Return-path: Received: from nbd.name ([46.4.11.11]:44229 "EHLO nbd.name" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752887Ab3KNKZ1 (ORCPT ); Thu, 14 Nov 2013 05:25:27 -0500 Message-ID: <5284A50E.4080702@openwrt.org> (sfid-20131114_112530_336904_A92AC477) Date: Thu, 14 Nov 2013 11:25:18 +0100 From: Felix Fietkau MIME-Version: 1.0 To: Bruno Randolf , linux-wireless@vger.kernel.org CC: Sujith Manoharan , Ingo Randolf , mcgrof@do-not-panic.com Subject: Re: ath9k AR9382 high latency ping References: <52492FB0.3010201@einfach.org> <5249443E.3080700@rempel-privat.de> <524954BE.8060603@einfach.org> <52497C2E.1060406@rempel-privat.de> <5249A2F7.5090908@einfach.org> <524DA3DC.6060806@einfach.org> <21087.18050.72635.617321@gargle.gargle.HOWL> <52849677.3090609@einfach.org> In-Reply-To: <52849677.3090609@einfach.org> Content-Type: text/plain; charset=ISO-8859-1 Sender: linux-wireless-owner@vger.kernel.org List-ID: On 2013-11-14 10:23, Bruno Randolf wrote: > Hello All! > > Here is another ath9k PCIe card I have problems with in a Intel Atom > N2600 embedded board (Advantech MIO-2261N). This is the same board I was > reporting problems with the AR9287 before. I exchanged the card for an > AR9382 and at least I can associate now. > > It's a AR9382 (Unex DNXA-116) and the PCI ID 168c:0030 subsystem > 168c:3116, reported as "AR9300 Rev:3"... > > The symptom are ping times of around 1000ms, when connected to an AP > (very close, I expect around 2ms), eg. > > 64 bytes from 192.168.5.1: icmp_seq=1 ttl=64 time=1237 ms > 64 bytes from 192.168.5.1: icmp_seq=2 ttl=64 time=1403 ms > 64 bytes from 192.168.5.1: icmp_seq=3 ttl=64 time=1567 ms > 64 bytes from 192.168.5.1: icmp_seq=4 ttl=64 time=1728 ms > > The same card works fine in another Laptop, BTW, so it must be a problem > specific to this board. > > I run wireless-testing from yesterday and I tried with different > combinations of enabling/disabling ASPM in the BIOS and booting with > pcie_aspm=off pcie_aspm=force all of which has the same result... Have you checked for IRQ issues? That seems likely given that packets are sent, but the tx path is still considered hung. - Felix