diff --git a/hypervisor/arch/x86/e820.c b/hypervisor/arch/x86/e820.c
index 8d410ed88..bdcf82ea7 100644
--- a/hypervisor/arch/x86/e820.c
+++ b/hypervisor/arch/x86/e820.c
@@ -111,11 +111,18 @@ void init_e820(void)
 	uint32_t i;
 
 	if (boot_regs[0] == MULTIBOOT_INFO_MAGIC) {
-		struct multiboot_info *mbi = (struct multiboot_info *)(hpa2hva((uint64_t)boot_regs[1]));
+		/*
+		 * Before installing new PML4 table in enable_paging(), HPA->HVA is always 1:1 mapping
+		 * and hpa2hva() can't be used to do the conversion. Here we simply treat boot_reg[1] as HPA.
+		 */
+		uint64_t hpa = (uint64_t)boot_regs[1];
+		struct multiboot_info *mbi = (struct multiboot_info *)hpa;
 
 		pr_info("Multiboot info detected\n");
 		if ((mbi->mi_flags & MULTIBOOT_INFO_HAS_MMAP) != 0U) {
-			struct multiboot_mmap *mmap = (struct multiboot_mmap *)hpa2hva((uint64_t)mbi->mi_mmap_addr);
+			/* HPA->HVA is always 1:1 mapping at this moment */
+			hpa = (uint64_t)mbi->mi_mmap_addr;
+			struct multiboot_mmap *mmap = (struct multiboot_mmap *)hpa;
 
 			e820_entries_count = mbi->mi_mmap_length / sizeof(struct multiboot_mmap);
 			if (e820_entries_count > E820_MAX_ENTRIES) {
diff --git a/hypervisor/arch/x86/guest/vm.c b/hypervisor/arch/x86/guest/vm.c
index 23b4e625c..9d30dceb6 100644
--- a/hypervisor/arch/x86/guest/vm.c
+++ b/hypervisor/arch/x86/guest/vm.c
@@ -187,7 +187,7 @@ static void create_sos_vm_e820(struct acrn_vm *vm)
 	uint32_t i;
 	uint64_t entry_start;
 	uint64_t entry_end;
-	uint64_t hv_start_pa = get_hv_image_base();
+	uint64_t hv_start_pa = hva2hpa((void *)(get_hv_image_base()));
 	uint64_t hv_end_pa  = hv_start_pa + CONFIG_HV_RAM_SIZE;
 	uint32_t entries_count = get_e820_entries_count();
 	struct e820_entry *entry, new_entry = {0};
@@ -301,7 +301,7 @@ static void prepare_sos_vm_memmap(struct acrn_vm *vm)
 	/* unmap hypervisor itself for safety
 	 * will cause EPT violation if sos accesses hv memory
 	 */
-	hv_hpa = get_hv_image_base();
+	hv_hpa = hva2hpa((void *)(get_hv_image_base()));
 	ept_mr_del(vm, pml4_page, hv_hpa, CONFIG_HV_RAM_SIZE);
 }
 
diff --git a/hypervisor/arch/x86/mmu.c b/hypervisor/arch/x86/mmu.c
index c0a4fb287..9e3e4701e 100644
--- a/hypervisor/arch/x86/mmu.c
+++ b/hypervisor/arch/x86/mmu.c
@@ -204,8 +204,8 @@ void enable_paging(void)
 	CPU_CR_READ(cr0, &tmp64);
 	CPU_CR_WRITE(cr0, tmp64 | CR0_WP);
 
-	CPU_CR_WRITE(cr3, hva2hpa(ppt_mmu_pml4_addr));
-
+	/* HPA->HVA is 1:1 mapping at this moment, simply treat ppt_mmu_pml4_addr as HPA. */
+	CPU_CR_WRITE(cr3, ppt_mmu_pml4_addr);
 }
 
 void enable_smep(void)
@@ -293,6 +293,9 @@ void init_paging(void)
 	/*
 	 * set the paging-structure entries' U/S flag to supervisor-mode for hypervisor owned memroy.
 	 * (exclude the memory reserve for trusty)
+	 *
+	 * Before the new PML4 take effect in enable_paging(), HPA->HVA is always 1:1 mapping,
+	 * simply treat the return value of get_hv_image_base() as HPA.
 	 */
 	hv_hpa = get_hv_image_base();
 	mmu_modify_or_del((uint64_t *)ppt_mmu_pml4_addr, hv_hpa & PDE_MASK,
diff --git a/hypervisor/boot/reloc.c b/hypervisor/boot/reloc.c
index 6c5b41ad1..c28a506f3 100644
--- a/hypervisor/boot/reloc.c
+++ b/hypervisor/boot/reloc.c
@@ -49,7 +49,7 @@ uint64_t get_hv_image_delta(void)
 	return addr;
 }
 
-/* get the actual Hypervisor load address */
+/* get the actual Hypervisor load address (HVA) */
 uint64_t get_hv_image_base(void)
 {
 	return (get_hv_image_delta() + CONFIG_HV_RAM_START);
diff --git a/hypervisor/common/hypercall.c b/hypervisor/common/hypercall.c
index e7d5d70b6..1c9acdbe4 100644
--- a/hypervisor/common/hypercall.c
+++ b/hypervisor/common/hypercall.c
@@ -571,7 +571,7 @@ static int32_t add_vm_memory_region(struct acrn_vm *vm, struct acrn_vm *target_v
 			__func__, vm->vm_id, region->sos_vm_gpa);
 		ret = -EINVAL;
 	} else {
-		base_paddr = get_hv_image_base();
+		base_paddr = hva2hpa((void *)(get_hv_image_base()));
 		if (((hpa <= base_paddr) && ((hpa + region->size) > base_paddr)) ||
 				((hpa >= base_paddr) && (hpa < (base_paddr + CONFIG_HV_RAM_SIZE)))) {
 			pr_err("%s: overlap the HV memory region.", __func__);
@@ -719,7 +719,7 @@ static int32_t write_protect_page(struct acrn_vm *vm,const struct wp_data *wp)
 		dev_dbg(ACRN_DBG_HYCALL, "[vm%d] gpa=0x%x hpa=0x%x",
 				vm->vm_id, wp->gpa, hpa);
 
-		base_paddr = get_hv_image_base();
+		base_paddr = hva2hpa((void *)(get_hv_image_base()));
 		if (((hpa <= base_paddr) && ((hpa + PAGE_SIZE) > base_paddr)) ||
 				((hpa >= base_paddr) &&
 				(hpa < (base_paddr + CONFIG_HV_RAM_SIZE)))) {