acpi=force safe?

acpi=force safe?

[Nick Bartos]

I manage a little router distro that uses acpi to do a clean shutdown when
the power button is tapped.  So really I am only using the powerdown &
button event features.

The problem I am having is that a lot of the boards I use are older, and
acpi does not get enabled by default.  In the distro I want to keep
everything consistent, so if I fix this problem by forcing acpi on, I will
want to do it globally (for all systems).

I can do acpi=force, but I worry about causing stability issues.

I posted this question on the kernel devel list, and it was suggested to
me that I compile in basic acpi support button support, and nothing else. 
Then I could do a acpi=force and just force on those features.

I am wondering if this will be safe, or if I could run into problems.  I
don't want to sacrifice stability to have these features, but I would like
all my boxes to be consistent.

At least one of the systems is an asus p4b266 (most boards are asus),
which is on the blacklist for acpi=ht.  I tried to force acpi on for this
board which did work, but it gave me a warning about overriding acpi=ht
(which I really don't know what that is supposed to do).

If this is OK, then what other features can I enable and safely force acpi
on?  I would like to eventually implement fan speed and processor temp
monitoring (possibly other health monitoring).  For some router only boxes
it may be good to reduce power consumption, but I don't want to push my
luck if it may not be safe.

I am using the latest 2.4.x pre at the moment.

Comments?





-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: acpi=force safe?

[Andi Kleen]

On Thu, 5 Feb 2004 23:14:15 -0600 (CST)
"Nick Bartos" <spam99-1EIjRwF9Gc5s5w19saIqfg@public.gmane.org> wrote:

> I can do acpi=force, but I worry about causing stability issues.

If the machine works with that it is safe. However many older boards
have broken ACPI, that is why it is disabled by default for older boards.

-Andi


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: acpi=force safe?

[Nick Bartos]

So if I have a /proc/acpi dir then it is safe?  What is the worst that
could happen if the acpi is broken in the board and I force it?  Just
courious, what if I forced it for a really old board that didn't have acpi
at all and thus technically couldn't work anyway (I wouldn't do that but I
do wonder)?  Would the acpi files in proc still exist or not?


> On Thu, 5 Feb 2004 23:14:15 -0600 (CST)
> "Nick Bartos" <spam99-1EIjRwF9Gc5s5w19saIqfg@public.gmane.org> wrote:
>
>> I can do acpi=force, but I worry about causing stability issues.
>
> If the machine works with that it is safe. However many older boards
> have broken ACPI, that is why it is disabled by default for older boards.
>
> -Andi
>
>
> -------------------------------------------------------
> The SF.Net email is sponsored by EclipseCon 2004
> Premiere Conference on Open Tools Development and Integration
> See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
> http://www.eclipsecon.org/osdn
> _______________________________________________
> Acpi-devel mailing list
> Acpi-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
> https://lists.sourceforge.net/lists/listinfo/acpi-devel
>



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

RE: acpi=force safe?

[Yu, Luming]

> So if I have a /proc/acpi dir then it is safe?  What is the worst that
> could happen if the acpi is broken in the board and I force it?  Just
> courious, what if I forced it for a really old board that 
> didn't have acpi
> at all and thus technically couldn't work anyway (I wouldn't 
> do that but I
> do wonder)?  Would the acpi files in proc still exist or not?
> 

It's not safe to determine that ACPI has fully enabled only through the 
existence of /proc/acpi. 

--Luming


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

RE: acpi=force safe?

[Nick Bartos]

Am I to understand that the worse that could happen is that the acpi
features I compiled in (just standard and the button option) just wouldn't
work?  If that is the case then that would be fine.

Really I just don't want things like kernel screw ups (which I thought I
might have had once due to acpi, but I don't remember), or other things
that affect system stability.

If I can compile in these things and force acpi on and not be any worse
off stability wise than if I didn't force it, then that willl be fine for
me.



>> So if I have a /proc/acpi dir then it is safe?  What is the worst that
>> could happen if the acpi is broken in the board and I force it?  Just
>> courious, what if I forced it for a really old board that
>> didn't have acpi
>> at all and thus technically couldn't work anyway (I wouldn't
>> do that but I
>> do wonder)?  Would the acpi files in proc still exist or not?
>>
>
> It's not safe to determine that ACPI has fully enabled only through the
> existence of /proc/acpi.
>
> --Luming
>



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: acpi=force safe?

[Bruno Ducrot]

On Fri, Feb 06, 2004 at 11:16:15AM -0600, Nick Bartos wrote:
> Am I to understand that the worse that could happen is that the acpi
> features I compiled in (just standard and the button option) just wouldn't
> work?  If that is the case then that would be fine.

Just standard mean that all the IRQ routing was done by ACPI
amongst other things.  And you need to do so if you want to receive
events from the power button.

That part of the firmware may be broken, and you may have hang on boot
(even though I think it is safe now without lapic support even on old laptops).

> Really I just don't want things like kernel screw ups (which I thought I
> might have had once due to acpi, but I don't remember), or other things
> that affect system stability.
> 
> If I can compile in these things and force acpi on and not be any worse
> off stability wise than if I didn't force it, then that willl be fine for
> me.

Why not let user decide at boot time?  That easy to do so with syslinux
for example.

-- 
Bruno Ducrot

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: acpi=force safe?

[Nick Bartos]

>
> Why not let user decide at boot time?  That easy to do so with syslinux
> for example.

I do use syslinux, but these boxes are supposed to be shrink rapped
solutions for customers where the user really just uses the web interface
(and some of the boxes are headless), so having the user modify boot parms
really isn't possible.  Also, the entire distro is based on a single
config and upgrading overrites the bootloader config, so setting user
defined defaults really wouldn't work.

As long as there aren't any changes in the linux acpi implementation that
make a board stop working on an upgrade (obviously it will be tested on
the inital install), then it should be fine.  How probable is it for that
to happen?  If the hardware is unchanged I would hope that updating the
kernel (with acpi=force) wouldn't prevent the system from booting.




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: acpi=force safe?

[Nick Bartos]

>
> That part of the firmware may be broken, and you may have hang on boot
> (even though I think it is safe now without lapic support even on old
> laptops).
>

I am compiling it with smp support, so I think it automatically puts in
lapic support.  Is it a bad idea to do that?  I don't have any smp boxes
yet but I wanted to support it if possible.


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: acpi=force safe?

[Bruno Ducrot]

On Fri, Feb 06, 2004 at 04:46:49PM -0600, Nick Bartos wrote:
> 
> >
> > That part of the firmware may be broken, and you may have hang on boot
> > (even though I think it is safe now without lapic support even on old
> > laptops).
> >
> 
> I am compiling it with smp support, so I think it automatically puts in
> lapic support.

lapic support and IO-apic is a good idea ihmo if you have good ethernet
cards especially if you intend to have a lot of traffics in your
router..


-- 
Bruno Ducrot

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: acpi=force safe?

[Bruno Ducrot]

On Fri, Feb 06, 2004 at 03:49:12PM -0600, Nick Bartos wrote:
> >
> > Why not let user decide at boot time?  That easy to do so with syslinux
> > for example.
> 
> I do use syslinux, but these boxes are supposed to be shrink rapped
> solutions for customers where the user really just uses the web interface
> (and some of the boxes are headless), so having the user modify boot parms
> really isn't possible.  Also, the entire distro is based on a single
> config and upgrading overrites the bootloader config, so setting user
> defined defaults really wouldn't work.
> 
> As long as there aren't any changes in the linux acpi implementation that
> make a board stop working on an upgrade (obviously it will be tested on
> the inital install), then it should be fine.  How probable is it for that
> to happen?  If the hardware is unchanged I would hope that updating the
> kernel (with acpi=force) wouldn't prevent the system from booting.
> 

Well, you just have to implement a quality policy then, but I guess
you already have one.

-- 
Bruno Ducrot

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn