[PATCH net-next v5 0/3] implement kthread based napi poll

[PATCH net-next v5 0/3] implement kthread based napi poll

[Wei Wang]

The idea of moving the napi poll process out of softirq context to a
kernel thread based context is not new.
Paolo Abeni and Hannes Frederic Sowa have proposed patches to move napi
poll to kthread back in 2016. And Felix Fietkau has also proposed
patches of similar ideas to use workqueue to process napi poll just a
few weeks ago.

The main reason we'd like to push forward with this idea is that the
scheduler has poor visibility into cpu cycles spent in softirq context,
and is not able to make optimal scheduling decisions of the user threads.
For example, we see in one of the application benchmark where network
load is high, the CPUs handling network softirqs has ~80% cpu util. And
user threads are still scheduled on those CPUs, despite other more idle
cpus available in the system. And we see very high tail latencies. In this
case, we have to explicitly pin away user threads from the CPUs handling
network softirqs to ensure good performance.
With napi poll moved to kthread, scheduler is in charge of scheduling both
the kthreads handling network load, and the user threads, and is able to
make better decisions. In the previous benchmark, if we do this and we
pin the kthreads processing napi poll to specific CPUs, scheduler is
able to schedule user threads away from these CPUs automatically.

And the reason we prefer 1 kthread per napi, instead of 1 workqueue
entity per host, is that kthread is more configurable than workqueue,
and we could leverage existing tuning tools for threads, like taskset,
chrt, etc to tune scheduling class and cpu set, etc. Another reason is
if we eventually want to provide busy poll feature using kernel threads
for napi poll, kthread seems to be more suitable than workqueue.
Furthermore, for large platforms with 2 NICs attached to 2 sockets,
kthread is more flexible to be pinned to different sets of CPUs.  

In this patch series, I revived Paolo and Hannes's patch in 2016 and
made modifications. Then there are changes proposed by Felix, Jakub,
Paolo and myself on top of those, with suggestions from Eric Dumazet.

In terms of performance, I ran tcp_rr tests with 1000 flows with
various request/response sizes, with RFS/RPS disabled, and compared
performance between softirq vs kthread vs workqueue (patchset proposed
by Felix Fietkau).
Host has 56 hyper threads and 100Gbps nic, 8 rx queues and only 1 numa
node. All threads are unpinned.

        req/resp   QPS   50%tile    90%tile    99%tile    99.9%tile
softirq   1B/1B   2.75M   337us       376us      1.04ms     3.69ms
kthread   1B/1B   2.67M   371us       408us      455us      550us
workq     1B/1B   2.56M   384us       435us      673us      822us

softirq 5KB/5KB   1.46M   678us       750us      969us      2.78ms
kthread 5KB/5KB   1.44M   695us       789us      891us      1.06ms
workq   5KB/5KB   1.34M   720us       905us     1.06ms      1.57ms

softirq 1MB/1MB   11.0K   79ms       166ms      306ms       630ms
kthread 1MB/1MB   11.0K   75ms       177ms      303ms       596ms
workq   1MB/1MB   11.0K   79ms       180ms      303ms       587ms

When running workqueue implementation, I found the number of threads
used is usually twice as much as kthread implementation. This probably
introduces higher scheduling cost, which results in higher tail
latencies in most cases.

I also ran an application benchmark, which performs fixed qps remote SSD
read/write operations, with various sizes. Again, both with RFS/RPS
disabled.
The result is as follows:
         op_size  QPS   50%tile 95%tile 99%tile 99.9%tile  
softirq   4K     572.6K   385us   1.5ms  3.16ms   6.41ms
kthread   4K     572.6K   390us   803us  2.21ms   6.83ms
workq     4k     572.6K   384us   763us  3.12ms   6.87ms

softirq   64K    157.9K   736us   1.17ms 3.40ms   13.75ms
kthread   64K    157.9K   745us   1.23ms 2.76ms    9.87ms 
workq     64K    157.9K   746us   1.23ms 2.76ms    9.96ms

softirq   1M     10.98K   2.03ms  3.10ms  3.7ms   11.56ms
kthread   1M     10.98K   2.13ms  3.21ms  4.02ms  13.3ms
workq     1M     10.98K   2.13ms  3.20ms  3.99ms  14.12ms

In this set of tests, the latency is predominant by the SSD operation.
Also, the user threads are much busier compared to tcp_rr tests. We have
to pin the kthreads/workqueue threads to limit to a few CPUs, to not
disturb user threads, and provide some isolation.

Changes since v4:
Recorded the threaded setting in dev and restore it in napi_enable().

Changes since v3:
Merged and rearranged patches in a logical order for easier review. 
Changed sysfs control to be per device.

Changes since v2:
Corrected typo in patch 1, and updated the cover letter with more
detailed and updated test results.

Changes since v1:
Replaced kthread_create() with kthread_run() in patch 5 as suggested by
Felix Fietkau.

Changes since RFC:
Renamed the kthreads to be napi/<dev>-<napi_id> in patch 5 as suggested
by Hannes Frederic Sowa.

Felix Fietkau (1):
  net: extract napi poll functionality to __napi_poll()

Wei Wang (2):
  net: implement threaded-able napi poll loop support
  net: add sysfs attribute to control napi threaded mode

 include/linux/netdevice.h |  14 +--
 net/core/dev.c            | 184 +++++++++++++++++++++++++++++++++++---
 net/core/net-sysfs.c      |  63 +++++++++++++
 3 files changed, 244 insertions(+), 17 deletions(-)

-- 
2.29.2.684.gfbc64c5ab5-goog

[PATCH net-next v5 1/3] net: extract napi poll functionality to __napi_poll()

[Wei Wang]

From: Felix Fietkau <nbd@nbd.name> 

This commit introduces a new function __napi_poll() which does the main
logic of the existing napi_poll() function, and will be called by other
functions in later commits.
This idea and implementation is done by Felix Fietkau <nbd@nbd.name> and
is proposed as part of the patch to move napi work to work_queue
context.
This commit by itself is a code restructure.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Wei Wang <weiwan@google.com>
---
 net/core/dev.c | 35 +++++++++++++++++++++++++----------
 1 file changed, 25 insertions(+), 10 deletions(-)

diff --git a/net/core/dev.c b/net/core/dev.c
index 0fd0d4eb678c..adf74573f51c 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -6765,15 +6765,10 @@ void __netif_napi_del(struct napi_struct *napi)
 }
 EXPORT_SYMBOL(__netif_napi_del);
 
-static int napi_poll(struct napi_struct *n, struct list_head *repoll)
+static int __napi_poll(struct napi_struct *n, bool *repoll)
 {
-	void *have;
 	int work, weight;
 
-	list_del_init(&n->poll_list);
-
-	have = netpoll_poll_lock(n);
-
 	weight = n->weight;
 
 	/* This NAPI_STATE_SCHED test is for avoiding a race
@@ -6793,7 +6788,7 @@ static int napi_poll(struct napi_struct *n, struct list_head *repoll)
 			    n->poll, work, weight);
 
 	if (likely(work < weight))
-		goto out_unlock;
+		return work;
 
 	/* Drivers must not modify the NAPI state if they
 	 * consume the entire weight.  In such cases this code
@@ -6802,7 +6797,7 @@ static int napi_poll(struct napi_struct *n, struct list_head *repoll)
 	 */
 	if (unlikely(napi_disable_pending(n))) {
 		napi_complete(n);
-		goto out_unlock;
+		return work;
 	}
 
 	/* The NAPI context has more processing work, but busy-polling
@@ -6815,7 +6810,7 @@ static int napi_poll(struct napi_struct *n, struct list_head *repoll)
 			 */
 			napi_schedule(n);
 		}
-		goto out_unlock;
+		return work;
 	}
 
 	if (n->gro_bitmask) {
@@ -6833,9 +6828,29 @@ static int napi_poll(struct napi_struct *n, struct list_head *repoll)
 	if (unlikely(!list_empty(&n->poll_list))) {
 		pr_warn_once("%s: Budget exhausted after napi rescheduled\n",
 			     n->dev ? n->dev->name : "backlog");
-		goto out_unlock;
+		return work;
 	}
 
+	*repoll = true;
+
+	return work;
+}
+
+static int napi_poll(struct napi_struct *n, struct list_head *repoll)
+{
+	bool do_repoll = false;
+	void *have;
+	int work;
+
+	list_del_init(&n->poll_list);
+
+	have = netpoll_poll_lock(n);
+
+	work = __napi_poll(n, &do_repoll);
+
+	if (!do_repoll)
+		goto out_unlock;
+
 	list_add_tail(&n->poll_list, repoll);
 
 out_unlock:
-- 
2.29.2.684.gfbc64c5ab5-goog

[PATCH net-next v5 2/3] net: implement threaded-able napi poll loop support

[Wei Wang]

This patch allows running each napi poll loop inside its own
kernel thread.
The threaded mode could be enabled through napi_set_threaded()
api, and does not require a device up/down. The kthread gets
created on demand when napi_set_threaded() is called, and gets
shut down eventually in napi_disable().

Once that threaded mode is enabled and the kthread is
started, napi_schedule() will wake-up such thread instead
of scheduling the softirq.

The threaded poll loop behaves quite likely the net_rx_action,
but it does not have to manipulate local irqs and uses
an explicit scheduling point based on netdev_budget.

Co-developed-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Co-developed-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Co-developed-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Wei Wang <weiwan@google.com>
---
 include/linux/netdevice.h |  12 ++--
 net/core/dev.c            | 121 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 126 insertions(+), 7 deletions(-)

diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 7bf167993c05..2cd1e3975103 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -347,6 +347,7 @@ struct napi_struct {
 	struct list_head	dev_list;
 	struct hlist_node	napi_hash_node;
 	unsigned int		napi_id;
+	struct task_struct	*thread;
 };
 
 enum {
@@ -358,6 +359,7 @@ enum {
 	NAPI_STATE_NO_BUSY_POLL,	/* Do not add in napi_hash, no busy polling */
 	NAPI_STATE_IN_BUSY_POLL,	/* sk_busy_loop() owns this NAPI */
 	NAPI_STATE_PREFER_BUSY_POLL,	/* prefer busy-polling over softirq processing*/
+	NAPI_STATE_THREADED,		/* The poll is performed inside its own thread*/
 };
 
 enum {
@@ -369,6 +371,7 @@ enum {
 	NAPIF_STATE_NO_BUSY_POLL	= BIT(NAPI_STATE_NO_BUSY_POLL),
 	NAPIF_STATE_IN_BUSY_POLL	= BIT(NAPI_STATE_IN_BUSY_POLL),
 	NAPIF_STATE_PREFER_BUSY_POLL	= BIT(NAPI_STATE_PREFER_BUSY_POLL),
+	NAPIF_STATE_THREADED		= BIT(NAPI_STATE_THREADED),
 };
 
 enum gro_result {
@@ -511,13 +514,7 @@ void napi_disable(struct napi_struct *n);
  * Resume NAPI from being scheduled on this context.
  * Must be paired with napi_disable.
  */
-static inline void napi_enable(struct napi_struct *n)
-{
-	BUG_ON(!test_bit(NAPI_STATE_SCHED, &n->state));
-	smp_mb__before_atomic();
-	clear_bit(NAPI_STATE_SCHED, &n->state);
-	clear_bit(NAPI_STATE_NPSVC, &n->state);
-}
+void napi_enable(struct napi_struct *n);
 
 /**
  *	napi_synchronize - wait until NAPI is not running
@@ -2158,6 +2155,7 @@ struct net_device {
 	struct lock_class_key	*qdisc_tx_busylock;
 	struct lock_class_key	*qdisc_running_key;
 	bool			proto_down;
+	bool			threaded;
 	unsigned		wol_enabled:1;
 
 	struct list_head	net_notifier_list;
diff --git a/net/core/dev.c b/net/core/dev.c
index adf74573f51c..47c33affaa80 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -91,6 +91,7 @@
 #include <linux/etherdevice.h>
 #include <linux/ethtool.h>
 #include <linux/skbuff.h>
+#include <linux/kthread.h>
 #include <linux/bpf.h>
 #include <linux/bpf_trace.h>
 #include <net/net_namespace.h>
@@ -1475,6 +1476,36 @@ void netdev_notify_peers(struct net_device *dev)
 }
 EXPORT_SYMBOL(netdev_notify_peers);
 
+static int napi_threaded_poll(void *data);
+
+static int napi_kthread_create(struct napi_struct *n)
+{
+	int err = 0;
+
+	/* Create and wake up the kthread once to put it in
+	 * TASK_INTERRUPTIBLE mode to avoid the blocked task
+	 * warning and work with loadavg.
+	 */
+	n->thread = kthread_run(napi_threaded_poll, n, "napi/%s-%d",
+				n->dev->name, n->napi_id);
+	if (IS_ERR(n->thread)) {
+		err = PTR_ERR(n->thread);
+		pr_err("kthread_run failed with err %d\n", err);
+		n->thread = NULL;
+	}
+
+	return err;
+}
+
+static void napi_kthread_stop(struct napi_struct *n)
+{
+	if (!n->thread)
+		return;
+	kthread_stop(n->thread);
+	clear_bit(NAPI_STATE_THREADED, &n->state);
+	n->thread = NULL;
+}
+
 static int __dev_open(struct net_device *dev, struct netlink_ext_ack *extack)
 {
 	const struct net_device_ops *ops = dev->netdev_ops;
@@ -4234,6 +4265,11 @@ int gro_normal_batch __read_mostly = 8;
 static inline void ____napi_schedule(struct softnet_data *sd,
 				     struct napi_struct *napi)
 {
+	if (test_bit(NAPI_STATE_THREADED, &napi->state)) {
+		wake_up_process(napi->thread);
+		return;
+	}
+
 	list_add_tail(&napi->poll_list, &sd->poll_list);
 	__raise_softirq_irqoff(NET_RX_SOFTIRQ);
 }
@@ -6690,6 +6726,29 @@ static void init_gro_hash(struct napi_struct *napi)
 	napi->gro_bitmask = 0;
 }
 
+static int napi_set_threaded(struct napi_struct *n, bool threaded)
+{
+	int err = 0;
+
+	ASSERT_RTNL();
+
+	if (threaded == !!test_bit(NAPI_STATE_THREADED, &n->state))
+		return 0;
+	if (threaded) {
+		if (!n->thread) {
+			err = napi_kthread_create(n);
+			if (err)
+				goto out;
+		}
+		set_bit(NAPI_STATE_THREADED, &n->state);
+	} else {
+		clear_bit(NAPI_STATE_THREADED, &n->state);
+	}
+
+out:
+	return err;
+}
+
 void netif_napi_add(struct net_device *dev, struct napi_struct *napi,
 		    int (*poll)(struct napi_struct *, int), int weight)
 {
@@ -6731,12 +6790,29 @@ void napi_disable(struct napi_struct *n)
 		msleep(1);
 
 	hrtimer_cancel(&n->timer);
+	napi_kthread_stop(n);
 
 	clear_bit(NAPI_STATE_PREFER_BUSY_POLL, &n->state);
 	clear_bit(NAPI_STATE_DISABLE, &n->state);
 }
 EXPORT_SYMBOL(napi_disable);
 
+void napi_enable(struct napi_struct *n)
+{
+	bool locked = rtnl_is_locked();
+
+	BUG_ON(!test_bit(NAPI_STATE_SCHED, &n->state));
+	smp_mb__before_atomic();
+	clear_bit(NAPI_STATE_SCHED, &n->state);
+	clear_bit(NAPI_STATE_NPSVC, &n->state);
+	if (!locked)
+		rtnl_lock();
+	WARN_ON(napi_set_threaded(n, n->dev->threaded));
+	if (!locked)
+		rtnl_unlock();
+}
+EXPORT_SYMBOL(napi_enable);
+
 static void flush_gro_hash(struct napi_struct *napi)
 {
 	int i;
@@ -6859,6 +6935,51 @@ static int napi_poll(struct napi_struct *n, struct list_head *repoll)
 	return work;
 }
 
+static int napi_thread_wait(struct napi_struct *napi)
+{
+	set_current_state(TASK_INTERRUPTIBLE);
+
+	while (!kthread_should_stop() && !napi_disable_pending(napi)) {
+		if (test_bit(NAPI_STATE_SCHED, &napi->state)) {
+			WARN_ON(!list_empty(&napi->poll_list));
+			__set_current_state(TASK_RUNNING);
+			return 0;
+		}
+
+		schedule();
+		set_current_state(TASK_INTERRUPTIBLE);
+	}
+	__set_current_state(TASK_RUNNING);
+	return -1;
+}
+
+static int napi_threaded_poll(void *data)
+{
+	struct napi_struct *napi = data;
+	void *have;
+
+	while (!napi_thread_wait(napi)) {
+		for (;;) {
+			bool repoll = false;
+
+			local_bh_disable();
+
+			have = netpoll_poll_lock(napi);
+			__napi_poll(napi, &repoll);
+			netpoll_poll_unlock(have);
+
+			__kfree_skb_flush();
+			local_bh_enable();
+
+			if (!repoll)
+				break;
+
+			cond_resched();
+		}
+	}
+	return 0;
+}
+
 static __latent_entropy void net_rx_action(struct softirq_action *h)
 {
 	struct softnet_data *sd = this_cpu_ptr(&softnet_data);
-- 
2.29.2.684.gfbc64c5ab5-goog

[PATCH net-next v5 3/3] net: add sysfs attribute to control napi threaded mode

[Wei Wang]

This patch adds a new sysfs attribute to the network device class.
Said attribute provides a per-device control to enable/disable the
threaded mode for all the napi instances of the given network device.

Co-developed-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Co-developed-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Co-developed-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Wei Wang <weiwan@google.com>
---
 include/linux/netdevice.h |  2 ++
 net/core/dev.c            | 28 +++++++++++++++++
 net/core/net-sysfs.c      | 63 +++++++++++++++++++++++++++++++++++++++
 3 files changed, 93 insertions(+)

diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 2cd1e3975103..8fd54a14cb78 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -498,6 +498,8 @@ static inline bool napi_complete(struct napi_struct *n)
 	return napi_complete_done(n, 0);
 }
 
+int dev_set_threaded(struct net_device *dev, bool threaded);
+
 /**
  *	napi_disable - prevent NAPI from scheduling
  *	@n: NAPI context
diff --git a/net/core/dev.c b/net/core/dev.c
index 47c33affaa80..9737f149ff50 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -6749,6 +6749,34 @@ static int napi_set_threaded(struct napi_struct *n, bool threaded)
 	return err;
 }
 
+static void dev_disable_threaded_all(struct net_device *dev)
+{
+	struct napi_struct *napi;
+
+	list_for_each_entry(napi, &dev->napi_list, dev_list)
+		napi_set_threaded(napi, false);
+}
+
+int dev_set_threaded(struct net_device *dev, bool threaded)
+{
+	struct napi_struct *napi;
+	int ret;
+
+	dev->threaded = threaded;
+	list_for_each_entry(napi, &dev->napi_list, dev_list) {
+		ret = napi_set_threaded(napi, threaded);
+		if (ret) {
+			/* Error occurred on one of the napi,
+			 * reset threaded mode on all napi.
+			 */
+			dev_disable_threaded_all(dev);
+			break;
+		}
+	}
+
+	return ret;
+}
+
 void netif_napi_add(struct net_device *dev, struct napi_struct *napi,
 		    int (*poll)(struct napi_struct *, int), int weight)
 {
diff --git a/net/core/net-sysfs.c b/net/core/net-sysfs.c
index 94fff0700bdd..3c9ef4692b3d 100644
--- a/net/core/net-sysfs.c
+++ b/net/core/net-sysfs.c
@@ -538,6 +538,68 @@ static ssize_t phys_switch_id_show(struct device *dev,
 }
 static DEVICE_ATTR_RO(phys_switch_id);
 
+static ssize_t threaded_show(struct device *dev,
+			     struct device_attribute *attr, char *buf)
+{
+	struct net_device *netdev = to_net_dev(dev);
+	struct napi_struct *n;
+	bool enabled;
+	int ret;
+
+	if (!rtnl_trylock())
+		return restart_syscall();
+
+	if (!dev_isalive(netdev)) {
+		ret = -EINVAL;
+		goto unlock;
+	}
+
+	if (list_empty(&netdev->napi_list)) {
+		ret = -EOPNOTSUPP;
+		goto unlock;
+	}
+
+	/* Only return true if all napi have threaded mode.
+	 * The inconsistency could happen when the device driver calls
+	 * napi_disable()/napi_enable() with dev->threaded set to true,
+	 * but napi_kthread_create() fails.
+	 * We return false in this case to remind the user that one or
+	 * more napi did not have threaded mode enabled properly.
+	 */
+	list_for_each_entry(n, &netdev->napi_list, dev_list) {
+		enabled = !!test_bit(NAPI_STATE_THREADED, &n->state);
+		if (!enabled)
+			break;
+	}
+
+	ret = sprintf(buf, fmt_dec, enabled);
+
+unlock:
+	rtnl_unlock();
+	return ret;
+}
+
+static int modify_napi_threaded(struct net_device *dev, unsigned long val)
+{
+	struct napi_struct *napi;
+	int ret;
+
+	if (list_empty(&dev->napi_list))
+		return -EOPNOTSUPP;
+
+	ret = dev_set_threaded(dev, !!val);
+
+	return ret;
+}
+
+static ssize_t threaded_store(struct device *dev,
+			      struct device_attribute *attr,
+			      const char *buf, size_t len)
+{
+	return netdev_store(dev, attr, buf, len, modify_napi_threaded);
+}
+static DEVICE_ATTR_RW(threaded);
+
 static struct attribute *net_class_attrs[] __ro_after_init = {
 	&dev_attr_netdev_group.attr,
 	&dev_attr_type.attr,
@@ -570,6 +632,7 @@ static struct attribute *net_class_attrs[] __ro_after_init = {
 	&dev_attr_proto_down.attr,
 	&dev_attr_carrier_up_count.attr,
 	&dev_attr_carrier_down_count.attr,
+	&dev_attr_threaded.attr,
 	NULL,
 };
 ATTRIBUTE_GROUPS(net_class);
-- 
2.29.2.684.gfbc64c5ab5-goog

Re: [PATCH net-next v5 2/3] net: implement threaded-able napi poll loop support

[Jakub Kicinski]

On Tue, 15 Dec 2020 17:25:14 -0800 Wei Wang wrote:
> +void napi_enable(struct napi_struct *n)
> +{
> +	bool locked = rtnl_is_locked();

Maybe you'll prove me wrong but I think this is never a correct
construct.

> +	BUG_ON(!test_bit(NAPI_STATE_SCHED, &n->state));
> +	smp_mb__before_atomic();
> +	clear_bit(NAPI_STATE_SCHED, &n->state);
> +	clear_bit(NAPI_STATE_NPSVC, &n->state);
> +	if (!locked)
> +		rtnl_lock();

Why do we need the lock? Can't we assume the caller of napi_enable()
has the sole ownership of the napi instance? Surely clearing the other
flags would be pretty broken as well, so the napi must had been
disabled when this is called by the driver.

> +	WARN_ON(napi_set_threaded(n, n->dev->threaded));
> +	if (!locked)
> +		rtnl_unlock();
> +}
> +EXPORT_SYMBOL(napi_enable);

Let's switch to RFC postings and get it in for 5.12 :(

Re: [PATCH net-next v5 2/3] net: implement threaded-able napi poll loop support

[Wei Wang]

On Tue, Dec 15, 2020 at 5:53 PM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Tue, 15 Dec 2020 17:25:14 -0800 Wei Wang wrote:
> > +void napi_enable(struct napi_struct *n)
> > +{
> > +     bool locked = rtnl_is_locked();
>
> Maybe you'll prove me wrong but I think this is never a correct
> construct.

Sorry, I don't get what you mean here.

>
> > +     BUG_ON(!test_bit(NAPI_STATE_SCHED, &n->state));
> > +     smp_mb__before_atomic();
> > +     clear_bit(NAPI_STATE_SCHED, &n->state);
> > +     clear_bit(NAPI_STATE_NPSVC, &n->state);
> > +     if (!locked)
> > +             rtnl_lock();
>
> Why do we need the lock? Can't we assume the caller of napi_enable()
> has the sole ownership of the napi instance? Surely clearing the other
> flags would be pretty broken as well, so the napi must had been
> disabled when this is called by the driver.
>

Hmm... OK. The reason I added this lock operation is that we have
ASSERT_RTNL() check in napi_set_threaded(), because it is necessary
from the net-sysfs path to grab rtnl lock when modifying the threaded
mode. Maybe it is not needed here.
And the reason I added a rtnl_is_locked() check is I found mlx driver
already holds rtnl lock before calling napi_enable(). Not sure about
other drivers though.

> > +     WARN_ON(napi_set_threaded(n, n->dev->threaded));
> > +     if (!locked)
> > +             rtnl_unlock();
> > +}
> > +EXPORT_SYMBOL(napi_enable);
>
> Let's switch to RFC postings and get it in for 5.12 :(

OK.

Re: [PATCH net-next v5 2/3] net: implement threaded-able napi poll loop support

[Jakub Kicinski]

On Tue, 15 Dec 2020 18:15:06 -0800 Wei Wang wrote:
> On Tue, Dec 15, 2020 at 5:53 PM Jakub Kicinski <kuba@kernel.org> wrote:
> > On Tue, 15 Dec 2020 17:25:14 -0800 Wei Wang wrote:  
> > > +void napi_enable(struct napi_struct *n)
> > > +{
> > > +     bool locked = rtnl_is_locked();  
> >
> > Maybe you'll prove me wrong but I think this is never a correct
> > construct.  
> 
> Sorry, I don't get what you mean here.

You're checking if the mutex is locked, not if the caller is holding
the mutex.

> > > +     BUG_ON(!test_bit(NAPI_STATE_SCHED, &n->state));
> > > +     smp_mb__before_atomic();
> > > +     clear_bit(NAPI_STATE_SCHED, &n->state);
> > > +     clear_bit(NAPI_STATE_NPSVC, &n->state);
> > > +     if (!locked)
> > > +             rtnl_lock();  
> >
> > Why do we need the lock? Can't we assume the caller of napi_enable()
> > has the sole ownership of the napi instance? Surely clearing the other
> > flags would be pretty broken as well, so the napi must had been
> > disabled when this is called by the driver.
> >  
> 
> Hmm... OK. The reason I added this lock operation is that we have
> ASSERT_RTNL() check in napi_set_threaded(), because it is necessary
> from the net-sysfs path to grab rtnl lock when modifying the threaded
> mode. Maybe it is not needed here.

Fair point, the sysfs write path could try to change the setting while
the NIC is starting.

> And the reason I added a rtnl_is_locked() check is I found mlx driver
> already holds rtnl lock before calling napi_enable(). Not sure about
> other drivers though.

I'd think most drivers will only mess around with napi under rtnl_lock.

> > > +     WARN_ON(napi_set_threaded(n, n->dev->threaded));
> > > +     if (!locked)
> > > +             rtnl_unlock();
> > > +}
> > > +EXPORT_SYMBOL(napi_enable);