On Wed, Nov 8, 2017 at 6:55 AM, Greentime Hu <green.hu@xxxxxxxxx> wrote:
> +#define get_user(x,p) \
> + ({ \
> + const register typeof(*(p)) __user *__p asm("$r0") = (p);\
> + register unsigned long __r2 asm("$r2"); \
> + register int __e asm("$r0"); \
> + switch (sizeof(*(__p))) { \
> + case 1: \
> + __get_user_x(__r2, __p, __e, 1, "$lp"); \
> + break; \
> + case 2: \
> + __get_user_x(__r2, __p, __e, 2, "$lp"); \
> + break; \
> + case 4: \
> + __get_user_x(__r2, __p, __e, 4, "$lp"); \
> + break; \
> + case 8: \
> + __get_user_x(__r2, __p, __e, 8, "$r3", "$lp"); \
> + break; \
> + default: __e = __get_user_bad(); break; \
> + } \
> + x = (typeof(*(p))) __r2; \
> + __e; \
> + })
Something looks odd here: __get_user_bad above looks like it is meant
to provide a link-time error
> +__get_user_bad_8:
> + move $r3, #0
> +__get_user_bad:
> + move $r2, #0
> + move $r0, #-EFAULT
> + ret
> +
but here you actually have a symbol with that name, it gets turned
into a runtime error. I think the first one needs to get renamed to
actually work as expected and force the link error in built-in code
(it works in modules ins __get_user_bad is not exported).
> +
> +__put_user_bad:
> + move $r0, #-EFAULT
> + ret
> +
same here.
Arnd
