Re: [PATCH 0/14] KVM: Kernel-based Virtual Machine (v4)

[Avi Kivity <avi@qumranet.com>]

Andrew Morton wrote:
> On Wed, 08 Nov 2006 09:14:50 +0200
> Avi Kivity <avi@qumranet.com> wrote:
>
>   
>> Roland Dreier wrote:
>>     
>>>  > That's gas 2.16.1.  I assume it needs some super-new binutils.
>>>  > 
>>>  > I'm not sure what to do about this.  What's the minimum version?
>>>
>>> According to http://kvm.sourceforge.net/howto.html :
>>>     A recent enough binutils (>= 2.16.91.0.2) for vmx instruction support
>>>   
>>>       
>> Either that or a bunch of ugly .byte macros.
>>
>>     
>
> I think we could live with the binutils requirement as long as we can find
> some automagic way of not breaking people's `make allmodconfig'.  Because
> quite a lot of those people who do cross-compilation tend to use older
> binutilses.
>   

These crosses are usually for $wierd target on x86 host, right?  But no 
one will compile kvm for non-x86:

        depends on X86 && EXPERIMENTAL

> But I don't know how to do that.  We _could_ do a trick similar to the
> `cc-version' make rule, and then use the new `as-version' to make the whole
> kvm.o compile down to an empty .o file.  But that's pretty hacky.  

Not only hacky, it will confuse the regular user who gets a nonworking 
kvm.ko due to old binutils.


> Would
> really prefer something at Kconfig-time, but we have no way of letting the
> assembler version feed into the Kconfig system (nor do we want it, I
> suspect).
>   

config AS_VERSION
        eval as --version | awk '{ ... }'

?

-- 
error compiling committee.c: too many arguments to function