On 28.08.2011, at 09:08, Peter Maydell <peter.mayd...@linaro.org> wrote:
> On 28 August 2011 14:47, Alexander Graf <ag...@suse.de> wrote: >> Well, if we want host machine dependent types we should use host >> dependent types. > > ...we couldn't decide what we wanted :-) > >> Apparently uint16 is expected to be uint32_t in the code IIUC. > > If the softfloat code relies on uint16 being exactly 32 bits > (or indeed on it being exactly 16 bits) then it has a bug in it. My point is that we can have it either host-dependent or not. Both apparently doesn't work :) So the agnostic deterministic way would be to always use the same static type on all hosts. Alex >
