Message-ID: <6d7386ba-0012-4f13-b31b-fcd7a15f173e@meta.com>
Date: Thu, 29 Jan 2026 09:48:42 -0800
From: Vishwanath Seshagiri <vishs@...a.com>
To: "Michael S. Tsirkin" <mst@...hat.com>
Cc: Jason Wang <jasowang@...hat.com>, Xuan Zhuo <xuanzhuo@...ux.alibaba.com>,
        Eugenio Pérez <eperezma@...hat.com>,
        Andrew Lunn <andrew+netdev@...n.ch>,
        "David S . Miller"
 <davem@...emloft.net>,
        Eric Dumazet <edumazet@...gle.com>, Jakub Kicinski <kuba@...nel.org>,
        Paolo Abeni <pabeni@...hat.com>, David Wei <dw@...idwei.uk>,
        netdev@...r.kernel.org, virtualization@...ts.linux.dev,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2 net-next 1/2] virtio_net: add page_pool support for
 buffer allocation

On 1/28/26 10:30 PM, Michael S. Tsirkin wrote:
> On Wed, Jan 28, 2026 at 01:20:30PM -0800, Vishwanath Seshagiri wrote:
>> Use page_pool for RX buffer allocation in mergeable and small buffer
>> modes to enable page recycling and avoid repeated page allocator calls.
>> skb_mark_for_recycle() enables page reuse in the network stack.
>>
>> Big packets mode is unchanged because it uses page->private for linked
>> list chaining of multiple pages per buffer, which conflicts with
>> page_pool's internal use of page->private.
>>
>> Implement conditional DMA premapping using virtqueue_dma_dev():
>> - When non-NULL (vhost, virtio-pci): use PP_FLAG_DMA_MAP with page_pool
>>    handling DMA mapping, submit via virtqueue_add_inbuf_premapped()
>> - When NULL (VDUSE, direct physical): page_pool handles allocation only,
>>    submit via virtqueue_add_inbuf_ctx()
>>
>> This preserves the DMA premapping optimization from commit 31f3cd4e5756b
>> ("virtio-net: rq submits premapped per-buffer") while adding page_pool
>> support as a prerequisite for future zero-copy features (devmem TCP,
>> io_uring ZCRX).
>>
>> Page pools are created in probe and destroyed in remove (not open/close),
>> following existing driver behavior where RX buffers remain in virtqueues
>> across interface state changes.
>>
>> The rx_mode_work_enabled flag prevents virtnet_rx_mode_work() from
>> sending control virtqueue commands while ndo_close is tearing down
>> device state, avoiding virtqueue corruption during concurrent operations.
>>
>> Signed-off-by: Vishwanath Seshagiri <vishs@...a.com>
>> ---
>>   drivers/net/Kconfig      |   1 +
>>   drivers/net/virtio_net.c | 353 ++++++++++++++++++++++-----------------
>>   2 files changed, 203 insertions(+), 151 deletions(-)
>>
>> diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
>> index ac12eaf11755..f1e6b6b0a86f 100644
>> --- a/drivers/net/Kconfig
>> +++ b/drivers/net/Kconfig
>> @@ -450,6 +450,7 @@ config VIRTIO_NET
>>   	depends on VIRTIO
>>   	select NET_FAILOVER
>>   	select DIMLIB
>> +	select PAGE_POOL
>>   	help
>>   	  This is the virtual network driver for virtio.  It can be used with
>>   	  QEMU based VMMs (like KVM or Xen).  Say Y or M.
>> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
>> index db88dcaefb20..df2a5fc5187e 100644
>> --- a/drivers/net/virtio_net.c
>> +++ b/drivers/net/virtio_net.c
>> @@ -26,6 +26,7 @@
>>   #include <net/netdev_rx_queue.h>
>>   #include <net/netdev_queues.h>
>>   #include <net/xdp_sock_drv.h>
>> +#include <net/page_pool/helpers.h>
>>   
>>   static int napi_weight = NAPI_POLL_WEIGHT;
>>   module_param(napi_weight, int, 0444);
>> @@ -359,6 +360,11 @@ struct receive_queue {
>>   	/* Page frag for packet buffer allocation. */
>>   	struct page_frag alloc_frag;
>>   
>> +	struct page_pool *page_pool;
>> +
>> +	/* True if page_pool handles DMA mapping via PP_FLAG_DMA_MAP */
>> +	bool use_page_pool_dma;
>> +
>>   	/* RX: fragments + linear part + virtio header */
>>   	struct scatterlist sg[MAX_SKB_FRAGS + 2];
>>   
>> @@ -521,11 +527,13 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
>>   			       struct virtnet_rq_stats *stats);
>>   static void virtnet_receive_done(struct virtnet_info *vi, struct receive_queue *rq,
>>   				 struct sk_buff *skb, u8 flags);
>> -static struct sk_buff *virtnet_skb_append_frag(struct sk_buff *head_skb,
>> +static struct sk_buff *virtnet_skb_append_frag(struct receive_queue *rq,
>> +					       struct sk_buff *head_skb,
>>   					       struct sk_buff *curr_skb,
>>   					       struct page *page, void *buf,
>>   					       int len, int truesize);
>>   static void virtnet_xsk_completed(struct send_queue *sq, int num);
>> +static void free_unused_bufs(struct virtnet_info *vi);
>>   
>>   enum virtnet_xmit_type {
>>   	VIRTNET_XMIT_TYPE_SKB,
>> @@ -706,15 +714,21 @@ static struct page *get_a_page(struct receive_queue *rq, gfp_t gfp_mask)
>>   	return p;
>>   }
>>   
>> +static void virtnet_put_page(struct receive_queue *rq, struct page *page,
>> +			     bool allow_direct)
>> +{
>> +	page_pool_put_page(rq->page_pool, page, -1, allow_direct);
>> +}
>> +
>>   static void virtnet_rq_free_buf(struct virtnet_info *vi,
>>   				struct receive_queue *rq, void *buf)
>>   {
>>   	if (vi->mergeable_rx_bufs)
>> -		put_page(virt_to_head_page(buf));
>> +		virtnet_put_page(rq, virt_to_head_page(buf), false);
>>   	else if (vi->big_packets)
>>   		give_pages(rq, buf);
>>   	else
>> -		put_page(virt_to_head_page(buf));
>> +		virtnet_put_page(rq, virt_to_head_page(buf), false);
>>   }
> 
> 
> what I dislike here is how big_packets mode still pokes
> at give_pages but other modes use the page pool.
> 
> Given all modes operate with struct page it's hard to
> shake the feeling we could be trying to put a page
> we did not get from the pool back into the pool,
> or vice versa.

The allocation and free paths are symmetric for each of the modes.
Although, I made an error in the give_pages() path of big buffers.
That being said, the dual system approach is fragile. and I excluded
big_packets because it chains via page->private, which conflicts with
page_pool's internal use of page->private.

Would you prefer I extend the page_pool to big_packets in v3 or
followup? This would require building an alternate chaining mechanism,
which conflicts with page_pool's internal use of page->private.

> 
> 
> 
>>   
>>   static void enable_rx_mode_work(struct virtnet_info *vi)
>> @@ -877,9 +891,6 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi,
>>   		if (unlikely(!skb))
>>   			return NULL;
>>   
>> -		page = (struct page *)page->private;
>> -		if (page)
>> -			give_pages(rq, page);
>>   		goto ok;
>>   	}
>>
> 
> For example above you did not touch give_pages, here
> you are ripping out give_pages. Superficially, weird.

This was done in error, I will fix it in v3.

> 
> 
> I ask myself whether page pool is not better than the
> homegrown linked list that give_pages uses, anyway.

Extending page pool support is not only for performance, but
it is also needed for future zero copy features across RX modes.

> 
> Will need some perf testing though.

If you want me to run specific performance tests, I can add it in v3.

> 

Powered by blists - more mailing lists