First, please send all wireless patches to
[email protected], and be sure to CC me as well...thanks!
On Sat, May 19, 2007 at 12:50:31AM +0800, Eugene Teo wrote:
> libertas_upload_rx_packet() calls netif_rx() before returning, and it always return 0.
> Also within libertas_upload_rx_packet(), it will initialize skb->protocol anyways.
>
> Spotted by the Coverity checker.
A nearly identical patch was posted by Florin Malita <[email protected]>
to netdev (also the wrong list) on Wednesday evening.
> done:
> LEAVE();
>
> - skb->protocol = __constant_htons(0x0019); /* ETH_P_80211_RAW */
> -
Except for this part...is this intentional?
John
--
John W. Linville
[email protected]
<quote sender="John W. Linville">
> On Sat, May 19, 2007 at 11:14:10AM +0800, Eugene Teo wrote:
> > John W. Linville wrote:
>
> > >> done:
> > >> LEAVE();
> > >>
> > >> - skb->protocol = __constant_htons(0x0019); /* ETH_P_80211_RAW */
> > >> -
> > >
> > > Except for this part...is this intentional?
> >
> > skb could have been freed by then. And, in libertas_upload_rx_packet(), skb->protocol
> > is initialized by eth_type_trans(skb, priv->wlan_dev.netdev).
>
> OK, I see that. Looks like Florin has reposted his patch, still
> without this hunk. Would you like to submit a patch for this hunk?
skb could have been freed by then. Also, in libertas_upload_rx_packet(),
skb->protocol is initialized by eth_type_trans().
Cc: John W. Linville <[email protected]>
Signed-off-by: Eugene Teo <[email protected]>
---
diff -uprN -X 2.6.22-rc2/Documentation/dontdiff 2.6.22-rc2.default/drivers/net/wireless/libertas/rx.c 2.6.22-rc2/drivers/net/wireless/libertas/rx.c
--- 2.6.22-rc2.default/drivers/net/wireless/libertas/rx.c 2007-05-21 22:07:50.000000000 +0800
+++ 2.6.22-rc2/drivers/net/wireless/libertas/rx.c 2007-05-21 22:08:44.000000000 +0800
@@ -453,7 +453,5 @@ static int process_rxed_802_11_packet(wl
done:
LEAVE();
- skb->protocol = __constant_htons(0x0019); /* ETH_P_80211_RAW */
-
return (ret);
}
--
1024D/58DF8823 print 47B9 90F6 AE4A 9C51 37E0 D6E1 EA84 C6A2 58DF 8823
main(i) { putchar(182623909 >> (i-1) * 5&31|!!(i<7)<<6) && main(++i); }
On Sat, May 19, 2007 at 11:14:10AM +0800, Eugene Teo wrote:
> John W. Linville wrote:
> >> done:
> >> LEAVE();
> >>
> >> - skb->protocol = __constant_htons(0x0019); /* ETH_P_80211_RAW */
> >> -
> >
> > Except for this part...is this intentional?
>
> skb could have been freed by then. And, in libertas_upload_rx_packet(), skb->protocol
> is initialized by eth_type_trans(skb, priv->wlan_dev.netdev).
OK, I see that. Looks like Florin has reposted his patch, still
without this hunk. Would you like to submit a patch for this hunk?
Thanks,
John
--
John W. Linville
[email protected]
John W. Linville wrote:
> First, please send all wireless patches to
> [email protected], and be sure to CC me as well...thanks!
>
> On Sat, May 19, 2007 at 12:50:31AM +0800, Eugene Teo wrote:
>> libertas_upload_rx_packet() calls netif_rx() before returning, and it always return 0.
>> Also within libertas_upload_rx_packet(), it will initialize skb->protocol anyways.
>>
>> Spotted by the Coverity checker.
>
> A nearly identical patch was posted by Florin Malita <[email protected]>
> to netdev (also the wrong list) on Wednesday evening.
Nod. I wasn't subscribed to netdev list.
>> done:
>> LEAVE();
>>
>> - skb->protocol = __constant_htons(0x0019); /* ETH_P_80211_RAW */
>> -
>
> Except for this part...is this intentional?
skb could have been freed by then. And, in libertas_upload_rx_packet(), skb->protocol
is initialized by eth_type_trans(skb, priv->wlan_dev.netdev).
Eugene