From owner-freebsd-wireless@FreeBSD.ORG Sat Feb 2 10:06:52 2013 Return-Path: Delivered-To: freebsd-wireless@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) by hub.freebsd.org (Postfix) with ESMTP id 350FDF8C for ; Sat, 2 Feb 2013 10:06:52 +0000 (UTC) (envelope-from adrian.chadd@gmail.com) Received: from mail-wi0-f175.google.com (mail-wi0-f175.google.com [209.85.212.175]) by mx1.freebsd.org (Postfix) with ESMTP id C1FDB8B5 for ; Sat, 2 Feb 2013 10:06:51 +0000 (UTC) Received: by mail-wi0-f175.google.com with SMTP id l13so897060wie.2 for ; Sat, 02 Feb 2013 02:06:44 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:x-received:in-reply-to:references:date:message-id :subject:from:to:cc:content-type:content-transfer-encoding; bh=JCQ7SL95BBpYMMUKRwZyebfPUT2iNEBNMHeIOZdqPSg=; b=fURRd9sOY7h0BywMsYQjmh7Wy3qvAvA194Y5ygBwnj5wkKPlVcA5eFsZOBNiv1FdGa gBSWlC31inffl+/iJZWsVMn0TlsmI+pNBzx3k661gKvXFi84rzw40Z/GTAiBSHYUQ2ZD 5i1AzK13mR4kgjt1EA9fB3RW6wd7/iFTH3Ds/iP+zTIu8h+N4Ubg4J7jt4sFuXUkwc/Q FHpYzkCvZgR0gYk9s8HFE2msUw8YRK5PXuHAAP08vgHOO4klbr1YI34Afgqwbr4Uv6rH NsMUB7O8sBe07h2bRuT02Gk24ooeZ/hzxGl9MCjdVAgUPllp6bDq9B3CieupiVdNdjzb OLpw== MIME-Version: 1.0 X-Received: by 10.194.172.197 with SMTP id be5mr26244410wjc.20.1359799604748; Sat, 02 Feb 2013 02:06:44 -0800 (PST) Received: by 10.216.98.5 with HTTP; Sat, 2 Feb 2013 02:06:44 -0800 (PST) In-Reply-To: References: <4f00f1d5.1446e70a.3836.ffffc25f@mx.google.com> Date: Sat, 2 Feb 2013 02:06:44 -0800 Message-ID: Subject: Re: ath "stops working" in hostap mode From: Adrian Chadd To: "Lay, Nathan" Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Cc: freebsd-wireless@freebsd.org X-BeenThere: freebsd-wireless@freebsd.org X-Mailman-Version: 2.1.14 Precedence: list List-Id: "Discussions of 802.11 stack, tools device driver development." List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 02 Feb 2013 10:06:52 -0000 Hi, So what happens when you turn off the devices in question? Do things settle down? Adrian On 1 February 2013 22:45, Lay, Nathan wrote: > One last observation, the Roku and the tablet both employ Broadcom Wi-Fi = chipsets ... > > Lastly, thanks for all your hard work on if_ath and FreeBSD. > > Nate > > ________________________________________ > From: Lay, Nathan > Sent: Saturday, February 02, 2013 1:26 AM > To: Adrian Chadd > Subject: RE: ath "stops working" in hostap mode > > Hi Adrian, > Ugh, I neglected to provide this updated information: > > FreeBSD RADIO.LOCAL 9.1-STABLE FreeBSD 9.1-STABLE #0 r245019: Fri Jan 4 = 00:02:39 EST 2013 root@RADIO.LOCAL:/usr/obj/usr/src/sys/RADIO amd64 > > ath0@pci0:1:0:0: class=3D0x028000 card=3D0x3a701186 chip=3D0x00241= 68c rev=3D0x01 hdr=3D0x00 > vendor =3D 'Atheros Communications Inc.' > device =3D 'AR5008 Wireless Network Adapter' > class =3D network > > ath0: mem 0xfe9f0000-0xfe9fffff irq 16 at device 0.0 on pc= i1 > ath0: AR5418 mac 12.10 RF2133 phy 8.1 > > NOTE: I claimed that it worked flawlessly in FreeBSD 8.x (and it did) ...= but I did not test 8.x with this kind of device. This bug may manifest its= elf in 8.x with Roku too. But it does seem to correlate with low power devi= ces. I had introduced a phone and tablet to the network when the problem fi= rst appeared (9.0-RELEASE had just been released or was released for a whil= e). I no longer use the tablet and haven't seen this problem in most of the= 2012 year. > > Best regards, > Nathan Lay > ________________________________________ > From: Lay, Nathan > Sent: Saturday, February 02, 2013 1:16 AM > To: Adrian Chadd > Subject: RE: ath "stops working" in hostap mode > > Hi Adrian, > The problem vanished for most of the year and has recently resurfaced. I = just introduced a Roku 2 XD onto the network two days ago. It is a low pow= er device and, as far as I can tell, sits idle on the Wi-Fi network most of= the time. The introduction of this device appears to be responsible for at= h "stopping." > > Just as a recap, the problem NIC is an AR2416 (D-Link DWA 556). It somet= imes drops off of Wi-Fi scans and is unresponsive and requires the driver i= tself being reloaded to restore service. > > One symptom of the problem is that bexmit (from athstats) drops below 5. = Once that happens, ath is "stuck" and the driver must be unloaded and loade= d again to "unstuck" it. This heuristic is so reliable that I employ a cron= script that checks bexmit every minute and it has triggered maybe twice in= the entire year of 2012. And it has triggered about 4-5 times since I intr= oduced Roku 2 two days ago. > > I hope that this new information is helpful. Since it is now quasi-reprod= ucible, I would be happy to help you collect debug information. > > Best regards, > Nathan Lay > > ________________________________________ > From: Lay, Nathan > Sent: Friday, January 06, 2012 9:01 PM > To: Adrian Chadd > Subject: RE: ath "stops working" in hostap mode > > It's fine. I'm horrendously busy too! As I said, I never expected immedia= te responses to my problem. > > I'll do what I can and let you know if I find anything. > > Best Regards, > Nathan Lay > > ________________________________________ > From: Adrian Chadd [adrian.chadd@gmail.com] > Sent: Friday, January 06, 2012 5:52 PM > To: Lay, Nathan > Subject: Re: ath "stops working" in hostap mode > > Hi, > > I've not forgotten about this, I'm just working on other things at the mo= ment. > > Are you able to try older snapshots of a -HEAD kernel to see where I > introduced it? I think that's going to be by far the easiest way to > establish what happened. > > Thanks, > > > > Adrian > > On 2 January 2012 14:51, Lay, Nathan wrote: >> You'll be delighted/disappointed to know that this still happens when co= nfigured to use channel 5:g (as opposed to 5:ht/40). >> >> Thought you would like to know. >> >> Best Regards, >> Nathan Lay >> ________________________________________ >> From: Adrian Chadd [adrian.chadd@gmail.com] >> Sent: Sunday, January 01, 2012 6:52 PM >> To: Lay, Nathan >> Subject: RE: ath "stops working" in hostap mode >> >> Hah it's fine! I want it fixed! >> >> Thanks for chasing this up with me. >> >> Where in the us are you, btw? >> >> >> >> Adrian >> >> >> >> Sent from my Palm Pre on AT&T >> >> ________________________________ >> On Jan 1, 2012 3:37 PM, Lay, Nathan wrote: >> >> Of course! I hope you don't have the impression that I expect immediate = responses! >> >> I'll create the PR now. >> >> Best Regards, >> Nathan Lay >> >> ________________________________ >> From: Adrian Chadd [adrian.chadd@gmail.com] >> Sent: Sunday, January 01, 2012 6:34 PM >> To: Lay, Nathan >> Subject: RE: ath "stops working" in hostap mode >> >> I'm currently out today, lets revisit this tuesday? >> >> Please create a pr. :) >> >> >> Adrian >> >> >> >> Sent from my Palm Pre on AT&T >> >> ________________________________ >> On Jan 1, 2012 3:32 PM, Lay, Nathan wrote: >> >> Shall I check out various revisions and try them out? >> >> Can you name any SVN revisions that are suspect? It also seems to happen= more frequently when debug options are turned on (which is a good thing). >> >> Best Regards, >> Nathan Lay >> >> ________________________________ >> From: Adrian Chadd [adrian.chadd@gmail.com] >> Sent: Sunday, January 01, 2012 6:31 PM >> To: Lay, Nathan >> Subject: RE: ath "stops working" in hostap mode >> >> Ah good! A regression we can track down. >> >> >> Adrian >> >> >> >> Sent from my Palm Pre on AT&T >> >> ________________________________ >> On Jan 1, 2012 2:15 PM, Lay, Nathan wrote: >> >> Of course! I'll do it today! I promise. >> >> However, I should also mention that this bug appears to be introduced. T= his card was very stable in hostap mode on 8.x. I'm glad that I ran into th= is problem if there are so few AR5416 users. >> >> Best Regards, >> Nathan Lay >> >> ________________________________________ >> From: Adrian Chadd [adrian.chadd@gmail.com] >> Sent: Sunday, January 01, 2012 5:00 PM >> To: Lay, Nathan >> Subject: Re: ath "stops working" in hostap mode >> >> On 1 January 2012 13:38, Lay, Nathan wrote: >>> Unloading/reloading the module seems to fix it. >>> >>> I'll just make a cron job that checks for bexmit <=3D 5 and fixes the w= lan if it happens. >> >> That's fine, just please don't forget to create a PR. I'd like to fix >> the driver so you don't _have_ to unload and reload it. >> It's also plausible this sort of bug affects later chips too, as the >> 11n NICs all share code with the AR5416 NIC. >> >> >> Adrian >> >> >> > >