HOWTO Fix A Buggy DSDT File

[peterwil]

@67GTA:
Like peterwil I've got a Toshiba L305D-S5881. Today I found out something somewhere is quite buggy: I tried to compile a vanilla kernel, but during the process the system decided it was getting too hot and turned off.

DSDT has one warning, pertaining to an AND operation that doesn't store its result. It's the same as the ACPI reference implementation, which on page 206 says this in sample code:

// Only allow native hot plug control if OS supports:
// * ASPM
// * Clock PM
// * MSI/MSI-X
If(LNotEqual(And(SUPP, 0x16), 0x16))
{
    And(CTRL,0x1E) // Mask bit 0 (and undefined bits)
}

Is it me, or does that masking not actually mask anything, because CTRL isn't being updated?
Would this change anything?

Also, it seems that the rather dramatic shut off wasn't really due to temperatures reaching 105C: the laptop felt pretty warm, but not nearly that hot. Do you (or anyone) know of anything that could alleviate this problem? I trust the preinstalled Windows doesn't have this problem, but I've never used it long enough to find out.

Lastly: I tried using toshset, but it complains of missing Toshiba functionality in the kernel. Enabling the module toshiba_acpi fails with the following message:

"FATAL: Error inserting toshiba_acpi (/lib/modules/2.6.28-11-generic/kernel/drivers/acpi/toshiba_acpi.ko): No such device"

Hope I didn't veer too much off topic...

Trebaruna,
Glad to know you have the same machine (Toshiba L305D-S5881) with Ubuntu Jaunty. My DSDT has a very similar warning on compilation with iasl..There are no errors.. Previously 67GTA concluded that getting rid of that warning may not help much with
fan overheating.. The good thing for me is that on jaunty so far
I haven't seen it go much higher than 55C unlike Hardy where it would jump to 75C within minutes.. Having said that I don't think my problem is fully resolved since the dmesg output does not seem to indicate
i) The fan default state to be ON
ii) Thermal zone temp. settings to be lower than what I have seen in this thread from the output of others where 67GTA has fixed the dsdt. Most of the postings have the AML binaries. It might be more useful to have the before and after DSDT's to get a deeper understanding. Assembler code is not my expertise .
One thing that comes to mind right away is my BIOS is Insyde v1.2. I think there is an update to this v1.5. Have you tried this update yet?

Comparing the dmesg output in hardy and jaunty seems like they are very similar. There is the usual line about acpi_os (linux) query ignored which seems to suggest that the linux tables are not being read right. Haven't tried sticking this as a boot option in Jaunty to see if it goes away.

On filing a bug in Launchpad I got a response from the kernel team that "Toshiba extensions are not enabled by default and you would have to create a custom distribution..." Also I came across numerous posts complaining of overheating laptops in Jaunty. It doesn't seem to be only Toshiba's..

I have a good mind to try Fedora 11 on the remaining free space partition on my laptop to see how it behaves..But that is for another day... ( Factory installed Vista takes up 2 ! )..

If you do end up resolving this completely pl. post as I am very interested in this as well.

Thanks
Peter

[peterwil]

Well...I don't know if mine would help (I'm guessing that mine was a special case), but I can attach it here. I'm still tinkering with it, looking and making sure that the IRQs get set properly.

I do NOT want you to use this for your own DSDT file since our two machines are different. You can look at this to simply gather an understanding though, which is absolutely fine.

Thanks Makoto.. I won't use your DSDT ( they are machine specific ) .. But will get some pointers from looking at similarities if there are any..I am particularly interested in the thermal zone settings .. which seem to have brought your temps down by 10C....

[kahukiloia]

67GTA,

First of all, thanks for the very informative howto. I have followed your instructions and got the output file without a hitch. The hitch popped up a when I got to this:

Maximum error count (200) exceeded
ASL Input: /home/kahu/dsdt.dsl - 7375 lines, 254663 bytes, 3490 keywords
Compilation complete. 201 Errors, 0 Warnings, 0 Remarks, 570 Optimizations

Maximum error count (200) exceeded
Segmentation fault

I am slightly over my head with this and stuck. I hate to do this to you or anyone, but..... If you could give me some idea how to proceed with making this file usable, I would be very grateful. I am on newest version of Linux-Mint (Gloria - 7) and since I have upgraded from Elissa, I have been experiencing issues with booting up (have to hold any key), and sleep (mainly waking up), additionally, it would seem that my fan is on an unauthorized overtime. Any help would be greatly appreciated.

[67GTA]

kahukiloia:

You now have zero errors. I fixed errors for suspend/hibernate, so with the nvidia drivers installed, you should be able to wake properly. The hanging at boot should stop. I also fixed errors with the temp readings. It should run cooler/quieter. Follow the rest of the how to with the dsdt.aml file. Be sure to add the acpi_osi="Linux" line to /boot/grub/menu.lst. It is asking for it in your dmesg output. After a reboot, send me your dmesg output again.

dsdt.aml.zip

[Trebaruna]

I've just checked my fiancee's laptop. Her machine's DSDT is completely screwed up, it makes iasl segfault! (though it seems that's the default at > 200 errors).
Most (all?) of the errors seem to be "object does not exist" though, so I've a feeling there's a pretty simple fix. Would anyone know what to do with this file? (see attachment).

Also, gta, did you get a chance to look at my question about the AND operation not storing its results? Is it significant at all, you think?

[67GTA]

Trebaruna:

I changed all instances of (^CPU0) to (\CPU0), and there are zero errors. Sorry, I missed your question. Can you post the error message?

dsdt.aml.tar.gz

[Trebaruna]

I changed all instances of (^CPU0) to (\CPU0), and there are zero errors.

Thanks! I guess that means there's nothing wrong with it (syntactically anyway)?

Sorry, I missed your question. Can you post the error message?

It's a warning, actually:

dsdt.dsl  2783:                         And (CTRL, 0x1E)
Warning  1105 -                                 ^ Result is not used, operator has no effect

The ACPI spec mentions exactly this piece of code on page 186, claiming it is to mask bit 0. Indeed, that's what this statement would do, if it stored the result back into CTRL. As it stands, though, only the AND is taking place, without the storing. Is the spec (and by extension this piece of code, which is taken directly from the reference) wrong?

[67GTA]

You can try adding CRTL back into the method to shut the error up, but it won't affect your system in any way since the control method isn't used. Try this:

And (CTRL, 0x1E, CTRL)

[peterwil]

Thanks! I guess that means there's nothing wrong with it (syntactically anyway)?


It's a warning, actually:

dsdt.dsl  2783:                         And (CTRL, 0x1E)
Warning  1105 -                                 ^ Result is not used, operator has no effect

The ACPI spec mentions exactly this piece of code on page 186, claiming it is to mask bit 0. Indeed, that's what this statement would do, if it stored the result back into CTRL. As it stands, though, only the AND is taking place, without the storing. Is the spec (and by extension this piece of code, which is taken directly from the reference) wrong?

Mine has exactly the same warning...

peter@neutrino:~/pw_acpi_issues/pw_modified$ iasl -tc ./dsdt.dsl

Intel ACPI Component Architecture
ASL Optimizing Compiler version 20081204 [Jan 10 2009]
Copyright (C) 2000 - 2008 Intel Corporation
Supports ACPI Specification Revision 3.0a

./dsdt.dsl  2803:                         And (CTRL, 0x1E)
Warning  1105 -                                   ^ Result is not used, operator has no effect

ASL Input:  ./dsdt.dsl - 7962 lines, 277321 bytes, 3515 keywords
AML Output: ./dsdt.aml - 30381 bytes, 838 named objects, 2677 executable opcodes

Compilation complete. 0 Errors, 1 Warnings, 0 Remarks, 18 Optimizations

--- No errors..
I haven't seen it shut down.. But today under slightly heavy processing..
(portal + servers ) I saw the temps spike to 75C and then drop down back again to around 56C..

[kahukiloia]

67GTA,

You are amazing, works just as advertised. Quiet, fast, does not hang up on boot while on the batteries; only thing I did not have a chance to fully test is the waking up from suspend, however, quick test already worked really well.

Thank you very much for your help, I was really stuck. Attached is the post update dmesg output. Once again, thank you.

kahukiloia