ACPI bug

[knudch]

After have tried a lot of different acpi drivers for my system without succes Here is a report for the latest ver.

Install is going fine After reboot the system stops just after showing the Kernel version

"Internal revision 14.105_SMP" "A disk-read error occurred" "Insert a system diskette and restart the system"

Keyboard is working C-A-D reboot

For last 3 or 4 version this has been the result

BR Knud

[knudch]

Log acpi test log tool attach

[David Azarewicz]

Please edit your CONFIG.SYS to add the /VW switch to the PSD command line so you have: "PSD=ACPI.PSD /VW". then reboot and see if the system will boot.

If it still won't boot, change the /VW to /PIC so you have "PSD=ACPI.PSD /PIC", then reboot and see if the system will boot.

If you can get the system to boot in one of these 2 ways (preferably with /VW) please generate another testlog and attach it to this ticket. The testlog is only useful if the PSD is running.

[knudch]

tried with these 3 combinations:

/VW

/PIC

/VW /PIC

same result, stop just after showing kernel version

Knud

[ecsnl]

If its the same defect I don't know. But in the Mensys office we also have a machine that when booting ACPI dies of very fast with the same error message. Its an HP machine.

Question 1. When you press the key combination ALT F2 at startup is there a driver that you see it hangs on ? (remming out that driver in the config.sys will not help BTW it will just hang on the next driver and so on)>

Question 2. Does this system have a serial port or can you make memory dump on hard disc ? If you can make a memory dump this would require a 2 GB of free disc space. But if you have more then 2 GB of RAM. There is also trick.

If you can tell me what device driver it hangs on during boot. We can also remove (for the dump) the line earlymeminit=true from the config.sys. If the system is still loading the BASEDEV drivers we should be able to get a dump of only 16 MB.

If you do not know how to make a memory dump partition thats no problem. But if you can answer the above questions.

It will BTW not mean that the problems can fixed with the memory dump. But it could give us some hints as to why the boot is not working.

Thanks,

Roderick Klein Mensys

[David Azarewicz]

Don't bother with ALT F2, and don't bother with a memory dump. We may still do that later but first please see the section "Collecting Logs" at http://svn.netlabs.org/acpi/wiki/Requirements and follow the directions for installing the debug version of the ACPI software. You must have the debug version of the PSD installed to get any useful information out of the system.

Once you have installed the debug version with Warpin and installed them with the Install script, but before you reboot, follow the directions under "Collecting a Video Log".

I will know more after I see the picture.

[knudch]

I succeded to generate a seriel log, is attached acpi pack was 3.21.04 but the error has been the same for several versions

Knud

[knudch]

seriel log

[David Azarewicz]

Thanks for the serial log.

Your system won't boot because of many of these type of errors in your Acpi Tables:

ACPI Error: [LNKC] Namespace lookup failure, AE_NOT_FOUND (20121114/dsobject-209)

These are critical resources that are required to operate the system.

Based on reading your specific ACPI code, if you add /OS="Microsoft Windows" to the PSD command line it will change slightly the execution of certain ACPI methods. It will probably not fix your problem but you can try it. The /OS switch must be typed *exactly* as I have written it.

[knudch]

With the /OS switch Then I got this after approx 20 sec. Attach picture Just for info..system is running perfect with WIN7/XP and Linux Fedora 17

Knud

[David Azarewicz]

The code in the picture indicates a hardware failure. I would suggest that you try booting after completely powering off the system.

[knudch]

Just for info.....

All test is made from cold boot (power switch has been off for 5-10sec)

Reason: I have had both on this HW and my previous HW that eCS is doing something about the HW so that harddisk did not operate correct(previous HW), and on new HW network function does not operate after eCS has been up. (Booting into WIN or Linux). Also after a clean shutdown in eCS.

Tomorrow I will try again and together with debug switch set and seriel log.

Knud

[knudch]

I the last test once more with the debug level=3 and OS="Micro switch It stopped again with "A disk-read error occurred" "Insert a system diskette and restart the system"

I have attached the seriel log from the test.

Knud

[David Azarewicz]

The /OS switch did what it was supposed to but it had no effect. There is clearly a problem with the ACPI in that system.

You can also try this: Remove the /OS switch and add /OIM=32

[knudch]

Same result as with /OS switch

Seriel log attached

Knud