Re: [RFC][PATCH] x86: Cleanup Kconfig

[Dave Jones <davej@codemonkey.org.uk>]

On Wed, Jun 03, 2015 at 06:19:22PM +0200, Ingo Molnar wrote:

 > > On Wed, Jun 03, 2015 at 09:41:32AM +0200, Ingo Molnar wrote:
 > >  > 
 > >  > > 	select GENERIC_CLOCKEVENTS_BROADCAST	if X86_64 || (X86_32 && X86_LOCAL_APIC)
 > >  > 
 > >  > Btw., could we (in a separate patch) turn this into:
 > >  > 
 > >  > > 	select GENERIC_CLOCKEVENTS_BROADCAST
 > >  > 
 > >  > x86 systems without an APIC are rare and rarely tested, we are better off 
 > >  > simplifying our clockevents layout.
 > > 
 > > There's a ton of code in arch/x86 for cpus that don't have apic. Anything Cyrix 
 > > pre-VIA buy out, Anything AMD pre Athlon, did Intel have apic on 486 ?  
 > > Transmeta ?
 > > 
 > > I'm all for abandoning support for 2 decade old junk, but given how long it took 
 > > to drop 386, I wouldn't be surprised if there are still a lot of 586 era people 
 > > still out there playing doom and wearing clothes that are about to come back 
 > > into fashion.
 > 
 > We are not desupporting them in any way - we just simplify a generic clockevents 
 > bit by always enabling it.

For some reason I thought the generic events code implied the need for an apic.
Probably the misleading dependancy.

	Dave