> On 28 Jan 2016, at 14:32, Neels Hofmeyr <[email protected]> wrote: > > Would anyone like to prevent this commit from reaching master? ;)
of course. I will even unplug your heated blanket! The silent_call handling doesn't fit from an architectural point of view. It is used to demonstrate a security implication of carrying a mobile phone. We simply open a channel and then don't do anything with it. At the same time one will receive measurement reports and could more easily triangulate or actively try to determine where the phone is. As such this feature needs to always "accept" a new "Complete layer3 information" (as of GSM 08.08). The original todo is indeed not a good one as it has absolutely no meta information. Thinking about the code my goal was to define a more clear ownership of the "MSC" side of an operation. So instead of starting a generic "close soon" timer (or the loc operation timer) have something that takes ownership of the first message. This would mean we dispatch "complete layer3 information" differently than follow up messages. As with many parts this part of the architecture has never been fixed/implemented/corrected as the current is working good enough(tm). We can either remove the comment, leave it as such or fix the above text and put it into the code as a fixme/todo. holger
