netdev - Re: [PATCH net-next v2 4/9] xsk: extend xsk_build_skb() to support passing an already allocated skb

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <aKzaVRQeeSuH155P@boxer>
Date: Mon, 25 Aug 2025 23:49:09 +0200
From: Maciej Fijalkowski <maciej.fijalkowski@...el.com>
To: Jason Xing <kerneljasonxing@...il.com>
CC: <davem@...emloft.net>, <edumazet@...gle.com>, <kuba@...nel.org>,
	<pabeni@...hat.com>, <bjorn@...nel.org>, <magnus.karlsson@...el.com>,
	<jonathan.lemon@...il.com>, <sdf@...ichev.me>, <ast@...nel.org>,
	<daniel@...earbox.net>, <hawk@...nel.org>, <john.fastabend@...il.com>,
	<horms@...nel.org>, <andrew+netdev@...n.ch>, <bpf@...r.kernel.org>,
	<netdev@...r.kernel.org>, Jason Xing <kernelxing@...cent.com>
Subject: Re: [PATCH net-next v2 4/9] xsk: extend xsk_build_skb() to support
 passing an already allocated skb

On Mon, Aug 25, 2025 at 09:53:37PM +0800, Jason Xing wrote:
> From: Jason Xing <kernelxing@...cent.com>
> 
> Batch xmit mode needs to allocate and build skbs at one time. To avoid
> reinvent the wheel, use xsk_build_skb() as the second half process of
> the whole initialization of each skb.
> 
> The original xsk_build_skb() itself allocates a new skb by calling
> sock_alloc_send_skb whether in copy mode or zerocopy mode. Add a new
> parameter allocated skb to let other callers to pass an already
> allocated skb to support later xmit batch feature. At that time,
> another building skb function will generate a new skb and pass it to
> xsk_build_skb() to finish the rest of building process, like
> initializing structures and copying data.

are you saying you were able to avoid sock_alloc_send_skb() calls for
batching approach and your socket memory accounting problems disappeared?

> 
> Signed-off-by: Jason Xing <kernelxing@...cent.com>
> ---
>  include/net/xdp_sock.h |  4 ++++
>  net/xdp/xsk.c          | 23 ++++++++++++++++-------
>  2 files changed, 20 insertions(+), 7 deletions(-)
> 
> diff --git a/include/net/xdp_sock.h b/include/net/xdp_sock.h
> index c2b05268b8ad..cbba880c27c3 100644
> --- a/include/net/xdp_sock.h
> +++ b/include/net/xdp_sock.h
> @@ -123,6 +123,10 @@ struct xsk_tx_metadata_ops {
>  	void	(*tmo_request_launch_time)(u64 launch_time, void *priv);
>  };
>  
> +
> +struct sk_buff *xsk_build_skb(struct xdp_sock *xs,
> +			      struct sk_buff *allocated_skb,
> +			      struct xdp_desc *desc);

why do you export this?

>  #ifdef CONFIG_XDP_SOCKETS
>  
>  int xsk_generic_rcv(struct xdp_sock *xs, struct xdp_buff *xdp);
> diff --git a/net/xdp/xsk.c b/net/xdp/xsk.c
> index 173ad49379c3..213d6100e405 100644
> --- a/net/xdp/xsk.c
> +++ b/net/xdp/xsk.c
> @@ -605,6 +605,7 @@ static void xsk_drop_skb(struct sk_buff *skb)
>  }
>  
>  static struct sk_buff *xsk_build_skb_zerocopy(struct xdp_sock *xs,
> +					      struct sk_buff *allocated_skb,
>  					      struct xdp_desc *desc)
>  {
>  	struct xsk_buff_pool *pool = xs->pool;
> @@ -618,7 +619,10 @@ static struct sk_buff *xsk_build_skb_zerocopy(struct xdp_sock *xs,
>  	if (!skb) {
>  		hr = max(NET_SKB_PAD, L1_CACHE_ALIGN(xs->dev->needed_headroom));
>  
> -		skb = sock_alloc_send_skb(&xs->sk, hr, 1, &err);
> +		if (!allocated_skb)
> +			skb = sock_alloc_send_skb(&xs->sk, hr, 1, &err);
> +		else
> +			skb = allocated_skb;
>  		if (unlikely(!skb))
>  			return ERR_PTR(err);
>  
> @@ -657,8 +661,9 @@ static struct sk_buff *xsk_build_skb_zerocopy(struct xdp_sock *xs,
>  	return skb;
>  }
>  
> -static struct sk_buff *xsk_build_skb(struct xdp_sock *xs,
> -				     struct xdp_desc *desc)
> +struct sk_buff *xsk_build_skb(struct xdp_sock *xs,
> +			      struct sk_buff *allocated_skb,
> +			      struct xdp_desc *desc)
>  {
>  	struct xsk_tx_metadata *meta = NULL;
>  	struct net_device *dev = xs->dev;
> @@ -667,7 +672,7 @@ static struct sk_buff *xsk_build_skb(struct xdp_sock *xs,
>  	int err;
>  
>  	if (dev->priv_flags & IFF_TX_SKB_NO_LINEAR) {
> -		skb = xsk_build_skb_zerocopy(xs, desc);
> +		skb = xsk_build_skb_zerocopy(xs, allocated_skb, desc);
>  		if (IS_ERR(skb)) {
>  			err = PTR_ERR(skb);
>  			goto free_err;
> @@ -683,8 +688,12 @@ static struct sk_buff *xsk_build_skb(struct xdp_sock *xs,
>  			first_frag = true;
>  
>  			hr = max(NET_SKB_PAD, L1_CACHE_ALIGN(dev->needed_headroom));
> -			tr = dev->needed_tailroom;
> -			skb = sock_alloc_send_skb(&xs->sk, hr + len + tr, 1, &err);
> +			if (!allocated_skb) {
> +				tr = dev->needed_tailroom;
> +				skb = sock_alloc_send_skb(&xs->sk, hr + len + tr, 1, &err);
> +			} else {
> +				skb = allocated_skb;
> +			}
>  			if (unlikely(!skb))
>  				goto free_err;
>  
> @@ -818,7 +827,7 @@ static int __xsk_generic_xmit(struct sock *sk)
>  			goto out;
>  		}
>  
> -		skb = xsk_build_skb(xs, &desc);
> +		skb = xsk_build_skb(xs, NULL, &desc);
>  		if (IS_ERR(skb)) {
>  			err = PTR_ERR(skb);
>  			if (err != -EOVERFLOW)
> -- 
> 2.41.3
>