This is from our modified 3.14.4+ kernel, but there are relatively few changes to ath9k. Test case is create 128 stations and run them against another ath9k AP, overdriving with UDP frames in bi-directional flows.
Both AP and station system show similar log messages. Some of the printouts below is from our debugging patches I believe... We have seen this since at least 3.7 era kernels, so it is probably not a regression, but it does seem to be easier to hit these days..maybe just some RF environment change though. My tree is here: http://dmz2.candelatech.com/git/gitweb.cgi?p=linux-3.14.dev.y/.git;a=summary This sounds a bit like the problems Dave Taht was reporting, though it appears my hack that resets the NIC when it is hung like this keeps it at least mostly working. Since it seems I have a good way to reproduce this now, I'm curious if anyone has suggestions for how to debug this better (or perhaps better work-arounds?) [313823.128460] ath: 1400503010.313367 wiphy0: BB_WATCHDOG: Skipping interrupts [313965.440167] ath: 1400503152.625074 wiphy0: BB_WATCHDOG: Skipping interrupts [314656.950849] ath: 1400503844.135752 wiphy0: Timeout while waiting for nf to load: AR_PHY_AGC_CONTROL=0x1d0dda [314656.951071] ath: 1400503844.135981 wiphy0: BB_WATCHDOG: Skipping interrupts [315343.351639] ath: 1400504530.536546 wiphy0: BB_WATCHDOG: Skipping interrupts [315661.015172] ath: 1400504848.200078 wiphy0: BB_WATCHDOG: Skipping interrupts [316949.624973] ath: 1400506136.809879 wiphy0: BB_WATCHDOG: Skipping interrupts [317259.273747] ath: 1400506446.458652 wiphy0: BB_WATCHDOG: Skipping interrupts [317601.176352] ath: 1400506788.361258 wiphy0: BB_WATCHDOG: Skipping interrupts [317727.457658] ath: 1400506914.642565 wiphy0: BB_WATCHDOG: Skipping interrupts [318416.860840] ath: 1400507604.45749 wiphy0: BB_WATCHDOG: Skipping interrupts [330581.299111] ath: 1400519768.484019 wiphy0: soft tx hang: queue: 2 pending-frames: 123, resetting chip [330581.309359] ath: 1400519768.494267 wiphy0: Pending frames still exist on txq: 2 after drain: 123 axq-depth: 0 ampdu-depth: 0 Thanks, Ben -- Ben Greear <gree...@candelatech.com> Candela Technologies Inc http://www.candelatech.com _______________________________________________ ath9k-devel mailing list ath9k-devel@lists.ath9k.org https://lists.ath9k.org/mailman/listinfo/ath9k-devel
