[xdp-hints] Re: [PATCH bpf-next v5 05/17] bpf: Introduce device-bound XDP programs

[Martin KaFai Lau <martin.lau@linux.dev>]

On 12/20/22 2:20 PM, Stanislav Fomichev wrote:
> -int bpf_prog_offload_init(struct bpf_prog *prog, union bpf_attr *attr)
> +int bpf_prog_dev_bound_init(struct bpf_prog *prog, union bpf_attr *attr)
>   {
>   	struct bpf_offload_netdev *ondev;
>   	struct bpf_prog_offload *offload;
> @@ -199,7 +197,7 @@ int bpf_prog_offload_init(struct bpf_prog *prog, union bpf_attr *attr)
>   	    attr->prog_type != BPF_PROG_TYPE_XDP)
>   		return -EINVAL;
>   
> -	if (attr->prog_flags)
> +	if (attr->prog_flags & ~BPF_F_XDP_DEV_BOUND_ONLY)
>   		return -EINVAL;
>   
>   	offload = kzalloc(sizeof(*offload), GFP_USER);
> @@ -214,11 +212,23 @@ int bpf_prog_offload_init(struct bpf_prog *prog, union bpf_attr *attr)
>   	if (err)
>   		goto err_maybe_put;
>   
> +	prog->aux->offload_requested = !(attr->prog_flags & BPF_F_XDP_DEV_BOUND_ONLY);

Just noticed bpf_prog_dev_bound_init() takes BPF_PROG_TYPE_SCHED_CLS.  Not sure 
if there is device match check when attaching BPF_PROG_TYPE_SCHED_CLS.  If not, 
does it make sense to reject dev bound only BPF_PROG_TYPE_SCHED_CLS?

> +
>   	down_write(&bpf_devs_lock);
>   	ondev = bpf_offload_find_netdev(offload->netdev);
>   	if (!ondev) {
> -		err = -EINVAL;
> -		goto err_unlock;
> +		if (bpf_prog_is_offloaded(prog->aux)) {
> +			err = -EINVAL;
> +			goto err_unlock;
> +		}
> +
> +		/* When only binding to the device, explicitly
> +		 * create an entry in the hashtable.
> +		 */
> +		err = __bpf_offload_dev_netdev_register(NULL, offload->netdev);
> +		if (err)
> +			goto err_unlock;
> +		ondev = bpf_offload_find_netdev(offload->netdev);
>   	}
>   	offload->offdev = ondev->offdev;
>   	prog->aux->offload = offload;
> @@ -321,12 +331,41 @@ bpf_prog_offload_remove_insns(struct bpf_verifier_env *env, u32 off, u32 cnt)
>   	up_read(&bpf_devs_lock);
>   }
>   
> -void bpf_prog_offload_destroy(struct bpf_prog *prog)
> +static void __bpf_prog_dev_bound_destroy(struct bpf_prog *prog)
> +{
> +	struct bpf_prog_offload *offload = prog->aux->offload;
> +
> +	if (offload->dev_state)
> +		offload->offdev->ops->destroy(prog);
> +
> +	/* Make sure BPF_PROG_GET_NEXT_ID can't find this dead program */
> +	bpf_prog_free_id(prog, true);
> +
> +	kfree(offload);
> +	prog->aux->offload = NULL;
> +}
> +
> +void bpf_prog_dev_bound_destroy(struct bpf_prog *prog)
>   {
> +	struct bpf_offload_netdev *ondev;
> +	struct net_device *netdev;
> +
> +	rtnl_lock();
>   	down_write(&bpf_devs_lock);
> -	if (prog->aux->offload)
> -		__bpf_prog_offload_destroy(prog);
> +	if (prog->aux->offload) {
> +		list_del_init(&prog->aux->offload->offloads);
> +
> +		netdev = prog->aux->offload->netdev;

After saving the netdev, would it work to call __bpf_prog_offload_destroy() here 
instead of creating an almost identical __bpf_prog_dev_bound_destroy().  The 
idea is to call list_del_init() first but does not need the "offload" around to 
do the __bpf_offload_dev_netdev_unregister()?

> +		if (netdev) {

I am thinking offload->netdev cannot be NULL.  Did I overlook places that reset 
offload->netdev back to NULL?  eg. In bpf_prog_offload_info_fill_ns(), it is not 
checking offload->netdev.

> +			ondev = bpf_offload_find_netdev(netdev);

and ondev should not be NULL too?

I am trying to ensure my understanding that all offload->netdev and ondev should 
be protected by bpf_devs_lock.

> +			if (ondev && !ondev->offdev && list_empty(&ondev->progs))
> +				__bpf_offload_dev_netdev_unregister(NULL, netdev);
> +		}
> +
> +		__bpf_prog_dev_bound_destroy(prog);
> +	}
>   	up_write(&bpf_devs_lock);
> +	rtnl_unlock();
>   }