Merge pull request wolfSSL#10756 from SparkiDev/aes_asm_ymm_zmm

Intel x64 ASM: Add new assembly for AES

Author: dgarske

.github/workflows/symbol-prefixes.yml

@@ -49,7 +49,7 @@ jobs:
           {
               if (($7 !~ /^[0-9]+$/) ||
                   ($8 ~ /^(wc_|wolf|WOLF|__pfx|fe_|sp_[a-zA-Z090-0_]*[0-9])/) ||
-                  ($8 ~ /(_avx[12]|_AVX[12]|_sse[12]|_SSE[12]|_aesni|_AESNI|_bmi2|_x64$)/))
+                  ($8 ~ /(_avx[12]|_AVX[12]|_sse[12]|_SSE[12]|_aesni|_AESNI|_vaes|_VAES|_avx512|_AVX512|_bmi2|_x64$)/))
               {
                   next;
               }

.wolfssl_known_macro_extras

@@ -381,6 +381,7 @@ NO_AES_DECRYPT
 NO_ARDUINO_DEFAULT
 NO_ASM
 NO_ASN_OLD_TYPE_NAMES
+NO_AVX512_SUPPORT
 NO_CAMELLIA_CBC
 NO_CERT
 NO_CERT_IN_TICKET
@@ -459,6 +460,7 @@ NO_STDIO_FGETS_REMAP
 NO_STM32_HMAC
 NO_TKERNEL_MEM_POOL
 NO_TLSX_PSKKEM_PLAIN_ANNOUNCE
+NO_VAES_SUPPORT
 NO_VERIFY_OID
 NO_WC_DHGENERATEPUBLIC
 NO_WC_SHE_GETUID

linuxkm/Kbuild

@@ -200,6 +200,8 @@ $(obj)/wolfcrypt/src/aes_gcm_asm.o: asflags-y := $(WOLFSSL_ASFLAGS) $(ASFLAGS_FP
 $(obj)/wolfcrypt/src/aes_gcm_asm.o: OBJECT_FILES_NON_STANDARD := y
 $(obj)/wolfcrypt/src/aes_xts_asm.o: asflags-y := $(WOLFSSL_ASFLAGS) $(ASFLAGS_FPU_DISABLE_SIMD_ENABLE)
 $(obj)/wolfcrypt/src/aes_xts_asm.o: OBJECT_FILES_NON_STANDARD := y
+$(obj)/wolfcrypt/src/aes_x86_64_asm.o: asflags-y := $(WOLFSSL_ASFLAGS) $(ASFLAGS_FPU_DISABLE_SIMD_ENABLE)
+$(obj)/wolfcrypt/src/aes_x86_64_asm.o: OBJECT_FILES_NON_STANDARD := y
 $(obj)/wolfcrypt/src/sp_x86_64_asm.o: asflags-y := $(WOLFSSL_ASFLAGS) $(ASFLAGS_FPU_DISABLE_SIMD_ENABLE)
 $(obj)/wolfcrypt/src/sp_x86_64_asm.o: OBJECT_FILES_NON_STANDARD := y
 $(obj)/wolfcrypt/src/sha256_asm.o: asflags-y := $(WOLFSSL_ASFLAGS) $(ASFLAGS_FPU_DISABLE_SIMD_ENABLE)

src/include.am

@@ -109,6 +109,7 @@ endif
 
 if BUILD_AESNI
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/aes_asm.S
+src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/aes_x86_64_asm.S
 if BUILD_X86_ASM
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/aes_gcm_x86_asm.S
 else
@@ -259,6 +260,7 @@ endif BUILD_PPC64_ASM
 
 if BUILD_AESNI
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/aes_asm.S
+src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/aes_x86_64_asm.S
 if BUILD_X86_ASM
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/aes_gcm_x86_asm.S
 else
@@ -532,6 +534,7 @@ endif BUILD_PPC64_ASM
 
 if BUILD_AESNI
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/aes_asm.S
+src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/aes_x86_64_asm.S
 if BUILD_X86_ASM
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/aes_gcm_x86_asm.S
 else
@@ -867,6 +870,7 @@ endif BUILD_AES
 
 if BUILD_AESNI
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/aes_asm.S
+src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/aes_x86_64_asm.S
 if BUILD_X86_ASM
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/aes_gcm_x86_asm.S
 else
@@ -1708,6 +1712,7 @@ endif
 if !BUILD_FIPS_V2_PLUS
 if BUILD_AESNI
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/aes_asm.S
+src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/aes_x86_64_asm.S
 if BUILD_X86_ASM
 src_libwolfssl@LIBSUFFIX@_la_SOURCES += wolfcrypt/src/aes_gcm_x86_asm.S
 else

wolfcrypt/benchmark/benchmark.c

@@ -4794,6 +4794,8 @@ static void print_cpu_features(void)
     if (IS_INTEL_MOVBE(cpuid_flags))  printf(" movbe");
     if (IS_INTEL_BMI1(cpuid_flags))   printf(" bmi1");
     if (IS_INTEL_SHA(cpuid_flags))    printf(" sha");
+    if (IS_INTEL_VAES(cpuid_flags))   printf(" vaes");
+    if (IS_INTEL_AVX512(cpuid_flags)) printf(" avx512");
 #endif
 #ifdef __aarch64__
     printf("Aarch64 -");