Re: [PATCH 2/9] x86: introduce a set of platform feature flags

[Ingo Molnar <mingo@elte.hu>]

* H. Peter Anvin <hpa@linux.intel.com> wrote:

> Ingo Molnar wrote:
> > 
> > Btw., this enumeration of basic PC features isnt bad in itself - and 
> > if there's a boot-flag based detection method (like on MRST) then 
> > this can convey a 'should this platform driver attempt to 
> > initialize' information to the driver, and rather cleanly so.
> > 
> 
> That is the whole point of this, yes. [...]

Except that about half of all the actual uses of platform_has() are 
all but related to the "should this driver initialize" question ;-)

They are more used as "fudge platform code a bit here and there to 
meet the limited constraints of the platform".

And that kind of fudging is a problem, obviously.

> [...] Furthermore, it is expected that we *also* would be able to 
> set platform flags based on early probes.

Yeah.

> > But there's bad uses of this as well, and those bad uses seem to 
> > dominate this patch-set.
> 
> Not entirely surprising, since this stuff is more or less randomly 
> sprinkled through the existing code.  We have a huge bunch of 
> legacy code and yes, it needs cleanup.
> 
> This is a huge opportunity, of course, but it's going to be a lot 
> of work.  A lot of this is likely to overlap directly with the 
> needs of Xen, too.

Correct.

	Ingo