[linux-pm] [RFC][PATCH] PM: Introduce new top level suspend and hibernation callbacks (rev. 2)

[Rafael J. Wysocki]

On Friday, 21 of March 2008, Sam Ravnborg wrote:
> Hi Rafael.

Hi Sam,

> Is it possible to extend this in some way so we avoid the
> #ifdef stuff in the drivers?

Well, I'd love to do something like this.

> We could introduce a few special sections that we discard if
> PM is not in use.
> We have a reliable build time infrastructure to detect
> inconsistencies if needed.
> 
> Something like:
> #define __suspend     __section(.suspend.text)
> #define __suspenddata __section(.suspend.data)
> 
> #define __hibernate     __section(.hibernate.text)
> #define __hibernatedata __section(.hibernate.data)
> 
> A few more tricks will be needed when we assign the functon pointers.
> We have __devexit_p(*) and we may use something similar.

Unfortunately, I have a little experience with linkers and I don't think I'll
be able to do anything like this in a reasonable time without any help.

Thanks,
Rafael