Re: [Fastboot] Fw: Problems with reboot/poweroff on SMP machine

Re: [Fastboot] Fw: Problems with reboot/poweroff on SMP machine

[Andrew Morton]

(Added acpi people)

ebiederm-aS9lmoZGLiVWk0Htik3J/w@public.gmane.org (Eric W. Biederman) wrote:
>
> ebiederm-aS9lmoZGLiVWk0Htik3J/w@public.gmane.org (Eric W. Biederman) writes:
> 
> > Martin MOKREJ___ <mmokrejs-FKSC99mem8wheYDy1lqhK0JFmxxWawaa@public.gmane.org> writes:
> >
> >> Hi Eric,
> >>   I don't know if I was epxected to test this right away, but
> >> I just tried over acpi patched sources on my i686:
> >
> > Thanks.  That is my bug. 
> >
> > I knew the code wasn't quite right but I missed how off it
> > was :)
> >
> > Just remove the BUG_ON(!shutdown_prepared) line and it should
> > compile.  I don't know why I though I could read a static
> > variable in another function...
> 
> I am now officially confused.  My patch might clean up a corner
> case or two but it really should not help, because the issue
> I was worried about is mostly fixed.
> 
> All of the pieces are there for the acpi power off case to work
> correctly so is beginning to look like a completely different bug
> then what I am suspecting.  Anyone who can testing of my patch
> would be a help.
> 
> The only other interesting data point I can think of at the moment
> is the attached patch below.  
> 
> I want to know if acpi_power_off is having problems because interrupts
> get disabled.

I have a vague feeling that adding local_irq_disable() to acpi_power_off()
is illegal.

void acpi_power_off(void)
{
	printk("%s called\n", __FUNCTION__);
	local_irq_disable();
	acpi_enter_sleep_state(ACPI_STATE_S5);
}

Can an ACPI person please comment?

> Eric
> 
> diff --git a/arch/i386/kernel/i8259.c b/arch/i386/kernel/i8259.c
> --- a/arch/i386/kernel/i8259.c
> +++ b/arch/i386/kernel/i8259.c
> @@ -283,7 +283,7 @@ static struct sysdev_class i8259_sysdev_
>         set_kset_name("i8259"),
>         .suspend = i8259A_suspend,
>         .resume = i8259A_resume,
> -       .shutdown = i8259A_shutdown,
> +//     .shutdown = i8259A_shutdown,
>  };
> 
>  static struct sys_device device_i8259A = {


-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_idt77&alloc_id\x16492&op=click

Re: [Fastboot] Fw: Problems with reboot/poweroff on SMP machine

[Eric W. Biederman]

Andrew Morton <akpm-3NddpPZAyC0@public.gmane.org> writes:

> (Added acpi people)
>
> I have a vague feeling that adding local_irq_disable() to acpi_power_off()
> is illegal.

I didn't my preliminary patch just fat fingered that line.
All I was really doing was removing acpi_sleep_prepare(),
to be certain we weren't calling it at the wrong time.


> void acpi_power_off(void)
> {
> 	printk("%s called\n", __FUNCTION__);
> 	local_irq_disable();
> 	acpi_enter_sleep_state(ACPI_STATE_S5);
> }
>
> Can an ACPI person please comment?

>From drivers/acpi/hwsleep.c
> /*******************************************************************************
>  *
>  * FUNCTION:    acpi_enter_sleep_state
>  *
>  * PARAMETERS:  sleep_state         - Which sleep state to enter
>  *
>  * RETURN:      Status
>  *
>  * DESCRIPTION: Enter a system sleep state (see ACPI 2.0 spec p 231)
>  *              THIS FUNCTION MUST BE CALLED WITH INTERRUPTS DISABLED
>  *
>  ******************************************************************************/
> 
> acpi_status asmlinkage
> acpi_enter_sleep_state (

So at least according to the comments it would be an error
to call that function with interrupts enabled.

So either there is some weird control flow going on that gets
acpi_sleep_prepare called with interrupts disabled.  Or there
is an entirely different problem showing up with acpi_power_off.

My current hunch is some very weird control flow possibly the system
devices not being shutdown in the expected order but I don't know
at the moment.  

I have one bug report that remove the i8259A shutdown fixes
the problem in 2.6.13-rc2.  What I am having a very hard time
seeing is how it could possibly make a difference.

Eric


-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click

Re: [Fastboot] Fw: Problems with reboot/poweroff on SMP machine

[Martin MOKREJŠ]

The patch below really doesn't help of ASUS P4C800E-Deluxe when
applied to plain 2.6.13-rc3-git7. The machines turns the power immediately
again.
Martin

Andrew Morton wrote:
> (Added acpi people)
> 
> ebiederm-aS9lmoZGLiVWk0Htik3J/w@public.gmane.org (Eric W. Biederman) wrote:
> 
>>ebiederm-aS9lmoZGLiVWk0Htik3J/w@public.gmane.org (Eric W. Biederman) writes:
>>
>>
>>>Martin MOKREJ___ <mmokrejs-FKSC99mem8wheYDy1lqhK0JFmxxWawaa@public.gmane.org> writes:
>>>
>>>
>>>>Hi Eric,
>>>>  I don't know if I was epxected to test this right away, but
>>>>I just tried over acpi patched sources on my i686:
>>>
>>>Thanks.  That is my bug. 
>>>
>>>I knew the code wasn't quite right but I missed how off it
>>>was :)
>>>
>>>Just remove the BUG_ON(!shutdown_prepared) line and it should
>>>compile.  I don't know why I though I could read a static
>>>variable in another function...
>>
>>I am now officially confused.  My patch might clean up a corner
>>case or two but it really should not help, because the issue
>>I was worried about is mostly fixed.
>>
>>All of the pieces are there for the acpi power off case to work
>>correctly so is beginning to look like a completely different bug
>>then what I am suspecting.  Anyone who can testing of my patch
>>would be a help.
>>
>>The only other interesting data point I can think of at the moment
>>is the attached patch below.  
>>
>>I want to know if acpi_power_off is having problems because interrupts
>>get disabled.
> 
> 
> I have a vague feeling that adding local_irq_disable() to acpi_power_off()
> is illegal.
> 
> void acpi_power_off(void)
> {
> 	printk("%s called\n", __FUNCTION__);
> 	local_irq_disable();
> 	acpi_enter_sleep_state(ACPI_STATE_S5);
> }
> 
> Can an ACPI person please comment?
> 
> 
>>Eric
>>
>>diff --git a/arch/i386/kernel/i8259.c b/arch/i386/kernel/i8259.c
>>--- a/arch/i386/kernel/i8259.c
>>+++ b/arch/i386/kernel/i8259.c
>>@@ -283,7 +283,7 @@ static struct sysdev_class i8259_sysdev_
>>        set_kset_name("i8259"),
>>        .suspend = i8259A_suspend,
>>        .resume = i8259A_resume,
>>-       .shutdown = i8259A_shutdown,
>>+//     .shutdown = i8259A_shutdown,
>> };
>>
>> static struct sys_device device_i8259A = {
> 
> 
> 

-- 
Martin Mokrejs
Email: 'bW9rcmVqc21Acmlib3NvbWUubmF0dXIuY3VuaS5jeg==\n'.decode('base64')
GPG key is at http://www.natur.cuni.cz/~mmokrejs


-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click