Differential D8365

[net80211] add initial support for HT upgrade of IBSS nodes. * add ability for IBSS nodes to be upgraded when it sees HT IEs * add HT IEs to probe requests/responses and beacons * send a probe request after bringing up a node so we can...
ClosedPublic
Actions

Authored by adrian on Oct 28 2016, 2:12 AM.

Details

Reviewers

avos

Commits

rS308949: [net80211] flesh out more IBSS 11n support

Summary

...potentially

upgrade it to 11n if we see a probe response with HE IEs.

The 802.11 spec notes that a subset of HTINFO is allowed for IBSS, so
ensure we add it. However, there's nothing in the spec that fully
explains how nodes negotiate IBSS. Eg, we may want to not start sending
traffic until a probe request/response occurs so we don't have to
re-init the node with updated rate control information.

Diff Detail

Repository

rS FreeBSD src repository - subversion

Lint

Lint Passed

Unit

No Test Coverage

Build Status

Buildable 5724
Build 6021: CI src build	Jenkins
Build 6020: arc lint + arc unit

Event Timeline

adrian updated this revision to Diff 21744.Oct 28 2016, 2:12 AM

adrian retitled this revision from to [net80211] add initial support for HT upgrade of IBSS nodes. * add ability for IBSS nodes to be upgraded when it sees HT IEs * add HT IEs to probe requests/responses and beacons * send a probe request after bringing up a node so we can....

adrian updated this object.

adrian edited the test plan for this revision. (Show Details)

Herald added a subscriber: imp. · View Herald TranscriptOct 28 2016, 2:12 AM

adrian added a reviewer: avos.Oct 28 2016, 2:12 AM

avos added inline comments.Oct 29 2016, 8:50 AM

sys/net80211/ieee80211_output.c
2077	As I can see, this variable is never used here. Is something missing?

ping? any comments?

avos accepted this revision.Nov 21 2016, 8:08 PM

avos edited edge metadata.

avos added inline comments.

sys/net80211/ieee80211_output.c
2085	Still have no idea why it is referenced here

This revision is now accepted and ready to land.Nov 21 2016, 8:08 PM

adrian added inline comments.Nov 21 2016, 8:34 PM

sys/net80211/ieee80211_output.c
2085	Because vap->iv_bss can be changed/recycled at any time, and unless you're holding the node table lock, you're not guaranteed to get the stable version. Now, this is also not guaranteed to get the stable version because of how the node reference counting is implemented - it's an atomic, and it's quite possible that something is freeing the node whilst you're busy getting a reference on it - but this reduces the window way way down and it's the "right" way to be getting access to iv_bss anyway.