Re: [PATCH ref0] net: add Faraday FTMAC100 10/100 Ethernet driver

Hi Eric,

On Fri, Feb 25, 2011 at 10:32 AM, Po-Yu Chuang [off-list ref] wrote:

On Fri, Feb 25, 2011 at 1:48 AM, Eric Dumazet [off-list ref] wrote:

quoted

Le jeudi 24 février 2011 à 18:39 +0100, Eric Dumazet a écrit :

quoted

Le jeudi 24 février 2011 à 17:29 +0800, Po-Yu Chuang a écrit :

quoted

From: Po-Yu Chuang <redacted>
+
+static bool ftmac100_rx_packet(struct ftmac100 *priv, int *processed)
+{
+   struct net_device *netdev = priv->netdev;
+   struct ftmac100_rxdes *rxdes;
+   struct sk_buff *skb;
+   struct page *page;
+   dma_addr_t map;
+   int length;
+
+   rxdes = ftmac100_rx_locate_first_segment(priv);
+   if (!rxdes)
+           return false;
+
+   if (unlikely(ftmac100_rx_packet_error(priv, rxdes))) {
+           ftmac100_rx_drop_packet(priv);
+           return true;
+   }
+
+   /*
+    * It is impossible to get multi-segment packets
+    * because we always provide big enough receive buffers.
+    */
+   if (unlikely(!ftmac100_rxdes_last_segment(rxdes)))
+           BUG();
+
+   /* start processing */
+   skb = netdev_alloc_skb_ip_align(netdev, ETH_HLEN);

Oh I see... You should allocate a bigger head (say... 128 bytes)

And copy in it up to 128 bytes of first part... this to avoid upper
stack to reallocate skb head (because IP/TCP processing need to get
their headers in skb head)

Take a look at drivers/net/niu.c :

#define RX_SKB_ALLOC_SIZE   128 + NET_IP_ALIGN

static int niu_process_rx_pkt(...)
{
       ...
       skb = netdev_alloc_skb(np->dev, RX_SKB_ALLOC_SIZE);
       ...
       while (1) {
               ...
               niu_rx_skb_append(skb, page, off, append_size);
       }
}

Oh I got it.

I will try this and redo the benchmarking.

It's a little faster than v5 now. Thanks.
I will submit the current version later.

One more question just curious, why 128 bytes?

best regards,
Po-Yu Chuang