On 10/04/14 09:59, Johannes Berg wrote: > On Thu, 2014-04-10 at 09:00 +0200, Antonio Quartulli wrote: >> On 09/04/14 22:39, Johannes Berg wrote: >>> diff --git a/net/wireless/scan.c b/net/wireless/scan.c >>> index 7d09a712cb1f..746c56ebd66c 100644 >>> --- a/net/wireless/scan.c >>> +++ b/net/wireless/scan.c >>> @@ -540,6 +540,8 @@ struct cfg80211_bss *cfg80211_get_bss(struct wiphy *wiphy, >>> continue; >>> if (channel && bss->pub.channel != channel) >>> continue; >>> + if (!is_valid_ether_addr(bss->pub.bssid)) >>> + continue; >> >> Wouldn't it be better to prevent such entry to end up in the bss list at >> all? (i.e. filtering during the scan?) > > I thought about that, but I'm not so sure. > > On the one hand, that would make sense since it's unusable, on the other > hand it could cause confusion if you don't see some other network that > you expect to show up ... I agree here. Showing the network is reasonable because the user expects to see what's around. Imagine a user that scans just to see which channel is less crowded... > > And then again, if you can click the network in the UI and then it won't > connect to it, that also causes confusion. I'm not sure what the best > way is ... > True. But here it's up to the UI to check the MAC and tell the user that this is an invalid BSSID, no? In the end I drop my original objection :) Cheers, -- Antonio Quartulli
Attachment:
signature.asc
Description: OpenPGP digital signature