Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 16 Mar 2013 04:41:37 -0700
From:      Adrian Chadd <adrian@freebsd.org>
To:        freebsd-wireless@freebsd.org
Subject:   [ath] If you're running -HEAD, please update
Message-ID:  <CAJ-Vmo=H7437bJibjwHtrpFRtBN77gZt_PBkh9Lj3Z2Hiiko4Q@mail.gmail.com>

next in thread | raw e-mail | index | archive | help
Hi all,

If you're running -HEAD and ath(4) devices, then please update.

I've fixed a handful of corner case bugs recently:

* The pre-EDMA chips require keeping the last TX descriptor processed
around until the next is processed - and it needs to be done per-QCU,
not globally. This fixes some traffic stalls.
* The EDMA chips likely require the above too in special cases (read
CABQ processing) but as I don't yet support AR9380 in hostap mode, I'm
not yet worried about this.
* .. and I then implemented missing locking for the above. Tsk me.
* I fixed a TX buffer leak in the aggregation cleanup path (ie, when a
TID that's currently doing aggregation is downgraded to
non-aggregation.)
* I fixed a TID traffic queue depth bug, again in the aggregation cleanup path.

Thanks to Rui for being really patient with me whilst I've been
tracking down these issues and providing me with lots of debug logs

There's still a few odd corner cases to do with BAR handling and queue
pause/unpause but now that I've fixed the above issues (which were
causing hangs as well :) it shouldn't be difficult to track down the
root cause.




Adrian



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAJ-Vmo=H7437bJibjwHtrpFRtBN77gZt_PBkh9Lj3Z2Hiiko4Q>