[OpenWrt-Devel] [PATCH] ath79: ag71xx: use netif_receive_skb_list on 4.19

Rosen Penev rosenp at gmail.com
Wed Jan 8 19:48:07 EST 2020



Sent from my iPhone

> On Jan 8, 2020, at 2:50 PM, Tom Psyborg <pozega.tomislav at gmail.com> wrote:
> 
> On 08/01/2020, Rosen Penev <rosenp at gmail.com> wrote:
>> From: Chuanhong Guo <gch981213 at gmail.com>
>> 
>> This new function make batch processing of network packets possible,
>> which slightly improves performance.
>> 
>> Signed-off-by: Chuanhong Guo <gch981213 at gmail.com>
>> Tested-by: Rosen Penev <rosenp at gmail.com>
>> ---
>> around a 20mbps improvement is measured on a TP-LINK Archer C7v2
>> .../net/ethernet/atheros/ag71xx/ag71xx_main.c | 20 ++++++++++++++++++-
>> 1 file changed, 19 insertions(+), 1 deletion(-)
>> 
>> diff --git
>> a/target/linux/ath79/files/drivers/net/ethernet/atheros/ag71xx/ag71xx_main.c
>> b/target/linux/ath79/files/drivers/net/ethernet/atheros/ag71xx/ag71xx_main.c
>> index 0924b81b92..8831a51acc 100644
>> ---
>> a/target/linux/ath79/files/drivers/net/ethernet/atheros/ag71xx/ag71xx_main.c
>> +++
>> b/target/linux/ath79/files/drivers/net/ethernet/atheros/ag71xx/ag71xx_main.c
>> @@ -1106,14 +1106,22 @@ static int ag71xx_rx_packets(struct ag71xx *ag, int
>> limit)
>>    unsigned int offset = ag->rx_buf_offset;
>>    int ring_mask = BIT(ring->order) - 1;
>>    int ring_size = BIT(ring->order);
>> +#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,19,0))
>> +    struct list_head rx_list;
>> +    struct sk_buff *next;
>> +#else
>>    struct sk_buff_head queue;
>> +#endif
>>    struct sk_buff *skb;
>>    int done = 0;
>> 
>>    DBG("%s: rx packets, limit=%d, curr=%u, dirty=%u\n",
>>            dev->name, limit, ring->curr, ring->dirty);
>> -
>> +#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,19,0))
>> +    INIT_LIST_HEAD(&rx_list);
>> +#else
>>    skb_queue_head_init(&queue);
>> +#endif
>> 
>>    while (done < limit) {
>>        unsigned int i = ring->curr & ring_mask;
>> @@ -1155,7 +1163,11 @@ static int ag71xx_rx_packets(struct ag71xx *ag, int
>> limit)
>>        } else {
>>            skb->dev = dev;
>>            skb->ip_summed = CHECKSUM_NONE;
>> +#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,19,0))
>> +            list_add_tail(&skb->list, &rx_list);
>> +#else
>>            __skb_queue_tail(&queue, skb);
>> +#endif
>>        }
>> 
>> next:
>> @@ -1167,10 +1179,16 @@ next:
>> 
>>    ag71xx_ring_rx_refill(ag);
>> 
>> +#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,19,0))
>> +    list_for_each_entry_safe(skb, next, &rx_list, list)
>> +        skb->protocol = eth_type_trans(skb, dev);
>> +    netif_receive_skb_list(&rx_list);
>> +#else
>>    while ((skb = __skb_dequeue(&queue)) != NULL) {
>>        skb->protocol = eth_type_trans(skb, dev);
>>        netif_receive_skb(skb);
>>    }
>> +#endif
>> 
>>    DBG("%s: rx finish, curr=%u, dirty=%u, done=%d\n",
>>        dev->name, ring->curr, ring->dirty, done);
>> --
>> 2.24.1
>> 
>> 
>> _______________________________________________
>> openwrt-devel mailing list
>> openwrt-devel at lists.openwrt.org
>> https://lists.openwrt.org/mailman/listinfo/openwrt-devel
>> 
> 
> An identical email has been submitted yesterday. Why are you sending it again?
Mistake on my end. I removed the duplicate from patchwork.
_______________________________________________
openwrt-devel mailing list
openwrt-devel at lists.openwrt.org
https://lists.openwrt.org/mailman/listinfo/openwrt-devel


More information about the openwrt-devel mailing list