Fix undefined behavior in hint::spin_loop for x86 targets without SSE2

The pause instruction requires SSE2 but was being unconditionally used
on targets without it, resulting in undefined behavior.

This PR fixes that by only using the pause intrinsic if SSE2 is available.

It also removes the inline assembly which was not required since these
instructions are available in core::arch, and extends support of
the spin_loop hint to arm targets with the v6 feature which also
support the yield instruction.

Closes #59237 .

src/libcore/hint.rs

@@ -2,7 +2,7 @@
 
 //! Hints to compiler that affects how code should be emitted or optimized.
 
-use intrinsics;
+use {arch, intrinsics};
 
 /// Informs the compiler that this point in the code is not reachable, enabling
 /// further optimizations.
@@ -62,13 +62,32 @@ pub unsafe fn unreachable_unchecked() -> ! {
 #[inline]
 #[unstable(feature = "renamed_spin_loop", issue = "55002")]
 pub fn spin_loop() {
-    #[cfg(any(target_arch = "x86", target_arch = "x86_64"))]
-    unsafe {
-        asm!("pause" ::: "memory" : "volatile");
+    #[cfg(
+        all(
+            any(target_arch = "x86", target_arch = "x86_64"),
+            target_feature = "sse2"
+        )
+    )] {
+        #[cfg(target_arch = "x86")] {
+            unsafe { arch::x86::_mm_pause() };
+        }
+
+        #[cfg(target_arch = "x86_64")] {
+            unsafe { arch::x86_64::_mm_pause() };
+        }
     }
 
-    #[cfg(target_arch = "aarch64")]
-    unsafe {
-        asm!("yield" ::: "memory" : "volatile");
+    #[cfg(
+        any(
+            target_arch = "aarch64",
+            all(target_arch = "arm", target_feature = "v6")
+        )
+    )] {
+        #[cfg(target_arch = "aarch64")] {
+            unsafe { arch::aarch64::__yield() };
+        }
+        #[cfg(target_arch = "arm")] {
+            unsafe { arch::arm::__yield() };
+        }
     }
 }