netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jakub Kicinski <kuba@kernel.org>
To: David Wei <dw@davidwei.uk>
Cc: Jiri Pirko <jiri@resnulli.us>,
	Sabrina Dubroca <sd@queasysnail.net>,
	netdev@vger.kernel.org, "David S. Miller" <davem@davemloft.net>,
	Eric Dumazet <edumazet@google.com>,
	Paolo Abeni <pabeni@redhat.com>
Subject: Re: [PATCH net-next v6 1/4] netdevsim: allow two netdevsim ports to be connected
Date: Thu, 25 Jan 2024 18:24:03 -0800	[thread overview]
Message-ID: <20240125182403.13c4475b@kernel.org> (raw)
In-Reply-To: <20240126012357.535494-2-dw@davidwei.uk>

On Thu, 25 Jan 2024 17:23:54 -0800 David Wei wrote:
> diff --git a/drivers/net/netdevsim/bus.c b/drivers/net/netdevsim/bus.c
> index bcbc1e19edde..be8ac2e60c69 100644
> --- a/drivers/net/netdevsim/bus.c
> +++ b/drivers/net/netdevsim/bus.c
> @@ -232,9 +232,81 @@ del_device_store(const struct bus_type *bus, const char *buf, size_t count)
>  }
>  static BUS_ATTR_WO(del_device);
>  
> +static ssize_t link_device_store(const struct bus_type *bus, const char *buf, size_t count)
> +{
> +	unsigned int netnsid_a, netnsid_b, ifidx_a, ifidx_b;
> +	struct netdevsim *nsim_a, *nsim_b;
> +	struct net_device *dev_a, *dev_b;
> +	struct net *ns_a, *ns_b;
> +	int err;
> +
> +	err = sscanf(buf, "%u %u %u %u", &netnsid_a, &ifidx_a, &netnsid_b, &ifidx_b);

I'd go for "%u:%u %u:%u" to make the 'grouping' of netns and ifindex
more obvious. But no strong feelings.

> +	if (err != 4) {
> +		pr_err("Format for linking two devices is \"netnsid_a ifidx_a netnsid_b ifidx_b\" (uint uint unit uint).\n");
> +		return -EINVAL;
> +	}
> +
> +	err = -EINVAL;
> +	rtnl_lock();
> +	ns_a = get_net_ns_by_id(current->nsproxy->net_ns, netnsid_a);
> +	if (!ns_a) {
> +		pr_err("Could not find netns with id: %d\n", netnsid_a);
> +		goto out_unlock_rtnl;
> +	}
> +
> +	dev_a = dev_get_by_index(ns_a, ifidx_a);

since you're under rtnl_lock you can use __get_device_by_index(),
it doesn't increase the refcount so you won't have to worry about
releasing it.

> +	if (!dev_a) {
> +		pr_err("Could not find device with ifindex %d in netnsid %d\n", ifidx_a, netnsid_a);
> +		goto out_put_netns_a;
> +	}
> +
> +	if (!netdev_is_nsim(dev_a)) {
> +		pr_err("Device with ifindex %d in netnsid %d is not a netdevsim\n", ifidx_a, netnsid_a);
> +		goto out_put_dev_a;
> +	}
> +
> +	ns_b = get_net_ns_by_id(current->nsproxy->net_ns, netnsid_b);
> +	if (!ns_b) {
> +		pr_err("Could not find netns with id: %d\n", netnsid_b);
> +		goto out_put_dev_a;
> +	}
> +
> +	dev_b = dev_get_by_index(ns_b, ifidx_b);
> +	if (!dev_b) {
> +		pr_err("Could not find device with ifindex %d in netnsid %d\n", ifidx_b, netnsid_b);
> +		goto out_put_netns_b;
> +	}
> +
> +	if (!netdev_is_nsim(dev_b)) {
> +		pr_err("Device with ifindex %d in netnsid %d is not a netdevsim\n", ifidx_b, netnsid_b);
> +		goto out_put_dev_b;
> +	}
> +
> +	err = 0;
> +	nsim_a = netdev_priv(dev_a);
> +	nsim_b = netdev_priv(dev_b);
> +	rcu_assign_pointer(nsim_a->peer, nsim_b);
> +	rcu_assign_pointer(nsim_b->peer, nsim_a);

Shouldn't we check if peer is NULL? Otherwise we can get into weird
situations where we link A<>B then B<>C and then the pointers look like
this A->B<>C. When B gets freed A's pointer won't get cleared.

> +out_put_dev_b:
> +	dev_put(dev_b);
> +out_put_netns_b:
> +	put_net(ns_b);
> +out_put_dev_a:
> +	dev_put(dev_a);
> +out_put_netns_a:
> +	put_net(ns_a);
> +out_unlock_rtnl:
> +	rtnl_unlock();
> +
> +	return !err ? count : err;
> +}
> +static BUS_ATTR_WO(link_device);

  reply	other threads:[~2024-01-26  2:24 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-01-26  1:23 [PATCH net-next v6 0/4] netdevsim: link and forward skbs between ports David Wei
2024-01-26  1:23 ` [PATCH net-next v6 1/4] netdevsim: allow two netdevsim ports to be connected David Wei
2024-01-26  2:24   ` Jakub Kicinski [this message]
2024-01-26 18:54     ` David Wei
2024-01-26 19:17       ` Jakub Kicinski
2024-01-26  1:23 ` [PATCH net-next v6 2/4] netdevsim: forward skbs from one connected port to another David Wei
2024-01-26  2:26   ` Jakub Kicinski
2024-01-26 18:55     ` David Wei
2024-01-26  1:23 ` [PATCH net-next v6 3/4] netdevsim: add selftest for forwarding skb between connected ports David Wei
2024-01-26  1:23 ` [PATCH net-next v6 4/4] netdevsim: add Makefile for selftests David Wei

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20240125182403.13c4475b@kernel.org \
    --to=kuba@kernel.org \
    --cc=davem@davemloft.net \
    --cc=dw@davidwei.uk \
    --cc=edumazet@google.com \
    --cc=jiri@resnulli.us \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    --cc=sd@queasysnail.net \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).