ACPI 3.05 traps on Shuttle Barebone with AMD Dual Core and VIA chipset

[Chuck McKinnis]

System boots with two processors but after loading of EVFS it traps in kernelmodule before WPS comes up. It makes no difference if I use /PIC or APIC. With ACPI 3.04 the system works well with parameter /PIC. ACPI 3.05 is tested on both rc3 and rc4 and behaves the same.

[Chuck McKinnis]

Attached is logs from using 1 CPU and original ACPI.PSD, also attached is log from using 2 CPU with debug-version of ACPI.PSD. A photo of the trap after first installation.

What surprices me is the system is working on 2 CPU's when using the debug-version? No traps, just working.

[pasha]

try remming uniaud, ntfs.ifs, fat32.ifs ?

[Chuck McKinnis]

I did rem out uniaud, makes no difference, it still traps. The other two is not installed on my system, it's an eCS only system.

[Chuck McKinnis]

I've got the same problem but with an ASUS Pundit P1-AH2 Booksize Barebone with nVidia nForce 410/430 chipset which has an Athlon64 X2 3800+ EE installed. I've filed the bug at ecomstation.com bugtracker (​http://bugs.ecomstation.nl/view.php?id=1838) but I attach the files here as well.

Last known good ACPI driver for SMP is v2.22...none of the 3.0x line was usable so far.

I've also tried other workarounds diabled USBRESMGR.SYS, UNIAUD and different ACPI.PSD parameters for SMP ("/SMP", "/SMP /APIC" and "/SMP /APIC /!NOD") but nothing worked so far.

Attached files: 1.) TRAP 8 screenshot using ACPI v3.05 in SMP mode 2.) ACPICA$ output using ACPI v3.05 in UNI (normal) mode with installed SMP-Kernel

I hope this helps to get rid of this annoying problem.

Kind regards, Juergen Ulbts

[Chuck McKinnis]

This problem was introduced with version 3.05 to me, on 3.04 and earlier 3.x's it has worked on both kernels without any traps. The only thing that has been different is shutdown of the computer. With 3.04 and 3.05 it shuts down but starts again as if had gone for a reboot. With 3.03 it did a shutdown and nothing else. That doesn't bother me much but it's a must to have the kernels running. On my laptops the shutdown works perfect but they have only one kernel so it's difficult to compare.

For the moment I am using the debug-version of 3.05 on my Shuttle with AMD Dual Core and VIA-chipset and both kernels is activated and runs perfect.

bjorso

[Chuck McKinnis]

Yes, your trap was a little bit different but it seems to happen at the same place (after EVFS). I now replaced the original Intel network drivers with nickk's nveth 0.0.4 driver for the integrated nforce nic chipset, removed the os2logo and added the Debug version of ACPI.PSD. I also modified the parameters again to the values that normally should work for the nForce chipset: /SMP /APIC The boot process stops a little bit after the nveth driver (I think) and I took a screenshot of that: ACPI_v305-DEBUG_(with-SMP-APIC)_Athlon64X2_ASUS-Pundit_P1-AH2_small.jpg

[Chuck McKinnis]

Tried it again with "/SMP /APIC /!NOD" and the nveth driver and this worked. Attaching log file!

[Chuck McKinnis]

Replying to guest:

Tried it again with "/SMP /APIC /!NOD" and the nveth driver and this worked. Attaching log file!

Update: It's only working when I use the DEBUG version of ACPI.PSD. It's still crashing with the normal ACPI.PSD.

Very interesting!

[pasha]

Read pls ​http://ru.ecomstation.ru/projects/acpitools/?action=testcase , then download file acpi-t-29122007.zip. Try boot with it. Intresting result. Have trap in boot, no trap in boot

[Chuck McKinnis]

I am afraid the testkernel did not make any difference. It still trapped in OS2KRNL so I had to revert to Debug-version.

bjorso

[Chuck McKinnis]

Ok, now I also had the time to test the testdriver dated 2007-12-29. It doesn't solve the problem here. Normal operation (without /SMP) is working but when I switch to "/SMP /..." it does crash.

I've attached the picture of the dump when it crashed with "/SMP /APIC /!NOD" using the test driver from 2007-12-29: ACPI_TESTDRIVER_2007-12-29_(with-SMP-APIC-noNOD)_Athlon64X2_ASUS-Pundit_P1-AH2_small.jpg

Then I also attached the log while using only one CPU core (ACPI.PSD without parameters): ACPI_TESTDRIVER_2007-12-29_UNI-Mode_Athlon64X2_ASUS-Pundit_P1-AH2.txt

This weekend I get my Null-Modem cable back, so I can log the boot process with my other machine. Just let me know if this will help you finding the problem.

Many thanks!

Kind regards, Juergen Ulbts

[Chuck McKinnis]

Ok, I've got the boot protocol transmitted over Null-Modem cable. Parameters used: /SMP /APIC /O1

File: ACPI_v305-DEBUG_(with-SMP-APIC)_Athlon64X2_ASUS-Pundit_P1-AH2.log

[pasha]

1. Add /ALS
2. if fail try rem ntfs and fat32

[Chuck McKinnis]

Ok, tried /SMP /APIC /ALS but this did hang the machine (I can still see the Kernel version number). I removed NTFS.IFS and FAT32.IFS already before I added the /ALS switch.

The logfile is attached.

So far I didn't install the SMP-Debug Kernel. If this is even better I'll add that one as well.

[pasha]

In log I see , that you are use SMP kernel. Check pls doscall1.dll. It must be from SMP kernel

[Chuck McKinnis]

Replying to pasha:

In log I see , that you are use SMP kernel. Check pls doscall1.dll. It must be from SMP kernel

I used the eCS Dialog to install the SMP Kernel, so I hope it did install the required DOSCALL1.DLL.

[Chuck McKinnis]

Replying to guest:

Replying to pasha:

In log I see , that you are use SMP kernel. Check pls doscall1.dll. It must be from SMP kernel

I used the eCS Dialog to install the SMP Kernel, so I hope it did install the required DOSCALL1.DLL.

Checked DOSCALL1.DLL installed: DOSCALL1.DLL - 144.631 bytes - 02.08.2005

I also downloaded SMP20050811D.ZIP from another site as the debug kernel 14.104a doesn't seem to be available directly from mensys. I'm installing this one now and create another log.

For faster feedback you can get to me via ICQ or Jabber. Just send me an email using the address on my website (see "Contact" on the menu): ​http://www.juergen-ulbts.de/

[pasha]

IRC is better ;-) irc.ecomstation.com -> #netlabs , my nick [Pasha]. Now I ending any changes and 2-3 hour after can give to you next acpi.psd

[Chuck McKinnis]

Replying to pasha:

IRC is better ;-) irc.ecomstation.com -> #netlabs , my nick [Pasha]. Now I ending any changes and 2-3 hour after can give to you next acpi.psd

Ok, I hate IRC but I'll show up there in a few minutes...just created the latest LOG with the debug kernel installed. Last driver loaded after EVFS is nickk's NVETH v0.0.4 then the boot process stops.

The log follows in a few moments...one your new build is available I'll try that one.

[Chuck McKinnis]

Ok, here the log from the latest 3.05 (2008-01-13) using the debug kernel 14.104a (one released on testcase) with /SMP /APIC /O1.

[Chuck McKinnis]

Another log with /SMP /APIC /ALS
The bootprocess stops and drivers are not loaded.
Last line at the console still shows: "Internal revision 14.104a_SMP"

[Chuck McKinnis]

Begin testing of ACPI.PSD first build from 2008-01-14... 1.) /SMP /APIC /ALS 2.) /SMP /APIC /ALS /TMR ... Logs follow...

[Chuck McKinnis]

Description of results...

With /SMP /APIC /ALS boot hangs while loading kernel. I can still see the "Internal revision 14.104a_SMP" line.

The same happens with /SMP /APIC /ALS /TMR but the log includes more information. It seems internally the boot process didn't get that far as without the /TMR parameter.

[Chuck McKinnis]

Description of the new round of tests:
Fresh install of eCS 2.0 rc4 on the ASUS P1-AH2...this time with SNAP installed.
Replaced ACPI.PSD/SYM with the build from 2008-01-16.
Replaced normal SMP kernel with debug kernel.

Parameters start with /SMP /APIC /O1

Log follows...

[Chuck McKinnis]

Both times it stopped while loading "screen01.sys".
I entered "db ebx" on both tests (/SMP /APIC and also with /SMP only).
When I did the /SMP test, I continued with the "G" command but the machine trapped later (as you can see at the log...that follows).

[Chuck McKinnis]

Tested the new 3.06 build on my Shuttle. Unfortunately the same traps as with 3.05.

bjorso

[pasha]

ACPI25012007.ZIP , I need terminal log

[Chuck McKinnis]

Replying to pasha:

ACPI25012007.ZIP , I need terminal log

This time it traps before the virtual driver for Samba filesystem is loaded. It still refuses to shut down. It reboots the computer every time. See attached picture and logs produced with version 3.06 and 3.06 debug. It works fine on two kernels when running on debug-version except for shut down.

bjorso

[pasha]

Do you use ntfs.ifs or fat32.ifs? If yes, try remming

[Chuck McKinnis]

Replying to pasha:

Do you use ntfs.ifs or fat32.ifs? If yes, try remming

No, this an eCS only system and ntfs.ifs or fat32.ifs is never installed.

bjorso

[Chuck McKinnis]

I'm uploading the results of PCI.EXE for my AMD Athlon64 X2 on my ASUS Pundit P1-AH2 (nVidia nForce 410/430) now.
Downloaded the latest PCIDEV-ID List.
Other tests with the latest ACPI at betazone follow.

[pasha]

Please download experimental ACPI build from Mensys site:

• Experimental build for you:

acpi31012008.zip

• Read instructions: ​http://ecomstation.ru/projects/acpitools/?action=testcase

[Chuck McKinnis]

Downloaded acpi31012008 and tried. Behaves fine, computer boots normal and both CPU-kernels loads, what is missing is a proper shut down, it still reboots the computer instead of the requested shut down.

bjorso

[Chuck McKinnis]

OK, please, create new ticket related to shutdown in 2-4 weeks. wait for official release of fixed ACPI, try different methods of shutdown.