Re: ahc reports lots of pci parity errors

Top Page
Attachments:
Message as email
+ (text/plain)
Delete this message
Reply to this message
Author: Michael Havens
Date:  
To: plug-discuss, benny
CC: freebsd-questions
Subject: Re: ahc reports lots of pci parity errors

Hey buddy, did you ever resolve the data parity error issue? While 
investigating your free-bsd list I discovered another person  had encountered 
the same problem (see 1- below). Their solution seems to implicate  an ISA 
card. However, I found a reason and a solution at the following address:
orange.kame.net/dev/cvsweb.cgi/kame/freebsd5/ sys/dev/aic7xxx/
aic7xxx_pci.c?rev=1.1.1.2
*
     * Disable PCI parity error checking.  Users typically
     * do this to work around broken PCI chipsets that get
     * the parity timing wrong and thus generate lots of spurious
     * errors.  The chip only allows us to disable *all* parity
     * error reporting when doing this, so CIO bus, scb ram, and
     * scratch ram parity errors will be ignored too.
     */


PLUG, so my question to you is how does one disable parity error reporting
and what could be detrimental in doing this?

As I was reviewing this mailing I took heed of the last message at thebottom:

    On lot of motherboards I've seen some PCI slots share a single 
    interrupt. I don't know if this is the case with your motherboard, 
    but if it is then I can certainly imagine it causing problems.


So I will save this message and restart and see if any devices have the same
IRQ......

I opened BIOS and found ACPI IRQ = 10; however that could be changed to 9 or
11. I also found IRQ 10 assigned to PCI/ISA PnP but found I could change that
to Legacy ISA (which I did).....and restarted the computer...... and got the
same error. (did I do this right?)

So now I will unplug the ISA card..... this isn't the problem.

I now am certain that the data parity timing is wrong.
--
<:-)Mike(-:>

1-
Hiho! :-)

I'm getting lots of those, but my system is stable.

ahc0: PCI error Interrupt at seqaddr = 0xffff
ahc0: Signaled a Target Abort
ahc0: WARNING WARNING WARNING WARNING
ahc0: Too many PCI parity errors observed as a target.
ahc0: Some device on this bus is generating bad parity.
ahc0: This is an error *observed by*, not *generated by*, this
controller.
ahc0: PCI parity error checking has been disabled.
ahc0: WARNING WARNING WARNING WARNING
ahc0: PCI error Interrupt at seqaddr = 0xffff


I'm running 5.2.1-RELEASE with ACPI disabled.

Usually I'm getting one or two errors over a period of two weeks or
longer, but since I'm using my old ISA Creative SB32, I'm getting lots
in a very short period of time.

What could cause that? A broken PCI-card? Motherboard?

TIA :-)

Bye
Marc

----------------------------

Maybe a problem with the ISA-to-PCI bridge on your motherboard, or else a
timing problem. You aren't playing with the bus frequency multipliers,
overclocking, or anything like that, right?

Dump the ISA soundcard for a PCI one and you'll probably do better...

--
-Chuck

----------------------------

> Maybe a problem with the ISA-to-PCI bridge on your motherboard, or
> else a timing problem. You aren't playing with the bus frequency
> multipliers, overclocking, or anything like that, right?


Nope, nothing at all. It's a Tyan Thunder X with two PII-Xeon 450Mhz,
nothing overclocked.

> Dump the ISA soundcard for a PCI one and you'll probably do better...


Yep, just removed it.

But as i wrote above I still get sporadic parity errors even without
that old isa-card.

Any idea what might cause those?

Bye
Marc

----------------------------

Hmm, I narrowed it down.

It starts going havoc if there are more than 4 PCI cards inserted. If
the fifth card is a PCI card, I get lots of errors and even lockups
really soon, if the fifth card is an ISA card, i get less errors and
I've only encountered one lockup until now.

I've two NICs and two graphic-cards (all PCI), ide-controller onboard
and scsi controller onboard.

Is this behaviour something that might be expected with this kind of
setup or is it a sign of defective/failing hardware (buggy/broken
southbridge, as Chuck suggested)?

TIA :-)

Bye
Marc

----------------------------

On lot of motherboards I've seen some PCI slots share a single interrupt. I
don't know if this is the case with your motherboard, but if it is then I can
certainly imagine it causing problems.
--
Toomas Aas






This message has been scanned for viruses by the VEI Internet
Automatic Email Spam and Virus Scanner, and is believed to be free of spam or viruses.
Please report spam to . If you would like 98.9 % spam blocked from your
E-mail then go to VEI Internet for details. Anti-spam/Anti-virus is FREE with every account.


http://www.vei.net/

---------------------------------------------------
PLUG-discuss mailing list -
To subscribe, unsubscribe, or to change you mail settings:
http://lists.PLUG.phoenix.az.us/mailman/listinfo/plug-discuss