[PATCH] ieee80211_geo.c: remove frivolous BUG_ON's

I have come to consider BUG_ON generally harmful. The idea of an assert is
to prevent a program to execute past a point where its state is known
erroneous, thus preventing it from dealing more damage to the data
(or hiding the traces of malfunction). The problem is, in kernel this harm
has to be balanced against the harm of forced reboot.

The last straw was our softmac tree, where "iwlist eth1 scan" causes
a lockup. It is absolutely frivolus and provides no advantages a normal
assert has to provide. In fact, doing this impedes debugging.

Signed-off-by: Pete Zaitcev <zaitcev@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

net/ieee80211/ieee80211_geo.c

@@ -50,7 +50,8 @@ int ieee80211_is_valid_channel(struct ieee80211_device *ieee, u8 channel)
 
 	/* Driver needs to initialize the geography map before using
 	 * these helper functions */
-	BUG_ON(ieee->geo.bg_channels == 0 && ieee->geo.a_channels == 0);
+	if (ieee->geo.bg_channels == 0 && ieee->geo.a_channels == 0)
+		return 0;
 
 	if (ieee->freq_band & IEEE80211_24GHZ_BAND)
 		for (i = 0; i < ieee->geo.bg_channels; i++)
@@ -78,7 +79,8 @@ int ieee80211_channel_to_index(struct ieee80211_device *ieee, u8 channel)
 
 	/* Driver needs to initialize the geography map before using
 	 * these helper functions */
-	BUG_ON(ieee->geo.bg_channels == 0 && ieee->geo.a_channels == 0);
+	if (ieee->geo.bg_channels == 0 && ieee->geo.a_channels == 0)
+		return -1;
 
 	if (ieee->freq_band & IEEE80211_24GHZ_BAND)
 		for (i = 0; i < ieee->geo.bg_channels; i++)
@@ -99,7 +101,8 @@ u8 ieee80211_freq_to_channel(struct ieee80211_device * ieee, u32 freq)
 
 	/* Driver needs to initialize the geography map before using
 	 * these helper functions */
-	BUG_ON(ieee->geo.bg_channels == 0 && ieee->geo.a_channels == 0);
+	if (ieee->geo.bg_channels == 0 && ieee->geo.a_channels == 0)
+		return 0;
 
 	freq /= 100000;