Procedure Linux follows for shutting down an x86 system when ACPI is not used

Procedure Linux follows for shutting down an x86 system when ACPI is not used

[limp]

Hi all,

I am trying to figure out the procedure that Linux follows for shutting down
an x86 based system when ACPI is not used. From what I've found, when we try
to shutdown the system from user-space, _reboot() system call is called with
"LINUX_REBOOT_CMD_POWER_OFF" as argument. By tracking down the steps
performed in the kernel when the aforementioned system call is involved, I
ended up in kernel_power_off(), unlock_kernel(), and finaly do_exit()
functions.

The kernel_power_off() ultimately calls the  native_machine_shutdown() form
which Linux disables LAPIC, I/O APIC, HPET, etc. Then, after calling the
unlock_kernel(), the do_exit() is called for ultimately shutting down the
system. In the end of this function, I can see that the processor enters an
infinite loop from which it repeatedly executes NOP instructions. That is
actually the behaviour that I experience when I am not using ACPI; the
system is not actually turning off but it goes into an idle state.

Is this how Linux should act when ACPI is not used? I would expect in this
case Linux to invoke an SMI so that the appropriate SMI function actually
shutdown the system but I can't see where (and if) that takes place.

Any help will be greatly appreciated.

John K.

Procedure Linux follows for shutting down an x86 system when ACPI is not used

[Peter Teoh]

To shutdown the CPU, it is just the command "halt" in assembly, but to
shutdown the system' powersupply, and all other component, nowdays it is
hardwired to ACPI, eg:

http://www.thinkwiki.org/wiki/How_to_make_APM_work

http://tuxmobil.org/apm_linux.html

http://www.linuxdocs.org/HOWTOs/mini/Battery-Powered-3.html

<http://forums.anandtech.com/showthread.php?t=487583>So without ACPI,  as
mentioned in URL above, alternative is APM.   Alternatively if u are talking
about SMM/SMI, yes, sounds logical, as that is how system can shutdown
itself after detecting overheating in motherboard (for example) -
temperature monitoring is part of SMM mode.   But I am not sure if u can
trigger SMI that specifically shutdown system or not. (software trigger for
SMI is possible ....port 0xb2)

http://en.wikipedia.org/wiki/System_Management_Mode

Do aware that Intel VT (VMM) can intercept these SMI too:

http://software.intel.com/en-us/forums/showthread.php?t=67301


On Wed, Jun 15, 2011 at 3:53 AM, limp <johnkyr83@hotmail.com> wrote:

> Hi all,
>
> I am trying to figure out the procedure that Linux follows for shutting
> down
> an x86 based system when ACPI is not used. From what I've found, when we
> try
> to shutdown the system from user-space, _reboot() system call is called
> with
> "LINUX_REBOOT_CMD_POWER_OFF" as argument. By tracking down the steps
> performed in the kernel when the aforementioned system call is involved, I
> ended up in kernel_power_off(), unlock_kernel(), and finaly do_exit()
> functions.
>
> The kernel_power_off() ultimately calls the  native_machine_shutdown() form
> which Linux disables LAPIC, I/O APIC, HPET, etc. Then, after calling the
> unlock_kernel(), the do_exit() is called for ultimately shutting down the
> system. In the end of this function, I can see that the processor enters an
> infinite loop from which it repeatedly executes NOP instructions. That is
> actually the behaviour that I experience when I am not using ACPI; the
> system is not actually turning off but it goes into an idle state.
>
> Is this how Linux should act when ACPI is not used? I would expect in this
> case Linux to invoke an SMI so that the appropriate SMI function actually
> shutdown the system but I can't see where (and if) that takes place.
>
> Any help will be greatly appreciated.
>
> John K.
>
>
> _______________________________________________
> Kernelnewbies mailing list
> Kernelnewbies at kernelnewbies.org
> http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
>



-- 
Regards,
Peter Teoh
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.kernelnewbies.org/pipermail/kernelnewbies/attachments/20110615/4a066476/attachment.html