Hi
I am trying to understand how the power button pressing event triggers the machine to shutdown. From the ACPI spec 4.7.2.2.1, it mentioned that the power button can be implemented either as a fixed hardware or a generic hardware programming method. So, for ASUS M2V-MX SE, how the power button is implemented?
I tested coreboot with Linux on M2V-MX SE. When the Linux stops at login window, pressing the power button can lead the machine to shutdown. So what are the concrete steps between the power button is pressed to the Linux begins to shutdown? Is it a AML code get executed? (but I read the M2V-MX SE dsdt source file, and didn't find anything related to power button.) Will the event go to the BIOS first? Then go to the OS? Or it is completely handled by OS?
The problem I try to solve is to find out whether BIOS can get the button event before the OS and do something without OS.
Also, for the power button override event, it is completely handled by hardware, right? Is it possible to trigger this event from the BIOS?
My questions may seem to be a little bit strange. Feel free to let me know, If you still not clear about the questions. Thanks.
Regards,
Jiang
* Jiang Wang jwangzju@gmail.com [110422 01:33]:
I am trying to understand how the power button pressing event triggers the machine to shutdown. From the ACPI spec 4.7.2.2.1, it mentioned that the power button can be implemented either as a fixed hardware or a generic hardware programming method. So, for ASUS M2V-MX SE, how the power button is implemented?
I tested coreboot with Linux on M2V-MX SE. When the Linux stops at login window, pressing the power button can lead the machine to shutdown. So what are the concrete steps between the power button is pressed to the Linux begins to shutdown? Is it a AML code get executed? (but I read the M2V-MX SE dsdt source file, and didn't find anything related to power button.) Will the event go to the BIOS first? Then go to the OS? Or it is completely handled by OS?
Check ICH7 sleepstates.asl and fadt.c Some values are taken from there to power off the machine. Look at FILO source code or ich7 SMI code to see how that can be done outside of ACPI.
The problem I try to solve is to find out whether BIOS can get the button event before the OS and do something without OS.
If the OS is an ACPI OS, it will (usually) get the event through a SCI. If not, a SMI can be caused and the SMI handler can handle the event. You could possibly hack something like having POWERBTN cause a SMI, then do something in the smi handler, then retrigger a SCI to the OS. It seems like the wrong thing though. If you want to execute "BIOS" code in the OS context, put it into the DSDT, or even better just put it in user land to begin with.
Also, for the power button override event, it is completely handled by hardware, right? Is it possible to trigger this event from the BIOS?
Yes, that is done in hardware. You can catch a power press in SMM and just switch the machine off immediately. In the ICH7 code this is done when ACPI is not active.
-
Jiang Wang -
Stefan Reinauer