net: make GRO aware of skb->head_frag

GRO can check if skb to be merged has its skb->head mapped to a page
fragment, instead of a kmalloc() area.

We 'upgrade' skb->head as a fragment in itself

This avoids the frag_list fallback, and permits to build true GRO skb
(one sk_buff and up to 16 fragments), using less memory.

This reduces number of cache misses when user makes its copy, since a
single sk_buff is fetched.

This is a followup of patch "net: allow skb->head to be a page fragment"

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Maciej Żenczykowski <maze@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Tom Herbert <therbert@google.com>
Cc: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Cc: Ben Hutchings <bhutchings@solarflare.com>
Cc: Matt Carlson <mcarlson@broadcom.com>
Cc: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

4 files changed, 33 insertions, 2 deletions

diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index e0b70e961e6..7f377fb8b52 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -1509,6 +1509,8 @@ struct napi_gro_cb {
 
 	/* Free the skb? */
 	int free;
+#define NAPI_GRO_FREE		  1
+#define NAPI_GRO_FREE_STOLEN_HEAD 2
 };
 
 #define NAPI_GRO_CB(skb) ((struct napi_gro_cb *)(skb)->cb)
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 9d28a22a855..2c75e98953b 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -561,6 +561,7 @@ static inline struct rtable *skb_rtable(const struct sk_buff *skb)
 extern void kfree_skb(struct sk_buff *skb);
 extern void consume_skb(struct sk_buff *skb);
 extern void	       __kfree_skb(struct sk_buff *skb);
+extern struct kmem_cache *skbuff_head_cache;
 extern struct sk_buff *__alloc_skb(unsigned int size,
 				   gfp_t priority, int fclone, int node);
 extern struct sk_buff *build_skb(void *data, unsigned int frag_size);
diff --git a/net/core/dev.c b/net/core/dev.c
index 501f3cc703d..a2be59fe6ab 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -3546,7 +3546,10 @@ gro_result_t napi_skb_finish(gro_result_t ret, struct sk_buff *skb)
 		break;
 
 	case GRO_MERGED_FREE:
-		consume_skb(skb);
+		if (NAPI_GRO_CB(skb)->free == NAPI_GRO_FREE_STOLEN_HEAD)
+			kmem_cache_free(skbuff_head_cache, skb);
+		else
+			__kfree_skb(skb);
 		break;
 
 	case GRO_HELD:
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index effa75d0e31..09cc38651b2 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -69,7 +69,7 @@
 #include <trace/events/skb.h>
 #include <linux/highmem.h>
 
-static struct kmem_cache *skbuff_head_cache __read_mostly;
+struct kmem_cache *skbuff_head_cache __read_mostly;
 static struct kmem_cache *skbuff_fclone_cache __read_mostly;
 
 static void sock_pipe_buf_release(struct pipe_inode_info *pipe,
@@ -2901,6 +2901,31 @@ int skb_gro_receive(struct sk_buff **head, struct sk_buff *skb)
 
 		NAPI_GRO_CB(skb)->free = 1;
 		goto done;
+	} else if (skb->head_frag) {
+		int nr_frags = pinfo->nr_frags;
+		skb_frag_t *frag = pinfo->frags + nr_frags;
+		struct page *page = virt_to_head_page(skb->head);
+		unsigned int first_size = headlen - offset;
+		unsigned int first_offset;
+
+		if (nr_frags + 1 + skbinfo->nr_frags > MAX_SKB_FRAGS)
+			return -E2BIG;
+
+		first_offset = skb->data -
+			       (unsigned char *)page_address(page) +
+			       offset;
+
+		pinfo->nr_frags = nr_frags + 1 + skbinfo->nr_frags;
+
+		frag->page.p	  = page;
+		frag->page_offset = first_offset;
+		skb_frag_size_set(frag, first_size);
+
+		memcpy(frag + 1, skbinfo->frags, sizeof(*frag) * skbinfo->nr_frags);
+		/* We dont need to clear skbinfo->nr_frags here */
+
+		NAPI_GRO_CB(skb)->free = NAPI_GRO_FREE_STOLEN_HEAD;
+		goto done;
 	} else if (skb_gro_len(p) != pinfo->gso_size)
 		return -E2BIG;