Re: i386 very early memory detection cleanup patch breaks the build

["H. Peter Anvin" <hpa@zytor.com>]

James Bottomley wrote:
> The attached should fix it again.

Could you perhaps describe which architecture this is a problem on, and 
what its entry condition looks like?

> This tampering with the trampoline was extraneous to the actual patch. 
> The rule should be that if you don't understand what something is doing,
> don't try to fix it.

I removed it because I removed the VISWS dependency, thus making it 
redundant.  What you seem to be saying is that the dependency should 
have been on SMP not X86_SMP; if that's the issue then please make it so.

I think you just needed to apply your own rule to the above statement.

> In this case CONFIG_X86_TRAMPOLINE is needed for the subarch's that
> provide their own SMP code but still use the standard trampoline.  I
> always thought the visws used the trampoline even in UP boot, but if it
> doesn't, just take out the X86_VISWS dependency.

It doesn't anymore.  The only reason it did was because of stupid 
partitioning between head.S and trampoline.S, which the patch cleans up.

	-hpa