NetBSD Problem Report #51660

From www@NetBSD.org  Sat Nov 26 23:20:28 2016
Return-Path: <www@NetBSD.org>
Received: from mail.netbsd.org (mail.netbsd.org [199.233.217.200])
	(using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits))
	(Client CN "mail.netbsd.org", Issuer "Postmaster NetBSD.org" (verified OK))
	by mollari.NetBSD.org (Postfix) with ESMTPS id 9C31F7A2D6
	for <gnats-bugs@gnats.NetBSD.org>; Sat, 26 Nov 2016 23:20:28 +0000 (UTC)
Message-Id: <20161126232027.757EF7A33E@mollari.NetBSD.org>
Date: Sat, 26 Nov 2016 23:20:27 +0000 (UTC)
From: venture37@geeklan.co.uk
Reply-To: venture37@geeklan.co.uk
To: gnats-bugs@NetBSD.org
Subject: an(4) fails to load with Aironet 340 or 350 
X-Send-Pr-Version: www-1.0

>Number:         51660
>Category:       kern
>Synopsis:       an(4) fails to load with Aironet 340 or 350
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sat Nov 26 23:25:00 +0000 2016
>Last-Modified:  Mon Nov 28 08:00:02 +0000 2016
>Originator:     Sevan Janiyan
>Release:        NetBSD-head
>Organization:
>Environment:
NetBSD 7.99.42 i386

>Description:
Attaching a PCMCIA Aironet 340 or 350 card results in the kernel reporting:
an0 at pcmcia0 function 0: <Cisco Systems, 340 Series Wireless LAN Adapter>
an0: failed to load config data
an0: failed to attach controller
an0: detached

>How-To-Repeat:
Insert a PCMCIA card into a system running NetBSD
>Fix:

>Audit-Trail:
From: Martin Husemann <martin@duskware.de>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: kern/51660: an(4) fails to load with Aironet 340 or 350
Date: Sun, 27 Nov 2016 17:06:10 +0100

 Do other pcmcia cards work with this particular machine?

 I have seen similar issues when the pcmcia bridge was using "bad"
 address ranges. Please show dmesg at least for the pcmcia bridge.

 Martin

From: Sevan Janiyan <venture37@geeklan.co.uk>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: kern/51660: an(4) fails to load with Aironet 340 or 350
Date: Sun, 27 Nov 2016 16:40:37 +0000

 Hi Martin,

 On 27/11/2016 16:10, Martin Husemann wrote:
 >  Do other pcmcia cards work with this particular machine?
 >  
 >  I have seen similar issues when the pcmcia bridge was using "bad"
 >  address ranges. Please show dmesg at least for the pcmcia bridge.


 Full dmesg is here
 http://www.netbsd.org/~sevan/x60s-dmesg.txt

 Bits I thought might be relevant:
 pci0 at mainbus0 bus 0: configuration mode 1
 pci0: i/o space, memory space enabled, rd/line, rd/mult, wr/inv ok
 ppb4 at pci0 dev 30 function 0: vendor 8086 product 2448 (rev. 0xe2)
 pci5 at ppb4 bus 5
 pci5: i/o space, memory space enabled
 cbb0 at pci5 dev 0 function 0: vendor 1180 product 0476 (rev. 0xb4)
 cbb0: cacheline 0x0 lattimer 0x40
 cbb0: bhlc 0x824000
 cbb0: interrupting at ioapic0 pin 16
 cardslot0 at cbb0
 cardbus0 at cardslot0: bus 6
 pcmcia0 at cardslot0


 Sevan

From: Sevan Janiyan <venture37@geeklan.co.uk>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: kern/51660: an(4) fails to load with Aironet 340 or 350
Date: Sun, 27 Nov 2016 16:49:28 +0000

 On 27/11/2016 16:10, Martin Husemann wrote:
 > Do other pcmcia cards work with this particular machine?

 All the cards I have a range of different problems from being
 unsupported, not working or causing system to lock up (ath(4) in hostap
 mode).

 I have another system running various different operating systems which
 I can test the cards on even. May have to resort to running XP for some
 the vintage hardware to work but I can give it a shot over the next week.

 Sevan

From: Sevan Janiyan <venture37@geeklan.co.uk>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: kern/51660: an(4) fails to load with Aironet 340 or 350
Date: Mon, 28 Nov 2016 02:43:15 +0000

 On 27/11/2016 16:10, Martin Husemann wrote:
 > Do other pcmcia cards work with this particular machine?

 I connected the cards to my HP Jornada 720 running the latest -HEAD
 hpcarm snapshot.

 There the an(4) & wi(4) drivers bind without issue. Cards are still not
 functioning however.
 bringing the interfaces up & issuing list scan resultes in to output
 about found access points. siginfo reports netio.

 I booted an old OpenBSD-current i386 snapshot on the ThinkPad, the cards
 work there & running a scan reports access points by both cards.



 Sevan

From: Robert Elz <kre@munnari.OZ.AU>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: kern/51660: an(4) fails to load with Aironet 340 or 350
Date: Mon, 28 Nov 2016 13:17:42 +0700

 My impression has been that support for (at least wireless) pcmcia
 broke sometime back about NetBSD 2 or 3 (4?), when cardbus support was added.

 I used to have some of the original (Lucent?) wireless pcmcia cards
 (used the wi driver - the erliest running BSDi in pre NetBSD days)
 which as I recall (not a reliable source) was the only way to do wireless
 back then (there weren't even competing vendors originally as I remember it).
 They worked fine in my (at the time) pcmcia only laptop, and still when the
 laptop changed to one that supported cardbus and pcmcia via a bridge (but
 with NetBSD still only supporting pcmcia, which was no problem to me.)

 I think that laptop (if not that one, then the next) also had built in
 wireless, so the pcmcia cards (which were 802.11b only I think) started
 getting used a lot less, but sometime later (with the same laptop, same
 cards, NetBSD all the time, but a newer version) simply didn't work any more.
 (I had a few of the cards, so it wasn't just that the card broke.)

 Since need for pcmcia wireless was very rare by that time - and even rarer
 now, my last 2, maybe 3, laptops haven't supported it (that is, pcmcia/cardbus,
 unless there's some USB dongle I could plug in, not that I want to) at all -
 this never really seemed very important.

 Not sure if I ever filed a PR about it - maybe - but it would have been more
 than 10 years ago if I did, maybe closer to 15, maybe more - but by now if
 there was one it would have been closed (I presume) as relating to a long
 unsupported kernel.  [Actually I did: kern/39838 (only 8 years ago) - still
 open - and that one was about 4.99.73 but apparently goes back into earlier
 4.99.* versions, I just did a search for pcmcia related PRs, there are LOTS,
 many still open.]

 I never recall seeing anything committed that sounded like it would
 be a fix.  I no longer have any way to test this (no idea where the cards
 are, nor a working laptop that they could be plugged into.)

 Whether the problem applied to all pcmcia, or only cardbus/pcmcia bridges,
 or only those implemented in some odd way, or only to wireless cards (never
 used any other kind of pcmcia card in any serious way) I have no idea, the
 point of this message is just that I suspect this problem is a very very old
 one.

 kre

From: Martin Husemann <martin@duskware.de>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: kern/51660: an(4) fails to load with Aironet 340 or 350
Date: Mon, 28 Nov 2016 08:56:40 +0100

 On Sun, Nov 27, 2016 at 04:50:00PM +0000, Sevan Janiyan wrote:
 >  All the cards I have a range of different problems from being
 >  unsupported, not working or causing system to lock up (ath(4) in hostap
 >  mode).
 >  
 >  I have another system running various different operating systems which
 >  I can test the cards on even. May have to resort to running XP for some
 >  the vintage hardware to work but I can give it a shot over the next week.

 The question was more: does the cardbus/pcmcia bridge work for pcmcia cards
 at all? The symptoms you saw could be explained if the pcmcia bus spaces
 were not mapped at the right place or inaccessible (and I have seen that
 happen in the distant past.

 Or put the other way around: are we looking at a problem in the pcmcia driver
 or the pcmcia/cardbus bridge driver. If the same card works better in your
 Jornada running NetBSD current this gives a strong hint.

 Martin

From: Martin Husemann <martin@duskware.de>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: kern/51660: an(4) fails to load with Aironet 340 or 350
Date: Mon, 28 Nov 2016 08:59:27 +0100

 wi0@pcmcia worked for me last I tried (maybe 1 year ago) in a sparc system.
 I guess the lossage you mention is cardbus/acpi bus space mapping related
 (and not directly connected to the wi driver) - probably the same issue
 Sevan is seeing.

 I'll retest my wi0 later this week.

 Martin

Home	PR Database Search