Re: [PATCH net-next v3] net: dqs: add NIC stall detector based on BQL

[Eric Dumazet]

On Wed, Feb 14, 2024 at 3:45 PM Breno Leitao <leitao@xxxxxxxxxx> wrote:
>
> On Tue, Feb 13, 2024 at 10:04:57AM -0800, Jakub Kicinski wrote:
> > On Tue, 13 Feb 2024 14:57:49 +0100 Eric Dumazet wrote:
> > > Please note that adding other sysfs entries is expensive for workloads
> > > creating/deleting netdev and netns often.
> > >
> > > I _think_ we should find a way for not creating
> > > /sys/class/net/<interface>/queues/tx-{Q}/byte_queue_limits  directory
> > > and files
> > > for non BQL enabled devices (like loopback !)
> >
> > We should try, see if anyone screams. We could use IFF_NO_QUEUE, and
> > NETIF_F_LLTX as a proxy for "device doesn't have a real queue so BQL
> > would be pointless"? Obviously better to annotate the drivers which
> > do have BQL support, but there's >50 of them on a quick count..
>
> Let me make sure I understand the suggestion above. We want to disable
> BQL completely for devices that has dev->features & NETIF_F_LLTX or
> dev->priv_flags & IFF_NO_QUEUE, right?
>
> Maybe we can add a ->enabled field in struct dql, and set it according
> to the features above. Then we can created the sysfs and process the dql
> operations based on that field. This should avoid some unnecessary calls
> also, if we are not display sysfs.
>
> Here is a very simple PoC to represent what I had in mind. Am I in the
> right direction?

No, this was really about sysfs entries (aka dql_group)

Partial patch would be:

diff --git a/net/core/net-sysfs.c b/net/core/net-sysfs.c
index a09d507c5b03d24a829bf7af0b7cf1e6a0bdb65a..094e3b2d78cca40d810b2fa3bd4393d22b30e6ad
100644
--- a/net/core/net-sysfs.c
+++ b/net/core/net-sysfs.c
@@ -1709,9 +1709,11 @@ static int netdev_queue_add_kobject(struct
net_device *dev, int index)
                goto err;

 #ifdef CONFIG_BQL
-       error = sysfs_create_group(kobj, &dql_group);
-       if (error)
-               goto err;
+       if (netdev_uses_bql(dev)) {
+               error = sysfs_create_group(kobj, &dql_group);
+               if (error)
+                       goto err;
+       }
 #endif

        kobject_uevent(kobj, KOBJ_ADD);
@@ -1734,7 +1736,8 @@ static int tx_queue_change_owner(struct
net_device *ndev, int index,
                return error;

 #ifdef CONFIG_BQL
-       error = sysfs_group_change_owner(kobj, &dql_group, kuid, kgid);
+       if (netdev_uses_bql(ndev))
+               error = sysfs_group_change_owner(kobj, &dql_group, kuid, kgid);
 #endif
        return error;
 }
@@ -1768,7 +1771,8 @@ netdev_queue_update_kobjects(struct net_device
*dev, int old_num, int new_num)
                if (!refcount_read(&dev_net(dev)->ns.count))
                        queue->kobj.uevent_suppress = 1;
 #ifdef CONFIG_BQL
-               sysfs_remove_group(&queue->kobj, &dql_group);
+               if (netdev_uses_bql(dev))
+                       sysfs_remove_group(&queue->kobj, &dql_group);
 #endif
                kobject_put(&queue->kobj);
        }