+    netdevnl = NetdevFamily()
+    queue_info = netdevnl.queue_get(
+        {"ifindex": nsim.ifindex, "id": src_queue, "type": "rx"}
+    )
+    ksft_in("lease", queue_info)
+    ksft_eq(queue_info["lease"]["queue"]["id"], result["id"])
+
+    # Delete netkit (virtual device removed first, physical stays)
+    cmd(f"ip link del dev {nk_host}")
+
+    # Verify lease is cleaned up on physical device
+    queue_info = netdevnl.queue_get(
+        {"ifindex": nsim.ifindex, "id": src_queue, "type": "rx"}
+    )
+    ksft_not_in("lease", queue_info)
+
+
+def test_multiple_leases(netns) -> None:
+    nsimdev = NetdevSimDev(port_count=1, queue_count=3)
+    defer(nsimdev.remove)
+    nsim = nsimdev.nsims[0]
+    ip(f"link set dev {nsim.ifname} up")
+
+    nk_host, _, nk_guest, nk_guest_idx = create_netkit(rxqueues=4)
+    defer(cmd, f"ip link del dev {nk_host}", fail=False)
+
+    ip(f"link set dev {nk_guest} netns {netns.name}")
+    ip(f"link set dev {nk_host} up")
+    ip(f"link set dev {nk_guest} up", ns=netns)
+
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        r1 = netdevnl.queue_create(
+            {
+                "ifindex": nk_guest_idx,
+                "type": "rx",
+                "lease": {
+                    "ifindex": nsim.ifindex,
+                    "queue": {"id": 1, "type": "rx"},
+                    "netns-id": 0,
+                },
+            }
+        )
+        r2 = netdevnl.queue_create(
+            {
+                "ifindex": nk_guest_idx,
+                "type": "rx",
+                "lease": {
+                    "ifindex": nsim.ifindex,
+                    "queue": {"id": 2, "type": "rx"},
+                    "netns-id": 0,
+                },
+            }
+        )
+
+    ksft_eq(r1["id"], 1)
+    ksft_eq(r2["id"], 2)
+
+    # Verify both leases visible on physical device
+    netdevnl = NetdevFamily()
+    q1 = netdevnl.queue_get(
+        {"ifindex": nsim.ifindex, "id": 1, "type": "rx"}
+    )
+    q2 = netdevnl.queue_get(
+        {"ifindex": nsim.ifindex, "id": 2, "type": "rx"}
+    )
+    ksft_in("lease", q1)
+    ksft_in("lease", q2)
+    ksft_eq(q1["lease"]["ifindex"], nk_guest_idx)
+    ksft_eq(q2["lease"]["ifindex"], nk_guest_idx)
+    ksft_eq(q1["lease"]["queue"]["id"], r1["id"])
+    ksft_eq(q2["lease"]["queue"]["id"], r2["id"])
+
+
+def test_lease_queue_tx_type(netns) -> None:
+    nsimdev = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev.remove)
+    nsim = nsimdev.nsims[0]
+    ip(f"link set dev {nsim.ifname} up")
+
+    nk_host, _, nk_guest, nk_guest_idx = create_netkit(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_host}", fail=False)
+
+    ip(f"link set dev {nk_guest} netns {netns.name}")
+    ip(f"link set dev {nk_host} up")
+    ip(f"link set dev {nk_guest} up", ns=netns)
+
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        with ksft_raises(NlError) as e:
+            netdevnl.queue_create(
+                {
+                    "ifindex": nk_guest_idx,
+                    "type": "rx",
+                    "lease": {
+                        "ifindex": nsim.ifindex,
+                        "queue": {"id": 1, "type": "tx"},
+                        "netns-id": 0,
+                    },
+                }
+            )
+        ksft_eq(e.exception.nl_msg.error, -errno.EINVAL)
+
+
+def test_invalid_netns(netns) -> None:
+    nk_host, _, nk_guest, nk_guest_idx = create_netkit(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_host}", fail=False)
+
+    ip(f"link set dev {nk_guest} netns {netns.name}")
+    ip(f"link set dev {nk_host} up")
+    ip(f"link set dev {nk_guest} up", ns=netns)
+
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        with ksft_raises(NlError) as e:
+            netdevnl.queue_create(
+                {
+                    "ifindex": nk_guest_idx,
+                    "type": "rx",
+                    "lease": {
+                        "ifindex": 1,
+                        "queue": {"id": 0, "type": "rx"},
+                        "netns-id": 999,
+                    },
+                }
+            )
+        ksft_eq(e.exception.nl_msg.error, -errno.ENONET)
+
+
+def test_invalid_phys_ifindex(netns) -> None:
+    nk_host, _, nk_guest, nk_guest_idx = create_netkit(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_host}", fail=False)
+
+    ip(f"link set dev {nk_guest} netns {netns.name}")
+    ip(f"link set dev {nk_host} up")
+    ip(f"link set dev {nk_guest} up", ns=netns)
+
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        with ksft_raises(NlError) as e:
+            netdevnl.queue_create(
+                {
+                    "ifindex": nk_guest_idx,
+                    "type": "rx",
+                    "lease": {
+                        "ifindex": 99999,
+                        "queue": {"id": 0, "type": "rx"},
+                        "netns-id": 0,
+                    },
+                }
+            )
+        ksft_eq(e.exception.nl_msg.error, -errno.ENODEV)
+
+
+def test_multi_netkit_remove_phys(netns) -> None:
+    nsimdev = NetdevSimDev(port_count=1, queue_count=3)
+    defer(nsimdev.remove)
+    nsim = nsimdev.nsims[0]
+    ip(f"link set dev {nsim.ifname} up")
+
+    # Create two netkit pairs, each leasing a different physical queue
+    nk_host_a, _, nk_guest_a, nk_guest_a_idx = create_netkit(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_host_a}", fail=False)
+
+    nk_host_b, _, nk_guest_b, nk_guest_b_idx = create_netkit(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_host_b}", fail=False)
+
+    ip(f"link set dev {nk_guest_a} netns {netns.name}")
+    ip(f"link set dev {nk_host_a} up")
+    ip(f"link set dev {nk_guest_a} up", ns=netns)
+
+    ip(f"link set dev {nk_guest_b} netns {netns.name}")
+    ip(f"link set dev {nk_host_b} up")
+    ip(f"link set dev {nk_guest_b} up", ns=netns)
+
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        netdevnl.queue_create(
+            {
+                "ifindex": nk_guest_a_idx,
+                "type": "rx",
+                "lease": {
+                    "ifindex": nsim.ifindex,
+                    "queue": {"id": 1, "type": "rx"},
+                    "netns-id": 0,
+                },
+            }
+        )
+        netdevnl.queue_create(
+            {
+                "ifindex": nk_guest_b_idx,
+                "type": "rx",
+                "lease": {
+                    "ifindex": nsim.ifindex,
+                    "queue": {"id": 2, "type": "rx"},
+                    "netns-id": 0,
+                },
+            }
+        )
+
+    # Removing the physical device should take down both netkit pairs
+    nsimdev.remove()
+    time.sleep(0.1)
+    ret = cmd(f"ip link show dev {nk_host_a}", fail=False)
+    ksft_ne(ret.ret, 0)
+    ret = cmd(f"ip link show dev {nk_host_b}", fail=False)
+    ksft_ne(ret.ret, 0)
+
+
+def test_single_remove_phys(_netns) -> None:
+    nsimdev = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev.remove)
+    nsim = nsimdev.nsims[0]
+    ip(f"link set dev {nsim.ifname} up")
+
+    nk_name, nk_idx = create_netkit_single(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_name}", fail=False)
+
+    ip(f"link set dev {nk_name} up")
+
+    netdevnl = NetdevFamily()
+    netdevnl.queue_create(
+        {
+            "ifindex": nk_idx,
+            "type": "rx",
+            "lease": {
+                "ifindex": nsim.ifindex,
+                "queue": {"id": 1, "type": "rx"},
+            },
+        }
+    )
+
+    # Removing the physical device should take down the single netkit device
+    nsimdev.remove()
+    time.sleep(0.1)
+    ret = cmd(f"ip link show dev {nk_name}", fail=False)
+    ksft_ne(ret.ret, 0)
+
+
+def test_link_flap_virt(netns) -> None:
+    nsimdev = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev.remove)
+    nsim = nsimdev.nsims[0]
+    ip(f"link set dev {nsim.ifname} up")
+
+    nk_host, _, nk_guest, nk_guest_idx = create_netkit(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_host}")
+
+    ip(f"link set dev {nk_guest} netns {netns.name}")
+    ip(f"link set dev {nk_host} up")
+    ip(f"link set dev {nk_guest} up", ns=netns)
+
+    src_queue = 1
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()
+        result = netdevnl.queue_create(
+            {
+                "ifindex": nk_guest_idx,
+                "type": "rx",
+                "lease": {
+                    "ifindex": nsim.ifindex,
+                    "queue": {"id": src_queue, "type": "rx"},
+                    "netns-id": 0,
+                },
+            }
+        )
+        nk_queue_id = result["id"]
+
+    netdevnl = NetdevFamily()
+    queue_info = netdevnl.queue_get(
+        {"ifindex": nsim.ifindex, "id": src_queue, "type": "rx"}
+    )
+    ksft_in("lease", queue_info)
+    ksft_eq(queue_info["lease"]["queue"]["id"], nk_queue_id)
+
+    # Link flap the virtual (netkit) device
+    ip(f"link set dev {nk_guest} down", ns=netns)
+    ip(f"link set dev {nk_guest} up", ns=netns)
+
+    # Verify lease survives the virtual device flap
+    queue_info = netdevnl.queue_get(
+        {"ifindex": nsim.ifindex, "id": src_queue, "type": "rx"}
+    )
+    ksft_in("lease", queue_info)
+    ksft_eq(queue_info["lease"]["queue"]["id"], nk_queue_id)
+
+
+def test_phys_queue_no_lease(netns) -> None:
+    nsimdev = NetdevSimDev(port_count=1, queue_count=2)
+    defer(nsimdev.remove)
+    nsim = nsimdev.nsims[0]
+    ip(f"link set dev {nsim.ifname} up")
+
+    nk_host, _, nk_guest, nk_guest_idx = create_netkit(rxqueues=2)
+    defer(cmd, f"ip link del dev {nk_host}")
+
+    ip(f"link set dev {nk_guest} netns {netns.name}")
+    ip(f"link set dev {nk_host} up")
+    ip(f"link set dev {nk_guest} up", ns=netns)
+
+    with NetNSEnter(str(netns)):
+        netdevnl = NetdevFamily()