>Number:         11299
>Category:       port-i386
>Synopsis:       PNPBIOS probe causes spontaneous reboots on Gateway E5200
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    port-i386-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Oct 23 13:06:01 PDT 2000
>Closed-Date:
>Last-Modified:
>Originator:     Rafal Boni
>Release:        NetBSD 1.5H, CVS updated Oct 16th, ~ 11:30PM EST
>Organization:
None
>Environment:
	Gateway E5200 desktop: Tyan Patriot motherboard (dual-PIII-capable,
	only one CPU installed), AMI BIOS (rev not available right now,
	will supply next time I reboot the box).

    NetBSD 1.5H (CYCLOPS.pnpbios) #11: Mon Oct 23 09:53:33 EDT 2000
    rafal@cyclops:/extra/src-current/sys/arch/i386/compile/CYCLOPS.pnpbios

>Description:
	The PNPBIOS probe causes a reboot of the machine (no panic, no
	messages on consol, just a reboot) while querying the BIOS for
	info on node #/index 14.  Tracing through the code shows that the
	reboot happens in the guts of pnpbios_getnode().

	Tracing through the PNPBIOS code under NT on the same box shows
	that the PNPBIOS code is attempting to grovel through the extended
	BIOS data segment at 0040:xxxx (in x86 segment-speak 8-).  

	Skipping index 14 allows the machine to boot without problems.
	Skipping any/all other indices doesn't affect the problem.  
	Commenting out the pnpbios_attachnode() or other code in the 
	loop doesn't affect the behaviour either.

>How-To-Repeat:
	Boot PNPBIOS-enabled kernel on my Gateway E5200, watch it reboot
	right after querying PNPBIOS index 13.  Track down the failure in
	GDB, see that it happens getting node data for index 14.

>Fix:
	Not yet 8-<

>Release-Note:
>Audit-Trail:
>Unformatted: