diff --git a/arch/x86/core/x86_mmu.c b/arch/x86/core/x86_mmu.c
index 4d1de69cf52..df00e57746e 100644
--- a/arch/x86/core/x86_mmu.c
+++ b/arch/x86/core/x86_mmu.c
@@ -1084,7 +1084,7 @@ static pentry_t flags_to_entry(uint32_t flags)
 	case K_MEM_CACHE_WB:
 		break;
 	default:
-		__ASSERT(false, "bad memory mapping flags 0x%x", flags);
+		return -ENOTSUP;
 	}
 
 	if ((flags & K_MEM_PERM_RW) != 0U) {
@@ -1103,10 +1103,12 @@ static pentry_t flags_to_entry(uint32_t flags)
 }
 
 /* map new region virt..virt+size to phys with provided arch-neutral flags */
-void arch_mem_map(void *virt, uintptr_t phys, size_t size, uint32_t flags)
+int arch_mem_map(void *virt, uintptr_t phys, size_t size, uint32_t flags)
 {
 	range_map_unlocked(virt, phys, size, flags_to_entry(flags),
 			   MASK_ALL, 0);
+
+	return 0;
 }
 
 #if CONFIG_X86_STACK_PROTECTION
diff --git a/kernel/include/kernel_arch_interface.h b/kernel/include/kernel_arch_interface.h
index 2cf4ac9925a..4e014d05619 100644
--- a/kernel/include/kernel_arch_interface.h
+++ b/kernel/include/kernel_arch_interface.h
@@ -241,13 +241,6 @@ static inline bool arch_is_in_isr(void);
  * to this API are assumed to be serialized, and indeed all usage will
  * originate from kernel/mm.c which handles virtual memory management.
  *
- * Architectures are expected to pre-allocate page tables for the entire
- * address space, as defined by CONFIG_KERNEL_VM_BASE and
- * CONFIG_KERNEL_VM_SIZE. This operation should never require any kind of
- * allocation for paging structures.
- *
- * Validation of arguments should be done via assertions.
- *
  * This API is part of infrastructure still under development and may
  * change.
  *
@@ -255,8 +248,12 @@ static inline bool arch_is_in_isr(void);
  * @param addr Page-aligned Source physical address to map
  * @param size Page-aligned size of the mapped memory region in bytes
  * @param flags Caching, access and control flags, see K_MAP_* macros
+ * @retval 0 Success
+ * @retval -ENOTSUP Unsupported cache mode with no suitable fallback, or
+ *	   unsupported flags
+ * @retval -ENOMEM Memory for additional paging structures unavailable
  */
-void arch_mem_map(void *dest, uintptr_t addr, size_t size, uint32_t flags);
+int arch_mem_map(void *dest, uintptr_t addr, size_t size, uint32_t flags);
 
 /**
  * Remove mappings for a provided virtual address range
diff --git a/kernel/mmu.c b/kernel/mmu.c
index 15d984cade0..7cf128f077a 100644
--- a/kernel/mmu.c
+++ b/kernel/mmu.c
@@ -276,6 +276,7 @@ static void frame_mapped_set(struct z_page_frame *pf, void *addr)
  */
 static int map_anon_page(void *addr, uint32_t flags)
 {
+	int ret;
 	struct z_page_frame *pf;
 	uintptr_t phys;
 	bool lock = (flags & K_MEM_MAP_LOCK) != 0;
@@ -284,10 +285,14 @@ static int map_anon_page(void *addr, uint32_t flags)
 	if (pf == NULL) {
 		return -ENOMEM;
 	}
-
 	phys = z_page_frame_to_phys(pf);
-	arch_mem_map(addr, phys, CONFIG_MMU_PAGE_SIZE, flags | K_MEM_CACHE_WB);
 
+	ret = arch_mem_map(addr, phys, CONFIG_MMU_PAGE_SIZE,
+			   flags | K_MEM_CACHE_WB);
+	if (ret != 0) {
+		free_page_frame_list_put(pf);
+		return -ENOMEM;
+	}
 	if (lock) {
 		pf->flags |= Z_PAGE_FRAME_PINNED;
 	}
@@ -380,6 +385,7 @@ void z_phys_map(uint8_t **virt_ptr, uintptr_t phys, size_t size, uint32_t flags)
 {
 	uintptr_t aligned_phys, addr_offset;
 	size_t aligned_size;
+	int ret;
 	k_spinlock_key_t key;
 	uint8_t *dest_addr;
 
@@ -407,7 +413,11 @@ void z_phys_map(uint8_t **virt_ptr, uintptr_t phys, size_t size, uint32_t flags)
 	LOG_DBG("arch_mem_map(%p, 0x%lx, %zu, %x) offset %lu", dest_addr,
 		aligned_phys, aligned_size, flags, addr_offset);
 
-	arch_mem_map(dest_addr, aligned_phys, aligned_size, flags);
+	ret = arch_mem_map(dest_addr, aligned_phys, aligned_size, flags);
+	if (ret != 0) {
+		LOG_ERR("arch_mem_map() failed with %d", ret);
+		goto fail;
+	}
 	k_spin_unlock(&z_mm_lock, key);
 
 	*virt_ptr = dest_addr + addr_offset;