2.6.13-rc: ACPI_INTERPRETER=y, PCI=n compile error

2.6.13-rc: ACPI_INTERPRETER=y, PCI=n compile error

[Adrian Bunk]

I got the following compile error in 2.6.13-rc6-mm2, but it seems to be 
a problem coming from Linus' tree introduced by the
  [ACPI] S3 resume: avoid kmalloc() might_sleep oops symptom
patch:

<--  snip  -->

...
  LD      .tmp_vmlinux1
drivers/built-in.o: In function `acpi_os_allocate':
: undefined reference to `acpi_in_resume'
make: *** [.tmp_vmlinux1] Error 1

<--  snip  -->

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

RE: 2.6.13-rc: ACPI_INTERPRETER=y, PCI=n compile error

[Brown, Len]

>Subject: 2.6.13-rc: ACPI_INTERPRETER=y, PCI=n compile error
>
>I got the following compile error in 2.6.13-rc6-mm2, but it 
>seems to be 
>a problem coming from Linus' tree introduced by the
>  [ACPI] S3 resume: avoid kmalloc() might_sleep oops symptom
>patch:
>
><--  snip  -->
>
>...
>  LD      .tmp_vmlinux1
>drivers/built-in.o: In function `acpi_os_allocate':
>: undefined reference to `acpi_in_resume'
>make: *** [.tmp_vmlinux1] Error 1
>
><--  snip  -->

Do you have an ACPI-enabled machine that has no PCI?
I'm not aware of any, and would be interested to know
if one exists.

We've had problems with this theoretical build config
for some time:
http://bugzilla.kernel.org/show_bug.cgi?id=1364
because nobody, including me, tests it.

Indeed, one possible fix would be to make CONFIG_ACPI
depend on CONFIG_PCI -- which brings me back to
my origianl question.

-Len

Re: 2.6.13-rc: ACPI_INTERPRETER=y, PCI=n compile error

[Adrian Bunk]

On Wed, Aug 24, 2005 at 11:13:05AM -0400, Brown, Len wrote:
> 
> >Subject: 2.6.13-rc: ACPI_INTERPRETER=y, PCI=n compile error
> >
> >I got the following compile error in 2.6.13-rc6-mm2, but it 
> >seems to be 
> >a problem coming from Linus' tree introduced by the
> >  [ACPI] S3 resume: avoid kmalloc() might_sleep oops symptom
> >patch:
> >
> ><--  snip  -->
> >
> >...
> >  LD      .tmp_vmlinux1
> >drivers/built-in.o: In function `acpi_os_allocate':
> >: undefined reference to `acpi_in_resume'
> >make: *** [.tmp_vmlinux1] Error 1
> >
> ><--  snip  -->
> 
> Do you have an ACPI-enabled machine that has no PCI?
> I'm not aware of any, and would be interested to know
> if one exists.

No, I don't have one.

> We've had problems with this theoretical build config
> for some time:
> http://bugzilla.kernel.org/show_bug.cgi?id=1364
> because nobody, including me, tests it.
> 
> Indeed, one possible fix would be to make CONFIG_ACPI
> depend on CONFIG_PCI -- which brings me back to
> my origianl question.

Making ACPI dependent on PCI is a possible solution for this problem.

> -Len

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed