On Sun, 2008-04-20 at 10:41 +0200, Sebastian Siewior wrote: > The function before lbs_ethtool_get_stats() (in net code that > determinate the number of elements, dunno the name right now) is able to > return an error. This one could check if the firmware has mesh support. That's what my second patch does. > Do I understand this correct: If the firmware has support for mesh > devices than we have ethX and mshX and only mshX should return the > statistics? Well, they are _mesh_ statistics :) > returning uninitialized nonsense is leaking kernel memory. Indeed. We need that memset there to handle errors, definitely. > What about changing lbs_ethtool_get_stats() from void to int? All > other drivers are reading memory to get this data, we have to go > through usb/cs/sdio layer and all of them may fail. Yeah, that might be worthwhile. > I will try to form a patch around Monday that fixes Dan's comments (if > nobody else is going to). Which comments? If we apply your original patch, followed by my patch in <1208509888.9212.291.camel@xxxxxxxxxxxxxxxxxx> (which for some reason only went to wireless-dev, not libertas-dev), then I think it's fine. -- dwmw2 -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
