SLPB problem - Daemon trouble

[David McKenna]

Hardware: MSI H55M-E33 board, Intel H55 chipset, Intel i3 CPU, 2GB DDR3 RAM, 500GB Drive, 1 JFS partition, ICH10-AHCI, Integrated Video, Realtek 8111C NIC, Intel HDA sound.

Software: eCS 2.1, Panorama, OS2AHCI-JFS, UniAud?, Multimac Realtek, ACPI 3.19.15.

Installed new ACPI while booted with 3.19.14 with no trouble. After, boot stopped at blank screen with blinking cursor right before WPS would start. Tried no switch and /VW with same result. With /PIC, computer would suspend at the same spot, but would not resume. CAD does not work. Booting to command line would work, but keyboard is unresponsive and can't type anything.

Booted with CD, REMmed out APM and ACPIDaemon and no switch on ACPI. Computer boots fine. Added APM back in, still boots fine. Added ACPIDaemon and no boot. I tried starting the Daemon after booting with ACPI and APM and the computer immediately suspends, but won't resume.

Not sure how to trouble-shoot this, but I'll try anything you like...

[David McKenna]

At one point I edited acpid.cfg to make sure nothing there would cause a suspend and set the log directory. Here is what was captured a couple of times:

06 Feb 17:43:54 AcpiDaemon? ValidateSleepState?: State '\_S3_' is acceptable

06 Feb 17:43:54 AcpiDaemon? ========== AcpiDaemon? Starting ==========

06 Feb 17:43:54 AcpiDaemon? WARNING! video save mode not set

06 Feb 17:43:54 AcpiDaemon? Error detecting video size

06 Feb 17:43:54 AcpiDaemon? Detected APM version 3.19

06 Feb 17:43:54 AcpiDaemon? LID set to Do Nothing, PowerButton? to Power Off

06 Feb 17:43:54 AcpiDaemon? Suspend support:On State:S3 [\_S3_] VBE:0x0

06 Feb 17:43:54 AcpiDaemon? Log to screen:Off

06 Feb 17:43:54 AcpiDaemon? NOTIFY: ========== Notify thread is starting ==========

06 Feb 17:43:54 AcpiDaemon? PWRBT: ========== Power Button thread is starting ==========

06 Feb 17:43:54 AcpiDaemon? PWRBT: Calling AcpiTkWaitPressButton?

06 Feb 17:43:54 AcpiDaemon? PM: ========== Power Manager thread is starting ==========

06 Feb 17:43:54 AcpiDaemon? APM: ========== APM thread is starting ==========

06 Feb 17:43:54 AcpiDaemon? SLPB Device [SLPB] found

06 Feb 17:43:54 AcpiDaemon? NOTIFY: Calling AcpiTkWaitNotify?()

06 Feb 17:43:56 AcpiDaemon? NOTIFY: Notify Event Device=SLPB Number=80

06 Feb 17:43:56 AcpiDaemon? DoSuspend?: Suspending to state 3 \_S3_

06 Feb 17:43:56 AcpiDaemon? DoSuspend?: VioPopUp? ProcType? 0x4

06 Feb 17:43:56 AcpiDaemon? DoSuspend?: VioPopUp?=480

06 Feb 17:43:56 AcpiDaemon? DoSuspend?: After VioPopUp? ProcType? 0x4

06 Feb 17:43:56 AcpiDaemon? DoSuspend?: KbdFlushBuffer?=464

06 Feb 17:43:56 AcpiDaemon? DoSuspend?: Enter Sleep State

06 Feb 17:46:46 AcpiDaemon? ValidateSleepState?: State '\_S3_' is acceptable

06 Feb 17:46:46 AcpiDaemon? ========== AcpiDaemon? Starting ==========

06 Feb 17:46:46 AcpiDaemon? WARNING! video save mode not set

06 Feb 17:46:46 AcpiDaemon? Error detecting video size

06 Feb 17:46:46 AcpiDaemon? Detected APM version 3.19

06 Feb 17:46:46 AcpiDaemon? LID set to Do Nothing, PowerButton? to Power Off

06 Feb 17:46:46 AcpiDaemon? Suspend support:On State:S3 [\_S3_] VBE:0x0

06 Feb 17:46:46 AcpiDaemon? Log to screen:Off

06 Feb 17:46:46 AcpiDaemon? NOTIFY: ========== Notify thread is starting ==========

06 Feb 17:46:46 AcpiDaemon? PWRBT: ========== Power Button thread is starting ==========

06 Feb 17:46:46 AcpiDaemon? PWRBT: Calling AcpiTkWaitPressButton?

06 Feb 17:46:46 AcpiDaemon? PM: ========== Power Manager thread is starting ==========

06 Feb 17:46:46 AcpiDaemon? APM: ========== APM thread is starting ==========

06 Feb 17:46:46 AcpiDaemon? SLPB Device [SLPB] found

06 Feb 17:46:46 AcpiDaemon? NOTIFY: Calling AcpiTkWaitNotify?()

06 Feb 17:46:48 AcpiDaemon? NOTIFY: Notify Event Device=SLPB Number=80

06 Feb 17:46:48 AcpiDaemon? DoSuspend?: Suspending to state 3 \_S3_

[David Azarewicz]

Very interesting. The data you provided indicates that the sleep button has been pressed and I'm assuming that you didn't press it. What kind of system is this? Laptop or desktop? If desktop does it have a sleep button? Is the wire for the sleep button connected to anything? Can the sleep button be disabled in the BIOS?

[David McKenna]

This is a desktop. No sleep button that I know about. I scoured the BIOS, but found no settings for a sleep button. The manual makes no mention of one either. Strange there is a SLPB device listed...

[David McKenna]

Testlog file

[David Azarewicz]

Well apparently the hardware supports a sleep button since the vendor put it in the ACPI. The reason I asked about a wire dangling in the case is because if it is it could be generating fake button pushes from the electrical noise inside the case. In fact, something like that appears to be what is happening from looking at the log you uploaded. It seems the sleep button is firing over and over. This generates an interrupt and causes the event code to run every time it fires, using cpu time unnecessarily. This interrupt processing is happening even if you ignore it and even if you don't run the daemon. It probably was happening even with 3.19.14, but the 3.19.14 daemon didn't have the code written to act on it.

We can come up with a way to ignore the sleep button, but for now it looks like a hardware problem. The hardware shouldn't be doing that. It will very likely cause you problems in the future, even if the daemon ignores it. For now, don't run the daemon. I'll have a look through your ACPI tables. If I think of anything else, I'll let you know.

[David McKenna]

Hmmm... no dangling wires. I opened the case and removed the board (it is small for the case) and cleaned it carefully both sides. I also inspected every header pin for a sleep button connection but they are all accounted for other things. Put it back together, but still ACPIDaemon sends it to suspend.

I see this in the testlog:

ACPI Error: Could not enable SleepButton? event (20110112/evxfevnt-288)

ACPI Warning: Could not enable fixed event 0x3 (20110112/evxface-295)

AcpiStartEvents?: Sleep Button Install Status=0xF9C0C8D4

Doesn't that imply it shouldn't work (go to sleep)?

[David Azarewicz]

No, that is just a different mechanism which is not implemented on most systems anymore. A vendor could implement the fixed sleep button event, or the SLPB device. Most use the SLPB device and not the fixed event.

[David McKenna]

Version 3.20.01 boots easily with ACPIDAEMON.EXE now.