Andrew said: > Here's an example: > > There's a function called control_hook. It is defined in a > > prototype > prior to its use in device_hooks. Since control_hook actually > performs a number of different behaviors, depending on the > argument, each was broken out into a separate function, which was > defined before control_hook. No other functions are going to call > these functions, so there's really no point to define prototypes for > them in the header or anywhere else. + > But this seems really stupid. Does anyone have a strong argument for > preserving this warning? > Perhaps there is an alternative warning we could use instead? For such case, aka static/local to module functions, prefixing them with "static", as it should have been BTW ;-), will silence the compiler. When we change to "all warnings" setup, I've done it in several places of network drivers and modules to silence compiler. > Or perhaps we could at least > remove it from CCFLAGS and keep it for C++FLAGS? I'm not for a solution *solving* warnings generation, I'm for solution fixing the warning *origin*. Compiler is your friend. Most of the time. That's true for warnings too. - Philippe -- Fortune Cookie Says: You have the body of a 19 year old. Please return it before it gets wrinkled.