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

[Avi Kivity <avi@qumranet.com>]

Arjan van de Ven wrote:
>>> 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 '{ ... }'
>>     
>
>
> config time is not possible (not to mention it's not that uncommon to
> config on a different box than you compile). 

If we add the 'eval' Kconfig keyword, it becomes possible, but I guess 
it isn't worth it if there are more users and anyway split 
configure/compile, as you note, breaks it.


> Makefile side is not that
> hard; in fact what you'd need is a very small check similar to
> scripts/gcc-x86_64-has-stack-protector.sh . While that checks a gcc
> feature, checking the VMX operations via a C program with inline asm is
> actually the most realistic test ANYWAY ...
>   

The problem with that is that the test comes too late: after we've 
configured.  Andrew wants to keep allmodconfig working, and for that we 
need to deselect CONFIG_KVM before compilation starts.

gcc.*protector.sh only affects the Makefile, not the configuration, AFAICT.

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