2013/6/17 Richard Henderson <r...@redhat.com>:
> On 06/16/2013 11:55 PM, Kai Tietz wrote:
>> +static bool
>> +ix86_function_attribute_inlinable_p (const_tree fndecl ATTRIBUTE_UNUSED)
>> +{
>> + return true;
>> +}
>
> This is hook_bool_const_tree_true.
Right, we could define macro to this ...
> I have an idea that perhaps the default ought to be true, and that the few
> targets (like mep) that have an interrupt attribute, etc ought to be the ones
> to actually implement this hook.
Yes, that was actual the cause why I did a RFC on that patch. I was
concerned to break some targets by changing the default. I admit that
changing default would be the preferred solution for me, too.
>
> r~
Kai
