2.6.25-gitX and pmd_bad() on x86_32

2.6.25-gitX and pmd_bad() on x86_32

[Joerg Friedrich]

Hi!

Yesterday I tried to compile 2.6.25-git5 but I ran into a "Unknown
Symbol" (pmd_bad)  warning from linker while compiling the nvidia-binary module.

I tracked this issue down to this git-commit:
9fc34113f6880b215cbea4e7017fc818700384c2

AFAICS, x86_32 is the only architecture where pmd_bad is neither a
define or a static inline. Including asm/pgtable.h is no longer
sufficient on x868_32. Is this behaviour intended or is there just a
EXPORT_SYMBOL missing?

-- 
Jörg Friedrich

There are only 10 types of people:
Those who understand binary and those who don't.

Re: 2.6.25-gitX and pmd_bad() on x86_32

[Andi Kleen]

Joerg Friedrich <Joerg.Friedrich@friedrich-kn.de> writes:

> Yesterday I tried to compile 2.6.25-git5 but I ran into a "Unknown
> Symbol" (pmd_bad)  warning from linker while compiling the nvidia-binary module.

pmd_bad is always optional. You can safely make it return 0 in some local
macro.

> AFAICS, x86_32 is the only architecture where pmd_bad is neither a
> define or a static inline. Including asm/pgtable.h is no longer
> sufficient on x868_32. Is this behaviour intended or is there just a
> EXPORT_SYMBOL missing?

Kernel changes generally don't care about out of tree modules.
Also pmd_bad() is kind of internal anyways.

-Andi