[PATCH] Better fixup for the orinoco driver
The latest kernel added a pretty ugly fix for the orinoco etherleak bug
which contains bogus skb->len checks already done by the caller and causes
copies of all odd sized frames (which are quite common)
While the skb->len check should be ripped out the other fix is harder to do
properly so I'm proposing for this the -mm tree only until next 2.6.x so
that it gets tested.
Instead of copying buffers around blindly this code implements a padding
aware version of the hermes buffer writing function which does padding as
the buffer is loaded and thus more cleanly and without bogus 1.5K copies.
Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
diff --git a/drivers/net/wireless/orinoco.c b/drivers/net/wireless/orinoco.c
index 70a3477..488ab06 100644
--- a/drivers/net/wireless/orinoco.c
+++ b/drivers/net/wireless/orinoco.c
@@ -542,14 +542,21 @@
stats->tx_errors++;
goto fail;
}
+ /* Actual xfer length - allow for padding */
+ len = ALIGN(data_len, 2);
+ if (len < ETH_ZLEN - ETH_HLEN)
+ len = ETH_ZLEN - ETH_HLEN;
} else { /* IEEE 802.3 frame */
data_len = len + ETH_HLEN;
data_off = HERMES_802_3_OFFSET;
p = skb->data;
+ /* Actual xfer length - round up for odd length packets */
+ len = ALIGN(data_len, 2);
+ if (len < ETH_ZLEN)
+ len = ETH_ZLEN;
}
- /* Round up for odd length packets */
- err = hermes_bap_pwrite(hw, USER_BAP, p, ALIGN(data_len, 2),
+ err = hermes_bap_pwrite_pad(hw, USER_BAP, p, data_len, len,
txfid, data_off);
if (err) {
printk(KERN_ERR "%s: Error %d writing packet to BAP\n",