ath9k: correct expected max RX buffer size

We should only tell the hardware its capable of DMA'ing
to us only what we asked dev_alloc_skb(). Prior to this
it is possible a large RX'd frame could have corrupted
DMA data but for us but we were saved only because we
were previously also pci_map_single()'ing the same large
value. The issue prior to this though was we were unmapping
a smaller amount which the prior DMA patch fixed.
Signed-off-by: Bennyam Malavazi <Bennyam.Malavazi@atheros.com>
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

drivers/net/wireless/ath9k/recv.c

@@ -49,10 +49,12 @@ static void ath_rx_buf_link(struct ath_softc *sc, struct ath_buf *bf)
 	ASSERT(skb != NULL);
 	ds->ds_vdata = skb->data;
 
-	/* setup rx descriptors */
+	/* setup rx descriptors. The sc_rxbufsize here tells the harware
+	 * how much data it can DMA to us and that we are prepared
+	 * to process */
 	ath9k_hw_setuprxdesc(ah,
 			     ds,
-			     skb_tailroom(skb),   /* buffer size */
+			     sc->sc_rxbufsize,
 			     0);
 
 	if (sc->sc_rxlink == NULL)
@@ -398,6 +400,13 @@ static struct sk_buff *ath_rxbuf_alloc(struct ath_softc *sc,
 	 * in rx'd frames.
 	 */
 
+	/* Note: the kernel can allocate a value greater than
+	 * what we ask it to give us. We really only need 4 KB as that
+	 * is this hardware supports and in fact we need at least 3849
+	 * as that is the MAX AMSDU size this hardware supports.
+	 * Unfortunately this means we may get 8 KB here from the
+	 * kernel... and that is actually what is observed on some
+	 * systems :( */
 	skb = dev_alloc_skb(len + sc->sc_cachelsz - 1);
 	if (skb != NULL) {
 		off = ((unsigned long) skb->data) % sc->sc_cachelsz;