[CASSINI]: Fix endianness bug.

Here's proposed fix for RX checksum handling in cassini; it affects
little-endian working with half-duplex gigabit, but obviously needs
testing on big-endian too.

The problem is, we need to convert checksum to fixed-endian *before*
correcting for (unstripped) FCS.  On big-endian it won't matter
(conversion is no-op), on little-endian it will, but only if FCS is
not stripped by hardware; i.e. in half-duplex gigabit mode when
->crc_size is set.

cassini.c part is that fix, cassini.h one consists of trivial
endianness annotations.  With that applied the sucker is endian-clean,
according to sparse.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/cassini.c

@@ -1979,6 +1979,7 @@ static int cas_rx_process_pkt(struct cas *cp, struct cas_rx_comp *rxc,
 	struct cas_page *page;
 	struct sk_buff *skb;
 	void *addr, *crcaddr;
+	__sum16 csum;
 	char *p;
 
 	hlen = CAS_VAL(RX_COMP2_HDR_SIZE, words[1]);
@@ -2158,14 +2159,15 @@ static int cas_rx_process_pkt(struct cas *cp, struct cas_rx_comp *rxc,
 		skb_put(skb, alloclen);
 	}
 
-	i = CAS_VAL(RX_COMP4_TCP_CSUM, words[3]);
+	csum = (__force __sum16)htons(CAS_VAL(RX_COMP4_TCP_CSUM, words[3]));
 	if (cp->crc_size) {
 		/* checksum includes FCS. strip it out. */
-		i = csum_fold(csum_partial(crcaddr, cp->crc_size, i));
+		csum = csum_fold(csum_partial(crcaddr, cp->crc_size,
+					      csum_unfold(csum)));
 		if (addr)
 			cas_page_unmap(addr);
 	}
-	skb->csum = ntohs(i ^ 0xffff);
+	skb->csum = csum_unfold(~csum);
 	skb->ip_summed = CHECKSUM_COMPLETE;
 	skb->protocol = eth_type_trans(skb, cp->dev);
 	return len;

drivers/net/cassini.h

@@ -4122,8 +4122,8 @@ cas_saturn_patch_t cas_saturn_patch[] = {
 							     inserted into
 							     outgoing frame. */
 struct cas_tx_desc {
-	u64     control;
-	u64     buffer;
+	__le64     control;
+	__le64     buffer;
 };
 
 /* descriptor ring for free buffers contains page-sized buffers. the index
@@ -4131,8 +4131,8 @@ struct cas_tx_desc {
  * the completion ring.
  */
 struct cas_rx_desc {
-	u64     index;
-	u64     buffer;
+	__le64     index;
+	__le64     buffer;
 };
 
 /* received packets are put on the completion ring. */
@@ -4210,10 +4210,10 @@ struct cas_rx_desc {
 #define RX_INDEX_RELEASE                  0x0000000000002000ULL
 
 struct cas_rx_comp {
-	u64     word1;
-	u64     word2;
-	u64     word3;
-	u64     word4;
+	__le64     word1;
+	__le64     word2;
+	__le64     word3;
+	__le64     word4;
 };
 
 enum link_state {
@@ -4252,7 +4252,7 @@ struct cas_init_block {
 	struct cas_rx_comp rxcs[N_RX_COMP_RINGS][INIT_BLOCK_RX_COMP];
 	struct cas_rx_desc rxds[N_RX_DESC_RINGS][INIT_BLOCK_RX_DESC];
 	struct cas_tx_desc txds[N_TX_RINGS][INIT_BLOCK_TX];
-	u64 tx_compwb;
+	__le64 tx_compwb;
 };
 
 /* tiny buffers to deal with target abort issue. we allocate a bit