Received-SPF: pass (google.com: domain of linux-kernel+bounces-95534-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.48.161 as permitted sender) client-ip=147.75.48.161;
Subject: Re: [RFC PATCH net-next v1 1/2] net: mirror skb frag ref/unref
 helpers
To: Mina Almasry <almasrymina@google.com>, <netdev@vger.kernel.org>,
	<linux-kernel@vger.kernel.org>, <linux-rdma@vger.kernel.org>
CC: Mirko Lindner <mlindner@marvell.com>, Stephen Hemminger
	<stephen@networkplumber.org>, "David S. Miller" <davem@davemloft.net>, Eric
 Dumazet <edumazet@google.com>, Jakub Kicinski <kuba@kernel.org>, Paolo Abeni
	<pabeni@redhat.com>, Tariq Toukan <tariqt@nvidia.com>, Boris Pismenny
	<borisp@nvidia.com>, John Fastabend <john.fastabend@gmail.com>, Dragos
 Tatulea <dtatulea@nvidia.com>
References: <20240306235922.282781-1-almasrymina@google.com>
 <20240306235922.282781-2-almasrymina@google.com>
From: Yunsheng Lin <linyunsheng@huawei.com>
Message-ID: <8ef17e46-3d14-e010-c721-6fca3d57f78f@huawei.com>
Date: Thu, 7 Mar 2024 20:28:53 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:52.0) Gecko/20100101
 Thunderbird/52.2.0
Precedence: bulk
MIME-Version: 1.0
In-Reply-To: <20240306235922.282781-2-almasrymina@google.com>
Content-Type: text/plain; charset="utf-8"
Content-Language: en-US
Content-Transfer-Encoding: 7bit

On 2024/3/7 7:59, Mina Almasry wrote:

..

>  
>  int skb_pp_cow_data(struct page_pool *pool, struct sk_buff **pskb,
> diff --git a/net/core/skbuff.c b/net/core/skbuff.c
> index 1f918e602bc4..6d234faa9d9e 100644
> --- a/net/core/skbuff.c
> +++ b/net/core/skbuff.c
> @@ -1006,6 +1006,21 @@ int skb_cow_data_for_xdp(struct page_pool *pool, struct sk_buff **pskb,
>  EXPORT_SYMBOL(skb_cow_data_for_xdp);
>  
>  #if IS_ENABLED(CONFIG_PAGE_POOL)
> +bool napi_pp_get_page(struct page *page)
> +{
> +
> +	struct page *head_page;
> +
> +	head_page = compound_head(page);
> +
> +	if (!is_pp_page(page))

I would use the head_page for is_pp_page(), I am not sure it
matters that much, but I believe it is the precedent.

Maybe do the below and remove head_page varible:
page = compound_head(page);

> +		return false;
> +
> +	page_pool_ref_page(head_page);
> +	return true;
> +}
> +EXPORT_SYMBOL(napi_pp_get_page);
> +

..

> -
>  static void skb_kfree_head(void *head, unsigned int end_offset)
>  {
>  	if (end_offset == SKB_SMALL_HEAD_HEADROOM)
> @@ -4199,7 +4183,7 @@ int skb_shift(struct sk_buff *tgt, struct sk_buff *skb, int shiftlen)
>  			to++;
>  
>  		} else {
> -			__skb_frag_ref(fragfrom);
> +			__skb_frag_ref(fragfrom, skb->pp_recycle);
>  			skb_frag_page_copy(fragto, fragfrom);
>  			skb_frag_off_copy(fragto, fragfrom);
>  			skb_frag_size_set(fragto, todo);
> @@ -4849,7 +4833,7 @@ struct sk_buff *skb_segment(struct sk_buff *head_skb,
>  			}
>  
>  			*nskb_frag = (i < 0) ? skb_head_frag_to_page_desc(frag_skb) : *frag;
> -			__skb_frag_ref(nskb_frag);
> +			__skb_frag_ref(nskb_frag, nskb->pp_recycle);
>  			size = skb_frag_size(nskb_frag);
>  
>  			if (pos < offset) {
> @@ -5980,10 +5964,8 @@ bool skb_try_coalesce(struct sk_buff *to, struct sk_buff *from,
>  	/* if the skb is not cloned this does nothing
>  	 * since we set nr_frags to 0.
>  	 */
> -	if (skb_pp_frag_ref(from)) {
I guess it worth mentioning that skb->pp_recycle is only checked once,
and skb->pp_recycle is checked for every frag after this patch.

> -		for (i = 0; i < from_shinfo->nr_frags; i++)
> -			__skb_frag_ref(&from_shinfo->frags[i]);
> -	}
> +	for (i = 0; i < from_shinfo->nr_frags; i++)
> +		__skb_frag_ref(&from_shinfo->frags[i], from->pp_recycle);
>  
>  	to->truesize += delta;
>  	to->len += len;
>