atomicity fix when enabling paging

- before enabling paging VM asks kernel to resize its segments. This
  may cause kernel to segfault if APIC is used and an interrupt
  happens between this and paging enabled. As these are 2 separate
  vmctl calls it is not atomic. This patch fixes this problem. VM does
  not ask kernel to resize the segments in a separate call anymore.
  The new segments limit is part of the "enable paging" call. It
  generalizes this call in such a way that more information can be
  passed as need be or the information may be completely different if
  another architecture requires this.

kernel/system/do_vmctl.c

@@ -127,15 +127,7 @@ PUBLIC int do_vmctl(struct proc * caller, message * m_ptr)
 		vm_init(p);
 		if(!vm_running)
 			panic("do_vmctl: paging enabling failed");
-		if ((err = arch_enable_paging()) != OK) {
-			return err;
-		}
-		if(newmap(caller, p, (struct mem_map *) m_ptr->SVMCTL_VALUE) != OK)
-			panic("do_vmctl: newmap failed");
-		FIXLINMSG(p);
-		assert(p->p_delivermsg_lin ==
-		  umap_local(p, D, p->p_delivermsg_vir, sizeof(message)));
-		return OK;
+		return arch_enable_paging(caller, m_ptr);
 	case VMCTL_KERN_PHYSMAP:
 	{
 		int i = m_ptr->SVMCTL_VALUE;