On Mon, 2014-05-19 at 16:45 +0400, Sergei Shtylyov wrote:

Hello.

On 19-05-2014 8:31, Julia Lawall wrote:

quoted

From: Julia Lawall <redacted>

quoted

Delete unnecessary local variable whose value is always 0 and that hides
the fact that the result is always 0.

quoted

A simplified version of the semantic patch that fixes this problem is as
follows: (http://coccinelle.lip6.fr/)

quoted

// <smpl>
@r exists@
local idexpression ret;
expression e;
position p;
@@

-ret = 0;
... when != ret = e
return
- ret
+ 0
   ;
// </smpl>

quoted

Signed-off-by: Julia Lawall <redacted>

quoted

---
Alternatively, was an error code intended in the bad length case, as is
done in process_brxed_802_11_packet?

quoted

  drivers/net/wireless/libertas/rx.c |    7 ++-----
  1 file changed, 2 insertions(+), 5 deletions(-)

quoted

diff --git a/drivers/net/wireless/libertas/rx.c b/drivers/net/wireless/libertas/rx.c
index c7366b0..807c5b8 100644
--- a/drivers/net/wireless/libertas/rx.c
+++ b/drivers/net/wireless/libertas/rx.c

[...]

quoted

@@ -154,10 +152,9 @@ int lbs_process_rxed_packet(struct lbs_private *priv, struct sk_buff *skb)
  	else
  		netif_rx_ni(skb);

-	ret = 0;
  done:
-	lbs_deb_leave_args(LBS_DEB_RX, "ret %d", ret);
-	return ret;
+	lbs_deb_leave_args(LBS_DEB_RX, "ret %d", 0);

    Why not just "ret 0"?

I think instead we want:

diff --git a/drivers/net/wireless/libertas/rx.c b/drivers/net/wireless/libertas/rx.c
index c7366b0..e446fed 100644
--- a/drivers/net/wireless/libertas/rx.c
+++ b/drivers/net/wireless/libertas/rx.c

@@ -67,30 +67,32 @@ int lbs_process_rxed_packet(struct lbs_private *priv, struct sk_buff *skb)
 
 	lbs_deb_enter(LBS_DEB_RX);
 
 	BUG_ON(!skb);
 
 	skb->ip_summed = CHECKSUM_NONE;
 
-	if (priv->wdev->iftype == NL80211_IFTYPE_MONITOR)
-		return process_rxed_802_11_packet(priv, skb);
+	if (priv->wdev->iftype == NL80211_IFTYPE_MONITOR) {
+		ret = process_rxed_802_11_packet(priv, skb);
+		goto done;
+	}
 
 	p_rx_pd = (struct rxpd *) skb->data;
 	p_rx_pkt = (struct rxpackethdr *) ((u8 *)p_rx_pd +
 		le32_to_cpu(p_rx_pd->pkt_ptr));
 
 	dev = lbs_mesh_set_dev(priv, dev, p_rx_pd);
 
 	lbs_deb_hex(LBS_DEB_RX, "RX Data: Before chop rxpd", skb->data,
 		 min_t(unsigned int, skb->len, 100));
 
 	if (skb->len < (ETH_HLEN + 8 + sizeof(struct rxpd))) {
 		lbs_deb_rx("rx err: frame received with bad length\n");
 		dev->stats.rx_length_errors++;
-		ret = 0;
+		ret = -EINVAL;
 		dev_kfree_skb(skb);
 		goto done;
 	}
 
 	lbs_deb_rx("rx data: skb->len - pkt_ptr = %d-%zd = %zd\n",
 		skb->len, (size_t)le32_to_cpu(p_rx_pd->pkt_ptr),
 		skb->len - (size_t)le32_to_cpu(p_rx_pd->pkt_ptr));

Since that (a) preserves the enter/leave balance when monitor mode is
enabled, and (b) fixes the return code that has been 0 since the driver
was added to the tree in early 2007.  Dave Woodhouse fixed the
process_rxed_802_11_packet() return code for EINVAL case in commit
7bf02c29 from late 2007, and I think the normal codepath should do
EINVAL too.  (though it turns out nothing cares about the return code
anyway, we should still fix it)

Dan