fxp0: device timeout | SCB already complete (me too)

Shaun Jurrens shaun.jurrens at skoleetaten.oslo.no
Thu Jun 5 02:15:50 PDT 2003 

On Wed, Jun 04, 2003 at 06:32:46PM +0200, Palle Girgensohn wrote:
#> Hi Shaun,
#> 
#> Thanks for the input! Glad to hear I'm not the only one
#> 
#> In my case, both the SCSI and NIC are integrated on the motherboard, so I 
#> cannot really move them around... :)
#> 
#> Also, as I mentioned, I tried a de0 (PCI card, not onboard, and it 
#> literally stopped the machine). Is the de0 driver also a problem?
#> 
#> /Palle

	I'm beginning to think it's a scsi problem of sorts as well so
	I clipped -hardware and Cc'd -scsi on this. I just happed to 
	(unfortunately) run into this on another box yesterday after
	four months of relative quiet. I happened to be moving an 
	interface over from some crap Nortel switch to a nice Cisco
	switch and promptly a different interface began to do it's dance.
	It's the same interface each time (and I've changed cards...)
	Anyway, for the record, a little from messages:

Jun  2 18:48:43 nol33n0x /kernel: fxp0: Microcode loaded, int_delay: 1000 usec  
bundle_max: 6
Jun  4 16:57:50 nol33n0x /kernel: fxp1: SCB timeout: 0x80 0xe0 0x50 0x0
Jun  4 16:57:51 nol33n0x last message repeated 4 times
Jun  4 16:57:51 nol33n0x /kernel: fxp1: SCB timeout: 0x80 0xe0 0x50 0x400
Jun  4 16:57:58 nol33n0x /kernel: fxp1: SCB timeout: 0x80 0xe0 0x50 0x0
Jun  4 16:57:58 nol33n0x last message repeated 3 times
Jun  4 16:57:58 nol33n0x /kernel: fxp1: SCB timeout: 0x80 0xe0 0x50 0x400
Jun  4 16:57:58 nol33n0x last message repeated 110 times
Jun  4 16:58:17 nol33n0x /kernel: fxp1: SCB timeout: 0x80 0xe0 0x90 0x400
Jun  4 16:58:20 nol33n0x last message repeated 17 times
Jun  4 17:09:04 nol33n0x /kernel: fxp1: SCB timeout: 0x80 0xe0 0x90 0x400
Jun  4 17:09:09 nol33n0x last message repeated 2 times
Jun  4 17:09:09 nol33n0x /kernel: fxp1: SCB timeout: 0x80 0xe0 0x90 0x0
Jun  4 17:09:12 nol33n0x last message repeated 3 times
Jun  4 17:09:12 nol33n0x /kernel: fxp1: SCB timeout: 0x80 0xe0 0x90 0x400
Jun  4 17:09:39 nol33n0x last message repeated 22 times
Jun  4 17:18:18 nol33n0x login: ROOT LOGIN (root) ON ttyv0
Jun  4 17:19:21 nol33n0x /kernel: fxp1: DMA timeout
Jun  4 17:19:21 nol33n0x /kernel: fxp1: Microcode loaded, int_delay: 1000 usec  
bundle_max: 6
Jun  4 17:19:21 nol33n0x /kernel: fxp1: DMA timeout
Jun  4 17:19:21 nol33n0x /kernel: fxp1: SCB timeout: 0x10 0x0 0x80 0x0
Jun  4 17:19:21 nol33n0x /kernel: fxp1: DMA timeout
Jun  4 17:19:21 nol33n0x /kernel: fxp1: SCB timeout: 0x10 0x0 0x80 0x0
Jun  4 17:19:21 nol33n0x /kernel: fxp1: DMA timeout
Jun  4 17:19:21 nol33n0x /kernel: fxp1: SCB timeout: 0x10 0x0 0x80 0x0
Jun  4 17:19:21 nol33n0x /kernel: fxp1: SCB timeout: 0x10 0x0 0x80 0x0
Jun  4 17:19:37 nol33n0x /kernel: fxp1: command queue timeout
Jun  4 17:19:46 nol33n0x /kernel: fxp1: SCB timeout: 0x1 0x0 0x80 0x400
Jun  4 17:19:46 nol33n0x /kernel: fxp1: SCB timeout: 0x81 0x0 0x80 0x400
Jun  4 17:19:58 nol33n0x last message repeated 37 times

...

Jun  4 17:24:21 nol33n0x /kernel: fxp1: SCB timeout: 0x80 0xe0 0x90 0x0
Jun  4 17:24:21 nol33n0x last message repeated 8 times
Jun  4 17:24:21 nol33n0x /kernel: fxp1: SCB timeout: 0x80 0xe0 0x90 0x400
Jun  4 17:24:37 nol33n0x last message repeated 115 times

	After that the box didn't find 3 of the 5 fxp nic's until a new
	boot and a cleared ESCD. Not sure why an fxp card should bitch 
	about SCB's anyway.  I'd be grateful for any pointers here. 
	FreeBSD is on it's way out on firewalls here otherwise because 
	I'm catching a good deal of heat about it.  More info is available 
	on request.

#> 
#> 
#> 
#> --On tisdag, juni 03, 2003 17.21.23 +0200 Shaun Jurrens 
#> <shaun.jurrens at skoleetaten.oslo.no> wrote:
#> 
#> >I hate to say it, but I've had these for months starting at 4.6-stable
#> >and continuing up to at least the latest 4.7-RRELEASE-p* . I have one
#> >dual -current box that has exibited the same behaviour as well.
#> >
#> >The boxes work just fine with the xl0 driver. Lots of different
#> >motherboards and processors (all PIII) and a number of different Intel
#> >card revisions. I can't run my squid boxes on fxp cards _at all_ for
#> >example, the fxp driver will take the box down with it. On my firewalls
#> >it's locked up the  interfaces numerous times.
#> >
#> >The only suggestion I can offer at the moment is to try various card
#> >placements over your PCI slots. I've found stability using one of the
#> >first two slots for my Adaptec controller (2940U[2]W, 29160[N]) and the
#> >rest for the Intel nics.  This happens both with or without POLLING
#> >enabled. I've tried a number of combinations of POLLING enabled/disabled,
#> >not  compiled in and different HZ settings. Obviously no POLLING on my
#> >SMP  boxes.
#> >
#> >I know one or two others that have had problems with this too, but
#> >haven't  had the time or equipment at hand to work with any developers on
#> >getting this fixed. I guess I got the equipment now (various PIII UP/SMP
#> >boards from Gigabyte, Asus) and a little time if anyone wants to bite.
#> >
#> >My guess is that the POLLING commits broke something, but that's just a
#> >guess. I don't have any dc cards here, and no one has ever complained
#> >about either them or the rl cards timing out.  There also seems to be
#> >a definite correlation between the fxp problem and the ahc driver.
#> >
#> >Ok, the rest of the "me too's" should now chime in with a bit of time
#> >and energy. There's also a PR open on this: kern/45568 .
#> >
#> >
#> >
#> >--
#> 

-- 
Med vennlig hilsen/Sincerely,

Shaun D. Jurrens
Drift og Sikkerhetskonsulent
IKT-Avdeling
Oslo Skoleetaten

gpg key fingerprint: 007A B6BD 8B1B BAB9 C583  2D19 3A7F 4A3E F83E 84AE
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-net/attachments/20030605/0213e3b9/attachment.bin

More information about the freebsd-net mailing list