Full arm cpu feature detection.

CMakeLists.txt

@@ -155,13 +155,17 @@ if (USE_CPU_EXTENSIONS)
                 )
         endif()
     elseif (AWS_ARCH_ARM64 OR AWS_ARCH_ARM32)
-        if (MSVC)
+        if (WINDOWS)
             file(GLOB AWS_COMMON_ARCH_SRC
-                "source/arch/arm/msvc/*.c"
+                "source/arch/arm/windows/*.c"
                 )
-        elseif (AWS_HAVE_AUXV)
+        elseif(APPLE)
+            file(GLOB AWS_COMMON_ARCH_SRC
+                    "source/arch/arm/darwin/*.c"
+                    )
+        else()
             file(GLOB AWS_COMMON_ARCH_SRC
-                "source/arch/arm/asm/*.c"
+                "source/arch/arm/auxv/*.c"
                 )
         endif()
     endif()

bin/system_info/print_system_info.c

@@ -3,6 +3,7 @@
 #include <aws/common/byte_buf.h>
 #include <aws/common/logging.h>
 #include <aws/common/system_info.h>
+#include <aws/common/cpuid.h>
 
 int main(void) {
     struct aws_allocator *allocator = aws_default_allocator();
@@ -39,6 +40,19 @@ int main(void) {
         fprintf(stdout, "  'numa architecture': 'false'\n");
     }
 
+    fprintf(stdout, "   'cpu_capabilities': {\n");
+    fprintf(stdout, "       'arm_crc': %s,\n", aws_cpu_has_feature(AWS_CPU_FEATURE_ARM_CRC) ? "true" : "false");
+    fprintf(stdout, "       'arm_pmull': %s,\n", aws_cpu_has_feature(AWS_CPU_FEATURE_ARM_PMULL) ? "true" : "false");
+    fprintf(stdout, "       'arm_crypto': %s,\n", aws_cpu_has_feature(AWS_CPU_FEATURE_ARM_CRYPTO) ? "true" : "false");
+    fprintf(stdout, "       'amd_sse4_1': %s,\n", aws_cpu_has_feature(AWS_CPU_FEATURE_SSE_4_1) ? "true" : "false");
+    fprintf(stdout, "       'amd_sse4_2': %s,\n", aws_cpu_has_feature(AWS_CPU_FEATURE_SSE_4_2) ? "true" : "false");
+    fprintf(stdout, "       'amd_clmul': %s,\n", aws_cpu_has_feature(AWS_CPU_FEATURE_CLMUL) ? "true" : "false");
+    fprintf(stdout, "       'amd_vpclmulqdq': %s,\n", aws_cpu_has_feature(AWS_CPU_FEATURE_VPCLMULQDQ) ? "true" : "false");
+    fprintf(stdout, "       'amd_avx2': %s,\n", aws_cpu_has_feature(AWS_CPU_FEATURE_AVX2) ? "true" : "false");
+    fprintf(stdout, "       'amd_avx512': %s,\n", aws_cpu_has_feature(AWS_CPU_FEATURE_AVX512) ? "true" : "false");
+    fprintf(stdout, "       'amd_bmi2': %s\n", aws_cpu_has_feature(AWS_CPU_FEATURE_BMI2) ? "true" : "false");
+    fprintf(stdout, "   }\n");
+
     fprintf(stdout, "}\n");
     aws_system_environment_release(env);
     aws_logger_clean_up(&logger);

include/aws/common/cpuid.h

@@ -18,6 +18,8 @@ enum aws_cpu_feature_name {
     AWS_CPU_FEATURE_ARM_CRC,
     AWS_CPU_FEATURE_BMI2,
     AWS_CPU_FEATURE_VPCLMULQDQ,
+    AWS_CPU_FEATURE_ARM_PMULL,
+    AWS_CPU_FEATURE_ARM_CRYPTO,
     AWS_CPU_FEATURE_COUNT,
 };
 

source/arch/arm/asm/cpuid.c → source/arch/arm/auxv/cpuid.c

@@ -29,7 +29,9 @@ struct cap_bits {
 
 #    if (defined(__aarch64__))
 struct cap_bits s_check_cap[AWS_CPU_FEATURE_COUNT] = {
-    [AWS_CPU_FEATURE_ARM_CRC] = {0, 1 << 7 /* HWCAP_CRC */},
+    [AWS_CPU_FEATURE_ARM_CRC] = {0, 1 << 7 /* HWCAP_CRC32 */},
+    [AWS_CPU_FEATURE_ARM_PMULL] = {0, 1 << 4 /* HWCAP_PMULL */},
+    [AWS_CPU_FEATURE_ARM_CRYPTO] = {0, 1 << 3 /* HWCAP_AES */},
 };
 #    else
 struct cap_bits s_check_cap[AWS_CPU_FEATURE_COUNT] = {

source/arch/arm/darwin/cpuid.c

@@ -0,0 +1,40 @@
+/*
+ * Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License").
+ * You may not use this file except in compliance with the License.
+ * A copy of the License is located at
+ *
+ *  http://aws.amazon.com/apache2.0
+ *
+ * or in the "license" file accompanying this file. This file 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.
+ */
+
+#include <aws/common/cpuid.h>
+
+#include <sys/sysctl.h>
+
+bool aws_cpu_has_feature(enum aws_cpu_feature_name feature_name) {
+    int64_t ret = 0;
+    size_t size = sizeof(ret);
+
+    switch (feature_name) {
+        case AWS_CPU_FEATURE_ARM_PMULL:
+            if (sysctlbyname("hw.optional.arm.FEAT_PMULL", &ret, &size, NULL, 0) != -1) {
+                return ret == 1;
+            }
+        case AWS_CPU_FEATURE_ARM_CRC:
+            if (sysctlbyname("hw.optional.armv8_crc32", &ret, &size, NULL, 0) != -1) {
+                return ret == 1;
+            }
+        case AWS_CPU_FEATURE_ARM_CRYPTO:
+            if (sysctlbyname("hw.optional.arm.FEAT_AES", &ret, &size, NULL, 0) != -1) {
+                return ret == 1;
+            }
+        default:
+            return false;
+    }
+}

source/arch/arm/msvc/cpuid.c → source/arch/arm/windows/cpuid.c

@@ -13,9 +13,18 @@
  * permissions and limitations under the License.
  */
 
+#include <Windows.h>
 #include <aws/common/cpuid.h>
-#include <stdlib.h>
 
 bool aws_cpu_has_feature(enum aws_cpu_feature_name feature_name) {
-    return false;
+    switch (feature_name) {
+        case AWS_CPU_FEATURE_ARM_CRC:
+            return IsProcessorFeaturePresent(PF_ARM_V8_CRC32_INSTRUCTIONS_AVAILABLE) != 0;
+        // this is the best we've got on windows as they don't separate PMULL and AES from each other.
+        case AWS_CPU_FEATURE_ARM_PMULL:
+        case AWS_CPU_FEATURE_ARM_CRYPTO:
+            return IsProcessorFeaturePresent(PF_ARM_V8_CRYPTO_INSTRUCTIONS_AVAILABLE) != 0;
+        default:
+            return false;
+    }
 }