From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from mail2.candelatech.com ([208.74.158.173]:52010 "EHLO mail2.candelatech.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932414AbcK1XwJ (ORCPT ); Mon, 28 Nov 2016 18:52:09 -0500 Received: from [192.168.43.60] (unknown [172.56.42.146]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by mail2.candelatech.com (Postfix) with ESMTPSA id EE73F40B24A for ; Mon, 28 Nov 2016 15:52:08 -0800 (PST) Message-ID: <583CC3CF.1040101@candelatech.com> (sfid-20161129_005213_046862_FC8BD006) Date: Mon, 28 Nov 2016 15:54:55 -0800 From: Ben Greear MIME-Version: 1.0 To: "linux-wireless@vger.kernel.org" Subject: OOM with pktgen when transmitting on ath10k station in 4.7.10+ kernel Content-Type: text/plain; charset=utf-8; format=flowed Sender: linux-wireless-owner@vger.kernel.org List-ID: When I over-drive ath10k station using pktgen, system goes OOM quickly. If I catch it in time and stop pktgen, then memory is recovered over the next few seconds. I assume that there is an un-bounded queue somewhere. But, I cannot find any queues in the mac80211 tx path that apply to ath10k (as far as I can tell). The one possibility is the pending queues, but printing them out with debugfs shows them all zeros, and I have some code to bound them at 1000 pkts anyway. I didn't see anything obvious in ath10k either, but I must be missing something... pktgen transmits under the queue logic, so it will likely be ignoring any queue-stopped signals from mac80211. Are there any other places that pkts can be queued()? Thanks, Ben -- Ben Greear Candela Technologies Inc http://www.candelatech.com