Bug 1843

Created attachment 1768 [details]
Patch (option 1)

Currently we have a glitch in IPv6 dropped packets handling.

If the packet is dropped "normally", then the Ipv6L3Protocol::m_dropTrace is called.
If the packet is dropped because an extension is "bad", then Ipv6Extension::m_dropTrace is called.

Now the problem is twofold.
1) it is not obvious that both traces should be hooked, and
2) Ipv6Extension::m_dropTrace does not have the same parameters as the other one.
In particular, the drop cause is missing, plus a lot of other things.

In order to fix it, two things might be done:
1) kill Ipv6Extension::m_dropTrace and have Ipv6L3Protocol::m_dropTrace called instead.
2) leave Ipv6Extension::m_dropTrace but call Ipv6L3Protocol::m_dropTrace as well.
3) modify Ipv6Extension::m_dropTrace to have the same info as Ipv6L3Protocol::m_dropTrace.

Cons:
1) might break existing simulations.
2) might produce funny results (i.e., drops counted twice)
3) is still not obvious. People might forget about drops in extensions.

Of the three options I prefer the first. It's intrusive, but at least people will be forced to use the right things.