Re: [PATCH net-next 1/7] net: ethtool: netlink: Allow per-netdevice DUMP operations

[Simon Horman <horms@kernel.org>]

On Wed, Mar 05, 2025 at 03:19:31PM +0100, Maxime Chevallier wrote:
> We have a number of netlink commands in the ethnl family that may have
> multiple objects to dump even for a single net_device, including :
> 
>  - PLCA, PSE-PD, phy: one message per PHY device
>  - tsinfo: one message per timestamp source (netdev + phys)
>  - rss: One per RSS context
> 
> To get this behaviour, these netlink commands need to roll a custom
> ->dumpit().
> 
> To prepare making per-netdev DUMP more generic in ethnl, introduce a
> member in the ethnl ops to indicate if a given command may allow
> pernetdev DUMPs (also referred to as filtered DUMPs).
> 
> Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
> ---
>  net/ethtool/netlink.c | 45 ++++++++++++++++++++++++++++---------------
>  net/ethtool/netlink.h |  1 +
>  2 files changed, 30 insertions(+), 16 deletions(-)
> 
> diff --git a/net/ethtool/netlink.c b/net/ethtool/netlink.c
> index 734849a57369..0815b28ba32f 100644
> --- a/net/ethtool/netlink.c
> +++ b/net/ethtool/netlink.c
> @@ -578,21 +578,34 @@ static int ethnl_default_dumpit(struct sk_buff *skb,
>  	int ret = 0;
>  
>  	rcu_read_lock();
> -	for_each_netdev_dump(net, dev, ctx->pos_ifindex) {
> -		dev_hold(dev);
> +	if (ctx->req_info->dev) {
> +		dev = ctx->req_info->dev;
>  		rcu_read_unlock();
> -
> -		ret = ethnl_default_dump_one(skb, dev, ctx, genl_info_dump(cb));
> -
> +		/* Filtered DUMP request targeted to a single netdev. We already
> +		 * hold a ref to the netdev from ->start()
> +		 */
> +		ret = ethnl_default_dump_one_dev(skb, dev, ctx,
> +						 genl_info_dump(cb));

Hi Maxime,

ethnl_default_dump_one_dev() is called here but it doesn't exist
until the following patch is applied, which breaks bisection.

>  		rcu_read_lock();
> -		dev_put(dev);
> -
> -		if (ret < 0 && ret != -EOPNOTSUPP) {
> -			if (likely(skb->len))
> -				ret = skb->len;
> -			break;
> +		netdev_put(ctx->req_info->dev, &ctx->req_info->dev_tracker);
> +	} else {
> +		for_each_netdev_dump(net, dev, ctx->pos_ifindex) {
> +			dev_hold(dev);
> +			rcu_read_unlock();
> +
> +			ret = ethnl_default_dump_one(skb, dev, ctx,
> +						     genl_info_dump(cb));
> +
> +			rcu_read_lock();
> +			dev_put(dev);
> +
> +			if (ret < 0 && ret != -EOPNOTSUPP) {
> +				if (likely(skb->len))
> +					ret = skb->len;
> +				break;
> +			}
> +			ret = 0;
>  		}
> -		ret = 0;
>  	}
>  	rcu_read_unlock();
>  

...

> diff --git a/net/ethtool/netlink.h b/net/ethtool/netlink.h
> index ec6ab5443a6f..4db27182741f 100644
> --- a/net/ethtool/netlink.h
> +++ b/net/ethtool/netlink.h
> @@ -388,6 +388,7 @@ struct ethnl_request_ops {
>  	unsigned int		req_info_size;
>  	unsigned int		reply_data_size;
>  	bool			allow_nodev_do;
> +	bool			allow_pernetdev_dump;

nit: allow_pernetdev_dump should also be added to the Kernel doc for
     struct ethnl_request_ops

     Flagged by ./scripts/kernel-doc -none

     There also appear to be similar minor issues with subsequent
     patches in this series.

>  	u8			set_ntf_cmd;
>  
>  	int (*parse_request)(struct ethnl_req_info *req_info,
> -- 
> 2.48.1
>