kern/51251: ACPI device probe raises NMI

[hannken%eis.cs.tu-bs.de@localhost]

>Number:         51251
>Category:       kern
>Synopsis:       ACPI device probe raises NMI
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu Jun 16 16:20:00 +0000 2016
>Originator:     Juergen Hannken-Illjes
>Release:        NetBSD 7.99.30
>Organization:
	
>Environment:
	
	
System: NetBSD toaster.dd 7.99.30 NetBSD 7.99.30 (work.amd64) #2: Thu Jun 16 17:38:28 MEST 2016  hannken%builder.isf.cs.tu-bs.de@localhost:/work/build/obj/obj.amd64/sys/arch/amd64/compile/work.amd64 amd64
Architecture: x86_64
Machine: amd64
>Description:
	
Booting an IBM eserver xSeries 336 -[883721Y]- I get a NMI during PCI bus probe:

total memory = 16383 MB
avail memory = 15871 MB
SMBIOS rev. 2.3 @ 0xf5f9e (52 entries)
mainbus0 (root)
efi: missing or invalid systbl
ACPI: RSDP 0x00000000000FDFB0 000024 (v02 IBM   )
ACPI: XSDT 0x00000000D7FCFF00 00003C (v01 IBM    SERONYXP 00001001 IBM  45444F43)
ACPI: FACP 0x00000000D7FCFE40 000084 (v02 IBM    SERONYXP 00001001 IBM  45444F43)
ACPI: DSDT 0x00000000D7FCCF00 002346 (v02 IBM    SERTURQU 00001000 INTL 20041203)
ACPI: FACS 0x00000000D7FCFD00 000040
ACPI: APIC 0x00000000D7FCFD80 00009C (v01 IBM    SERONYXP 00001001 IBM  45444F43)
ACPI: MCFG 0x00000000D7FCFD40 00003C (v01 IBM    SERONYXP 00001001 IBM  45444F43)
ACPI: 1 ACPI AML tables successfully acquired and loaded

efi: missing or invalid systbl
ioapic0 at mainbus0 apid 14
ioapic1 at mainbus0 apid 13
ioapic2 at mainbus0 apid 12
cpu0 at mainbus0 apid 0
cpu0: 32 page colors
cpu0: calibrating local timer
cpu0: apic clock running at 200 MHz
cpu0: Intel(R) Xeon(TM) CPU 3.20GHz, id 0xf41
cpu0: PAT enabled
cpu1 at mainbus0 apid 6
cpu1: 2 page colors
x86_ipi_init: ESR 00000040
cpu1: Intel(R) Xeon(TM) CPU 3.20GHz, id 0xf41
cpu1: PAT enabled
cpu2 at mainbus0 apid 1
cpu2: 2 page colors
cpu2: Intel(R) Xeon(TM) CPU 3.20GHz, id 0xf41
cpu2: PAT enabled
cpu3 at mainbus0 apid 7
cpu3: 2 page colors
cpu3: Intel(R) Xeon(TM) CPU 3.20GHz, id 0xf41
cpu3: PAT enabled
acpi0 at mainbus0: Intel ACPICA 20160422
efi: missing or invalid systbl
acpi0: MCFG: segment 0, bus 0-7, address 0x00000000e0000000
MCFG: MEMMAP: 0x0000000000000000-0x000000000009d3ff, size=0x000000000009d400, type=1(Memory)
MCFG: MEMMAP: 0x000000000009d400-0x000000000009ffff, size=0x0000000000002c00, type=2(Reserved)
MCFG: MEMMAP: 0x00000000000e0000-0x00000000000fffff, size=0x0000000000020000, type=2(Reserved)
MCFG: MEMMAP: 0x0000000000100000-0x00000000d7fcceff, size=0x00000000d7eccf00, type=1(Memory)
MCFG: MEMMAP: 0x00000000d7fd0000-0x00000000d7ffffff, size=0x0000000000030000, type=2(Reserved)
MCFG: MEMMAP: 0x00000000d7fccf00-0x00000000d7fcffff, size=0x0000000000003100, type=3(ACPI)
MCFG: MEMMAP: 0x0000000100000000-0x0000000427ffffff, size=0x0000000328000000, type=1(Memory)
MCFG: MEMMAP: 0x00000000e0000000-0x00000000efffffff, size=0x0000000010000000, type=2(Reserved)
acpi0: fixed power button present
acpi0: ACPI-Fast 24-bit timer
SIOM (PNP0C02) at acpi0 not configured
pckbc1 at acpi0 (PS2K, PNP0303) (kbd port): io 0x64,0x60 irq 1
pckbc2 at acpi0 (PS2M, PNP0F13) (aux port): irq 12
COM2 (PNP0501) at acpi0 not configured
PIC2 (PNP0003) at acpi0 not configured
PIC3 (PNP0003) at acpi0 not configured
PIC4 (PNP0003) at acpi0 not configured
attimer1 at acpi0 (TMR, PNP0100): io 0x40-0x43 irq 0
pcppi1 at acpi0 (SPKR, PNP0800): io 0x61
midi0 at pcppi1: PC speaker
sysbeep0 at pcppi1
SBD1 (PNP0C02) at acpi0 not configured
pckbdprobe: reset error 5
pmsprobe: reset error 5
attimer1: attached to pcppi1
ipmi0 at mainbus0
ipmi0: WARNING: power management not supported
pci0 at mainbus0 bus 0: configuration mode 1
[ probe 000:00:0 ]
[ ffff80011e331000 + 100 ]
[ ffff80011e331000 + 100 ]
acpi0: MCFG: 000:00:0: Ok (cfg[0x100]=0x00000000)
[ probe 000:00:1 ]
[ ffff80011e332000 + 100 ]
[ ffff80011e332000 + 100 ]
acpi0: MCFG: 000:00:1: Ok (cfg[0x100]=0x00000000)
[ probe 000:02:0 ]
[ ffff80011e341000 + 100 ]
[ ffff80011e341000 + 100 ]
acpi0: MCFG: 000:02:0: Ok (cfg[0x100]=0x00010001)
[ probe 000:04:0 ]
[ ffff80011e351000 + 100 ]
[ ffff80011e351000 + 100 ]
acpi0: MCFG: 000:04:0: Ok (cfg[0x100]=0x00010001)
[ probe 000:06:0 ]
[ ffff80011e361000 + 100 ]
[ ffff80011e361000 + 100 ]
acpi0: MCFG: 000:06:0: Ok (cfg[0x100]=0x00010001)
[ probe 000:07:0 ]
[ ffff80011e369000 + 100 ]
[ ffff80011e369000 + 100 ]
acpi0: MCFG: 000:07:0: Ok (cfg[0x100]=0x00010001)
[ probe 000:08:0 ]
[ ffff80011e371000 + 100 ]
[ ffff80011e371000 + 100 ]
acpi0: MCFG: 000:08:0: Ok (cfg[0x100]=0x00000000)
[ probe 000:29:0 ]
[ ffff80011e419000 + 100 ]
NMI ... going to debugger

The lines reading "[ probe ..." come from a printf in acpimcfg_device_probe()
before it probes extended configuration space.

The lines reading "[ fff..." come from a printf in acpi_md_mcfg_read()
before it does "__asm(...", they print "bsh" and "addr".

With "pcictl pci0 list" I get:

000:00:0: Intel E7525 Memory Controller Hub (host bridge, revision 0x0c)
000:00:1: Intel E7525 Error Reporting Device (undefined, subclass 0x00, revision 0x0c)
000:02:0: Intel E7525 PCI Express Port A (PCI bridge, revision 0x0c)
000:04:0: Intel E7525 PCI Express Port B (PCI bridge, revision 0x0c)
000:06:0: Intel E7520 PCI Express Port C (PCI bridge, revision 0x0c)
000:07:0: Intel E7520 PCI Express Port C1 (PCI bridge, revision 0x0c)
000:08:0: Intel E7520 Extended Configuration (miscellaneous system, revision 0x0c)
000:29:0: Intel 82801EB/ER USB UHCI Controller (USB serial bus, UHCI, revision 0x02)
...

>How-To-Repeat:
	
Boot a -current kernel on an IBM x336.
>Fix:
	

>Unformatted: