mmu: refine function break_page_table

function break_page_table should return next_level_page_size, if
something wrong, it return 0.

the change is valid for release version, as at that time ASSERT()
in break_page_table is empty.

Signed-off-by: Jason Chen CJ <jason.cj.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>

Author: JasonChenCJ

hypervisor/arch/x86/mmu.c

@@ -787,12 +787,12 @@ static uint64_t break_page_table(struct map_params *map_params, void *paddr,
 		/* New entry present - need to allocate a new table */
 		sub_tab_addr = alloc_paging_struct();
 		/* Check to ensure memory available for this structure */
-		if (sub_tab_addr == 0) {
+		if (sub_tab_addr == NULL) {
 			/* Error:
 			 * Unable to find table memory necessary to map memory
 			 */
 			pr_err("Fail to find table memory for map memory");
-			ASSERT(sub_tab_addr == 0, "");
+			ASSERT(0, "fail to alloc table memory for map memory");
 			return 0;
 		}
 
@@ -900,6 +900,8 @@ static int modify_paging(struct map_params *map_params, void *paddr,
 				 */
 				page_size = break_page_table(map_params,
 					paddr, vaddr, page_size, direct);
+				if (page_size == 0)
+					return -EINVAL;
 			}
 		} else {
 			page_size = ((uint64_t)remaining_size < page_size)