Re: [PATCH] x86/uaccess: Implement get_kernel()

[Peter Zijlstra]

On Fri, Apr 10, 2015 at 01:14:27PM +0200, Ingo Molnar wrote:
> +/*
> + * Simple copy-from-possibly-faulting-kernel-addresses method that
> + * avoids the STAC/CLAC SMAP overhead.
> + *
> + * NOTE: this does not propagate the error code of faulting kernel
> + *	 addresses properly. You can recover it via uaccess_catch()
> + *       if you really need to.
> + */
> +#define get_kernel(dst, src)						\
> +do {									\
> +	typeof(*(src)) __val;						\

Should we make that:

	typeof(*(src)) __val = (dst);

> +									\
> +	switch (sizeof(__val)) {					\
> +	case 1:	__get_kernel_asm_ex(__val, src, "b", "b", "=q"); break;	\
> +	case 2: __get_kernel_asm_ex(__val, src, "w", "w", "=r"); break;	\
> +	case 4: __get_kernel_asm_ex(__val, src, "l", "k", "=r"); break;	\
> +	case 8: __get_kernel_asm_ex(__val, src, "q", " ", "=r"); break;	\
> +	default: __get_kernel_BUILD_ERROR();				\
> +	}								\
> +	(dst) = __val;							\
> +} while (0)

Such that when we fault, the value is unmodified? The way it is we'll
assign whatever was on stack for __val, which seems undesirable, no?
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/