tcp: fix tcp_md5_hash_skb_data()

TCP md5 communications fail [1] for some devices, because sg/crypto code
assume page offsets are below PAGE_SIZE.

This was discovered using mlx4 driver [2], but I suspect loopback
might trigger the same bug now we use order-3 pages in tcp_sendmsg()

[1] Failure is giving following messages.

huh, entered softirq 3 NET_RX ffffffff806ad230 preempt_count 00000100,
exited with 00000101?

[2] mlx4 driver uses order-2 pages to allocate RX frags

Reported-by: Matt Schnall <mischnal@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Bernhard Beck <bbeck@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/ipv4/tcp.c

@@ -3269,8 +3269,11 @@ int tcp_md5_hash_skb_data(struct tcp_md5sig_pool *hp,
 
 	for (i = 0; i < shi->nr_frags; ++i) {
 		const struct skb_frag_struct *f = &shi->frags[i];
-		struct page *page = skb_frag_page(f);
-		sg_set_page(&sg, page, skb_frag_size(f), f->page_offset);
+		unsigned int offset = f->page_offset;
+		struct page *page = skb_frag_page(f) + (offset >> PAGE_SHIFT);
+
+		sg_set_page(&sg, page, skb_frag_size(f),
+			    offset_in_page(offset));
 		if (crypto_hash_update(desc, &sg, skb_frag_size(f)))
 			return 1;
 	}