svn commit: r226933 - user/adrian/if_ath_tx/sys/dev/ath/ath_hal/ar5416

[Adrian Chadd]

Author: adrian
Date: Sun Oct 30 12:11:12 2011
New Revision: 226933
URL: http://svn.freebsd.org/changeset/base/226933

Log:
  For now, flip off 11n chipset OFDM weak signal detection by default.
  
  I've noticed in very congested environments that the AR9220's I have
  get very upset and spew out tens of thousands of OFDM PHY errors a second
  (OFDM restart, mostly.) The AR9160 in the same environment doesn't log
  anywhere near as many OFDM errors and does not at all get "stuck beacon"
  issues.
  
  I however think the stuck beacon condition that I and testers have been
  seeing is due to other issues, rather than specifically caused by
  lots of PHY errors. I think the PHY errors are just setting it off.
  
  Merlin does seem to have a "baseband hang" bit in the hang capability
  set, I have no idea what the baseband hangs are or what the workarounds
  are. In any case, I'll keep digging.
  
  Newma has this disabled in hostap mode and leaves it default in STA mode
  (I think this is 'whatever is in the ini file values.')
  It will be enabled if the beacon RSSI is strong enough.

Modified:
  user/adrian/if_ath_tx/sys/dev/ath/ath_hal/ar5416/ar5416_ani.c

Modified: user/adrian/if_ath_tx/sys/dev/ath/ath_hal/ar5416/ar5416_ani.c
==============================================================================
--- user/adrian/if_ath_tx/sys/dev/ath/ath_hal/ar5416/ar5416_ani.c	Sun Oct 30 11:43:12 2011	(r226932)
+++ user/adrian/if_ath_tx/sys/dev/ath/ath_hal/ar5416/ar5416_ani.c	Sun Oct 30 12:11:12 2011	(r226933)
@@ -611,7 +611,7 @@ ar5416AniReset(struct ath_hal *ah, const
 		ar5416AniControl(ah, HAL_ANI_NOISE_IMMUNITY_LEVEL, 0);
 		ar5416AniControl(ah, HAL_ANI_SPUR_IMMUNITY_LEVEL, 0);
 		ar5416AniControl(ah, HAL_ANI_OFDM_WEAK_SIGNAL_DETECTION,
-			AH_TRUE);
+			AH_FALSE);
 		ar5416AniControl(ah, HAL_ANI_CCK_WEAK_SIGNAL_THR, AH_FALSE);
 		ar5416AniControl(ah, HAL_ANI_FIRSTEP_LEVEL, 0);
 		ichan->privFlags |= CHANNEL_ANI_SETUP;