amd64/93002: amd64 (6.0) coredumps at unpredictable times

[Doug White]

On Tue, 7 Feb 2006, Steve Rieger wrote:

> Fatal trap 1: privileged instruction fault while in kernel mode
> instruction pointer     = 0x8:0xffffffff8040d5ea
> stack pointer           = 0x10:0xffffffffb54df6d0
> frame pointer           = 0x10:0xffffffffa5171000
> code segment            = base 0x0, limit 0xfffff, type 0x1b
>                          = DPL 0, pres 1, long 1, def32 0, gran 1
> processor eflags        = interrupt enabled, resume, IOPL = 0
> current process         = 60492 (as)
> trap number             = 1
> panic: privileged instruction fault

This is a very atypical trap. If you were running -CURRENT it'd indicate a
bug, but the location of the trap isn't around any sort of privilieged
instruction which leads me to believe you have a hardware issue.

To check, can you please load this dump back into gdb and run:

disass 0xffffffff8040d5ea

This will print a disassembly of the function at that point. Please post
the output of the first screen or so. That may help us identify what the
faulting instruction was. If that instruction can't throw that type of
exception then we can eliminate a software bug.

> pcib0: <ACPI Host-PCI bridge> port 0xcf8-0xcff,0xcf0-0xcf3 on acpi0
> pci_link29: BIOS IRQ 12 for -2145774616.1.INTA is invalid
> pci_link23: BIOS IRQ 10 for -2145774616.2.INTA is invalid
> pci_link24: BIOS IRQ 10 for -2145774616.2.INTB is invalid
> pci_link30: BIOS IRQ 10 for -2145774616.2.INTC is invalid

This is additionally scary, as it looks like there is bad bugs in the ACPI
tables in the BIOS.

If you haven't already, please upgrade the BIOS on your system to the
latest, and verify that the memory installed in the system is the type
specified by the board vendor.

-- 
Doug White                    |  FreeBSD: The Power to Serve
dwhite at gumbysoft.com          |  www.FreeBSD.org