Re: Toshiba Satellite ACPI problems

[Robert Hancock <hancockrwd@gmail.com>]

On 11/23/2009 06:37 PM, Rafael J. Wysocki wrote:
> [Adding linux-acpi to the CC list.]
>
> On Tuesday 24 November 2009, Mike Smith wrote:
>> I have a brand new Satellite L505D-S5983 which I'm trying to install
>> Linux on. Whenever I boot a Linux kernel on it with no special
>> parameters it enters an in(de)finite loop dumping messages like this:
>>
>> ACPI Error (uteval-0313): Return object type is incorrect
>> [\_SB_.HSB1._STA] (Node ffff8800af8121a0), AE_TYPE
>> ACPI Error: Type returned from _STA was incorrect: Device, expected
>> Btypes: 1 20090521 uteval-319
>> ACPI Error (psargs-0359): [DR ] Namespace lookup failure, AE_NOT_FOUND
>> ACPI Error (psparse-0537): Method parse/execution failed
>> [\_SB_.VALZ._STA] (Node ffff8800af812540), AE_NOT_FOUND
>> ACPI Error: Found unknown opcode 39 at AML address ffffc900000132cd
>> offset 0, ignoring 20090521 psloop-137
>> [...]

Hmm, that looks really unhappy..

>>
>> If I boot with acpi=noirq these messages only continue for a few
>> seconds, although they cause a couple of initscripts to hang later on.
>> With ACPI fully enabled they keep appearing forever and the system
>> doesn't finish booting. With acpi=off they don't appear at all (but
>> the fan stays at low speed and the laptop gets quite hot if run for
>> more than a few minutes).
>>
>> I've found only one other mention of this problem on the internet
>> (http://ubuntuforums.org/showthread.php?p=8358864#post8358864) from
>> someone with the same laptop as mine, but there was no working fix
>> posted to that thread beyond completely disabling ACPI. More research
>> told me that there used to be a way of fixing up the DSDT and having
>> it loaded from the initrd but the new policy is to just ask the
>> manufacturer to fix it and the kernel developers to work around it,
>> so... here it is. (Angry emails to Toshiba and Best Buy pending.)
>>
>> Full dmesg output can be found at http://slexy.org/view/s25zU08AKJ
>> (By the way, what's the correct way to include dmesg output in a lkml
>> message if you don't know for sure which parts are relevant? Dumping
>> all 50k or so of it into the email seems like the wrong thing to
>> do...)
>
> I think it's better if you file a bug report against ACPI at bugzilla.kernel.org.

Can you extract and post the DSDT in either raw or decompiled form, as 
described here:

http://smackerelofopinion.blogspot.com/2009/10/dumping-acpi-tables-using-acpidump-and.html