acpi=off otherwise machine doesnt boot

[Maaz]

I have installed a SUSE Linux 10 Desktop on a home-made(un-branded) desktop machine
After installation i tried to boot the machine... but machine doesnt. To boot the system i pass acpi=off kenel parameter... and machine boots succesfuly.

#uname -r
2.6.16.21-0.8-default

output of dmidecode is attached

here i wana asked that .. is there any proper way to fix this issue... because i think that acpi=off is a workaround, but not a proper sol.

Regards
Maaz

[Oviwan]

Hey

had to do the same on my notebook. have no other solution found but also not long searched ;)

Regards

[Rob Leadbeater]

I guess you might want to check how the power options are configured in the BIOS of the machine...

Cheers,

Rob

[Court Campbell]

acpi=off is a solution. I would think that's why the parameter exists. Who needs power management anyway.

[Maaz]

Thanks all
>Who needs power management anyway.
So why its there ?... so why it(acpi) works on other and not this particular machine.

Regards
Maaz

[skt_skt]

ACPI puts the OS in control of system configuration and power management.
Further, it acts as a hardware abstraction layer between the OS and the platform BIOS -- allowing the OS and the platform to evolve independently. See the Documentation section for further information.



The Linux/ACPI project is focused on making Linux run well on all ACPI-enabled platforms.
Linux/ACPI is based on Intel's ACPI Component Architecture (ACPICA) reference implementation.
http://www.intel.com/technology/iapc/acpi/downloads.htm



A central goal of the project is that any computer whose ACPI implementation boots and runs Windows properly, should also boot and run Linux properly.
This means that if Linux needs any ACPI-related command-line parameters or a modified BIOS DSDT to boot and run properly, then you should file a Linux bug.



To participate in the project, the latest information on filing bugs and submitting patches is here:
http://ftp.kernel.org/pub/linux/kernel/people/lenb/acpi/patches/README.ACPI

[Court Campbell]

Why is it there? That is a great question that I cannot answer. You are not the first nor the last to have issues with acpi. Microsoft was one of the promoters if that gives you any idea of how useless it is. No, I am not a MS hater. But just because someone or some company has an idea, it doesn't make it a good one.

[dirk dierickx]

it _is_ a proper solution, not a workaround. for so many things that linux supports there are some bad implementations in the hardware out there.

so, you have to turn acpi off, blame it on the HW vendor for being incompliant to the acpi standard. it's a lousy excuse but it is true.

on the other hand, were you planning on using acpi? you won't be missing out on anything major, it's a desktop pc after all...

[Fred Ruffet]

ACPI is a good thing. Maybe you don't think of any advantage because you use it on standard PC as workstation.

I use Linux on servers and laptops. Battery management, fans and cpu supervisions are great for those kind of environments. As for low energy consuption, that saves computer components.

Managing this points with a single interface on many machines is also a good point.

Really, ACPI is our friend.

Regards,

Fred

[Mike Stroyan]

You should try updating to the newest motherboard BIOS available from
http://www.intel.com/products/motherboard/D945GCCR/tools.htm
That might correct some problem with the ACPI data provided by the BIOS.

You could also try a less drastic boot option such as pci=nommconf which disables attempts to use the memory mapped IO mechanism for PCI. That is a common cure for problems with recent Intel motherboards.

[Maaz]

Thanks Dear Mike Stroyan .
whats this kernel parameter does pci=nommconf... ? what is it for ?

[Mike Stroyan · Accepted solution]

The pci=nommconf boot option disables one specific feature that is among the many that are disabled by the acpi=off option.
The MMCONFIG feature uses a recently added ACPI MCFG table that provides another way to describe PCI devices' configuration to the OS. MMCONFIG changes from IO mapping PCI devices to memory mapping PCI devices.
There are other older ways to identify and access PCI devices. It seems that some systems and BIOS revisions have trouble with linux trying to use MMCONFIG, while others have trouble with trying to use the "bios" or "conf1" or "conf2" methods. The MMCONFIG mechanism is still fairly new in linux. And the ACPI MCFG feature itself is new. Because ACPI MCFG is ignored by windows releases prior to vista it may not have had much testing. So trouble with MMCONFIG may be due to problems in the kernel or problems in the BIOS, or disagreement in interpreting specifications.
Forcing pci=nommconf means that the PCI devices are iomapped in the old way instead of memory mapped in the new way.