- From: Michael Welzl <
>
- To:
- Subject: Bugs in my code
- Date: Tue, 19 Dec 2017 11:08:01 +0100
Hi,
If anyone stumbles over bugs in my code when testing it, I’d greatly
appreciate being told !!
I noticed the following things when testing it "in real life”:
- a problem with SACK: this assert can fail:
NS_ASSERT ((*(m_sentList.begin ()))->m_sacked == false);
… in the file tcp-tx-buffer.cc.
- when pacingDelay becomes zero, we get an endless loop (or, in real-life
tests, a simple crash) - because ns-3 wants to endlessly schedule something
at the same time.
This can happen because the calculation yields something close to zero (few
packets over a large interval, something like that). As a quick fix, I added
this line in my test script just above Simulator::Run():
Time::SetResolution(ns3::Time::PS);
… but this doesn’t really solve the problem for all cases. I think I need to
add a rule in the pacing logic to say, when the result is zero, simply don’t
delay anything instead of trying to schedule.
I’ll look at these bugs as soon as I can, but if someone else feels like
fixing, by all means go ahead!
…and if you detect more, please do tell me!
Meanwhile, I think for tests it’s safer to turn off SACK.
Cheers,
Michael
- Bugs in my code, Michael Welzl
Archivio con motore MhonArc 2.6.16.