Date: Tue, 27 Feb 2001 08:47:12 +0100 (CET) From: Remy Nonnenmacher <remy@boostworks.com> To: evenson@panix.com Cc: stable@freebsd.org Subject: Re: fxp0: SCB timeout Message-ID: <200102270745.IAA79608@luxren2.boostworks.com> In-Reply-To: <a0k86dulmq.fsf@panix6.panix.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 26 Feb, Mark Evenson wrote:
> Remy Nonnenmacher <remy@boostworks.com> writes:
>
>> Hello,
>>
>> I recently upgraded a machine from 4.2-RELEASE (was working fine) to
>> 4.2-STABLE and i now get messages from fxp driver:
>>
>> fxp0: SCB timeout
>>
>> this hangs nfs but network keeps going on (with some short pauses
>> anyway). The motherboard is an Intel STL2, running SMP, with integrated
>> 82559 chip.
>>
>
> I ran into the same problem on a Dell Dimension 4100 whose onboard Ethernet
> is reported via dmesg as
>
> fxp0: <Intel PLC 10/100 Ethernet> port 0xdf00-0xdf3f mem 0xfeaff000-0xfeafffff irq 3 at device 8.0 on pci2
> fxp0: Ethernet address 00:03:47:0e:70:be
>
> And by support.dell.com with my service tag as:
>
> Integrated Intel 10/100 Ethernet Controller Driver (Intel 82562ET chip set)
>
>
> This error would occur at a random interval after the network interface was
> brought up and used, usually within ten minutes. The failure seemed to
> correspond loosely to the number of packets that were pumped through the
> interface. The failure was a total lack of IP traffic until rebooted, not
> just some network traffic.
>
> I slapped a rather old PCI fxp card in the box that I had lying around my
> house to get around the problem, which works fine.
>
> Since I am running this hardware as a contract Java programmer, I can't
> really justify spending the time to go through the drivers to figure out
> what is wrong.
>
> This error occurs both on 4_2_RELEASE (which I installed) and 4_STABLE
> (which I am tracking currently).
>
> I would gladly serve as a test point if someone develops code that they
> think addresses this problem.
>
I also added an 82558 based card and the problem got fixed. After some
checking, it appears that the CSB timeout problem raises in combination
with nfs:
It can have the machine receive any amount of data via network with
high disk activity (like de-tarring an rsh-ed flow) but is soon as i
mount a remote nfs tree (like /usr/ports/distfiles) and try to tar -x
something from there, the message appears and the nfs mount locks
solid. Machine remains available from network for normal operation with
some utilities being trapped by the lock (df, for example).
In fact, I get the problem with 4.2-RELEASE now. What is strange is
that i'm _sure_ it worked soon after installing a fresh CD image and
just recompiling a kernel for SMP. The problem appeared after going
back and forth between 4.2-STABLE and -RELEASE with a full world build.
FYI: net cards here:
STL2 integrated fxp: (82559 based, Problem with NFS)
fxp0: <Intel Pro 10/100B/100+ Ethernet> port 0x5400-0x543f
mem 0xfb000000-0xfb0fffff,0xfb101000-0xfb101fff irq 9 at device 3.0 on pci0
Added card: (82558 based, Works fine with NFS)
pcib2: <DEC 21152 PCI-PCI bridge> at device 10.0 on pci1
pci2: <PCI bus> on pcib2
fxp0: <Intel Pro 10/100B/100+ Ethernet> port 0x7000-0x701f
mem 0xfd100000-0xfd1fffff,0xfd500000-0xfd500fff irq 5 at device 4.0 on pci2
fxp0: Ethernet address 00:03:47:42:53:43
fxp1: <Intel Pro 10/100B/100+ Ethernet> port 0x7020-0x703f
mem 0xfd200000-0xfd2fffff,0xfd501000-0xfd501fff irq 2 at device 5.0 on pci2
fxp1: Ethernet address 00:03:47:42:53:44
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200102270745.IAA79608>