Re: Trouble with NetBSD 4.0 MP on a Asus CUR-DLS

["Greg A. Woods; Planix, Inc." <woods%planix.ca@localhost>]

On 14-Feb-08, at 6:58 PM, Pierre Pronchery wrote:
> I have trouble booting NetBSD 4.0 MP on the Asus CUR-DLS mainboard,
> actually extracted from a HP Netserver E800. It runs the latest BIOS I
> could find on HP website.
>
> As dumped below, when dealing with ACPI the system hits a  
> breakpoint (if
> I understand correctly, it comes from the system's ACPI, not the
> kernel). If I tell the kernel to continue anyway, I get a page fault
> shortly before starting init.
>
> Is there anything I can do to fix this?

That looks a little bit like the crashes (and maybe also the regular  
nightly hangs) I've been seeing with my Dell PE2650 of late.  See PR#  
38019.

It's just that maybe on your system whatever causes the hang instead  
immediately triggers the supervisor trap and you're dead in the water.

On the other hand it could also be completely unrelated.  :-)

Note that on my machine sometimes a supervisor trap is triggered by  
interactive usage attempts once the system has begun to get wedged  
(e.g. hitting <CTRL-T> on the console).

Today for the first time this strange message was on the console with  
a syslog timestamp some five hours after the hang began:

        sa_switch(1647.1): no upcall data.

(I have added "options DIAGNOSTIC" to my kernel -- I'm thinking I  
need to add "LOCKDEBUG" too.)

More often I don't get any kind of error or panic until after I try  
to reboot from within DDB, at which point it almost always does this:

        [halt sent]
        Stopped at      netbsd:cpu_Debugger+0x4:        popl    %ebp
        db{0}> trace
	cpu_Debugger(c52a8600,0,ce9dff38,c033f390,c52a8660) at  
netbsd:cpu_Debugger+0x4
        comintr(c52a8600,0,ce9d0010,30,10) at netbsd:comintr+0x6f0
        Xintr_ioapic_edge4() at netbsd:Xintr_ioapic_edge4+0x9c
        --- interrupt ---
        netbsd:cpu_switch+0xab:
        db{0}> reboot
        syncing disks... done
        unmounting file systems...panic: TLB IPI rendezvous failed (mask e)
        Stopped at      netbsd:cpu_Debugger+0x4:        popl    %ebp
        db{0}>

> root file system type: ffs
> kernel: supervisor trap page fault, code=0
> Stopped at      netbsd:i686_mtrr_reload+0x14a:  popl    %ebx
> db{0}> tr
> i686_mtrr_reload(c2cc3800,1,0,c2cc3800,ccaf1000) at  
> netbsd:i686_mtrr_reload+0x14
> a
> cpu_init(c2cc3800,0,0,0,0) at netbsd:cpu_init+0x62
> cpu_hatch(c2cc3800,0,0,0,0) at netbsd:cpu_hatch+0xa3

--
                                        Greg A. Woods; Planix, Inc.
                                        <woods%planix.ca@localhost>