On 07/01/2013 10:51 AM, Andreas Färber wrote: > I wonder, all targets seem to implement the same pattern of passing a > hard-coded bool argument to their internal function. > > Is there any reason not to have translate-all.c call the function with > that true/false directly?
The idea is to avoid runtime checks for the rare search_pc case. Instead we pass constants to inline functions and transform what would be runtime checks into compile-time optimized code paths. r~