Re: [PATCH 1/3] PAD helper for native and paravirt platform

[Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>]

On Fri, Feb 17, 2012 at 05:59:56PM +0000, Liu, Jinsong wrote:
> >> 
> >> +static inline int __acpi_pad_init(void)
> >> +{
> >> +	return PVOP_CALL0(int, pv_pad_ops.acpi_pad_init); +}
> >> +
> >> +static inline void __acpi_pad_exit(void)
> >> +{
> >> +	PVOP_VCALL0(pv_pad_ops.acpi_pad_exit);
> >> +}
> > 
> > With this you, aiui, you aim at getting the calls patched. Are the
> > callers of this really on performance critical paths? If not, the
> > simpler approach of having an ops structure the fields of which get
> > overwritten by 
> > Xen initialization would seem a more appropriate approach.
> > 
> 
> Yes, I agree. I code in this way just want to keep same coding style as other pv functions of paravirt.h.
> I update the patch w/ a simpler approach, and will post later.
> Of course, we need Konrad's comments.

The thing is that the paravirt approach also impacts lguests. While I don't
think your patch will affect it, it just doesn't seem like the right place.

It seems that a more general approach, like the x86 one would be appropiate.
Or since this is ACPI related - perhaps in the drivers/acpi/osl.c ? - That is
all "OS dependent functions" and seem proper?