[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-smp
Subject:    IRQ deadlocks, PR440FX motherboard, 2.0.35
From:       System Administrator <admin () intergrafix ! net>
Date:       1998-09-22 19:24:21
[Download RAW message or body]


Hi,

I'm moving this over to linux-smp after its brief discussion on the 
aic7xxx mailing list.

I'm running caldera linux with kernel 2.0.35 on an Intel Providence
Motherboard with 2 PPro 200s, adaptec 7880 SCSI chipset with doug's latest
pre10 driver, and the onboard etherexpress pro100B. Also running the
latest Intel BIOS, set for the MP 1.4 spec, and I/O APICs

My machine has been freezing up (almost daily now) with 
 IRQ DEADLOCK DETECTED BY CPU 0 and
 IRQ DEADLOCK DETECTED BY CPU 12

Syslog doesn't report anything. This has been ongoing (kernel 2.0.31, .33,
.34)

procinfo reports the following for the interrupts:
irq  0:   3350878   timer               irq  8:         0 + rtc                
irq  1:        46   keyboard            irq  9:         0                      
irq  2:         0   cascade             irq 10:         0                      
irq  3:         0                       irq 11:         0                      
irq  4:         0                       irq 12:         0   PS/2 Mouse         
irq  5:    148884   aic7xxx             irq 13:         1   math error         
irq  6:         2                       irq 14:    804596   Intel
EtherExpress 
irq  7:         0                       irq 15:         0 ress                 

dmesg reports for the processors:
Sep 22 04:19:58 cygnus kernel: Cannot find map file.
Sep 22 04:19:58 cygnus kernel: Intel MultiProcessor Specification v1.4
Sep 22 04:19:58 cygnus kernel:     Virtual Wire compatibility mode.
Sep 22 04:19:58 cygnus kernel: OEM ID: INTEL    Product ID: PR440FX
APIC at: 0xFEC08000
Sep 22 04:19:58 cygnus kernel: Processor #0 Pentium(tm) Pro APIC version 17
Sep 22 04:19:58 cygnus kernel: Processor #12 Pentium(tm) Pro APIC version 17
Sep 22 04:19:58 cygnus kernel: I/O APIC #13 Version 17 at 0xFEC00000.
Sep 22 04:19:58 cygnus kernel: Processors: 2
Sep 22 04:19:58 cygnus kernel: pcibios_init : BIOS32 Service Directory
structure at 0x000fd9e0
Sep 22 04:19:58 cygnus kernel: pcibios_init : BIOS32 Service Directory
entry at 0xfd9f0
Sep 22 04:19:58 cygnus kernel: pcibios_init : PCI BIOS revision 2.10 entry
at 0xfda11

The 2 servers this affects seem to freeze under heavy loads (news, which
always has a high load, and www, USUALLY during backups to a SCSI tape
drive, nfs is also a big reliance)

my latest theorey comes from Documentation/rtc.txt under the kernel source
tree which talks about SMP being affected by the system clock and how it's
run. This concerns me because I sync my time via Caldera's server via
xntpd/ntpdate. Perhaps this is causing the problem.


Thanx,

-Tony
.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-.
Anthony J. Biacco                           Network Administrator/Engineer
admin@intergrafix.net                        Intergrafix Internet Services

          "The best way to predict the future, is to invent it."
http://cygnus.ncohafmuta.com                    http://www.intergrafix.net
.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-.

[prev in list] [next in list] [prev in thread] [next in thread]