net: replace percpu_xxx funcs with this_cpu_xxx or __this_cpu_xxx

percpu_xxx funcs are duplicated with this_cpu_xxx funcs, so replace
them for further code clean up.

And in preempt safe scenario, __this_cpu_xxx funcs may has a bit
better performance since __this_cpu_xxx has no redundant
preempt_enable/preempt_disable on some architectures.

Signed-off-by: Alex Shi <[e-mail]>
Acked-by: Eric Dumazet <[e-mail]>
Acked-by: David S. Miller <[e-mail]>
Cc: Patrick McHardy <[e-mail]>
Signed-off-by: Andrew Morton <[e-mail]>
Signed-off-by: Tejun Heo <[e-mail]>

net/netfilter/xt_TEE.c

···
 	const struct xt_tee_tginfo *info = par->targinfo;
 	struct iphdr *iph;
 
-	if (percpu_read(tee_active))
+	if (__this_cpu_read(tee_active))
 		return XT_CONTINUE;
 	/*
 	 * Copy the skb, and route the copy. Will later return %XT_CONTINUE for
···
 	ip_send_check(iph);
 
 	if (tee_tg_route4(skb, info)) {
-		percpu_write(tee_active, true);
+		__this_cpu_write(tee_active, true);
 		ip_local_out(skb);
-		percpu_write(tee_active, false);
+		__this_cpu_write(tee_active, false);
 	} else {
 		kfree_skb(skb);
 	}
···
 {
 	const struct xt_tee_tginfo *info = par->targinfo;
 
-	if (percpu_read(tee_active))
+	if (__this_cpu_read(tee_active))
 		return XT_CONTINUE;
 	skb = pskb_copy(skb, GFP_ATOMIC);
 	if (skb == NULL)
···
 		--iph->hop_limit;
 	}
 	if (tee_tg_route6(skb, info)) {
-		percpu_write(tee_active, true);
+		__this_cpu_write(tee_active, true);
 		ip6_local_out(skb);
-		percpu_write(tee_active, false);
+		__this_cpu_write(tee_active, false);
 	} else {
 		kfree_skb(skb);
 	}

net/socket.c

···
 	inode->i_uid = current_fsuid();
 	inode->i_gid = current_fsgid();
 
-	percpu_add(sockets_in_use, 1);
+	this_cpu_add(sockets_in_use, 1);
 	return sock;
 }
 
···
 	if (rcu_dereference_protected(sock->wq, 1)->fasync_list)
 		printk(KERN_ERR "sock_release: fasync list not empty!\n");
 
-	percpu_sub(sockets_in_use, 1);
+	this_cpu_sub(sockets_in_use, 1);
 	if (!sock->file) {
 		iput(SOCK_INODE(sock));
 		return;