On Fri, Dec 18, 2015 at 03:42:05PM +0100, Stefan Sperling wrote: > On Fri, Dec 18, 2015 at 03:16:52PM +0100, David Coppa wrote: > > On Thu, Dec 17, 2015 at 10:34 PM, Stefan Sperling <s...@stsp.name> wrote: > > > On Thu, Dec 17, 2015 at 08:13:03PM +0100, Stefan Sperling wrote: > > >> This should fix the infinite scanning loops people have been > > >> reporting with 11n-enabled iwn(4), as well as the issue where > > >> clients associating to 11g APs end up in 11b mode and can't > > >> use OFDM data rates. > > >> > > >> ok? > > > > > > Updated diff which allows scanning to work after somebody decided > > > to run 'ifconfig iwn0 media autoselect mode 11n' for some reason. > > > (Don't do that, it won't enforce 11n -- 11n cannot really be set > > > a priori because it's negotiated during association.) > > > > > > Also delete a bit of code from ieee80211_newstate() in ieee80211_proto.c > > > which is pointless now that ieee80211_chan2mode() won't return MODE_11N. > > > > A minor problem I've noticed was dhclient being slower to get the link > > from my AP. > > The patch below fixed it. > > Here's yet another update. > > I hope this also fixes a transmit problem on 2GHz reported by sthen@. > For some reason ieee80211_end_scan() handled some parts of setting > up 11g ERP but that's now much too early in the association process. > I've moved this code into ieee80211_recv_assoc_resp() because before > then we just don't know whether we're going to end up in 11g or 11n mode.
With this diff on top of the other iwn diff, my x200s correctly gets 11g instead of falling back to 11b: media: IEEE802.11 autoselect (OFDM6 mode 11g) iwn0 at pci2 dev 0 function 0 "Intel WiFi Link 5100" rev 0x00: msi, MIMO 1T2R, MoW time find /usr/src/sys/arch/amd64 (woooo benchmarks!) over nfs isnt a regression anymore :) Landry
