Merge pull request #6731 from SparkiDev/aes_arm32_thumb2

AES ARM32 and Thumb2 ASM: fixup ARM32 and add Thumb2

Author: JacobBarthelmeh

configure.ac

@@ -2275,7 +2275,7 @@ AC_ARG_ENABLE([aescbc],
 if test "$ENABLED_AESCBC" = "no"
 then
     AM_CFLAGS="$AM_CFLAGS -DNO_AES_CBC"
-    AM_CCASFLAGS="$AM_CCASFLAGS -DHAVE_AES_CBC"
+    AM_CCASFLAGS="$AM_CCASFLAGS -DNO_AES_CBC"
 fi
 
 # AES-CBC length checks (checks that input lengths are multiples of block size)
@@ -2589,7 +2589,7 @@ then
             AC_MSG_NOTICE([64bit ARMv8 found, setting mcpu to generic+crypto])
             ;;
         armv7a*)
-            AM_CPPFLAGS="$AM_CPPFLAGS -march=armv7-a -mfpu=neon -DWOLFSSL_ARM_ARCH=7"
+            AM_CPPFLAGS="$AM_CPPFLAGS -march=armv7-a -mfpu=neon -DWOLFSSL_ARM_ARCH=7 -marm"
             # Include options.h
             AM_CCASFLAGS="$AM_CCASFLAGS -DEXTERNAL_OPTS_OPENVPN"
             ENABLED_ARMASM_CRYPTO=no

src/include.am

@@ -157,16 +157,26 @@ endif
 
 if BUILD_AES
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/aes.c
-if BUILD_ARMASM_NEON
+if BUILD_ARMASM
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/port/arm/armv8-aes.c
+endif BUILD_ARMASM
+if BUILD_ARMASM_NEON
 if !BUILD_ARMASM_CRYPTO
 if BUILD_ARMASM_INLINE
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/port/arm/armv8-32-aes-asm_c.c
 else
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/port/arm/armv8-32-aes-asm.S
 endif !BUILD_ARMASM_INLINE
 endif !BUILD_ARMASM_CRYPTO
-endif BUILD_ARMASM_NEON
+else
+if BUILD_ARMASM
+if BUILD_ARMASM_INLINE
+src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/port/arm/thumb2-aes-asm_c.c
+else
+src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/port/arm/thumb2-aes-asm.S
+endif !BUILD_ARMASM_INLINE
+endif BUILD_ARMASM
+endif !BUILD_ARMASM_NEON
 endif BUILD_AES
 
 if BUILD_AESNI
@@ -401,16 +411,28 @@ endif
 if !BUILD_FIPS_CURRENT
 if BUILD_AES
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/aes.c
-if BUILD_ARMASM_NEON
+if BUILD_ARMASM
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/port/arm/armv8-aes.c
+endif BUILD_ARMASM
+if BUILD_ARMASM_NEON
 if !BUILD_ARMASM_CRYPTO
 if BUILD_ARMASM_INLINE
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/port/arm/armv8-32-aes-asm_c.c
+src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/port/arm/thumb2-aes-asm_c.c
 else
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/port/arm/armv8-32-aes-asm.S
+src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/port/arm/thumb2-aes-asm.S
 endif !BUILD_ARMASM_INLINE
 endif !BUILD_ARMASM_CRYPTO
-endif BUILD_ARMASM_NEON
+else
+if BUILD_ARMASM
+if BUILD_ARMASM_INLINE
+src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/port/arm/thumb2-aes-asm_c.c
+else
+src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/port/arm/thumb2-aes-asm.S
+endif !BUILD_ARMASM_INLINE
+endif BUILD_ARMASM
+endif !BUILD_ARMASM_NEON
 if BUILD_AFALG
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/port/af_alg/afalg_aes.c
 endif BUILD_AFALG

wolfcrypt/src/aes.c

@@ -97,7 +97,7 @@ block cipher mechanism that uses n-bit binary string parameter key with 128-bits
     #include <wolfcrypt/src/misc.c>
 #endif
 
-#if !defined(WOLFSSL_ARMASM) || defined(WOLFSSL_ARMASM_NO_NEON)
+#ifndef WOLFSSL_ARMASM
 
 #ifdef WOLFSSL_IMX6_CAAM_BLOB
     /* case of possibly not using hardware acceleration for AES but using key
@@ -4573,7 +4573,7 @@ int wc_AesSetIV(Aes* aes, const byte* iv)
     #endif /* NEED_AES_CTR_SOFT */
 
 #endif /* WOLFSSL_AES_COUNTER */
-#endif /* !WOLFSSL_ARMASM || WOLFSSL_ARMASM_NO_NEON */
+#endif /* !WOLFSSL_ARMASM */
 
 
 /*
@@ -4620,7 +4620,7 @@ static WC_INLINE void IncCtr(byte* ctr, word32 ctrSz)
 
 #endif
 
-#if defined(WOLFSSL_ARMASM) && !defined(WOLFSSL_ARMASM_NO_NEON)
+#ifdef WOLFSSL_ARMASM
     /* implementation is located in wolfcrypt/src/port/arm/armv8-aes.c */
 
 #elif defined(WOLFSSL_AFALG)
@@ -8851,7 +8851,7 @@ int wc_AesCcmCheckTagSize(int sz)
     return 0;
 }
 
-#if defined(WOLFSSL_ARMASM) && !defined(WOLFSSL_ARMASM_NO_NEON)
+#ifdef WOLFSSL_ARMASM
     /* implementation located in wolfcrypt/src/port/arm/armv8-aes.c */
 
 #elif defined(HAVE_COLDFIRE_SEC)

wolfcrypt/src/curve25519.c

@@ -58,7 +58,13 @@ const curve25519_set_type curve25519_sets[] = {
     }
 };
 
-static const unsigned char kCurve25519BasePoint[CURVE25519_KEYSIZE] = {9};
+static const word32 kCurve25519BasePoint[CURVE25519_KEYSIZE/sizeof(word32)] = {
+#ifdef BIG_ENDIAN_ORDER
+    0x09000000
+#else
+    9
+#endif
+};
 
 /* Curve25519 private key must be less than order */
 /* These functions clamp private k and check it */
@@ -133,7 +139,7 @@ int wc_curve25519_make_pub(int public_size, byte* pub, int private_size,
 
     SAVE_VECTOR_REGISTERS(return _svr_ret;);
 
-    ret = curve25519(pub, priv, kCurve25519BasePoint);
+    ret = curve25519(pub, priv, (byte*)kCurve25519BasePoint);
 
     RESTORE_VECTOR_REGISTERS();
 #endif

wolfcrypt/src/ge_operations.c

@@ -921,10 +921,14 @@ int ge_compress_key(byte* out, const byte* xIn, const byte* yIn, word32 keySz)
 {
     ge_p2  g;
     byte   bArray[ED25519_KEY_SIZE];
+    byte   x[ED25519_KEY_SIZE];
+    byte   y[ED25519_KEY_SIZE];
     word32 i;
 
-    fe_frombytes(g.X, xIn);
-    fe_frombytes(g.Y, yIn);
+    XMEMCPY(x, xIn, ED25519_KEY_SIZE);
+    XMEMCPY(y, yIn, ED25519_KEY_SIZE);
+    fe_frombytes(g.X, x);
+    fe_frombytes(g.Y, y);
     fe_1(g.Z);
 
     ge_tobytes(bArray, &g);