On Wed, May 11, 2022 at 05:27:46AM +0300, Kirill A. Shutemov wrote:
> +#define __untagged_addr(addr, n) \
> + ((__force __typeof__(addr))sign_extend64((__force u64)(addr), n))
> +
> +#define untagged_addr(addr) ({ \
> + u64 __addr = (__force u64)(addr); \
> + if (__addr >> 63 == 0) { \
> + if (current->thread.features & X86_THREAD_LAM_U57) \
> + __addr &= __untagged_addr(__addr, 56); \
> + else if (current->thread.features & X86_THREAD_LAM_U48) \
> + __addr &= __untagged_addr(__addr, 47); \
> + } \
> + (__force __typeof__(addr))__addr; \
> +})
Assuming you got your bits in hardware order:
u64 __addr = addr;
if ((s64)__addr >= 0) {
int lam = (current->thread.features >> X86_THREAD_LAM_U57) & 3;
if (lam)
__addr &= sign_extend64(__addr, 65 - 9*lam);
}
__addr;
has less branches on and should definitely result in better code (or I
need more morning juice).
> +
> +#define untagged_ptr(ptr) ({ \
> + u64 __ptrval = (__force u64)(ptr); \
> + __ptrval = untagged_addr(__ptrval); \
> + (__force __typeof__(*(ptr)) *)__ptrval; \
> +})
> #endif /* !__ASSEMBLY__ */
>
> #ifdef CONFIG_X86_VSYSCALL_EMULATION
> --
> 2.35.1
>
