On Fri, Sep 2, 2011 at 9:47 PM, Anthony Liguori <aligu...@us.ibm.com> wrote:
> Hi,
>
> There have been a few attempts in the past to allow TCG to be disabled
> at build time. Recently, Alex made the suggestion that we could do it by
> using
> the same trick that we used to introduce kvm support. That involves
> introducing
> a tcg_enabled() macro that will be (0) if TCG is disabled in the build.
>
> GCC is smart enough to do dead code elimination if it sees an if (0) and the
> result is that if you can do:
Is this also true for gcc optimization -O0? Didn't we have breakages
because of similar issues recently?
>
> if (tcg_enabled()) {
> foo();
> }
>
> and it's more or less equivalent to:
>
> #ifdef CONFIG_TCG
> foo();
> #endif
>
> Without the ugliness that comes from using the preprocessor. I think this
> ended
> up being pretty straight forward. exec.c could use a fair bit of cleanup but
> other than that, this pretty much eliminates all of the TCG code from the
> build.
>
> This absolutely is going to break non-x86 KVM builds if they use the
> --disable-tcg flag as I haven't tested those yet. The normal TCG build
> shouldn't be affected at all though.
>
> In principle, the code assumes that you need KVM if you don't have TCG. Of
> course, some extra logic could be added to allow for Xen if TCG isn't present.
>
>
>
