[AVX512] Added load/store from BW/VL subsets to Register2Memory opcode tables.

[oota-llvm.git] / lib / Support / Host.cpp

diff --git a/lib/Support/Host.cpp b/lib/Support/Host.cpp
index d883c05887283c6e29970e56ecb8074558f0abc3..e2dd6d522bb7f19863125128cf9d9b5f3259ee42 100644 (file)
--- a/lib/Support/Host.cpp
+++ b/lib/Support/Host.cpp
@@ -7,7 +7,7 @@
 //
 //===----------------------------------------------------------------------===//
 //
-//  This header file implements the operating system Host concept.
+//  This file implements the operating system Host concept.
 //
 //===----------------------------------------------------------------------===//
 
@@ -39,6 +39,8 @@
 #include <mach/machine.h>
 #endif
 
+#define DEBUG_TYPE "host-detection"
+
 //===----------------------------------------------------------------------===//
 //
 //  Implementations of the CPU detection routines
@@ -221,6 +223,7 @@ StringRef sys::getHostCPUName() {
                  (EBX & 0x20);
   GetX86CpuIDAndInfo(0x80000001, &EAX, &EBX, &ECX, &EDX);
   bool Em64T = (EDX >> 29) & 0x1;
+  bool HasTBM = (ECX >> 21) & 0x1;
 
   if (memcmp(text.c, "GenuineIntel", 12) == 0) {
     switch (Family) {
@@ -433,9 +436,11 @@ StringRef sys::getHostCPUName() {
       case 21:
         if (!HasAVX) // If the OS doesn't support AVX provide a sane fallback.
           return "btver1";
+        if (Model >= 0x50)
+          return "bdver4"; // 50h-6Fh: Excavator
         if (Model >= 0x30)
           return "bdver3"; // 30h-3Fh: Steamroller
-        if (Model >= 0x10)
+        if (Model >= 0x10 || HasTBM)
           return "bdver2"; // 10h-1Fh: Piledriver
         return "bdver1";   // 00h-0Fh: Bulldozer
       case 22:
@@ -449,7 +454,7 @@ StringRef sys::getHostCPUName() {
   return "generic";
 }
 #elif defined(__APPLE__) && (defined(__ppc__) || defined(__powerpc__))
-std::string sys::getHostCPUName() {
+StringRef sys::getHostCPUName() {
   host_basic_info_data_t hostInfo;
   mach_msg_type_number_t infoCount;
 
@@ -478,7 +483,7 @@ std::string sys::getHostCPUName() {
   return "generic";
 }
 #elif defined(__linux__) && (defined(__ppc__) || defined(__powerpc__))
-std::string sys::getHostCPUName() {
+StringRef sys::getHostCPUName() {
   // Access to the Processor Version Register (PVR) on PowerPC is privileged,
   // and so we must use an operating-system interface to determine the current
   // processor type. On Linux, this is exposed through the /proc/cpuinfo file.
@@ -565,10 +570,12 @@ std::string sys::getHostCPUName() {
     .Case("A2", "a2")
     .Case("POWER6", "pwr6")
     .Case("POWER7", "pwr7")
+    .Case("POWER8", "pwr8")
+    .Case("POWER8E", "pwr8")
     .Default(generic);
 }
 #elif defined(__linux__) && defined(__arm__)
-std::string sys::getHostCPUName() {
+StringRef sys::getHostCPUName() {
   // The cpuid register on arm is not accessible from user space. On Linux,
   // it is exposed through the /proc/cpuinfo file.
   // Note: We cannot mmap /proc/cpuinfo here and then process the resulting
@@ -620,10 +627,21 @@ std::string sys::getHostCPUName() {
           .Case("0xc24", "cortex-m4")
           .Default("generic");
 
+  if (Implementer == "0x51") // Qualcomm Technologies, Inc.
+    // Look for the CPU part line.
+    for (unsigned I = 0, E = Lines.size(); I != E; ++I)
+      if (Lines[I].startswith("CPU part"))
+        // The CPU part is a 3 digit hexadecimal number with a 0x prefix. The
+        // values correspond to the "Part number" in the CP15/c0 register. The
+        // contents are specified in the various processor manuals.
+        return StringSwitch<const char *>(Lines[I].substr(8).ltrim("\t :"))
+          .Case("0x06f", "krait") // APQ8064
+          .Default("generic");
+
   return "generic";
 }
 #elif defined(__linux__) && defined(__s390x__)
-std::string sys::getHostCPUName() {
+StringRef sys::getHostCPUName() {
   // STIDP is a privileged operation, so use /proc/cpuinfo instead.
   // Note: We cannot mmap /proc/cpuinfo here and then process the resulting
   // memory buffer because the 'file' has 0 size (it can be read from only
@@ -665,12 +683,12 @@ std::string sys::getHostCPUName() {
   return "generic";
 }
 #else
-std::string sys::getHostCPUName() {
+StringRef sys::getHostCPUName() {
   return "generic";
 }
 #endif
 
-#if defined(__linux__) && defined(__arm__)
+#if defined(__linux__) && (defined(__arm__) || defined(__aarch64__))
 bool sys::getHostCPUFeatures(StringMap<bool> &Features) {
   std::string Err;
   DataStreamer *DS = getDataFileStreamer("/proc/cpuinfo", &Err);
@@ -699,8 +717,24 @@ bool sys::getHostCPUFeatures(StringMap<bool> &Features) {
       break;
     }
 
+#if defined(__aarch64__)
+  // Keep track of which crypto features we have seen
+  enum {
+    CAP_AES   = 0x1,
+    CAP_PMULL = 0x2,
+    CAP_SHA1  = 0x4,
+    CAP_SHA2  = 0x8
+  };
+  uint32_t crypto = 0;
+#endif
+
   for (unsigned I = 0, E = CPUFeatures.size(); I != E; ++I) {
     StringRef LLVMFeatureStr = StringSwitch<StringRef>(CPUFeatures[I])
+#if defined(__aarch64__)
+      .Case("asimd", "neon")
+      .Case("fp", "fp-armv8")
+      .Case("crc32", "crc")
+#else
       .Case("half", "fp16")
       .Case("neon", "neon")
       .Case("vfpv3", "vfp3")
@@ -708,12 +742,32 @@ bool sys::getHostCPUFeatures(StringMap<bool> &Features) {
       .Case("vfpv4", "vfp4")
       .Case("idiva", "hwdiv-arm")
       .Case("idivt", "hwdiv")
+#endif
       .Default("");
 
+#if defined(__aarch64__)
+    // We need to check crypto separately since we need all of the crypto
+    // extensions to enable the subtarget feature
+    if (CPUFeatures[I] == "aes")
+      crypto |= CAP_AES;
+    else if (CPUFeatures[I] == "pmull")
+      crypto |= CAP_PMULL;
+    else if (CPUFeatures[I] == "sha1")
+      crypto |= CAP_SHA1;
+    else if (CPUFeatures[I] == "sha2")
+      crypto |= CAP_SHA2;
+#endif
+
     if (LLVMFeatureStr != "")
       Features.GetOrCreateValue(LLVMFeatureStr).setValue(true);
   }
 
+#if defined(__aarch64__)
+  // If we have all crypto bits we can add the feature
+  if (crypto == (CAP_AES | CAP_PMULL | CAP_SHA1 | CAP_SHA2))
+    Features.GetOrCreateValue("crypto").setValue(true);
+#endif
+
   return true;
 }
 #else