2.6.0-pre4 hangs if acpi enabled

2.6.0-pre4 hangs if acpi enabled

[Éric Brunet]

Hi,

When booting 2.6.0-pre4 on my intel P4 with a shuttle motherboard, the
computer hangs after printing

	hda: max request size: 128KiB
	hda: 234441648 sectors (120034 MB) w/2048KiB Cache, CHS=65535/16/63, UDMA(100)
	 hda: hda1 hda2 hda4 < hda5 hda6 hda7 >
	mice: PS/2 mouse device common for all mice

No scroll from the keyboard, no sys-rq, Caps-lock doesn't lit the led on
the keyboard. The computer is completely frozen.

On a normal boot, the following lines are

	serio: i8042 AUX port at 0x60,0x64 irq 12
	input: AT Set 2 keyboard on isa0060/serio0
	serio: i8042 KBD port at 0x60,0x64 irq 1

2.6.0-pre1 was working nicely. I haven't tried 2.6.0-pre2 or -pre3.
2.6.0-pre4 works if I append acpi=off of pci=noacpi at the boot
commandline.

You will find on <http://perso.nerim.net/~tudia/bug-reports/> the
complete boot messages, the configuration of the kernel, the dsdt, the
output of lspci -vv, etc. Please tell me if some relevant information is
missing.

Regards,

	Éric Brunet

Re: 2.6.0-pre4 hangs if acpi enabled

[Felipe Alfaro Solana]

On Mon, 2003-08-25 at 13:33, Éric Brunet wrote:
> Hi,
> 
> When booting 2.6.0-pre4 on my intel P4 with a shuttle motherboard, the
> computer hangs after printing
> 
> 	hda: max request size: 128KiB
> 	hda: 234441648 sectors (120034 MB) w/2048KiB Cache, CHS=65535/16/63, UDMA(100)
> 	 hda: hda1 hda2 hda4 < hda5 hda6 hda7 >
> 	mice: PS/2 mouse device common for all mice
> 
> No scroll from the keyboard, no sys-rq, Caps-lock doesn't lit the led on
> the keyboard. The computer is completely frozen.

Well, if you are using ACPI, then this is a know behaviour with -test4
releases. It's ACPI related. Please, take a look at the following bug
report:

http://bugzilla.kernel.org/show_bug.cgi?id=1123

During boot, the kernel checks for i8042 AUX/MUX ports, tries
registering IRQ #12 and then hangs. I fixed the problem on my i845DE
motherboard by flashing a new BIOS with APIC and IOAPIC support. Then, I
recompiled 2.6.0-test4 with APIC and IOAPIC support and the hangs went
away.

If you can't use/compile APIC and IOAPIC for your motherboard, try
booting the kernel with "pci=noacpi" as a workaround. This will prevent
from using ACPI IRQ routing and use standard PCI IRQ routing. Also, if
you have the time, please, attach your "dmidecode" and "acpidmp" to the
bug report at the above URL.

Thanks!

Re: Re: 2.6.0-pre4 hangs if acpi enabled

[Éric Brunet]

On Mon, Aug 25, 2003 at 02:18:33PM +0200, Felipe Alfaro Solana wrote:
> http://bugzilla.kernel.org/show_bug.cgi?id=1123
> 
> During boot, the kernel checks for i8042 AUX/MUX ports, tries
> registering IRQ #12 and then hangs. I fixed the problem on my i845DE
> motherboard by flashing a new BIOS with APIC and IOAPIC support. Then, I
> recompiled 2.6.0-test4 with APIC and IOAPIC support and the hangs went
> away.

Indeed, compiling in APIC and IO-APIC hides the problem and lets the
computer boot and work.

By the way, why should people use APIC and IO-APIC on UP computers ? Is
it faster or more efficient in some way ? Kernel documentation says that
the APIC is a ``replacement for PC-style interrupt controller''. Is there
anything the APIC can do that the traditionnal interupt controller can't
do ? Kernel documentation is not very incentive for wasting an extra 7.5
ko in the kernel image...
> 
> you have the time, please, attach your "dmidecode" and "acpidmp" to the
> bug report at the above URL.
> 
I don't have any dmi and the output of acpidmp is now on the web page
given in my bug report.

Éric


-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines
at the same time. Free trial click here:http://www.vmware.com/wl/offer/358/0

Re: Re: 2.6.0-pre4 hangs if acpi enabled

[Felipe Alfaro Solana]

On Tue, 2003-08-26 at 16:39, Éric Brunet wrote:

> By the way, why should people use APIC and IO-APIC on UP computers ? Is
> it faster or more efficient in some way ? Kernel documentation says that
> the APIC is a ``replacement for PC-style interrupt controller''. Is there
> anything the APIC can do that the traditionnal interupt controller can't
> do ? Kernel documentation is not very incentive for wasting an extra 7.5
> ko in the kernel image...

It seems that IO-APIC removes the 15 IRQ limit that is present when
using XT-PIC-style interrupt controllers. For example, on my P4 box with
IO-APIC enabled, IRQ9 is assigned exclusively to ACPI and is not shared
between multiple hardware devices. Also, USB UHCI is mapped to IRQ 23.



-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines
at the same time. Free trial click here:http://www.vmware.com/wl/offer/358/0