Align kernel size up to 512 boundary instead of loading one block extra

Might fix phil-opp/blog_os#409.

Author: phil-opp

src/second_stage.s

@@ -30,8 +30,8 @@ load_kernel_from_disk:
 
     # block count
     mov ecx, _kib_kernel_size
+    add ecx, 511 # align up
     shr ecx, 9
-    add ecx, 1
 
 load_next_kernel_block_from_disk:
     # load block from disk