tx flush bugs, at least in my firmware.

[Ben Greear]

Just FYI, I have no idea if this problem is with my patches, my
firwmare, or what.  At any rate, it is unlikely to be reproducible on
upstream firmware due to its limitations in numbers of VIFS supported.

Here is the problem:

Previous to today, I was using wpa_supplicant that had sharing of scan results
mostly disabled.  It would only associate a single station VIF at a time, and
mostly things were working up to 36 stations (yeah, I managed to eke out another
4 stations :)).

But today, I started fixing the scan result sharing to work like it used to work
back when I mostly tested ath9k systems.

Now, after 20-25 or so stations vifs are associated (and most of the first 20 would
be associating concurrently as possible), I get the errors below and ath10k is
effectively hung:

ath10k: failed to flush transmit queue (skip 0 ar-state 1): 0

This is very repeatable.

My kernel does not have the latest two series of patches that Michal
posted that deal with reset issues, so maybe they hold the key.

So, I'm going to dig into this, but thought I'd throw this out in case
anyone else is seeing something similar.

Thanks,
Ben

-- 
Ben Greear <greearb at candelatech.com>
Candela Technologies Inc  http://www.candelatech.com