Bug 2306

Summary:	wifi Dcf busy if CCA is busy
Product:	ns-3	Reporter:	Tom Henderson <tomh>
Component:	wifi	Assignee:	sebastien.deronne
Status:	RESOLVED FIXED
Severity:	normal	CC:	matis18, ns-bugs
Priority:	P3
Version:	ns-3.25
Hardware:	All
OS:	All
See Also:	https://www.nsnam.org/bugzilla/show_bug.cgi?id=2314
Attachments:	patch to fix

Description Tom Henderson 2016-02-22 21:45:57 EST

DcfManager has a method that checks whether the medium is busy for reasons of RXing, TXing, or NAV, but not for CCA busy for some reason.

The attached patch will raise DcfManager::IsBusy() when the phy state is in CCA busy.  It causes pcap-trace based regression traces to change, and one DcfManagerTest test to fail (change included in the patch) but other tests pass with this change.

Comment 1 Tom Henderson 2016-02-22 21:48:34 EST

Created attachment 2290 [details]
patch to fix

Comment 2 sebastien.deronne 2016-04-24 05:27:54 EDT

Looks ok for me at first sight.
Does anyone has some doubts on the proposed change?

Comment 3 Matías Richart 2016-04-28 10:11:02 EDT

I'm surprised of this bug and makes me wonder if it was done this way for some reason. 
But in my opinion the solution looks good and needs to be applied.

For the case of power adaptation this is an important change and makes some results of experiments I've made in the past to change a lot.

Comment 4 sebastien.deronne 2016-05-16 15:42:42 EDT

It looks like this patch makes routing-aodv-regression test fail.
I assume we should simply rescan pcap traces, to be checked though.

Comment 5 sebastien.deronne 2016-06-09 04:45:37 EDT

(In reply to sebastien.deronne from comment #4)
> It looks like this patch makes routing-aodv-regression test fail.
> I assume we should simply rescan pcap traces, to be checked though.

Tom will have a look.

Comment 6 Tom Henderson 2016-09-28 00:20:34 EDT

pushed in changeset 12342:2e7e4f29be88; all tests were reviewed and updated as needed