For me, I still stick to uint8_t, since all callers and callee always treat it as uint8_t. It will make the code more clearer for readers.
> Date: Mon, 11 May 2015 15:06:48 -0700 > From: r...@twiddle.net > To: xili_gchen_5...@hotmail.com; peter.mayd...@linaro.org; afaer...@suse.de; > cmetc...@ezchip.com > CC: riku.voi...@iki.fi; w...@tilera.com; qemu-devel@nongnu.org > Subject: Re: [PATCH 05/10 v10] target-tilegx/opcode_tilegx.h: Modify it to > fit qemu using > > On 05/11/2015 02:06 PM, Chen Gang wrote: > > On 5/12/15 00:01, Richard Henderson wrote: > >> On 05/10/2015 03:42 PM, Chen Gang wrote: > >>> -static __inline unsigned int > >>> +static inline uint8_t > >>> get_BFEnd_X0(tilegx_bundle_bits num) > >> > >> Do not change these casts to uint8_t. It's unnecessary churn. > >> > > > > For me, it is enough to return uint8_t, and the caller really treats it > > as uint8_t. So for the function declaration, uint8_t is more precise > > than unsigned int for return type. > > I don't want to argue about this anymore. Drop all the uint8_t and uint16_t. > > > r~