Merge branch 'netdevsim-add-napi-support'

David Wei says:

====================
netdevsim: add NAPI support

Add NAPI support to netdevsim and register its Rx queues with NAPI
instances. Then add a selftest using the new netdev Python selftest
infra to exercise the existing Netdev Netlink API, specifically the
queue-get API.

This expands test coverage and further fleshes out netdevsim as a test
device. It's still my goal to make it useful for testing things like
flow steering and ZC Rx.
====================

Link: https://lore.kernel.org/r/20240507163228.2066817-1-dw@davidwei.ukSigned-off-by: Jakub Kicinski <kuba@kernel.org>

drivers/net/netdevsim/netdev.c

@@ -28,11 +28,33 @@
 
 #include "netdevsim.h"
 
+#define NSIM_RING_SIZE		256
+
+static int nsim_napi_rx(struct nsim_rq *rq, struct sk_buff *skb)
+{
+	if (skb_queue_len(&rq->skb_queue) > NSIM_RING_SIZE) {
+		dev_kfree_skb_any(skb);
+		return NET_RX_DROP;
+	}
+
+	skb_queue_tail(&rq->skb_queue, skb);
+	return NET_RX_SUCCESS;
+}
+
+static int nsim_forward_skb(struct net_device *dev, struct sk_buff *skb,
+			    struct nsim_rq *rq)
+{
+	return __dev_forward_skb(dev, skb) ?: nsim_napi_rx(rq, skb);
+}
+
 static netdev_tx_t nsim_start_xmit(struct sk_buff *skb, struct net_device *dev)
 {
 	struct netdevsim *ns = netdev_priv(dev);
+	struct net_device *peer_dev;
 	unsigned int len = skb->len;
 	struct netdevsim *peer_ns;
+	struct nsim_rq *rq;
+	int rxq;
 
 	rcu_read_lock();
 	if (!nsim_ipsec_tx(ns, skb))
@@ -42,10 +64,18 @@ static netdev_tx_t nsim_start_xmit(struct sk_buff *skb, struct net_device *dev)
 	if (!peer_ns)
 		goto out_drop_free;
 
+	peer_dev = peer_ns->netdev;
+	rxq = skb_get_queue_mapping(skb);
+	if (rxq >= peer_dev->num_rx_queues)
+		rxq = rxq % peer_dev->num_rx_queues;
+	rq = &peer_ns->rq[rxq];
+
 	skb_tx_timestamp(skb);
-	if (unlikely(dev_forward_skb(peer_ns->netdev, skb) == NET_RX_DROP))
+	if (unlikely(nsim_forward_skb(peer_dev, skb, rq) == NET_RX_DROP))
 		goto out_drop_cnt;
 
+	napi_schedule(&rq->napi);
+
 	rcu_read_unlock();
 	u64_stats_update_begin(&ns->syncp);
 	ns->tx_packets++;
@@ -300,25 +330,146 @@ static int nsim_get_iflink(const struct net_device *dev)
 	return iflink;
 }
 
+static int nsim_rcv(struct nsim_rq *rq, int budget)
+{
+	struct sk_buff *skb;
+	int i;
+
+	for (i = 0; i < budget; i++) {
+		if (skb_queue_empty(&rq->skb_queue))
+			break;
+
+		skb = skb_dequeue(&rq->skb_queue);
+		netif_receive_skb(skb);
+	}
+
+	return i;
+}
+
+static int nsim_poll(struct napi_struct *napi, int budget)
+{
+	struct nsim_rq *rq = container_of(napi, struct nsim_rq, napi);
+	int done;
+
+	done = nsim_rcv(rq, budget);
+	napi_complete(napi);
+
+	return done;
+}
+
+static int nsim_create_page_pool(struct nsim_rq *rq)
+{
+	struct page_pool_params p = {
+		.order = 0,
+		.pool_size = NSIM_RING_SIZE,
+		.nid = NUMA_NO_NODE,
+		.dev = &rq->napi.dev->dev,
+		.napi = &rq->napi,
+		.dma_dir = DMA_BIDIRECTIONAL,
+		.netdev = rq->napi.dev,
+	};
+
+	rq->page_pool = page_pool_create(&p);
+	if (IS_ERR(rq->page_pool)) {
+		int err = PTR_ERR(rq->page_pool);
+
+		rq->page_pool = NULL;
+		return err;
+	}
+	return 0;
+}
+
+static int nsim_init_napi(struct netdevsim *ns)
+{
+	struct net_device *dev = ns->netdev;
+	struct nsim_rq *rq;
+	int err, i;
+
+	for (i = 0; i < dev->num_rx_queues; i++) {
+		rq = &ns->rq[i];
+
+		netif_napi_add(dev, &rq->napi, nsim_poll);
+	}
+
+	for (i = 0; i < dev->num_rx_queues; i++) {
+		rq = &ns->rq[i];
+
+		err = nsim_create_page_pool(rq);
+		if (err)
+			goto err_pp_destroy;
+	}
+
+	return 0;
+
+err_pp_destroy:
+	while (i--) {
+		page_pool_destroy(ns->rq[i].page_pool);
+		ns->rq[i].page_pool = NULL;
+	}
+
+	for (i = 0; i < dev->num_rx_queues; i++)
+		__netif_napi_del(&ns->rq[i].napi);
+
+	return err;
+}
+
+static void nsim_enable_napi(struct netdevsim *ns)
+{
+	struct net_device *dev = ns->netdev;
+	int i;
+
+	for (i = 0; i < dev->num_rx_queues; i++) {
+		struct nsim_rq *rq = &ns->rq[i];
+
+		netif_queue_set_napi(dev, i, NETDEV_QUEUE_TYPE_RX, &rq->napi);
+		napi_enable(&rq->napi);
+	}
+}
+
 static int nsim_open(struct net_device *dev)
 {
 	struct netdevsim *ns = netdev_priv(dev);
-	struct page_pool_params pp = { 0 };
+	int err;
 
-	pp.pool_size = 128;
-	pp.dev = &dev->dev;
-	pp.dma_dir = DMA_BIDIRECTIONAL;
-	pp.netdev = dev;
+	err = nsim_init_napi(ns);
+	if (err)
+		return err;
 
-	ns->pp = page_pool_create(&pp);
-	return PTR_ERR_OR_ZERO(ns->pp);
+	nsim_enable_napi(ns);
+
+	return 0;
+}
+
+static void nsim_del_napi(struct netdevsim *ns)
+{
+	struct net_device *dev = ns->netdev;
+	int i;
+
+	for (i = 0; i < dev->num_rx_queues; i++) {
+		struct nsim_rq *rq = &ns->rq[i];
+
+		napi_disable(&rq->napi);
+		__netif_napi_del(&rq->napi);
+	}
+	synchronize_net();
+
+	for (i = 0; i < dev->num_rx_queues; i++) {
+		page_pool_destroy(ns->rq[i].page_pool);
+		ns->rq[i].page_pool = NULL;
+	}
 }
 
 static int nsim_stop(struct net_device *dev)
 {
 	struct netdevsim *ns = netdev_priv(dev);
+	struct netdevsim *peer;
+
+	netif_carrier_off(dev);
+	peer = rtnl_dereference(ns->peer);
+	if (peer)
+		netif_carrier_off(peer->netdev);
 
-	page_pool_destroy(ns->pp);
+	nsim_del_napi(ns);
 
 	return 0;
 }
@@ -437,7 +588,7 @@ nsim_pp_hold_write(struct file *file, const char __user *data,
 	if (!netif_running(ns->netdev) && val) {
 		ret = -ENETDOWN;
 	} else if (val) {
-		ns->page = page_pool_dev_alloc_pages(ns->pp);
+		ns->page = page_pool_dev_alloc_pages(ns->rq[0].page_pool);
 		if (!ns->page)
 			ret = -ENOMEM;
 	} else {
@@ -477,6 +628,35 @@ static void nsim_setup(struct net_device *dev)
 	dev->xdp_features = NETDEV_XDP_ACT_HW_OFFLOAD;
 }
 
+static int nsim_queue_init(struct netdevsim *ns)
+{
+	struct net_device *dev = ns->netdev;
+	int i;
+
+	ns->rq = kvcalloc(dev->num_rx_queues, sizeof(*ns->rq),
+			  GFP_KERNEL_ACCOUNT | __GFP_RETRY_MAYFAIL);
+	if (!ns->rq)
+		return -ENOMEM;
+
+	for (i = 0; i < dev->num_rx_queues; i++)
+		skb_queue_head_init(&ns->rq[i].skb_queue);
+
+	return 0;
+}
+
+static void nsim_queue_free(struct netdevsim *ns)
+{
+	struct net_device *dev = ns->netdev;
+	int i;
+
+	for (i = 0; i < dev->num_rx_queues; i++)
+		skb_queue_purge_reason(&ns->rq[i].skb_queue,
+				       SKB_DROP_REASON_QUEUE_PURGE);
+
+	kvfree(ns->rq);
+	ns->rq = NULL;
+}
+
 static int nsim_init_netdevsim(struct netdevsim *ns)
 {
 	struct mock_phc *phc;
@@ -495,10 +675,14 @@ static int nsim_init_netdevsim(struct netdevsim *ns)
 		goto err_phc_destroy;
 
 	rtnl_lock();
-	err = nsim_bpf_init(ns);
+	err = nsim_queue_init(ns);
 	if (err)
 		goto err_utn_destroy;
 
+	err = nsim_bpf_init(ns);
+	if (err)
+		goto err_rq_destroy;
+
 	nsim_macsec_init(ns);
 	nsim_ipsec_init(ns);
 
@@ -512,6 +696,8 @@ static int nsim_init_netdevsim(struct netdevsim *ns)
 	nsim_ipsec_teardown(ns);
 	nsim_macsec_teardown(ns);
 	nsim_bpf_uninit(ns);
+err_rq_destroy:
+	nsim_queue_free(ns);
 err_utn_destroy:
 	rtnl_unlock();
 	nsim_udp_tunnels_info_destroy(ns->netdev);
@@ -593,6 +779,7 @@ void nsim_destroy(struct netdevsim *ns)
 		nsim_macsec_teardown(ns);
 		nsim_ipsec_teardown(ns);
 		nsim_bpf_uninit(ns);
+		nsim_queue_free(ns);
 	}
 	rtnl_unlock();
 	if (nsim_dev_port_is_pf(ns->nsim_dev_port))

drivers/net/netdevsim/netdevsim.h

@@ -90,11 +90,18 @@ struct nsim_ethtool {
 	struct ethtool_fecparam fec;
 };
 
+struct nsim_rq {
+	struct napi_struct napi;
+	struct sk_buff_head skb_queue;
+	struct page_pool *page_pool;
+};
+
 struct netdevsim {
 	struct net_device *netdev;
 	struct nsim_dev *nsim_dev;
 	struct nsim_dev_port *nsim_dev_port;
 	struct mock_phc *phc;
+	struct nsim_rq *rq;
 
 	u64 tx_packets;
 	u64 tx_bytes;
@@ -125,7 +132,6 @@ struct netdevsim {
 		struct debugfs_u32_array dfs_ports[2];
 	} udp_ports;
 
-	struct page_pool *pp;
 	struct page *page;
 	struct dentry *pp_dfs;
 

tools/testing/selftests/drivers/net/Makefile

@@ -4,6 +4,7 @@ TEST_INCLUDES := $(wildcard lib/py/*.py)
 
 TEST_PROGS := \
 	ping.py \
+	queues.py \
 	stats.py \
 # end of TEST_PROGS
 

tools/testing/selftests/drivers/net/lib/py/env.py

@@ -36,7 +36,7 @@ class NetDrvEnv:
     """
     Class for a single NIC / host env, with no remote end
     """
-    def __init__(self, src_path):
+    def __init__(self, src_path, **kwargs):
         self._ns = None
 
         self.env = _load_env_file(src_path)
@@ -44,11 +44,13 @@ class NetDrvEnv:
         if 'NETIF' in self.env:
             self.dev = ip("link show dev " + self.env['NETIF'], json=True)[0]
         else:
-            self._ns = NetdevSimDev()
+            self._ns = NetdevSimDev(**kwargs)
             self.dev = self._ns.nsims[0].dev
         self.ifindex = self.dev['ifindex']
 
     def __enter__(self):
+        ip(f"link set dev {self.dev['ifname']} up")
+
         return self
 
     def __exit__(self, ex_type, ex_value, ex_tb):

tools/testing/selftests/drivers/net/queues.py

+#!/usr/bin/env python3
+# SPDX-License-Identifier: GPL-2.0
+
+from lib.py import ksft_run, ksft_exit, ksft_eq, KsftSkipEx
+from lib.py import EthtoolFamily, NetdevFamily
+from lib.py import NetDrvEnv
+from lib.py import cmd
+import glob
+
+
+def sys_get_queues(ifname) -> int:
+    folders = glob.glob(f'/sys/class/net/{ifname}/queues/rx-*')
+    return len(folders)
+
+
+def nl_get_queues(cfg, nl):
+    queues = nl.queue_get({'ifindex': cfg.ifindex}, dump=True)
+    if queues:
+        return len([q for q in queues if q['type'] == 'rx'])
+    return None
+
+
+def get_queues(cfg, nl) -> None:
+    queues = nl_get_queues(cfg, nl)
+    if not queues:
+        raise KsftSkipEx('queue-get not supported by device')
+
+    expected = sys_get_queues(cfg.dev['ifname'])
+    ksft_eq(queues, expected)
+
+
+def addremove_queues(cfg, nl) -> None:
+    queues = nl_get_queues(cfg, nl)
+    if not queues:
+        raise KsftSkipEx('queue-get not supported by device')
+
+    curr_queues = sys_get_queues(cfg.dev['ifname'])
+    if curr_queues == 1:
+        raise KsftSkipEx('cannot decrement queue: already at 1')
+
+    netnl = EthtoolFamily()
+    channels = netnl.channels_get({'header': {'dev-index': cfg.ifindex}})
+    if channels['combined-count'] == 0:
+        rx_type = 'rx'
+    else:
+        rx_type = 'combined'
+
+    expected = curr_queues - 1
+    cmd(f"ethtool -L {cfg.dev['ifname']} {rx_type} {expected}", timeout=10)
+    queues = nl_get_queues(cfg, nl)
+    ksft_eq(queues, expected)
+
+    expected = curr_queues
+    cmd(f"ethtool -L {cfg.dev['ifname']} {rx_type} {expected}", timeout=10)
+    queues = nl_get_queues(cfg, nl)
+    ksft_eq(queues, expected)
+
+
+def main() -> None:
+    with NetDrvEnv(__file__, queue_count=3) as cfg:
+        ksft_run([get_queues, addremove_queues], args=(cfg, NetdevFamily()))
+    ksft_exit()
+
+
+if __name__ == "__main__":
+    main()

tools/testing/selftests/net/lib/py/nsim.py

@@ -49,7 +49,7 @@ class NetdevSimDev:
         with open(fullpath, "w") as f:
             f.write(val)
 
-    def __init__(self, port_count=1, ns=None):
+    def __init__(self, port_count=1, queue_count=1, ns=None):
         # nsim will spawn in init_net, we'll set to actual ns once we switch it there
         self.ns = None
 
@@ -59,7 +59,7 @@ class NetdevSimDev:
         addr = random.randrange(1 << 15)
         while True:
             try:
-                self.ctrl_write("new_device", "%u %u" % (addr, port_count))
+                self.ctrl_write("new_device", "%u %u %u" % (addr, port_count, queue_count))
             except OSError as e:
                 if e.errno == errno.ENOSPC:
                     addr = random.randrange(1 << 15)

tools/testing/selftests/net/lib/py/utils.py

@@ -8,7 +8,7 @@ import time
 
 
 class cmd:
-    def __init__(self, comm, shell=True, fail=True, ns=None, background=False, host=None):
+    def __init__(self, comm, shell=True, fail=True, ns=None, background=False, host=None, timeout=5):
         if ns:
             comm = f'ip netns exec {ns} ' + comm
 
@@ -23,15 +23,15 @@ class cmd:
             self.proc = subprocess.Popen(comm, shell=shell, stdout=subprocess.PIPE,
                                          stderr=subprocess.PIPE)
         if not background:
-            self.process(terminate=False, fail=fail)
+            self.process(terminate=False, fail=fail, timeout=timeout)
 
-    def process(self, terminate=True, fail=None):
+    def process(self, terminate=True, fail=None, timeout=5):
         if fail is None:
             fail = not terminate
 
         if terminate:
             self.proc.terminate()
-        stdout, stderr = self.proc.communicate(timeout=5)
+        stdout, stderr = self.proc.communicate(timeout)
         self.stdout = stdout.decode("utf-8")
         self.stderr = stderr.decode("utf-8")
         self.proc.stdout.close()