diff --git a/include/target.mk b/include/target.mk
index 530e711fa5..dfa46f462f 100644
--- a/include/target.mk
+++ b/include/target.mk
@@ -209,6 +209,9 @@ ifeq ($(DUMP),1)
   endif
   ifeq ($(ARCH),aarch64)
     CPU_TYPE ?= generic
+    ifneq ($(filter $(FEATURES),fpu),)
+      CPU_CFLAGS += -march=armv8-a+crypto
+    endif
     CPU_CFLAGS_generic = -mcpu=generic
     CPU_CFLAGS_cortex-a53 = -mcpu=cortex-a53
   endif
diff --git a/package/ctcgfw/naiveproxy/Makefile b/package/ctcgfw/naiveproxy/Makefile
index 65b53c5316..1940a5abc7 100644
--- a/package/ctcgfw/naiveproxy/Makefile
+++ b/package/ctcgfw/naiveproxy/Makefile
@@ -8,13 +8,12 @@
 include $(TOPDIR)/rules.mk
 
 PKG_NAME:=naiveproxy
-PKG_VERSION:=88.0.4324.96-1
-PKG_RELEASE:=2
+PKG_VERSION:=89.0.4389.72-1
+PKG_RELEASE:=1
 
 PKG_SOURCE:=$(PKG_NAME)-$(PKG_VERSION).tar.gz
 PKG_SOURCE_URL:=https://codeload.github.com/klzgrad/naiveproxy/tar.gz/v$(PKG_VERSION)?
-PKG_HASH:=287c5291d0371a0d1e06f08fcb29c1d80da4a07e557fb8d2bf770b357a8eb29a
-PKG_BUILD_DIR:=$(BUILD_DIR)/$(PKG_NAME)-$(PKG_VERSION)
+PKG_HASH:=94b1c3f1e63ecab43eb8006a38ed1a76f1e510364ba651abaef0ec575449f687
 
 PKG_LICENSE:=BSD 3-Clause
 PKG_LICENSE_FILES:=LICENSE
@@ -63,32 +62,32 @@ ifneq ($(CONFIG_CCACHE),)
 	export naive_ccache_flags=cc_wrapper="ccache"
 endif
 
-AFDO_VER:=88.0.4324.91_rc-r1-merged
-AFDO_FILE:=chromeos-chrome-amd64-$$(AFDO_VER).afdo.bz2
+AFDO_VER:=89.0.4389.65_rc-r1-merged
+AFDO_FILE:=chromeos-chrome-amd64-$(AFDO_VER).afdo.bz2
 define Download/AFDO_PROF
 	URL:=https://storage.googleapis.com.cnpmjs.org/chromeos-prebuilt/afdo-job/llvm \
 	     https://storage.googleapis.com/chromeos-prebuilt/afdo-job/llvm
 	URL_FILE:=$(AFDO_FILE)
 	FILE:=$(AFDO_FILE)
-	HASH:=d918ae36f535ffd7a3c36c7d06a0a58a3d93f03d07a670cee0f7cb4538925b34
+	HASH:=75a176d57c9b5fb12bff7e9442ddf21226df8aa25de19266f7ceb426eafd3421
 endef
 
-CLANG_VER:=llvmorg-12-init-11462-g418f18c6-1
-CLANG_FILE:=clang-$$(CLANG_VER).tgz
+CLANG_VER:=llvmorg-12-init-16296-g5e476061-1
+CLANG_FILE:=clang-$(CLANG_VER).tgz
 define Download/CLANG
 	URL:=https://commondatastorage.googleapis.com/chromium-browser-clang/Linux_x64
 	URL_FILE:=$(CLANG_FILE)
 	FILE:=$(CLANG_FILE)
-	HASH:=59e4682179bbda9b1cf7fcc5c6e3e24445d6e9e8411bd1c893882f63c6b65308
+	HASH:=78729dfab39fb9d52ede2841e408cf505959b26a5d0b086ed7cf4f784f2b393a
 endef
 
-GN_VER:=53d92014bf94c3893886470a1c7c1289f8818db0
+GN_VER:=595e3be7c8381d4eeefce62a63ec12bae9ce5140
 GN_FILE:=gn-git_revision-$(GN_VER).zip
 define Download/GN_TOOL
 	URL:=https://chrome-infra-packages.appspot.com/dl/gn/gn/linux-amd64/+
 	URL_FILE:=git_revision:$(GN_VER)
 	FILE:=$(GN_FILE)
-	HASH:=e39d3da74cb2802614c0edc4877483ff50045abedb98adb40eaadb68ab4066ff
+	HASH:=9d61dcab3814868786768c9dfab71e0cd850a61dbad9a049536d2f1aae784fe4
 endef
 
 PGO_VER:=4324-1610733451-8576f5ff23113bddfb13d417df9ef2a9be1f740d
diff --git a/package/ctcgfw/naiveproxy/patches/010-Fix-OpenWrt-x86-builds.patch b/package/ctcgfw/naiveproxy/patches/010-Fix-OpenWrt-x86-builds.patch
deleted file mode 100644
index ab85c7b063..0000000000
--- a/package/ctcgfw/naiveproxy/patches/010-Fix-OpenWrt-x86-builds.patch
+++ /dev/null
@@ -1,22 +0,0 @@
-From 63f89cd454bcd1f2cdd14e63de0587a2f692cd07 Mon Sep 17 00:00:00 2001
-From: klzgrad <kizdiv@gmail.com>
-Date: Sun, 31 Jan 2021 20:15:28 +0800
-Subject: [PATCH] Fix OpenWrt x86 builds
-
----
- src/build/toolchain/linux/BUILD.gn | 6 +++---
- 1 file changed, 3 insertions(+), 3 deletions(-)
-
---- a/src/build/toolchain/linux/BUILD.gn
-+++ b/src/build/toolchain/linux/BUILD.gn
-@@ -99,8 +99,8 @@ clang_toolchain("clang_x86") {
- clang_toolchain("clang_x86_openwrt") {
-   # Output linker map files for binary size analysis.
-   enable_linker_map = true
--  extra_cppflags = "--target=i386-openwrt-linux-musl -D_LIBCPP_HAS_MUSL_LIBC -D__UCLIBC__"
--  extra_ldflags = "--target=i386-openwrt-linux-musl"
-+  extra_cppflags = "--target=i486-openwrt-linux-musl -D_LIBCPP_HAS_MUSL_LIBC -D__UCLIBC__"
-+  extra_ldflags = "--target=i486-openwrt-linux-musl"
- 
-   toolchain_args = {
-     current_cpu = "x86"
diff --git a/package/ctcgfw/naiveproxy/src/init_env.sh b/package/ctcgfw/naiveproxy/src/init_env.sh
index 1aaf1760f7..0260616f02 100755
--- a/package/ctcgfw/naiveproxy/src/init_env.sh
+++ b/package/ctcgfw/naiveproxy/src/init_env.sh
@@ -16,7 +16,7 @@ naive_arch="${target_arch}"
 [ "${target_arch}" == "i386" ] && naive_arch="x86"
 [ "${target_arch}" == "x86_64" ] && naive_arch="x64"
 [ "${target_arch}" == "aarch64" ] && naive_arch="arm64"
-ldso_path="/lib/$(find "${toolchain_dir}/" | grep -Eo "ld-musl-[a-z0-9_-]+\\.so\\.1")"
+# ldso_path="/lib/$(find "${toolchain_dir}/" | grep -Eo "ld-musl-[a-z0-9_-]+\\.so\\.1")"
 
 # OS detection
 [ "$(uname)" != "Linux" -o "$(uname -m)" != "x86_64" ] && { echo -e "Support Linux AMD64 only."; exit 1; }
@@ -81,8 +81,8 @@ ozone_platform_headless=true
 current_os=\"linux\"
 current_cpu=\"${naive_arch}\"
 sysroot=\"${toolchain_dir}\"
-custom_toolchain=\"//build/toolchain/linux:clang_${naive_arch}_openwrt\"
-ldso_path=\"${ldso_path}\""
+custom_toolchain=\"//build/toolchain/linux:clang_${naive_arch}_openwrt\""
+# ldso_path=\"${ldso_path}\""
 [ "${target_arch}" == "arm" ] && {
 	naive_flags="${naive_flags} arm_version=0 arm_cpu=\"${cpu_type}\""
 	[ -n "${cpu_subtype}" ] && { echo "${cpu_subtype}" | grep -q "neon" && neon_flag="arm_use_neon=true" || neon_flag="arm_use_neon=false"; naive_flags="${naive_flags} arm_fpu=\"${cpu_subtype}\" arm_float_abi=\"hard\" ${neon_flag}"; } || naive_flags="${naive_flags} arm_float_abi=\"soft\" arm_use_neon=false"
diff --git a/package/libs/mbedtls/Makefile b/package/libs/mbedtls/Makefile
index 4fcef3dd9c..69632a77d0 100644
--- a/package/libs/mbedtls/Makefile
+++ b/package/libs/mbedtls/Makefile
@@ -23,6 +23,7 @@ PKG_CPE_ID:=cpe:/a:arm:mbed_tls
 
 PKG_CONFIG_DEPENDS := \
 	CONFIG_LIBMBEDTLS_DEBUG_C \
+	CONFIG_LIBMBEDTLS_HAVE_ARMV8CE_AES \
 	CONFIG_LIBMBEDTLS_HAVE_SSE2 \
 	CONFIG_LIBMBEDTLS_HKDF_C
 
@@ -61,6 +62,20 @@ config LIBMBEDTLS_DEBUG_C
 
 	 Usually, you don't need this, so don't select this if you're unsure.
 
+config LIBMBEDTLS_HAVE_ARMV8CE_AES
+	depends on PACKAGE_libmbedtls
+	bool
+	default y
+	prompt "Enable use of the ARMv8 Crypto Extensions"
+	depends on aarch64 && !TARGET_bcm27xx
+	help
+	 Use of the ARMv8 Crypto Extensions greatly increase performance
+	 (up to 4x faster on AES-GCM while 10x faster on raw AES).
+
+	 Related instructions should be included in all modern Aarch64
+	 devices, except some wastes like Broadcom.
+	 If you don't sure, say Y here.
+
 config LIBMBEDTLS_HAVE_SSE2
 	depends on PACKAGE_libmbedtls
 	bool
@@ -68,12 +83,12 @@ config LIBMBEDTLS_HAVE_SSE2
 	prompt "Enable use of x86 SSE2 instructions"
 	depends on x86_64 || i386
 	help
-		Use of SSE2 instructions greatly increase performance (up to
-		3x faster) with a minimum (~0.2%, or 23KB) increase in package
-		size, but it will bring no benefit if your hardware does not
-		support them, such as Geode GX and LX.  In this case you may
-		save 23KB by saying yes here.  AMD Geode NX, and Intel
-		Pentium 4 and above support SSE2.
+	 Use of SSE2 instructions greatly increase performance (up to
+	 3x faster) with a minimum (~0.2%, or 23KB) increase in package
+	 size, but it will bring no benefit if your hardware does not
+	 support them, such as Geode GX and LX.  In this case you may
+	 save 23KB by saying yes here.  AMD Geode NX, and Intel
+	 Pentium 4 and above support SSE2.
 
 config LIBMBEDTLS_HKDF_C
 	depends on PACKAGE_libmbedtls
@@ -118,6 +133,7 @@ define Build/Configure
 
 	awk 'BEGIN { rc = 1 } \
 	     /#define MBEDTLS_DEBUG_C/ { $$$$0 = "$(if $(CONFIG_LIBMBEDTLS_DEBUG_C),,// )#define MBEDTLS_DEBUG_C"; rc = 0 } \
+	     /#define MBEDTLS_ARMV8CE_AES_C/ { $$$$0 = "$(if $(CONFIG_LIBMBEDTLS_HAVE_ARMV8CE_AES),,// )#define MBEDTLS_ARMV8CE_AES_C"; rc = 0 } \
 	     /#define MBEDTLS_HAVE_SSE2/ { $$$$0 = "$(if $(CONFIG_LIBMBEDTLS_HAVE_SSE2),,// )#define MBEDTLS_HAVE_SSE2"; rc = 0 } \
 	     { print } \
 	     END { exit(rc) }' $(PKG_BUILD_DIR)/include/mbedtls/config.h \
diff --git a/package/libs/mbedtls/patches/100-Implements-AES-and-GCM-with-ARMv8-Crypto-Extensions.patch b/package/libs/mbedtls/patches/100-Implements-AES-and-GCM-with-ARMv8-Crypto-Extensions.patch
new file mode 100644
index 0000000000..b6531181f0
--- /dev/null
+++ b/package/libs/mbedtls/patches/100-Implements-AES-and-GCM-with-ARMv8-Crypto-Extensions.patch
@@ -0,0 +1,401 @@
+From dfb6015ca79a9fee28f7fcb0af7e350a83574b83 Mon Sep 17 00:00:00 2001
+From: "Markku-Juhani O. Saarinen" <mjos@mjos.fi>
+Date: Mon, 20 Nov 2017 14:58:41 +0000
+Subject: Implements AES and GCM with ARMv8 Crypto Extensions
+
+A compact patch that provides AES and GCM implementations that utilize the
+ARMv8 Crypto Extensions. The config flag is MBEDTLS_ARMV8CE_AES_C, which
+is disabled by default as we don't do runtime checking for the feature.
+The new implementation lives in armv8ce_aes.c.
+
+Provides similar functionality to https://github.com/ARMmbed/mbedtls/pull/432
+Thanks to Barry O'Rourke and others for that contribtion.
+
+Tested on a Cortex A53 device and QEMU. On a midrange phone the real AES-GCM
+throughput increases about 4x, while raw AES speed is up to 10x faster.
+
+When cross-compiling, you want to set something like:
+
+  export CC='aarch64-linux-gnu-gcc'
+  export CFLAGS='-Ofast -march=armv8-a+crypto'
+  scripts/config.pl set MBEDTLS_ARMV8CE_AES_C
+
+QEMU seems to also need
+
+  export LDFLAGS='-static'
+
+Then run normal make or cmake etc.
+---
+
+diff -ruNa --binary a/ChangeLog.d/armv8_crypto_extensions.txt b/ChangeLog.d/armv8_crypto_extensions.txt
+--- a/ChangeLog.d/armv8_crypto_extensions.txt	1970-01-01 08:00:00.000000000 +0800
++++ b/ChangeLog.d/armv8_crypto_extensions.txt	2021-03-07 15:07:17.781911791 +0800
+@@ -0,0 +1,2 @@
++Features
++    * Support ARMv8 Cryptography Extensions for AES and GCM.
+diff -ruNa --binary a/include/mbedtls/armv8ce_aes.h b/include/mbedtls/armv8ce_aes.h
+--- a/include/mbedtls/armv8ce_aes.h	1970-01-01 08:00:00.000000000 +0800
++++ b/include/mbedtls/armv8ce_aes.h	2021-03-07 15:07:17.781911791 +0800
+@@ -0,0 +1,63 @@
++/**
++ * \file armv8ce_aes.h
++ *
++ * \brief ARMv8 Cryptography Extensions -- Optimized code for AES and GCM
++ */
++
++/*
++ *
++ *  Copyright (C) 2006-2017, ARM Limited, All Rights Reserved
++ *  SPDX-License-Identifier: Apache-2.0
++ *
++ *  Licensed under the Apache License, Version 2.0 (the "License"); you may
++ *  not use this file except in compliance with the License.
++ *  You may obtain a copy of the License at
++ *
++ *  http://www.apache.org/licenses/LICENSE-2.0
++ *
++ *  Unless required by applicable law or agreed to in writing, software
++ *  distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
++ *  WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
++ *  See the License for the specific language governing permissions and
++ *  limitations under the License.
++ *
++ *  This file is part of mbed TLS (https://tls.mbed.org)
++ */
++
++#ifndef MBEDTLS_ARMV8CE_AES_H
++#define MBEDTLS_ARMV8CE_AES_H
++
++#include "aes.h"
++
++/**
++ * \brief          [ARMv8 Crypto Extensions] AES-ECB block en(de)cryption
++ *
++ * \param ctx      AES context
++ * \param mode     MBEDTLS_AES_ENCRYPT or MBEDTLS_AES_DECRYPT
++ * \param input    16-byte input block
++ * \param output   16-byte output block
++ *
++ * \return         0 on success (cannot fail)
++ */
++
++int mbedtls_armv8ce_aes_crypt_ecb( mbedtls_aes_context *ctx,
++                                   int mode,
++                                   const unsigned char input[16],
++                                   unsigned char output[16] );
++
++/**
++ * \brief          [ARMv8 Crypto Extensions]  Multiply in GF(2^128) for GCM
++ *
++ * \param c        Result
++ * \param a        First operand
++ * \param b        Second operand
++ *
++ * \note           Both operands and result are bit strings interpreted as
++ *                 elements of GF(2^128) as per the GCM spec.
++ */
++
++void mbedtls_armv8ce_gcm_mult( unsigned char c[16],
++                               const unsigned char a[16],
++                               const unsigned char b[16] );
++
++#endif /* MBEDTLS_ARMV8CE_AES_H */
+diff -ruNa --binary a/include/mbedtls/check_config.h b/include/mbedtls/check_config.h
+--- a/include/mbedtls/check_config.h	2020-12-10 20:54:15.000000000 +0800
++++ b/include/mbedtls/check_config.h	2021-03-07 15:06:45.625543309 +0800
+@@ -95,6 +95,10 @@
+ #error "MBEDTLS_AESNI_C defined, but not all prerequisites"
+ #endif
+ 
++#if defined(MBEDTLS_ARMV8CE_AES_C) && !defined(MBEDTLS_HAVE_ASM)
++#error "MBEDTLS_ARMV8CE_AES_C defined, but not all prerequisites"
++#endif
++
+ #if defined(MBEDTLS_CTR_DRBG_C) && !defined(MBEDTLS_AES_C)
+ #error "MBEDTLS_CTR_DRBG_C defined, but not all prerequisites"
+ #endif
+@@ -772,3 +776,4 @@
+ typedef int mbedtls_iso_c_forbids_empty_translation_units;
+ 
+ #endif /* MBEDTLS_CHECK_CONFIG_H */
++
+diff -ruNa --binary a/include/mbedtls/config.h b/include/mbedtls/config.h
+--- a/include/mbedtls/config.h	2020-12-10 20:54:15.000000000 +0800
++++ b/include/mbedtls/config.h	2021-03-07 15:14:27.957855484 +0800
+@@ -73,6 +73,7 @@
+  * Requires support for asm() in compiler.
+  *
+  * Used in:
++ *      library/armv8ce_aes.c
+  *      library/aria.c
+  *      library/timing.c
+  *      include/mbedtls/bn_mul.h
+@@ -1888,6 +1889,21 @@
+ #define MBEDTLS_AESNI_C
+ 
+ /**
++ * \def MBEDTLS_ARMV8CE_AES_C
++ *
++ * Enable ARMv8 Crypto Extensions for AES and GCM
++ *
++ * Module:  library/armv8ce_aes.c
++ * Caller:  library/aes.c
++ *          library/gcm.c
++ *
++ * Requires: MBEDTLS_HAVE_ASM
++ *
++ * This module adds support for Armv8 Cryptography Extensions for AES and GCM.
++ */
++//#define MBEDTLS_ARMV8CE_AES_C
++
++/**
+  * \def MBEDTLS_AES_C
+  *
+  * Enable the AES block cipher.
+diff -ruNa --binary a/library/aes.c b/library/aes.c
+--- a/library/aes.c	2020-12-10 20:54:15.000000000 +0800
++++ b/library/aes.c	2021-03-07 15:06:45.625543309 +0800
+@@ -69,7 +69,9 @@
+ #if defined(MBEDTLS_AESNI_C)
+ #include "mbedtls/aesni.h"
+ #endif
+-
++#if defined(MBEDTLS_ARMV8CE_AES_C)
++#include "mbedtls/armv8ce_aes.h"
++#endif
+ #if defined(MBEDTLS_SELF_TEST)
+ #if defined(MBEDTLS_PLATFORM_C)
+ #include "mbedtls/platform.h"
+@@ -1052,6 +1054,11 @@
+         return( mbedtls_aesni_crypt_ecb( ctx, mode, input, output ) );
+ #endif
+ 
++#if defined(MBEDTLS_ARMV8CE_AES_C)
++	// We don't do runtime checking for ARMv8 Crypto Extensions
++	return mbedtls_armv8ce_aes_crypt_ecb( ctx, mode, input, output );
++#endif
++
+ #if defined(MBEDTLS_PADLOCK_C) && defined(MBEDTLS_HAVE_X86)
+     if( aes_padlock_ace )
+     {
+diff -ruNa --binary a/library/armv8ce_aes.c b/library/armv8ce_aes.c
+--- a/library/armv8ce_aes.c	1970-01-01 08:00:00.000000000 +0800
++++ b/library/armv8ce_aes.c	2021-03-07 15:07:17.781911791 +0800
+@@ -0,0 +1,142 @@
++/*
++ *  ARMv8 Cryptography Extensions -- Optimized code for AES and GCM
++ *
++ *  Copyright (C) 2006-2017, ARM Limited, All Rights Reserved
++ *  SPDX-License-Identifier: Apache-2.0
++ *
++ *  Licensed under the Apache License, Version 2.0 (the "License"); you may
++ *  not use this file except in compliance with the License.
++ *  You may obtain a copy of the License at
++ *
++ *  http://www.apache.org/licenses/LICENSE-2.0
++ *
++ *  Unless required by applicable law or agreed to in writing, software
++ *  distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
++ *  WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
++ *  See the License for the specific language governing permissions and
++ *  limitations under the License.
++ *
++ *  This file is part of mbed TLS (https://tls.mbed.org)
++ */
++
++#if !defined(MBEDTLS_CONFIG_FILE)
++#include "mbedtls/config.h"
++#else
++#include MBEDTLS_CONFIG_FILE
++#endif
++
++#if defined(MBEDTLS_ARMV8CE_AES_C)
++
++#include <arm_neon.h>
++#include "mbedtls/armv8ce_aes.h"
++
++#ifndef asm
++#define asm __asm
++#endif
++
++/*
++ *  [Armv8 Cryptography Extensions]  AES-ECB block en(de)cryption
++ */
++
++#if defined(MBEDTLS_AES_C)
++
++int mbedtls_armv8ce_aes_crypt_ecb( mbedtls_aes_context *ctx,
++                                   int mode,
++                                   const unsigned char input[16],
++                                   unsigned char output[16] )
++{
++    unsigned int i;
++    const uint8_t *rk;
++    uint8x16_t x, k;
++
++    x = vld1q_u8( input );                          /* input block */
++    rk = (const uint8_t *) ctx->rk;                 /* round keys  */
++
++    if( mode == MBEDTLS_AES_ENCRYPT )
++    {
++        for( i = ctx->nr - 1; i != 0; i-- )         /* encryption loop */
++        {
++            k = vld1q_u8( rk );
++            rk += 16;
++            x = vaeseq_u8( x, k );
++            x = vaesmcq_u8( x );
++        }
++        k = vld1q_u8( rk );
++        rk += 16;
++        x = vaeseq_u8( x, k );
++    }
++    else
++    {
++        for( i = ctx->nr - 1; i != 0 ; i-- )         /* decryption loop */
++        {
++            k = vld1q_u8( rk );
++            rk += 16;
++            x = vaesdq_u8( x, k );
++            x = vaesimcq_u8( x );
++        }
++        k = vld1q_u8( rk );
++        rk += 16;
++        x = vaesdq_u8( x, k );
++    }
++
++    k = vld1q_u8( rk );                             /* final key just XORed */
++    x = veorq_u8( x, k );
++    vst1q_u8( output, x );                          /* write out */
++
++    return ( 0 );
++}
++
++#endif /* MBEDTLS_AES_C */
++
++
++/*
++ *  [Armv8 Cryptography Extensions]  Multiply in GF(2^128) for GCM
++ */
++
++#if defined(MBEDTLS_GCM_C)
++
++void mbedtls_armv8ce_gcm_mult( unsigned char c[16],
++                               const unsigned char a[16],
++                               const unsigned char b[16] )
++{
++    /* GCM's GF(2^128) polynomial basis is x^128 + x^7 + x^2 + x + 1 */
++    const uint64x2_t base = { 0, 0x86 };            /* note missing LS bit */
++
++    register uint8x16_t vc asm( "v0" );             /* named registers */
++    register uint8x16_t va asm( "v1" );             /* (to avoid conflict) */
++    register uint8x16_t vb asm( "v2" );
++    register uint64x2_t vp asm( "v3" );
++
++    va = vld1q_u8( a );                             /* load inputs */
++    vb = vld1q_u8( b );
++    vp = base;
++
++    asm (
++        "rbit    %1.16b, %1.16b             \n\t"   /* reverse bit order */
++        "rbit    %2.16b, %2.16b             \n\t"
++        "pmull2  %0.1q,  %1.2d,  %2.2d      \n\t"   /* v0 = a.hi * b.hi */
++        "pmull2  v4.1q,  %0.2d,  %3.2d      \n\t"   /* mul v0 by x^64, reduce */
++        "ext     %0.16b, %0.16b, %0.16b, #8 \n\t"
++        "eor     %0.16b, %0.16b, v4.16b     \n\t"
++        "ext     v5.16b, %2.16b, %2.16b, #8 \n\t"   /* (swap hi and lo in b) */
++        "pmull   v4.1q,  %1.1d,  v5.1d      \n\t"   /* v0 ^= a.lo * b.hi */
++        "eor     %0.16b, %0.16b, v4.16b     \n\t"
++        "pmull2  v4.1q,  %1.2d,  v5.2d      \n\t"   /* v0 ^= a.hi * b.lo */
++        "eor     %0.16b, %0.16b, v4.16b     \n\t"
++        "pmull2  v4.1q,  %0.2d,  %3.2d      \n\t"   /* mul v0 by x^64, reduce */
++        "ext     %0.16b, %0.16b, %0.16b, #8 \n\t"
++        "eor     %0.16b, %0.16b, v4.16b     \n\t"
++        "pmull   v4.1q,  %1.1d,  %2.1d      \n\t"   /* v0 ^= a.lo * b.lo */
++        "eor     %0.16b, %0.16b, v4.16b     \n\t"
++        "rbit    %0.16b, %0.16b             \n\t"   /* reverse bits for output */
++        : "=w" (vc)                                 /* q0:      output */
++        : "w" (va), "w" (vb), "w" (vp)              /* q1, q2:  input */
++        : "v4", "v5"                                /* q4, q5:  clobbered */
++    );
++
++    vst1q_u8( c, vc );                              /* write out */
++}
++
++#endif /* MBEDTLS_GCM_C */
++
++#endif /* MBEDTLS_ARMV8CE_AES_C */
+diff -ruNa --binary a/library/CMakeLists.txt b/library/CMakeLists.txt
+--- a/library/CMakeLists.txt	2020-12-10 20:54:15.000000000 +0800
++++ b/library/CMakeLists.txt	2021-03-07 15:06:45.625543309 +0800
+@@ -7,6 +7,7 @@
+     aesni.c
+     arc4.c
+     aria.c
++    armv8ce_aes.c
+     asn1parse.c
+     asn1write.c
+     base64.c
+diff -ruNa --binary a/library/gcm.c b/library/gcm.c
+--- a/library/gcm.c	2020-12-10 20:54:15.000000000 +0800
++++ b/library/gcm.c	2021-03-07 15:06:45.625543309 +0800
+@@ -71,6 +71,10 @@
+ #include "mbedtls/aesni.h"
+ #endif
+ 
++#if defined(MBEDTLS_ARMV8CE_AES_C)
++#include "mbedtls/armv8ce_aes.h"
++#endif
++
+ #if defined(MBEDTLS_SELF_TEST) && defined(MBEDTLS_AES_C)
+ #include "mbedtls/aes.h"
+ #include "mbedtls/platform.h"
+@@ -140,6 +144,12 @@
+     if( ( ret = mbedtls_cipher_update( &ctx->cipher_ctx, h, 16, h, &olen ) ) != 0 )
+         return( ret );
+ 
++#if defined(MBEDTLS_ARMV8CE_AES_C)
++	// we don't do feature testing with ARMv8 cryptography extensions
++    memcpy( ctx ->HL, h, 16 );          // put H at the beginning of buffer
++    return( 0 );                        // that's all we need
++#endif
++
+     /* pack h as two 64-bits ints, big-endian */
+     GET_UINT32_BE( hi, h,  0  );
+     GET_UINT32_BE( lo, h,  4  );
+@@ -248,6 +258,11 @@
+     unsigned char lo, hi, rem;
+     uint64_t zh, zl;
+ 
++#if defined(MBEDTLS_ARMV8CE_AES_C)
++	mbedtls_armv8ce_gcm_mult( output, x, (const unsigned char *) ctx->HL );
++	return;
++#endif
++
+ #if defined(MBEDTLS_AESNI_C) && defined(MBEDTLS_HAVE_X86_64)
+     if( mbedtls_aesni_has_support( MBEDTLS_AESNI_CLMUL ) ) {
+         unsigned char h[16];
+diff -ruNa --binary a/library/Makefile b/library/Makefile
+--- a/library/Makefile	2020-12-10 20:54:15.000000000 +0800
++++ b/library/Makefile	2021-03-07 15:12:49.277078224 +0800
+@@ -65,6 +65,7 @@
+ 
+ OBJS_CRYPTO=	aes.o		aesni.o		arc4.o		\
+ 		aria.o		asn1parse.o	asn1write.o	\
++		armv8ce_aes.o					\
+ 		base64.o	bignum.o	blowfish.o	\
+ 		camellia.o	ccm.o		chacha20.o	\
+ 		chachapoly.o	cipher.o	cipher_wrap.o	\
+diff -ruNa --binary a/library/version_features.c b/library/version_features.c
+--- a/library/version_features.c	2020-12-10 20:54:15.000000000 +0800
++++ b/library/version_features.c	2021-03-07 15:06:45.625543309 +0800
+@@ -583,6 +583,9 @@
+ #if defined(MBEDTLS_AESNI_C)
+     "MBEDTLS_AESNI_C",
+ #endif /* MBEDTLS_AESNI_C */
++#if defined(MBEDTLS_ARMV8CE_AES_C)
++    "MBEDTLS_ARMV8CE_AES_C",
++#endif /* MBEDTLS_ARMV8CE_AES_C */
+ #if defined(MBEDTLS_AES_C)
+     "MBEDTLS_AES_C",
+ #endif /* MBEDTLS_AES_C */
+