From owner-freebsd-hackers  Mon Jul  7 01:29:42 1997
Return-Path: <owner-hackers>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.5/8.8.5) id BAA02628
          for hackers-outgoing; Mon, 7 Jul 1997 01:29:42 -0700 (PDT)
Received: from implode.root.com (implode.root.com [198.145.90.17])
          by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id BAA02622
          for <FreeBSD-Hackers@FreeBSD.ORG>; Mon, 7 Jul 1997 01:29:38 -0700 (PDT)
Received: from implode.root.com (localhost [127.0.0.1])
	by implode.root.com (8.8.5/8.8.5) with ESMTP id BAA23800;
	Mon, 7 Jul 1997 01:31:15 -0700 (PDT)
Message-Id: <199707070831.BAA23800@implode.root.com>
To: Simon Shapiro <Shimon@i-Connect.Net>
cc: FreeBSD-Hackers@FreeBSD.ORG
Subject: Re: PCI Bridge Question 
In-reply-to: Your message of "Mon, 07 Jul 1997 00:16:52 PDT."
             <XFMail.970707001652.Shimon@i-Connect.Net> 
From: David Greenman <dg@root.com>
Reply-To: dg@root.com
Date: Mon, 07 Jul 1997 01:31:15 -0700
Sender: owner-hackers@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

>> >fxp1 <Intel EtherExpress Pro 10/100B Ethernet> rev 2 int a irq 10 on
>> pci1:0
>> >pci_map_mem failed: device's memrange 0xff9ff000-0xff9fffff is
>> incompatible
>> >with
>> > its bridge's memrange 0xfd200000-0xfd6fffff
>> >fxp1: Ethernet address 00:a0:c9:6b:8e:b6
>> >
>> >    What can I do to make this work?
>> 
>>    Did you verify that it actually doesn't work? From the above, the
>> device appears to have been mapped and attached okay.
>
>Yup.  Dead as a dead horse.  The other card is very alive.
>This problem does not exist under 3.0-current 9as of today).

   Then what version was it failing under?

>BTW, under SMP, this driver panics.  This happens under NFS load.
>What one sees is:
>
>Page fault on cpuid = 0;
>
>It blows on fxp_add_rfabuf + 0xc7.
>
>We get there via:
>
>fxp_add_rfabuf
>fxp_intr

   Hmmm. That function just allocates mbufs/mbuf clusters and adds them to the
end of a chain. It appears that the index into the mclrefcnt array is causing
the panic - apparantly because the mbuf cluster pointer value, which was just
gotten from mclfree, is bogus. This might indicate that something that wasn't
an mbuf cluster had been freed onto the mclfree list. In any case, this
appears to be a much more generic problem - not specific to the fxp device
driver.

-DG

David Greenman
Core-team/Principal Architect, The FreeBSD Project