x86/asm: Use "m" operand constraint in WRUSSQ asm template

The WRUSSQ instruction uses a memory operand, so use "m"
operand constraint instead of forcing usage of pointer
register using "r" constraint. The generated assembly
code improves from:

   6ece3:	48 8d 43 f8          	lea    -0x8(%rbx),%rax
   ...
   6eceb:	66 48 0f 38 f5 18    	wrussq %rbx,(%rax)

to:

   6ecea:	66 48 0f 38 f5 43 f8 	wrussq %rax,-0x8(%rbx)

Signed-off-by: Uros Bizjak <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: Josh Poimboeuf <[email protected]>
Cc: Ard Biesheuvel <[email protected]>
Cc: "H. Peter Anvin" <[email protected]>
Link: https://lore.kernel.org/r/[email protected]

Author: ubizjak

arch/x86/include/asm/special_insns.h

@@ -205,9 +205,9 @@ static inline void clwb(volatile void *__p)
 #ifdef CONFIG_X86_USER_SHADOW_STACK
 static inline int write_user_shstk_64(u64 __user *addr, u64 val)
 {
-	asm goto("1: wrussq %[val], (%[addr])\n"
+	asm goto("1: wrussq %[val], %[addr]\n"
 			  _ASM_EXTABLE(1b, %l[fail])
-			  :: [addr] "r" (addr), [val] "r" (val)
+			  :: [addr] "m" (*addr), [val] "r" (val)
 			  :: fail);
 	return 0;
 fail: