compiler-rt: prevent llvm making memcpy recursive

before this commit, when compiling for target riscv64-linux-none and
with `-mcpu basline`, recursive jalr instructions were generated that
caused infinite looping and a stack overflow for copy lengths between 16
and 31 (inclusive).

This was caused by LLVM optimisations - in particular LLVM has an
optimisation that can turn copies of a sufficient length into a call to
memcpy. This change breaks up some copies in order to try and prevent
this optimisation from occuring.

lib/compiler_rt/memcpy.zig

@@ -143,8 +143,23 @@ inline fn memcpy_range2(
     comptime std.debug.assert(std.math.isPowerOfTwo(min));
     const copy_len = min;
     const last = len - copy_len;
-    dest[0..copy_len].* = src[0..copy_len].*;
-    dest[last..][0..copy_len].* = src[last..][0..copy_len].*;
+    if (copy_len > size) { // comptime-known
+        // we do these copies 1 CopyType at a time to prevent llvm turning this into a call to memcpy
+        const d: [*]align(1) CopyType = @ptrCast(dest);
+        const s: [*]align(1) const CopyType = @ptrCast(src);
+        const count = @divExact(copy_len, size);
+        inline for (d[0..count], s[0..count]) |*r, v| {
+            r.* = v;
+        }
+        const dl: [*]align(1) CopyType = @ptrCast(dest + last);
+        const sl: [*]align(1) const CopyType = @ptrCast(src + last);
+        inline for (dl[0..count], sl[0..count]) |*r, v| {
+            r.* = v;
+        }
+    } else {
+        dest[0..copy_len].* = src[0..copy_len].*;
+        dest[last..][0..copy_len].* = src[last..][0..copy_len].*;
+    }
 }
 
 test "aligned" {