acpi 3.21.07 + Gigabyte EP45-DS3 - no poweroff with this version

[hakiko]

Hello, I have a problem since acpi version 3.21.0x .. 3.21.07 On my Gigabyte EP45-DS3 don't works power off.

Before with acpi 3.18 it runs perfectly.

Other things with acpi 3.21.07 works good:

• Intel Core 2 Quad - SMP - 4 cores
• PCI Graphic Card ATI Radeon 9250 - SNAP accelerated 1600x1200 16K
• Intel Pro/1000 CT PCIe
• Audio works also

However, it is not possible to Power off the system with this build of ACPI installed. Shutdown gets as far as closing the Desktop and, after a few seconds the system issues 3 beeps. Pressing the Power Off button on the system box at this point does not cause any problems on next boot.

Which files do you need? Which tests should I run.

With "acpistat poweroff" the system freezes - no power off.

greetings Heiko

[robh]

Exactly the same here. I don't exactly remember when the power-off started to fail. Gigabyte Z68X-UD3H-B3. Intel I5 4 core (no hyperthreading), 8 GB RAM, onboard video. eCS 2.1 and latest 2.2beta, ACPI 3.22.03, AHCI 1.32, USB 11.5 (2 x USBEHCD) , Panorama 1.06. Don't know what other info might be useful, I'll try to attach the output of PCI. Regards, Rob.

[robh]

Output of PCI for Gigabyte Z68X-UD3H-B3

[David Azarewicz]

Version 3.21.07 is quite old and not supported. Version 3.22.03 is current.

To power off the system, the PSD calls a BIOS function. The only way it can fail is if the BIOS returns an error or if the BIOS function just fails completely.

The only way to see the status coming back from the BIOS function (if it fails) is with a serial port log. So if you have a serial port connector plugged into the serial port on your motherboard, and if you capture at least the last few lines of a serial port log when you try to power off (see http://svn.netlabs.org/acpi/wiki/Requirements), I'll have a look at it.

[robh]

I've installed the debug versions of ACPI.PSD version 3.22.03 (the 3.21.07 was from the original report by hakiko with another Gigabyte motherboard) The ACPI line in my config.sys:

PSD=ACPI.PSD /DBGLVL=3 /O1

The log was captured via the serial port and a cross-over cable on another system. I've copied the last lines of the log below. If you would the startup phase too I can attach the whole log.

IPIPsdInt CPU 2 OemHlpXHandler: PSDCMD_STRATGY: Shutdown 0 OemHlpXHandler: PSDCMD_STRATGY: Shutdown 0 OemHlpXHandler: PSDCMD_STRATGY: Shutdown 1 OemHlpXHandler: PSDCMD_STRATGY: Shutdown 1 PSD_IRQ_MASK: mask Mode=2 IRQ=16 IRQMask=FF73FCFC IOAPICMask=FF73FCFC PSSD_PROCHLT: CPU 1 to offfline PSD_APP_COMM: Set Sleep State to 5

[David Azarewicz]

Replying to robh:

I've installed the debug versions of ACPI.PSD version 3.22.03 (the 3.21.07 was from the original report by hakiko with another Gigabyte motherboard) The ACPI line in my config.sys:

PSD=ACPI.PSD /DBGLVL=3 /O1

The log was captured via the serial port and a cross-over cable on another system. I've copied the last lines of the log below. If you would the startup phase too I can attach the whole log.

IPIPsdInt CPU 2 OemHlpXHandler: PSDCMD_STRATGY: Shutdown 0 OemHlpXHandler: PSDCMD_STRATGY: Shutdown 0 OemHlpXHandler: PSDCMD_STRATGY: Shutdown 1 OemHlpXHandler: PSDCMD_STRATGY: Shutdown 1 PSD_IRQ_MASK: mask Mode=2 IRQ=16 IRQMask=FF73FCFC IOAPICMask=FF73FCFC PSSD_PROCHLT: CPU 1 to offfline PSD_APP_COMM: Set Sleep State to 5

Why are you posting your information on someone else's ticket?

In your case, since there is nothing after the "PSD_APP_COMM: Set Sleep State to 5" line, that means that there was no error status. It also means that the Set Sleep State call is not returning at all, which is correct. It should power off the system and never return. From the point of view of the PSD, it has completed everything it is supposed to do and has executed the final step to power off the system -- call the ACPI method in the BIOS to do the actual power off. If your system has not powered off, this means that something in your BIOS has failed to do what it was supposed to do. Unfortunately, there is no way to find out what went wrong in your BIOS.

[robh]

Replying to dazarewicz:

Why are you posting your information on someone else's ticket?

Well, before opening a new ticket I searched for a similar ticket to avoid duplicates and this one seemed to match: same motherboard brand (probably same BIOS) and same issue (no power off).

In your case, since there is nothing after the "PSD_APP_COMM: Set Sleep State to 5" line, that means that there was no error status. It also means that the Set Sleep State call is not returning at all, which is correct. It should power off the system and never return. From the point of view of the PSD, it has completed everything it is supposed to do and has executed the final step to power off the system -- call the ACPI method in the BIOS to do the actual power off. If your system has not powered off, this means that something in your BIOS has failed to do what it was supposed to do. Unfortunately, there is no way to find out what went wrong in your BIOS.

Thank you for your explanation. It seems I'll have to live with this inconvenience. I've searched BIOS setup and only in the power management section I found one parameter mentioning ACPI: 'ACPI Suspend Type' with the choice between S3(STR) and S1(POS), of which I think S3 the most obvious.

You would not want to comment on the remark that W7 is capable of powering off this system?

[David Azarewicz]

Please attach the zip file created by

testlog acpi

See http://svn.netlabs.org/acpi/wiki/Requirements for details.

[robh]

output of testlog dd 2013/12/30

[robh]

Due to an issue with my disk I did a fresh install of eCS22-betaII (Dec 23). Today I installed the debug version of ACPI.PSD and ran testlog (the version which is present in the ACPI debug directory).