Support for Runtime CPU type detection via builtins (issue5754058)

Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 24695 bytes --]

Subject:Support for Runtime CPU type detection via builtins

Hi,

    I have made a new patch to only have two builtins :

* __builtin_cpu_is ("<CPUNAME>")
* __builtin_cpu_supports ("<FEATURE>")

apart from the cpu init builtin, __builtin_cpu_init.

List of CPU names :

* "amd"
* "intel"
* "atom"
* "core2"
* "corei7"
* "nehalem"
* "westmere"
* "sandybridge"
* "amdfam10h"
* "barcelona"
* "shanghai"
* "istanbul"
* "bdver1"
* "bdver2"

List of CPU features :

* "cmov"
* "mmx"
* "popcnt"
* "sse"
* "sse2"
* "sse3"
* "ssse3"
* "sse4.1"
* "sse4.2"

As an example, to check if CPU is corei7, call __builtin_cpu_is ("corei7")

Comments?
Thanks.



	* config/i386/i386.c (build_struct_with_one_bit_fields): New function.
	(make_var_decl): New function.
	(get_field_from_struct): New function.
	(fold_builtin_target): New function.
	(ix86_fold_builtin): New function.
	(ix86_expand_builtin): Expand new builtins by folding them.
	(make_cpu_type_builtin): New functions.
	(ix86_init_platform_type_builtins): Make the new builtins.
	(ix86_init_builtins): Make new builtins to detect CPU type.
	(TARGET_FOLD_BUILTIN): New macro.
	(IX86_BUILTIN_CPU_INIT): New enum value.
	(IX86_BUILTIN_CPU_IS): New enum value.
	(IX86_BUILTIN_CPU_SUPPORTS): New enum value.
	* config/i386/i386-builtin-types.def: New function type.
	* testsuite/gcc.target/builtin_target.c: New testcase.
	
	* libgcc/config/i386/i386-cpuinfo.c: New file.
	* libgcc/config/i386/t-cpuinfo: New file.
	* libgcc/config.host: Include t-cpuinfo.
	* libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
	and __cpu_features.

Index: libgcc/config.host
===================================================================
--- libgcc/config.host	(revision 185898)
+++ libgcc/config.host	(working copy)
@@ -1130,7 +1130,7 @@ i[34567]86-*-linux* | x86_64-*-linux* | \
   i[34567]86-*-kfreebsd*-gnu | x86_64-*-kfreebsd*-gnu | \
   i[34567]86-*-knetbsd*-gnu | \
   i[34567]86-*-gnu*)
-	tmake_file="${tmake_file} t-tls i386/t-linux"
+	tmake_file="${tmake_file} t-tls i386/t-linux i386/t-cpuinfo"
 	if test "$libgcc_cv_cfi" = "yes"; then
 		tmake_file="${tmake_file} t-stack i386/t-stack-i386"
 	fi
Index: libgcc/config/i386/t-cpuinfo
===================================================================
--- libgcc/config/i386/t-cpuinfo	(revision 0)
+++ libgcc/config/i386/t-cpuinfo	(revision 0)
@@ -0,0 +1 @@
+LIB2ADD += $(srcdir)/config/i386/i386-cpuinfo.c
Index: libgcc/config/i386/libgcc-glibc.ver
===================================================================
--- libgcc/config/i386/libgcc-glibc.ver	(revision 185898)
+++ libgcc/config/i386/libgcc-glibc.ver	(working copy)
@@ -147,6 +147,11 @@ GCC_4.3.0 {
   __trunctfxf2
   __unordtf2
 }
+
+GCC_4.8.0 {
+  __cpu_model
+  __cpu_features
+}
 %else
 GCC_4.4.0 {
   __addtf3
@@ -183,4 +188,9 @@ GCC_4.4.0 {
 GCC_4.5.0 {
   __extendxftf2
 }
+
+GCC_4.8.0 {
+  __cpu_model
+  __cpu_features
+}
 %endif
Index: libgcc/config/i386/i386-cpuinfo.c
===================================================================
--- libgcc/config/i386/i386-cpuinfo.c	(revision 0)
+++ libgcc/config/i386/i386-cpuinfo.c	(revision 0)
@@ -0,0 +1,306 @@
+/* Get CPU type and Features for x86 processors.
+   Copyright (C) 2012 Free Software Foundation, Inc.
+   Contributed by Sriraman Tallam (tmsriram@google.com)
+
+This file is part of GCC.
+
+GCC is free software; you can redistribute it and/or modify it under
+the terms of the GNU General Public License as published by the Free
+Software Foundation; either version 3, or (at your option) any later
+version.
+
+GCC is distributed in the hope that it will be useful, but WITHOUT ANY
+WARRANTY; without even the implied warranty of MERCHANTABILITY or
+FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+for more details.
+
+You should have received a copy of the GNU General Public License
+along with GCC; see the file COPYING3.  If not see
+<http://www.gnu.org/licenses/>. */
+
+#include "cpuid.h"
+#include "tsystem.h"
+
+int __cpu_indicator_init (void) __attribute__ ((constructor (101)));
+
+enum vendor_signatures
+{
+  SIG_INTEL =	0x756e6547 /* Genu */,
+  SIG_AMD =	0x68747541 /* Auth */
+};
+
+/* ISA Features supported. */
+
+struct __processor_features
+{
+  unsigned int __cpu_cmov : 1;
+  unsigned int __cpu_mmx : 1;
+  unsigned int __cpu_popcnt : 1;
+  unsigned int __cpu_sse : 1;
+  unsigned int __cpu_sse2 : 1;
+  unsigned int __cpu_sse3 : 1;
+  unsigned int __cpu_ssse3 : 1;
+  unsigned int __cpu_sse4_1 : 1;
+  unsigned int __cpu_sse4_2 : 1;
+} __cpu_features;
+
+/* Processor Model. */
+
+struct __processor_model
+{
+  /* Vendor. */
+  unsigned int __cpu_is_amd : 1;
+  unsigned int __cpu_is_intel : 1;
+  /* CPU type. */
+  unsigned int __cpu_is_intel_atom : 1;
+  unsigned int __cpu_is_intel_core2 : 1;
+  unsigned int __cpu_is_intel_corei7 : 1;
+  unsigned int __cpu_is_intel_corei7_nehalem : 1;
+  unsigned int __cpu_is_intel_corei7_westmere : 1;
+  unsigned int __cpu_is_intel_corei7_sandybridge : 1;
+  unsigned int __cpu_is_amdfam10h : 1;
+  unsigned int __cpu_is_amdfam10h_barcelona : 1;
+  unsigned int __cpu_is_amdfam10h_shanghai : 1;
+  unsigned int __cpu_is_amdfam10h_istanbul : 1;
+  unsigned int __cpu_is_amdfam15h_bdver1 : 1;
+  unsigned int __cpu_is_amdfam15h_bdver2 : 1;
+} __cpu_model;
+
+/* Get the specific type of AMD CPU.  */
+
+static void
+get_amd_cpu (unsigned int family, unsigned int model)
+{
+  switch (family)
+    {
+    /* AMD Family 10h.  */
+    case 0x10:
+      switch (model)
+	{
+	case 0x2:
+	  /* Barcelona.  */
+	  __cpu_model.__cpu_is_amdfam10h = 1;
+	  __cpu_model.__cpu_is_amdfam10h_barcelona = 1;
+	  break;
+	case 0x4:
+	  /* Shanghai.  */
+	  __cpu_model.__cpu_is_amdfam10h = 1;
+	  __cpu_model.__cpu_is_amdfam10h_shanghai = 1;
+	  break;
+	case 0x8:
+	  /* Istanbul.  */
+	  __cpu_model.__cpu_is_amdfam10h = 1;
+	  __cpu_model.__cpu_is_amdfam10h_istanbul = 1;
+	  break;
+	default:
+	  break;
+	}
+      break;
+    /* AMD Family 15h.  */
+    case 0x15:
+      /* Bulldozer version 1.  */
+      if (model >= 0 && model <= 0xf)
+        __cpu_model.__cpu_is_amdfam15h_bdver1 = 1;
+      /* Bulldozer version 2.  */
+      if (model >= 0x10 && model <= 0x1f)
+        __cpu_model.__cpu_is_amdfam15h_bdver2 = 1;
+      break;
+    default:
+      break;
+    }
+}
+
+/* Get the specific type of Intel CPU.  */
+
+static void
+get_intel_cpu (unsigned int family, unsigned int model, unsigned int brand_id)
+{
+  /* Parse family and model only if brand ID is 0. */
+  if (brand_id == 0)
+    {
+      switch (family)
+	{
+	case 0x5:
+	  /* Pentium.  */
+	  break;
+	case 0x6:
+	  switch (model)
+	    {
+	    case 0x1c:
+	    case 0x26:
+	      /* Atom.  */
+	      __cpu_model.__cpu_is_intel_atom = 1;
+	      break;
+	    case 0x1a:
+	    case 0x1e:
+	    case 0x1f:
+	    case 0x2e:
+	      /* Nehalem.  */
+	      __cpu_model.__cpu_is_intel_corei7 = 1;
+	      __cpu_model.__cpu_is_intel_corei7_nehalem = 1;
+	      break;
+	    case 0x25:
+	    case 0x2c:
+	    case 0x2f:
+	      /* Westmere.  */
+	      __cpu_model.__cpu_is_intel_corei7 = 1;
+	      __cpu_model.__cpu_is_intel_corei7_westmere = 1;
+	      break;
+	    case 0x2a:
+	      /* Sandy Bridge.  */
+	      __cpu_model.__cpu_is_intel_corei7 = 1;
+	      __cpu_model.__cpu_is_intel_corei7_sandybridge = 1;
+	      break;
+	    case 0x17:
+	    case 0x1d:
+	      /* Penryn.  */
+	    case 0x0f:
+	      /* Merom.  */
+	      __cpu_model.__cpu_is_intel_core2 = 1;
+	      break;
+	    default:
+	      break;
+	    }
+	  break;
+	default:
+	  /* We have no idea.  */
+	  break;
+	}
+    }
+}	             	
+
+static void
+get_available_features (unsigned int ecx, unsigned int edx)
+{
+  __cpu_features.__cpu_cmov = (edx & bit_CMOV) ? 1 : 0;
+  __cpu_features.__cpu_mmx = (edx & bit_MMX) ? 1 : 0;
+  __cpu_features.__cpu_sse = (edx & bit_SSE) ? 1 : 0;
+  __cpu_features.__cpu_sse2 = (edx & bit_SSE2) ? 1 : 0;
+  __cpu_features.__cpu_popcnt = (ecx & bit_POPCNT) ? 1 : 0;
+  __cpu_features.__cpu_sse3 = (ecx & bit_SSE3) ? 1 : 0;
+  __cpu_features.__cpu_ssse3 = (ecx & bit_SSSE3) ? 1 : 0;
+  __cpu_features.__cpu_sse4_1 = (ecx & bit_SSE4_1) ? 1 : 0;
+  __cpu_features.__cpu_sse4_2 = (ecx & bit_SSE4_2) ? 1 : 0;
+}
+
+
+/* Sanity check for the vendor and cpu type flags.  */
+
+static int
+sanity_check (void)
+{
+  unsigned int one_type = 0;
+
+  /* Vendor cannot be Intel and AMD. */
+  gcc_assert((__cpu_model.__cpu_is_intel == 0)
+             || (__cpu_model.__cpu_is_amd == 0));
+ 
+  /* Only one CPU type can be set. */
+  one_type = (__cpu_model.__cpu_is_intel_atom
+	      + __cpu_model.__cpu_is_intel_core2
+	      + __cpu_model.__cpu_is_intel_corei7_nehalem
+	      + __cpu_model.__cpu_is_intel_corei7_westmere
+	      + __cpu_model.__cpu_is_intel_corei7_sandybridge
+	      + __cpu_model.__cpu_is_amdfam10h_barcelona
+	      + __cpu_model.__cpu_is_amdfam10h_shanghai
+	      + __cpu_model.__cpu_is_amdfam10h_istanbul
+	      + __cpu_model.__cpu_is_amdfam15h_bdver1
+	      + __cpu_model.__cpu_is_amdfam15h_bdver2);
+
+  gcc_assert (one_type <= 1);
+  return 0;
+}
+
+/* A noinline function calling __get_cpuid. Having many calls to
+   cpuid in one function in 32-bit mode causes GCC to complain:
+   "canâ??t find a register in class â??CLOBBERED_REGSâ??".  This is
+   related to PR rtl-optimization 44174. */
+
+static int __attribute__ ((noinline))
+__get_cpuid_output (unsigned int __level,
+		    unsigned int *__eax, unsigned int *__ebx,
+		    unsigned int *__ecx, unsigned int *__edx)
+{
+  return __get_cpuid (__level, __eax, __ebx, __ecx, __edx);
+}
+
+
+/* A constructor function that is sets __cpu_model and __cpu_features with
+   the right values.  This needs to run only once.  This constructor is
+   given the highest priority and it should run before constructors without
+   the priority set.  However, it still runs after ifunc initializers and
+   needs to be called explicitly there.  */
+
+int __attribute__ ((constructor (101)))
+__cpu_indicator_init (void)
+{
+  unsigned int eax, ebx, ecx, edx;
+
+  int max_level = 5;
+  unsigned int vendor;
+  unsigned int model, family, brand_id;
+  unsigned int extended_model, extended_family;
+  static int called = 0;
+
+  /* This function needs to run just once.  */
+  if (called)
+    return 0;
+  else
+    called = 1;
+
+  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
+  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
+    return -1;
+
+  vendor = ebx;
+  max_level = eax;
+
+  if (max_level < 1)
+    return -1;
+
+  if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
+    return -1;
+
+  model = (eax >> 4) & 0x0f;
+  family = (eax >> 8) & 0x0f;
+  brand_id = ebx & 0xff;
+  extended_model = (eax >> 12) & 0xf0;
+  extended_family = (eax >> 20) & 0xff;
+
+  if (vendor == SIG_INTEL)
+    {
+      /* Adjust model and family for Intel CPUS. */
+      if (family == 0x0f)
+	{
+	  family += extended_family;
+	  model += extended_model;
+	}
+      else if (family == 0x06)
+	model += extended_model;
+
+      /* Get CPU type.  */
+      __cpu_model.__cpu_is_intel = 1;
+      get_intel_cpu (family, model, brand_id);
+    }
+
+  if (vendor == SIG_AMD)
+    {
+      /* Adjust model and family for AMD CPUS. */
+      if (family == 0x0f)
+	{
+	  family += extended_family;
+	  model += (extended_model << 4);
+	}
+
+      /* Get CPU type.  */
+      __cpu_model.__cpu_is_amd = 1;
+      get_amd_cpu (family, model);
+    }
+
+  /* Find available features. */
+  get_available_features (ecx, edx);
+
+  sanity_check ();
+
+  return 0;
+}
Index: gcc/testsuite/gcc.target/i386/builtin_target.c
===================================================================
--- gcc/testsuite/gcc.target/i386/builtin_target.c	(revision 0)
+++ gcc/testsuite/gcc.target/i386/builtin_target.c	(revision 0)
@@ -0,0 +1,63 @@
+/* This test checks if the __builtin_cpu_is and __builtin_cpu_supports calls
+   are recognized. */
+
+/* { dg-do run } */
+
+int
+fn1 ()
+{
+  if (__builtin_cpu_supports ("cmov") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("mmx") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("popcnt") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse2") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse3") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("ssse3") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse4.1") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse4.2") < 0)
+    return -1;
+  if (__builtin_cpu_is ("amd") < 0)
+    return -1;
+  if (__builtin_cpu_is ("intel") < 0)
+    return -1;
+  if (__builtin_cpu_is ("atom") < 0)
+    return -1;
+  if (__builtin_cpu_is ("core2") < 0)
+    return -1;
+  if (__builtin_cpu_is ("corei7") < 0)
+    return -1;
+  if (__builtin_cpu_is ("nehalem") < 0)
+    return -1;
+  if (__builtin_cpu_is ("westmere") < 0)
+    return -1;
+  if (__builtin_cpu_is ("sandybridge") < 0)
+    return -1;
+  if (__builtin_cpu_is ("amdfam10h") < 0)
+    return -1;
+  if (__builtin_cpu_is ("barcelona") < 0)
+    return -1;
+  if (__builtin_cpu_is ("shanghai") < 0)
+    return -1;
+  if (__builtin_cpu_is ("istanbul") < 0)
+    return -1;
+  if (__builtin_cpu_is ("bdver1") < 0)
+    return -1;
+  if (__builtin_cpu_is ("bdver2") < 0)
+    return -1;
+
+  return 0;
+}
+
+int main ()
+{
+  __builtin_cpu_init ();
+  return fn1 ();
+}
Index: gcc/config/i386/i386-builtin-types.def
===================================================================
--- gcc/config/i386/i386-builtin-types.def	(revision 185898)
+++ gcc/config/i386/i386-builtin-types.def	(working copy)
@@ -155,6 +155,7 @@ DEF_FUNCTION_TYPE (INT, V4SF)
 DEF_FUNCTION_TYPE (INT, V8QI)
 DEF_FUNCTION_TYPE (INT, V8SF)
 DEF_FUNCTION_TYPE (INT, V32QI)
+DEF_FUNCTION_TYPE (INT, PCCHAR)
 DEF_FUNCTION_TYPE (INT64, INT64)
 DEF_FUNCTION_TYPE (INT64, V2DF)
 DEF_FUNCTION_TYPE (INT64, V4SF)
Index: gcc/config/i386/i386.c
===================================================================
--- gcc/config/i386/i386.c	(revision 185898)
+++ gcc/config/i386/i386.c	(working copy)
@@ -25809,6 +25809,11 @@ enum ix86_builtins
   /* CFString built-in for darwin */
   IX86_BUILTIN_CFSTRING,
 
+  /* Builtins to get CPU type and supported features. */
+  IX86_BUILTIN_CPU_INIT,
+  IX86_BUILTIN_CPU_IS,
+  IX86_BUILTIN_CPU_SUPPORTS,
+
   IX86_BUILTIN_MAX
 };
 
@@ -27627,6 +27632,308 @@ ix86_init_mmx_sse_builtins (void)
     }
 }
 
+/* Returns a struct type with name NAME and number of fields equal to
+   NUM_FIELDS.  Each field is a unsigned int bit field of length 1 bit. */
+
+static tree
+build_struct_with_one_bit_fields (int num_fields, const char *name)
+{
+  int i;
+  char field_name [10];
+  tree field = NULL_TREE, field_chain = NULL_TREE;
+  tree type = make_node (RECORD_TYPE);
+
+  strcpy (field_name, "kk_field");
+
+  for (i = 0; i < num_fields; i++)
+    {
+      /* Name the fields, 0_field, 1_field, ... */
+      field_name [0] = '0' + (i / 10);
+      field_name [1] = '0' + (i % 10);
+      field = build_decl (UNKNOWN_LOCATION, FIELD_DECL,
+			  get_identifier (field_name), unsigned_type_node);
+      DECL_BIT_FIELD (field) = 1;
+      DECL_SIZE (field) = bitsize_one_node;
+      if (field_chain != NULL_TREE)
+	DECL_CHAIN (field) = field_chain;
+      field_chain = field;
+    }
+  finish_builtin_struct (type, name, field_chain, NULL_TREE);
+  return type;
+}
+
+/* Returns a extern, comdat VAR_DECL of type TYPE and name NAME. */
+
+static tree
+make_var_decl (tree type, const char *name)
+{
+  tree new_decl;
+
+  new_decl = build_decl (UNKNOWN_LOCATION,
+	                 VAR_DECL,
+	  	         get_identifier(name),
+		         type);
+
+  DECL_EXTERNAL (new_decl) = 1;
+  TREE_STATIC (new_decl) = 1;
+  TREE_PUBLIC (new_decl) = 1;
+  DECL_INITIAL (new_decl) = 0;
+  DECL_ARTIFICIAL (new_decl) = 0;
+  DECL_PRESERVE_P (new_decl) = 1;
+
+  make_decl_one_only (new_decl, DECL_ASSEMBLER_NAME (new_decl));
+  assemble_variable (new_decl, 0, 0, 0);
+
+  return new_decl;
+}
+
+/* Traverses the chain of fields in STRUCT_TYPE and returns the FIELD_NUM
+   numbered field. */
+
+static tree
+get_field_from_struct (tree struct_type, int field_num)
+{
+  int i;
+  tree field = TYPE_FIELDS (struct_type);
+
+  for (i = 0; i < field_num; i++, field = DECL_CHAIN(field))
+    {
+      gcc_assert (field != NULL_TREE);
+    }
+
+  return field;
+}
+
+/* FNDECL is a __builtin_cpu_is or a __builtin_cpu_supports call that is folded
+   into an integer defined in libgcc/config/i386/i386-cpuinfo.c */
+
+static tree 
+fold_builtin_cpu (tree fndecl, tree *args)
+{
+  int i;
+  enum ix86_builtins fn_code = (enum ix86_builtins)
+				DECL_FUNCTION_CODE (fndecl);
+  tree param_string_cst = NULL;
+
+  /* This is the order of bit-fields in __processor_features in
+     i386-cpuinfo.c */
+  enum processor_features
+  {
+    F_CMOV = 0,
+    F_MMX,
+    F_POPCNT,
+    F_SSE,
+    F_SSE2,
+    F_SSE3,
+    F_SSSE3,
+    F_SSE4_1,
+    F_SSE4_2,
+    F_MAX
+  };
+
+  /* This is the order of bit-fields in __processor_model in
+     i386-cpuinfo.c */
+  enum processor_model
+  {
+    M_AMD = 0,
+    M_INTEL,
+    M_INTEL_ATOM,
+    M_INTEL_CORE2,
+    M_INTEL_COREI7,
+    M_INTEL_COREI7_NEHALEM,
+    M_INTEL_COREI7_WESTMERE,
+    M_INTEL_COREI7_SANDYBRIDGE,
+    M_AMDFAM10H,
+    M_AMDFAM10H_BARCELONA,
+    M_AMDFAM10H_SHANGHAI,
+    M_AMDFAM10H_ISTANBUL,
+    M_AMDFAM15H_BDVER1,
+    M_AMDFAM15H_BDVER2,
+    M_MAX
+  };
+
+  static struct _arch_names_table
+    {
+      const char *const name;
+      const enum processor_model model;
+    }
+  const arch_names_table[M_MAX] =
+    {
+      {"amd", M_AMD},
+      {"intel", M_INTEL},
+      {"atom", M_INTEL_ATOM},
+      {"core2", M_INTEL_CORE2},
+      {"corei7", M_INTEL_COREI7},
+      {"nehalem", M_INTEL_COREI7_NEHALEM},
+      {"westmere", M_INTEL_COREI7_WESTMERE},
+      {"sandybridge", M_INTEL_COREI7_SANDYBRIDGE},
+      {"amdfam10h", M_AMDFAM10H},
+      {"barcelona", M_AMDFAM10H_BARCELONA},
+      {"shanghai", M_AMDFAM10H_SHANGHAI},
+      {"istanbul", M_AMDFAM10H_ISTANBUL},
+      {"bdver1", M_AMDFAM15H_BDVER1},
+      {"bdver2", M_AMDFAM15H_BDVER2},
+    };
+
+  static struct _isa_names_table
+    {
+      const char *const name;
+      const enum processor_features feature;
+    }
+  const isa_names_table[F_MAX] =
+    {
+      {"cmov",   F_CMOV},
+      {"mmx",    F_MMX},
+      {"popcnt", F_POPCNT},
+      {"sse",    F_SSE},
+      {"sse2",   F_SSE2},
+      {"sse3",   F_SSE3},
+      {"ssse3",  F_SSSE3},
+      {"sse4.1", F_SSE4_1},
+      {"sse4.2", F_SSE4_2}
+    };
+
+  static tree __processor_features_type = NULL_TREE;
+  static tree __cpu_features_var = NULL_TREE;
+  static tree __processor_model_type = NULL_TREE;
+  static tree __cpu_model_var = NULL_TREE;
+  static tree field;
+  static tree which_struct;
+
+  if (__processor_features_type == NULL_TREE)
+    __processor_features_type = build_struct_with_one_bit_fields (F_MAX,
+ 			          "__processor_features");
+
+  if (__processor_model_type == NULL_TREE)
+    __processor_model_type = build_struct_with_one_bit_fields (M_MAX,
+ 			          "__processor_model");
+
+  if (__cpu_features_var == NULL_TREE)
+    __cpu_features_var = make_var_decl (__processor_features_type,
+					"__cpu_features");
+
+  if (__cpu_model_var == NULL_TREE)
+    __cpu_model_var = make_var_decl (__processor_model_type,
+				     "__cpu_model");
+
+  gcc_assert ((args != NULL) && (*args != NULL));
+
+  param_string_cst = *args;
+  while (param_string_cst
+	 && TREE_CODE (param_string_cst) !=  STRING_CST)
+    {
+      /* *args must be a expr that can contain other EXPRS leading to a
+	 STRING_CST.   */
+      if (!EXPR_P (param_string_cst))
+ 	{
+	  error ("Parameter to builtin must be a string constant or literal");
+	  return integer_zero_node;
+	}
+      param_string_cst = TREE_OPERAND (EXPR_CHECK (param_string_cst), 0);
+    }
+ 
+  gcc_assert (param_string_cst);
+
+  if (fn_code == IX86_BUILTIN_CPU_IS)
+    {
+
+      for (i = 0; i < M_MAX; i++)
+	if (strcmp (arch_names_table[i].name,
+	    TREE_STRING_POINTER (param_string_cst)) == 0)
+	  break;
+
+      if (i == M_MAX)
+	{
+	  error ("Parameter to builtin not valid: %s",
+	         TREE_STRING_POINTER (param_string_cst));
+	  return integer_zero_node;
+	}
+
+      field = get_field_from_struct (__processor_model_type,
+				     arch_names_table[i].model);
+      which_struct = __cpu_model_var;
+    }
+
+  if (fn_code == IX86_BUILTIN_CPU_SUPPORTS)
+    {
+      for (i = 0; i < F_MAX; i++)
+	if (strcmp (isa_names_table[i].name,
+	    TREE_STRING_POINTER (param_string_cst)) == 0)
+	  break;
+
+      if (i == F_MAX)
+	{
+	  error ("Parameter to builtin not valid: %s",
+	       	 TREE_STRING_POINTER (param_string_cst));
+	  return integer_zero_node;
+	}
+
+      field = get_field_from_struct (__processor_features_type,
+				     isa_names_table[i].feature);
+      which_struct = __cpu_features_var;
+    }
+
+  return build3 (COMPONENT_REF, TREE_TYPE (field), which_struct,
+		 field, NULL_TREE);
+}
+
+static tree
+ix86_fold_builtin (tree fndecl, int n_args,
+		   tree *args, bool ignore ATTRIBUTE_UNUSED)
+{
+  if (DECL_BUILT_IN_CLASS (fndecl) == BUILT_IN_MD)
+    {
+      enum ix86_builtins fn_code = (enum ix86_builtins)
+				   DECL_FUNCTION_CODE (fndecl);
+      if (fn_code ==  IX86_BUILTIN_CPU_IS
+	  || fn_code == IX86_BUILTIN_CPU_SUPPORTS)
+	{
+	  gcc_assert (n_args == 1);
+          return fold_builtin_cpu (fndecl, args);
+	}
+    }
+
+  return NULL_TREE;
+}
+
+/* Make builtins to detect cpu type and features supported.  NAME is
+   the builtin name, CODE is the builtin code, and FTYPE is the function
+   type of the builtin.  */
+
+static void
+make_cpu_type_builtin (const char* name, int code,
+		       enum ix86_builtin_func_type ftype)
+{
+  tree decl;
+  tree type;
+
+  type = ix86_get_builtin_func_type (ftype);
+  decl = add_builtin_function (name, type, code, BUILT_IN_MD,
+			       NULL, NULL_TREE);
+  gcc_assert (decl != NULL_TREE);
+  ix86_builtins[(int) code] = decl;
+  TREE_READONLY (decl) = 1;
+}
+
+/* Make builtins to get CPU type and features supported.  The created
+   builtins are :
+
+   __builtin_cpu_init (), to detect cpu type and features,
+   __builtin_cpu_is ("<CPUNAME>"), to check if cpu is of type <CPUNAME>,
+   __builtin_cpu_supports ("<FEATURE>"), to check if cpu supports <FEATURE>
+   */
+
+static void
+ix86_init_platform_type_builtins (void)
+{
+  make_cpu_type_builtin ("__builtin_cpu_init", IX86_BUILTIN_CPU_INIT,
+			 INT_FTYPE_VOID);
+  make_cpu_type_builtin ("__builtin_cpu_is", IX86_BUILTIN_CPU_IS,
+			 INT_FTYPE_PCCHAR);
+  make_cpu_type_builtin ("__builtin_cpu_supports", IX86_BUILTIN_CPU_SUPPORTS,
+			 INT_FTYPE_PCCHAR);
+}
+
 /* Internal method for ix86_init_builtins.  */
 
 static void
@@ -27710,6 +28017,9 @@ ix86_init_builtins (void)
 
   ix86_init_builtin_types ();
 
+  /* Builtins to get CPU type and features. */
+  ix86_init_platform_type_builtins ();
+
   /* TFmode support builtins.  */
   def_builtin_const (0, "__builtin_infq",
 		     FLOAT128_FTYPE_VOID, IX86_BUILTIN_INFQ);
@@ -29328,6 +29638,28 @@ ix86_expand_builtin (tree exp, rtx target, rtx sub
   enum machine_mode mode0, mode1, mode2, mode3, mode4;
   unsigned int fcode = DECL_FUNCTION_CODE (fndecl);
 
+  /* For CPU builtins that can be folded, fold first and expand the fold.  */
+  switch (fcode)
+    {
+    case IX86_BUILTIN_CPU_INIT:
+      {
+	/* Make it call __cpu_indicator_init in libgcc. */
+	tree call_expr, fndecl, type;
+        type = build_function_type_list (integer_type_node, NULL_TREE); 
+	fndecl = build_fn_decl ("__cpu_indicator_init", type);
+	call_expr = build_call_expr (fndecl, 0); 
+	return expand_expr (call_expr, target, mode, EXPAND_NORMAL);
+      }
+    case IX86_BUILTIN_CPU_IS:
+    case IX86_BUILTIN_CPU_SUPPORTS:
+      {
+	tree arg0 = CALL_EXPR_ARG (exp, 0);
+	tree fold_expr = fold_builtin_cpu (fndecl, &arg0);
+	gcc_assert (fold_expr != NULL_TREE);
+        return expand_expr (fold_expr, target, mode, EXPAND_NORMAL);
+      }
+    }
+
   /* Determine whether the builtin function is available under the current ISA.
      Originally the builtin was not created if it wasn't applicable to the
      current ISA based on the command line switches.  With function specific
@@ -39058,6 +39390,9 @@ ix86_autovectorize_vector_sizes (void)
 #undef TARGET_BUILD_BUILTIN_VA_LIST
 #define TARGET_BUILD_BUILTIN_VA_LIST ix86_build_builtin_va_list
 
+#undef TARGET_FOLD_BUILTIN
+#define TARGET_FOLD_BUILTIN ix86_fold_builtin
+
 #undef TARGET_ENUM_VA_LIST_P
 #define TARGET_ENUM_VA_LIST_P ix86_enum_va_list

--
This patch is available for review at http://codereview.appspot.com/5754058

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Michael Matz]

Hi,

On Thu, 29 Mar 2012, Sriraman Tallam wrote:

> +struct __processor_model
> +{
> +  /* Vendor. */
> +  unsigned int __cpu_is_amd : 1;
> +  unsigned int __cpu_is_intel : 1;
> +  /* CPU type. */
> +  unsigned int __cpu_is_intel_atom : 1;
> +  unsigned int __cpu_is_intel_core2 : 1;
> +  unsigned int __cpu_is_intel_corei7 : 1;
> +  unsigned int __cpu_is_intel_corei7_nehalem : 1;
> +  unsigned int __cpu_is_intel_corei7_westmere : 1;
> +  unsigned int __cpu_is_intel_corei7_sandybridge : 1;
> +  unsigned int __cpu_is_amdfam10h : 1;
> +  unsigned int __cpu_is_amdfam10h_barcelona : 1;
> +  unsigned int __cpu_is_amdfam10h_shanghai : 1;
> +  unsigned int __cpu_is_amdfam10h_istanbul : 1;
> +  unsigned int __cpu_is_amdfam15h_bdver1 : 1;
> +  unsigned int __cpu_is_amdfam15h_bdver2 : 1;
> +} __cpu_model;

It doesn't make sense for the model to be a bitfield, a processor will 
have only ever exactly one model.  Just make it an enum or even just an 
int.


Ciao,
Michael.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

[-- Attachment #1: Type: text/plain, Size: 2496 bytes --]

On Fri, Mar 30, 2012 at 5:47 AM, Michael Matz <matz@suse.de> wrote:
> Hi,
>
> On Thu, 29 Mar 2012, Sriraman Tallam wrote:
>
>> +struct __processor_model
>> +{
>> +  /* Vendor. */
>> +  unsigned int __cpu_is_amd : 1;
>> +  unsigned int __cpu_is_intel : 1;
>> +  /* CPU type. */
>> +  unsigned int __cpu_is_intel_atom : 1;
>> +  unsigned int __cpu_is_intel_core2 : 1;
>> +  unsigned int __cpu_is_intel_corei7 : 1;
>> +  unsigned int __cpu_is_intel_corei7_nehalem : 1;
>> +  unsigned int __cpu_is_intel_corei7_westmere : 1;
>> +  unsigned int __cpu_is_intel_corei7_sandybridge : 1;
>> +  unsigned int __cpu_is_amdfam10h : 1;
>> +  unsigned int __cpu_is_amdfam10h_barcelona : 1;
>> +  unsigned int __cpu_is_amdfam10h_shanghai : 1;
>> +  unsigned int __cpu_is_amdfam10h_istanbul : 1;
>> +  unsigned int __cpu_is_amdfam15h_bdver1 : 1;
>> +  unsigned int __cpu_is_amdfam15h_bdver2 : 1;
>> +} __cpu_model;
>
> It doesn't make sense for the model to be a bitfield, a processor will
> have only ever exactly one model.  Just make it an enum or even just an
> int.

Not entirely true, nehalem and corei7 can be both set. However, I
modified this by dividing it into types and sub types and then did
what you said.

       * config/i386/i386.c (build_processor_features_struct): New function.
       (build_processor_model_struct): New function.
       (make_var_decl): New function.
       (get_field_from_struct): New function.
       (fold_builtin_target): New function.
       (ix86_fold_builtin): New function.
       (ix86_expand_builtin): Expand new builtins by folding them.
       (make_cpu_type_builtin): New functions.
       (ix86_init_platform_type_builtins): Make the new builtins.
       (ix86_init_builtins): Make new builtins to detect CPU type.
       (TARGET_FOLD_BUILTIN): New macro.
       (IX86_BUILTIN_CPU_INIT): New enum value.
       (IX86_BUILTIN_CPU_IS): New enum value.
       (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
       * config/i386/i386-builtin-types.def: New function type.
       * testsuite/gcc.target/builtin_target.c: New testcase.

       * libgcc/config/i386/i386-cpuinfo.c: New file.
       * libgcc/config/i386/t-cpuinfo: New file.
       * libgcc/config.host: Include t-cpuinfo.
       * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
       and __cpu_features.

Patch available for review here:
http://codereview.appspot.com/5754058

Thanks,
-Sri.


>
>
> Ciao,
> Michael.

[-- Attachment #2: cpu_runtime_patch.txt --]
[-- Type: text/plain, Size: 24167 bytes --]

Index: libgcc/config.host
===================================================================
--- libgcc/config.host	(revision 185898)
+++ libgcc/config.host	(working copy)
@@ -1130,7 +1130,7 @@ i[34567]86-*-linux* | x86_64-*-linux* | \
   i[34567]86-*-kfreebsd*-gnu | x86_64-*-kfreebsd*-gnu | \
   i[34567]86-*-knetbsd*-gnu | \
   i[34567]86-*-gnu*)
-	tmake_file="${tmake_file} t-tls i386/t-linux"
+	tmake_file="${tmake_file} t-tls i386/t-linux i386/t-cpuinfo"
 	if test "$libgcc_cv_cfi" = "yes"; then
 		tmake_file="${tmake_file} t-stack i386/t-stack-i386"
 	fi
Index: libgcc/config/i386/t-cpuinfo
===================================================================
--- libgcc/config/i386/t-cpuinfo	(revision 0)
+++ libgcc/config/i386/t-cpuinfo	(revision 0)
@@ -0,0 +1 @@
+LIB2ADD += $(srcdir)/config/i386/i386-cpuinfo.c
Index: libgcc/config/i386/i386-cpuinfo.c
===================================================================
--- libgcc/config/i386/i386-cpuinfo.c	(revision 0)
+++ libgcc/config/i386/i386-cpuinfo.c	(revision 0)
@@ -0,0 +1,298 @@
+/* Get CPU type and Features for x86 processors.
+   Copyright (C) 2012 Free Software Foundation, Inc.
+   Contributed by Sriraman Tallam (tmsriram@google.com)
+
+This file is part of GCC.
+
+GCC is free software; you can redistribute it and/or modify it under
+the terms of the GNU General Public License as published by the Free
+Software Foundation; either version 3, or (at your option) any later
+version.
+
+GCC is distributed in the hope that it will be useful, but WITHOUT ANY
+WARRANTY; without even the implied warranty of MERCHANTABILITY or
+FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+for more details.
+
+You should have received a copy of the GNU General Public License
+along with GCC; see the file COPYING3.  If not see
+<http://www.gnu.org/licenses/>. */
+
+#include "cpuid.h"
+#include "tsystem.h"
+
+int __cpu_indicator_init (void) __attribute__ ((constructor (101)));
+
+enum vendor_signatures
+{
+  SIG_INTEL =	0x756e6547 /* Genu */,
+  SIG_AMD =	0x68747541 /* Auth */
+};
+
+/* ISA Features supported. */
+
+struct __processor_features
+{
+  unsigned int __cpu_cmov : 1;
+  unsigned int __cpu_mmx : 1;
+  unsigned int __cpu_popcnt : 1;
+  unsigned int __cpu_sse : 1;
+  unsigned int __cpu_sse2 : 1;
+  unsigned int __cpu_sse3 : 1;
+  unsigned int __cpu_ssse3 : 1;
+  unsigned int __cpu_sse4_1 : 1;
+  unsigned int __cpu_sse4_2 : 1;
+} __cpu_features;
+
+/* Processor Vendor and Models. */
+
+enum processor_vendor
+{
+  VENDOR_INTEL = 1,
+  VENDOR_AMD,
+  VENDOR_MAX
+};
+
+enum processor_types
+{
+  INTEL_ATOM = 1,
+  INTEL_CORE2,
+  INTEL_COREI7,
+  AMDFAM10H,
+  AMDFAM15H,
+  CPU_TYPE_MAX
+};
+
+enum processor_subtypes
+{
+  INTEL_COREI7_NEHALEM,
+  INTEL_COREI7_WESTMERE,
+  INTEL_COREI7_SANDYBRIDGE,
+  AMDFAM10H_BARCELONA,
+  AMDFAM10H_SHANGHAI,
+  AMDFAM10H_ISTANBUL,
+  AMDFAM15H_BDVER1,
+  AMDFAM15H_BDVER2,
+  CPU_SUBTYPE_MAX
+};
+
+struct __processor_model
+{
+  unsigned int __cpu_vendor;
+  unsigned int __cpu_type;
+  unsigned int __cpu_subtype;
+} __cpu_model;
+
+/* Get the specific type of AMD CPU.  */
+
+static void
+get_amd_cpu (unsigned int family, unsigned int model)
+{
+  __cpu_model.__cpu_vendor = VENDOR_AMD;
+  switch (family)
+    {
+    /* AMD Family 10h.  */
+    case 0x10:
+      switch (model)
+	{
+	case 0x2:
+	  /* Barcelona.  */
+	  __cpu_model.__cpu_type = AMDFAM10H;
+	  __cpu_model.__cpu_subtype = AMDFAM10H_BARCELONA;
+	  break;
+	case 0x4:
+	  /* Shanghai.  */
+	  __cpu_model.__cpu_type = AMDFAM10H;
+	  __cpu_model.__cpu_subtype = AMDFAM10H_SHANGHAI;
+	  break;
+	case 0x8:
+	  /* Istanbul.  */
+	  __cpu_model.__cpu_type = AMDFAM10H;
+	  __cpu_model.__cpu_subtype = AMDFAM10H_ISTANBUL;
+	  break;
+	default:
+	  break;
+	}
+      break;
+    /* AMD Family 15h.  */
+    case 0x15:
+      __cpu_model.__cpu_type = AMDFAM15H;
+      /* Bulldozer version 1.  */
+      if ( model <= 0xf)
+	__cpu_model.__cpu_subtype = AMDFAM15H_BDVER1;
+      /* Bulldozer version 2.  */
+      if (model >= 0x10 && model <= 0x1f)
+	__cpu_model.__cpu_subtype = AMDFAM15H_BDVER2;
+      break;
+    default:
+      break;
+    }
+}
+
+/* Get the specific type of Intel CPU.  */
+
+static void
+get_intel_cpu (unsigned int family, unsigned int model, unsigned int brand_id)
+{
+  __cpu_model.__cpu_vendor = VENDOR_INTEL;
+  /* Parse family and model only if brand ID is 0. */
+  if (brand_id == 0)
+    {
+      switch (family)
+	{
+	case 0x5:
+	  /* Pentium.  */
+	  break;
+	case 0x6:
+	  switch (model)
+	    {
+	    case 0x1c:
+	    case 0x26:
+	      /* Atom.  */
+	      __cpu_model.__cpu_type = INTEL_ATOM;
+	      break;
+	    case 0x1a:
+	    case 0x1e:
+	    case 0x1f:
+	    case 0x2e:
+	      /* Nehalem.  */
+	      __cpu_model.__cpu_type = INTEL_COREI7;
+	      __cpu_model.__cpu_subtype = INTEL_COREI7_NEHALEM;
+	      break;
+	    case 0x25:
+	    case 0x2c:
+	    case 0x2f:
+	      /* Westmere.  */
+	      __cpu_model.__cpu_type = INTEL_COREI7;
+	      __cpu_model.__cpu_subtype = INTEL_COREI7_WESTMERE;
+	      break;
+	    case 0x2a:
+	      /* Sandy Bridge.  */
+	      __cpu_model.__cpu_type = INTEL_COREI7;
+	      __cpu_model.__cpu_subtype = INTEL_COREI7_SANDYBRIDGE;
+	      break;
+	    case 0x17:
+	    case 0x1d:
+	      /* Penryn.  */
+	    case 0x0f:
+	      /* Merom.  */
+	      __cpu_model.__cpu_type = INTEL_CORE2;
+	      break;
+	    default:
+	      break;
+	    }
+	  break;
+	default:
+	  /* We have no idea.  */
+	  break;
+	}
+    }
+}	             	
+
+static void
+get_available_features (unsigned int ecx, unsigned int edx)
+{
+  __cpu_features.__cpu_cmov = (edx & bit_CMOV) ? 1 : 0;
+  __cpu_features.__cpu_mmx = (edx & bit_MMX) ? 1 : 0;
+  __cpu_features.__cpu_sse = (edx & bit_SSE) ? 1 : 0;
+  __cpu_features.__cpu_sse2 = (edx & bit_SSE2) ? 1 : 0;
+  __cpu_features.__cpu_popcnt = (ecx & bit_POPCNT) ? 1 : 0;
+  __cpu_features.__cpu_sse3 = (ecx & bit_SSE3) ? 1 : 0;
+  __cpu_features.__cpu_ssse3 = (ecx & bit_SSSE3) ? 1 : 0;
+  __cpu_features.__cpu_sse4_1 = (ecx & bit_SSE4_1) ? 1 : 0;
+  __cpu_features.__cpu_sse4_2 = (ecx & bit_SSE4_2) ? 1 : 0;
+}
+
+/* A noinline function calling __get_cpuid. Having many calls to
+   cpuid in one function in 32-bit mode causes GCC to complain:
+   "can;t find a register in class CLOBBERED_REGS".  This is
+   related to PR rtl-optimization 44174. */
+
+static int __attribute__ ((noinline))
+__get_cpuid_output (unsigned int __level,
+		    unsigned int *__eax, unsigned int *__ebx,
+		    unsigned int *__ecx, unsigned int *__edx)
+{
+  return __get_cpuid (__level, __eax, __ebx, __ecx, __edx);
+}
+
+
+/* A constructor function that is sets __cpu_model and __cpu_features with
+   the right values.  This needs to run only once.  This constructor is
+   given the highest priority and it should run before constructors without
+   the priority set.  However, it still runs after ifunc initializers and
+   needs to be called explicitly there.  */
+
+int __attribute__ ((constructor (101)))
+__cpu_indicator_init (void)
+{
+  unsigned int eax, ebx, ecx, edx;
+
+  int max_level = 5;
+  unsigned int vendor;
+  unsigned int model, family, brand_id;
+  unsigned int extended_model, extended_family;
+  static int called = 0;
+
+  /* This function needs to run just once.  */
+  if (called)
+    return 0;
+  else
+    called = 1;
+
+  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
+  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
+    return -1;
+
+  vendor = ebx;
+  max_level = eax;
+
+  if (max_level < 1)
+    return -1;
+
+  if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
+    return -1;
+
+  model = (eax >> 4) & 0x0f;
+  family = (eax >> 8) & 0x0f;
+  brand_id = ebx & 0xff;
+  extended_model = (eax >> 12) & 0xf0;
+  extended_family = (eax >> 20) & 0xff;
+
+  if (vendor == SIG_INTEL)
+    {
+      /* Adjust model and family for Intel CPUS. */
+      if (family == 0x0f)
+	{
+	  family += extended_family;
+	  model += extended_model;
+	}
+      else if (family == 0x06)
+	model += extended_model;
+
+      /* Get CPU type.  */
+      get_intel_cpu (family, model, brand_id);
+    }
+
+  if (vendor == SIG_AMD)
+    {
+      /* Adjust model and family for AMD CPUS. */
+      if (family == 0x0f)
+	{
+	  family += extended_family;
+	  model += (extended_model << 4);
+	}
+
+      /* Get CPU type.  */
+      get_amd_cpu (family, model);
+    }
+
+  /* Find available features. */
+  get_available_features (ecx, edx);
+
+  gcc_assert (__cpu_model.__cpu_vendor < VENDOR_MAX);
+  gcc_assert (__cpu_model.__cpu_type < CPU_TYPE_MAX);
+  gcc_assert (__cpu_model.__cpu_subtype < CPU_SUBTYPE_MAX);
+
+  return 0;
+}
Index: libgcc/config/i386/libgcc-glibc.ver
===================================================================
--- libgcc/config/i386/libgcc-glibc.ver	(revision 185898)
+++ libgcc/config/i386/libgcc-glibc.ver	(working copy)
@@ -147,6 +147,11 @@ GCC_4.3.0 {
   __trunctfxf2
   __unordtf2
 }
+
+GCC_4.8.0 {
+  __cpu_model
+  __cpu_features
+}
 %else
 GCC_4.4.0 {
   __addtf3
@@ -183,4 +188,9 @@ GCC_4.4.0 {
 GCC_4.5.0 {
   __extendxftf2
 }
+
+GCC_4.8.0 {
+  __cpu_model
+  __cpu_features
+}
 %endif
Index: gcc/testsuite/gcc.target/i386/builtin_target.c
===================================================================
--- gcc/testsuite/gcc.target/i386/builtin_target.c	(revision 0)
+++ gcc/testsuite/gcc.target/i386/builtin_target.c	(revision 0)
@@ -0,0 +1,65 @@
+/* This test checks if the __builtin_cpu_is and __builtin_cpu_supports calls
+   are recognized. */
+
+/* { dg-do run } */
+
+int
+fn1 ()
+{
+  if (__builtin_cpu_supports ("cmov") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("mmx") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("popcnt") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse2") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse3") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("ssse3") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse4.1") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse4.2") < 0)
+    return -1;
+  if (__builtin_cpu_is ("amd") < 0)
+    return -1;
+  if (__builtin_cpu_is ("intel") < 0)
+    return -1;
+  if (__builtin_cpu_is ("atom") < 0)
+    return -1;
+  if (__builtin_cpu_is ("core2") < 0)
+    return -1;
+  if (__builtin_cpu_is ("corei7") < 0)
+    return -1;
+  if (__builtin_cpu_is ("nehalem") < 0)
+    return -1;
+  if (__builtin_cpu_is ("westmere") < 0)
+    return -1;
+  if (__builtin_cpu_is ("sandybridge") < 0)
+    return -1;
+  if (__builtin_cpu_is ("amdfam10h") < 0)
+    return -1;
+  if (__builtin_cpu_is ("barcelona") < 0)
+    return -1;
+  if (__builtin_cpu_is ("shanghai") < 0)
+    return -1;
+  if (__builtin_cpu_is ("istanbul") < 0)
+    return -1;
+  if (__builtin_cpu_is ("amdfam15h") < 0)
+    return -1;
+  if (__builtin_cpu_is ("bdver1") < 0)
+    return -1;
+  if (__builtin_cpu_is ("bdver2") < 0)
+    return -1;
+
+  return 0;
+}
+
+int main ()
+{
+  __builtin_cpu_init ();
+  return fn1 ();
+}
Index: gcc/config/i386/i386-builtin-types.def
===================================================================
--- gcc/config/i386/i386-builtin-types.def	(revision 185898)
+++ gcc/config/i386/i386-builtin-types.def	(working copy)
@@ -155,6 +155,7 @@ DEF_FUNCTION_TYPE (INT, V4SF)
 DEF_FUNCTION_TYPE (INT, V8QI)
 DEF_FUNCTION_TYPE (INT, V8SF)
 DEF_FUNCTION_TYPE (INT, V32QI)
+DEF_FUNCTION_TYPE (INT, PCCHAR)
 DEF_FUNCTION_TYPE (INT64, INT64)
 DEF_FUNCTION_TYPE (INT64, V2DF)
 DEF_FUNCTION_TYPE (INT64, V4SF)
Index: gcc/config/i386/i386.c
===================================================================
--- gcc/config/i386/i386.c	(revision 185898)
+++ gcc/config/i386/i386.c	(working copy)
@@ -25809,6 +25809,11 @@ enum ix86_builtins
   /* CFString built-in for darwin */
   IX86_BUILTIN_CFSTRING,
 
+  /* Builtins to get CPU type and supported features. */
+  IX86_BUILTIN_CPU_INIT,
+  IX86_BUILTIN_CPU_IS,
+  IX86_BUILTIN_CPU_SUPPORTS,
+
   IX86_BUILTIN_MAX
 };
 
@@ -27627,6 +27632,361 @@ ix86_init_mmx_sse_builtins (void)
     }
 }
 
+/* This builds the processor_model struct type defined in
+   libgcc/config/i386/i386-cpuinfo.c  */
+static tree
+build_processor_model_struct (void)
+{
+  const char *field_name[] = {"__cpu_vendor", "__cpu_type", "__cpu_subtype"};
+  tree field = NULL_TREE, field_chain = NULL_TREE;
+  int i;
+  tree type = make_node (RECORD_TYPE);
+  int num_fields = sizeof (field_name) / sizeof (const char *);
+
+  for (i = 0; i < num_fields; ++i)
+    {
+      field = build_decl (UNKNOWN_LOCATION, FIELD_DECL,
+			  get_identifier (field_name[i]), unsigned_type_node);
+      if (field_chain != NULL_TREE)
+	DECL_CHAIN (field) = field_chain;
+      field_chain = field; 
+    }
+  finish_builtin_struct (type, "__processor_model", field_chain, NULL_TREE);
+  return type;
+}
+
+/* This builds the processor_features struct type defined in
+   libgcc/config/i386/i386-cpuinfo.c  */
+
+static tree
+build_processor_features_struct (void)
+{
+  int i;
+  const char *field_name[] = {"__cpu_cmov", "__cpu_mmx", "__cpu_popcnt",
+			      "__cpu_sse", "__cpu_sse2", "__cpu_sse3",
+			      "__cpu_ssse3", "__cpu_sse4_1", "__cpu_sse4_2"};
+
+  tree field = NULL_TREE, field_chain = NULL_TREE;
+  tree type = make_node (RECORD_TYPE);
+  int num_fields = sizeof (field_name) / sizeof (const char *);
+  
+  for (i = 0; i < num_fields; i++)
+    {
+      field = build_decl (UNKNOWN_LOCATION, FIELD_DECL,
+			  get_identifier (field_name[i]), unsigned_type_node);
+      DECL_BIT_FIELD (field) = 1;
+      DECL_SIZE (field) = bitsize_one_node;
+      if (field_chain != NULL_TREE)
+	DECL_CHAIN (field) = field_chain;
+      field_chain = field;
+    }
+  finish_builtin_struct (type, "__processor_features", field_chain, NULL_TREE);
+  return type;
+}
+
+/* Returns a extern, comdat VAR_DECL of type TYPE and name NAME. */
+
+static tree
+make_var_decl (tree type, const char *name)
+{
+  tree new_decl;
+
+  new_decl = build_decl (UNKNOWN_LOCATION,
+	                 VAR_DECL,
+	  	         get_identifier(name),
+		         type);
+
+  DECL_EXTERNAL (new_decl) = 1;
+  TREE_STATIC (new_decl) = 1;
+  TREE_PUBLIC (new_decl) = 1;
+  DECL_INITIAL (new_decl) = 0;
+  DECL_ARTIFICIAL (new_decl) = 0;
+  DECL_PRESERVE_P (new_decl) = 1;
+
+  make_decl_one_only (new_decl, DECL_ASSEMBLER_NAME (new_decl));
+  assemble_variable (new_decl, 0, 0, 0);
+
+  return new_decl;
+}
+
+/* Traverses the chain of fields in STRUCT_TYPE and returns the FIELD_NUM
+   numbered field. */
+
+static tree
+get_field_from_struct (tree struct_type, int field_num)
+{
+  int i;
+  tree field = TYPE_FIELDS (struct_type);
+
+  for (i = 0; i < field_num; i++, field = DECL_CHAIN(field))
+    {
+      gcc_assert (field != NULL_TREE);
+    }
+
+  return field;
+}
+
+/* FNDECL is a __builtin_cpu_is or a __builtin_cpu_supports call that is folded
+   into an integer defined in libgcc/config/i386/i386-cpuinfo.c */
+
+static tree 
+fold_builtin_cpu (tree fndecl, tree *args)
+{
+  unsigned int i;
+  enum ix86_builtins fn_code = (enum ix86_builtins)
+				DECL_FUNCTION_CODE (fndecl);
+  tree param_string_cst = NULL;
+
+  /* This is the order of bit-fields in __processor_features in
+     i386-cpuinfo.c */
+  enum processor_features
+  {
+    F_CMOV = 0,
+    F_MMX,
+    F_POPCNT,
+    F_SSE,
+    F_SSE2,
+    F_SSE3,
+    F_SSSE3,
+    F_SSE4_1,
+    F_SSE4_2,
+    F_MAX
+  };
+
+  /* These are the values for vendor types and cpu types  and subtypes
+     in i386-cpuinfo.c.  Cpu types and subtypes should be subtracted by
+     the corresponding start value.  */
+  enum processor_model
+  {
+    M_INTEL = 1,
+    M_AMD,
+    M_CPU_TYPE_START,
+    M_INTEL_ATOM,
+    M_INTEL_CORE2,
+    M_INTEL_COREI7,
+    M_AMDFAM10H,
+    M_AMDFAM15H,
+    M_CPU_SUBTYPE_START,
+    M_INTEL_COREI7_NEHALEM,
+    M_INTEL_COREI7_WESTMERE,
+    M_INTEL_COREI7_SANDYBRIDGE,
+    M_AMDFAM10H_BARCELONA,
+    M_AMDFAM10H_SHANGHAI,
+    M_AMDFAM10H_ISTANBUL,
+    M_AMDFAM15H_BDVER1,
+    M_AMDFAM15H_BDVER2
+  };
+
+  static struct _arch_names_table
+    {
+      const char *const name;
+      const enum processor_model model;
+    }
+  const arch_names_table[] =
+    {
+      {"amd", M_AMD},
+      {"intel", M_INTEL},
+      {"atom", M_INTEL_ATOM},
+      {"core2", M_INTEL_CORE2},
+      {"corei7", M_INTEL_COREI7},
+      {"nehalem", M_INTEL_COREI7_NEHALEM},
+      {"westmere", M_INTEL_COREI7_WESTMERE},
+      {"sandybridge", M_INTEL_COREI7_SANDYBRIDGE},
+      {"amdfam10h", M_AMDFAM10H},
+      {"barcelona", M_AMDFAM10H_BARCELONA},
+      {"shanghai", M_AMDFAM10H_SHANGHAI},
+      {"istanbul", M_AMDFAM10H_ISTANBUL},
+      {"amdfam15h", M_AMDFAM15H},
+      {"bdver1", M_AMDFAM15H_BDVER1},
+      {"bdver2", M_AMDFAM15H_BDVER2},
+    };
+
+  static struct _isa_names_table
+    {
+      const char *const name;
+      const enum processor_features feature;
+    }
+  const isa_names_table[] =
+    {
+      {"cmov",   F_CMOV},
+      {"mmx",    F_MMX},
+      {"popcnt", F_POPCNT},
+      {"sse",    F_SSE},
+      {"sse2",   F_SSE2},
+      {"sse3",   F_SSE3},
+      {"ssse3",  F_SSSE3},
+      {"sse4.1", F_SSE4_1},
+      {"sse4.2", F_SSE4_2}
+    };
+
+  static tree __processor_features_type = NULL_TREE;
+  static tree __cpu_features_var = NULL_TREE;
+  static tree __processor_model_type = NULL_TREE;
+  static tree __cpu_model_var = NULL_TREE;
+  static tree field;
+  static tree which_struct;
+
+  if (__processor_features_type == NULL_TREE)
+    __processor_features_type = build_processor_features_struct ();
+
+  if (__processor_model_type == NULL_TREE)
+    __processor_model_type = build_processor_model_struct ();
+
+  if (__cpu_features_var == NULL_TREE)
+    __cpu_features_var = make_var_decl (__processor_features_type,
+					"__cpu_features");
+
+  if (__cpu_model_var == NULL_TREE)
+    __cpu_model_var = make_var_decl (__processor_model_type,
+				     "__cpu_model");
+
+  gcc_assert ((args != NULL) && (*args != NULL));
+
+  param_string_cst = *args;
+  while (param_string_cst
+	 && TREE_CODE (param_string_cst) !=  STRING_CST)
+    {
+      /* *args must be a expr that can contain other EXPRS leading to a
+	 STRING_CST.   */
+      if (!EXPR_P (param_string_cst))
+ 	{
+	  error ("Parameter to builtin must be a string constant or literal");
+	  return integer_zero_node;
+	}
+      param_string_cst = TREE_OPERAND (EXPR_CHECK (param_string_cst), 0);
+    }
+ 
+  gcc_assert (param_string_cst);
+
+  if (fn_code == IX86_BUILTIN_CPU_IS)
+    {
+      tree ref, expr;
+      unsigned int field_val = 0;
+      unsigned int NUM_ARCH_NAMES
+	= sizeof (arch_names_table) / sizeof (struct _arch_names_table);
+
+      for (i = 0; i < NUM_ARCH_NAMES; i++)
+	if (strcmp (arch_names_table[i].name,
+	    TREE_STRING_POINTER (param_string_cst)) == 0)
+	  break;
+
+      if (i == NUM_ARCH_NAMES)
+	{
+	  error ("Parameter to builtin not valid: %s",
+	         TREE_STRING_POINTER (param_string_cst));
+	  return integer_zero_node;
+	}
+
+      field = TYPE_FIELDS (__processor_model_type);
+      field_val = arch_names_table[i].model;
+
+      /* CPU types are stored in the next field.  */
+      if (field_val > M_CPU_TYPE_START
+	  && field_val < M_CPU_SUBTYPE_START)
+	{
+	  field = DECL_CHAIN (field);
+	  field_val -= M_CPU_TYPE_START;
+	}
+
+      /* CPU subtypes are stored in the last field.  */
+      if (field_val > M_CPU_SUBTYPE_START)
+	{
+	  field = DECL_CHAIN ( DECL_CHAIN (field));
+	  field_val -= M_CPU_SUBTYPE_START;
+	}
+
+      /* Get the appropriate field in __cpu_model.  */
+      ref =  build3 (COMPONENT_REF, TREE_TYPE (field), __cpu_model_var,
+		     field, NULL_TREE);
+
+      /* Check the value.  */
+      return build2 (EQ_EXPR, unsigned_type_node, ref,
+		     build_int_cstu (unsigned_type_node, field_val));
+    }
+  else if (fn_code == IX86_BUILTIN_CPU_SUPPORTS)
+    {
+      unsigned int NUM_ISA_NAMES
+	= sizeof (isa_names_table) / sizeof (struct _isa_names_table);
+      for (i = 0; i < NUM_ISA_NAMES; i++)
+	if (strcmp (isa_names_table[i].name,
+	    TREE_STRING_POINTER (param_string_cst)) == 0)
+	  break;
+
+      if (i == NUM_ISA_NAMES)
+	{
+	  error ("Parameter to builtin not valid: %s",
+	       	 TREE_STRING_POINTER (param_string_cst));
+	  return integer_zero_node;
+	}
+
+      field = get_field_from_struct (__processor_features_type,
+				     isa_names_table[i].feature);
+      which_struct = __cpu_features_var;
+      return build3 (COMPONENT_REF, TREE_TYPE (field), which_struct,
+		     field, NULL_TREE);
+    }
+
+  gcc_unreachable ();
+
+}
+
+static tree
+ix86_fold_builtin (tree fndecl, int n_args,
+		   tree *args, bool ignore ATTRIBUTE_UNUSED)
+{
+  if (DECL_BUILT_IN_CLASS (fndecl) == BUILT_IN_MD)
+    {
+      enum ix86_builtins fn_code = (enum ix86_builtins)
+				   DECL_FUNCTION_CODE (fndecl);
+      if (fn_code ==  IX86_BUILTIN_CPU_IS
+	  || fn_code == IX86_BUILTIN_CPU_SUPPORTS)
+	{
+	  gcc_assert (n_args == 1);
+          return fold_builtin_cpu (fndecl, args);
+	}
+    }
+
+  return NULL_TREE;
+}
+
+/* Make builtins to detect cpu type and features supported.  NAME is
+   the builtin name, CODE is the builtin code, and FTYPE is the function
+   type of the builtin.  */
+
+static void
+make_cpu_type_builtin (const char* name, int code,
+		       enum ix86_builtin_func_type ftype)
+{
+  tree decl;
+  tree type;
+
+  type = ix86_get_builtin_func_type (ftype);
+  decl = add_builtin_function (name, type, code, BUILT_IN_MD,
+			       NULL, NULL_TREE);
+  gcc_assert (decl != NULL_TREE);
+  ix86_builtins[(int) code] = decl;
+  TREE_READONLY (decl) = 1;
+}
+
+/* Make builtins to get CPU type and features supported.  The created
+   builtins are :
+
+   __builtin_cpu_init (), to detect cpu type and features,
+   __builtin_cpu_is ("<CPUNAME>"), to check if cpu is of type <CPUNAME>,
+   __builtin_cpu_supports ("<FEATURE>"), to check if cpu supports <FEATURE>
+   */
+
+static void
+ix86_init_platform_type_builtins (void)
+{
+  make_cpu_type_builtin ("__builtin_cpu_init", IX86_BUILTIN_CPU_INIT,
+			 INT_FTYPE_VOID);
+  make_cpu_type_builtin ("__builtin_cpu_is", IX86_BUILTIN_CPU_IS,
+			 INT_FTYPE_PCCHAR);
+  make_cpu_type_builtin ("__builtin_cpu_supports", IX86_BUILTIN_CPU_SUPPORTS,
+			 INT_FTYPE_PCCHAR);
+}
+
 /* Internal method for ix86_init_builtins.  */
 
 static void
@@ -27710,6 +28070,9 @@ ix86_init_builtins (void)
 
   ix86_init_builtin_types ();
 
+  /* Builtins to get CPU type and features. */
+  ix86_init_platform_type_builtins ();
+
   /* TFmode support builtins.  */
   def_builtin_const (0, "__builtin_infq",
 		     FLOAT128_FTYPE_VOID, IX86_BUILTIN_INFQ);
@@ -29328,6 +29691,28 @@ ix86_expand_builtin (tree exp, rtx target, rtx sub
   enum machine_mode mode0, mode1, mode2, mode3, mode4;
   unsigned int fcode = DECL_FUNCTION_CODE (fndecl);
 
+  /* For CPU builtins that can be folded, fold first and expand the fold.  */
+  switch (fcode)
+    {
+    case IX86_BUILTIN_CPU_INIT:
+      {
+	/* Make it call __cpu_indicator_init in libgcc. */
+	tree call_expr, fndecl, type;
+        type = build_function_type_list (integer_type_node, NULL_TREE); 
+	fndecl = build_fn_decl ("__cpu_indicator_init", type);
+	call_expr = build_call_expr (fndecl, 0); 
+	return expand_expr (call_expr, target, mode, EXPAND_NORMAL);
+      }
+    case IX86_BUILTIN_CPU_IS:
+    case IX86_BUILTIN_CPU_SUPPORTS:
+      {
+	tree arg0 = CALL_EXPR_ARG (exp, 0);
+	tree fold_expr = fold_builtin_cpu (fndecl, &arg0);
+	gcc_assert (fold_expr != NULL_TREE);
+        return expand_expr (fold_expr, target, mode, EXPAND_NORMAL);
+      }
+    }
+
   /* Determine whether the builtin function is available under the current ISA.
      Originally the builtin was not created if it wasn't applicable to the
      current ISA based on the command line switches.  With function specific
@@ -39058,6 +39443,9 @@ ix86_autovectorize_vector_sizes (void)
 #undef TARGET_BUILD_BUILTIN_VA_LIST
 #define TARGET_BUILD_BUILTIN_VA_LIST ix86_build_builtin_va_list
 
+#undef TARGET_FOLD_BUILTIN
+#define TARGET_FOLD_BUILTIN ix86_fold_builtin
+
 #undef TARGET_ENUM_VA_LIST_P
 #define TARGET_ENUM_VA_LIST_P ix86_enum_va_list
 

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Richard Guenther]

On Sat, Mar 31, 2012 at 1:03 AM, Sriraman Tallam <tmsriram@google.com> wrote:
> On Fri, Mar 30, 2012 at 5:47 AM, Michael Matz <matz@suse.de> wrote:
>> Hi,
>>
>> On Thu, 29 Mar 2012, Sriraman Tallam wrote:
>>
>>> +struct __processor_model
>>> +{
>>> +  /* Vendor. */
>>> +  unsigned int __cpu_is_amd : 1;
>>> +  unsigned int __cpu_is_intel : 1;
>>> +  /* CPU type. */
>>> +  unsigned int __cpu_is_intel_atom : 1;
>>> +  unsigned int __cpu_is_intel_core2 : 1;
>>> +  unsigned int __cpu_is_intel_corei7 : 1;
>>> +  unsigned int __cpu_is_intel_corei7_nehalem : 1;
>>> +  unsigned int __cpu_is_intel_corei7_westmere : 1;
>>> +  unsigned int __cpu_is_intel_corei7_sandybridge : 1;
>>> +  unsigned int __cpu_is_amdfam10h : 1;
>>> +  unsigned int __cpu_is_amdfam10h_barcelona : 1;
>>> +  unsigned int __cpu_is_amdfam10h_shanghai : 1;
>>> +  unsigned int __cpu_is_amdfam10h_istanbul : 1;
>>> +  unsigned int __cpu_is_amdfam15h_bdver1 : 1;
>>> +  unsigned int __cpu_is_amdfam15h_bdver2 : 1;
>>> +} __cpu_model;
>>
>> It doesn't make sense for the model to be a bitfield, a processor will
>> have only ever exactly one model.  Just make it an enum or even just an
>> int.
>
> Not entirely true, nehalem and corei7 can be both set. However, I
> modified this by dividing it into types and sub types and then did
> what you said.

Uh...  then I suppose you need to document somewhere what names
match to what cpuid family/model (supposedly thats where your two-layer
hierarchy comes from, which incidentially misses one layer, the vendor?)

Richard.

>       * config/i386/i386.c (build_processor_features_struct): New function.
>       (build_processor_model_struct): New function.
>       (make_var_decl): New function.
>       (get_field_from_struct): New function.
>       (fold_builtin_target): New function.
>       (ix86_fold_builtin): New function.
>       (ix86_expand_builtin): Expand new builtins by folding them.
>       (make_cpu_type_builtin): New functions.
>       (ix86_init_platform_type_builtins): Make the new builtins.
>       (ix86_init_builtins): Make new builtins to detect CPU type.
>       (TARGET_FOLD_BUILTIN): New macro.
>       (IX86_BUILTIN_CPU_INIT): New enum value.
>       (IX86_BUILTIN_CPU_IS): New enum value.
>       (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>       * config/i386/i386-builtin-types.def: New function type.
>       * testsuite/gcc.target/builtin_target.c: New testcase.
>
>       * libgcc/config/i386/i386-cpuinfo.c: New file.
>       * libgcc/config/i386/t-cpuinfo: New file.
>       * libgcc/config.host: Include t-cpuinfo.
>       * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
>       and __cpu_features.
>
> Patch available for review here:
> http://codereview.appspot.com/5754058
>
> Thanks,
> -Sri.
>
>
>>
>>
>> Ciao,
>> Michael.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

On Mon, Apr 2, 2012 at 5:38 AM, Richard Guenther
<richard.guenther@gmail.com> wrote:
> On Sat, Mar 31, 2012 at 1:03 AM, Sriraman Tallam <tmsriram@google.com> wrote:
>> On Fri, Mar 30, 2012 at 5:47 AM, Michael Matz <matz@suse.de> wrote:
>>> Hi,
>>>
>>> On Thu, 29 Mar 2012, Sriraman Tallam wrote:
>>>
>>>> +struct __processor_model
>>>> +{
>>>> +  /* Vendor. */
>>>> +  unsigned int __cpu_is_amd : 1;
>>>> +  unsigned int __cpu_is_intel : 1;
>>>> +  /* CPU type. */
>>>> +  unsigned int __cpu_is_intel_atom : 1;
>>>> +  unsigned int __cpu_is_intel_core2 : 1;
>>>> +  unsigned int __cpu_is_intel_corei7 : 1;
>>>> +  unsigned int __cpu_is_intel_corei7_nehalem : 1;
>>>> +  unsigned int __cpu_is_intel_corei7_westmere : 1;
>>>> +  unsigned int __cpu_is_intel_corei7_sandybridge : 1;
>>>> +  unsigned int __cpu_is_amdfam10h : 1;
>>>> +  unsigned int __cpu_is_amdfam10h_barcelona : 1;
>>>> +  unsigned int __cpu_is_amdfam10h_shanghai : 1;
>>>> +  unsigned int __cpu_is_amdfam10h_istanbul : 1;
>>>> +  unsigned int __cpu_is_amdfam15h_bdver1 : 1;
>>>> +  unsigned int __cpu_is_amdfam15h_bdver2 : 1;
>>>> +} __cpu_model;
>>>
>>> It doesn't make sense for the model to be a bitfield, a processor will
>>> have only ever exactly one model.  Just make it an enum or even just an
>>> int.
>>
>> Not entirely true, nehalem and corei7 can be both set. However, I
>> modified this by dividing it into types and sub types and then did
>> what you said.
>
> Uh...  then I suppose you need to document somewhere what names
> match to what cpuid family/model (supposedly thats where your two-layer
> hierarchy comes from, which incidentially misses one layer, the vendor?)

Yes, it is 3-layer. Vendor, family and model.

What is the right place to document these?

>
> Richard.
>
>>       * config/i386/i386.c (build_processor_features_struct): New function.
>>       (build_processor_model_struct): New function.
>>       (make_var_decl): New function.
>>       (get_field_from_struct): New function.
>>       (fold_builtin_target): New function.
>>       (ix86_fold_builtin): New function.
>>       (ix86_expand_builtin): Expand new builtins by folding them.
>>       (make_cpu_type_builtin): New functions.
>>       (ix86_init_platform_type_builtins): Make the new builtins.
>>       (ix86_init_builtins): Make new builtins to detect CPU type.
>>       (TARGET_FOLD_BUILTIN): New macro.
>>       (IX86_BUILTIN_CPU_INIT): New enum value.
>>       (IX86_BUILTIN_CPU_IS): New enum value.
>>       (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>>       * config/i386/i386-builtin-types.def: New function type.
>>       * testsuite/gcc.target/builtin_target.c: New testcase.
>>
>>       * libgcc/config/i386/i386-cpuinfo.c: New file.
>>       * libgcc/config/i386/t-cpuinfo: New file.
>>       * libgcc/config.host: Include t-cpuinfo.
>>       * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
>>       and __cpu_features.
>>
>> Patch available for review here:
>> http://codereview.appspot.com/5754058
>>
>> Thanks,
>> -Sri.
>>
>>
>>>
>>>
>>> Ciao,
>>> Michael.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

[-- Attachment #1: Type: text/plain, Size: 3176 bytes --]

On Mon, Apr 2, 2012 at 5:38 AM, Richard Guenther
<richard.guenther@gmail.com> wrote:
> On Sat, Mar 31, 2012 at 1:03 AM, Sriraman Tallam <tmsriram@google.com> wrote:
>> On Fri, Mar 30, 2012 at 5:47 AM, Michael Matz <matz@suse.de> wrote:
>>> Hi,
>>>
>>> On Thu, 29 Mar 2012, Sriraman Tallam wrote:
>>>
>>>> +struct __processor_model
>>>> +{
>>>> +  /* Vendor. */
>>>> +  unsigned int __cpu_is_amd : 1;
>>>> +  unsigned int __cpu_is_intel : 1;
>>>> +  /* CPU type. */
>>>> +  unsigned int __cpu_is_intel_atom : 1;
>>>> +  unsigned int __cpu_is_intel_core2 : 1;
>>>> +  unsigned int __cpu_is_intel_corei7 : 1;
>>>> +  unsigned int __cpu_is_intel_corei7_nehalem : 1;
>>>> +  unsigned int __cpu_is_intel_corei7_westmere : 1;
>>>> +  unsigned int __cpu_is_intel_corei7_sandybridge : 1;
>>>> +  unsigned int __cpu_is_amdfam10h : 1;
>>>> +  unsigned int __cpu_is_amdfam10h_barcelona : 1;
>>>> +  unsigned int __cpu_is_amdfam10h_shanghai : 1;
>>>> +  unsigned int __cpu_is_amdfam10h_istanbul : 1;
>>>> +  unsigned int __cpu_is_amdfam15h_bdver1 : 1;
>>>> +  unsigned int __cpu_is_amdfam15h_bdver2 : 1;
>>>> +} __cpu_model;
>>>
>>> It doesn't make sense for the model to be a bitfield, a processor will
>>> have only ever exactly one model.  Just make it an enum or even just an
>>> int.
>>
>> Not entirely true, nehalem and corei7 can be both set. However, I
>> modified this by dividing it into types and sub types and then did
>> what you said.
>
> Uh...  then I suppose you need to document somewhere what names
> match to what cpuid family/model (supposedly thats where your two-layer
> hierarchy comes from, which incidentially misses one layer, the vendor?)

Added documentation to extend.texi

Patch available for review here:
http://codereview.appspot.com/5754058

Thanks,
-Sri.


>
> Richard.
>
>>       * config/i386/i386.c (build_processor_features_struct): New function.
>>       (build_processor_model_struct): New function.
>>       (make_var_decl): New function.
>>       (get_field_from_struct): New function.
>>       (fold_builtin_target): New function.
>>       (ix86_fold_builtin): New function.
>>       (ix86_expand_builtin): Expand new builtins by folding them.
>>       (make_cpu_type_builtin): New functions.
>>       (ix86_init_platform_type_builtins): Make the new builtins.
>>       (ix86_init_builtins): Make new builtins to detect CPU type.
>>       (TARGET_FOLD_BUILTIN): New macro.
>>       (IX86_BUILTIN_CPU_INIT): New enum value.
>>       (IX86_BUILTIN_CPU_IS): New enum value.
>>       (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>>       * config/i386/i386-builtin-types.def: New function type.
>>       * testsuite/gcc.target/builtin_target.c: New testcase.
>>
>>       * libgcc/config/i386/i386-cpuinfo.c: New file.
>>       * libgcc/config/i386/t-cpuinfo: New file.
>>       * libgcc/config.host: Include t-cpuinfo.
>>       * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
>>       and __cpu_features.
>>
>> Patch available for review here:
>> http://codereview.appspot.com/5754058
>>
>> Thanks,
>> -Sri.
>>
>>
>>>
>>>
>>> Ciao,
>>> Michael.

[-- Attachment #2: cpu_runtime_patch.txt --]
[-- Type: text/plain, Size: 28968 bytes --]


       * config/i386/i386.c (build_processor_features_struct): New function.
       (build_processor_model_struct): New function.
       (make_var_decl): New function.
       (get_field_from_struct): New function.
       (fold_builtin_target): New function.
       (ix86_fold_builtin): New function.
       (ix86_expand_builtin): Expand new builtins by folding them.
       (make_cpu_type_builtin): New functions.
       (ix86_init_platform_type_builtins): Make the new builtins.
       (ix86_init_builtins): Make new builtins to detect CPU type.
       (TARGET_FOLD_BUILTIN): New macro.
       (IX86_BUILTIN_CPU_INIT): New enum value.
       (IX86_BUILTIN_CPU_IS): New enum value.
       (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
       * config/i386/i386-builtin-types.def: New function type.
       * testsuite/gcc.target/builtin_target.c: New testcase.
       * doc/extend.texi: Document builtins.

       * libgcc/config/i386/i386-cpuinfo.c: New file.
       * libgcc/config/i386/t-cpuinfo: New file.
       * libgcc/config.host: Include t-cpuinfo.
       * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
       and __cpu_features.


Index: libgcc/config.host
===================================================================
--- libgcc/config.host	(revision 185898)
+++ libgcc/config.host	(working copy)
@@ -1130,7 +1130,7 @@ i[34567]86-*-linux* | x86_64-*-linux* | \
   i[34567]86-*-kfreebsd*-gnu | x86_64-*-kfreebsd*-gnu | \
   i[34567]86-*-knetbsd*-gnu | \
   i[34567]86-*-gnu*)
-	tmake_file="${tmake_file} t-tls i386/t-linux"
+	tmake_file="${tmake_file} t-tls i386/t-linux i386/t-cpuinfo"
 	if test "$libgcc_cv_cfi" = "yes"; then
 		tmake_file="${tmake_file} t-stack i386/t-stack-i386"
 	fi
Index: libgcc/config/i386/t-cpuinfo
===================================================================
--- libgcc/config/i386/t-cpuinfo	(revision 0)
+++ libgcc/config/i386/t-cpuinfo	(revision 0)
@@ -0,0 +1 @@
+LIB2ADD += $(srcdir)/config/i386/i386-cpuinfo.c
Index: libgcc/config/i386/i386-cpuinfo.c
===================================================================
--- libgcc/config/i386/i386-cpuinfo.c	(revision 0)
+++ libgcc/config/i386/i386-cpuinfo.c	(revision 0)
@@ -0,0 +1,298 @@
+/* Get CPU type and Features for x86 processors.
+   Copyright (C) 2012 Free Software Foundation, Inc.
+   Contributed by Sriraman Tallam (tmsriram@google.com)
+
+This file is part of GCC.
+
+GCC is free software; you can redistribute it and/or modify it under
+the terms of the GNU General Public License as published by the Free
+Software Foundation; either version 3, or (at your option) any later
+version.
+
+GCC is distributed in the hope that it will be useful, but WITHOUT ANY
+WARRANTY; without even the implied warranty of MERCHANTABILITY or
+FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+for more details.
+
+You should have received a copy of the GNU General Public License
+along with GCC; see the file COPYING3.  If not see
+<http://www.gnu.org/licenses/>. */
+
+#include "cpuid.h"
+#include "tsystem.h"
+
+int __cpu_indicator_init (void) __attribute__ ((constructor (101)));
+
+enum vendor_signatures
+{
+  SIG_INTEL =	0x756e6547 /* Genu */,
+  SIG_AMD =	0x68747541 /* Auth */
+};
+
+/* ISA Features supported. */
+
+struct __processor_features
+{
+  unsigned int __cpu_cmov : 1;
+  unsigned int __cpu_mmx : 1;
+  unsigned int __cpu_popcnt : 1;
+  unsigned int __cpu_sse : 1;
+  unsigned int __cpu_sse2 : 1;
+  unsigned int __cpu_sse3 : 1;
+  unsigned int __cpu_ssse3 : 1;
+  unsigned int __cpu_sse4_1 : 1;
+  unsigned int __cpu_sse4_2 : 1;
+} __cpu_features;
+
+/* Processor Vendor and Models. */
+
+enum processor_vendor
+{
+  VENDOR_INTEL = 1,
+  VENDOR_AMD,
+  VENDOR_MAX
+};
+
+enum processor_types
+{
+  INTEL_ATOM = 1,
+  INTEL_CORE2,
+  INTEL_COREI7,
+  AMDFAM10H,
+  AMDFAM15H,
+  CPU_TYPE_MAX
+};
+
+enum processor_subtypes
+{
+  INTEL_COREI7_NEHALEM,
+  INTEL_COREI7_WESTMERE,
+  INTEL_COREI7_SANDYBRIDGE,
+  AMDFAM10H_BARCELONA,
+  AMDFAM10H_SHANGHAI,
+  AMDFAM10H_ISTANBUL,
+  AMDFAM15H_BDVER1,
+  AMDFAM15H_BDVER2,
+  CPU_SUBTYPE_MAX
+};
+
+struct __processor_model
+{
+  unsigned int __cpu_vendor;
+  unsigned int __cpu_type;
+  unsigned int __cpu_subtype;
+} __cpu_model;
+
+/* Get the specific type of AMD CPU.  */
+
+static void
+get_amd_cpu (unsigned int family, unsigned int model)
+{
+  __cpu_model.__cpu_vendor = VENDOR_AMD;
+  switch (family)
+    {
+    /* AMD Family 10h.  */
+    case 0x10:
+      switch (model)
+	{
+	case 0x2:
+	  /* Barcelona.  */
+	  __cpu_model.__cpu_type = AMDFAM10H;
+	  __cpu_model.__cpu_subtype = AMDFAM10H_BARCELONA;
+	  break;
+	case 0x4:
+	  /* Shanghai.  */
+	  __cpu_model.__cpu_type = AMDFAM10H;
+	  __cpu_model.__cpu_subtype = AMDFAM10H_SHANGHAI;
+	  break;
+	case 0x8:
+	  /* Istanbul.  */
+	  __cpu_model.__cpu_type = AMDFAM10H;
+	  __cpu_model.__cpu_subtype = AMDFAM10H_ISTANBUL;
+	  break;
+	default:
+	  break;
+	}
+      break;
+    /* AMD Family 15h.  */
+    case 0x15:
+      __cpu_model.__cpu_type = AMDFAM15H;
+      /* Bulldozer version 1.  */
+      if ( model <= 0xf)
+	__cpu_model.__cpu_subtype = AMDFAM15H_BDVER1;
+      /* Bulldozer version 2.  */
+      if (model >= 0x10 && model <= 0x1f)
+	__cpu_model.__cpu_subtype = AMDFAM15H_BDVER2;
+      break;
+    default:
+      break;
+    }
+}
+
+/* Get the specific type of Intel CPU.  */
+
+static void
+get_intel_cpu (unsigned int family, unsigned int model, unsigned int brand_id)
+{
+  __cpu_model.__cpu_vendor = VENDOR_INTEL;
+  /* Parse family and model only if brand ID is 0. */
+  if (brand_id == 0)
+    {
+      switch (family)
+	{
+	case 0x5:
+	  /* Pentium.  */
+	  break;
+	case 0x6:
+	  switch (model)
+	    {
+	    case 0x1c:
+	    case 0x26:
+	      /* Atom.  */
+	      __cpu_model.__cpu_type = INTEL_ATOM;
+	      break;
+	    case 0x1a:
+	    case 0x1e:
+	    case 0x1f:
+	    case 0x2e:
+	      /* Nehalem.  */
+	      __cpu_model.__cpu_type = INTEL_COREI7;
+	      __cpu_model.__cpu_subtype = INTEL_COREI7_NEHALEM;
+	      break;
+	    case 0x25:
+	    case 0x2c:
+	    case 0x2f:
+	      /* Westmere.  */
+	      __cpu_model.__cpu_type = INTEL_COREI7;
+	      __cpu_model.__cpu_subtype = INTEL_COREI7_WESTMERE;
+	      break;
+	    case 0x2a:
+	      /* Sandy Bridge.  */
+	      __cpu_model.__cpu_type = INTEL_COREI7;
+	      __cpu_model.__cpu_subtype = INTEL_COREI7_SANDYBRIDGE;
+	      break;
+	    case 0x17:
+	    case 0x1d:
+	      /* Penryn.  */
+	    case 0x0f:
+	      /* Merom.  */
+	      __cpu_model.__cpu_type = INTEL_CORE2;
+	      break;
+	    default:
+	      break;
+	    }
+	  break;
+	default:
+	  /* We have no idea.  */
+	  break;
+	}
+    }
+}	             	
+
+static void
+get_available_features (unsigned int ecx, unsigned int edx)
+{
+  __cpu_features.__cpu_cmov = (edx & bit_CMOV) ? 1 : 0;
+  __cpu_features.__cpu_mmx = (edx & bit_MMX) ? 1 : 0;
+  __cpu_features.__cpu_sse = (edx & bit_SSE) ? 1 : 0;
+  __cpu_features.__cpu_sse2 = (edx & bit_SSE2) ? 1 : 0;
+  __cpu_features.__cpu_popcnt = (ecx & bit_POPCNT) ? 1 : 0;
+  __cpu_features.__cpu_sse3 = (ecx & bit_SSE3) ? 1 : 0;
+  __cpu_features.__cpu_ssse3 = (ecx & bit_SSSE3) ? 1 : 0;
+  __cpu_features.__cpu_sse4_1 = (ecx & bit_SSE4_1) ? 1 : 0;
+  __cpu_features.__cpu_sse4_2 = (ecx & bit_SSE4_2) ? 1 : 0;
+}
+
+/* A noinline function calling __get_cpuid. Having many calls to
+   cpuid in one function in 32-bit mode causes GCC to complain:
+   "can;t find a register in class CLOBBERED_REGS".  This is
+   related to PR rtl-optimization 44174. */
+
+static int __attribute__ ((noinline))
+__get_cpuid_output (unsigned int __level,
+		    unsigned int *__eax, unsigned int *__ebx,
+		    unsigned int *__ecx, unsigned int *__edx)
+{
+  return __get_cpuid (__level, __eax, __ebx, __ecx, __edx);
+}
+
+
+/* A constructor function that is sets __cpu_model and __cpu_features with
+   the right values.  This needs to run only once.  This constructor is
+   given the highest priority and it should run before constructors without
+   the priority set.  However, it still runs after ifunc initializers and
+   needs to be called explicitly there.  */
+
+int __attribute__ ((constructor (101)))
+__cpu_indicator_init (void)
+{
+  unsigned int eax, ebx, ecx, edx;
+
+  int max_level = 5;
+  unsigned int vendor;
+  unsigned int model, family, brand_id;
+  unsigned int extended_model, extended_family;
+  static int called = 0;
+
+  /* This function needs to run just once.  */
+  if (called)
+    return 0;
+  else
+    called = 1;
+
+  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
+  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
+    return -1;
+
+  vendor = ebx;
+  max_level = eax;
+
+  if (max_level < 1)
+    return -1;
+
+  if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
+    return -1;
+
+  model = (eax >> 4) & 0x0f;
+  family = (eax >> 8) & 0x0f;
+  brand_id = ebx & 0xff;
+  extended_model = (eax >> 12) & 0xf0;
+  extended_family = (eax >> 20) & 0xff;
+
+  if (vendor == SIG_INTEL)
+    {
+      /* Adjust model and family for Intel CPUS. */
+      if (family == 0x0f)
+	{
+	  family += extended_family;
+	  model += extended_model;
+	}
+      else if (family == 0x06)
+	model += extended_model;
+
+      /* Get CPU type.  */
+      get_intel_cpu (family, model, brand_id);
+    }
+
+  if (vendor == SIG_AMD)
+    {
+      /* Adjust model and family for AMD CPUS. */
+      if (family == 0x0f)
+	{
+	  family += extended_family;
+	  model += (extended_model << 4);
+	}
+
+      /* Get CPU type.  */
+      get_amd_cpu (family, model);
+    }
+
+  /* Find available features. */
+  get_available_features (ecx, edx);
+
+  gcc_assert (__cpu_model.__cpu_vendor < VENDOR_MAX);
+  gcc_assert (__cpu_model.__cpu_type < CPU_TYPE_MAX);
+  gcc_assert (__cpu_model.__cpu_subtype < CPU_SUBTYPE_MAX);
+
+  return 0;
+}
Index: libgcc/config/i386/libgcc-glibc.ver
===================================================================
--- libgcc/config/i386/libgcc-glibc.ver	(revision 185898)
+++ libgcc/config/i386/libgcc-glibc.ver	(working copy)
@@ -147,6 +147,11 @@ GCC_4.3.0 {
   __trunctfxf2
   __unordtf2
 }
+
+GCC_4.8.0 {
+  __cpu_model
+  __cpu_features
+}
 %else
 GCC_4.4.0 {
   __addtf3
@@ -183,4 +188,9 @@ GCC_4.4.0 {
 GCC_4.5.0 {
   __extendxftf2
 }
+
+GCC_4.8.0 {
+  __cpu_model
+  __cpu_features
+}
 %endif
Index: gcc/testsuite/gcc.target/i386/builtin_target.c
===================================================================
--- gcc/testsuite/gcc.target/i386/builtin_target.c	(revision 0)
+++ gcc/testsuite/gcc.target/i386/builtin_target.c	(revision 0)
@@ -0,0 +1,65 @@
+/* This test checks if the __builtin_cpu_is and __builtin_cpu_supports calls
+   are recognized. */
+
+/* { dg-do run } */
+
+int
+fn1 ()
+{
+  if (__builtin_cpu_supports ("cmov") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("mmx") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("popcnt") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse2") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse3") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("ssse3") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse4.1") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse4.2") < 0)
+    return -1;
+  if (__builtin_cpu_is ("amd") < 0)
+    return -1;
+  if (__builtin_cpu_is ("intel") < 0)
+    return -1;
+  if (__builtin_cpu_is ("atom") < 0)
+    return -1;
+  if (__builtin_cpu_is ("core2") < 0)
+    return -1;
+  if (__builtin_cpu_is ("corei7") < 0)
+    return -1;
+  if (__builtin_cpu_is ("nehalem") < 0)
+    return -1;
+  if (__builtin_cpu_is ("westmere") < 0)
+    return -1;
+  if (__builtin_cpu_is ("sandybridge") < 0)
+    return -1;
+  if (__builtin_cpu_is ("amdfam10h") < 0)
+    return -1;
+  if (__builtin_cpu_is ("barcelona") < 0)
+    return -1;
+  if (__builtin_cpu_is ("shanghai") < 0)
+    return -1;
+  if (__builtin_cpu_is ("istanbul") < 0)
+    return -1;
+  if (__builtin_cpu_is ("amdfam15h") < 0)
+    return -1;
+  if (__builtin_cpu_is ("bdver1") < 0)
+    return -1;
+  if (__builtin_cpu_is ("bdver2") < 0)
+    return -1;
+
+  return 0;
+}
+
+int main ()
+{
+  __builtin_cpu_init ();
+  return fn1 ();
+}
Index: gcc/config/i386/i386-builtin-types.def
===================================================================
--- gcc/config/i386/i386-builtin-types.def	(revision 185898)
+++ gcc/config/i386/i386-builtin-types.def	(working copy)
@@ -155,6 +155,7 @@ DEF_FUNCTION_TYPE (INT, V4SF)
 DEF_FUNCTION_TYPE (INT, V8QI)
 DEF_FUNCTION_TYPE (INT, V8SF)
 DEF_FUNCTION_TYPE (INT, V32QI)
+DEF_FUNCTION_TYPE (INT, PCCHAR)
 DEF_FUNCTION_TYPE (INT64, INT64)
 DEF_FUNCTION_TYPE (INT64, V2DF)
 DEF_FUNCTION_TYPE (INT64, V4SF)
Index: gcc/config/i386/i386.c
===================================================================
--- gcc/config/i386/i386.c	(revision 185898)
+++ gcc/config/i386/i386.c	(working copy)
@@ -25809,6 +25809,11 @@ enum ix86_builtins
   /* CFString built-in for darwin */
   IX86_BUILTIN_CFSTRING,
 
+  /* Builtins to get CPU type and supported features. */
+  IX86_BUILTIN_CPU_INIT,
+  IX86_BUILTIN_CPU_IS,
+  IX86_BUILTIN_CPU_SUPPORTS,
+
   IX86_BUILTIN_MAX
 };
 
@@ -27627,6 +27632,361 @@ ix86_init_mmx_sse_builtins (void)
     }
 }
 
+/* This builds the processor_model struct type defined in
+   libgcc/config/i386/i386-cpuinfo.c  */
+static tree
+build_processor_model_struct (void)
+{
+  const char *field_name[] = {"__cpu_vendor", "__cpu_type", "__cpu_subtype"};
+  tree field = NULL_TREE, field_chain = NULL_TREE;
+  int i;
+  tree type = make_node (RECORD_TYPE);
+  int num_fields = sizeof (field_name) / sizeof (const char *);
+
+  for (i = 0; i < num_fields; ++i)
+    {
+      field = build_decl (UNKNOWN_LOCATION, FIELD_DECL,
+			  get_identifier (field_name[i]), unsigned_type_node);
+      if (field_chain != NULL_TREE)
+	DECL_CHAIN (field) = field_chain;
+      field_chain = field; 
+    }
+  finish_builtin_struct (type, "__processor_model", field_chain, NULL_TREE);
+  return type;
+}
+
+/* This builds the processor_features struct type defined in
+   libgcc/config/i386/i386-cpuinfo.c  */
+
+static tree
+build_processor_features_struct (void)
+{
+  int i;
+  const char *field_name[] = {"__cpu_cmov", "__cpu_mmx", "__cpu_popcnt",
+			      "__cpu_sse", "__cpu_sse2", "__cpu_sse3",
+			      "__cpu_ssse3", "__cpu_sse4_1", "__cpu_sse4_2"};
+
+  tree field = NULL_TREE, field_chain = NULL_TREE;
+  tree type = make_node (RECORD_TYPE);
+  int num_fields = sizeof (field_name) / sizeof (const char *);
+  
+  for (i = 0; i < num_fields; i++)
+    {
+      field = build_decl (UNKNOWN_LOCATION, FIELD_DECL,
+			  get_identifier (field_name[i]), unsigned_type_node);
+      DECL_BIT_FIELD (field) = 1;
+      DECL_SIZE (field) = bitsize_one_node;
+      if (field_chain != NULL_TREE)
+	DECL_CHAIN (field) = field_chain;
+      field_chain = field;
+    }
+  finish_builtin_struct (type, "__processor_features", field_chain, NULL_TREE);
+  return type;
+}
+
+/* Returns a extern, comdat VAR_DECL of type TYPE and name NAME. */
+
+static tree
+make_var_decl (tree type, const char *name)
+{
+  tree new_decl;
+
+  new_decl = build_decl (UNKNOWN_LOCATION,
+	                 VAR_DECL,
+	  	         get_identifier(name),
+		         type);
+
+  DECL_EXTERNAL (new_decl) = 1;
+  TREE_STATIC (new_decl) = 1;
+  TREE_PUBLIC (new_decl) = 1;
+  DECL_INITIAL (new_decl) = 0;
+  DECL_ARTIFICIAL (new_decl) = 0;
+  DECL_PRESERVE_P (new_decl) = 1;
+
+  make_decl_one_only (new_decl, DECL_ASSEMBLER_NAME (new_decl));
+  assemble_variable (new_decl, 0, 0, 0);
+
+  return new_decl;
+}
+
+/* Traverses the chain of fields in STRUCT_TYPE and returns the FIELD_NUM
+   numbered field. */
+
+static tree
+get_field_from_struct (tree struct_type, int field_num)
+{
+  int i;
+  tree field = TYPE_FIELDS (struct_type);
+
+  for (i = 0; i < field_num; i++, field = DECL_CHAIN(field))
+    {
+      gcc_assert (field != NULL_TREE);
+    }
+
+  return field;
+}
+
+/* FNDECL is a __builtin_cpu_is or a __builtin_cpu_supports call that is folded
+   into an integer defined in libgcc/config/i386/i386-cpuinfo.c */
+
+static tree 
+fold_builtin_cpu (tree fndecl, tree *args)
+{
+  unsigned int i;
+  enum ix86_builtins fn_code = (enum ix86_builtins)
+				DECL_FUNCTION_CODE (fndecl);
+  tree param_string_cst = NULL;
+
+  /* This is the order of bit-fields in __processor_features in
+     i386-cpuinfo.c */
+  enum processor_features
+  {
+    F_CMOV = 0,
+    F_MMX,
+    F_POPCNT,
+    F_SSE,
+    F_SSE2,
+    F_SSE3,
+    F_SSSE3,
+    F_SSE4_1,
+    F_SSE4_2,
+    F_MAX
+  };
+
+  /* These are the values for vendor types and cpu types  and subtypes
+     in i386-cpuinfo.c.  Cpu types and subtypes should be subtracted by
+     the corresponding start value.  */
+  enum processor_model
+  {
+    M_INTEL = 1,
+    M_AMD,
+    M_CPU_TYPE_START,
+    M_INTEL_ATOM,
+    M_INTEL_CORE2,
+    M_INTEL_COREI7,
+    M_AMDFAM10H,
+    M_AMDFAM15H,
+    M_CPU_SUBTYPE_START,
+    M_INTEL_COREI7_NEHALEM,
+    M_INTEL_COREI7_WESTMERE,
+    M_INTEL_COREI7_SANDYBRIDGE,
+    M_AMDFAM10H_BARCELONA,
+    M_AMDFAM10H_SHANGHAI,
+    M_AMDFAM10H_ISTANBUL,
+    M_AMDFAM15H_BDVER1,
+    M_AMDFAM15H_BDVER2
+  };
+
+  static struct _arch_names_table
+    {
+      const char *const name;
+      const enum processor_model model;
+    }
+  const arch_names_table[] =
+    {
+      {"amd", M_AMD},
+      {"intel", M_INTEL},
+      {"atom", M_INTEL_ATOM},
+      {"core2", M_INTEL_CORE2},
+      {"corei7", M_INTEL_COREI7},
+      {"nehalem", M_INTEL_COREI7_NEHALEM},
+      {"westmere", M_INTEL_COREI7_WESTMERE},
+      {"sandybridge", M_INTEL_COREI7_SANDYBRIDGE},
+      {"amdfam10h", M_AMDFAM10H},
+      {"barcelona", M_AMDFAM10H_BARCELONA},
+      {"shanghai", M_AMDFAM10H_SHANGHAI},
+      {"istanbul", M_AMDFAM10H_ISTANBUL},
+      {"amdfam15h", M_AMDFAM15H},
+      {"bdver1", M_AMDFAM15H_BDVER1},
+      {"bdver2", M_AMDFAM15H_BDVER2},
+    };
+
+  static struct _isa_names_table
+    {
+      const char *const name;
+      const enum processor_features feature;
+    }
+  const isa_names_table[] =
+    {
+      {"cmov",   F_CMOV},
+      {"mmx",    F_MMX},
+      {"popcnt", F_POPCNT},
+      {"sse",    F_SSE},
+      {"sse2",   F_SSE2},
+      {"sse3",   F_SSE3},
+      {"ssse3",  F_SSSE3},
+      {"sse4.1", F_SSE4_1},
+      {"sse4.2", F_SSE4_2}
+    };
+
+  static tree __processor_features_type = NULL_TREE;
+  static tree __cpu_features_var = NULL_TREE;
+  static tree __processor_model_type = NULL_TREE;
+  static tree __cpu_model_var = NULL_TREE;
+  static tree field;
+  static tree which_struct;
+
+  if (__processor_features_type == NULL_TREE)
+    __processor_features_type = build_processor_features_struct ();
+
+  if (__processor_model_type == NULL_TREE)
+    __processor_model_type = build_processor_model_struct ();
+
+  if (__cpu_features_var == NULL_TREE)
+    __cpu_features_var = make_var_decl (__processor_features_type,
+					"__cpu_features");
+
+  if (__cpu_model_var == NULL_TREE)
+    __cpu_model_var = make_var_decl (__processor_model_type,
+				     "__cpu_model");
+
+  gcc_assert ((args != NULL) && (*args != NULL));
+
+  param_string_cst = *args;
+  while (param_string_cst
+	 && TREE_CODE (param_string_cst) !=  STRING_CST)
+    {
+      /* *args must be a expr that can contain other EXPRS leading to a
+	 STRING_CST.   */
+      if (!EXPR_P (param_string_cst))
+ 	{
+	  error ("Parameter to builtin must be a string constant or literal");
+	  return integer_zero_node;
+	}
+      param_string_cst = TREE_OPERAND (EXPR_CHECK (param_string_cst), 0);
+    }
+ 
+  gcc_assert (param_string_cst);
+
+  if (fn_code == IX86_BUILTIN_CPU_IS)
+    {
+      tree ref, expr;
+      unsigned int field_val = 0;
+      unsigned int NUM_ARCH_NAMES
+	= sizeof (arch_names_table) / sizeof (struct _arch_names_table);
+
+      for (i = 0; i < NUM_ARCH_NAMES; i++)
+	if (strcmp (arch_names_table[i].name,
+	    TREE_STRING_POINTER (param_string_cst)) == 0)
+	  break;
+
+      if (i == NUM_ARCH_NAMES)
+	{
+	  error ("Parameter to builtin not valid: %s",
+	         TREE_STRING_POINTER (param_string_cst));
+	  return integer_zero_node;
+	}
+
+      field = TYPE_FIELDS (__processor_model_type);
+      field_val = arch_names_table[i].model;
+
+      /* CPU types are stored in the next field.  */
+      if (field_val > M_CPU_TYPE_START
+	  && field_val < M_CPU_SUBTYPE_START)
+	{
+	  field = DECL_CHAIN (field);
+	  field_val -= M_CPU_TYPE_START;
+	}
+
+      /* CPU subtypes are stored in the last field.  */
+      if (field_val > M_CPU_SUBTYPE_START)
+	{
+	  field = DECL_CHAIN ( DECL_CHAIN (field));
+	  field_val -= M_CPU_SUBTYPE_START;
+	}
+
+      /* Get the appropriate field in __cpu_model.  */
+      ref =  build3 (COMPONENT_REF, TREE_TYPE (field), __cpu_model_var,
+		     field, NULL_TREE);
+
+      /* Check the value.  */
+      return build2 (EQ_EXPR, unsigned_type_node, ref,
+		     build_int_cstu (unsigned_type_node, field_val));
+    }
+  else if (fn_code == IX86_BUILTIN_CPU_SUPPORTS)
+    {
+      unsigned int NUM_ISA_NAMES
+	= sizeof (isa_names_table) / sizeof (struct _isa_names_table);
+      for (i = 0; i < NUM_ISA_NAMES; i++)
+	if (strcmp (isa_names_table[i].name,
+	    TREE_STRING_POINTER (param_string_cst)) == 0)
+	  break;
+
+      if (i == NUM_ISA_NAMES)
+	{
+	  error ("Parameter to builtin not valid: %s",
+	       	 TREE_STRING_POINTER (param_string_cst));
+	  return integer_zero_node;
+	}
+
+      field = get_field_from_struct (__processor_features_type,
+				     isa_names_table[i].feature);
+      which_struct = __cpu_features_var;
+      return build3 (COMPONENT_REF, TREE_TYPE (field), which_struct,
+		     field, NULL_TREE);
+    }
+
+  gcc_unreachable ();
+
+}
+
+static tree
+ix86_fold_builtin (tree fndecl, int n_args,
+		   tree *args, bool ignore ATTRIBUTE_UNUSED)
+{
+  if (DECL_BUILT_IN_CLASS (fndecl) == BUILT_IN_MD)
+    {
+      enum ix86_builtins fn_code = (enum ix86_builtins)
+				   DECL_FUNCTION_CODE (fndecl);
+      if (fn_code ==  IX86_BUILTIN_CPU_IS
+	  || fn_code == IX86_BUILTIN_CPU_SUPPORTS)
+	{
+	  gcc_assert (n_args == 1);
+          return fold_builtin_cpu (fndecl, args);
+	}
+    }
+
+  return NULL_TREE;
+}
+
+/* Make builtins to detect cpu type and features supported.  NAME is
+   the builtin name, CODE is the builtin code, and FTYPE is the function
+   type of the builtin.  */
+
+static void
+make_cpu_type_builtin (const char* name, int code,
+		       enum ix86_builtin_func_type ftype)
+{
+  tree decl;
+  tree type;
+
+  type = ix86_get_builtin_func_type (ftype);
+  decl = add_builtin_function (name, type, code, BUILT_IN_MD,
+			       NULL, NULL_TREE);
+  gcc_assert (decl != NULL_TREE);
+  ix86_builtins[(int) code] = decl;
+  TREE_READONLY (decl) = 1;
+}
+
+/* Make builtins to get CPU type and features supported.  The created
+   builtins are :
+
+   __builtin_cpu_init (), to detect cpu type and features,
+   __builtin_cpu_is ("<CPUNAME>"), to check if cpu is of type <CPUNAME>,
+   __builtin_cpu_supports ("<FEATURE>"), to check if cpu supports <FEATURE>
+   */
+
+static void
+ix86_init_platform_type_builtins (void)
+{
+  make_cpu_type_builtin ("__builtin_cpu_init", IX86_BUILTIN_CPU_INIT,
+			 INT_FTYPE_VOID);
+  make_cpu_type_builtin ("__builtin_cpu_is", IX86_BUILTIN_CPU_IS,
+			 INT_FTYPE_PCCHAR);
+  make_cpu_type_builtin ("__builtin_cpu_supports", IX86_BUILTIN_CPU_SUPPORTS,
+			 INT_FTYPE_PCCHAR);
+}
+
 /* Internal method for ix86_init_builtins.  */
 
 static void
@@ -27710,6 +28070,9 @@ ix86_init_builtins (void)
 
   ix86_init_builtin_types ();
 
+  /* Builtins to get CPU type and features. */
+  ix86_init_platform_type_builtins ();
+
   /* TFmode support builtins.  */
   def_builtin_const (0, "__builtin_infq",
 		     FLOAT128_FTYPE_VOID, IX86_BUILTIN_INFQ);
@@ -29328,6 +29691,28 @@ ix86_expand_builtin (tree exp, rtx target, rtx sub
   enum machine_mode mode0, mode1, mode2, mode3, mode4;
   unsigned int fcode = DECL_FUNCTION_CODE (fndecl);
 
+  /* For CPU builtins that can be folded, fold first and expand the fold.  */
+  switch (fcode)
+    {
+    case IX86_BUILTIN_CPU_INIT:
+      {
+	/* Make it call __cpu_indicator_init in libgcc. */
+	tree call_expr, fndecl, type;
+        type = build_function_type_list (integer_type_node, NULL_TREE); 
+	fndecl = build_fn_decl ("__cpu_indicator_init", type);
+	call_expr = build_call_expr (fndecl, 0); 
+	return expand_expr (call_expr, target, mode, EXPAND_NORMAL);
+      }
+    case IX86_BUILTIN_CPU_IS:
+    case IX86_BUILTIN_CPU_SUPPORTS:
+      {
+	tree arg0 = CALL_EXPR_ARG (exp, 0);
+	tree fold_expr = fold_builtin_cpu (fndecl, &arg0);
+	gcc_assert (fold_expr != NULL_TREE);
+        return expand_expr (fold_expr, target, mode, EXPAND_NORMAL);
+      }
+    }
+
   /* Determine whether the builtin function is available under the current ISA.
      Originally the builtin was not created if it wasn't applicable to the
      current ISA based on the command line switches.  With function specific
@@ -39058,6 +39443,9 @@ ix86_autovectorize_vector_sizes (void)
 #undef TARGET_BUILD_BUILTIN_VA_LIST
 #define TARGET_BUILD_BUILTIN_VA_LIST ix86_build_builtin_va_list
 
+#undef TARGET_FOLD_BUILTIN
+#define TARGET_FOLD_BUILTIN ix86_fold_builtin
+
 #undef TARGET_ENUM_VA_LIST_P
 #define TARGET_ENUM_VA_LIST_P ix86_enum_va_list
 
Index: gcc/doc/extend.texi
===================================================================
--- gcc/doc/extend.texi	(revision 185898)
+++ gcc/doc/extend.texi	(working copy)
@@ -9413,6 +9413,140 @@ Similar to @code{__builtin_huge_val}, except the r
 @findex __builtin_huge_valq
 @end table
 
+The following built-in functions are always available and can be used to
+check the target platform type.
+
+@deftypefn {Built-in Function} void __builtin_cpu_init (void)
+This function runs the CPU detection code to check the type of CPU and the features
+supported. This builtin needs to be invoked along with the builtins to check CPU type
+and features, @code{__builtin_cpu_is} and @code{__builtin_cpu_supports}, only when used
+used in a function that will be executed before any constructors are called.  The
+CPU detection code is automatically executed in a very high priority constructor.
+
+For example, this function has to be used in @code{ifunc} resolvers which check for
+CPU type using the builtins, @code{__builtin_cpu_is} and @code{__builtin_cpu_supports}.
+@smallexample
+
+static void (*resolve_memcpy (void)) (void)
+@{
+  // ifunc resolvers fire before constructors, explicitly call the init
+  // function.
+  __builtin_cpu_init ();
+  if (__builtin_cpu_supports ("ssse3"))
+    return ssse3_memcpy; // super fast memcpy with ssse3 instructions.
+  else
+    return default_memcpy;
+@}
+
+void *memcpy (void *, const void *, size_t)
+     __attribute__ ((ifunc ("resolve_memcpy")));
+@end smallexample
+
+@end deftypefn
+
+@deftypefn {Built-in Function} int __builtin_cpu_is (const char *@var{cpuname})
+This function returns @code{1}, if the runtime cpu is of type @var{cpuname}
+ and returns @code{0} otherwise. The following cpu names can be detected:
+
+@table @samp
+@item intel
+Intel CPU.
+
+@item atom
+Intel ATOM CPU.
+
+@item core2
+Intel Core2 CPU.
+
+@item corei7
+Intel Corei7 CPU.
+
+@item nehalem
+Intel Corei7 Nehalem CPU.
+
+@item westmere
+Intel Corei7 Westmere CPU.
+
+@item sandybridge
+Intel Corei7 Sandybridge CPU.
+
+@item amd
+AMD CPU.
+
+@item amdfam10h
+AMD family 10h CPU.
+
+@item barcelona
+AMD family 10h Barcelona CPU.
+
+@item shanghai
+AMD family 10h Shanghai CPU.
+
+@item istanbul
+AMD family 15h Istanbul CPU.
+
+@item amdfam15h
+AMD family 15h CPU.
+
+@item bdver1
+AMD family 15h Bulldozer version 1.
+
+@item bdver2
+AMD family 15h Bulldozer version 1.
+@end table 
+
+Here is an example:
+@smallexample
+if (__builtin_cpu_is ("corei7"))
+  @{
+     do_corei7 (); //Corei7 specific implementation.
+  @}
+else
+  @{
+     do_generic (); //Generic implementation.
+  @}  
+@end smallexample
+@end deftypefn
+
+@deftypefn {Built-in Function} int __builtin_cpu_supports (const char *@var{feature})
+This function returns @code{1}, if the runtime cpu supports @var{feature}
+ and returns @code{0} otherwise. The following features can be detected:
+
+@table @samp
+@item cmov
+CMOV instruction.
+@item mmx
+MMX instructions.
+@item popcnt
+POPCNT instruction.
+@item sse
+SSE instructions.
+@item sse2
+SSE2 instructions.
+@item sse3
+SSE3 instructions.
+@item ssse3
+SSSE3 instructions.
+@item sse4.1
+SSE4.1 instructions.
+@item sse4.2
+SSE4.2 instructions.
+@end table
+
+Here is an example:
+@smallexample
+if (__builtin_cpu_supports ("popcnt"))
+  @{
+     asm("popcnt %1,%0" : "=r"(count) : "rm"(n) : "cc");
+  @}
+else
+  @{
+     count = generic_countbits (n); //generic implementation.
+  @}  
+@end smallexample
+@end deftypefn
+
+
 The following built-in functions are made available by @option{-mmmx}.
 All of them generate the machine instruction that is part of the name.
 

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

Hi,

i386 maintainers -  Is this patch ok?

Thanks,
-Sri.

On Mon, Apr 2, 2012 at 5:48 PM, Sriraman Tallam <tmsriram@google.com> wrote:
> On Mon, Apr 2, 2012 at 5:38 AM, Richard Guenther
> <richard.guenther@gmail.com> wrote:
>> On Sat, Mar 31, 2012 at 1:03 AM, Sriraman Tallam <tmsriram@google.com> wrote:
>>> On Fri, Mar 30, 2012 at 5:47 AM, Michael Matz <matz@suse.de> wrote:
>>>> Hi,
>>>>
>>>> On Thu, 29 Mar 2012, Sriraman Tallam wrote:
>>>>
>>>>> +struct __processor_model
>>>>> +{
>>>>> +  /* Vendor. */
>>>>> +  unsigned int __cpu_is_amd : 1;
>>>>> +  unsigned int __cpu_is_intel : 1;
>>>>> +  /* CPU type. */
>>>>> +  unsigned int __cpu_is_intel_atom : 1;
>>>>> +  unsigned int __cpu_is_intel_core2 : 1;
>>>>> +  unsigned int __cpu_is_intel_corei7 : 1;
>>>>> +  unsigned int __cpu_is_intel_corei7_nehalem : 1;
>>>>> +  unsigned int __cpu_is_intel_corei7_westmere : 1;
>>>>> +  unsigned int __cpu_is_intel_corei7_sandybridge : 1;
>>>>> +  unsigned int __cpu_is_amdfam10h : 1;
>>>>> +  unsigned int __cpu_is_amdfam10h_barcelona : 1;
>>>>> +  unsigned int __cpu_is_amdfam10h_shanghai : 1;
>>>>> +  unsigned int __cpu_is_amdfam10h_istanbul : 1;
>>>>> +  unsigned int __cpu_is_amdfam15h_bdver1 : 1;
>>>>> +  unsigned int __cpu_is_amdfam15h_bdver2 : 1;
>>>>> +} __cpu_model;
>>>>
>>>> It doesn't make sense for the model to be a bitfield, a processor will
>>>> have only ever exactly one model.  Just make it an enum or even just an
>>>> int.
>>>
>>> Not entirely true, nehalem and corei7 can be both set. However, I
>>> modified this by dividing it into types and sub types and then did
>>> what you said.
>>
>> Uh...  then I suppose you need to document somewhere what names
>> match to what cpuid family/model (supposedly thats where your two-layer
>> hierarchy comes from, which incidentially misses one layer, the vendor?)
>
> Added documentation to extend.texi
>
> Patch available for review here:
> http://codereview.appspot.com/5754058
>
> Thanks,
> -Sri.
>
>
>>
>> Richard.
>>
>>>       * config/i386/i386.c (build_processor_features_struct): New function.
>>>       (build_processor_model_struct): New function.
>>>       (make_var_decl): New function.
>>>       (get_field_from_struct): New function.
>>>       (fold_builtin_target): New function.
>>>       (ix86_fold_builtin): New function.
>>>       (ix86_expand_builtin): Expand new builtins by folding them.
>>>       (make_cpu_type_builtin): New functions.
>>>       (ix86_init_platform_type_builtins): Make the new builtins.
>>>       (ix86_init_builtins): Make new builtins to detect CPU type.
>>>       (TARGET_FOLD_BUILTIN): New macro.
>>>       (IX86_BUILTIN_CPU_INIT): New enum value.
>>>       (IX86_BUILTIN_CPU_IS): New enum value.
>>>       (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>>>       * config/i386/i386-builtin-types.def: New function type.
>>>       * testsuite/gcc.target/builtin_target.c: New testcase.
>>>
>>>       * libgcc/config/i386/i386-cpuinfo.c: New file.
>>>       * libgcc/config/i386/t-cpuinfo: New file.
>>>       * libgcc/config.host: Include t-cpuinfo.
>>>       * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
>>>       and __cpu_features.
>>>
>>> Patch available for review here:
>>> http://codereview.appspot.com/5754058
>>>
>>> Thanks,
>>> -Sri.
>>>
>>>
>>>>
>>>>
>>>> Ciao,
>>>> Michael.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Gerald Pfeifer]

On Thu, 29 Mar 2012, Sriraman Tallam wrote:
> Hi,
> 
>     I have made a new patch to only have two builtins :
> 
> * __builtin_cpu_is ("<CPUNAME>")
> * __builtin_cpu_supports ("<FEATURE>")
> 
> apart from the cpu init builtin, __builtin_cpu_init.

I don't see any .texi file as part of this change.  Shouldn't this
be documented (and also added to the release notes gcc-4.8/changes.html)?

> List of CPU names :
> 
> * "amd"
> * "intel"

Are company names really suitable here?  Intel is also still producing
ia64 aka Itanium, and in the future AMD might do some ARM-based designs.
AMD64 and Intel64 might work.

Gerald

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

[-- Attachment #1: Type: text/plain, Size: 1001 bytes --]

On Sun, Apr 8, 2012 at 6:17 AM, Gerald Pfeifer <gerald@pfeifer.com> wrote:
> On Thu, 29 Mar 2012, Sriraman Tallam wrote:
>> Hi,
>>
>>     I have made a new patch to only have two builtins :
>>
>> * __builtin_cpu_is ("<CPUNAME>")
>> * __builtin_cpu_supports ("<FEATURE>")
>>
>> apart from the cpu init builtin, __builtin_cpu_init.
>
> I don't see any .texi file as part of this change.  Shouldn't this
> be documented (and also added to the release notes gcc-4.8/changes.html)?

I modified extend.texi for documenting the builtins. I guess you
missed it. I am reattaching the patch just in case.

>
>> List of CPU names :
>>
>> * "amd"
>> * "intel"
>
> Are company names really suitable here?  Intel is also still producing
> ia64 aka Itanium, and in the future AMD might do some ARM-based designs.
> AMD64 and Intel64 might work.

This is basically the vendor info from CPUID. I dont mind changing it
but I do not understand why Intel or AMD is unsuitable.

>
> Gerald

[-- Attachment #2: cpu_runtime_patch.txt --]
[-- Type: text/plain, Size: 28968 bytes --]


       * config/i386/i386.c (build_processor_features_struct): New function.
       (build_processor_model_struct): New function.
       (make_var_decl): New function.
       (get_field_from_struct): New function.
       (fold_builtin_target): New function.
       (ix86_fold_builtin): New function.
       (ix86_expand_builtin): Expand new builtins by folding them.
       (make_cpu_type_builtin): New functions.
       (ix86_init_platform_type_builtins): Make the new builtins.
       (ix86_init_builtins): Make new builtins to detect CPU type.
       (TARGET_FOLD_BUILTIN): New macro.
       (IX86_BUILTIN_CPU_INIT): New enum value.
       (IX86_BUILTIN_CPU_IS): New enum value.
       (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
       * config/i386/i386-builtin-types.def: New function type.
       * testsuite/gcc.target/builtin_target.c: New testcase.
       * doc/extend.texi: Document builtins.

       * libgcc/config/i386/i386-cpuinfo.c: New file.
       * libgcc/config/i386/t-cpuinfo: New file.
       * libgcc/config.host: Include t-cpuinfo.
       * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
       and __cpu_features.


Index: libgcc/config.host
===================================================================
--- libgcc/config.host	(revision 185898)
+++ libgcc/config.host	(working copy)
@@ -1130,7 +1130,7 @@ i[34567]86-*-linux* | x86_64-*-linux* | \
   i[34567]86-*-kfreebsd*-gnu | x86_64-*-kfreebsd*-gnu | \
   i[34567]86-*-knetbsd*-gnu | \
   i[34567]86-*-gnu*)
-	tmake_file="${tmake_file} t-tls i386/t-linux"
+	tmake_file="${tmake_file} t-tls i386/t-linux i386/t-cpuinfo"
 	if test "$libgcc_cv_cfi" = "yes"; then
 		tmake_file="${tmake_file} t-stack i386/t-stack-i386"
 	fi
Index: libgcc/config/i386/t-cpuinfo
===================================================================
--- libgcc/config/i386/t-cpuinfo	(revision 0)
+++ libgcc/config/i386/t-cpuinfo	(revision 0)
@@ -0,0 +1 @@
+LIB2ADD += $(srcdir)/config/i386/i386-cpuinfo.c
Index: libgcc/config/i386/i386-cpuinfo.c
===================================================================
--- libgcc/config/i386/i386-cpuinfo.c	(revision 0)
+++ libgcc/config/i386/i386-cpuinfo.c	(revision 0)
@@ -0,0 +1,298 @@
+/* Get CPU type and Features for x86 processors.
+   Copyright (C) 2012 Free Software Foundation, Inc.
+   Contributed by Sriraman Tallam (tmsriram@google.com)
+
+This file is part of GCC.
+
+GCC is free software; you can redistribute it and/or modify it under
+the terms of the GNU General Public License as published by the Free
+Software Foundation; either version 3, or (at your option) any later
+version.
+
+GCC is distributed in the hope that it will be useful, but WITHOUT ANY
+WARRANTY; without even the implied warranty of MERCHANTABILITY or
+FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+for more details.
+
+You should have received a copy of the GNU General Public License
+along with GCC; see the file COPYING3.  If not see
+<http://www.gnu.org/licenses/>. */
+
+#include "cpuid.h"
+#include "tsystem.h"
+
+int __cpu_indicator_init (void) __attribute__ ((constructor (101)));
+
+enum vendor_signatures
+{
+  SIG_INTEL =	0x756e6547 /* Genu */,
+  SIG_AMD =	0x68747541 /* Auth */
+};
+
+/* ISA Features supported. */
+
+struct __processor_features
+{
+  unsigned int __cpu_cmov : 1;
+  unsigned int __cpu_mmx : 1;
+  unsigned int __cpu_popcnt : 1;
+  unsigned int __cpu_sse : 1;
+  unsigned int __cpu_sse2 : 1;
+  unsigned int __cpu_sse3 : 1;
+  unsigned int __cpu_ssse3 : 1;
+  unsigned int __cpu_sse4_1 : 1;
+  unsigned int __cpu_sse4_2 : 1;
+} __cpu_features;
+
+/* Processor Vendor and Models. */
+
+enum processor_vendor
+{
+  VENDOR_INTEL = 1,
+  VENDOR_AMD,
+  VENDOR_MAX
+};
+
+enum processor_types
+{
+  INTEL_ATOM = 1,
+  INTEL_CORE2,
+  INTEL_COREI7,
+  AMDFAM10H,
+  AMDFAM15H,
+  CPU_TYPE_MAX
+};
+
+enum processor_subtypes
+{
+  INTEL_COREI7_NEHALEM,
+  INTEL_COREI7_WESTMERE,
+  INTEL_COREI7_SANDYBRIDGE,
+  AMDFAM10H_BARCELONA,
+  AMDFAM10H_SHANGHAI,
+  AMDFAM10H_ISTANBUL,
+  AMDFAM15H_BDVER1,
+  AMDFAM15H_BDVER2,
+  CPU_SUBTYPE_MAX
+};
+
+struct __processor_model
+{
+  unsigned int __cpu_vendor;
+  unsigned int __cpu_type;
+  unsigned int __cpu_subtype;
+} __cpu_model;
+
+/* Get the specific type of AMD CPU.  */
+
+static void
+get_amd_cpu (unsigned int family, unsigned int model)
+{
+  __cpu_model.__cpu_vendor = VENDOR_AMD;
+  switch (family)
+    {
+    /* AMD Family 10h.  */
+    case 0x10:
+      switch (model)
+	{
+	case 0x2:
+	  /* Barcelona.  */
+	  __cpu_model.__cpu_type = AMDFAM10H;
+	  __cpu_model.__cpu_subtype = AMDFAM10H_BARCELONA;
+	  break;
+	case 0x4:
+	  /* Shanghai.  */
+	  __cpu_model.__cpu_type = AMDFAM10H;
+	  __cpu_model.__cpu_subtype = AMDFAM10H_SHANGHAI;
+	  break;
+	case 0x8:
+	  /* Istanbul.  */
+	  __cpu_model.__cpu_type = AMDFAM10H;
+	  __cpu_model.__cpu_subtype = AMDFAM10H_ISTANBUL;
+	  break;
+	default:
+	  break;
+	}
+      break;
+    /* AMD Family 15h.  */
+    case 0x15:
+      __cpu_model.__cpu_type = AMDFAM15H;
+      /* Bulldozer version 1.  */
+      if ( model <= 0xf)
+	__cpu_model.__cpu_subtype = AMDFAM15H_BDVER1;
+      /* Bulldozer version 2.  */
+      if (model >= 0x10 && model <= 0x1f)
+	__cpu_model.__cpu_subtype = AMDFAM15H_BDVER2;
+      break;
+    default:
+      break;
+    }
+}
+
+/* Get the specific type of Intel CPU.  */
+
+static void
+get_intel_cpu (unsigned int family, unsigned int model, unsigned int brand_id)
+{
+  __cpu_model.__cpu_vendor = VENDOR_INTEL;
+  /* Parse family and model only if brand ID is 0. */
+  if (brand_id == 0)
+    {
+      switch (family)
+	{
+	case 0x5:
+	  /* Pentium.  */
+	  break;
+	case 0x6:
+	  switch (model)
+	    {
+	    case 0x1c:
+	    case 0x26:
+	      /* Atom.  */
+	      __cpu_model.__cpu_type = INTEL_ATOM;
+	      break;
+	    case 0x1a:
+	    case 0x1e:
+	    case 0x1f:
+	    case 0x2e:
+	      /* Nehalem.  */
+	      __cpu_model.__cpu_type = INTEL_COREI7;
+	      __cpu_model.__cpu_subtype = INTEL_COREI7_NEHALEM;
+	      break;
+	    case 0x25:
+	    case 0x2c:
+	    case 0x2f:
+	      /* Westmere.  */
+	      __cpu_model.__cpu_type = INTEL_COREI7;
+	      __cpu_model.__cpu_subtype = INTEL_COREI7_WESTMERE;
+	      break;
+	    case 0x2a:
+	      /* Sandy Bridge.  */
+	      __cpu_model.__cpu_type = INTEL_COREI7;
+	      __cpu_model.__cpu_subtype = INTEL_COREI7_SANDYBRIDGE;
+	      break;
+	    case 0x17:
+	    case 0x1d:
+	      /* Penryn.  */
+	    case 0x0f:
+	      /* Merom.  */
+	      __cpu_model.__cpu_type = INTEL_CORE2;
+	      break;
+	    default:
+	      break;
+	    }
+	  break;
+	default:
+	  /* We have no idea.  */
+	  break;
+	}
+    }
+}	             	
+
+static void
+get_available_features (unsigned int ecx, unsigned int edx)
+{
+  __cpu_features.__cpu_cmov = (edx & bit_CMOV) ? 1 : 0;
+  __cpu_features.__cpu_mmx = (edx & bit_MMX) ? 1 : 0;
+  __cpu_features.__cpu_sse = (edx & bit_SSE) ? 1 : 0;
+  __cpu_features.__cpu_sse2 = (edx & bit_SSE2) ? 1 : 0;
+  __cpu_features.__cpu_popcnt = (ecx & bit_POPCNT) ? 1 : 0;
+  __cpu_features.__cpu_sse3 = (ecx & bit_SSE3) ? 1 : 0;
+  __cpu_features.__cpu_ssse3 = (ecx & bit_SSSE3) ? 1 : 0;
+  __cpu_features.__cpu_sse4_1 = (ecx & bit_SSE4_1) ? 1 : 0;
+  __cpu_features.__cpu_sse4_2 = (ecx & bit_SSE4_2) ? 1 : 0;
+}
+
+/* A noinline function calling __get_cpuid. Having many calls to
+   cpuid in one function in 32-bit mode causes GCC to complain:
+   "can;t find a register in class CLOBBERED_REGS".  This is
+   related to PR rtl-optimization 44174. */
+
+static int __attribute__ ((noinline))
+__get_cpuid_output (unsigned int __level,
+		    unsigned int *__eax, unsigned int *__ebx,
+		    unsigned int *__ecx, unsigned int *__edx)
+{
+  return __get_cpuid (__level, __eax, __ebx, __ecx, __edx);
+}
+
+
+/* A constructor function that is sets __cpu_model and __cpu_features with
+   the right values.  This needs to run only once.  This constructor is
+   given the highest priority and it should run before constructors without
+   the priority set.  However, it still runs after ifunc initializers and
+   needs to be called explicitly there.  */
+
+int __attribute__ ((constructor (101)))
+__cpu_indicator_init (void)
+{
+  unsigned int eax, ebx, ecx, edx;
+
+  int max_level = 5;
+  unsigned int vendor;
+  unsigned int model, family, brand_id;
+  unsigned int extended_model, extended_family;
+  static int called = 0;
+
+  /* This function needs to run just once.  */
+  if (called)
+    return 0;
+  else
+    called = 1;
+
+  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
+  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
+    return -1;
+
+  vendor = ebx;
+  max_level = eax;
+
+  if (max_level < 1)
+    return -1;
+
+  if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
+    return -1;
+
+  model = (eax >> 4) & 0x0f;
+  family = (eax >> 8) & 0x0f;
+  brand_id = ebx & 0xff;
+  extended_model = (eax >> 12) & 0xf0;
+  extended_family = (eax >> 20) & 0xff;
+
+  if (vendor == SIG_INTEL)
+    {
+      /* Adjust model and family for Intel CPUS. */
+      if (family == 0x0f)
+	{
+	  family += extended_family;
+	  model += extended_model;
+	}
+      else if (family == 0x06)
+	model += extended_model;
+
+      /* Get CPU type.  */
+      get_intel_cpu (family, model, brand_id);
+    }
+
+  if (vendor == SIG_AMD)
+    {
+      /* Adjust model and family for AMD CPUS. */
+      if (family == 0x0f)
+	{
+	  family += extended_family;
+	  model += (extended_model << 4);
+	}
+
+      /* Get CPU type.  */
+      get_amd_cpu (family, model);
+    }
+
+  /* Find available features. */
+  get_available_features (ecx, edx);
+
+  gcc_assert (__cpu_model.__cpu_vendor < VENDOR_MAX);
+  gcc_assert (__cpu_model.__cpu_type < CPU_TYPE_MAX);
+  gcc_assert (__cpu_model.__cpu_subtype < CPU_SUBTYPE_MAX);
+
+  return 0;
+}
Index: libgcc/config/i386/libgcc-glibc.ver
===================================================================
--- libgcc/config/i386/libgcc-glibc.ver	(revision 185898)
+++ libgcc/config/i386/libgcc-glibc.ver	(working copy)
@@ -147,6 +147,11 @@ GCC_4.3.0 {
   __trunctfxf2
   __unordtf2
 }
+
+GCC_4.8.0 {
+  __cpu_model
+  __cpu_features
+}
 %else
 GCC_4.4.0 {
   __addtf3
@@ -183,4 +188,9 @@ GCC_4.4.0 {
 GCC_4.5.0 {
   __extendxftf2
 }
+
+GCC_4.8.0 {
+  __cpu_model
+  __cpu_features
+}
 %endif
Index: gcc/testsuite/gcc.target/i386/builtin_target.c
===================================================================
--- gcc/testsuite/gcc.target/i386/builtin_target.c	(revision 0)
+++ gcc/testsuite/gcc.target/i386/builtin_target.c	(revision 0)
@@ -0,0 +1,65 @@
+/* This test checks if the __builtin_cpu_is and __builtin_cpu_supports calls
+   are recognized. */
+
+/* { dg-do run } */
+
+int
+fn1 ()
+{
+  if (__builtin_cpu_supports ("cmov") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("mmx") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("popcnt") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse2") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse3") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("ssse3") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse4.1") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse4.2") < 0)
+    return -1;
+  if (__builtin_cpu_is ("amd") < 0)
+    return -1;
+  if (__builtin_cpu_is ("intel") < 0)
+    return -1;
+  if (__builtin_cpu_is ("atom") < 0)
+    return -1;
+  if (__builtin_cpu_is ("core2") < 0)
+    return -1;
+  if (__builtin_cpu_is ("corei7") < 0)
+    return -1;
+  if (__builtin_cpu_is ("nehalem") < 0)
+    return -1;
+  if (__builtin_cpu_is ("westmere") < 0)
+    return -1;
+  if (__builtin_cpu_is ("sandybridge") < 0)
+    return -1;
+  if (__builtin_cpu_is ("amdfam10h") < 0)
+    return -1;
+  if (__builtin_cpu_is ("barcelona") < 0)
+    return -1;
+  if (__builtin_cpu_is ("shanghai") < 0)
+    return -1;
+  if (__builtin_cpu_is ("istanbul") < 0)
+    return -1;
+  if (__builtin_cpu_is ("amdfam15h") < 0)
+    return -1;
+  if (__builtin_cpu_is ("bdver1") < 0)
+    return -1;
+  if (__builtin_cpu_is ("bdver2") < 0)
+    return -1;
+
+  return 0;
+}
+
+int main ()
+{
+  __builtin_cpu_init ();
+  return fn1 ();
+}
Index: gcc/config/i386/i386-builtin-types.def
===================================================================
--- gcc/config/i386/i386-builtin-types.def	(revision 185898)
+++ gcc/config/i386/i386-builtin-types.def	(working copy)
@@ -155,6 +155,7 @@ DEF_FUNCTION_TYPE (INT, V4SF)
 DEF_FUNCTION_TYPE (INT, V8QI)
 DEF_FUNCTION_TYPE (INT, V8SF)
 DEF_FUNCTION_TYPE (INT, V32QI)
+DEF_FUNCTION_TYPE (INT, PCCHAR)
 DEF_FUNCTION_TYPE (INT64, INT64)
 DEF_FUNCTION_TYPE (INT64, V2DF)
 DEF_FUNCTION_TYPE (INT64, V4SF)
Index: gcc/config/i386/i386.c
===================================================================
--- gcc/config/i386/i386.c	(revision 185898)
+++ gcc/config/i386/i386.c	(working copy)
@@ -25809,6 +25809,11 @@ enum ix86_builtins
   /* CFString built-in for darwin */
   IX86_BUILTIN_CFSTRING,
 
+  /* Builtins to get CPU type and supported features. */
+  IX86_BUILTIN_CPU_INIT,
+  IX86_BUILTIN_CPU_IS,
+  IX86_BUILTIN_CPU_SUPPORTS,
+
   IX86_BUILTIN_MAX
 };
 
@@ -27627,6 +27632,361 @@ ix86_init_mmx_sse_builtins (void)
     }
 }
 
+/* This builds the processor_model struct type defined in
+   libgcc/config/i386/i386-cpuinfo.c  */
+static tree
+build_processor_model_struct (void)
+{
+  const char *field_name[] = {"__cpu_vendor", "__cpu_type", "__cpu_subtype"};
+  tree field = NULL_TREE, field_chain = NULL_TREE;
+  int i;
+  tree type = make_node (RECORD_TYPE);
+  int num_fields = sizeof (field_name) / sizeof (const char *);
+
+  for (i = 0; i < num_fields; ++i)
+    {
+      field = build_decl (UNKNOWN_LOCATION, FIELD_DECL,
+			  get_identifier (field_name[i]), unsigned_type_node);
+      if (field_chain != NULL_TREE)
+	DECL_CHAIN (field) = field_chain;
+      field_chain = field; 
+    }
+  finish_builtin_struct (type, "__processor_model", field_chain, NULL_TREE);
+  return type;
+}
+
+/* This builds the processor_features struct type defined in
+   libgcc/config/i386/i386-cpuinfo.c  */
+
+static tree
+build_processor_features_struct (void)
+{
+  int i;
+  const char *field_name[] = {"__cpu_cmov", "__cpu_mmx", "__cpu_popcnt",
+			      "__cpu_sse", "__cpu_sse2", "__cpu_sse3",
+			      "__cpu_ssse3", "__cpu_sse4_1", "__cpu_sse4_2"};
+
+  tree field = NULL_TREE, field_chain = NULL_TREE;
+  tree type = make_node (RECORD_TYPE);
+  int num_fields = sizeof (field_name) / sizeof (const char *);
+  
+  for (i = 0; i < num_fields; i++)
+    {
+      field = build_decl (UNKNOWN_LOCATION, FIELD_DECL,
+			  get_identifier (field_name[i]), unsigned_type_node);
+      DECL_BIT_FIELD (field) = 1;
+      DECL_SIZE (field) = bitsize_one_node;
+      if (field_chain != NULL_TREE)
+	DECL_CHAIN (field) = field_chain;
+      field_chain = field;
+    }
+  finish_builtin_struct (type, "__processor_features", field_chain, NULL_TREE);
+  return type;
+}
+
+/* Returns a extern, comdat VAR_DECL of type TYPE and name NAME. */
+
+static tree
+make_var_decl (tree type, const char *name)
+{
+  tree new_decl;
+
+  new_decl = build_decl (UNKNOWN_LOCATION,
+	                 VAR_DECL,
+	  	         get_identifier(name),
+		         type);
+
+  DECL_EXTERNAL (new_decl) = 1;
+  TREE_STATIC (new_decl) = 1;
+  TREE_PUBLIC (new_decl) = 1;
+  DECL_INITIAL (new_decl) = 0;
+  DECL_ARTIFICIAL (new_decl) = 0;
+  DECL_PRESERVE_P (new_decl) = 1;
+
+  make_decl_one_only (new_decl, DECL_ASSEMBLER_NAME (new_decl));
+  assemble_variable (new_decl, 0, 0, 0);
+
+  return new_decl;
+}
+
+/* Traverses the chain of fields in STRUCT_TYPE and returns the FIELD_NUM
+   numbered field. */
+
+static tree
+get_field_from_struct (tree struct_type, int field_num)
+{
+  int i;
+  tree field = TYPE_FIELDS (struct_type);
+
+  for (i = 0; i < field_num; i++, field = DECL_CHAIN(field))
+    {
+      gcc_assert (field != NULL_TREE);
+    }
+
+  return field;
+}
+
+/* FNDECL is a __builtin_cpu_is or a __builtin_cpu_supports call that is folded
+   into an integer defined in libgcc/config/i386/i386-cpuinfo.c */
+
+static tree 
+fold_builtin_cpu (tree fndecl, tree *args)
+{
+  unsigned int i;
+  enum ix86_builtins fn_code = (enum ix86_builtins)
+				DECL_FUNCTION_CODE (fndecl);
+  tree param_string_cst = NULL;
+
+  /* This is the order of bit-fields in __processor_features in
+     i386-cpuinfo.c */
+  enum processor_features
+  {
+    F_CMOV = 0,
+    F_MMX,
+    F_POPCNT,
+    F_SSE,
+    F_SSE2,
+    F_SSE3,
+    F_SSSE3,
+    F_SSE4_1,
+    F_SSE4_2,
+    F_MAX
+  };
+
+  /* These are the values for vendor types and cpu types  and subtypes
+     in i386-cpuinfo.c.  Cpu types and subtypes should be subtracted by
+     the corresponding start value.  */
+  enum processor_model
+  {
+    M_INTEL = 1,
+    M_AMD,
+    M_CPU_TYPE_START,
+    M_INTEL_ATOM,
+    M_INTEL_CORE2,
+    M_INTEL_COREI7,
+    M_AMDFAM10H,
+    M_AMDFAM15H,
+    M_CPU_SUBTYPE_START,
+    M_INTEL_COREI7_NEHALEM,
+    M_INTEL_COREI7_WESTMERE,
+    M_INTEL_COREI7_SANDYBRIDGE,
+    M_AMDFAM10H_BARCELONA,
+    M_AMDFAM10H_SHANGHAI,
+    M_AMDFAM10H_ISTANBUL,
+    M_AMDFAM15H_BDVER1,
+    M_AMDFAM15H_BDVER2
+  };
+
+  static struct _arch_names_table
+    {
+      const char *const name;
+      const enum processor_model model;
+    }
+  const arch_names_table[] =
+    {
+      {"amd", M_AMD},
+      {"intel", M_INTEL},
+      {"atom", M_INTEL_ATOM},
+      {"core2", M_INTEL_CORE2},
+      {"corei7", M_INTEL_COREI7},
+      {"nehalem", M_INTEL_COREI7_NEHALEM},
+      {"westmere", M_INTEL_COREI7_WESTMERE},
+      {"sandybridge", M_INTEL_COREI7_SANDYBRIDGE},
+      {"amdfam10h", M_AMDFAM10H},
+      {"barcelona", M_AMDFAM10H_BARCELONA},
+      {"shanghai", M_AMDFAM10H_SHANGHAI},
+      {"istanbul", M_AMDFAM10H_ISTANBUL},
+      {"amdfam15h", M_AMDFAM15H},
+      {"bdver1", M_AMDFAM15H_BDVER1},
+      {"bdver2", M_AMDFAM15H_BDVER2},
+    };
+
+  static struct _isa_names_table
+    {
+      const char *const name;
+      const enum processor_features feature;
+    }
+  const isa_names_table[] =
+    {
+      {"cmov",   F_CMOV},
+      {"mmx",    F_MMX},
+      {"popcnt", F_POPCNT},
+      {"sse",    F_SSE},
+      {"sse2",   F_SSE2},
+      {"sse3",   F_SSE3},
+      {"ssse3",  F_SSSE3},
+      {"sse4.1", F_SSE4_1},
+      {"sse4.2", F_SSE4_2}
+    };
+
+  static tree __processor_features_type = NULL_TREE;
+  static tree __cpu_features_var = NULL_TREE;
+  static tree __processor_model_type = NULL_TREE;
+  static tree __cpu_model_var = NULL_TREE;
+  static tree field;
+  static tree which_struct;
+
+  if (__processor_features_type == NULL_TREE)
+    __processor_features_type = build_processor_features_struct ();
+
+  if (__processor_model_type == NULL_TREE)
+    __processor_model_type = build_processor_model_struct ();
+
+  if (__cpu_features_var == NULL_TREE)
+    __cpu_features_var = make_var_decl (__processor_features_type,
+					"__cpu_features");
+
+  if (__cpu_model_var == NULL_TREE)
+    __cpu_model_var = make_var_decl (__processor_model_type,
+				     "__cpu_model");
+
+  gcc_assert ((args != NULL) && (*args != NULL));
+
+  param_string_cst = *args;
+  while (param_string_cst
+	 && TREE_CODE (param_string_cst) !=  STRING_CST)
+    {
+      /* *args must be a expr that can contain other EXPRS leading to a
+	 STRING_CST.   */
+      if (!EXPR_P (param_string_cst))
+ 	{
+	  error ("Parameter to builtin must be a string constant or literal");
+	  return integer_zero_node;
+	}
+      param_string_cst = TREE_OPERAND (EXPR_CHECK (param_string_cst), 0);
+    }
+ 
+  gcc_assert (param_string_cst);
+
+  if (fn_code == IX86_BUILTIN_CPU_IS)
+    {
+      tree ref, expr;
+      unsigned int field_val = 0;
+      unsigned int NUM_ARCH_NAMES
+	= sizeof (arch_names_table) / sizeof (struct _arch_names_table);
+
+      for (i = 0; i < NUM_ARCH_NAMES; i++)
+	if (strcmp (arch_names_table[i].name,
+	    TREE_STRING_POINTER (param_string_cst)) == 0)
+	  break;
+
+      if (i == NUM_ARCH_NAMES)
+	{
+	  error ("Parameter to builtin not valid: %s",
+	         TREE_STRING_POINTER (param_string_cst));
+	  return integer_zero_node;
+	}
+
+      field = TYPE_FIELDS (__processor_model_type);
+      field_val = arch_names_table[i].model;
+
+      /* CPU types are stored in the next field.  */
+      if (field_val > M_CPU_TYPE_START
+	  && field_val < M_CPU_SUBTYPE_START)
+	{
+	  field = DECL_CHAIN (field);
+	  field_val -= M_CPU_TYPE_START;
+	}
+
+      /* CPU subtypes are stored in the last field.  */
+      if (field_val > M_CPU_SUBTYPE_START)
+	{
+	  field = DECL_CHAIN ( DECL_CHAIN (field));
+	  field_val -= M_CPU_SUBTYPE_START;
+	}
+
+      /* Get the appropriate field in __cpu_model.  */
+      ref =  build3 (COMPONENT_REF, TREE_TYPE (field), __cpu_model_var,
+		     field, NULL_TREE);
+
+      /* Check the value.  */
+      return build2 (EQ_EXPR, unsigned_type_node, ref,
+		     build_int_cstu (unsigned_type_node, field_val));
+    }
+  else if (fn_code == IX86_BUILTIN_CPU_SUPPORTS)
+    {
+      unsigned int NUM_ISA_NAMES
+	= sizeof (isa_names_table) / sizeof (struct _isa_names_table);
+      for (i = 0; i < NUM_ISA_NAMES; i++)
+	if (strcmp (isa_names_table[i].name,
+	    TREE_STRING_POINTER (param_string_cst)) == 0)
+	  break;
+
+      if (i == NUM_ISA_NAMES)
+	{
+	  error ("Parameter to builtin not valid: %s",
+	       	 TREE_STRING_POINTER (param_string_cst));
+	  return integer_zero_node;
+	}
+
+      field = get_field_from_struct (__processor_features_type,
+				     isa_names_table[i].feature);
+      which_struct = __cpu_features_var;
+      return build3 (COMPONENT_REF, TREE_TYPE (field), which_struct,
+		     field, NULL_TREE);
+    }
+
+  gcc_unreachable ();
+
+}
+
+static tree
+ix86_fold_builtin (tree fndecl, int n_args,
+		   tree *args, bool ignore ATTRIBUTE_UNUSED)
+{
+  if (DECL_BUILT_IN_CLASS (fndecl) == BUILT_IN_MD)
+    {
+      enum ix86_builtins fn_code = (enum ix86_builtins)
+				   DECL_FUNCTION_CODE (fndecl);
+      if (fn_code ==  IX86_BUILTIN_CPU_IS
+	  || fn_code == IX86_BUILTIN_CPU_SUPPORTS)
+	{
+	  gcc_assert (n_args == 1);
+          return fold_builtin_cpu (fndecl, args);
+	}
+    }
+
+  return NULL_TREE;
+}
+
+/* Make builtins to detect cpu type and features supported.  NAME is
+   the builtin name, CODE is the builtin code, and FTYPE is the function
+   type of the builtin.  */
+
+static void
+make_cpu_type_builtin (const char* name, int code,
+		       enum ix86_builtin_func_type ftype)
+{
+  tree decl;
+  tree type;
+
+  type = ix86_get_builtin_func_type (ftype);
+  decl = add_builtin_function (name, type, code, BUILT_IN_MD,
+			       NULL, NULL_TREE);
+  gcc_assert (decl != NULL_TREE);
+  ix86_builtins[(int) code] = decl;
+  TREE_READONLY (decl) = 1;
+}
+
+/* Make builtins to get CPU type and features supported.  The created
+   builtins are :
+
+   __builtin_cpu_init (), to detect cpu type and features,
+   __builtin_cpu_is ("<CPUNAME>"), to check if cpu is of type <CPUNAME>,
+   __builtin_cpu_supports ("<FEATURE>"), to check if cpu supports <FEATURE>
+   */
+
+static void
+ix86_init_platform_type_builtins (void)
+{
+  make_cpu_type_builtin ("__builtin_cpu_init", IX86_BUILTIN_CPU_INIT,
+			 INT_FTYPE_VOID);
+  make_cpu_type_builtin ("__builtin_cpu_is", IX86_BUILTIN_CPU_IS,
+			 INT_FTYPE_PCCHAR);
+  make_cpu_type_builtin ("__builtin_cpu_supports", IX86_BUILTIN_CPU_SUPPORTS,
+			 INT_FTYPE_PCCHAR);
+}
+
 /* Internal method for ix86_init_builtins.  */
 
 static void
@@ -27710,6 +28070,9 @@ ix86_init_builtins (void)
 
   ix86_init_builtin_types ();
 
+  /* Builtins to get CPU type and features. */
+  ix86_init_platform_type_builtins ();
+
   /* TFmode support builtins.  */
   def_builtin_const (0, "__builtin_infq",
 		     FLOAT128_FTYPE_VOID, IX86_BUILTIN_INFQ);
@@ -29328,6 +29691,28 @@ ix86_expand_builtin (tree exp, rtx target, rtx sub
   enum machine_mode mode0, mode1, mode2, mode3, mode4;
   unsigned int fcode = DECL_FUNCTION_CODE (fndecl);
 
+  /* For CPU builtins that can be folded, fold first and expand the fold.  */
+  switch (fcode)
+    {
+    case IX86_BUILTIN_CPU_INIT:
+      {
+	/* Make it call __cpu_indicator_init in libgcc. */
+	tree call_expr, fndecl, type;
+        type = build_function_type_list (integer_type_node, NULL_TREE); 
+	fndecl = build_fn_decl ("__cpu_indicator_init", type);
+	call_expr = build_call_expr (fndecl, 0); 
+	return expand_expr (call_expr, target, mode, EXPAND_NORMAL);
+      }
+    case IX86_BUILTIN_CPU_IS:
+    case IX86_BUILTIN_CPU_SUPPORTS:
+      {
+	tree arg0 = CALL_EXPR_ARG (exp, 0);
+	tree fold_expr = fold_builtin_cpu (fndecl, &arg0);
+	gcc_assert (fold_expr != NULL_TREE);
+        return expand_expr (fold_expr, target, mode, EXPAND_NORMAL);
+      }
+    }
+
   /* Determine whether the builtin function is available under the current ISA.
      Originally the builtin was not created if it wasn't applicable to the
      current ISA based on the command line switches.  With function specific
@@ -39058,6 +39443,9 @@ ix86_autovectorize_vector_sizes (void)
 #undef TARGET_BUILD_BUILTIN_VA_LIST
 #define TARGET_BUILD_BUILTIN_VA_LIST ix86_build_builtin_va_list
 
+#undef TARGET_FOLD_BUILTIN
+#define TARGET_FOLD_BUILTIN ix86_fold_builtin
+
 #undef TARGET_ENUM_VA_LIST_P
 #define TARGET_ENUM_VA_LIST_P ix86_enum_va_list
 
Index: gcc/doc/extend.texi
===================================================================
--- gcc/doc/extend.texi	(revision 185898)
+++ gcc/doc/extend.texi	(working copy)
@@ -9413,6 +9413,140 @@ Similar to @code{__builtin_huge_val}, except the r
 @findex __builtin_huge_valq
 @end table
 
+The following built-in functions are always available and can be used to
+check the target platform type.
+
+@deftypefn {Built-in Function} void __builtin_cpu_init (void)
+This function runs the CPU detection code to check the type of CPU and the features
+supported. This builtin needs to be invoked along with the builtins to check CPU type
+and features, @code{__builtin_cpu_is} and @code{__builtin_cpu_supports}, only when used
+used in a function that will be executed before any constructors are called.  The
+CPU detection code is automatically executed in a very high priority constructor.
+
+For example, this function has to be used in @code{ifunc} resolvers which check for
+CPU type using the builtins, @code{__builtin_cpu_is} and @code{__builtin_cpu_supports}.
+@smallexample
+
+static void (*resolve_memcpy (void)) (void)
+@{
+  // ifunc resolvers fire before constructors, explicitly call the init
+  // function.
+  __builtin_cpu_init ();
+  if (__builtin_cpu_supports ("ssse3"))
+    return ssse3_memcpy; // super fast memcpy with ssse3 instructions.
+  else
+    return default_memcpy;
+@}
+
+void *memcpy (void *, const void *, size_t)
+     __attribute__ ((ifunc ("resolve_memcpy")));
+@end smallexample
+
+@end deftypefn
+
+@deftypefn {Built-in Function} int __builtin_cpu_is (const char *@var{cpuname})
+This function returns @code{1}, if the runtime cpu is of type @var{cpuname}
+ and returns @code{0} otherwise. The following cpu names can be detected:
+
+@table @samp
+@item intel
+Intel CPU.
+
+@item atom
+Intel ATOM CPU.
+
+@item core2
+Intel Core2 CPU.
+
+@item corei7
+Intel Corei7 CPU.
+
+@item nehalem
+Intel Corei7 Nehalem CPU.
+
+@item westmere
+Intel Corei7 Westmere CPU.
+
+@item sandybridge
+Intel Corei7 Sandybridge CPU.
+
+@item amd
+AMD CPU.
+
+@item amdfam10h
+AMD family 10h CPU.
+
+@item barcelona
+AMD family 10h Barcelona CPU.
+
+@item shanghai
+AMD family 10h Shanghai CPU.
+
+@item istanbul
+AMD family 15h Istanbul CPU.
+
+@item amdfam15h
+AMD family 15h CPU.
+
+@item bdver1
+AMD family 15h Bulldozer version 1.
+
+@item bdver2
+AMD family 15h Bulldozer version 1.
+@end table 
+
+Here is an example:
+@smallexample
+if (__builtin_cpu_is ("corei7"))
+  @{
+     do_corei7 (); //Corei7 specific implementation.
+  @}
+else
+  @{
+     do_generic (); //Generic implementation.
+  @}  
+@end smallexample
+@end deftypefn
+
+@deftypefn {Built-in Function} int __builtin_cpu_supports (const char *@var{feature})
+This function returns @code{1}, if the runtime cpu supports @var{feature}
+ and returns @code{0} otherwise. The following features can be detected:
+
+@table @samp
+@item cmov
+CMOV instruction.
+@item mmx
+MMX instructions.
+@item popcnt
+POPCNT instruction.
+@item sse
+SSE instructions.
+@item sse2
+SSE2 instructions.
+@item sse3
+SSE3 instructions.
+@item ssse3
+SSSE3 instructions.
+@item sse4.1
+SSE4.1 instructions.
+@item sse4.2
+SSE4.2 instructions.
+@end table
+
+Here is an example:
+@smallexample
+if (__builtin_cpu_supports ("popcnt"))
+  @{
+     asm("popcnt %1,%0" : "=r"(count) : "rm"(n) : "cc");
+  @}
+else
+  @{
+     count = generic_countbits (n); //generic implementation.
+  @}  
+@end smallexample
+@end deftypefn
+
+
 The following built-in functions are made available by @option{-mmmx}.
 All of them generate the machine instruction that is part of the name.
 

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

Ping.

On Tue, Apr 3, 2012 at 12:47 PM, Sriraman Tallam <tmsriram@google.com> wrote:
> Hi,
>
> i386 maintainers -  Is this patch ok?
>
> Thanks,
> -Sri.
>
> On Mon, Apr 2, 2012 at 5:48 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>> On Mon, Apr 2, 2012 at 5:38 AM, Richard Guenther
>> <richard.guenther@gmail.com> wrote:
>>> On Sat, Mar 31, 2012 at 1:03 AM, Sriraman Tallam <tmsriram@google.com> wrote:
>>>> On Fri, Mar 30, 2012 at 5:47 AM, Michael Matz <matz@suse.de> wrote:
>>>>> Hi,
>>>>>
>>>>> On Thu, 29 Mar 2012, Sriraman Tallam wrote:
>>>>>
>>>>>> +struct __processor_model
>>>>>> +{
>>>>>> +  /* Vendor. */
>>>>>> +  unsigned int __cpu_is_amd : 1;
>>>>>> +  unsigned int __cpu_is_intel : 1;
>>>>>> +  /* CPU type. */
>>>>>> +  unsigned int __cpu_is_intel_atom : 1;
>>>>>> +  unsigned int __cpu_is_intel_core2 : 1;
>>>>>> +  unsigned int __cpu_is_intel_corei7 : 1;
>>>>>> +  unsigned int __cpu_is_intel_corei7_nehalem : 1;
>>>>>> +  unsigned int __cpu_is_intel_corei7_westmere : 1;
>>>>>> +  unsigned int __cpu_is_intel_corei7_sandybridge : 1;
>>>>>> +  unsigned int __cpu_is_amdfam10h : 1;
>>>>>> +  unsigned int __cpu_is_amdfam10h_barcelona : 1;
>>>>>> +  unsigned int __cpu_is_amdfam10h_shanghai : 1;
>>>>>> +  unsigned int __cpu_is_amdfam10h_istanbul : 1;
>>>>>> +  unsigned int __cpu_is_amdfam15h_bdver1 : 1;
>>>>>> +  unsigned int __cpu_is_amdfam15h_bdver2 : 1;
>>>>>> +} __cpu_model;
>>>>>
>>>>> It doesn't make sense for the model to be a bitfield, a processor will
>>>>> have only ever exactly one model.  Just make it an enum or even just an
>>>>> int.
>>>>
>>>> Not entirely true, nehalem and corei7 can be both set. However, I
>>>> modified this by dividing it into types and sub types and then did
>>>> what you said.
>>>
>>> Uh...  then I suppose you need to document somewhere what names
>>> match to what cpuid family/model (supposedly thats where your two-layer
>>> hierarchy comes from, which incidentially misses one layer, the vendor?)
>>
>> Added documentation to extend.texi
>>
>> Patch available for review here:
>> http://codereview.appspot.com/5754058
>>
>> Thanks,
>> -Sri.
>>
>>
>>>
>>> Richard.
>>>
>>>>       * config/i386/i386.c (build_processor_features_struct): New function.
>>>>       (build_processor_model_struct): New function.
>>>>       (make_var_decl): New function.
>>>>       (get_field_from_struct): New function.
>>>>       (fold_builtin_target): New function.
>>>>       (ix86_fold_builtin): New function.
>>>>       (ix86_expand_builtin): Expand new builtins by folding them.
>>>>       (make_cpu_type_builtin): New functions.
>>>>       (ix86_init_platform_type_builtins): Make the new builtins.
>>>>       (ix86_init_builtins): Make new builtins to detect CPU type.
>>>>       (TARGET_FOLD_BUILTIN): New macro.
>>>>       (IX86_BUILTIN_CPU_INIT): New enum value.
>>>>       (IX86_BUILTIN_CPU_IS): New enum value.
>>>>       (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>>>>       * config/i386/i386-builtin-types.def: New function type.
>>>>       * testsuite/gcc.target/builtin_target.c: New testcase.
>>>>
>>>>       * libgcc/config/i386/i386-cpuinfo.c: New file.
>>>>       * libgcc/config/i386/t-cpuinfo: New file.
>>>>       * libgcc/config.host: Include t-cpuinfo.
>>>>       * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
>>>>       and __cpu_features.
>>>>
>>>> Patch available for review here:
>>>> http://codereview.appspot.com/5754058
>>>>
>>>> Thanks,
>>>> -Sri.
>>>>
>>>>
>>>>>
>>>>>
>>>>> Ciao,
>>>>> Michael.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

Ping.

On Thu, Apr 12, 2012 at 4:14 PM, Sriraman Tallam <tmsriram@google.com> wrote:
> Ping.
>
> On Tue, Apr 3, 2012 at 12:47 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>> Hi,
>>
>> i386 maintainers -  Is this patch ok?
>>
>> Thanks,
>> -Sri.
>>
>> On Mon, Apr 2, 2012 at 5:48 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>>> On Mon, Apr 2, 2012 at 5:38 AM, Richard Guenther
>>> <richard.guenther@gmail.com> wrote:
>>>> On Sat, Mar 31, 2012 at 1:03 AM, Sriraman Tallam <tmsriram@google.com> wrote:
>>>>> On Fri, Mar 30, 2012 at 5:47 AM, Michael Matz <matz@suse.de> wrote:
>>>>>> Hi,
>>>>>>
>>>>>> On Thu, 29 Mar 2012, Sriraman Tallam wrote:
>>>>>>
>>>>>>> +struct __processor_model
>>>>>>> +{
>>>>>>> +  /* Vendor. */
>>>>>>> +  unsigned int __cpu_is_amd : 1;
>>>>>>> +  unsigned int __cpu_is_intel : 1;
>>>>>>> +  /* CPU type. */
>>>>>>> +  unsigned int __cpu_is_intel_atom : 1;
>>>>>>> +  unsigned int __cpu_is_intel_core2 : 1;
>>>>>>> +  unsigned int __cpu_is_intel_corei7 : 1;
>>>>>>> +  unsigned int __cpu_is_intel_corei7_nehalem : 1;
>>>>>>> +  unsigned int __cpu_is_intel_corei7_westmere : 1;
>>>>>>> +  unsigned int __cpu_is_intel_corei7_sandybridge : 1;
>>>>>>> +  unsigned int __cpu_is_amdfam10h : 1;
>>>>>>> +  unsigned int __cpu_is_amdfam10h_barcelona : 1;
>>>>>>> +  unsigned int __cpu_is_amdfam10h_shanghai : 1;
>>>>>>> +  unsigned int __cpu_is_amdfam10h_istanbul : 1;
>>>>>>> +  unsigned int __cpu_is_amdfam15h_bdver1 : 1;
>>>>>>> +  unsigned int __cpu_is_amdfam15h_bdver2 : 1;
>>>>>>> +} __cpu_model;
>>>>>>
>>>>>> It doesn't make sense for the model to be a bitfield, a processor will
>>>>>> have only ever exactly one model.  Just make it an enum or even just an
>>>>>> int.
>>>>>
>>>>> Not entirely true, nehalem and corei7 can be both set. However, I
>>>>> modified this by dividing it into types and sub types and then did
>>>>> what you said.
>>>>
>>>> Uh...  then I suppose you need to document somewhere what names
>>>> match to what cpuid family/model (supposedly thats where your two-layer
>>>> hierarchy comes from, which incidentially misses one layer, the vendor?)
>>>
>>> Added documentation to extend.texi
>>>
>>> Patch available for review here:
>>> http://codereview.appspot.com/5754058
>>>
>>> Thanks,
>>> -Sri.
>>>
>>>
>>>>
>>>> Richard.
>>>>
>>>>>       * config/i386/i386.c (build_processor_features_struct): New function.
>>>>>       (build_processor_model_struct): New function.
>>>>>       (make_var_decl): New function.
>>>>>       (get_field_from_struct): New function.
>>>>>       (fold_builtin_target): New function.
>>>>>       (ix86_fold_builtin): New function.
>>>>>       (ix86_expand_builtin): Expand new builtins by folding them.
>>>>>       (make_cpu_type_builtin): New functions.
>>>>>       (ix86_init_platform_type_builtins): Make the new builtins.
>>>>>       (ix86_init_builtins): Make new builtins to detect CPU type.
>>>>>       (TARGET_FOLD_BUILTIN): New macro.
>>>>>       (IX86_BUILTIN_CPU_INIT): New enum value.
>>>>>       (IX86_BUILTIN_CPU_IS): New enum value.
>>>>>       (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>>>>>       * config/i386/i386-builtin-types.def: New function type.
>>>>>       * testsuite/gcc.target/builtin_target.c: New testcase.
>>>>>
>>>>>       * libgcc/config/i386/i386-cpuinfo.c: New file.
>>>>>       * libgcc/config/i386/t-cpuinfo: New file.
>>>>>       * libgcc/config.host: Include t-cpuinfo.
>>>>>       * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
>>>>>       and __cpu_features.
>>>>>
>>>>> Patch available for review here:
>>>>> http://codereview.appspot.com/5754058
>>>>>
>>>>> Thanks,
>>>>> -Sri.
>>>>>
>>>>>
>>>>>>
>>>>>>
>>>>>> Ciao,
>>>>>> Michael.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Uros Bizjak]

On Tue, Apr 3, 2012 at 9:47 PM, Sriraman Tallam <tmsriram@google.com> wrote:

> i386 maintainers -  Is this patch ok?

Has the community reached the consensus on how this kind of
functionality has to be implemented? I have followed the discussion a
bit, but IIRC, there was no clear decision. Without this decision, I
am not able to review the _implementation_ of agreed functionality for
x86 target.

(I apologize if I have missed the decision, please point me to the
discussion in this case.)

>>>> Patch available for review here:
>>>> http://codereview.appspot.com/5754058

Please attach patches or inline it in the message itself for a review.
Please see [1] for further instructions.

[1] http://gcc.gnu.org/contribute.html#patches

Uros.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

[-- Attachment #1: Type: text/plain, Size: 2733 bytes --]

Hi,

On Mon, Apr 23, 2012 at 1:19 AM, Uros Bizjak <ubizjak@gmail.com> wrote:
> On Tue, Apr 3, 2012 at 9:47 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>
>> i386 maintainers -  Is this patch ok?
>
> Has the community reached the consensus on how this kind of
> functionality has to be implemented? I have followed the discussion a
> bit, but IIRC, there was no clear decision. Without this decision, I
> am not able to review the _implementation_ of agreed functionality for
> x86 target.
>
> (I apologize if I have missed the decision, please point me to the
> discussion in this case.)

The discussions are here:

http://gcc.gnu.org/ml/gcc-patches/2011-08/msg01446.html
and follow-ups to this.

I am not sure about consensus, but the important points raised were:

1) Constructor ordering: What if some constructors fire before
cpu_indicator_init?, which determines the CPU. I addressed this
problem by making the priority of cpu_indicator_init to be the highest
possible. Still, IFUNC initializers will fire before and they have to
explicitly call __builtin_cpu_init() before checking the CPU type.
2) Reducing the number of builtins : It is only two now.


>
>>>>> Patch available for review here:
>>>>> http://codereview.appspot.com/5754058
>
> Please attach patches or inline it in the message itself for a review.
> Please see [1] for further instructions.

Patch attached, tested on x86_64 and all tests pass.


       * config/i386/i386.c (build_processor_features_struct): New function.
       (build_processor_model_struct): New function.
       (make_var_decl): New function.
       (get_field_from_struct): New function.
       (fold_builtin_target): New function.
       (ix86_fold_builtin): New function.
       (ix86_expand_builtin): Expand new builtins by folding them.
       (make_cpu_type_builtin): New functions.
       (ix86_init_platform_type_builtins): Make the new builtins.
       (ix86_init_builtins): Make new builtins to detect CPU type.
       (TARGET_FOLD_BUILTIN): New macro.
       (IX86_BUILTIN_CPU_INIT): New enum value.
       (IX86_BUILTIN_CPU_IS): New enum value.
       (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
       * config/i386/i386-builtin-types.def: New function type.
       * testsuite/gcc.target/builtin_target.c: New testcase.
       * doc/extend.texi: Document builtins.

       * libgcc/config/i386/i386-cpuinfo.c: New file.
       * libgcc/config/i386/t-cpuinfo: New file.
       * libgcc/config.host: Include t-cpuinfo.
       * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
       and __cpu_features.

Thanks,
-Sri.


>
> [1] http://gcc.gnu.org/contribute.html#patches
>
> Uros.

[-- Attachment #2: cpu_runtime_patch.txt --]
[-- Type: text/plain, Size: 28968 bytes --]


       * config/i386/i386.c (build_processor_features_struct): New function.
       (build_processor_model_struct): New function.
       (make_var_decl): New function.
       (get_field_from_struct): New function.
       (fold_builtin_target): New function.
       (ix86_fold_builtin): New function.
       (ix86_expand_builtin): Expand new builtins by folding them.
       (make_cpu_type_builtin): New functions.
       (ix86_init_platform_type_builtins): Make the new builtins.
       (ix86_init_builtins): Make new builtins to detect CPU type.
       (TARGET_FOLD_BUILTIN): New macro.
       (IX86_BUILTIN_CPU_INIT): New enum value.
       (IX86_BUILTIN_CPU_IS): New enum value.
       (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
       * config/i386/i386-builtin-types.def: New function type.
       * testsuite/gcc.target/builtin_target.c: New testcase.
       * doc/extend.texi: Document builtins.

       * libgcc/config/i386/i386-cpuinfo.c: New file.
       * libgcc/config/i386/t-cpuinfo: New file.
       * libgcc/config.host: Include t-cpuinfo.
       * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
       and __cpu_features.


Index: libgcc/config.host
===================================================================
--- libgcc/config.host	(revision 185898)
+++ libgcc/config.host	(working copy)
@@ -1130,7 +1130,7 @@ i[34567]86-*-linux* | x86_64-*-linux* | \
   i[34567]86-*-kfreebsd*-gnu | x86_64-*-kfreebsd*-gnu | \
   i[34567]86-*-knetbsd*-gnu | \
   i[34567]86-*-gnu*)
-	tmake_file="${tmake_file} t-tls i386/t-linux"
+	tmake_file="${tmake_file} t-tls i386/t-linux i386/t-cpuinfo"
 	if test "$libgcc_cv_cfi" = "yes"; then
 		tmake_file="${tmake_file} t-stack i386/t-stack-i386"
 	fi
Index: libgcc/config/i386/t-cpuinfo
===================================================================
--- libgcc/config/i386/t-cpuinfo	(revision 0)
+++ libgcc/config/i386/t-cpuinfo	(revision 0)
@@ -0,0 +1 @@
+LIB2ADD += $(srcdir)/config/i386/i386-cpuinfo.c
Index: libgcc/config/i386/i386-cpuinfo.c
===================================================================
--- libgcc/config/i386/i386-cpuinfo.c	(revision 0)
+++ libgcc/config/i386/i386-cpuinfo.c	(revision 0)
@@ -0,0 +1,298 @@
+/* Get CPU type and Features for x86 processors.
+   Copyright (C) 2012 Free Software Foundation, Inc.
+   Contributed by Sriraman Tallam (tmsriram@google.com)
+
+This file is part of GCC.
+
+GCC is free software; you can redistribute it and/or modify it under
+the terms of the GNU General Public License as published by the Free
+Software Foundation; either version 3, or (at your option) any later
+version.
+
+GCC is distributed in the hope that it will be useful, but WITHOUT ANY
+WARRANTY; without even the implied warranty of MERCHANTABILITY or
+FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+for more details.
+
+You should have received a copy of the GNU General Public License
+along with GCC; see the file COPYING3.  If not see
+<http://www.gnu.org/licenses/>. */
+
+#include "cpuid.h"
+#include "tsystem.h"
+
+int __cpu_indicator_init (void) __attribute__ ((constructor (101)));
+
+enum vendor_signatures
+{
+  SIG_INTEL =	0x756e6547 /* Genu */,
+  SIG_AMD =	0x68747541 /* Auth */
+};
+
+/* ISA Features supported. */
+
+struct __processor_features
+{
+  unsigned int __cpu_cmov : 1;
+  unsigned int __cpu_mmx : 1;
+  unsigned int __cpu_popcnt : 1;
+  unsigned int __cpu_sse : 1;
+  unsigned int __cpu_sse2 : 1;
+  unsigned int __cpu_sse3 : 1;
+  unsigned int __cpu_ssse3 : 1;
+  unsigned int __cpu_sse4_1 : 1;
+  unsigned int __cpu_sse4_2 : 1;
+} __cpu_features;
+
+/* Processor Vendor and Models. */
+
+enum processor_vendor
+{
+  VENDOR_INTEL = 1,
+  VENDOR_AMD,
+  VENDOR_MAX
+};
+
+enum processor_types
+{
+  INTEL_ATOM = 1,
+  INTEL_CORE2,
+  INTEL_COREI7,
+  AMDFAM10H,
+  AMDFAM15H,
+  CPU_TYPE_MAX
+};
+
+enum processor_subtypes
+{
+  INTEL_COREI7_NEHALEM,
+  INTEL_COREI7_WESTMERE,
+  INTEL_COREI7_SANDYBRIDGE,
+  AMDFAM10H_BARCELONA,
+  AMDFAM10H_SHANGHAI,
+  AMDFAM10H_ISTANBUL,
+  AMDFAM15H_BDVER1,
+  AMDFAM15H_BDVER2,
+  CPU_SUBTYPE_MAX
+};
+
+struct __processor_model
+{
+  unsigned int __cpu_vendor;
+  unsigned int __cpu_type;
+  unsigned int __cpu_subtype;
+} __cpu_model;
+
+/* Get the specific type of AMD CPU.  */
+
+static void
+get_amd_cpu (unsigned int family, unsigned int model)
+{
+  __cpu_model.__cpu_vendor = VENDOR_AMD;
+  switch (family)
+    {
+    /* AMD Family 10h.  */
+    case 0x10:
+      switch (model)
+	{
+	case 0x2:
+	  /* Barcelona.  */
+	  __cpu_model.__cpu_type = AMDFAM10H;
+	  __cpu_model.__cpu_subtype = AMDFAM10H_BARCELONA;
+	  break;
+	case 0x4:
+	  /* Shanghai.  */
+	  __cpu_model.__cpu_type = AMDFAM10H;
+	  __cpu_model.__cpu_subtype = AMDFAM10H_SHANGHAI;
+	  break;
+	case 0x8:
+	  /* Istanbul.  */
+	  __cpu_model.__cpu_type = AMDFAM10H;
+	  __cpu_model.__cpu_subtype = AMDFAM10H_ISTANBUL;
+	  break;
+	default:
+	  break;
+	}
+      break;
+    /* AMD Family 15h.  */
+    case 0x15:
+      __cpu_model.__cpu_type = AMDFAM15H;
+      /* Bulldozer version 1.  */
+      if ( model <= 0xf)
+	__cpu_model.__cpu_subtype = AMDFAM15H_BDVER1;
+      /* Bulldozer version 2.  */
+      if (model >= 0x10 && model <= 0x1f)
+	__cpu_model.__cpu_subtype = AMDFAM15H_BDVER2;
+      break;
+    default:
+      break;
+    }
+}
+
+/* Get the specific type of Intel CPU.  */
+
+static void
+get_intel_cpu (unsigned int family, unsigned int model, unsigned int brand_id)
+{
+  __cpu_model.__cpu_vendor = VENDOR_INTEL;
+  /* Parse family and model only if brand ID is 0. */
+  if (brand_id == 0)
+    {
+      switch (family)
+	{
+	case 0x5:
+	  /* Pentium.  */
+	  break;
+	case 0x6:
+	  switch (model)
+	    {
+	    case 0x1c:
+	    case 0x26:
+	      /* Atom.  */
+	      __cpu_model.__cpu_type = INTEL_ATOM;
+	      break;
+	    case 0x1a:
+	    case 0x1e:
+	    case 0x1f:
+	    case 0x2e:
+	      /* Nehalem.  */
+	      __cpu_model.__cpu_type = INTEL_COREI7;
+	      __cpu_model.__cpu_subtype = INTEL_COREI7_NEHALEM;
+	      break;
+	    case 0x25:
+	    case 0x2c:
+	    case 0x2f:
+	      /* Westmere.  */
+	      __cpu_model.__cpu_type = INTEL_COREI7;
+	      __cpu_model.__cpu_subtype = INTEL_COREI7_WESTMERE;
+	      break;
+	    case 0x2a:
+	      /* Sandy Bridge.  */
+	      __cpu_model.__cpu_type = INTEL_COREI7;
+	      __cpu_model.__cpu_subtype = INTEL_COREI7_SANDYBRIDGE;
+	      break;
+	    case 0x17:
+	    case 0x1d:
+	      /* Penryn.  */
+	    case 0x0f:
+	      /* Merom.  */
+	      __cpu_model.__cpu_type = INTEL_CORE2;
+	      break;
+	    default:
+	      break;
+	    }
+	  break;
+	default:
+	  /* We have no idea.  */
+	  break;
+	}
+    }
+}	             	
+
+static void
+get_available_features (unsigned int ecx, unsigned int edx)
+{
+  __cpu_features.__cpu_cmov = (edx & bit_CMOV) ? 1 : 0;
+  __cpu_features.__cpu_mmx = (edx & bit_MMX) ? 1 : 0;
+  __cpu_features.__cpu_sse = (edx & bit_SSE) ? 1 : 0;
+  __cpu_features.__cpu_sse2 = (edx & bit_SSE2) ? 1 : 0;
+  __cpu_features.__cpu_popcnt = (ecx & bit_POPCNT) ? 1 : 0;
+  __cpu_features.__cpu_sse3 = (ecx & bit_SSE3) ? 1 : 0;
+  __cpu_features.__cpu_ssse3 = (ecx & bit_SSSE3) ? 1 : 0;
+  __cpu_features.__cpu_sse4_1 = (ecx & bit_SSE4_1) ? 1 : 0;
+  __cpu_features.__cpu_sse4_2 = (ecx & bit_SSE4_2) ? 1 : 0;
+}
+
+/* A noinline function calling __get_cpuid. Having many calls to
+   cpuid in one function in 32-bit mode causes GCC to complain:
+   "can;t find a register in class CLOBBERED_REGS".  This is
+   related to PR rtl-optimization 44174. */
+
+static int __attribute__ ((noinline))
+__get_cpuid_output (unsigned int __level,
+		    unsigned int *__eax, unsigned int *__ebx,
+		    unsigned int *__ecx, unsigned int *__edx)
+{
+  return __get_cpuid (__level, __eax, __ebx, __ecx, __edx);
+}
+
+
+/* A constructor function that is sets __cpu_model and __cpu_features with
+   the right values.  This needs to run only once.  This constructor is
+   given the highest priority and it should run before constructors without
+   the priority set.  However, it still runs after ifunc initializers and
+   needs to be called explicitly there.  */
+
+int __attribute__ ((constructor (101)))
+__cpu_indicator_init (void)
+{
+  unsigned int eax, ebx, ecx, edx;
+
+  int max_level = 5;
+  unsigned int vendor;
+  unsigned int model, family, brand_id;
+  unsigned int extended_model, extended_family;
+  static int called = 0;
+
+  /* This function needs to run just once.  */
+  if (called)
+    return 0;
+  else
+    called = 1;
+
+  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
+  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
+    return -1;
+
+  vendor = ebx;
+  max_level = eax;
+
+  if (max_level < 1)
+    return -1;
+
+  if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
+    return -1;
+
+  model = (eax >> 4) & 0x0f;
+  family = (eax >> 8) & 0x0f;
+  brand_id = ebx & 0xff;
+  extended_model = (eax >> 12) & 0xf0;
+  extended_family = (eax >> 20) & 0xff;
+
+  if (vendor == SIG_INTEL)
+    {
+      /* Adjust model and family for Intel CPUS. */
+      if (family == 0x0f)
+	{
+	  family += extended_family;
+	  model += extended_model;
+	}
+      else if (family == 0x06)
+	model += extended_model;
+
+      /* Get CPU type.  */
+      get_intel_cpu (family, model, brand_id);
+    }
+
+  if (vendor == SIG_AMD)
+    {
+      /* Adjust model and family for AMD CPUS. */
+      if (family == 0x0f)
+	{
+	  family += extended_family;
+	  model += (extended_model << 4);
+	}
+
+      /* Get CPU type.  */
+      get_amd_cpu (family, model);
+    }
+
+  /* Find available features. */
+  get_available_features (ecx, edx);
+
+  gcc_assert (__cpu_model.__cpu_vendor < VENDOR_MAX);
+  gcc_assert (__cpu_model.__cpu_type < CPU_TYPE_MAX);
+  gcc_assert (__cpu_model.__cpu_subtype < CPU_SUBTYPE_MAX);
+
+  return 0;
+}
Index: libgcc/config/i386/libgcc-glibc.ver
===================================================================
--- libgcc/config/i386/libgcc-glibc.ver	(revision 185898)
+++ libgcc/config/i386/libgcc-glibc.ver	(working copy)
@@ -147,6 +147,11 @@ GCC_4.3.0 {
   __trunctfxf2
   __unordtf2
 }
+
+GCC_4.8.0 {
+  __cpu_model
+  __cpu_features
+}
 %else
 GCC_4.4.0 {
   __addtf3
@@ -183,4 +188,9 @@ GCC_4.4.0 {
 GCC_4.5.0 {
   __extendxftf2
 }
+
+GCC_4.8.0 {
+  __cpu_model
+  __cpu_features
+}
 %endif
Index: gcc/testsuite/gcc.target/i386/builtin_target.c
===================================================================
--- gcc/testsuite/gcc.target/i386/builtin_target.c	(revision 0)
+++ gcc/testsuite/gcc.target/i386/builtin_target.c	(revision 0)
@@ -0,0 +1,65 @@
+/* This test checks if the __builtin_cpu_is and __builtin_cpu_supports calls
+   are recognized. */
+
+/* { dg-do run } */
+
+int
+fn1 ()
+{
+  if (__builtin_cpu_supports ("cmov") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("mmx") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("popcnt") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse2") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse3") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("ssse3") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse4.1") < 0)
+    return -1;
+  if (__builtin_cpu_supports ("sse4.2") < 0)
+    return -1;
+  if (__builtin_cpu_is ("amd") < 0)
+    return -1;
+  if (__builtin_cpu_is ("intel") < 0)
+    return -1;
+  if (__builtin_cpu_is ("atom") < 0)
+    return -1;
+  if (__builtin_cpu_is ("core2") < 0)
+    return -1;
+  if (__builtin_cpu_is ("corei7") < 0)
+    return -1;
+  if (__builtin_cpu_is ("nehalem") < 0)
+    return -1;
+  if (__builtin_cpu_is ("westmere") < 0)
+    return -1;
+  if (__builtin_cpu_is ("sandybridge") < 0)
+    return -1;
+  if (__builtin_cpu_is ("amdfam10h") < 0)
+    return -1;
+  if (__builtin_cpu_is ("barcelona") < 0)
+    return -1;
+  if (__builtin_cpu_is ("shanghai") < 0)
+    return -1;
+  if (__builtin_cpu_is ("istanbul") < 0)
+    return -1;
+  if (__builtin_cpu_is ("amdfam15h") < 0)
+    return -1;
+  if (__builtin_cpu_is ("bdver1") < 0)
+    return -1;
+  if (__builtin_cpu_is ("bdver2") < 0)
+    return -1;
+
+  return 0;
+}
+
+int main ()
+{
+  __builtin_cpu_init ();
+  return fn1 ();
+}
Index: gcc/config/i386/i386-builtin-types.def
===================================================================
--- gcc/config/i386/i386-builtin-types.def	(revision 185898)
+++ gcc/config/i386/i386-builtin-types.def	(working copy)
@@ -155,6 +155,7 @@ DEF_FUNCTION_TYPE (INT, V4SF)
 DEF_FUNCTION_TYPE (INT, V8QI)
 DEF_FUNCTION_TYPE (INT, V8SF)
 DEF_FUNCTION_TYPE (INT, V32QI)
+DEF_FUNCTION_TYPE (INT, PCCHAR)
 DEF_FUNCTION_TYPE (INT64, INT64)
 DEF_FUNCTION_TYPE (INT64, V2DF)
 DEF_FUNCTION_TYPE (INT64, V4SF)
Index: gcc/config/i386/i386.c
===================================================================
--- gcc/config/i386/i386.c	(revision 185898)
+++ gcc/config/i386/i386.c	(working copy)
@@ -25809,6 +25809,11 @@ enum ix86_builtins
   /* CFString built-in for darwin */
   IX86_BUILTIN_CFSTRING,
 
+  /* Builtins to get CPU type and supported features. */
+  IX86_BUILTIN_CPU_INIT,
+  IX86_BUILTIN_CPU_IS,
+  IX86_BUILTIN_CPU_SUPPORTS,
+
   IX86_BUILTIN_MAX
 };
 
@@ -27627,6 +27632,361 @@ ix86_init_mmx_sse_builtins (void)
     }
 }
 
+/* This builds the processor_model struct type defined in
+   libgcc/config/i386/i386-cpuinfo.c  */
+static tree
+build_processor_model_struct (void)
+{
+  const char *field_name[] = {"__cpu_vendor", "__cpu_type", "__cpu_subtype"};
+  tree field = NULL_TREE, field_chain = NULL_TREE;
+  int i;
+  tree type = make_node (RECORD_TYPE);
+  int num_fields = sizeof (field_name) / sizeof (const char *);
+
+  for (i = 0; i < num_fields; ++i)
+    {
+      field = build_decl (UNKNOWN_LOCATION, FIELD_DECL,
+			  get_identifier (field_name[i]), unsigned_type_node);
+      if (field_chain != NULL_TREE)
+	DECL_CHAIN (field) = field_chain;
+      field_chain = field; 
+    }
+  finish_builtin_struct (type, "__processor_model", field_chain, NULL_TREE);
+  return type;
+}
+
+/* This builds the processor_features struct type defined in
+   libgcc/config/i386/i386-cpuinfo.c  */
+
+static tree
+build_processor_features_struct (void)
+{
+  int i;
+  const char *field_name[] = {"__cpu_cmov", "__cpu_mmx", "__cpu_popcnt",
+			      "__cpu_sse", "__cpu_sse2", "__cpu_sse3",
+			      "__cpu_ssse3", "__cpu_sse4_1", "__cpu_sse4_2"};
+
+  tree field = NULL_TREE, field_chain = NULL_TREE;
+  tree type = make_node (RECORD_TYPE);
+  int num_fields = sizeof (field_name) / sizeof (const char *);
+  
+  for (i = 0; i < num_fields; i++)
+    {
+      field = build_decl (UNKNOWN_LOCATION, FIELD_DECL,
+			  get_identifier (field_name[i]), unsigned_type_node);
+      DECL_BIT_FIELD (field) = 1;
+      DECL_SIZE (field) = bitsize_one_node;
+      if (field_chain != NULL_TREE)
+	DECL_CHAIN (field) = field_chain;
+      field_chain = field;
+    }
+  finish_builtin_struct (type, "__processor_features", field_chain, NULL_TREE);
+  return type;
+}
+
+/* Returns a extern, comdat VAR_DECL of type TYPE and name NAME. */
+
+static tree
+make_var_decl (tree type, const char *name)
+{
+  tree new_decl;
+
+  new_decl = build_decl (UNKNOWN_LOCATION,
+	                 VAR_DECL,
+	  	         get_identifier(name),
+		         type);
+
+  DECL_EXTERNAL (new_decl) = 1;
+  TREE_STATIC (new_decl) = 1;
+  TREE_PUBLIC (new_decl) = 1;
+  DECL_INITIAL (new_decl) = 0;
+  DECL_ARTIFICIAL (new_decl) = 0;
+  DECL_PRESERVE_P (new_decl) = 1;
+
+  make_decl_one_only (new_decl, DECL_ASSEMBLER_NAME (new_decl));
+  assemble_variable (new_decl, 0, 0, 0);
+
+  return new_decl;
+}
+
+/* Traverses the chain of fields in STRUCT_TYPE and returns the FIELD_NUM
+   numbered field. */
+
+static tree
+get_field_from_struct (tree struct_type, int field_num)
+{
+  int i;
+  tree field = TYPE_FIELDS (struct_type);
+
+  for (i = 0; i < field_num; i++, field = DECL_CHAIN(field))
+    {
+      gcc_assert (field != NULL_TREE);
+    }
+
+  return field;
+}
+
+/* FNDECL is a __builtin_cpu_is or a __builtin_cpu_supports call that is folded
+   into an integer defined in libgcc/config/i386/i386-cpuinfo.c */
+
+static tree 
+fold_builtin_cpu (tree fndecl, tree *args)
+{
+  unsigned int i;
+  enum ix86_builtins fn_code = (enum ix86_builtins)
+				DECL_FUNCTION_CODE (fndecl);
+  tree param_string_cst = NULL;
+
+  /* This is the order of bit-fields in __processor_features in
+     i386-cpuinfo.c */
+  enum processor_features
+  {
+    F_CMOV = 0,
+    F_MMX,
+    F_POPCNT,
+    F_SSE,
+    F_SSE2,
+    F_SSE3,
+    F_SSSE3,
+    F_SSE4_1,
+    F_SSE4_2,
+    F_MAX
+  };
+
+  /* These are the values for vendor types and cpu types  and subtypes
+     in i386-cpuinfo.c.  Cpu types and subtypes should be subtracted by
+     the corresponding start value.  */
+  enum processor_model
+  {
+    M_INTEL = 1,
+    M_AMD,
+    M_CPU_TYPE_START,
+    M_INTEL_ATOM,
+    M_INTEL_CORE2,
+    M_INTEL_COREI7,
+    M_AMDFAM10H,
+    M_AMDFAM15H,
+    M_CPU_SUBTYPE_START,
+    M_INTEL_COREI7_NEHALEM,
+    M_INTEL_COREI7_WESTMERE,
+    M_INTEL_COREI7_SANDYBRIDGE,
+    M_AMDFAM10H_BARCELONA,
+    M_AMDFAM10H_SHANGHAI,
+    M_AMDFAM10H_ISTANBUL,
+    M_AMDFAM15H_BDVER1,
+    M_AMDFAM15H_BDVER2
+  };
+
+  static struct _arch_names_table
+    {
+      const char *const name;
+      const enum processor_model model;
+    }
+  const arch_names_table[] =
+    {
+      {"amd", M_AMD},
+      {"intel", M_INTEL},
+      {"atom", M_INTEL_ATOM},
+      {"core2", M_INTEL_CORE2},
+      {"corei7", M_INTEL_COREI7},
+      {"nehalem", M_INTEL_COREI7_NEHALEM},
+      {"westmere", M_INTEL_COREI7_WESTMERE},
+      {"sandybridge", M_INTEL_COREI7_SANDYBRIDGE},
+      {"amdfam10h", M_AMDFAM10H},
+      {"barcelona", M_AMDFAM10H_BARCELONA},
+      {"shanghai", M_AMDFAM10H_SHANGHAI},
+      {"istanbul", M_AMDFAM10H_ISTANBUL},
+      {"amdfam15h", M_AMDFAM15H},
+      {"bdver1", M_AMDFAM15H_BDVER1},
+      {"bdver2", M_AMDFAM15H_BDVER2},
+    };
+
+  static struct _isa_names_table
+    {
+      const char *const name;
+      const enum processor_features feature;
+    }
+  const isa_names_table[] =
+    {
+      {"cmov",   F_CMOV},
+      {"mmx",    F_MMX},
+      {"popcnt", F_POPCNT},
+      {"sse",    F_SSE},
+      {"sse2",   F_SSE2},
+      {"sse3",   F_SSE3},
+      {"ssse3",  F_SSSE3},
+      {"sse4.1", F_SSE4_1},
+      {"sse4.2", F_SSE4_2}
+    };
+
+  static tree __processor_features_type = NULL_TREE;
+  static tree __cpu_features_var = NULL_TREE;
+  static tree __processor_model_type = NULL_TREE;
+  static tree __cpu_model_var = NULL_TREE;
+  static tree field;
+  static tree which_struct;
+
+  if (__processor_features_type == NULL_TREE)
+    __processor_features_type = build_processor_features_struct ();
+
+  if (__processor_model_type == NULL_TREE)
+    __processor_model_type = build_processor_model_struct ();
+
+  if (__cpu_features_var == NULL_TREE)
+    __cpu_features_var = make_var_decl (__processor_features_type,
+					"__cpu_features");
+
+  if (__cpu_model_var == NULL_TREE)
+    __cpu_model_var = make_var_decl (__processor_model_type,
+				     "__cpu_model");
+
+  gcc_assert ((args != NULL) && (*args != NULL));
+
+  param_string_cst = *args;
+  while (param_string_cst
+	 && TREE_CODE (param_string_cst) !=  STRING_CST)
+    {
+      /* *args must be a expr that can contain other EXPRS leading to a
+	 STRING_CST.   */
+      if (!EXPR_P (param_string_cst))
+ 	{
+	  error ("Parameter to builtin must be a string constant or literal");
+	  return integer_zero_node;
+	}
+      param_string_cst = TREE_OPERAND (EXPR_CHECK (param_string_cst), 0);
+    }
+ 
+  gcc_assert (param_string_cst);
+
+  if (fn_code == IX86_BUILTIN_CPU_IS)
+    {
+      tree ref, expr;
+      unsigned int field_val = 0;
+      unsigned int NUM_ARCH_NAMES
+	= sizeof (arch_names_table) / sizeof (struct _arch_names_table);
+
+      for (i = 0; i < NUM_ARCH_NAMES; i++)
+	if (strcmp (arch_names_table[i].name,
+	    TREE_STRING_POINTER (param_string_cst)) == 0)
+	  break;
+
+      if (i == NUM_ARCH_NAMES)
+	{
+	  error ("Parameter to builtin not valid: %s",
+	         TREE_STRING_POINTER (param_string_cst));
+	  return integer_zero_node;
+	}
+
+      field = TYPE_FIELDS (__processor_model_type);
+      field_val = arch_names_table[i].model;
+
+      /* CPU types are stored in the next field.  */
+      if (field_val > M_CPU_TYPE_START
+	  && field_val < M_CPU_SUBTYPE_START)
+	{
+	  field = DECL_CHAIN (field);
+	  field_val -= M_CPU_TYPE_START;
+	}
+
+      /* CPU subtypes are stored in the last field.  */
+      if (field_val > M_CPU_SUBTYPE_START)
+	{
+	  field = DECL_CHAIN ( DECL_CHAIN (field));
+	  field_val -= M_CPU_SUBTYPE_START;
+	}
+
+      /* Get the appropriate field in __cpu_model.  */
+      ref =  build3 (COMPONENT_REF, TREE_TYPE (field), __cpu_model_var,
+		     field, NULL_TREE);
+
+      /* Check the value.  */
+      return build2 (EQ_EXPR, unsigned_type_node, ref,
+		     build_int_cstu (unsigned_type_node, field_val));
+    }
+  else if (fn_code == IX86_BUILTIN_CPU_SUPPORTS)
+    {
+      unsigned int NUM_ISA_NAMES
+	= sizeof (isa_names_table) / sizeof (struct _isa_names_table);
+      for (i = 0; i < NUM_ISA_NAMES; i++)
+	if (strcmp (isa_names_table[i].name,
+	    TREE_STRING_POINTER (param_string_cst)) == 0)
+	  break;
+
+      if (i == NUM_ISA_NAMES)
+	{
+	  error ("Parameter to builtin not valid: %s",
+	       	 TREE_STRING_POINTER (param_string_cst));
+	  return integer_zero_node;
+	}
+
+      field = get_field_from_struct (__processor_features_type,
+				     isa_names_table[i].feature);
+      which_struct = __cpu_features_var;
+      return build3 (COMPONENT_REF, TREE_TYPE (field), which_struct,
+		     field, NULL_TREE);
+    }
+
+  gcc_unreachable ();
+
+}
+
+static tree
+ix86_fold_builtin (tree fndecl, int n_args,
+		   tree *args, bool ignore ATTRIBUTE_UNUSED)
+{
+  if (DECL_BUILT_IN_CLASS (fndecl) == BUILT_IN_MD)
+    {
+      enum ix86_builtins fn_code = (enum ix86_builtins)
+				   DECL_FUNCTION_CODE (fndecl);
+      if (fn_code ==  IX86_BUILTIN_CPU_IS
+	  || fn_code == IX86_BUILTIN_CPU_SUPPORTS)
+	{
+	  gcc_assert (n_args == 1);
+          return fold_builtin_cpu (fndecl, args);
+	}
+    }
+
+  return NULL_TREE;
+}
+
+/* Make builtins to detect cpu type and features supported.  NAME is
+   the builtin name, CODE is the builtin code, and FTYPE is the function
+   type of the builtin.  */
+
+static void
+make_cpu_type_builtin (const char* name, int code,
+		       enum ix86_builtin_func_type ftype)
+{
+  tree decl;
+  tree type;
+
+  type = ix86_get_builtin_func_type (ftype);
+  decl = add_builtin_function (name, type, code, BUILT_IN_MD,
+			       NULL, NULL_TREE);
+  gcc_assert (decl != NULL_TREE);
+  ix86_builtins[(int) code] = decl;
+  TREE_READONLY (decl) = 1;
+}
+
+/* Make builtins to get CPU type and features supported.  The created
+   builtins are :
+
+   __builtin_cpu_init (), to detect cpu type and features,
+   __builtin_cpu_is ("<CPUNAME>"), to check if cpu is of type <CPUNAME>,
+   __builtin_cpu_supports ("<FEATURE>"), to check if cpu supports <FEATURE>
+   */
+
+static void
+ix86_init_platform_type_builtins (void)
+{
+  make_cpu_type_builtin ("__builtin_cpu_init", IX86_BUILTIN_CPU_INIT,
+			 INT_FTYPE_VOID);
+  make_cpu_type_builtin ("__builtin_cpu_is", IX86_BUILTIN_CPU_IS,
+			 INT_FTYPE_PCCHAR);
+  make_cpu_type_builtin ("__builtin_cpu_supports", IX86_BUILTIN_CPU_SUPPORTS,
+			 INT_FTYPE_PCCHAR);
+}
+
 /* Internal method for ix86_init_builtins.  */
 
 static void
@@ -27710,6 +28070,9 @@ ix86_init_builtins (void)
 
   ix86_init_builtin_types ();
 
+  /* Builtins to get CPU type and features. */
+  ix86_init_platform_type_builtins ();
+
   /* TFmode support builtins.  */
   def_builtin_const (0, "__builtin_infq",
 		     FLOAT128_FTYPE_VOID, IX86_BUILTIN_INFQ);
@@ -29328,6 +29691,28 @@ ix86_expand_builtin (tree exp, rtx target, rtx sub
   enum machine_mode mode0, mode1, mode2, mode3, mode4;
   unsigned int fcode = DECL_FUNCTION_CODE (fndecl);
 
+  /* For CPU builtins that can be folded, fold first and expand the fold.  */
+  switch (fcode)
+    {
+    case IX86_BUILTIN_CPU_INIT:
+      {
+	/* Make it call __cpu_indicator_init in libgcc. */
+	tree call_expr, fndecl, type;
+        type = build_function_type_list (integer_type_node, NULL_TREE); 
+	fndecl = build_fn_decl ("__cpu_indicator_init", type);
+	call_expr = build_call_expr (fndecl, 0); 
+	return expand_expr (call_expr, target, mode, EXPAND_NORMAL);
+      }
+    case IX86_BUILTIN_CPU_IS:
+    case IX86_BUILTIN_CPU_SUPPORTS:
+      {
+	tree arg0 = CALL_EXPR_ARG (exp, 0);
+	tree fold_expr = fold_builtin_cpu (fndecl, &arg0);
+	gcc_assert (fold_expr != NULL_TREE);
+        return expand_expr (fold_expr, target, mode, EXPAND_NORMAL);
+      }
+    }
+
   /* Determine whether the builtin function is available under the current ISA.
      Originally the builtin was not created if it wasn't applicable to the
      current ISA based on the command line switches.  With function specific
@@ -39058,6 +39443,9 @@ ix86_autovectorize_vector_sizes (void)
 #undef TARGET_BUILD_BUILTIN_VA_LIST
 #define TARGET_BUILD_BUILTIN_VA_LIST ix86_build_builtin_va_list
 
+#undef TARGET_FOLD_BUILTIN
+#define TARGET_FOLD_BUILTIN ix86_fold_builtin
+
 #undef TARGET_ENUM_VA_LIST_P
 #define TARGET_ENUM_VA_LIST_P ix86_enum_va_list
 
Index: gcc/doc/extend.texi
===================================================================
--- gcc/doc/extend.texi	(revision 185898)
+++ gcc/doc/extend.texi	(working copy)
@@ -9413,6 +9413,140 @@ Similar to @code{__builtin_huge_val}, except the r
 @findex __builtin_huge_valq
 @end table
 
+The following built-in functions are always available and can be used to
+check the target platform type.
+
+@deftypefn {Built-in Function} void __builtin_cpu_init (void)
+This function runs the CPU detection code to check the type of CPU and the features
+supported. This builtin needs to be invoked along with the builtins to check CPU type
+and features, @code{__builtin_cpu_is} and @code{__builtin_cpu_supports}, only when used
+used in a function that will be executed before any constructors are called.  The
+CPU detection code is automatically executed in a very high priority constructor.
+
+For example, this function has to be used in @code{ifunc} resolvers which check for
+CPU type using the builtins, @code{__builtin_cpu_is} and @code{__builtin_cpu_supports}.
+@smallexample
+
+static void (*resolve_memcpy (void)) (void)
+@{
+  // ifunc resolvers fire before constructors, explicitly call the init
+  // function.
+  __builtin_cpu_init ();
+  if (__builtin_cpu_supports ("ssse3"))
+    return ssse3_memcpy; // super fast memcpy with ssse3 instructions.
+  else
+    return default_memcpy;
+@}
+
+void *memcpy (void *, const void *, size_t)
+     __attribute__ ((ifunc ("resolve_memcpy")));
+@end smallexample
+
+@end deftypefn
+
+@deftypefn {Built-in Function} int __builtin_cpu_is (const char *@var{cpuname})
+This function returns @code{1}, if the runtime cpu is of type @var{cpuname}
+ and returns @code{0} otherwise. The following cpu names can be detected:
+
+@table @samp
+@item intel
+Intel CPU.
+
+@item atom
+Intel ATOM CPU.
+
+@item core2
+Intel Core2 CPU.
+
+@item corei7
+Intel Corei7 CPU.
+
+@item nehalem
+Intel Corei7 Nehalem CPU.
+
+@item westmere
+Intel Corei7 Westmere CPU.
+
+@item sandybridge
+Intel Corei7 Sandybridge CPU.
+
+@item amd
+AMD CPU.
+
+@item amdfam10h
+AMD family 10h CPU.
+
+@item barcelona
+AMD family 10h Barcelona CPU.
+
+@item shanghai
+AMD family 10h Shanghai CPU.
+
+@item istanbul
+AMD family 15h Istanbul CPU.
+
+@item amdfam15h
+AMD family 15h CPU.
+
+@item bdver1
+AMD family 15h Bulldozer version 1.
+
+@item bdver2
+AMD family 15h Bulldozer version 1.
+@end table 
+
+Here is an example:
+@smallexample
+if (__builtin_cpu_is ("corei7"))
+  @{
+     do_corei7 (); //Corei7 specific implementation.
+  @}
+else
+  @{
+     do_generic (); //Generic implementation.
+  @}  
+@end smallexample
+@end deftypefn
+
+@deftypefn {Built-in Function} int __builtin_cpu_supports (const char *@var{feature})
+This function returns @code{1}, if the runtime cpu supports @var{feature}
+ and returns @code{0} otherwise. The following features can be detected:
+
+@table @samp
+@item cmov
+CMOV instruction.
+@item mmx
+MMX instructions.
+@item popcnt
+POPCNT instruction.
+@item sse
+SSE instructions.
+@item sse2
+SSE2 instructions.
+@item sse3
+SSE3 instructions.
+@item ssse3
+SSSE3 instructions.
+@item sse4.1
+SSE4.1 instructions.
+@item sse4.2
+SSE4.2 instructions.
+@end table
+
+Here is an example:
+@smallexample
+if (__builtin_cpu_supports ("popcnt"))
+  @{
+     asm("popcnt %1,%0" : "=r"(count) : "rm"(n) : "cc");
+  @}
+else
+  @{
+     count = generic_countbits (n); //generic implementation.
+  @}  
+@end smallexample
+@end deftypefn
+
+
 The following built-in functions are made available by @option{-mmmx}.
 All of them generate the machine instruction that is part of the name.
 

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

On Mon, Apr 23, 2012 at 1:19 AM, Uros Bizjak <ubizjak@gmail.com> wrote:
> On Tue, Apr 3, 2012 at 9:47 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>
>> i386 maintainers -  Is this patch ok?
>
> Has the community reached the consensus on how this kind of
> functionality has to be implemented? I have followed the discussion a
> bit, but IIRC, there was no clear decision. Without this decision, I
> am not able to review the _implementation_ of agreed functionality for
> x86 target.
>
> (I apologize if I have missed the decision, please point me to the
> discussion in this case.)

Also, Richard's ok on just using 2 builtins: __builtin_cpu_is and
__builtin_cpu_supports

http://gcc.gnu.org/ml/gcc-patches/2012-03/msg00752.html

>
>>>>> Patch available for review here:
>>>>> http://codereview.appspot.com/5754058
>
> Please attach patches or inline it in the message itself for a review.
> Please see [1] for further instructions.
>
> [1] http://gcc.gnu.org/contribute.html#patches
>
> Uros.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Uros Bizjak]

On Mon, Apr 23, 2012 at 6:59 PM, Sriraman Tallam <tmsriram@google.com> wrote:

>>> i386 maintainers -  Is this patch ok?
>>
>> Has the community reached the consensus on how this kind of
>> functionality has to be implemented? I have followed the discussion a
>> bit, but IIRC, there was no clear decision. Without this decision, I
>> am not able to review the _implementation_ of agreed functionality for
>> x86 target.
>>
>> (I apologize if I have missed the decision, please point me to the
>> discussion in this case.)
>
> The discussions are here:
>
> http://gcc.gnu.org/ml/gcc-patches/2011-08/msg01446.html
> and follow-ups to this.
>
> I am not sure about consensus, but the important points raised were:
>
> 1) Constructor ordering: What if some constructors fire before
> cpu_indicator_init?, which determines the CPU. I addressed this
> problem by making the priority of cpu_indicator_init to be the highest
> possible. Still, IFUNC initializers will fire before and they have to
> explicitly call __builtin_cpu_init() before checking the CPU type.
> 2) Reducing the number of builtins : It is only two now.

>       * config/i386/i386.c (build_processor_features_struct): New function.
>       (build_processor_model_struct): New function.
>       (make_var_decl): New function.
>       (get_field_from_struct): New function.
>       (fold_builtin_target): New function.
>       (ix86_fold_builtin): New function.
>       (ix86_expand_builtin): Expand new builtins by folding them.
>       (make_cpu_type_builtin): New functions.
>       (ix86_init_platform_type_builtins): Make the new builtins.
>       (ix86_init_builtins): Make new builtins to detect CPU type.
>       (TARGET_FOLD_BUILTIN): New macro.
>       (IX86_BUILTIN_CPU_INIT): New enum value.
>       (IX86_BUILTIN_CPU_IS): New enum value.
>       (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>       * config/i386/i386-builtin-types.def: New function type.
>       * testsuite/gcc.target/builtin_target.c: New testcase.
>       * doc/extend.texi: Document builtins.
>
>       * libgcc/config/i386/i386-cpuinfo.c: New file.
>       * libgcc/config/i386/t-cpuinfo: New file.
>       * libgcc/config.host: Include t-cpuinfo.
>       * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
>       and __cpu_features.

The patch is OK.

I guess that AVX is left as an exercise for a x86 maintainer ;)

(I have also CC'd H.J. for his opinion.)

Thanks,
Uros.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Uros Bizjak]

On Mon, Apr 23, 2012 at 6:59 PM, Sriraman Tallam <tmsriram@google.com> wrote:

>       * config/i386/i386.c (build_processor_features_struct): New function.
>       (build_processor_model_struct): New function.
>       (make_var_decl): New function.
>       (get_field_from_struct): New function.
>       (fold_builtin_target): New function.
>       (ix86_fold_builtin): New function.
>       (ix86_expand_builtin): Expand new builtins by folding them.
>       (make_cpu_type_builtin): New functions.
>       (ix86_init_platform_type_builtins): Make the new builtins.
>       (ix86_init_builtins): Make new builtins to detect CPU type.
>       (TARGET_FOLD_BUILTIN): New macro.
>       (IX86_BUILTIN_CPU_INIT): New enum value.
>       (IX86_BUILTIN_CPU_IS): New enum value.
>       (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>       * config/i386/i386-builtin-types.def: New function type.
>       * testsuite/gcc.target/builtin_target.c: New testcase.
>       * doc/extend.texi: Document builtins.
>
>       * libgcc/config/i386/i386-cpuinfo.c: New file.
>       * libgcc/config/i386/t-cpuinfo: New file.
>       * libgcc/config.host: Include t-cpuinfo.
>       * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
>       and __cpu_features.

A couple of typos in the documentation:

+@deftypefn {Built-in Function} void __builtin_cpu_init (void)
+This function runs the CPU detection code to check the type of CPU
and the features
+supported. This builtin needs to be invoked along with the builtins
to check CPU type
+and features, @code{__builtin_cpu_is} and
@code{__builtin_cpu_supports}, only when used
+used in a function that will be executed before any constructors are
called.  The

used used

+@item bdver1
+AMD family 15h Bulldozer version 1.
+
+@item bdver2
+AMD family 15h Bulldozer version 1.

Ehm ....

+@deftypefn {Built-in Function} int __builtin_cpu_supports (const char
*@var{feature})
+This function returns @code{1}, if the runtime cpu supports @var{feature}
+ and returns @code{0} otherwise. The following features can be detected:

In the testcases, return value of the builtin is checked for < 0. If
this signals an error, then please mention negative return and its
meaning.

+@deftypefn {Built-in Function} int __builtin_cpu_is (const char *@var{cpuname})
+This function returns @code{1}, if the runtime cpu is of type @var{cpuname}
+ and returns @code{0} otherwise. The following cpu names can be detected:

Also here.

I also think that this is very useful feature and deserves an entry in
gcc-4.8 changes at http://gcc.gnu.org/gcc-4.8/changes.html .

Uros.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[H.J. Lu]

On Mon, Apr 23, 2012 at 9:59 AM, Sriraman Tallam <tmsriram@google.com> wrote:
> Hi,
>
> On Mon, Apr 23, 2012 at 1:19 AM, Uros Bizjak <ubizjak@gmail.com> wrote:
>> On Tue, Apr 3, 2012 at 9:47 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>>
>>> i386 maintainers -  Is this patch ok?
>>
>> Has the community reached the consensus on how this kind of
>> functionality has to be implemented? I have followed the discussion a
>> bit, but IIRC, there was no clear decision. Without this decision, I
>> am not able to review the _implementation_ of agreed functionality for
>> x86 target.
>>
>> (I apologize if I have missed the decision, please point me to the
>> discussion in this case.)
>
> The discussions are here:
>
> http://gcc.gnu.org/ml/gcc-patches/2011-08/msg01446.html
> and follow-ups to this.
>
> I am not sure about consensus, but the important points raised were:
>
> 1) Constructor ordering: What if some constructors fire before
> cpu_indicator_init?, which determines the CPU. I addressed this
> problem by making the priority of cpu_indicator_init to be the highest
> possible. Still, IFUNC initializers will fire before and they have to
> explicitly call __builtin_cpu_init() before checking the CPU type.
> 2) Reducing the number of builtins : It is only two now.
>
>
>>
>>>>>> Patch available for review here:
>>>>>> http://codereview.appspot.com/5754058
>>
>> Please attach patches or inline it in the message itself for a review.
>> Please see [1] for further instructions.
>
> Patch attached, tested on x86_64 and all tests pass.
>
>
>       * config/i386/i386.c (build_processor_features_struct): New function.
>       (build_processor_model_struct): New function.
>       (make_var_decl): New function.
>       (get_field_from_struct): New function.
>       (fold_builtin_target): New function.
>       (ix86_fold_builtin): New function.
>       (ix86_expand_builtin): Expand new builtins by folding them.
>       (make_cpu_type_builtin): New functions.
>       (ix86_init_platform_type_builtins): Make the new builtins.
>       (ix86_init_builtins): Make new builtins to detect CPU type.
>       (TARGET_FOLD_BUILTIN): New macro.
>       (IX86_BUILTIN_CPU_INIT): New enum value.
>       (IX86_BUILTIN_CPU_IS): New enum value.
>       (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>       * config/i386/i386-builtin-types.def: New function type.
>       * testsuite/gcc.target/builtin_target.c: New testcase.
>       * doc/extend.texi: Document builtins.
>
>       * libgcc/config/i386/i386-cpuinfo.c: New file.
>       * libgcc/config/i386/t-cpuinfo: New file.
>       * libgcc/config.host: Include t-cpuinfo.
>       * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
>       and __cpu_features.
>
> Thanks,
> -Sri.
>
>

I have 2 comments:

1.  You should remove

static int called = 0;
if (called)
    return 0;
 else
    called = 1;

Instead, you can just do

if (_cpu_model.__cpu_vendor)
  return 0;

2.  You can replace

if (vendor == SIG_AMD)

with

else if (vendor == SIG_AMD)

Thanks.

-- 
H.J.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

On Mon, Apr 23, 2012 at 12:16 PM, Uros Bizjak <ubizjak@gmail.com> wrote:
> On Mon, Apr 23, 2012 at 6:59 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>
>>>> i386 maintainers -  Is this patch ok?
>>>
>>> Has the community reached the consensus on how this kind of
>>> functionality has to be implemented? I have followed the discussion a
>>> bit, but IIRC, there was no clear decision. Without this decision, I
>>> am not able to review the _implementation_ of agreed functionality for
>>> x86 target.
>>>
>>> (I apologize if I have missed the decision, please point me to the
>>> discussion in this case.)
>>
>> The discussions are here:
>>
>> http://gcc.gnu.org/ml/gcc-patches/2011-08/msg01446.html
>> and follow-ups to this.
>>
>> I am not sure about consensus, but the important points raised were:
>>
>> 1) Constructor ordering: What if some constructors fire before
>> cpu_indicator_init?, which determines the CPU. I addressed this
>> problem by making the priority of cpu_indicator_init to be the highest
>> possible. Still, IFUNC initializers will fire before and they have to
>> explicitly call __builtin_cpu_init() before checking the CPU type.
>> 2) Reducing the number of builtins : It is only two now.
>
>>       * config/i386/i386.c (build_processor_features_struct): New function.
>>       (build_processor_model_struct): New function.
>>       (make_var_decl): New function.
>>       (get_field_from_struct): New function.
>>       (fold_builtin_target): New function.
>>       (ix86_fold_builtin): New function.
>>       (ix86_expand_builtin): Expand new builtins by folding them.
>>       (make_cpu_type_builtin): New functions.
>>       (ix86_init_platform_type_builtins): Make the new builtins.
>>       (ix86_init_builtins): Make new builtins to detect CPU type.
>>       (TARGET_FOLD_BUILTIN): New macro.
>>       (IX86_BUILTIN_CPU_INIT): New enum value.
>>       (IX86_BUILTIN_CPU_IS): New enum value.
>>       (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>>       * config/i386/i386-builtin-types.def: New function type.
>>       * testsuite/gcc.target/builtin_target.c: New testcase.
>>       * doc/extend.texi: Document builtins.
>>
>>       * libgcc/config/i386/i386-cpuinfo.c: New file.
>>       * libgcc/config/i386/t-cpuinfo: New file.
>>       * libgcc/config.host: Include t-cpuinfo.
>>       * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
>>       and __cpu_features.
>
> The patch is OK.
>
> I guess that AVX is left as an exercise for a x86 maintainer ;)

I will add AVX, and make all the changes.

Thanks,
-Sri.

>
> (I have also CC'd H.J. for his opinion.)
>
> Thanks,
> Uros.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[H.J. Lu]

On Mon, Apr 23, 2012 at 1:43 PM, Sriraman Tallam <tmsriram@google.com> wrote:
> On Mon, Apr 23, 2012 at 12:16 PM, Uros Bizjak <ubizjak@gmail.com> wrote:
>> On Mon, Apr 23, 2012 at 6:59 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>>
>>>>> i386 maintainers -  Is this patch ok?
>>>>
>>>> Has the community reached the consensus on how this kind of
>>>> functionality has to be implemented? I have followed the discussion a
>>>> bit, but IIRC, there was no clear decision. Without this decision, I
>>>> am not able to review the _implementation_ of agreed functionality for
>>>> x86 target.
>>>>
>>>> (I apologize if I have missed the decision, please point me to the
>>>> discussion in this case.)
>>>
>>> The discussions are here:
>>>
>>> http://gcc.gnu.org/ml/gcc-patches/2011-08/msg01446.html
>>> and follow-ups to this.
>>>
>>> I am not sure about consensus, but the important points raised were:
>>>
>>> 1) Constructor ordering: What if some constructors fire before
>>> cpu_indicator_init?, which determines the CPU. I addressed this
>>> problem by making the priority of cpu_indicator_init to be the highest
>>> possible. Still, IFUNC initializers will fire before and they have to
>>> explicitly call __builtin_cpu_init() before checking the CPU type.
>>> 2) Reducing the number of builtins : It is only two now.
>>
>>>       * config/i386/i386.c (build_processor_features_struct): New function.
>>>       (build_processor_model_struct): New function.
>>>       (make_var_decl): New function.
>>>       (get_field_from_struct): New function.
>>>       (fold_builtin_target): New function.
>>>       (ix86_fold_builtin): New function.
>>>       (ix86_expand_builtin): Expand new builtins by folding them.
>>>       (make_cpu_type_builtin): New functions.
>>>       (ix86_init_platform_type_builtins): Make the new builtins.
>>>       (ix86_init_builtins): Make new builtins to detect CPU type.
>>>       (TARGET_FOLD_BUILTIN): New macro.
>>>       (IX86_BUILTIN_CPU_INIT): New enum value.
>>>       (IX86_BUILTIN_CPU_IS): New enum value.
>>>       (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>>>       * config/i386/i386-builtin-types.def: New function type.
>>>       * testsuite/gcc.target/builtin_target.c: New testcase.
>>>       * doc/extend.texi: Document builtins.
>>>
>>>       * libgcc/config/i386/i386-cpuinfo.c: New file.
>>>       * libgcc/config/i386/t-cpuinfo: New file.
>>>       * libgcc/config.host: Include t-cpuinfo.
>>>       * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
>>>       and __cpu_features.
>>
>> The patch is OK.
>>
>> I guess that AVX is left as an exercise for a x86 maintainer ;)
>
> I will add AVX, and make all the changes.
>

3 more comments:

1.  Your implementation isn't thread-safe. With my suggestion of

if (_cpu_model.__cpu_vendor)
 return 0;

you should change get_intel_cpu to return processor_vendor and
set _cpu_model.__cpu_vendor after setting up all other bits.

2. Bitfields

struct __processor_features
+{
+  unsigned int __cpu_cmov : 1;
+  unsigned int __cpu_mmx : 1;
+  unsigned int __cpu_popcnt : 1;
+  unsigned int __cpu_sse : 1;
+  unsigned int __cpu_sse2 : 1;
+  unsigned int __cpu_sse3 : 1;
+  unsigned int __cpu_ssse3 : 1;
+  unsigned int __cpu_sse4_1 : 1;
+  unsigned int __cpu_sse4_2 : 1;
+} __cpu_features;

may not be thread-safe since only char/short/int/long long
stores are atomic and it isn't extensible. You can use

unsigned int __cpu_features[1];

to make it extensible and store atomic.

3. You may move

unsigned int __cpu_features[1];

into struct __processor_model to only export one symbol
instead of 2.

-- 
H.J.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

[-- Attachment #1: Type: text/plain, Size: 5343 bytes --]

Hi,

   Thanks for all the comments. I have made all the changes as
mentioned and submiited the patch. Summary of changes made:

* Add support for AVX
* Fix documentation in extend.texi
* Make it thread-safe according to H.J.'s comments.

I have attached the patch. Boot-strapped and checked for test parity
with pristine build.

       * config/i386/i386.c (build_processor_model_struct): New function.
        (make_var_decl): New function.
        (fold_builtin_cpu): New function.
        (ix86_fold_builtin): New function.
        (make_cpu_type_builtin): New function.
        (ix86_init_platform_type_builtins): New function.
        (ix86_expand_builtin): Expand new builtins by folding them.
        (ix86_init_builtins): Make new builtins to detect CPU type.
        (TARGET_FOLD_BUILTIN): New macro.
        (IX86_BUILTIN_CPU_INIT): New enum value.
        (IX86_BUILTIN_CPU_IS): New enum value.
        (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
        * config/i386/i386-builtin-types.def: New function type.
        * testsuite/gcc.target/builtin_target.c: New testcase.
        * doc/extend.texi: Document builtins.

        * libgcc/config/i386/i386-cpuinfo.c: New file.
        * libgcc/config/i386/t-cpuinfo: New file.
        * libgcc/config.host: Include t-cpuinfo.
        * libgcc/config/i386/libgcc-glibc.ver: Version symbol __cpu_model.


Thanks,
-Sri.

On Mon, Apr 23, 2012 at 5:46 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
> On Mon, Apr 23, 2012 at 1:43 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>> On Mon, Apr 23, 2012 at 12:16 PM, Uros Bizjak <ubizjak@gmail.com> wrote:
>>> On Mon, Apr 23, 2012 at 6:59 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>>>
>>>>>> i386 maintainers -  Is this patch ok?
>>>>>
>>>>> Has the community reached the consensus on how this kind of
>>>>> functionality has to be implemented? I have followed the discussion a
>>>>> bit, but IIRC, there was no clear decision. Without this decision, I
>>>>> am not able to review the _implementation_ of agreed functionality for
>>>>> x86 target.
>>>>>
>>>>> (I apologize if I have missed the decision, please point me to the
>>>>> discussion in this case.)
>>>>
>>>> The discussions are here:
>>>>
>>>> http://gcc.gnu.org/ml/gcc-patches/2011-08/msg01446.html
>>>> and follow-ups to this.
>>>>
>>>> I am not sure about consensus, but the important points raised were:
>>>>
>>>> 1) Constructor ordering: What if some constructors fire before
>>>> cpu_indicator_init?, which determines the CPU. I addressed this
>>>> problem by making the priority of cpu_indicator_init to be the highest
>>>> possible. Still, IFUNC initializers will fire before and they have to
>>>> explicitly call __builtin_cpu_init() before checking the CPU type.
>>>> 2) Reducing the number of builtins : It is only two now.
>>>
>>>>       * config/i386/i386.c (build_processor_features_struct): New function.
>>>>       (build_processor_model_struct): New function.
>>>>       (make_var_decl): New function.
>>>>       (get_field_from_struct): New function.
>>>>       (fold_builtin_target): New function.
>>>>       (ix86_fold_builtin): New function.
>>>>       (ix86_expand_builtin): Expand new builtins by folding them.
>>>>       (make_cpu_type_builtin): New functions.
>>>>       (ix86_init_platform_type_builtins): Make the new builtins.
>>>>       (ix86_init_builtins): Make new builtins to detect CPU type.
>>>>       (TARGET_FOLD_BUILTIN): New macro.
>>>>       (IX86_BUILTIN_CPU_INIT): New enum value.
>>>>       (IX86_BUILTIN_CPU_IS): New enum value.
>>>>       (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>>>>       * config/i386/i386-builtin-types.def: New function type.
>>>>       * testsuite/gcc.target/builtin_target.c: New testcase.
>>>>       * doc/extend.texi: Document builtins.
>>>>
>>>>       * libgcc/config/i386/i386-cpuinfo.c: New file.
>>>>       * libgcc/config/i386/t-cpuinfo: New file.
>>>>       * libgcc/config.host: Include t-cpuinfo.
>>>>       * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
>>>>       and __cpu_features.
>>>
>>> The patch is OK.
>>>
>>> I guess that AVX is left as an exercise for a x86 maintainer ;)
>>
>> I will add AVX, and make all the changes.
>>
>
> 3 more comments:
>
> 1.  Your implementation isn't thread-safe. With my suggestion of
>
> if (_cpu_model.__cpu_vendor)
>  return 0;
>
> you should change get_intel_cpu to return processor_vendor and
> set _cpu_model.__cpu_vendor after setting up all other bits.
>
> 2. Bitfields
>
> struct __processor_features
> +{
> +  unsigned int __cpu_cmov : 1;
> +  unsigned int __cpu_mmx : 1;
> +  unsigned int __cpu_popcnt : 1;
> +  unsigned int __cpu_sse : 1;
> +  unsigned int __cpu_sse2 : 1;
> +  unsigned int __cpu_sse3 : 1;
> +  unsigned int __cpu_ssse3 : 1;
> +  unsigned int __cpu_sse4_1 : 1;
> +  unsigned int __cpu_sse4_2 : 1;
> +} __cpu_features;
>
> may not be thread-safe since only char/short/int/long long
> stores are atomic and it isn't extensible. You can use
>
> unsigned int __cpu_features[1];
>
> to make it extensible and store atomic.
>
> 3. You may move
>
> unsigned int __cpu_features[1];
>
> into struct __processor_model to only export one symbol
> instead of 2.
>
> --
> H.J.

[-- Attachment #2: cpu_runtime_patch.txt --]
[-- Type: text/plain, Size: 30382 bytes --]

2012-04-24  Sriraman Tallam  <tmsriram@google.com>

This patch adds new builtins to check for cpu type and features. 

* __builtin_cpu_is ("<CPUNAME>")
* __builtin_cpu_supports ("<FEATURE>")

apart from the cpu init builtin, __builtin_cpu_init.

List of CPU names :

* "amd"
* "intel"
* "atom"
* "core2"
* "corei7"
* "nehalem"
* "westmere"
* "sandybridge"
* "amdfam10h"
* "barcelona"
* "shanghai"
* "istanbul"
* "bdver1"
* "bdver2"

List of CPU features :

* "cmov"
* "mmx"
* "popcnt"
* "sse"
* "sse2"
* "sse3"
* "ssse3"
* "sse4.1"
* "sse4.2"
* "avx"

        * config/i386/i386.c (build_processor_model_struct): New function.
        (make_var_decl): New function.
        (fold_builtin_cpu): New function.
        (ix86_fold_builtin): New function.
        (make_cpu_type_builtin): New function.
        (ix86_init_platform_type_builtins): New function.
        (ix86_expand_builtin): Expand new builtins by folding them.
        (ix86_init_builtins): Make new builtins to detect CPU type.
        (TARGET_FOLD_BUILTIN): New macro.
        (IX86_BUILTIN_CPU_INIT): New enum value.
        (IX86_BUILTIN_CPU_IS): New enum value.
        (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
        * config/i386/i386-builtin-types.def: New function type.
        * testsuite/gcc.target/builtin_target.c: New testcase.
        * doc/extend.texi: Document builtins.

        * libgcc/config/i386/i386-cpuinfo.c: New file.
        * libgcc/config/i386/t-cpuinfo: New file.
        * libgcc/config.host: Include t-cpuinfo.
        * libgcc/config/i386/libgcc-glibc.ver: Version symbol __cpu_model.


Index: libgcc/config.host
===================================================================
--- libgcc/config.host        (revision 186788)
+++ libgcc/config.host        (working copy)
@@ -1130,7 +1130,7 @@ i[34567]86-*-linux* | x86_64-*-linux* | \
   i[34567]86-*-kfreebsd*-gnu | x86_64-*-kfreebsd*-gnu | \
   i[34567]86-*-knetbsd*-gnu | \
   i[34567]86-*-gnu*)
-        tmake_file="${tmake_file} t-tls i386/t-linux"
+        tmake_file="${tmake_file} t-tls i386/t-linux i386/t-cpuinfo"
         if test "$libgcc_cv_cfi" = "yes"; then
                 tmake_file="${tmake_file} t-stack i386/t-stack-i386"
         fi
Index: libgcc/config/i386/t-cpuinfo
===================================================================
--- libgcc/config/i386/t-cpuinfo        (revision 0)
+++ libgcc/config/i386/t-cpuinfo        (revision 0)
@@ -0,0 +1 @@
+LIB2ADD += $(srcdir)/config/i386/i386-cpuinfo.c
Index: libgcc/config/i386/i386-cpuinfo.c
===================================================================
--- libgcc/config/i386/i386-cpuinfo.c        (revision 0)
+++ libgcc/config/i386/i386-cpuinfo.c        (revision 0)
@@ -0,0 +1,316 @@
+/* Get CPU type and Features for x86 processors.
+   Copyright (C) 2012 Free Software Foundation, Inc.
+   Contributed by Sriraman Tallam (tmsriram@google.com)
+
+This file is part of GCC.
+
+GCC is free software; you can redistribute it and/or modify it under
+the terms of the GNU General Public License as published by the Free
+Software Foundation; either version 3, or (at your option) any later
+version.
+
+GCC is distributed in the hope that it will be useful, but WITHOUT ANY
+WARRANTY; without even the implied warranty of MERCHANTABILITY or
+FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+for more details.
+
+You should have received a copy of the GNU General Public License
+along with GCC; see the file COPYING3.  If not see
+<http://www.gnu.org/licenses/>. */
+
+#include "cpuid.h"
+#include "tsystem.h"
+
+int __cpu_indicator_init (void) __attribute__ ((constructor (101)));
+
+enum vendor_signatures
+{
+  SIG_INTEL =        0x756e6547 /* Genu */,
+  SIG_AMD =        0x68747541 /* Auth */
+};
+
+/* Processor Vendor and Models. */
+
+enum processor_vendor
+{
+  VENDOR_INTEL = 1,
+  VENDOR_AMD,
+  VENDOR_OTHER,
+  VENDOR_MAX
+};
+
+enum processor_types
+{
+  INTEL_ATOM = 1,
+  INTEL_CORE2,
+  INTEL_COREI7,
+  AMDFAM10H,
+  AMDFAM15H,
+  CPU_TYPE_MAX
+};
+
+enum processor_subtypes
+{
+  INTEL_COREI7_NEHALEM = 1,
+  INTEL_COREI7_WESTMERE,
+  INTEL_COREI7_SANDYBRIDGE,
+  AMDFAM10H_BARCELONA,
+  AMDFAM10H_SHANGHAI,
+  AMDFAM10H_ISTANBUL,
+  AMDFAM15H_BDVER1,
+  AMDFAM15H_BDVER2,
+  CPU_SUBTYPE_MAX
+};
+
+/* ISA Features supported. */
+
+enum processor_features
+{
+  FEATURE_CMOV = 0,
+  FEATURE_MMX,
+  FEATURE_POPCNT,
+  FEATURE_SSE,
+  FEATURE_SSE2,
+  FEATURE_SSE3,
+  FEATURE_SSSE3,
+  FEATURE_SSE4_1,
+  FEATURE_SSE4_2,
+  FEATURE_AVX
+};
+
+struct __processor_model
+{
+  unsigned int __cpu_vendor;
+  unsigned int __cpu_type;
+  unsigned int __cpu_subtype;
+  unsigned int __cpu_features[1];
+} __cpu_model;
+
+
+/* Get the specific type of AMD CPU.  */
+
+static void
+get_amd_cpu (unsigned int family, unsigned int model)
+{
+  switch (family)
+    {
+    /* AMD Family 10h.  */
+    case 0x10:
+      switch (model)
+        {
+        case 0x2:
+          /* Barcelona.  */
+          __cpu_model.__cpu_type = AMDFAM10H;
+          __cpu_model.__cpu_subtype = AMDFAM10H_BARCELONA;
+          break;
+        case 0x4:
+          /* Shanghai.  */
+          __cpu_model.__cpu_type = AMDFAM10H;
+          __cpu_model.__cpu_subtype = AMDFAM10H_SHANGHAI;
+          break;
+        case 0x8:
+          /* Istanbul.  */
+          __cpu_model.__cpu_type = AMDFAM10H;
+          __cpu_model.__cpu_subtype = AMDFAM10H_ISTANBUL;
+          break;
+        default:
+          break;
+        }
+      break;
+    /* AMD Family 15h.  */
+    case 0x15:
+      __cpu_model.__cpu_type = AMDFAM15H;
+      /* Bulldozer version 1.  */
+      if ( model <= 0xf)
+        __cpu_model.__cpu_subtype = AMDFAM15H_BDVER1;
+      /* Bulldozer version 2.  */
+      if (model >= 0x10 && model <= 0x1f)
+        __cpu_model.__cpu_subtype = AMDFAM15H_BDVER2;
+      break;
+    default:
+      break;
+    }
+}
+
+/* Get the specific type of Intel CPU.  */
+
+static void
+get_intel_cpu (unsigned int family, unsigned int model, unsigned int brand_id)
+{
+  /* Parse family and model only if brand ID is 0. */
+  if (brand_id == 0)
+    {
+      switch (family)
+        {
+        case 0x5:
+          /* Pentium.  */
+          break;
+        case 0x6:
+          switch (model)
+            {
+            case 0x1c:
+            case 0x26:
+              /* Atom.  */
+              __cpu_model.__cpu_type = INTEL_ATOM;
+              break;
+            case 0x1a:
+            case 0x1e:
+            case 0x1f:
+            case 0x2e:
+              /* Nehalem.  */
+              __cpu_model.__cpu_type = INTEL_COREI7;
+              __cpu_model.__cpu_subtype = INTEL_COREI7_NEHALEM;
+              break;
+            case 0x25:
+            case 0x2c:
+            case 0x2f:
+              /* Westmere.  */
+              __cpu_model.__cpu_type = INTEL_COREI7;
+              __cpu_model.__cpu_subtype = INTEL_COREI7_WESTMERE;
+              break;
+            case 0x2a:
+              /* Sandy Bridge.  */
+              __cpu_model.__cpu_type = INTEL_COREI7;
+              __cpu_model.__cpu_subtype = INTEL_COREI7_SANDYBRIDGE;
+              break;
+            case 0x17:
+            case 0x1d:
+              /* Penryn.  */
+            case 0x0f:
+              /* Merom.  */
+              __cpu_model.__cpu_type = INTEL_CORE2;
+              break;
+            default:
+              break;
+            }
+          break;
+        default:
+          /* We have no idea.  */
+          break;
+        }
+    }
+}                             
+
+static void
+get_available_features (unsigned int ecx, unsigned int edx)
+{
+  unsigned int features = 0;
+
+  if (edx & bit_CMOV)
+    features |= (1 << FEATURE_CMOV);
+  if (edx & bit_MMX)
+    features |= (1 << FEATURE_MMX);
+  if (edx & bit_SSE)
+    features |= (1 << FEATURE_SSE);
+  if (edx & bit_SSE2)
+    features |= (1 << FEATURE_SSE2);
+  if (ecx & bit_POPCNT)
+    features |= (1 << FEATURE_POPCNT);
+  if (ecx & bit_SSE3)
+    features |= (1 << FEATURE_SSE3);
+  if (ecx & bit_SSSE3)
+    features |= (1 << FEATURE_SSSE3);
+  if (ecx & bit_SSE4_1)
+    features |= (1 << FEATURE_SSE4_1);
+  if (ecx & bit_SSE4_2)
+    features |= (1 << FEATURE_SSE4_2);
+  if (ecx & bit_AVX)
+    features |= (1 << FEATURE_AVX);
+
+  __cpu_model.__cpu_features[0] = features;
+}
+
+/* A noinline function calling __get_cpuid. Having many calls to
+   cpuid in one function in 32-bit mode causes GCC to complain:
+   "can't find a register in class CLOBBERED_REGS".  This is
+   related to PR rtl-optimization 44174. */
+
+static int __attribute__ ((noinline))
+__get_cpuid_output (unsigned int __level,
+                    unsigned int *__eax, unsigned int *__ebx,
+                    unsigned int *__ecx, unsigned int *__edx)
+{
+  return __get_cpuid (__level, __eax, __ebx, __ecx, __edx);
+}
+
+
+/* A constructor function that is sets __cpu_model and __cpu_features with
+   the right values.  This needs to run only once.  This constructor is
+   given the highest priority and it should run before constructors without
+   the priority set.  However, it still runs after ifunc initializers and
+   needs to be called explicitly there.  */
+
+int __attribute__ ((constructor (101)))
+__cpu_indicator_init (void)
+{
+  unsigned int eax, ebx, ecx, edx;
+
+  int max_level = 5;
+  unsigned int vendor;
+  unsigned int model, family, brand_id;
+  unsigned int extended_model, extended_family;
+
+  /* This function needs to run just once.  */
+  if (__cpu_model.__cpu_vendor)
+    return 0;
+
+  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
+  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
+    return -1;
+
+  vendor = ebx;
+  max_level = eax;
+
+  if (max_level < 1)
+    return -1;
+
+  if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
+    return -1;
+
+  model = (eax >> 4) & 0x0f;
+  family = (eax >> 8) & 0x0f;
+  brand_id = ebx & 0xff;
+  extended_model = (eax >> 12) & 0xf0;
+  extended_family = (eax >> 20) & 0xff;
+
+  if (vendor == SIG_INTEL)
+    {
+      /* Adjust model and family for Intel CPUS. */
+      if (family == 0x0f)
+        {
+          family += extended_family;
+          model += extended_model;
+        }
+      else if (family == 0x06)
+        model += extended_model;
+
+      /* Get CPU type.  */
+      get_intel_cpu (family, model, brand_id);
+      /* Find available features. */
+      get_available_features (ecx, edx);
+      __cpu_model.__cpu_vendor = VENDOR_INTEL;
+    }
+  else if (vendor == SIG_AMD)
+    {
+      /* Adjust model and family for AMD CPUS. */
+      if (family == 0x0f)
+        {
+          family += extended_family;
+          model += (extended_model << 4);
+        }
+
+      /* Get CPU type.  */
+      get_amd_cpu (family, model);
+      /* Find available features. */
+      get_available_features (ecx, edx);
+      __cpu_model.__cpu_vendor = VENDOR_AMD;
+    }
+  else
+    __cpu_model.__cpu_vendor = VENDOR_OTHER;
+
+  gcc_assert (__cpu_model.__cpu_vendor < VENDOR_MAX);
+  gcc_assert (__cpu_model.__cpu_type < CPU_TYPE_MAX);
+  gcc_assert (__cpu_model.__cpu_subtype < CPU_SUBTYPE_MAX);
+
+  return 0;
+}
Index: libgcc/config/i386/libgcc-glibc.ver
===================================================================
--- libgcc/config/i386/libgcc-glibc.ver        (revision 186788)
+++ libgcc/config/i386/libgcc-glibc.ver        (working copy)
@@ -147,6 +147,10 @@ GCC_4.3.0 {
   __trunctfxf2
   __unordtf2
 }
+
+GCC_4.8.0 {
+  __cpu_model
+}
 %else
 GCC_4.4.0 {
   __addtf3
@@ -183,4 +187,8 @@ GCC_4.4.0 {
 GCC_4.5.0 {
   __extendxftf2
 }
+
+GCC_4.8.0 {
+  __cpu_model
+}
 %endif
Index: gcc/doc/extend.texi
===================================================================
--- gcc/doc/extend.texi        (revision 186788)
+++ gcc/doc/extend.texi        (working copy)
@@ -9419,6 +9419,144 @@ Similar to @code{__builtin_huge_val}, except the r
 @findex __builtin_huge_valq
 @end table
 
+The following built-in functions are always available and can be used to
+check the target platform type.
+
+@deftypefn {Built-in Function} void __builtin_cpu_init (void)
+This function runs the CPU detection code to check the type of CPU and the
+features supported.  This builtin needs to be invoked along with the builtins
+to check CPU type and features, @code{__builtin_cpu_is} and
+@code{__builtin_cpu_supports}, only when used in a function that will be
+executed before any constructors are called.  The CPU detection code is
+automatically executed in a very high priority constructor.
+
+For example, this function has to be used in @code{ifunc} resolvers which
+check for CPU type using the builtins, @code{__builtin_cpu_is}
+and @code{__builtin_cpu_supports}.
+@smallexample
+
+static void (*resolve_memcpy (void)) (void)
+@{
+  // ifunc resolvers fire before constructors, explicitly call the init
+  // function.
+  __builtin_cpu_init ();
+  if (__builtin_cpu_supports ("ssse3"))
+    return ssse3_memcpy; // super fast memcpy with ssse3 instructions.
+  else
+    return default_memcpy;
+@}
+
+void *memcpy (void *, const void *, size_t)
+     __attribute__ ((ifunc ("resolve_memcpy")));
+@end smallexample
+
+@end deftypefn
+
+@deftypefn {Built-in Function} int __builtin_cpu_is (const char *@var{cpuname})
+This function returns a positive integer if the runtime cpu is of type @var{cpuname}
+ and returns @code{0} otherwise. The following cpu names can be detected:
+
+@table @samp
+@item intel
+Intel CPU.
+
+@item atom
+Intel ATOM CPU.
+
+@item core2
+Intel Core2 CPU.
+
+@item corei7
+Intel Corei7 CPU.
+
+@item nehalem
+Intel Corei7 Nehalem CPU.
+
+@item westmere
+Intel Corei7 Westmere CPU.
+
+@item sandybridge
+Intel Corei7 Sandybridge CPU.
+
+@item amd
+AMD CPU.
+
+@item amdfam10h
+AMD family 10h CPU.
+
+@item barcelona
+AMD family 10h Barcelona CPU.
+
+@item shanghai
+AMD family 10h Shanghai CPU.
+
+@item istanbul
+AMD family 10h Istanbul CPU.
+
+@item amdfam15h
+AMD family 15h CPU.
+
+@item bdver1
+AMD family 15h Bulldozer version 1.
+
+@item bdver2
+AMD family 15h Bulldozer version 2.
+@end table
+
+Here is an example:
+@smallexample
+if (__builtin_cpu_is ("corei7"))
+  @{
+     do_corei7 (); //Corei7 specific implementation.
+  @}
+else
+  @{
+     do_generic (); //Generic implementation.
+  @}
+@end smallexample
+@end deftypefn
+
+@deftypefn {Built-in Function} int __builtin_cpu_supports (const char *@var{feature})
+This function returns a postive integer if the runtime cpu supports @var{feature}
+ and returns @code{0} otherwise. The following features can be detected:
+
+@table @samp
+@item cmov
+CMOV instruction.
+@item mmx
+MMX instructions.
+@item popcnt
+POPCNT instruction.
+@item sse
+SSE instructions.
+@item sse2
+SSE2 instructions.
+@item sse3
+SSE3 instructions.
+@item ssse3
+SSSE3 instructions.
+@item sse4.1
+SSE4.1 instructions.
+@item sse4.2
+SSE4.2 instructions.
+@item avx
+AVX instructions.
+@end table
+
+Here is an example:
+@smallexample
+if (__builtin_cpu_supports ("popcnt"))
+  @{
+     asm("popcnt %1,%0" : "=r"(count) : "rm"(n) : "cc");
+  @}
+else
+  @{
+     count = generic_countbits (n); //generic implementation.
+  @}
+@end smallexample
+@end deftypefn
+
+
 The following built-in functions are made available by @option{-mmmx}.
 All of them generate the machine instruction that is part of the name.
 
Index: gcc/testsuite/gcc.target/i386/builtin_target.c
===================================================================
--- gcc/testsuite/gcc.target/i386/builtin_target.c        (revision 0)
+++ gcc/testsuite/gcc.target/i386/builtin_target.c        (revision 0)
@@ -0,0 +1,70 @@
+/* This test checks if the __builtin_cpu_is and __builtin_cpu_supports calls
+   are recognized. */
+
+/* { dg-do run } */
+
+#include <assert.h>
+
+int
+fn1 ()
+{
+  /* Check CPU Features.  */
+  assert (__builtin_cpu_supports ("cmov") >= 0);
+
+  assert (__builtin_cpu_supports ("mmx") >= 0);
+
+  assert (__builtin_cpu_supports ("popcnt") >= 0);
+
+  assert (__builtin_cpu_supports ("sse") >= 0);
+
+  assert (__builtin_cpu_supports ("sse2") >= 0);
+
+  assert (__builtin_cpu_supports ("sse3") >= 0);
+
+  assert (__builtin_cpu_supports ("ssse3") >= 0);
+
+  assert (__builtin_cpu_supports ("sse4.1") >= 0);
+
+  assert (__builtin_cpu_supports ("sse4.2") >= 0);
+
+  assert (__builtin_cpu_supports ("avx") >= 0);
+
+  /* Check CPU type.  */
+  assert (__builtin_cpu_is ("amd") >= 0);
+
+  assert (__builtin_cpu_is ("intel") >= 0);
+
+  assert (__builtin_cpu_is ("atom") >= 0);
+
+  assert (__builtin_cpu_is ("core2") >= 0);
+
+  assert (__builtin_cpu_is ("corei7") >= 0);
+
+  assert (__builtin_cpu_is ("nehalem") >= 0);
+
+  assert (__builtin_cpu_is ("westmere") >= 0);
+
+  assert (__builtin_cpu_is ("sandybridge") >= 0);
+
+  assert (__builtin_cpu_is ("amdfam10h") >= 0);
+
+  assert (__builtin_cpu_is ("barcelona") >= 0);
+
+  assert (__builtin_cpu_is ("shanghai") >= 0);
+
+  assert (__builtin_cpu_is ("istanbul") >= 0);
+
+  assert (__builtin_cpu_is ("amdfam15h") >= 0);
+
+  assert (__builtin_cpu_is ("bdver1") >= 0);
+
+  assert (__builtin_cpu_is ("bdver2") >= 0);
+
+  return 0;
+}
+
+int main ()
+{
+  __builtin_cpu_init ();
+  return fn1 ();
+}
Index: gcc/config/i386/i386-builtin-types.def
===================================================================
--- gcc/config/i386/i386-builtin-types.def        (revision 186788)
+++ gcc/config/i386/i386-builtin-types.def        (working copy)
@@ -155,6 +155,7 @@ DEF_FUNCTION_TYPE (INT, V4SF)
 DEF_FUNCTION_TYPE (INT, V8QI)
 DEF_FUNCTION_TYPE (INT, V8SF)
 DEF_FUNCTION_TYPE (INT, V32QI)
+DEF_FUNCTION_TYPE (INT, PCCHAR)
 DEF_FUNCTION_TYPE (INT64, INT64)
 DEF_FUNCTION_TYPE (INT64, V2DF)
 DEF_FUNCTION_TYPE (INT64, V4SF)
Index: gcc/config/i386/i386.c
===================================================================
--- gcc/config/i386/i386.c        (revision 186788)
+++ gcc/config/i386/i386.c        (working copy)
@@ -25855,6 +25855,11 @@ enum ix86_builtins
   /* CFString built-in for darwin */
   IX86_BUILTIN_CFSTRING,
 
+  /* Builtins to get CPU type and supported features. */
+  IX86_BUILTIN_CPU_INIT,
+  IX86_BUILTIN_CPU_IS,
+  IX86_BUILTIN_CPU_SUPPORTS,
+
   IX86_BUILTIN_MAX
 };
 
@@ -27673,6 +27678,334 @@ ix86_init_mmx_sse_builtins (void)
     }
 }
 
+/* This builds the processor_model struct type defined in
+   libgcc/config/i386/i386-cpuinfo.c  */
+
+static tree
+build_processor_model_struct (void)
+{
+  const char *field_name[] = {"__cpu_vendor", "__cpu_type", "__cpu_subtype",
+                              "__cpu_features"};
+  tree field = NULL_TREE, field_chain = NULL_TREE;
+  int i;
+  tree type = make_node (RECORD_TYPE);
+
+  /* The first 3 fields are unsigned int.  */
+  for (i = 0; i < 3; ++i)
+    {
+      field = build_decl (UNKNOWN_LOCATION, FIELD_DECL,
+                          get_identifier (field_name[i]), unsigned_type_node);
+      if (field_chain != NULL_TREE)
+        DECL_CHAIN (field) = field_chain;
+      field_chain = field;
+    }
+
+  /* The last field is an array of unsigned integers of size one.  */
+  field = build_decl (UNKNOWN_LOCATION, FIELD_DECL,
+                      get_identifier (field_name[3]),
+                      build_array_type (unsigned_type_node,
+                                        build_index_type (size_one_node)));
+  if (field_chain != NULL_TREE)
+    DECL_CHAIN (field) = field_chain;
+  field_chain = field;
+
+  finish_builtin_struct (type, "__processor_model", field_chain, NULL_TREE);
+  return type;
+}
+
+/* Returns a extern, comdat VAR_DECL of type TYPE and name NAME. */
+
+static tree
+make_var_decl (tree type, const char *name)
+{
+  tree new_decl;
+
+  new_decl = build_decl (UNKNOWN_LOCATION,
+                         VAR_DECL,
+                           get_identifier(name),
+                         type);
+
+  DECL_EXTERNAL (new_decl) = 1;
+  TREE_STATIC (new_decl) = 1;
+  TREE_PUBLIC (new_decl) = 1;
+  DECL_INITIAL (new_decl) = 0;
+  DECL_ARTIFICIAL (new_decl) = 0;
+  DECL_PRESERVE_P (new_decl) = 1;
+
+  make_decl_one_only (new_decl, DECL_ASSEMBLER_NAME (new_decl));
+  assemble_variable (new_decl, 0, 0, 0);
+
+  return new_decl;
+}
+
+/* FNDECL is a __builtin_cpu_is or a __builtin_cpu_supports call that is folded
+   into an integer defined in libgcc/config/i386/i386-cpuinfo.c */
+
+static tree
+fold_builtin_cpu (tree fndecl, tree *args)
+{
+  unsigned int i;
+  enum ix86_builtins fn_code = (enum ix86_builtins)
+                                DECL_FUNCTION_CODE (fndecl);
+  tree param_string_cst = NULL;
+
+  /* This is the order of bit-fields in __processor_features in
+     i386-cpuinfo.c */
+  enum processor_features
+  {
+    F_CMOV = 0,
+    F_MMX,
+    F_POPCNT,
+    F_SSE,
+    F_SSE2,
+    F_SSE3,
+    F_SSSE3,
+    F_SSE4_1,
+    F_SSE4_2,
+    F_AVX,
+    F_MAX
+  };
+
+  /* These are the values for vendor types and cpu types  and subtypes
+     in i386-cpuinfo.c.  Cpu types and subtypes should be subtracted by
+     the corresponding start value.  */
+  enum processor_model
+  {
+    M_INTEL = 1,
+    M_AMD,
+    M_CPU_TYPE_START,
+    M_INTEL_ATOM,
+    M_INTEL_CORE2,
+    M_INTEL_COREI7,
+    M_AMDFAM10H,
+    M_AMDFAM15H,
+    M_CPU_SUBTYPE_START,
+    M_INTEL_COREI7_NEHALEM,
+    M_INTEL_COREI7_WESTMERE,
+    M_INTEL_COREI7_SANDYBRIDGE,
+    M_AMDFAM10H_BARCELONA,
+    M_AMDFAM10H_SHANGHAI,
+    M_AMDFAM10H_ISTANBUL,
+    M_AMDFAM15H_BDVER1,
+    M_AMDFAM15H_BDVER2
+  };
+
+  static struct _arch_names_table
+    {
+      const char *const name;
+      const enum processor_model model;
+    }
+  const arch_names_table[] =
+    {
+      {"amd", M_AMD},
+      {"intel", M_INTEL},
+      {"atom", M_INTEL_ATOM},
+      {"core2", M_INTEL_CORE2},
+      {"corei7", M_INTEL_COREI7},
+      {"nehalem", M_INTEL_COREI7_NEHALEM},
+      {"westmere", M_INTEL_COREI7_WESTMERE},
+      {"sandybridge", M_INTEL_COREI7_SANDYBRIDGE},
+      {"amdfam10h", M_AMDFAM10H},
+      {"barcelona", M_AMDFAM10H_BARCELONA},
+      {"shanghai", M_AMDFAM10H_SHANGHAI},
+      {"istanbul", M_AMDFAM10H_ISTANBUL},
+      {"amdfam15h", M_AMDFAM15H},
+      {"bdver1", M_AMDFAM15H_BDVER1},
+      {"bdver2", M_AMDFAM15H_BDVER2},
+    };
+
+  static struct _isa_names_table
+    {
+      const char *const name;
+      const enum processor_features feature;
+    }
+  const isa_names_table[] =
+    {
+      {"cmov",   F_CMOV},
+      {"mmx",    F_MMX},
+      {"popcnt", F_POPCNT},
+      {"sse",    F_SSE},
+      {"sse2",   F_SSE2},
+      {"sse3",   F_SSE3},
+      {"ssse3",  F_SSSE3},
+      {"sse4.1", F_SSE4_1},
+      {"sse4.2", F_SSE4_2},
+      {"avx",    F_AVX}
+    };
+
+  static tree __processor_model_type = NULL_TREE;
+  static tree __cpu_model_var = NULL_TREE;
+
+  if (__processor_model_type == NULL_TREE)
+    __processor_model_type = build_processor_model_struct ();
+
+  if (__cpu_model_var == NULL_TREE)
+    __cpu_model_var = make_var_decl (__processor_model_type,
+                                     "__cpu_model");
+
+  gcc_assert ((args != NULL) && (*args != NULL));
+
+  param_string_cst = *args;
+  while (param_string_cst
+         && TREE_CODE (param_string_cst) !=  STRING_CST)
+    {
+      /* *args must be a expr that can contain other EXPRS leading to a
+         STRING_CST.   */
+      if (!EXPR_P (param_string_cst))
+         {
+          error ("Parameter to builtin must be a string constant or literal");
+          return integer_zero_node;
+        }
+      param_string_cst = TREE_OPERAND (EXPR_CHECK (param_string_cst), 0);
+    }
+
+  gcc_assert (param_string_cst);
+
+  if (fn_code == IX86_BUILTIN_CPU_IS)
+    {
+      tree ref;
+      tree field;
+      unsigned int field_val = 0;
+      unsigned int NUM_ARCH_NAMES
+        = sizeof (arch_names_table) / sizeof (struct _arch_names_table);
+
+      for (i = 0; i < NUM_ARCH_NAMES; i++)
+        if (strcmp (arch_names_table[i].name,
+            TREE_STRING_POINTER (param_string_cst)) == 0)
+          break;
+
+      if (i == NUM_ARCH_NAMES)
+        {
+          error ("Parameter to builtin not valid: %s",
+                 TREE_STRING_POINTER (param_string_cst));
+          return integer_zero_node;
+        }
+
+      field = TYPE_FIELDS (__processor_model_type);
+      field_val = arch_names_table[i].model;
+
+      /* CPU types are stored in the next field.  */
+      if (field_val > M_CPU_TYPE_START
+          && field_val < M_CPU_SUBTYPE_START)
+        {
+          field = DECL_CHAIN (field);
+          field_val -= M_CPU_TYPE_START;
+        }
+
+      /* CPU subtypes are stored in the next field.  */
+      if (field_val > M_CPU_SUBTYPE_START)
+        {
+          field = DECL_CHAIN ( DECL_CHAIN (field));
+          field_val -= M_CPU_SUBTYPE_START;
+        }
+
+      /* Get the appropriate field in __cpu_model.  */
+      ref =  build3 (COMPONENT_REF, TREE_TYPE (field), __cpu_model_var,
+                     field, NULL_TREE);
+
+      /* Check the value.  */
+      return build2 (EQ_EXPR, unsigned_type_node, ref,
+                     build_int_cstu (unsigned_type_node, field_val));
+    }
+  else if (fn_code == IX86_BUILTIN_CPU_SUPPORTS)
+    {
+      tree ref;
+      tree array_elt;
+      tree field;
+      unsigned int field_val = 0;
+      unsigned int NUM_ISA_NAMES
+        = sizeof (isa_names_table) / sizeof (struct _isa_names_table);
+
+      for (i = 0; i < NUM_ISA_NAMES; i++)
+        if (strcmp (isa_names_table[i].name,
+            TREE_STRING_POINTER (param_string_cst)) == 0)
+          break;
+
+      if (i == NUM_ISA_NAMES)
+        {
+          error ("Parameter to builtin not valid: %s",
+                        TREE_STRING_POINTER (param_string_cst));
+          return integer_zero_node;
+        }
+
+      field = TYPE_FIELDS (__processor_model_type);
+      /* Get the last field, which is __cpu_features.  */
+      while (DECL_CHAIN (field))
+        field = DECL_CHAIN (field);
+
+      /* Get the appropriate field: __cpu_model.__cpu_features  */
+      ref =  build3 (COMPONENT_REF, TREE_TYPE (field), __cpu_model_var,
+                     field, NULL_TREE);
+
+      /* Access the 0th element of __cpu_features array.  */
+      array_elt = build4 (ARRAY_REF, unsigned_type_node, ref,
+                          integer_zero_node, NULL_TREE, NULL_TREE);
+
+      field_val = (1 << isa_names_table[i].feature);
+      /* Return __cpu_model.__cpu_features[0] & field_val  */
+      return build2 (BIT_AND_EXPR, unsigned_type_node, array_elt,
+                     build_int_cstu (unsigned_type_node, field_val));
+    }
+  gcc_unreachable ();
+}
+
+static tree
+ix86_fold_builtin (tree fndecl, int n_args,
+                   tree *args, bool ignore ATTRIBUTE_UNUSED)
+{
+  if (DECL_BUILT_IN_CLASS (fndecl) == BUILT_IN_MD)
+    {
+      enum ix86_builtins fn_code = (enum ix86_builtins)
+                                   DECL_FUNCTION_CODE (fndecl);
+      if (fn_code ==  IX86_BUILTIN_CPU_IS
+          || fn_code == IX86_BUILTIN_CPU_SUPPORTS)
+        {
+          gcc_assert (n_args == 1);
+          return fold_builtin_cpu (fndecl, args);
+        }
+    }
+
+  return NULL_TREE;
+}
+
+/* Make builtins to detect cpu type and features supported.  NAME is
+   the builtin name, CODE is the builtin code, and FTYPE is the function
+   type of the builtin.  */
+
+static void
+make_cpu_type_builtin (const char* name, int code,
+                       enum ix86_builtin_func_type ftype, bool is_const)
+{
+  tree decl;
+  tree type;
+
+  type = ix86_get_builtin_func_type (ftype);
+  decl = add_builtin_function (name, type, code, BUILT_IN_MD,
+                               NULL, NULL_TREE);
+  gcc_assert (decl != NULL_TREE);
+  ix86_builtins[(int) code] = decl;
+  TREE_READONLY (decl) = is_const;
+}
+
+/* Make builtins to get CPU type and features supported.  The created
+   builtins are :
+
+   __builtin_cpu_init (), to detect cpu type and features,
+   __builtin_cpu_is ("<CPUNAME>"), to check if cpu is of type <CPUNAME>,
+   __builtin_cpu_supports ("<FEATURE>"), to check if cpu supports <FEATURE>
+   */
+
+static void
+ix86_init_platform_type_builtins (void)
+{
+  make_cpu_type_builtin ("__builtin_cpu_init", IX86_BUILTIN_CPU_INIT,
+                         INT_FTYPE_VOID, false);
+  make_cpu_type_builtin ("__builtin_cpu_is", IX86_BUILTIN_CPU_IS,
+                         INT_FTYPE_PCCHAR, true);
+  make_cpu_type_builtin ("__builtin_cpu_supports", IX86_BUILTIN_CPU_SUPPORTS,
+                         INT_FTYPE_PCCHAR, true);
+}
+
 /* Internal method for ix86_init_builtins.  */
 
 static void
@@ -27756,6 +28089,9 @@ ix86_init_builtins (void)
 
   ix86_init_builtin_types ();
 
+  /* Builtins to get CPU type and features. */
+  ix86_init_platform_type_builtins ();
+
   /* TFmode support builtins.  */
   def_builtin_const (0, "__builtin_infq",
                      FLOAT128_FTYPE_VOID, IX86_BUILTIN_INFQ);
@@ -29374,6 +29710,28 @@ ix86_expand_builtin (tree exp, rtx target, rtx sub
   enum machine_mode mode0, mode1, mode2, mode3, mode4;
   unsigned int fcode = DECL_FUNCTION_CODE (fndecl);
 
+  /* For CPU builtins that can be folded, fold first and expand the fold.  */
+  switch (fcode)
+    {
+    case IX86_BUILTIN_CPU_INIT:
+      {
+        /* Make it call __cpu_indicator_init in libgcc. */
+        tree call_expr, fndecl, type;
+        type = build_function_type_list (integer_type_node, NULL_TREE); 
+        fndecl = build_fn_decl ("__cpu_indicator_init", type);
+        call_expr = build_call_expr (fndecl, 0); 
+        return expand_expr (call_expr, target, mode, EXPAND_NORMAL);
+      }
+    case IX86_BUILTIN_CPU_IS:
+    case IX86_BUILTIN_CPU_SUPPORTS:
+      {
+        tree arg0 = CALL_EXPR_ARG (exp, 0);
+        tree fold_expr = fold_builtin_cpu (fndecl, &arg0);
+        gcc_assert (fold_expr != NULL_TREE);
+        return expand_expr (fold_expr, target, mode, EXPAND_NORMAL);
+      }
+    }
+
   /* Determine whether the builtin function is available under the current ISA.
      Originally the builtin was not created if it wasn't applicable to the
      current ISA based on the command line switches.  With function specific
@@ -39100,6 +39458,9 @@ ix86_autovectorize_vector_sizes (void)
 #undef TARGET_BUILD_BUILTIN_VA_LIST
 #define TARGET_BUILD_BUILTIN_VA_LIST ix86_build_builtin_va_list
 
+#undef TARGET_FOLD_BUILTIN
+#define TARGET_FOLD_BUILTIN ix86_fold_builtin
+
 #undef TARGET_ENUM_VA_LIST_P
 #define TARGET_ENUM_VA_LIST_P ix86_enum_va_list
 

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[H.J. Lu]

On Tue, Apr 24, 2012 at 5:10 PM, Sriraman Tallam <tmsriram@google.com> wrote:
> Hi,
>
>   Thanks for all the comments. I have made all the changes as
> mentioned and submiited the patch. Summary of changes made:
>
> * Add support for AVX
> * Fix documentation in extend.texi
> * Make it thread-safe according to H.J.'s comments.
>
> I have attached the patch. Boot-strapped and checked for test parity
> with pristine build.
>
>       * config/i386/i386.c (build_processor_model_struct): New function.
>        (make_var_decl): New function.
>        (fold_builtin_cpu): New function.
>        (ix86_fold_builtin): New function.
>        (make_cpu_type_builtin): New function.
>        (ix86_init_platform_type_builtins): New function.
>        (ix86_expand_builtin): Expand new builtins by folding them.
>        (ix86_init_builtins): Make new builtins to detect CPU type.
>        (TARGET_FOLD_BUILTIN): New macro.
>        (IX86_BUILTIN_CPU_INIT): New enum value.
>        (IX86_BUILTIN_CPU_IS): New enum value.
>        (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>        * config/i386/i386-builtin-types.def: New function type.
>        * testsuite/gcc.target/builtin_target.c: New testcase.
>        * doc/extend.texi: Document builtins.
>
>        * libgcc/config/i386/i386-cpuinfo.c: New file.
>        * libgcc/config/i386/t-cpuinfo: New file.
>        * libgcc/config.host: Include t-cpuinfo.
>        * libgcc/config/i386/libgcc-glibc.ver: Version symbol __cpu_model.
>
>

+  /* This function needs to run just once.  */
+  if (__cpu_model.__cpu_vendor)
+    return 0;
+
+  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
+  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
+    return -1;

If __get_cpuid_output returns non-zero, it will be called
repeatedly.  I think you should set __cpu_model.__cpu_vendor
to non-zero in this case.

Otherwise, it looks good to me.

Thanks.


-- 
H.J.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

On Tue, Apr 24, 2012 at 5:24 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
> On Tue, Apr 24, 2012 at 5:10 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>> Hi,
>>
>>   Thanks for all the comments. I have made all the changes as
>> mentioned and submiited the patch. Summary of changes made:
>>
>> * Add support for AVX
>> * Fix documentation in extend.texi
>> * Make it thread-safe according to H.J.'s comments.
>>
>> I have attached the patch. Boot-strapped and checked for test parity
>> with pristine build.
>>
>>       * config/i386/i386.c (build_processor_model_struct): New function.
>>        (make_var_decl): New function.
>>        (fold_builtin_cpu): New function.
>>        (ix86_fold_builtin): New function.
>>        (make_cpu_type_builtin): New function.
>>        (ix86_init_platform_type_builtins): New function.
>>        (ix86_expand_builtin): Expand new builtins by folding them.
>>        (ix86_init_builtins): Make new builtins to detect CPU type.
>>        (TARGET_FOLD_BUILTIN): New macro.
>>        (IX86_BUILTIN_CPU_INIT): New enum value.
>>        (IX86_BUILTIN_CPU_IS): New enum value.
>>        (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>>        * config/i386/i386-builtin-types.def: New function type.
>>        * testsuite/gcc.target/builtin_target.c: New testcase.
>>        * doc/extend.texi: Document builtins.
>>
>>        * libgcc/config/i386/i386-cpuinfo.c: New file.
>>        * libgcc/config/i386/t-cpuinfo: New file.
>>        * libgcc/config.host: Include t-cpuinfo.
>>        * libgcc/config/i386/libgcc-glibc.ver: Version symbol __cpu_model.
>>
>>
>
> +  /* This function needs to run just once.  */
> +  if (__cpu_model.__cpu_vendor)
> +    return 0;
> +
> +  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
> +  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
> +    return -1;
>
> If __get_cpuid_output returns non-zero, it will be called
> repeatedly.  I think you should set __cpu_model.__cpu_vendor
> to non-zero in this case.

Done now.

2012-04-24  Sriraman Tallam  <tmsriram@google.com>

	* libgcc/config/i386/i386-cpuinfo.c: Set __cpu_vendor always.


Index: libgcc/config/i386/i386-cpuinfo.c
===================================================================
--- libgcc/config/i386/i386-cpuinfo.c	(revision 186789)
+++ libgcc/config/i386/i386-cpuinfo.c	(working copy)
@@ -256,16 +256,25 @@ __cpu_indicator_init (void)

   /* Assume cpuid insn present. Run in level 0 to get vendor id. */
   if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
-    return -1;
+    {
+      __cpu_model.__cpu_vendor = VENDOR_OTHER;
+      return -1;
+    }

   vendor = ebx;
   max_level = eax;

   if (max_level < 1)
-    return -1;
+    {
+      __cpu_model.__cpu_vendor = VENDOR_OTHER;
+      return -1;
+    }

   if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
-    return -1;
+    {
+      __cpu_model.__cpu_vendor = VENDOR_OTHER;
+      return -1;
+    }

   model = (eax >> 4) & 0x0f;
   family = (eax >> 8) & 0x0f;


Thanks,
-Sri.


>
> Otherwise, it looks good to me.
>
> Thanks.
>
>
> --
> H.J.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[H.J. Lu]

On Tue, Apr 24, 2012 at 7:06 PM, Sriraman Tallam <tmsriram@google.com> wrote:
> On Tue, Apr 24, 2012 at 5:24 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
>> On Tue, Apr 24, 2012 at 5:10 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>>> Hi,
>>>
>>>   Thanks for all the comments. I have made all the changes as
>>> mentioned and submiited the patch. Summary of changes made:
>>>
>>> * Add support for AVX
>>> * Fix documentation in extend.texi
>>> * Make it thread-safe according to H.J.'s comments.
>>>
>>> I have attached the patch. Boot-strapped and checked for test parity
>>> with pristine build.
>>>
>>>       * config/i386/i386.c (build_processor_model_struct): New function.
>>>        (make_var_decl): New function.
>>>        (fold_builtin_cpu): New function.
>>>        (ix86_fold_builtin): New function.
>>>        (make_cpu_type_builtin): New function.
>>>        (ix86_init_platform_type_builtins): New function.
>>>        (ix86_expand_builtin): Expand new builtins by folding them.
>>>        (ix86_init_builtins): Make new builtins to detect CPU type.
>>>        (TARGET_FOLD_BUILTIN): New macro.
>>>        (IX86_BUILTIN_CPU_INIT): New enum value.
>>>        (IX86_BUILTIN_CPU_IS): New enum value.
>>>        (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>>>        * config/i386/i386-builtin-types.def: New function type.
>>>        * testsuite/gcc.target/builtin_target.c: New testcase.
>>>        * doc/extend.texi: Document builtins.
>>>
>>>        * libgcc/config/i386/i386-cpuinfo.c: New file.
>>>        * libgcc/config/i386/t-cpuinfo: New file.
>>>        * libgcc/config.host: Include t-cpuinfo.
>>>        * libgcc/config/i386/libgcc-glibc.ver: Version symbol __cpu_model.
>>>
>>>
>>
>> +  /* This function needs to run just once.  */
>> +  if (__cpu_model.__cpu_vendor)
>> +    return 0;
>> +
>> +  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
>> +  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
>> +    return -1;
>>
>> If __get_cpuid_output returns non-zero, it will be called
>> repeatedly.  I think you should set __cpu_model.__cpu_vendor
>> to non-zero in this case.
>
> Done now.
>
> 2012-04-24  Sriraman Tallam  <tmsriram@google.com>
>
>        * libgcc/config/i386/i386-cpuinfo.c: Set __cpu_vendor always.
>
>
> Index: libgcc/config/i386/i386-cpuinfo.c
> ===================================================================
> --- libgcc/config/i386/i386-cpuinfo.c   (revision 186789)
> +++ libgcc/config/i386/i386-cpuinfo.c   (working copy)
> @@ -256,16 +256,25 @@ __cpu_indicator_init (void)
>
>   /* Assume cpuid insn present. Run in level 0 to get vendor id. */
>   if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
> -    return -1;
> +    {
> +      __cpu_model.__cpu_vendor = VENDOR_OTHER;
> +      return -1;
> +    }
>
>   vendor = ebx;
>   max_level = eax;
>
>   if (max_level < 1)
> -    return -1;
> +    {
> +      __cpu_model.__cpu_vendor = VENDOR_OTHER;
> +      return -1;
> +    }
>
>   if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
> -    return -1;
> +    {
> +      __cpu_model.__cpu_vendor = VENDOR_OTHER;
> +      return -1;
> +    }
>
>   model = (eax >> 4) & 0x0f;
>   family = (eax >> 8) & 0x0f;
>
>
> Thanks,

Should you also handle AVX2?


-- 
H.J.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

On Tue, Apr 24, 2012 at 7:39 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
> On Tue, Apr 24, 2012 at 7:06 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>> On Tue, Apr 24, 2012 at 5:24 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
>>> On Tue, Apr 24, 2012 at 5:10 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>>>> Hi,
>>>>
>>>>   Thanks for all the comments. I have made all the changes as
>>>> mentioned and submiited the patch. Summary of changes made:
>>>>
>>>> * Add support for AVX
>>>> * Fix documentation in extend.texi
>>>> * Make it thread-safe according to H.J.'s comments.
>>>>
>>>> I have attached the patch. Boot-strapped and checked for test parity
>>>> with pristine build.
>>>>
>>>>       * config/i386/i386.c (build_processor_model_struct): New function.
>>>>        (make_var_decl): New function.
>>>>        (fold_builtin_cpu): New function.
>>>>        (ix86_fold_builtin): New function.
>>>>        (make_cpu_type_builtin): New function.
>>>>        (ix86_init_platform_type_builtins): New function.
>>>>        (ix86_expand_builtin): Expand new builtins by folding them.
>>>>        (ix86_init_builtins): Make new builtins to detect CPU type.
>>>>        (TARGET_FOLD_BUILTIN): New macro.
>>>>        (IX86_BUILTIN_CPU_INIT): New enum value.
>>>>        (IX86_BUILTIN_CPU_IS): New enum value.
>>>>        (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>>>>        * config/i386/i386-builtin-types.def: New function type.
>>>>        * testsuite/gcc.target/builtin_target.c: New testcase.
>>>>        * doc/extend.texi: Document builtins.
>>>>
>>>>        * libgcc/config/i386/i386-cpuinfo.c: New file.
>>>>        * libgcc/config/i386/t-cpuinfo: New file.
>>>>        * libgcc/config.host: Include t-cpuinfo.
>>>>        * libgcc/config/i386/libgcc-glibc.ver: Version symbol __cpu_model.
>>>>
>>>>
>>>
>>> +  /* This function needs to run just once.  */
>>> +  if (__cpu_model.__cpu_vendor)
>>> +    return 0;
>>> +
>>> +  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
>>> +  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
>>> +    return -1;
>>>
>>> If __get_cpuid_output returns non-zero, it will be called
>>> repeatedly.  I think you should set __cpu_model.__cpu_vendor
>>> to non-zero in this case.
>>
>> Done now.
>>
>> 2012-04-24  Sriraman Tallam  <tmsriram@google.com>
>>
>>        * libgcc/config/i386/i386-cpuinfo.c: Set __cpu_vendor always.
>>
>>
>> Index: libgcc/config/i386/i386-cpuinfo.c
>> ===================================================================
>> --- libgcc/config/i386/i386-cpuinfo.c   (revision 186789)
>> +++ libgcc/config/i386/i386-cpuinfo.c   (working copy)
>> @@ -256,16 +256,25 @@ __cpu_indicator_init (void)
>>
>>   /* Assume cpuid insn present. Run in level 0 to get vendor id. */
>>   if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
>> -    return -1;
>> +    {
>> +      __cpu_model.__cpu_vendor = VENDOR_OTHER;
>> +      return -1;
>> +    }
>>
>>   vendor = ebx;
>>   max_level = eax;
>>
>>   if (max_level < 1)
>> -    return -1;
>> +    {
>> +      __cpu_model.__cpu_vendor = VENDOR_OTHER;
>> +      return -1;
>> +    }
>>
>>   if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
>> -    return -1;
>> +    {
>> +      __cpu_model.__cpu_vendor = VENDOR_OTHER;
>> +      return -1;
>> +    }
>>
>>   model = (eax >> 4) & 0x0f;
>>   family = (eax >> 8) & 0x0f;
>>
>>
>> Thanks,
>
> Should you also handle AVX2?

 I cannot test it and thought will wait till I get access to a
processor with AVX2.

Thanks,
-Sri.


>
>
> --
> H.J.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[H.J. Lu]

On Wed, Apr 25, 2012 at 2:25 PM, Sriraman Tallam <tmsriram@google.com> wrote:
> On Tue, Apr 24, 2012 at 7:39 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
>> On Tue, Apr 24, 2012 at 7:06 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>>> On Tue, Apr 24, 2012 at 5:24 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
>>>> On Tue, Apr 24, 2012 at 5:10 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>>>>> Hi,
>>>>>
>>>>>   Thanks for all the comments. I have made all the changes as
>>>>> mentioned and submiited the patch. Summary of changes made:
>>>>>
>>>>> * Add support for AVX
>>>>> * Fix documentation in extend.texi
>>>>> * Make it thread-safe according to H.J.'s comments.
>>>>>
>>>>> I have attached the patch. Boot-strapped and checked for test parity
>>>>> with pristine build.
>>>>>
>>>>>       * config/i386/i386.c (build_processor_model_struct): New function.
>>>>>        (make_var_decl): New function.
>>>>>        (fold_builtin_cpu): New function.
>>>>>        (ix86_fold_builtin): New function.
>>>>>        (make_cpu_type_builtin): New function.
>>>>>        (ix86_init_platform_type_builtins): New function.
>>>>>        (ix86_expand_builtin): Expand new builtins by folding them.
>>>>>        (ix86_init_builtins): Make new builtins to detect CPU type.
>>>>>        (TARGET_FOLD_BUILTIN): New macro.
>>>>>        (IX86_BUILTIN_CPU_INIT): New enum value.
>>>>>        (IX86_BUILTIN_CPU_IS): New enum value.
>>>>>        (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>>>>>        * config/i386/i386-builtin-types.def: New function type.
>>>>>        * testsuite/gcc.target/builtin_target.c: New testcase.
>>>>>        * doc/extend.texi: Document builtins.
>>>>>
>>>>>        * libgcc/config/i386/i386-cpuinfo.c: New file.
>>>>>        * libgcc/config/i386/t-cpuinfo: New file.
>>>>>        * libgcc/config.host: Include t-cpuinfo.
>>>>>        * libgcc/config/i386/libgcc-glibc.ver: Version symbol __cpu_model.
>>>>>
>>>>>
>>>>
>>>> +  /* This function needs to run just once.  */
>>>> +  if (__cpu_model.__cpu_vendor)
>>>> +    return 0;
>>>> +
>>>> +  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
>>>> +  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
>>>> +    return -1;
>>>>
>>>> If __get_cpuid_output returns non-zero, it will be called
>>>> repeatedly.  I think you should set __cpu_model.__cpu_vendor
>>>> to non-zero in this case.
>>>
>>> Done now.
>>>
>>> 2012-04-24  Sriraman Tallam  <tmsriram@google.com>
>>>
>>>        * libgcc/config/i386/i386-cpuinfo.c: Set __cpu_vendor always.
>>>
>>>
>>> Index: libgcc/config/i386/i386-cpuinfo.c
>>> ===================================================================
>>> --- libgcc/config/i386/i386-cpuinfo.c   (revision 186789)
>>> +++ libgcc/config/i386/i386-cpuinfo.c   (working copy)
>>> @@ -256,16 +256,25 @@ __cpu_indicator_init (void)
>>>
>>>   /* Assume cpuid insn present. Run in level 0 to get vendor id. */
>>>   if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
>>> -    return -1;
>>> +    {
>>> +      __cpu_model.__cpu_vendor = VENDOR_OTHER;
>>> +      return -1;
>>> +    }
>>>
>>>   vendor = ebx;
>>>   max_level = eax;
>>>
>>>   if (max_level < 1)
>>> -    return -1;
>>> +    {
>>> +      __cpu_model.__cpu_vendor = VENDOR_OTHER;
>>> +      return -1;
>>> +    }
>>>
>>>   if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
>>> -    return -1;
>>> +    {
>>> +      __cpu_model.__cpu_vendor = VENDOR_OTHER;
>>> +      return -1;
>>> +    }
>>>
>>>   model = (eax >> 4) & 0x0f;
>>>   family = (eax >> 8) & 0x0f;
>>>
>>>
>>> Thanks,
>>
>> Should you also handle AVX2?
>
>  I cannot test it and thought will wait till I get access to a
> processor with AVX2.
>

You can download an AVX2 emulator (SDE) from

http://software.intel.com/en-us/avx/

to test AVX2 binaries.

-- 
H.J.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

On Wed, Apr 25, 2012 at 2:28 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
> On Wed, Apr 25, 2012 at 2:25 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>> On Tue, Apr 24, 2012 at 7:39 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
>>> On Tue, Apr 24, 2012 at 7:06 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>>>> On Tue, Apr 24, 2012 at 5:24 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
>>>>> On Tue, Apr 24, 2012 at 5:10 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>>>>>> Hi,
>>>>>>
>>>>>>   Thanks for all the comments. I have made all the changes as
>>>>>> mentioned and submiited the patch. Summary of changes made:
>>>>>>
>>>>>> * Add support for AVX
>>>>>> * Fix documentation in extend.texi
>>>>>> * Make it thread-safe according to H.J.'s comments.
>>>>>>
>>>>>> I have attached the patch. Boot-strapped and checked for test parity
>>>>>> with pristine build.
>>>>>>
>>>>>>       * config/i386/i386.c (build_processor_model_struct): New function.
>>>>>>        (make_var_decl): New function.
>>>>>>        (fold_builtin_cpu): New function.
>>>>>>        (ix86_fold_builtin): New function.
>>>>>>        (make_cpu_type_builtin): New function.
>>>>>>        (ix86_init_platform_type_builtins): New function.
>>>>>>        (ix86_expand_builtin): Expand new builtins by folding them.
>>>>>>        (ix86_init_builtins): Make new builtins to detect CPU type.
>>>>>>        (TARGET_FOLD_BUILTIN): New macro.
>>>>>>        (IX86_BUILTIN_CPU_INIT): New enum value.
>>>>>>        (IX86_BUILTIN_CPU_IS): New enum value.
>>>>>>        (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>>>>>>        * config/i386/i386-builtin-types.def: New function type.
>>>>>>        * testsuite/gcc.target/builtin_target.c: New testcase.
>>>>>>        * doc/extend.texi: Document builtins.
>>>>>>
>>>>>>        * libgcc/config/i386/i386-cpuinfo.c: New file.
>>>>>>        * libgcc/config/i386/t-cpuinfo: New file.
>>>>>>        * libgcc/config.host: Include t-cpuinfo.
>>>>>>        * libgcc/config/i386/libgcc-glibc.ver: Version symbol __cpu_model.
>>>>>>
>>>>>>
>>>>>
>>>>> +  /* This function needs to run just once.  */
>>>>> +  if (__cpu_model.__cpu_vendor)
>>>>> +    return 0;
>>>>> +
>>>>> +  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
>>>>> +  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
>>>>> +    return -1;
>>>>>
>>>>> If __get_cpuid_output returns non-zero, it will be called
>>>>> repeatedly.  I think you should set __cpu_model.__cpu_vendor
>>>>> to non-zero in this case.
>>>>
>>>> Done now.
>>>>
>>>> 2012-04-24  Sriraman Tallam  <tmsriram@google.com>
>>>>
>>>>        * libgcc/config/i386/i386-cpuinfo.c: Set __cpu_vendor always.
>>>>
>>>>
>>>> Index: libgcc/config/i386/i386-cpuinfo.c
>>>> ===================================================================
>>>> --- libgcc/config/i386/i386-cpuinfo.c   (revision 186789)
>>>> +++ libgcc/config/i386/i386-cpuinfo.c   (working copy)
>>>> @@ -256,16 +256,25 @@ __cpu_indicator_init (void)
>>>>
>>>>   /* Assume cpuid insn present. Run in level 0 to get vendor id. */
>>>>   if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
>>>> -    return -1;
>>>> +    {
>>>> +      __cpu_model.__cpu_vendor = VENDOR_OTHER;
>>>> +      return -1;
>>>> +    }
>>>>
>>>>   vendor = ebx;
>>>>   max_level = eax;
>>>>
>>>>   if (max_level < 1)
>>>> -    return -1;
>>>> +    {
>>>> +      __cpu_model.__cpu_vendor = VENDOR_OTHER;
>>>> +      return -1;
>>>> +    }
>>>>
>>>>   if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
>>>> -    return -1;
>>>> +    {
>>>> +      __cpu_model.__cpu_vendor = VENDOR_OTHER;
>>>> +      return -1;
>>>> +    }
>>>>
>>>>   model = (eax >> 4) & 0x0f;
>>>>   family = (eax >> 8) & 0x0f;
>>>>
>>>>
>>>> Thanks,
>>>
>>> Should you also handle AVX2?
>>
>>  I cannot test it and thought will wait till I get access to a
>> processor with AVX2.
>>
>
> You can download an AVX2 emulator (SDE) from
>
> http://software.intel.com/en-us/avx/
>
> to test AVX2 binaries.

Ok thanks, I will prepare a patch.

-Sri.

>
> --
> H.J.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

[-- Attachment #1: Type: text/plain, Size: 4763 bytes --]

Hi H.J,

   Could you please review this patch for AVX2 check?

	* config/i386/i386-cpuinfo.c (FEATURE_AVX2): New enum value.
	(get_available_features): New argument. Check for AVX2.
	(__cpu_indicator_init): Modify call to get_available_features
.
	* doc/extend.texi: Document avx2 support.
	* testsuite/gcc.target/i386/builtin_target.c: Check avx2.
	* config/i386/i386.c (fold_builtin_cpu): Add avx2.

Thanks,
-Sri.

On Wed, Apr 25, 2012 at 2:45 PM, Sriraman Tallam <tmsriram@google.com> wrote:
> On Wed, Apr 25, 2012 at 2:28 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
>> On Wed, Apr 25, 2012 at 2:25 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>>> On Tue, Apr 24, 2012 at 7:39 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
>>>> On Tue, Apr 24, 2012 at 7:06 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>>>>> On Tue, Apr 24, 2012 at 5:24 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
>>>>>> On Tue, Apr 24, 2012 at 5:10 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>>>>>>> Hi,
>>>>>>>
>>>>>>>   Thanks for all the comments. I have made all the changes as
>>>>>>> mentioned and submiited the patch. Summary of changes made:
>>>>>>>
>>>>>>> * Add support for AVX
>>>>>>> * Fix documentation in extend.texi
>>>>>>> * Make it thread-safe according to H.J.'s comments.
>>>>>>>
>>>>>>> I have attached the patch. Boot-strapped and checked for test parity
>>>>>>> with pristine build.
>>>>>>>
>>>>>>>       * config/i386/i386.c (build_processor_model_struct): New function.
>>>>>>>        (make_var_decl): New function.
>>>>>>>        (fold_builtin_cpu): New function.
>>>>>>>        (ix86_fold_builtin): New function.
>>>>>>>        (make_cpu_type_builtin): New function.
>>>>>>>        (ix86_init_platform_type_builtins): New function.
>>>>>>>        (ix86_expand_builtin): Expand new builtins by folding them.
>>>>>>>        (ix86_init_builtins): Make new builtins to detect CPU type.
>>>>>>>        (TARGET_FOLD_BUILTIN): New macro.
>>>>>>>        (IX86_BUILTIN_CPU_INIT): New enum value.
>>>>>>>        (IX86_BUILTIN_CPU_IS): New enum value.
>>>>>>>        (IX86_BUILTIN_CPU_SUPPORTS): New enum value.
>>>>>>>        * config/i386/i386-builtin-types.def: New function type.
>>>>>>>        * testsuite/gcc.target/builtin_target.c: New testcase.
>>>>>>>        * doc/extend.texi: Document builtins.
>>>>>>>
>>>>>>>        * libgcc/config/i386/i386-cpuinfo.c: New file.
>>>>>>>        * libgcc/config/i386/t-cpuinfo: New file.
>>>>>>>        * libgcc/config.host: Include t-cpuinfo.
>>>>>>>        * libgcc/config/i386/libgcc-glibc.ver: Version symbol __cpu_model.
>>>>>>>
>>>>>>>
>>>>>>
>>>>>> +  /* This function needs to run just once.  */
>>>>>> +  if (__cpu_model.__cpu_vendor)
>>>>>> +    return 0;
>>>>>> +
>>>>>> +  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
>>>>>> +  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
>>>>>> +    return -1;
>>>>>>
>>>>>> If __get_cpuid_output returns non-zero, it will be called
>>>>>> repeatedly.  I think you should set __cpu_model.__cpu_vendor
>>>>>> to non-zero in this case.
>>>>>
>>>>> Done now.
>>>>>
>>>>> 2012-04-24  Sriraman Tallam  <tmsriram@google.com>
>>>>>
>>>>>        * libgcc/config/i386/i386-cpuinfo.c: Set __cpu_vendor always.
>>>>>
>>>>>
>>>>> Index: libgcc/config/i386/i386-cpuinfo.c
>>>>> ===================================================================
>>>>> --- libgcc/config/i386/i386-cpuinfo.c   (revision 186789)
>>>>> +++ libgcc/config/i386/i386-cpuinfo.c   (working copy)
>>>>> @@ -256,16 +256,25 @@ __cpu_indicator_init (void)
>>>>>
>>>>>   /* Assume cpuid insn present. Run in level 0 to get vendor id. */
>>>>>   if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
>>>>> -    return -1;
>>>>> +    {
>>>>> +      __cpu_model.__cpu_vendor = VENDOR_OTHER;
>>>>> +      return -1;
>>>>> +    }
>>>>>
>>>>>   vendor = ebx;
>>>>>   max_level = eax;
>>>>>
>>>>>   if (max_level < 1)
>>>>> -    return -1;
>>>>> +    {
>>>>> +      __cpu_model.__cpu_vendor = VENDOR_OTHER;
>>>>> +      return -1;
>>>>> +    }
>>>>>
>>>>>   if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
>>>>> -    return -1;
>>>>> +    {
>>>>> +      __cpu_model.__cpu_vendor = VENDOR_OTHER;
>>>>> +      return -1;
>>>>> +    }
>>>>>
>>>>>   model = (eax >> 4) & 0x0f;
>>>>>   family = (eax >> 8) & 0x0f;
>>>>>
>>>>>
>>>>> Thanks,
>>>>
>>>> Should you also handle AVX2?
>>>
>>>  I cannot test it and thought will wait till I get access to a
>>> processor with AVX2.
>>>
>>
>> You can download an AVX2 emulator (SDE) from
>>
>> http://software.intel.com/en-us/avx/
>>
>> to test AVX2 binaries.
>
> Ok thanks, I will prepare a patch.
>
> -Sri.
>
>>
>> --
>> H.J.

[-- Attachment #2: avx2_patch.txt --]
[-- Type: text/plain, Size: 3610 bytes --]

	* config/i386/i386-cpuinfo.c (FEATURE_AVX2): New enum value.
	(get_available_features): New argument. Check for AVX2.
	(__cpu_indicator_init): Modify call to get_available_features.
	* doc/extend.texi: Document avx2 support.
	* testsuite/gcc.target/i386/builtin_target.c: Check avx2.
	* config/i386/i386.c (fold_builtin_cpu): Add avx2.




Index: libgcc/config/i386/i386-cpuinfo.c
===================================================================
--- libgcc/config/i386/i386-cpuinfo.c	(revision 186848)
+++ libgcc/config/i386/i386-cpuinfo.c	(working copy)
@@ -75,7 +75,8 @@ enum processor_features
   FEATURE_SSSE3,
   FEATURE_SSE4_1,
   FEATURE_SSE4_2,
-  FEATURE_AVX
+  FEATURE_AVX,
+  FEATURE_AVX2
 };
 
 struct __processor_model
@@ -191,8 +192,11 @@ get_intel_cpu (unsigned int family, unsigned int m
     }
 }	             	
 
+/* ECX and EDX are output of CPUID at level one.  MAX_CPUID_LEVEL is
+   the max possible level of CPUID insn.  */
 static void
-get_available_features (unsigned int ecx, unsigned int edx)
+get_available_features (unsigned int ecx, unsigned int edx,
+			int max_cpuid_level)
 {
   unsigned int features = 0;
 
@@ -217,6 +221,15 @@ static void
   if (ecx & bit_AVX)
     features |= (1 << FEATURE_AVX);
 
+  /* Get Advanced Features at level 7 (eax = 7, ecx = 0). */
+  if (max_cpuid_level >= 7)
+    {
+      unsigned int eax, ebx, ecx, edx;
+      __cpuid_count (7, 0, eax, ebx, ecx, edx);
+      if (ebx & bit_AVX2)
+	features |= (1 << FEATURE_AVX2);
+    }
+
   __cpu_model.__cpu_features[0] = features;
 }
 
@@ -296,7 +309,7 @@ __cpu_indicator_init (void)
       /* Get CPU type.  */
       get_intel_cpu (family, model, brand_id);
       /* Find available features. */
-      get_available_features (ecx, edx);
+      get_available_features (ecx, edx, max_level);
       __cpu_model.__cpu_vendor = VENDOR_INTEL;
     }
   else if (vendor == SIG_AMD)
@@ -311,7 +324,7 @@ __cpu_indicator_init (void)
       /* Get CPU type.  */
       get_amd_cpu (family, model);
       /* Find available features. */
-      get_available_features (ecx, edx);
+      get_available_features (ecx, edx, max_level);
       __cpu_model.__cpu_vendor = VENDOR_AMD;
     }
   else
Index: gcc/doc/extend.texi
===================================================================
--- gcc/doc/extend.texi	(revision 186848)
+++ gcc/doc/extend.texi	(working copy)
@@ -9541,6 +9541,8 @@ SSE4.1 instructions.
 SSE4.2 instructions.
 @item avx
 AVX instructions.
+@item avx2
+AVX2 instructions.
 @end table
 
 Here is an example:
Index: gcc/testsuite/gcc.target/i386/builtin_target.c
===================================================================
--- gcc/testsuite/gcc.target/i386/builtin_target.c	(revision 186848)
+++ gcc/testsuite/gcc.target/i386/builtin_target.c	(working copy)
@@ -29,6 +29,8 @@ fn1 ()
 
   assert (__builtin_cpu_supports ("avx") >= 0);
 
+  assert (__builtin_cpu_supports ("avx2") >= 0);
+
   /* Check CPU type.  */
   assert (__builtin_cpu_is ("amd") >= 0);
 
Index: gcc/config/i386/i386.c
===================================================================
--- gcc/config/i386/i386.c	(revision 186848)
+++ gcc/config/i386/i386.c	(working copy)
@@ -27763,6 +27763,7 @@ fold_builtin_cpu (tree fndecl, tree *args)
     F_SSE4_1,
     F_SSE4_2,
     F_AVX,
+    F_AVX2,
     F_MAX
   };
 
@@ -27830,7 +27831,8 @@ fold_builtin_cpu (tree fndecl, tree *args)
       {"ssse3",  F_SSSE3},
       {"sse4.1", F_SSE4_1},
       {"sse4.2", F_SSE4_2},
-      {"avx",    F_AVX}
+      {"avx",    F_AVX},
+      {"avx2",   F_AVX2}
     };
 
   static tree __processor_model_type = NULL_TREE;

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[H.J. Lu]

On Wed, Apr 25, 2012 at 4:38 PM, Sriraman Tallam <tmsriram@google.com> wrote:
> Hi H.J,
>
>   Could you please review this patch for AVX2 check?
>
>        * config/i386/i386-cpuinfo.c (FEATURE_AVX2): New enum value.
>        (get_available_features): New argument. Check for AVX2.
>        (__cpu_indicator_init): Modify call to get_available_features
> .
>        * doc/extend.texi: Document avx2 support.
>        * testsuite/gcc.target/i386/builtin_target.c: Check avx2.
>        * config/i386/i386.c (fold_builtin_cpu): Add avx2.
>

It looks good to me.

Thanks.

-- 
H.J.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

Patch committed.

Thanks,
-Sri.

On Wed, Apr 25, 2012 at 4:52 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
> On Wed, Apr 25, 2012 at 4:38 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>> Hi H.J,
>>
>>   Could you please review this patch for AVX2 check?
>>
>>        * config/i386/i386-cpuinfo.c (FEATURE_AVX2): New enum value.
>>        (get_available_features): New argument. Check for AVX2.
>>        (__cpu_indicator_init): Modify call to get_available_features
>> .
>>        * doc/extend.texi: Document avx2 support.
>>        * testsuite/gcc.target/i386/builtin_target.c: Check avx2.
>>        * config/i386/i386.c (fold_builtin_cpu): Add avx2.
>>
>
> It looks good to me.
>
> Thanks.
>
> --
> H.J.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Marc Glisse]

On Wed, 25 Apr 2012, H.J. Lu wrote:

> You can download an AVX2 emulator (SDE) from
>
> http://software.intel.com/en-us/avx/
>
> to test AVX2 binaries.

Hello,

not related to this specific patch, but do you have a dejagnu board 
description that you could share that enables running the compiler 
normally and running the compiled test programs through SDE? Doing make 
check inside the emulator is possible but takes forever, while using the 
emulator just for execution tests seems like it should be manageable.

-- 
Marc Glisse

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

[-- Attachment #1: Type: text/plain, Size: 1659 bytes --]

Hi H.J.,

I am attaching a patch to add __cpu_indicator_init to the list of
symbols to be versioned and exported in libgcc_s.so. Also, updating
builtin_target.c test to explicitly do a CPUID and check if the
features are identified correctly like you had suggested earlier.

Patch ok?


	* config/i386/libgcc-bsd.ver: Version symbol __cpu_indicator_init.
	* config/i386/libgcc-sol2.ver: Ditto.
	* config/i386/libgcc-glibc.ver: Ditto.


	* gcc.target/i386/builtin_target.c (vendor_signatures): New enum.
	(check_intel_cpu_model): New function.
	(check_amd_cpu_model): New function.
	(check_features): New function.
	(__get_cpuid_output): New function.
	(check_detailed): New function.
	(fn1): Rename to quick_check.
	(main): Update to call quick_check and call check_detailed.

Thanks,
-Sri.

On Wed, Apr 25, 2012 at 5:52 PM, Sriraman Tallam <tmsriram@google.com> wrote:
> Patch committed.
>
> Thanks,
> -Sri.
>
> On Wed, Apr 25, 2012 at 4:52 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
>> On Wed, Apr 25, 2012 at 4:38 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>>> Hi H.J,
>>>
>>>   Could you please review this patch for AVX2 check?
>>>
>>>        * config/i386/i386-cpuinfo.c (FEATURE_AVX2): New enum value.
>>>        (get_available_features): New argument. Check for AVX2.
>>>        (__cpu_indicator_init): Modify call to get_available_features
>>> .
>>>        * doc/extend.texi: Document avx2 support.
>>>        * testsuite/gcc.target/i386/builtin_target.c: Check avx2.
>>>        * config/i386/i386.c (fold_builtin_cpu): Add avx2.
>>>
>>
>> It looks good to me.
>>
>> Thanks.
>>
>> --
>> H.J.

[-- Attachment #2: cpu_runtime_patch.txt --]
[-- Type: text/plain, Size: 7245 bytes --]

Index: libgcc/config/i386/libgcc-bsd.ver
===================================================================
--- libgcc/config/i386/libgcc-bsd.ver	(revision 188246)
+++ libgcc/config/i386/libgcc-bsd.ver	(working copy)
@@ -109,4 +109,5 @@ GCC_4.6.0 {
 
 GCC_4.8.0 {
   __cpu_model
+  __cpu_indicator_init
 }
Index: libgcc/config/i386/libgcc-sol2.ver
===================================================================
--- libgcc/config/i386/libgcc-sol2.ver	(revision 188246)
+++ libgcc/config/i386/libgcc-sol2.ver	(working copy)
@@ -109,4 +109,5 @@ GCC_4.5.0 {
 
 GCC_4.8.0 {
   __cpu_model
+  __cpu_indicator_init
 }
Index: libgcc/config/i386/libgcc-glibc.ver
===================================================================
--- libgcc/config/i386/libgcc-glibc.ver	(revision 188246)
+++ libgcc/config/i386/libgcc-glibc.ver	(working copy)
@@ -150,6 +150,7 @@ GCC_4.3.0 {
 
 GCC_4.8.0 {
   __cpu_model
+  __cpu_indicator_init
 }
 %else
 GCC_4.4.0 {
@@ -190,5 +191,6 @@ GCC_4.5.0 {
 
 GCC_4.8.0 {
   __cpu_model
+  __cpu_indicator_init
 }
 %endif
Index: gcc/testsuite/gcc.target/i386/builtin_target.c
===================================================================
--- gcc/testsuite/gcc.target/i386/builtin_target.c	(revision 188246)
+++ gcc/testsuite/gcc.target/i386/builtin_target.c	(working copy)
@@ -1,13 +1,229 @@
 /* This test checks if the __builtin_cpu_is and __builtin_cpu_supports calls
-   are recognized. */
+   are recognized.  It also independently uses CPUID to get cpu type and
+   features supported and checks if the builtins correctly identify the
+   platform.  The code to do the identification is adapted from
+   libgcc/config/i386/cpuinfo.c.  */
 
 /* { dg-do run } */
 
 #include <assert.h>
+#include "cpuid.h"
 
-int
-fn1 ()
+enum vendor_signatures
 {
+  SIG_INTEL =	0x756e6547 /* Genu */,
+  SIG_AMD =	0x68747541 /* Auth */
+};
+
+/* Check if the Intel CPU model and sub-model are identified.  */
+static void
+check_intel_cpu_model (unsigned int family, unsigned int model,
+		       unsigned int brand_id)
+{
+  /* Parse family and model only if brand ID is 0. */
+  if (brand_id == 0)
+    {
+      switch (family)
+	{
+	case 0x5:
+	  /* Pentium.  */
+	  break;
+	case 0x6:
+	  switch (model)
+	    {
+	    case 0x1c:
+	    case 0x26:
+	      /* Atom.  */
+	      assert (__builtin_cpu_is ("atom"));
+	      break;
+	    case 0x1a:
+	    case 0x1e:
+	    case 0x1f:
+	    case 0x2e:
+	      /* Nehalem.  */
+	      assert (__builtin_cpu_is ("corei7"));
+	      assert (__builtin_cpu_is ("nehalem"));
+	      break;
+	    case 0x25:
+	    case 0x2c:
+	    case 0x2f:
+	      /* Westmere.  */
+	      assert (__builtin_cpu_is ("corei7"));
+	      assert (__builtin_cpu_is ("westmere"));
+	      break;
+	    case 0x2a:
+	      /* Sandy Bridge.  */
+	      assert (__builtin_cpu_is ("corei7"));
+	      assert (__builtin_cpu_is ("sandybridge"));
+	      break;
+	    case 0x17:
+	    case 0x1d:
+	      /* Penryn.  */
+	    case 0x0f:
+	      /* Merom.  */
+	      assert (__builtin_cpu_is ("core2"));
+	      break;
+	    default:
+	      break;
+	    }
+	  break;
+	default:
+	  /* We have no idea.  */
+	  break;
+	}
+    }
+}
+
+/* Check if the AMD CPU model and sub-model are identified.  */
+static void
+check_amd_cpu_model (unsigned int family, unsigned int model)
+{
+  switch (family)
+    {
+    /* AMD Family 10h.  */
+    case 0x10:
+      switch (model)
+	{
+	case 0x2:
+	  /* Barcelona.  */
+	  assert (__builtin_cpu_is ("amdfam10h"));
+	  assert (__builtin_cpu_is ("barcelona"));
+	  break;
+	case 0x4:
+	  /* Shanghai.  */
+	  assert (__builtin_cpu_is ("amdfam10h"));
+	  assert (__builtin_cpu_is ("shanghai"));
+	  break;
+	case 0x8:
+	  /* Istanbul.  */
+	  assert (__builtin_cpu_is ("amdfam10h"));
+	  assert (__builtin_cpu_is ("istanbul"));
+	  break;
+	default:
+	  break;
+	}
+      break;
+    /* AMD Family 15h.  */
+    case 0x15:
+      assert (__builtin_cpu_is ("amdfam15h"));
+      /* Bulldozer version 1.  */
+      if ( model <= 0xf)
+	assert (__builtin_cpu_is ("bdver1"));
+      /* Bulldozer version 2.  */
+      if (model >= 0x10 && model <= 0x1f)
+	assert (__builtin_cpu_is ("bdver2"));
+      break;
+    default:
+      break;
+    }
+}
+
+/* Check if the ISA features are identified.  */
+static void
+check_features (unsigned int ecx, unsigned int edx,
+		int max_cpuid_level)
+{
+  if (edx & bit_CMOV)
+    assert (__builtin_cpu_supports ("cmov"));
+  if (edx & bit_MMX)
+    assert (__builtin_cpu_supports ("mmx"));
+  if (edx & bit_SSE)
+    assert (__builtin_cpu_supports ("sse"));
+  if (edx & bit_SSE2)
+    assert (__builtin_cpu_supports ("sse2"));
+  if (ecx & bit_POPCNT)
+    assert (__builtin_cpu_supports ("popcnt"));
+  if (ecx & bit_SSE3)
+    assert (__builtin_cpu_supports ("sse3"));
+  if (ecx & bit_SSSE3)
+    assert (__builtin_cpu_supports ("ssse3"));
+  if (ecx & bit_SSE4_1)
+    assert (__builtin_cpu_supports ("sse4.1"));
+  if (ecx & bit_SSE4_2)
+    assert (__builtin_cpu_supports ("sse4.2"));
+  if (ecx & bit_AVX)
+    assert (__builtin_cpu_supports ("avx"));
+
+  /* Get advanced features at level 7 (eax = 7, ecx = 0).  */
+  if (max_cpuid_level >= 7)
+    {
+      unsigned int eax, ebx, ecx, edx;
+      __cpuid_count (7, 0, eax, ebx, ecx, edx);
+      if (ebx & bit_AVX2)
+	assert (__builtin_cpu_supports ("avx2"));
+    }
+}
+
+static int __attribute__ ((noinline))
+__get_cpuid_output (unsigned int __level,
+		    unsigned int *__eax, unsigned int *__ebx,
+		    unsigned int *__ecx, unsigned int *__edx)
+{
+  return __get_cpuid (__level, __eax, __ebx, __ecx, __edx);
+}
+
+static int
+check_detailed ()
+{
+  unsigned int eax, ebx, ecx, edx;
+
+  int max_level; 
+  unsigned int vendor;
+  unsigned int model, family, brand_id;
+  unsigned int extended_model, extended_family;
+
+  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
+  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
+    return 0;
+
+  vendor = ebx;
+  max_level = eax;
+
+  if (max_level < 1)
+    return 0;
+
+  if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
+    return 0;
+
+  model = (eax >> 4) & 0x0f;
+  family = (eax >> 8) & 0x0f;
+  brand_id = ebx & 0xff;
+  extended_model = (eax >> 12) & 0xf0;
+  extended_family = (eax >> 20) & 0xff;
+
+  if (vendor == SIG_INTEL)
+    {
+      assert (__builtin_cpu_is ("intel"));
+      /* Adjust family and model for Intel CPUs.  */
+      if (family == 0x0f)
+	{
+	  family += extended_family;
+	  model += extended_model;
+	}
+      else if (family == 0x06)
+	model += extended_model;
+      check_intel_cpu_model (family, model, brand_id);
+      check_features (ecx, edx, max_level);
+    }
+  else if (vendor == SIG_AMD)
+    {
+      assert (__builtin_cpu_is ("amd"));
+      /* Adjust model and family for AMD CPUS. */
+      if (family == 0x0f)
+	{
+	  family += extended_family;
+	  model += (extended_model << 4);
+	}
+      check_amd_cpu_model (family, model);
+      check_features (ecx, edx, max_level);
+    }
+
+  return 0;
+}
+
+static int
+quick_check ()
+{
   /* Check CPU Features.  */
   assert (__builtin_cpu_supports ("cmov") >= 0);
 
@@ -68,5 +284,7 @@
 int main ()
 {
   __builtin_cpu_init ();
-  return fn1 ();
+  quick_check ();
+  check_detailed ();
+  return 0;
 }

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[H.J. Lu]

On Tue, Jun 5, 2012 at 3:00 PM, Sriraman Tallam <tmsriram@google.com> wrote:
> Hi H.J.,
>
> I am attaching a patch to add __cpu_indicator_init to the list of
> symbols to be versioned and exported in libgcc_s.so. Also, updating
> builtin_target.c test to explicitly do a CPUID and check if the
> features are identified correctly like you had suggested earlier.
>
> Patch ok?
>
>
>        * config/i386/libgcc-bsd.ver: Version symbol __cpu_indicator_init.
>        * config/i386/libgcc-sol2.ver: Ditto.
>        * config/i386/libgcc-glibc.ver: Ditto.
>
>
>        * gcc.target/i386/builtin_target.c (vendor_signatures): New enum.
>        (check_intel_cpu_model): New function.
>        (check_amd_cpu_model): New function.
>        (check_features): New function.
>        (__get_cpuid_output): New function.
>        (check_detailed): New function.
>        (fn1): Rename to quick_check.
>        (main): Update to call quick_check and call check_detailed.
>

It looks good.  The only problem is for C programs,  __cpu_model and
__cpu_indicator_init in libgcc_s.so aren't used at all.  I suggested
in

http://gcc.gnu.org/ml/gcc-patches/2012-05/msg01816.html

We can do one
of 3 things:

1. Abuse libgcc_eh.a by moving __cpu_model and __cpu_indicator_init
from libgcc.a to libgcc_eh.a.
2. Rename libgcc_eh.a to libgcc_static.a and move __cpu_model and
__cpu_indicator_init from libgcc.a to libgcc_static.a.
3. Add  libgcc_static.a and move __cpu_model and __cpu_indicator_ini
 from libgcc.a to libgcc_static.a.  We treat libgcc_static.a similar to
libgcc_eh.a.


-- 
H.J.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

On Wed, Jun 6, 2012 at 6:52 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
> On Tue, Jun 5, 2012 at 3:00 PM, Sriraman Tallam <tmsriram@google.com> wrote:
>> Hi H.J.,
>>
>> I am attaching a patch to add __cpu_indicator_init to the list of
>> symbols to be versioned and exported in libgcc_s.so. Also, updating
>> builtin_target.c test to explicitly do a CPUID and check if the
>> features are identified correctly like you had suggested earlier.
>>
>> Patch ok?
>>
>>
>>        * config/i386/libgcc-bsd.ver: Version symbol __cpu_indicator_init.
>>        * config/i386/libgcc-sol2.ver: Ditto.
>>        * config/i386/libgcc-glibc.ver: Ditto.
>>
>>
>>        * gcc.target/i386/builtin_target.c (vendor_signatures): New enum.
>>        (check_intel_cpu_model): New function.
>>        (check_amd_cpu_model): New function.
>>        (check_features): New function.
>>        (__get_cpuid_output): New function.
>>        (check_detailed): New function.
>>        (fn1): Rename to quick_check.
>>        (main): Update to call quick_check and call check_detailed.
>>
>
> It looks good.

I submitted this patch. I am working on fixing the problem mentioned
below about cpu indicator symbols from libgcc_s.so being used in C
programs.

Thanks,
-Sri.


The only problem is for C programs,  __cpu_model and
> __cpu_indicator_init in libgcc_s.so aren't used at all.  I suggested
> in
>
> http://gcc.gnu.org/ml/gcc-patches/2012-05/msg01816.html
>
> We can do one
> of 3 things:
>
> 1. Abuse libgcc_eh.a by moving __cpu_model and __cpu_indicator_init
> from libgcc.a to libgcc_eh.a.
> 2. Rename libgcc_eh.a to libgcc_static.a and move __cpu_model and
> __cpu_indicator_init from libgcc.a to libgcc_static.a.
> 3. Add  libgcc_static.a and move __cpu_model and __cpu_indicator_ini
>  from libgcc.a to libgcc_static.a.  We treat libgcc_static.a similar to
> libgcc_eh.a.
>
>
> --
> H.J.

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

Subject:Support for Runtime CPU type detection via builtins

Hi,

   I have uploaded a new patch to only have two builtins :

* __builtin_cpu_is ("<CPUNAME>")
* __builtin_cpu_supports ("<FEATURE>")

apart from the cpu init builtin, __builtin_cpu_init.

List of CPU names :

* "amd"
* "intel"
* "atom"
* "core2"
* "corei7"
* "nehalem"
* "westmere"
* "sandybridge"
* "amdfam10h"
* "barcelona"
* "shanghai"
* "istanbul"
* "bdver1"
* "bdver2"

List of CPU features :

* "cmov"
* "mmx"
* "popcnt"
* "sse"
* "sse2"
* "sse3"
* "ssse3"
* "sse4.1"
* "sse4.2"

As an example, to check if CPU is corei7, call __builtin_cpu_is ("corei7")

Comments?
Thanks.

On Mon, Mar 12, 2012 at 4:16 AM, Richard Guenther
<richard.guenther@gmail.com> wrote:
> On Thu, Mar 8, 2012 at 9:35 PM, Xinliang David Li <davidxl@google.com> wrote:
>> On Wed, Mar 7, 2012 at 5:51 AM, Richard Guenther
>> <richard.guenther@gmail.com> wrote:
>>> On Wed, Mar 7, 2012 at 1:49 AM, Sriraman Tallam <tmsriram@google.com> wrote:
>>>> Patch for CPU detection at run-time.
>>>> ===================================
>>>>
>>>> Patch for CPU detection at run-time, to be used in dispatching of
>>>> multi-versioned functions.   Please see this discussion:
>>>> http://gcc.gnu.org/ml/gcc-patches/2011-08/msg01355.html
>>>> when this patch for reviewed the last time.
>>>>
>>>> For more detailed description:
>>>> http://gcc.gnu.org/ml/gcc/2012-03/msg00074.html
>>>>
>>>> One of the main concerns was about making CPU detection initialization a
>>>> constructor. The main point raised was about constructor ordering. I have
>>>> added a priority value to the CPU detection constructor to make it very high
>>>> priority so that it is guaranteed to fire before every constructor without
>>>> an explicitly marked priority value of 101.  However, IFUNC initializers
>>>> will still fire before this constructor, so the cpu initialization routine
>>>> has to be explicitly called in such initializers for which I have added a
>>>> builtin: __builtin_cpu_init ().
>>>>
>>>> This patch adds the following new builtins:
>>>>
>>>> * __builtin_cpu_init
>>>> * __builtin_cpu_supports_cmov
>>>> * __builtin_cpu_supports_mmx
>>>> * __builtin_cpu_supports_popcount
>>>> * __builtin_cpu_supports_sse
>>>> * __builtin_cpu_supports_sse2
>>>> * __builtin_cpu_supports_sse3
>>>> * __builtin_cpu_supports_ssse3
>>>> * __builtin_cpu_supports_sse4_1
>>>> * __builtin_cpu_supports_sse4_2
>>>> * __builtin_cpu_is_amd
>>>> * __builtin_cpu_is_intel_atom
>>>> * __builtin_cpu_is_intel_core2
>>>> * __builtin_cpu_is_intel
>>>> * __builtin_cpu_is_intel_corei7
>>>> * __builtin_cpu_is_intel_corei7_nehalem
>>>> * __builtin_cpu_is_intel_corei7_westmere
>>>> * __builtin_cpu_is_intel_corei7_sandybridge
>>>> * __builtin_cpu_is_amdfam10
>>>> * __builtin_cpu_is_amdfam10_barcelona
>>>> * __builtin_cpu_is_amdfam10_shanghai
>>>> * __builtin_cpu_is_amdfam10_istanbul
>>>> * __builtin_cpu_is_amdfam15_bdver1
>>>> * __builtin_cpu_is_amdfam15_bdver2
>>>
>>> I think the non-feature detection functions are not necessary at all.
>>
>> They are useful if compiler needs to do auto versioning based on cpu model.
>>
>>> Builtin functions are not exactly cheap, nor is the scheme you invent
>>> backward/forward compatible.  Instead, why not add a single builtin
>>> function, __builtin_cpu_supports(const char *), and decode from
>>> a comma-separated list of features?  Unknown features are simply
>>> "not present".  So I can write code with only a single configure check,
>>
>> This is a good idea.
>>
>> __builtin_is_cpu (const char* );
>> __builtin_cpu_supports (char char*);
>
> That looks good to me.
>
> Richard.
>
>> thanks,
>>
>> David
>>
>>
>>> for __builtin_cpu_supports, and cater for future features or older compilers.
>>>
>>> And of course that builtin would be even cross-platform.
>>>
>>> Implementation-wise I'll leave this to x86 maintainers to comment on.
>>>
>>> Richard.
>>>
>>>>
>>>>        * config/i386/i386.c (build_struct_with_one_bit_fields): New function.
>>>>        (make_var_decl): New function.
>>>>        (get_field_from_struct): New function.
>>>>        (fold_builtin_target): New function.
>>>>        (ix86_fold_builtin): New function.
>>>>        (ix86_expand_builtin): Expand new builtins by folding them.
>>>>        (make_platform_builtin): New functions.
>>>>        (ix86_init_platform_type_builtins): Make the new builtins.
>>>>        (ix86_init_builtins): Make new builtins to detect CPU type.
>>>>        (TARGET_FOLD_BUILTIN): New macro.
>>>>        (IX86_BUILTIN_CPU_SUPPORTS_CMOV): New enum value.
>>>>        (IX86_BUILTIN_CPU_SUPPORTS_MMX): New enum value.
>>>>        (IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT): New enum value.
>>>>        (IX86_BUILTIN_CPU_SUPPORTS_SSE): New enum value.
>>>>        (IX86_BUILTIN_CPU_SUPPORTS_SSE2): New enum value.
>>>>        (IX86_BUILTIN_CPU_SUPPORTS_SSE3): New enum value.
>>>>        (IX86_BUILTIN_CPU_SUPPORTS_SSSE3): New enum value.
>>>>        (IX86_BUILTIN_CPU_SUPPORTS_SSE4_1): New enum value.
>>>>        (IX86_BUILTIN_CPU_SUPPORTS_SSE4_2): New enum value.
>>>>        (IX86_BUILTIN_CPU_INIT): New enum value.
>>>>        (IX86_BUILTIN_CPU_IS_AMD): New enum value.
>>>>        (IX86_BUILTIN_CPU_IS_INTEL): New enum value.
>>>>        (IX86_BUILTIN_CPU_IS_INTEL_ATOM): New enum value.
>>>>        (IX86_BUILTIN_CPU_IS_INTEL_CORE2): New enum value.
>>>>        (IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM): New enum value.
>>>>        (IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE): New enum value.
>>>>        (IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE): New enum value.
>>>>        (IX86_BUILTIN_CPU_IS_AMDFAM10_BARCELONA): New enum value.
>>>>        (IX86_BUILTIN_CPU_IS_AMDFAM10_SHANGHAI): New enum value.
>>>>        (IX86_BUILTIN_CPU_IS_AMDFAM10_ISTANBUL): New enum value.
>>>>        (IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1): New enum value.
>>>>        (IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2): New enum value.
>>>>        * config/i386/i386-builtin-types.def: New function type.
>>>>        * testsuite/gcc.target/builtin_target.c: New testcase.
>>>>
>>>>        * libgcc/config/i386/i386-cpuinfo.c: New file.
>>>>        * libgcc/config/i386/t-cpuinfo: New file.
>>>>        * libgcc/config.host: Include t-cpuinfo.
>>>>        * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
>>>>        and __cpu_features.
>>>>
>>>> Index: libgcc/config.host
>>>> ===================================================================
>>>> --- libgcc/config.host  (revision 184971)
>>>> +++ libgcc/config.host  (working copy)
>>>> @@ -1142,7 +1142,7 @@ i[34567]86-*-linux* | x86_64-*-linux* | \
>>>>   i[34567]86-*-kfreebsd*-gnu | x86_64-*-kfreebsd*-gnu | \
>>>>   i[34567]86-*-knetbsd*-gnu | \
>>>>   i[34567]86-*-gnu*)
>>>> -       tmake_file="${tmake_file} t-tls i386/t-linux"
>>>> +       tmake_file="${tmake_file} t-tls i386/t-linux i386/t-cpuinfo"
>>>>        if test "$libgcc_cv_cfi" = "yes"; then
>>>>                tmake_file="${tmake_file} t-stack i386/t-stack-i386"
>>>>        fi
>>>> Index: libgcc/config/i386/t-cpuinfo
>>>> ===================================================================
>>>> --- libgcc/config/i386/t-cpuinfo        (revision 0)
>>>> +++ libgcc/config/i386/t-cpuinfo        (revision 0)
>>>> @@ -0,0 +1 @@
>>>> +LIB2ADD += $(srcdir)/config/i386/i386-cpuinfo.c
>>>> Index: libgcc/config/i386/i386-cpuinfo.c
>>>> ===================================================================
>>>> --- libgcc/config/i386/i386-cpuinfo.c   (revision 0)
>>>> +++ libgcc/config/i386/i386-cpuinfo.c   (revision 0)
>>>> @@ -0,0 +1,306 @@
>>>> +/* Get CPU type and Features for x86 processors.
>>>> +   Copyright (C) 2011 Free Software Foundation, Inc.
>>>> +   Contributed by Sriraman Tallam (tmsriram@google.com)
>>>> +
>>>> +This file is part of GCC.
>>>> +
>>>> +GCC is free software; you can redistribute it and/or modify it under
>>>> +the terms of the GNU General Public License as published by the Free
>>>> +Software Foundation; either version 3, or (at your option) any later
>>>> +version.
>>>> +
>>>> +GCC is distributed in the hope that it will be useful, but WITHOUT ANY
>>>> +WARRANTY; without even the implied warranty of MERCHANTABILITY or
>>>> +FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
>>>> +for more details.
>>>> +
>>>> +You should have received a copy of the GNU General Public License
>>>> +along with GCC; see the file COPYING3.  If not see
>>>> +<http://www.gnu.org/licenses/>. */
>>>> +
>>>> +#include "cpuid.h"
>>>> +#include "tsystem.h"
>>>> +
>>>> +int __cpu_indicator_init (void) __attribute__ ((constructor (101)));
>>>> +
>>>> +enum vendor_signatures
>>>> +{
>>>> +  SIG_INTEL =  0x756e6547 /* Genu */,
>>>> +  SIG_AMD =    0x68747541 /* Auth */
>>>> +};
>>>> +
>>>> +/* ISA Features supported. */
>>>> +
>>>> +struct __processor_features
>>>> +{
>>>> +  unsigned int __cpu_cmov : 1;
>>>> +  unsigned int __cpu_mmx : 1;
>>>> +  unsigned int __cpu_popcnt : 1;
>>>> +  unsigned int __cpu_sse : 1;
>>>> +  unsigned int __cpu_sse2 : 1;
>>>> +  unsigned int __cpu_sse3 : 1;
>>>> +  unsigned int __cpu_ssse3 : 1;
>>>> +  unsigned int __cpu_sse4_1 : 1;
>>>> +  unsigned int __cpu_sse4_2 : 1;
>>>> +} __cpu_features;
>>>> +
>>>> +/* Processor Model. */
>>>> +
>>>> +struct __processor_model
>>>> +{
>>>> +  /* Vendor. */
>>>> +  unsigned int __cpu_is_amd : 1;
>>>> +  unsigned int __cpu_is_intel : 1;
>>>> +  /* CPU type. */
>>>> +  unsigned int __cpu_is_intel_atom : 1;
>>>> +  unsigned int __cpu_is_intel_core2 : 1;
>>>> +  unsigned int __cpu_is_intel_corei7 : 1;
>>>> +  unsigned int __cpu_is_intel_corei7_nehalem : 1;
>>>> +  unsigned int __cpu_is_intel_corei7_westmere : 1;
>>>> +  unsigned int __cpu_is_intel_corei7_sandybridge : 1;
>>>> +  unsigned int __cpu_is_amdfam10h : 1;
>>>> +  unsigned int __cpu_is_amdfam10h_barcelona : 1;
>>>> +  unsigned int __cpu_is_amdfam10h_shanghai : 1;
>>>> +  unsigned int __cpu_is_amdfam10h_istanbul : 1;
>>>> +  unsigned int __cpu_is_amdfam15h_bdver1 : 1;
>>>> +  unsigned int __cpu_is_amdfam15h_bdver2 : 1;
>>>> +} __cpu_model;
>>>> +
>>>> +/* Get the specific type of AMD CPU.  */
>>>> +
>>>> +static void
>>>> +get_amd_cpu (unsigned int family, unsigned int model)
>>>> +{
>>>> +  switch (family)
>>>> +    {
>>>> +    /* AMD Family 10h.  */
>>>> +    case 0x10:
>>>> +      switch (model)
>>>> +       {
>>>> +       case 0x2:
>>>> +         /* Barcelona.  */
>>>> +         __cpu_model.__cpu_is_amdfam10h = 1;
>>>> +         __cpu_model.__cpu_is_amdfam10h_barcelona = 1;
>>>> +         break;
>>>> +       case 0x4:
>>>> +         /* Shanghai.  */
>>>> +         __cpu_model.__cpu_is_amdfam10h = 1;
>>>> +         __cpu_model.__cpu_is_amdfam10h_shanghai = 1;
>>>> +         break;
>>>> +       case 0x8:
>>>> +         /* Istanbul.  */
>>>> +         __cpu_model.__cpu_is_amdfam10h = 1;
>>>> +         __cpu_model.__cpu_is_amdfam10h_istanbul = 1;
>>>> +         break;
>>>> +       default:
>>>> +         break;
>>>> +       }
>>>> +      break;
>>>> +    /* AMD Family 15h.  */
>>>> +    case 0x15:
>>>> +      /* Bulldozer version 1.  */
>>>> +      if (model >= 0 && model <= 0xf)
>>>> +        __cpu_model.__cpu_is_amdfam15h_bdver1 = 1;
>>>> +      /* Bulldozer version 2.  */
>>>> +      if (model >= 0x10 && model <= 0x1f)
>>>> +        __cpu_model.__cpu_is_amdfam15h_bdver2 = 1;
>>>> +      break;
>>>> +    default:
>>>> +      break;
>>>> +    }
>>>> +}
>>>> +
>>>> +/* Get the specific type of Intel CPU.  */
>>>> +
>>>> +static void
>>>> +get_intel_cpu (unsigned int family, unsigned int model, unsigned int brand_id)
>>>> +{
>>>> +  /* Parse family and model only if brand ID is 0. */
>>>> +  if (brand_id == 0)
>>>> +    {
>>>> +      switch (family)
>>>> +       {
>>>> +       case 0x5:
>>>> +         /* Pentium.  */
>>>> +         break;
>>>> +       case 0x6:
>>>> +         switch (model)
>>>> +           {
>>>> +           case 0x1c:
>>>> +           case 0x26:
>>>> +             /* Atom.  */
>>>> +             __cpu_model.__cpu_is_intel_atom = 1;
>>>> +             break;
>>>> +           case 0x1a:
>>>> +           case 0x1e:
>>>> +           case 0x1f:
>>>> +           case 0x2e:
>>>> +             /* Nehalem.  */
>>>> +             __cpu_model.__cpu_is_intel_corei7 = 1;
>>>> +             __cpu_model.__cpu_is_intel_corei7_nehalem = 1;
>>>> +             break;
>>>> +           case 0x25:
>>>> +           case 0x2c:
>>>> +           case 0x2f:
>>>> +             /* Westmere.  */
>>>> +             __cpu_model.__cpu_is_intel_corei7 = 1;
>>>> +             __cpu_model.__cpu_is_intel_corei7_westmere = 1;
>>>> +             break;
>>>> +           case 0x2a:
>>>> +             /* Sandy Bridge.  */
>>>> +             __cpu_model.__cpu_is_intel_corei7 = 1;
>>>> +             __cpu_model.__cpu_is_intel_corei7_sandybridge = 1;
>>>> +             break;
>>>> +           case 0x17:
>>>> +           case 0x1d:
>>>> +             /* Penryn.  */
>>>> +           case 0x0f:
>>>> +             /* Merom.  */
>>>> +             __cpu_model.__cpu_is_intel_core2 = 1;
>>>> +             break;
>>>> +           default:
>>>> +             break;
>>>> +           }
>>>> +         break;
>>>> +       default:
>>>> +         /* We have no idea.  */
>>>> +         break;
>>>> +       }
>>>> +    }
>>>> +}
>>>> +
>>>> +static void
>>>> +get_available_features (unsigned int ecx, unsigned int edx)
>>>> +{
>>>> +  __cpu_features.__cpu_cmov = (edx & bit_CMOV) ? 1 : 0;
>>>> +  __cpu_features.__cpu_mmx = (edx & bit_MMX) ? 1 : 0;
>>>> +  __cpu_features.__cpu_sse = (edx & bit_SSE) ? 1 : 0;
>>>> +  __cpu_features.__cpu_sse2 = (edx & bit_SSE2) ? 1 : 0;
>>>> +  __cpu_features.__cpu_popcnt = (ecx & bit_POPCNT) ? 1 : 0;
>>>> +  __cpu_features.__cpu_sse3 = (ecx & bit_SSE3) ? 1 : 0;
>>>> +  __cpu_features.__cpu_ssse3 = (ecx & bit_SSSE3) ? 1 : 0;
>>>> +  __cpu_features.__cpu_sse4_1 = (ecx & bit_SSE4_1) ? 1 : 0;
>>>> +  __cpu_features.__cpu_sse4_2 = (ecx & bit_SSE4_2) ? 1 : 0;
>>>> +}
>>>> +
>>>> +
>>>> +/* Sanity check for the vendor and cpu type flags.  */
>>>> +
>>>> +static int
>>>> +sanity_check (void)
>>>> +{
>>>> +  unsigned int one_type = 0;
>>>> +
>>>> +  /* Vendor cannot be Intel and AMD. */
>>>> +  gcc_assert((__cpu_model.__cpu_is_intel == 0)
>>>> +             || (__cpu_model.__cpu_is_amd == 0));
>>>> +
>>>> +  /* Only one CPU type can be set. */
>>>> +  one_type = (__cpu_model.__cpu_is_intel_atom
>>>> +             + __cpu_model.__cpu_is_intel_core2
>>>> +             + __cpu_model.__cpu_is_intel_corei7_nehalem
>>>> +             + __cpu_model.__cpu_is_intel_corei7_westmere
>>>> +             + __cpu_model.__cpu_is_intel_corei7_sandybridge
>>>> +             + __cpu_model.__cpu_is_amdfam10h_barcelona
>>>> +             + __cpu_model.__cpu_is_amdfam10h_shanghai
>>>> +             + __cpu_model.__cpu_is_amdfam10h_istanbul
>>>> +             + __cpu_model.__cpu_is_amdfam15h_bdver1
>>>> +             + __cpu_model.__cpu_is_amdfam15h_bdver2);
>>>> +
>>>> +  gcc_assert (one_type <= 1);
>>>> +  return 0;
>>>> +}
>>>> +
>>>> +/* A noinline function calling __get_cpuid. Having many calls to
>>>> +   cpuid in one function in 32-bit mode causes GCC to complain:
>>>> +   "can’t find a register in class ‘CLOBBERED_REGS’".  This is
>>>> +   related to PR rtl-optimization 44174. */
>>>> +
>>>> +static int __attribute__ ((noinline))
>>>> +__get_cpuid_output (unsigned int __level,
>>>> +                   unsigned int *__eax, unsigned int *__ebx,
>>>> +                   unsigned int *__ecx, unsigned int *__edx)
>>>> +{
>>>> +  return __get_cpuid (__level, __eax, __ebx, __ecx, __edx);
>>>> +}
>>>> +
>>>> +
>>>> +/* A constructor function that is sets __cpu_model and __cpu_features with
>>>> +   the right values.  This needs to run only once.  This constructor is
>>>> +   given the highest priority and it should run before constructors without
>>>> +   the priority set.  However, it still runs after ifunc initializers and
>>>> +   needs to be called explicitly there.  */
>>>> +
>>>> +int __attribute__ ((constructor (101)))
>>>> +__cpu_indicator_init (void)
>>>> +{
>>>> +  unsigned int eax, ebx, ecx, edx;
>>>> +
>>>> +  int max_level = 5;
>>>> +  unsigned int vendor;
>>>> +  unsigned int model, family, brand_id;
>>>> +  unsigned int extended_model, extended_family;
>>>> +  static int called = 0;
>>>> +
>>>> +  /* This function needs to run just once.  */
>>>> +  if (called)
>>>> +    return 0;
>>>> +  else
>>>> +    called = 1;
>>>> +
>>>> +  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
>>>> +  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
>>>> +    return -1;
>>>> +
>>>> +  vendor = ebx;
>>>> +  max_level = eax;
>>>> +
>>>> +  if (max_level < 1)
>>>> +    return -1;
>>>> +
>>>> +  if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
>>>> +    return -1;
>>>> +
>>>> +  model = (eax >> 4) & 0x0f;
>>>> +  family = (eax >> 8) & 0x0f;
>>>> +  brand_id = ebx & 0xff;
>>>> +  extended_model = (eax >> 12) & 0xf0;
>>>> +  extended_family = (eax >> 20) & 0xff;
>>>> +
>>>> +  if (vendor == SIG_INTEL)
>>>> +    {
>>>> +      /* Adjust model and family for Intel CPUS. */
>>>> +      if (family == 0x0f)
>>>> +       {
>>>> +         family += extended_family;
>>>> +         model += extended_model;
>>>> +       }
>>>> +      else if (family == 0x06)
>>>> +       model += extended_model;
>>>> +
>>>> +      /* Get CPU type.  */
>>>> +      __cpu_model.__cpu_is_intel = 1;
>>>> +      get_intel_cpu (family, model, brand_id);
>>>> +    }
>>>> +
>>>> +  if (vendor == SIG_AMD)
>>>> +    {
>>>> +      /* Adjust model and family for AMD CPUS. */
>>>> +      if (family == 0x0f)
>>>> +       {
>>>> +         family += extended_family;
>>>> +         model += (extended_model << 4);
>>>> +       }
>>>> +
>>>> +      /* Get CPU type.  */
>>>> +      __cpu_model.__cpu_is_amd = 1;
>>>> +      get_amd_cpu (family, model);
>>>> +    }
>>>> +
>>>> +  /* Find available features. */
>>>> +  get_available_features (ecx, edx);
>>>> +
>>>> +  sanity_check ();
>>>> +
>>>> +  return 0;
>>>> +}
>>>> Index: libgcc/config/i386/libgcc-glibc.ver
>>>> ===================================================================
>>>> --- libgcc/config/i386/libgcc-glibc.ver (revision 184971)
>>>> +++ libgcc/config/i386/libgcc-glibc.ver (working copy)
>>>> @@ -147,6 +147,11 @@ GCC_4.3.0 {
>>>>   __trunctfxf2
>>>>   __unordtf2
>>>>  }
>>>> +
>>>> +GCC_4.8.0 {
>>>> +  __cpu_model
>>>> +  __cpu_features
>>>> +}
>>>>  %else
>>>>  GCC_4.4.0 {
>>>>   __addtf3
>>>> @@ -183,4 +188,9 @@ GCC_4.4.0 {
>>>>  GCC_4.5.0 {
>>>>   __extendxftf2
>>>>  }
>>>> +
>>>> +GCC_4.8.0 {
>>>> +  __cpu_model
>>>> +  __cpu_features
>>>> +}
>>>>  %endif
>>>> Index: gcc/config/i386/i386-builtin-types.def
>>>> ===================================================================
>>>> --- gcc/config/i386/i386-builtin-types.def      (revision 184971)
>>>> +++ gcc/config/i386/i386-builtin-types.def      (working copy)
>>>> @@ -143,6 +143,7 @@ DEF_FUNCTION_TYPE (UINT64)
>>>>  DEF_FUNCTION_TYPE (UNSIGNED)
>>>>  DEF_FUNCTION_TYPE (VOID)
>>>>  DEF_FUNCTION_TYPE (PVOID)
>>>> +DEF_FUNCTION_TYPE (INT)
>>>>
>>>>  DEF_FUNCTION_TYPE (FLOAT, FLOAT)
>>>>  DEF_FUNCTION_TYPE (FLOAT128, FLOAT128)
>>>> Index: gcc/config/i386/i386.c
>>>> ===================================================================
>>>> --- gcc/config/i386/i386.c      (revision 184971)
>>>> +++ gcc/config/i386/i386.c      (working copy)
>>>> @@ -25637,6 +25637,33 @@ enum ix86_builtins
>>>>   /* CFString built-in for darwin */
>>>>   IX86_BUILTIN_CFSTRING,
>>>>
>>>> +  /* Builtins to get CPU features. */
>>>> +  IX86_BUILTIN_CPU_SUPPORTS_CMOV,
>>>> +  IX86_BUILTIN_CPU_SUPPORTS_MMX,
>>>> +  IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT,
>>>> +  IX86_BUILTIN_CPU_SUPPORTS_SSE,
>>>> +  IX86_BUILTIN_CPU_SUPPORTS_SSE2,
>>>> +  IX86_BUILTIN_CPU_SUPPORTS_SSE3,
>>>> +  IX86_BUILTIN_CPU_SUPPORTS_SSSE3,
>>>> +  IX86_BUILTIN_CPU_SUPPORTS_SSE4_1,
>>>> +  IX86_BUILTIN_CPU_SUPPORTS_SSE4_2,
>>>> +  /* Builtins to get CPU type. */
>>>> +  IX86_BUILTIN_CPU_INIT,
>>>> +  IX86_BUILTIN_CPU_IS_AMD,
>>>> +  IX86_BUILTIN_CPU_IS_INTEL,
>>>> +  IX86_BUILTIN_CPU_IS_INTEL_ATOM,
>>>> +  IX86_BUILTIN_CPU_IS_INTEL_CORE2,
>>>> +  IX86_BUILTIN_CPU_IS_INTEL_COREI7,
>>>> +  IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM,
>>>> +  IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE,
>>>> +  IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE,
>>>> +  IX86_BUILTIN_CPU_IS_AMDFAM10H,
>>>> +  IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA,
>>>> +  IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI,
>>>> +  IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL,
>>>> +  IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1,
>>>> +  IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2,
>>>> +
>>>>   IX86_BUILTIN_MAX
>>>>  };
>>>>
>>>> @@ -27446,6 +27473,593 @@ ix86_init_mmx_sse_builtins (void)
>>>>     }
>>>>  }
>>>>
>>>> +/* Returns a struct type with name NAME and number of fields equal to
>>>> +   NUM_FIELDS.  Each field is a unsigned int bit field of length 1 bit. */
>>>> +
>>>> +static tree
>>>> +build_struct_with_one_bit_fields (int num_fields, const char *name)
>>>> +{
>>>> +  int i;
>>>> +  char field_name [10];
>>>> +  tree field = NULL_TREE, field_chain = NULL_TREE;
>>>> +  tree type = make_node (RECORD_TYPE);
>>>> +
>>>> +  strcpy (field_name, "k_field");
>>>> +
>>>> +  for (i = 0; i < num_fields; i++)
>>>> +    {
>>>> +      /* Name the fields, 0_field, 1_field, ... */
>>>> +      field_name [0] = '0' + i;
>>>> +      field = build_decl (UNKNOWN_LOCATION, FIELD_DECL,
>>>> +                         get_identifier (field_name), unsigned_type_node);
>>>> +      DECL_BIT_FIELD (field) = 1;
>>>> +      DECL_SIZE (field) = bitsize_one_node;
>>>> +      if (field_chain != NULL_TREE)
>>>> +       DECL_CHAIN (field) = field_chain;
>>>> +      field_chain = field;
>>>> +    }
>>>> +  finish_builtin_struct (type, name, field_chain, NULL_TREE);
>>>> +  return type;
>>>> +}
>>>> +
>>>> +/* Returns a extern, comdat VAR_DECL of type TYPE and name NAME. */
>>>> +
>>>> +static tree
>>>> +make_var_decl (tree type, const char *name)
>>>> +{
>>>> +  tree new_decl;
>>>> +  struct varpool_node *vnode;
>>>> +
>>>> +  new_decl = build_decl (UNKNOWN_LOCATION,
>>>> +                        VAR_DECL,
>>>> +                        get_identifier(name),
>>>> +                        type);
>>>> +
>>>> +  DECL_EXTERNAL (new_decl) = 1;
>>>> +  TREE_STATIC (new_decl) = 1;
>>>> +  TREE_PUBLIC (new_decl) = 1;
>>>> +  DECL_INITIAL (new_decl) = 0;
>>>> +  DECL_ARTIFICIAL (new_decl) = 0;
>>>> +  DECL_PRESERVE_P (new_decl) = 1;
>>>> +
>>>> +  make_decl_one_only (new_decl, DECL_ASSEMBLER_NAME (new_decl));
>>>> +  assemble_variable (new_decl, 0, 0, 0);
>>>> +
>>>> +  vnode = varpool_node (new_decl);
>>>> +  gcc_assert (vnode != NULL);
>>>> +  /* Set finalized to 1, otherwise it asserts in function "write_symbol" in
>>>> +     lto-streamer-out.c. */
>>>> +  vnode->finalized = 1;
>>>> +
>>>> +  return new_decl;
>>>> +}
>>>> +
>>>> +/* Traverses the chain of fields in STRUCT_TYPE and returns the FIELD_NUM
>>>> +   numbered field. */
>>>> +
>>>> +static tree
>>>> +get_field_from_struct (tree struct_type, int field_num)
>>>> +{
>>>> +  int i;
>>>> +  tree field = TYPE_FIELDS (struct_type);
>>>> +
>>>> +  for (i = 0; i < field_num; i++, field = DECL_CHAIN(field))
>>>> +    {
>>>> +      gcc_assert (field != NULL_TREE);
>>>> +    }
>>>> +
>>>> +  return field;
>>>> +}
>>>> +
>>>> +/* FNDECL is a __builtin_cpu_* call that is folded into an integer defined
>>>> +   in libgcc/config/i386/i386-cpuinfo.c */
>>>> +
>>>> +static tree
>>>> +fold_builtin_cpu (enum ix86_builtins fn_code)
>>>> +{
>>>> +  /* This is the order of bit-fields in __processor_features in
>>>> +     i386-cpuinfo.c */
>>>> +  enum processor_features
>>>> +  {
>>>> +    F_CMOV = 0,
>>>> +    F_MMX,
>>>> +    F_POPCNT,
>>>> +    F_SSE,
>>>> +    F_SSE2,
>>>> +    F_SSE3,
>>>> +    F_SSSE3,
>>>> +    F_SSE4_1,
>>>> +    F_SSE4_2,
>>>> +    F_MAX
>>>> +  };
>>>> +
>>>> +  /* This is the order of bit-fields in __processor_model in
>>>> +     i386-cpuinfo.c */
>>>> +  enum processor_model
>>>> +  {
>>>> +    M_AMD = 0,
>>>> +    M_INTEL,
>>>> +    M_INTEL_ATOM,
>>>> +    M_INTEL_CORE2,
>>>> +    M_INTEL_COREI7,
>>>> +    M_INTEL_COREI7_NEHALEM,
>>>> +    M_INTEL_COREI7_WESTMERE,
>>>> +    M_INTEL_COREI7_SANDYBRIDGE,
>>>> +    M_AMDFAM10H,
>>>> +    M_AMDFAM10H_BARCELONA,
>>>> +    M_AMDFAM10H_SHANGHAI,
>>>> +    M_AMDFAM10H_ISTANBUL,
>>>> +    M_AMDFAM15H_BDVER1,
>>>> +    M_AMDFAM15H_BDVER2,
>>>> +    M_MAX
>>>> +  };
>>>> +
>>>> +  static tree __processor_features_type = NULL_TREE;
>>>> +  static tree __cpu_features_var = NULL_TREE;
>>>> +  static tree __processor_model_type = NULL_TREE;
>>>> +  static tree __cpu_model_var = NULL_TREE;
>>>> +  static tree field;
>>>> +  static tree which_struct;
>>>> +
>>>> +  if (__processor_features_type == NULL_TREE)
>>>> +    __processor_features_type = build_struct_with_one_bit_fields (F_MAX,
>>>> +                                 "__processor_features");
>>>> +
>>>> +  if (__processor_model_type == NULL_TREE)
>>>> +    __processor_model_type = build_struct_with_one_bit_fields (M_MAX,
>>>> +                                 "__processor_model");
>>>> +
>>>> +  if (__cpu_features_var == NULL_TREE)
>>>> +    __cpu_features_var = make_var_decl (__processor_features_type,
>>>> +                                       "__cpu_features");
>>>> +
>>>> +  if (__cpu_model_var == NULL_TREE)
>>>> +    __cpu_model_var = make_var_decl (__processor_model_type,
>>>> +                                    "__cpu_model");
>>>> +
>>>> +  /* Look at the code to identify the field requested. */
>>>> +  switch (fn_code)
>>>> +    {
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_CMOV:
>>>> +      field = get_field_from_struct (__processor_features_type, F_CMOV);
>>>> +      which_struct = __cpu_features_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_MMX:
>>>> +      field = get_field_from_struct (__processor_features_type, F_MMX);
>>>> +      which_struct = __cpu_features_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT:
>>>> +      field = get_field_from_struct (__processor_features_type, F_POPCNT);
>>>> +      which_struct = __cpu_features_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE:
>>>> +      field = get_field_from_struct (__processor_features_type, F_SSE);
>>>> +      which_struct = __cpu_features_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE2:
>>>> +      field = get_field_from_struct (__processor_features_type, F_SSE2);
>>>> +      which_struct = __cpu_features_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE3:
>>>> +      field = get_field_from_struct (__processor_features_type, F_SSE3);
>>>> +      which_struct = __cpu_features_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSSE3:
>>>> +      field = get_field_from_struct (__processor_features_type, F_SSSE3);
>>>> +      which_struct = __cpu_features_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_1:
>>>> +      field = get_field_from_struct (__processor_features_type, F_SSE4_1);
>>>> +      which_struct = __cpu_features_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_2:
>>>> +      field = get_field_from_struct (__processor_features_type, F_SSE4_2);
>>>> +      which_struct = __cpu_features_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_IS_AMD:
>>>> +      field = get_field_from_struct (__processor_model_type, M_AMD);
>>>> +      which_struct = __cpu_model_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_IS_INTEL:
>>>> +      field = get_field_from_struct (__processor_model_type, M_INTEL);
>>>> +      which_struct = __cpu_model_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_IS_INTEL_ATOM:
>>>> +      field = get_field_from_struct (__processor_model_type, M_INTEL_ATOM);
>>>> +      which_struct = __cpu_model_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_IS_INTEL_CORE2:
>>>> +      field = get_field_from_struct (__processor_model_type, M_INTEL_CORE2);
>>>> +      which_struct = __cpu_model_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7:
>>>> +      field = get_field_from_struct (__processor_model_type,
>>>> +                                    M_INTEL_COREI7);
>>>> +      which_struct = __cpu_model_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM:
>>>> +      field = get_field_from_struct (__processor_model_type,
>>>> +                                    M_INTEL_COREI7_NEHALEM);
>>>> +      which_struct = __cpu_model_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE:
>>>> +      field = get_field_from_struct (__processor_model_type,
>>>> +                                    M_INTEL_COREI7_WESTMERE);
>>>> +      which_struct = __cpu_model_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE:
>>>> +      field = get_field_from_struct (__processor_model_type,
>>>> +                                    M_INTEL_COREI7_SANDYBRIDGE);
>>>> +      which_struct = __cpu_model_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H:
>>>> +      field = get_field_from_struct (__processor_model_type,
>>>> +                                    M_AMDFAM10H);
>>>> +      which_struct = __cpu_model_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA:
>>>> +      field = get_field_from_struct (__processor_model_type,
>>>> +                                    M_AMDFAM10H_BARCELONA);
>>>> +      which_struct = __cpu_model_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI:
>>>> +      field = get_field_from_struct (__processor_model_type,
>>>> +                                    M_AMDFAM10H_SHANGHAI);
>>>> +      which_struct = __cpu_model_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL:
>>>> +      field = get_field_from_struct (__processor_model_type,
>>>> +                                    M_AMDFAM10H_ISTANBUL);
>>>> +      which_struct = __cpu_model_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1:
>>>> +      field = get_field_from_struct (__processor_model_type,
>>>> +                                    M_AMDFAM15H_BDVER1);
>>>> +      which_struct = __cpu_model_var;
>>>> +      break;
>>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2:
>>>> +      field = get_field_from_struct (__processor_model_type,
>>>> +                                    M_AMDFAM15H_BDVER2);
>>>> +      which_struct = __cpu_model_var;
>>>> +      break;
>>>> +    default:
>>>> +      return NULL_TREE;
>>>> +    }
>>>> +
>>>> +  return build3 (COMPONENT_REF, TREE_TYPE (field), which_struct, field, NULL_TREE);
>>>> +}
>>>> +
>>>> +static tree
>>>> +ix86_fold_builtin (tree fndecl, int n_args ATTRIBUTE_UNUSED,
>>>> +                  tree *args ATTRIBUTE_UNUSED, bool ignore ATTRIBUTE_UNUSED)
>>>> +{
>>>> +  const char* decl_name = IDENTIFIER_POINTER (DECL_NAME (fndecl));
>>>> +  if (DECL_BUILT_IN_CLASS (fndecl) == BUILT_IN_MD
>>>> +      && strstr(decl_name, "__builtin_cpu") != NULL)
>>>> +    {
>>>> +      enum ix86_builtins code = (enum ix86_builtins)
>>>> +                               DECL_FUNCTION_CODE (fndecl);
>>>> +      return fold_builtin_cpu (code);
>>>> +    }
>>>> +  return NULL_TREE;
>>>> +}
>>>> +
>>>> +/* A builtin to init/return the cpu type or feature.  Returns an
>>>> +   integer and the type is a const if IS_CONST is set. */
>>>> +
>>>> +static void
>>>> +make_platform_builtin (const char* name, int code, int is_const)
>>>> +{
>>>> +  tree decl;
>>>> +  tree type;
>>>> +
>>>> +  type = ix86_get_builtin_func_type (INT_FTYPE_VOID);
>>>> +  decl = add_builtin_function (name, type, code, BUILT_IN_MD,
>>>> +                              NULL, NULL_TREE);
>>>> +  gcc_assert (decl != NULL_TREE);
>>>> +  ix86_builtins[(int) code] = decl;
>>>> +  if (is_const)
>>>> +    TREE_READONLY (decl) = 1;
>>>> +}
>>>> +
>>>> +/* Builtins to get CPU type and features supported. */
>>>> +
>>>> +static void
>>>> +ix86_init_platform_type_builtins (void)
>>>> +{
>>>> +  make_platform_builtin ("__builtin_cpu_init",
>>>> +                        IX86_BUILTIN_CPU_INIT, 0);
>>>> +  make_platform_builtin ("__builtin_cpu_supports_cmov",
>>>> +                        IX86_BUILTIN_CPU_SUPPORTS_CMOV, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_supports_mmx",
>>>> +                        IX86_BUILTIN_CPU_SUPPORTS_MMX, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_supports_popcount",
>>>> +                        IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_supports_sse",
>>>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_supports_sse2",
>>>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE2, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_supports_sse3",
>>>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE3, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_supports_ssse3",
>>>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSSE3, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_supports_sse4_1",
>>>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE4_1, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_supports_sse4_2",
>>>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE4_2, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_is_amd",
>>>> +                        IX86_BUILTIN_CPU_IS_AMD, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_is_intel_atom",
>>>> +                        IX86_BUILTIN_CPU_IS_INTEL_ATOM, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_is_intel_core2",
>>>> +                        IX86_BUILTIN_CPU_IS_INTEL_CORE2, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_is_intel",
>>>> +                        IX86_BUILTIN_CPU_IS_INTEL, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_is_intel_corei7",
>>>> +                        IX86_BUILTIN_CPU_IS_INTEL_COREI7, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_is_intel_corei7_nehalem",
>>>> +                        IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_is_intel_corei7_westmere",
>>>> +                        IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_is_intel_corei7_sandybridge",
>>>> +                        IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_is_amdfam10",
>>>> +                        IX86_BUILTIN_CPU_IS_AMDFAM10H, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_is_amdfam10_barcelona",
>>>> +                        IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_is_amdfam10_shanghai",
>>>> +                        IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_is_amdfam10_istanbul",
>>>> +                        IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_is_amdfam15_bdver1",
>>>> +                        IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1, 1);
>>>> +  make_platform_builtin ("__builtin_cpu_is_amdfam15_bdver2",
>>>> +                        IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2, 1);
>>>> +}
>>>> +
>>>>  /* Internal method for ix86_init_builtins.  */
>>>>
>>>>  static void
>>>> @@ -27529,6 +28143,9 @@ ix86_init_builtins (void)
>>>>
>>>>   ix86_init_builtin_types ();
>>>>
>>>> +  /* Builtins to get CPU type and features. */
>>>> +  ix86_init_platform_type_builtins ();
>>>> +
>>>>   /* TFmode support builtins.  */
>>>>   def_builtin_const (0, "__builtin_infq",
>>>>                     FLOAT128_FTYPE_VOID, IX86_BUILTIN_INFQ);
>>>> @@ -29145,6 +29762,48 @@ ix86_expand_builtin (tree exp, rtx target, rtx sub
>>>>   enum machine_mode mode0, mode1, mode2, mode3, mode4;
>>>>   unsigned int fcode = DECL_FUNCTION_CODE (fndecl);
>>>>
>>>> +  /* For CPU builtins that can be folded, fold first and expand the fold.  */
>>>> +  switch (fcode)
>>>> +    {
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_CMOV:
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_MMX:
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT:
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE:
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE2:
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE3:
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSSE3:
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_1:
>>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_2:
>>>> +    case IX86_BUILTIN_CPU_IS_AMD:
>>>> +    case IX86_BUILTIN_CPU_IS_INTEL:
>>>> +    case IX86_BUILTIN_CPU_IS_INTEL_ATOM:
>>>> +    case IX86_BUILTIN_CPU_IS_INTEL_CORE2:
>>>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7:
>>>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM:
>>>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE:
>>>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE:
>>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H:
>>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA:
>>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI:
>>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL:
>>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1:
>>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2:
>>>> +      {
>>>> +        tree fold_expr = fold_builtin_cpu ((enum ix86_builtins) fcode);
>>>> +       gcc_assert (fold_expr != NULL_TREE);
>>>> +        return expand_expr (fold_expr, target, mode, EXPAND_NORMAL);
>>>> +      }
>>>> +    case IX86_BUILTIN_CPU_INIT:
>>>> +      {
>>>> +       /* Make it call __cpu_indicator_init in libgcc. */
>>>> +       tree call_expr, fndecl, type;
>>>> +        type = build_function_type_list (integer_type_node, NULL_TREE);
>>>> +       fndecl = build_fn_decl ("__cpu_indicator_init", type);
>>>> +       call_expr = build_call_expr (fndecl, 0);
>>>> +       return expand_expr (call_expr, target, mode, EXPAND_NORMAL);
>>>> +      }
>>>> +    }
>>>> +
>>>>   /* Determine whether the builtin function is available under the current ISA.
>>>>      Originally the builtin was not created if it wasn't applicable to the
>>>>      current ISA based on the command line switches.  With function specific
>>>> @@ -38610,6 +39269,12 @@ ix86_autovectorize_vector_sizes (void)
>>>>  #undef TARGET_BUILD_BUILTIN_VA_LIST
>>>>  #define TARGET_BUILD_BUILTIN_VA_LIST ix86_build_builtin_va_list
>>>>
>>>> +#undef TARGET_FOLD_BUILTIN
>>>> +#define TARGET_FOLD_BUILTIN ix86_fold_builtin
>>>> +
>>>>  #undef TARGET_ENUM_VA_LIST_P
>>>>  #define TARGET_ENUM_VA_LIST_P ix86_enum_va_list
>>>> Index: gcc/testsuite/gcc.target/i386/builtin_target.c
>>>> ===================================================================
>>>> --- gcc/testsuite/gcc.target/i386/builtin_target.c      (revision 0)
>>>> +++ gcc/testsuite/gcc.target/i386/builtin_target.c      (revision 0)
>>>> @@ -0,0 +1,61 @@
>>>> +/* This test checks if the __builtin_cpu_* calls are recognized. */
>>>> +
>>>> +/* { dg-do run } */
>>>> +
>>>> +int
>>>> +fn1 ()
>>>> +{
>>>> +  if (__builtin_cpu_supports_cmov () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_supports_mmx () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_supports_popcount () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_supports_sse () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_supports_sse2 () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_supports_sse3 () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_supports_ssse3 () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_supports_sse4_1 () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_supports_sse4_2 () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_is_amd () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_is_intel () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_is_intel_atom () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_is_intel_core2 () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_is_intel_corei7 () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_is_intel_corei7_nehalem () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_is_intel_corei7_westmere () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_is_intel_corei7_sandybridge () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_is_amdfam10 () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_is_amdfam10_barcelona () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_is_amdfam10_shanghai () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_is_amdfam10_istanbul () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_is_amdfam15_bdver1 () < 0)
>>>> +    return -1;
>>>> +  if (__builtin_cpu_is_amdfam15_bdver2 () < 0)
>>>> +    return -1;
>>>> +
>>>> +  return 0;
>>>> +}
>>>> +
>>>> +int main ()
>>>> +{
>>>> +  return fn1 ();
>>>> +}
>>>>
>>>> --
>>>> This patch is available for review at http://codereview.appspot.com/5754058

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Richard Guenther]

On Thu, Mar 8, 2012 at 9:35 PM, Xinliang David Li <davidxl@google.com> wrote:
> On Wed, Mar 7, 2012 at 5:51 AM, Richard Guenther
> <richard.guenther@gmail.com> wrote:
>> On Wed, Mar 7, 2012 at 1:49 AM, Sriraman Tallam <tmsriram@google.com> wrote:
>>> Patch for CPU detection at run-time.
>>> ===================================
>>>
>>> Patch for CPU detection at run-time, to be used in dispatching of
>>> multi-versioned functions.   Please see this discussion:
>>> http://gcc.gnu.org/ml/gcc-patches/2011-08/msg01355.html
>>> when this patch for reviewed the last time.
>>>
>>> For more detailed description:
>>> http://gcc.gnu.org/ml/gcc/2012-03/msg00074.html
>>>
>>> One of the main concerns was about making CPU detection initialization a
>>> constructor. The main point raised was about constructor ordering. I have
>>> added a priority value to the CPU detection constructor to make it very high
>>> priority so that it is guaranteed to fire before every constructor without
>>> an explicitly marked priority value of 101.  However, IFUNC initializers
>>> will still fire before this constructor, so the cpu initialization routine
>>> has to be explicitly called in such initializers for which I have added a
>>> builtin: __builtin_cpu_init ().
>>>
>>> This patch adds the following new builtins:
>>>
>>> * __builtin_cpu_init
>>> * __builtin_cpu_supports_cmov
>>> * __builtin_cpu_supports_mmx
>>> * __builtin_cpu_supports_popcount
>>> * __builtin_cpu_supports_sse
>>> * __builtin_cpu_supports_sse2
>>> * __builtin_cpu_supports_sse3
>>> * __builtin_cpu_supports_ssse3
>>> * __builtin_cpu_supports_sse4_1
>>> * __builtin_cpu_supports_sse4_2
>>> * __builtin_cpu_is_amd
>>> * __builtin_cpu_is_intel_atom
>>> * __builtin_cpu_is_intel_core2
>>> * __builtin_cpu_is_intel
>>> * __builtin_cpu_is_intel_corei7
>>> * __builtin_cpu_is_intel_corei7_nehalem
>>> * __builtin_cpu_is_intel_corei7_westmere
>>> * __builtin_cpu_is_intel_corei7_sandybridge
>>> * __builtin_cpu_is_amdfam10
>>> * __builtin_cpu_is_amdfam10_barcelona
>>> * __builtin_cpu_is_amdfam10_shanghai
>>> * __builtin_cpu_is_amdfam10_istanbul
>>> * __builtin_cpu_is_amdfam15_bdver1
>>> * __builtin_cpu_is_amdfam15_bdver2
>>
>> I think the non-feature detection functions are not necessary at all.
>
> They are useful if compiler needs to do auto versioning based on cpu model.
>
>> Builtin functions are not exactly cheap, nor is the scheme you invent
>> backward/forward compatible.  Instead, why not add a single builtin
>> function, __builtin_cpu_supports(const char *), and decode from
>> a comma-separated list of features?  Unknown features are simply
>> "not present".  So I can write code with only a single configure check,
>
> This is a good idea.
>
> __builtin_is_cpu (const char* );
> __builtin_cpu_supports (char char*);

That looks good to me.

Richard.

> thanks,
>
> David
>
>
>> for __builtin_cpu_supports, and cater for future features or older compilers.
>>
>> And of course that builtin would be even cross-platform.
>>
>> Implementation-wise I'll leave this to x86 maintainers to comment on.
>>
>> Richard.
>>
>>>
>>>        * config/i386/i386.c (build_struct_with_one_bit_fields): New function.
>>>        (make_var_decl): New function.
>>>        (get_field_from_struct): New function.
>>>        (fold_builtin_target): New function.
>>>        (ix86_fold_builtin): New function.
>>>        (ix86_expand_builtin): Expand new builtins by folding them.
>>>        (make_platform_builtin): New functions.
>>>        (ix86_init_platform_type_builtins): Make the new builtins.
>>>        (ix86_init_builtins): Make new builtins to detect CPU type.
>>>        (TARGET_FOLD_BUILTIN): New macro.
>>>        (IX86_BUILTIN_CPU_SUPPORTS_CMOV): New enum value.
>>>        (IX86_BUILTIN_CPU_SUPPORTS_MMX): New enum value.
>>>        (IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT): New enum value.
>>>        (IX86_BUILTIN_CPU_SUPPORTS_SSE): New enum value.
>>>        (IX86_BUILTIN_CPU_SUPPORTS_SSE2): New enum value.
>>>        (IX86_BUILTIN_CPU_SUPPORTS_SSE3): New enum value.
>>>        (IX86_BUILTIN_CPU_SUPPORTS_SSSE3): New enum value.
>>>        (IX86_BUILTIN_CPU_SUPPORTS_SSE4_1): New enum value.
>>>        (IX86_BUILTIN_CPU_SUPPORTS_SSE4_2): New enum value.
>>>        (IX86_BUILTIN_CPU_INIT): New enum value.
>>>        (IX86_BUILTIN_CPU_IS_AMD): New enum value.
>>>        (IX86_BUILTIN_CPU_IS_INTEL): New enum value.
>>>        (IX86_BUILTIN_CPU_IS_INTEL_ATOM): New enum value.
>>>        (IX86_BUILTIN_CPU_IS_INTEL_CORE2): New enum value.
>>>        (IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM): New enum value.
>>>        (IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE): New enum value.
>>>        (IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE): New enum value.
>>>        (IX86_BUILTIN_CPU_IS_AMDFAM10_BARCELONA): New enum value.
>>>        (IX86_BUILTIN_CPU_IS_AMDFAM10_SHANGHAI): New enum value.
>>>        (IX86_BUILTIN_CPU_IS_AMDFAM10_ISTANBUL): New enum value.
>>>        (IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1): New enum value.
>>>        (IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2): New enum value.
>>>        * config/i386/i386-builtin-types.def: New function type.
>>>        * testsuite/gcc.target/builtin_target.c: New testcase.
>>>
>>>        * libgcc/config/i386/i386-cpuinfo.c: New file.
>>>        * libgcc/config/i386/t-cpuinfo: New file.
>>>        * libgcc/config.host: Include t-cpuinfo.
>>>        * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
>>>        and __cpu_features.
>>>
>>> Index: libgcc/config.host
>>> ===================================================================
>>> --- libgcc/config.host  (revision 184971)
>>> +++ libgcc/config.host  (working copy)
>>> @@ -1142,7 +1142,7 @@ i[34567]86-*-linux* | x86_64-*-linux* | \
>>>   i[34567]86-*-kfreebsd*-gnu | x86_64-*-kfreebsd*-gnu | \
>>>   i[34567]86-*-knetbsd*-gnu | \
>>>   i[34567]86-*-gnu*)
>>> -       tmake_file="${tmake_file} t-tls i386/t-linux"
>>> +       tmake_file="${tmake_file} t-tls i386/t-linux i386/t-cpuinfo"
>>>        if test "$libgcc_cv_cfi" = "yes"; then
>>>                tmake_file="${tmake_file} t-stack i386/t-stack-i386"
>>>        fi
>>> Index: libgcc/config/i386/t-cpuinfo
>>> ===================================================================
>>> --- libgcc/config/i386/t-cpuinfo        (revision 0)
>>> +++ libgcc/config/i386/t-cpuinfo        (revision 0)
>>> @@ -0,0 +1 @@
>>> +LIB2ADD += $(srcdir)/config/i386/i386-cpuinfo.c
>>> Index: libgcc/config/i386/i386-cpuinfo.c
>>> ===================================================================
>>> --- libgcc/config/i386/i386-cpuinfo.c   (revision 0)
>>> +++ libgcc/config/i386/i386-cpuinfo.c   (revision 0)
>>> @@ -0,0 +1,306 @@
>>> +/* Get CPU type and Features for x86 processors.
>>> +   Copyright (C) 2011 Free Software Foundation, Inc.
>>> +   Contributed by Sriraman Tallam (tmsriram@google.com)
>>> +
>>> +This file is part of GCC.
>>> +
>>> +GCC is free software; you can redistribute it and/or modify it under
>>> +the terms of the GNU General Public License as published by the Free
>>> +Software Foundation; either version 3, or (at your option) any later
>>> +version.
>>> +
>>> +GCC is distributed in the hope that it will be useful, but WITHOUT ANY
>>> +WARRANTY; without even the implied warranty of MERCHANTABILITY or
>>> +FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
>>> +for more details.
>>> +
>>> +You should have received a copy of the GNU General Public License
>>> +along with GCC; see the file COPYING3.  If not see
>>> +<http://www.gnu.org/licenses/>. */
>>> +
>>> +#include "cpuid.h"
>>> +#include "tsystem.h"
>>> +
>>> +int __cpu_indicator_init (void) __attribute__ ((constructor (101)));
>>> +
>>> +enum vendor_signatures
>>> +{
>>> +  SIG_INTEL =  0x756e6547 /* Genu */,
>>> +  SIG_AMD =    0x68747541 /* Auth */
>>> +};
>>> +
>>> +/* ISA Features supported. */
>>> +
>>> +struct __processor_features
>>> +{
>>> +  unsigned int __cpu_cmov : 1;
>>> +  unsigned int __cpu_mmx : 1;
>>> +  unsigned int __cpu_popcnt : 1;
>>> +  unsigned int __cpu_sse : 1;
>>> +  unsigned int __cpu_sse2 : 1;
>>> +  unsigned int __cpu_sse3 : 1;
>>> +  unsigned int __cpu_ssse3 : 1;
>>> +  unsigned int __cpu_sse4_1 : 1;
>>> +  unsigned int __cpu_sse4_2 : 1;
>>> +} __cpu_features;
>>> +
>>> +/* Processor Model. */
>>> +
>>> +struct __processor_model
>>> +{
>>> +  /* Vendor. */
>>> +  unsigned int __cpu_is_amd : 1;
>>> +  unsigned int __cpu_is_intel : 1;
>>> +  /* CPU type. */
>>> +  unsigned int __cpu_is_intel_atom : 1;
>>> +  unsigned int __cpu_is_intel_core2 : 1;
>>> +  unsigned int __cpu_is_intel_corei7 : 1;
>>> +  unsigned int __cpu_is_intel_corei7_nehalem : 1;
>>> +  unsigned int __cpu_is_intel_corei7_westmere : 1;
>>> +  unsigned int __cpu_is_intel_corei7_sandybridge : 1;
>>> +  unsigned int __cpu_is_amdfam10h : 1;
>>> +  unsigned int __cpu_is_amdfam10h_barcelona : 1;
>>> +  unsigned int __cpu_is_amdfam10h_shanghai : 1;
>>> +  unsigned int __cpu_is_amdfam10h_istanbul : 1;
>>> +  unsigned int __cpu_is_amdfam15h_bdver1 : 1;
>>> +  unsigned int __cpu_is_amdfam15h_bdver2 : 1;
>>> +} __cpu_model;
>>> +
>>> +/* Get the specific type of AMD CPU.  */
>>> +
>>> +static void
>>> +get_amd_cpu (unsigned int family, unsigned int model)
>>> +{
>>> +  switch (family)
>>> +    {
>>> +    /* AMD Family 10h.  */
>>> +    case 0x10:
>>> +      switch (model)
>>> +       {
>>> +       case 0x2:
>>> +         /* Barcelona.  */
>>> +         __cpu_model.__cpu_is_amdfam10h = 1;
>>> +         __cpu_model.__cpu_is_amdfam10h_barcelona = 1;
>>> +         break;
>>> +       case 0x4:
>>> +         /* Shanghai.  */
>>> +         __cpu_model.__cpu_is_amdfam10h = 1;
>>> +         __cpu_model.__cpu_is_amdfam10h_shanghai = 1;
>>> +         break;
>>> +       case 0x8:
>>> +         /* Istanbul.  */
>>> +         __cpu_model.__cpu_is_amdfam10h = 1;
>>> +         __cpu_model.__cpu_is_amdfam10h_istanbul = 1;
>>> +         break;
>>> +       default:
>>> +         break;
>>> +       }
>>> +      break;
>>> +    /* AMD Family 15h.  */
>>> +    case 0x15:
>>> +      /* Bulldozer version 1.  */
>>> +      if (model >= 0 && model <= 0xf)
>>> +        __cpu_model.__cpu_is_amdfam15h_bdver1 = 1;
>>> +      /* Bulldozer version 2.  */
>>> +      if (model >= 0x10 && model <= 0x1f)
>>> +        __cpu_model.__cpu_is_amdfam15h_bdver2 = 1;
>>> +      break;
>>> +    default:
>>> +      break;
>>> +    }
>>> +}
>>> +
>>> +/* Get the specific type of Intel CPU.  */
>>> +
>>> +static void
>>> +get_intel_cpu (unsigned int family, unsigned int model, unsigned int brand_id)
>>> +{
>>> +  /* Parse family and model only if brand ID is 0. */
>>> +  if (brand_id == 0)
>>> +    {
>>> +      switch (family)
>>> +       {
>>> +       case 0x5:
>>> +         /* Pentium.  */
>>> +         break;
>>> +       case 0x6:
>>> +         switch (model)
>>> +           {
>>> +           case 0x1c:
>>> +           case 0x26:
>>> +             /* Atom.  */
>>> +             __cpu_model.__cpu_is_intel_atom = 1;
>>> +             break;
>>> +           case 0x1a:
>>> +           case 0x1e:
>>> +           case 0x1f:
>>> +           case 0x2e:
>>> +             /* Nehalem.  */
>>> +             __cpu_model.__cpu_is_intel_corei7 = 1;
>>> +             __cpu_model.__cpu_is_intel_corei7_nehalem = 1;
>>> +             break;
>>> +           case 0x25:
>>> +           case 0x2c:
>>> +           case 0x2f:
>>> +             /* Westmere.  */
>>> +             __cpu_model.__cpu_is_intel_corei7 = 1;
>>> +             __cpu_model.__cpu_is_intel_corei7_westmere = 1;
>>> +             break;
>>> +           case 0x2a:
>>> +             /* Sandy Bridge.  */
>>> +             __cpu_model.__cpu_is_intel_corei7 = 1;
>>> +             __cpu_model.__cpu_is_intel_corei7_sandybridge = 1;
>>> +             break;
>>> +           case 0x17:
>>> +           case 0x1d:
>>> +             /* Penryn.  */
>>> +           case 0x0f:
>>> +             /* Merom.  */
>>> +             __cpu_model.__cpu_is_intel_core2 = 1;
>>> +             break;
>>> +           default:
>>> +             break;
>>> +           }
>>> +         break;
>>> +       default:
>>> +         /* We have no idea.  */
>>> +         break;
>>> +       }
>>> +    }
>>> +}
>>> +
>>> +static void
>>> +get_available_features (unsigned int ecx, unsigned int edx)
>>> +{
>>> +  __cpu_features.__cpu_cmov = (edx & bit_CMOV) ? 1 : 0;
>>> +  __cpu_features.__cpu_mmx = (edx & bit_MMX) ? 1 : 0;
>>> +  __cpu_features.__cpu_sse = (edx & bit_SSE) ? 1 : 0;
>>> +  __cpu_features.__cpu_sse2 = (edx & bit_SSE2) ? 1 : 0;
>>> +  __cpu_features.__cpu_popcnt = (ecx & bit_POPCNT) ? 1 : 0;
>>> +  __cpu_features.__cpu_sse3 = (ecx & bit_SSE3) ? 1 : 0;
>>> +  __cpu_features.__cpu_ssse3 = (ecx & bit_SSSE3) ? 1 : 0;
>>> +  __cpu_features.__cpu_sse4_1 = (ecx & bit_SSE4_1) ? 1 : 0;
>>> +  __cpu_features.__cpu_sse4_2 = (ecx & bit_SSE4_2) ? 1 : 0;
>>> +}
>>> +
>>> +
>>> +/* Sanity check for the vendor and cpu type flags.  */
>>> +
>>> +static int
>>> +sanity_check (void)
>>> +{
>>> +  unsigned int one_type = 0;
>>> +
>>> +  /* Vendor cannot be Intel and AMD. */
>>> +  gcc_assert((__cpu_model.__cpu_is_intel == 0)
>>> +             || (__cpu_model.__cpu_is_amd == 0));
>>> +
>>> +  /* Only one CPU type can be set. */
>>> +  one_type = (__cpu_model.__cpu_is_intel_atom
>>> +             + __cpu_model.__cpu_is_intel_core2
>>> +             + __cpu_model.__cpu_is_intel_corei7_nehalem
>>> +             + __cpu_model.__cpu_is_intel_corei7_westmere
>>> +             + __cpu_model.__cpu_is_intel_corei7_sandybridge
>>> +             + __cpu_model.__cpu_is_amdfam10h_barcelona
>>> +             + __cpu_model.__cpu_is_amdfam10h_shanghai
>>> +             + __cpu_model.__cpu_is_amdfam10h_istanbul
>>> +             + __cpu_model.__cpu_is_amdfam15h_bdver1
>>> +             + __cpu_model.__cpu_is_amdfam15h_bdver2);
>>> +
>>> +  gcc_assert (one_type <= 1);
>>> +  return 0;
>>> +}
>>> +
>>> +/* A noinline function calling __get_cpuid. Having many calls to
>>> +   cpuid in one function in 32-bit mode causes GCC to complain:
>>> +   "can’t find a register in class ‘CLOBBERED_REGS’".  This is
>>> +   related to PR rtl-optimization 44174. */
>>> +
>>> +static int __attribute__ ((noinline))
>>> +__get_cpuid_output (unsigned int __level,
>>> +                   unsigned int *__eax, unsigned int *__ebx,
>>> +                   unsigned int *__ecx, unsigned int *__edx)
>>> +{
>>> +  return __get_cpuid (__level, __eax, __ebx, __ecx, __edx);
>>> +}
>>> +
>>> +
>>> +/* A constructor function that is sets __cpu_model and __cpu_features with
>>> +   the right values.  This needs to run only once.  This constructor is
>>> +   given the highest priority and it should run before constructors without
>>> +   the priority set.  However, it still runs after ifunc initializers and
>>> +   needs to be called explicitly there.  */
>>> +
>>> +int __attribute__ ((constructor (101)))
>>> +__cpu_indicator_init (void)
>>> +{
>>> +  unsigned int eax, ebx, ecx, edx;
>>> +
>>> +  int max_level = 5;
>>> +  unsigned int vendor;
>>> +  unsigned int model, family, brand_id;
>>> +  unsigned int extended_model, extended_family;
>>> +  static int called = 0;
>>> +
>>> +  /* This function needs to run just once.  */
>>> +  if (called)
>>> +    return 0;
>>> +  else
>>> +    called = 1;
>>> +
>>> +  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
>>> +  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
>>> +    return -1;
>>> +
>>> +  vendor = ebx;
>>> +  max_level = eax;
>>> +
>>> +  if (max_level < 1)
>>> +    return -1;
>>> +
>>> +  if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
>>> +    return -1;
>>> +
>>> +  model = (eax >> 4) & 0x0f;
>>> +  family = (eax >> 8) & 0x0f;
>>> +  brand_id = ebx & 0xff;
>>> +  extended_model = (eax >> 12) & 0xf0;
>>> +  extended_family = (eax >> 20) & 0xff;
>>> +
>>> +  if (vendor == SIG_INTEL)
>>> +    {
>>> +      /* Adjust model and family for Intel CPUS. */
>>> +      if (family == 0x0f)
>>> +       {
>>> +         family += extended_family;
>>> +         model += extended_model;
>>> +       }
>>> +      else if (family == 0x06)
>>> +       model += extended_model;
>>> +
>>> +      /* Get CPU type.  */
>>> +      __cpu_model.__cpu_is_intel = 1;
>>> +      get_intel_cpu (family, model, brand_id);
>>> +    }
>>> +
>>> +  if (vendor == SIG_AMD)
>>> +    {
>>> +      /* Adjust model and family for AMD CPUS. */
>>> +      if (family == 0x0f)
>>> +       {
>>> +         family += extended_family;
>>> +         model += (extended_model << 4);
>>> +       }
>>> +
>>> +      /* Get CPU type.  */
>>> +      __cpu_model.__cpu_is_amd = 1;
>>> +      get_amd_cpu (family, model);
>>> +    }
>>> +
>>> +  /* Find available features. */
>>> +  get_available_features (ecx, edx);
>>> +
>>> +  sanity_check ();
>>> +
>>> +  return 0;
>>> +}
>>> Index: libgcc/config/i386/libgcc-glibc.ver
>>> ===================================================================
>>> --- libgcc/config/i386/libgcc-glibc.ver (revision 184971)
>>> +++ libgcc/config/i386/libgcc-glibc.ver (working copy)
>>> @@ -147,6 +147,11 @@ GCC_4.3.0 {
>>>   __trunctfxf2
>>>   __unordtf2
>>>  }
>>> +
>>> +GCC_4.8.0 {
>>> +  __cpu_model
>>> +  __cpu_features
>>> +}
>>>  %else
>>>  GCC_4.4.0 {
>>>   __addtf3
>>> @@ -183,4 +188,9 @@ GCC_4.4.0 {
>>>  GCC_4.5.0 {
>>>   __extendxftf2
>>>  }
>>> +
>>> +GCC_4.8.0 {
>>> +  __cpu_model
>>> +  __cpu_features
>>> +}
>>>  %endif
>>> Index: gcc/config/i386/i386-builtin-types.def
>>> ===================================================================
>>> --- gcc/config/i386/i386-builtin-types.def      (revision 184971)
>>> +++ gcc/config/i386/i386-builtin-types.def      (working copy)
>>> @@ -143,6 +143,7 @@ DEF_FUNCTION_TYPE (UINT64)
>>>  DEF_FUNCTION_TYPE (UNSIGNED)
>>>  DEF_FUNCTION_TYPE (VOID)
>>>  DEF_FUNCTION_TYPE (PVOID)
>>> +DEF_FUNCTION_TYPE (INT)
>>>
>>>  DEF_FUNCTION_TYPE (FLOAT, FLOAT)
>>>  DEF_FUNCTION_TYPE (FLOAT128, FLOAT128)
>>> Index: gcc/config/i386/i386.c
>>> ===================================================================
>>> --- gcc/config/i386/i386.c      (revision 184971)
>>> +++ gcc/config/i386/i386.c      (working copy)
>>> @@ -25637,6 +25637,33 @@ enum ix86_builtins
>>>   /* CFString built-in for darwin */
>>>   IX86_BUILTIN_CFSTRING,
>>>
>>> +  /* Builtins to get CPU features. */
>>> +  IX86_BUILTIN_CPU_SUPPORTS_CMOV,
>>> +  IX86_BUILTIN_CPU_SUPPORTS_MMX,
>>> +  IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT,
>>> +  IX86_BUILTIN_CPU_SUPPORTS_SSE,
>>> +  IX86_BUILTIN_CPU_SUPPORTS_SSE2,
>>> +  IX86_BUILTIN_CPU_SUPPORTS_SSE3,
>>> +  IX86_BUILTIN_CPU_SUPPORTS_SSSE3,
>>> +  IX86_BUILTIN_CPU_SUPPORTS_SSE4_1,
>>> +  IX86_BUILTIN_CPU_SUPPORTS_SSE4_2,
>>> +  /* Builtins to get CPU type. */
>>> +  IX86_BUILTIN_CPU_INIT,
>>> +  IX86_BUILTIN_CPU_IS_AMD,
>>> +  IX86_BUILTIN_CPU_IS_INTEL,
>>> +  IX86_BUILTIN_CPU_IS_INTEL_ATOM,
>>> +  IX86_BUILTIN_CPU_IS_INTEL_CORE2,
>>> +  IX86_BUILTIN_CPU_IS_INTEL_COREI7,
>>> +  IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM,
>>> +  IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE,
>>> +  IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE,
>>> +  IX86_BUILTIN_CPU_IS_AMDFAM10H,
>>> +  IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA,
>>> +  IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI,
>>> +  IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL,
>>> +  IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1,
>>> +  IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2,
>>> +
>>>   IX86_BUILTIN_MAX
>>>  };
>>>
>>> @@ -27446,6 +27473,593 @@ ix86_init_mmx_sse_builtins (void)
>>>     }
>>>  }
>>>
>>> +/* Returns a struct type with name NAME and number of fields equal to
>>> +   NUM_FIELDS.  Each field is a unsigned int bit field of length 1 bit. */
>>> +
>>> +static tree
>>> +build_struct_with_one_bit_fields (int num_fields, const char *name)
>>> +{
>>> +  int i;
>>> +  char field_name [10];
>>> +  tree field = NULL_TREE, field_chain = NULL_TREE;
>>> +  tree type = make_node (RECORD_TYPE);
>>> +
>>> +  strcpy (field_name, "k_field");
>>> +
>>> +  for (i = 0; i < num_fields; i++)
>>> +    {
>>> +      /* Name the fields, 0_field, 1_field, ... */
>>> +      field_name [0] = '0' + i;
>>> +      field = build_decl (UNKNOWN_LOCATION, FIELD_DECL,
>>> +                         get_identifier (field_name), unsigned_type_node);
>>> +      DECL_BIT_FIELD (field) = 1;
>>> +      DECL_SIZE (field) = bitsize_one_node;
>>> +      if (field_chain != NULL_TREE)
>>> +       DECL_CHAIN (field) = field_chain;
>>> +      field_chain = field;
>>> +    }
>>> +  finish_builtin_struct (type, name, field_chain, NULL_TREE);
>>> +  return type;
>>> +}
>>> +
>>> +/* Returns a extern, comdat VAR_DECL of type TYPE and name NAME. */
>>> +
>>> +static tree
>>> +make_var_decl (tree type, const char *name)
>>> +{
>>> +  tree new_decl;
>>> +  struct varpool_node *vnode;
>>> +
>>> +  new_decl = build_decl (UNKNOWN_LOCATION,
>>> +                        VAR_DECL,
>>> +                        get_identifier(name),
>>> +                        type);
>>> +
>>> +  DECL_EXTERNAL (new_decl) = 1;
>>> +  TREE_STATIC (new_decl) = 1;
>>> +  TREE_PUBLIC (new_decl) = 1;
>>> +  DECL_INITIAL (new_decl) = 0;
>>> +  DECL_ARTIFICIAL (new_decl) = 0;
>>> +  DECL_PRESERVE_P (new_decl) = 1;
>>> +
>>> +  make_decl_one_only (new_decl, DECL_ASSEMBLER_NAME (new_decl));
>>> +  assemble_variable (new_decl, 0, 0, 0);
>>> +
>>> +  vnode = varpool_node (new_decl);
>>> +  gcc_assert (vnode != NULL);
>>> +  /* Set finalized to 1, otherwise it asserts in function "write_symbol" in
>>> +     lto-streamer-out.c. */
>>> +  vnode->finalized = 1;
>>> +
>>> +  return new_decl;
>>> +}
>>> +
>>> +/* Traverses the chain of fields in STRUCT_TYPE and returns the FIELD_NUM
>>> +   numbered field. */
>>> +
>>> +static tree
>>> +get_field_from_struct (tree struct_type, int field_num)
>>> +{
>>> +  int i;
>>> +  tree field = TYPE_FIELDS (struct_type);
>>> +
>>> +  for (i = 0; i < field_num; i++, field = DECL_CHAIN(field))
>>> +    {
>>> +      gcc_assert (field != NULL_TREE);
>>> +    }
>>> +
>>> +  return field;
>>> +}
>>> +
>>> +/* FNDECL is a __builtin_cpu_* call that is folded into an integer defined
>>> +   in libgcc/config/i386/i386-cpuinfo.c */
>>> +
>>> +static tree
>>> +fold_builtin_cpu (enum ix86_builtins fn_code)
>>> +{
>>> +  /* This is the order of bit-fields in __processor_features in
>>> +     i386-cpuinfo.c */
>>> +  enum processor_features
>>> +  {
>>> +    F_CMOV = 0,
>>> +    F_MMX,
>>> +    F_POPCNT,
>>> +    F_SSE,
>>> +    F_SSE2,
>>> +    F_SSE3,
>>> +    F_SSSE3,
>>> +    F_SSE4_1,
>>> +    F_SSE4_2,
>>> +    F_MAX
>>> +  };
>>> +
>>> +  /* This is the order of bit-fields in __processor_model in
>>> +     i386-cpuinfo.c */
>>> +  enum processor_model
>>> +  {
>>> +    M_AMD = 0,
>>> +    M_INTEL,
>>> +    M_INTEL_ATOM,
>>> +    M_INTEL_CORE2,
>>> +    M_INTEL_COREI7,
>>> +    M_INTEL_COREI7_NEHALEM,
>>> +    M_INTEL_COREI7_WESTMERE,
>>> +    M_INTEL_COREI7_SANDYBRIDGE,
>>> +    M_AMDFAM10H,
>>> +    M_AMDFAM10H_BARCELONA,
>>> +    M_AMDFAM10H_SHANGHAI,
>>> +    M_AMDFAM10H_ISTANBUL,
>>> +    M_AMDFAM15H_BDVER1,
>>> +    M_AMDFAM15H_BDVER2,
>>> +    M_MAX
>>> +  };
>>> +
>>> +  static tree __processor_features_type = NULL_TREE;
>>> +  static tree __cpu_features_var = NULL_TREE;
>>> +  static tree __processor_model_type = NULL_TREE;
>>> +  static tree __cpu_model_var = NULL_TREE;
>>> +  static tree field;
>>> +  static tree which_struct;
>>> +
>>> +  if (__processor_features_type == NULL_TREE)
>>> +    __processor_features_type = build_struct_with_one_bit_fields (F_MAX,
>>> +                                 "__processor_features");
>>> +
>>> +  if (__processor_model_type == NULL_TREE)
>>> +    __processor_model_type = build_struct_with_one_bit_fields (M_MAX,
>>> +                                 "__processor_model");
>>> +
>>> +  if (__cpu_features_var == NULL_TREE)
>>> +    __cpu_features_var = make_var_decl (__processor_features_type,
>>> +                                       "__cpu_features");
>>> +
>>> +  if (__cpu_model_var == NULL_TREE)
>>> +    __cpu_model_var = make_var_decl (__processor_model_type,
>>> +                                    "__cpu_model");
>>> +
>>> +  /* Look at the code to identify the field requested. */
>>> +  switch (fn_code)
>>> +    {
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_CMOV:
>>> +      field = get_field_from_struct (__processor_features_type, F_CMOV);
>>> +      which_struct = __cpu_features_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_MMX:
>>> +      field = get_field_from_struct (__processor_features_type, F_MMX);
>>> +      which_struct = __cpu_features_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT:
>>> +      field = get_field_from_struct (__processor_features_type, F_POPCNT);
>>> +      which_struct = __cpu_features_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE:
>>> +      field = get_field_from_struct (__processor_features_type, F_SSE);
>>> +      which_struct = __cpu_features_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE2:
>>> +      field = get_field_from_struct (__processor_features_type, F_SSE2);
>>> +      which_struct = __cpu_features_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE3:
>>> +      field = get_field_from_struct (__processor_features_type, F_SSE3);
>>> +      which_struct = __cpu_features_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSSE3:
>>> +      field = get_field_from_struct (__processor_features_type, F_SSSE3);
>>> +      which_struct = __cpu_features_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_1:
>>> +      field = get_field_from_struct (__processor_features_type, F_SSE4_1);
>>> +      which_struct = __cpu_features_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_2:
>>> +      field = get_field_from_struct (__processor_features_type, F_SSE4_2);
>>> +      which_struct = __cpu_features_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_IS_AMD:
>>> +      field = get_field_from_struct (__processor_model_type, M_AMD);
>>> +      which_struct = __cpu_model_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_IS_INTEL:
>>> +      field = get_field_from_struct (__processor_model_type, M_INTEL);
>>> +      which_struct = __cpu_model_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_IS_INTEL_ATOM:
>>> +      field = get_field_from_struct (__processor_model_type, M_INTEL_ATOM);
>>> +      which_struct = __cpu_model_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_IS_INTEL_CORE2:
>>> +      field = get_field_from_struct (__processor_model_type, M_INTEL_CORE2);
>>> +      which_struct = __cpu_model_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7:
>>> +      field = get_field_from_struct (__processor_model_type,
>>> +                                    M_INTEL_COREI7);
>>> +      which_struct = __cpu_model_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM:
>>> +      field = get_field_from_struct (__processor_model_type,
>>> +                                    M_INTEL_COREI7_NEHALEM);
>>> +      which_struct = __cpu_model_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE:
>>> +      field = get_field_from_struct (__processor_model_type,
>>> +                                    M_INTEL_COREI7_WESTMERE);
>>> +      which_struct = __cpu_model_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE:
>>> +      field = get_field_from_struct (__processor_model_type,
>>> +                                    M_INTEL_COREI7_SANDYBRIDGE);
>>> +      which_struct = __cpu_model_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H:
>>> +      field = get_field_from_struct (__processor_model_type,
>>> +                                    M_AMDFAM10H);
>>> +      which_struct = __cpu_model_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA:
>>> +      field = get_field_from_struct (__processor_model_type,
>>> +                                    M_AMDFAM10H_BARCELONA);
>>> +      which_struct = __cpu_model_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI:
>>> +      field = get_field_from_struct (__processor_model_type,
>>> +                                    M_AMDFAM10H_SHANGHAI);
>>> +      which_struct = __cpu_model_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL:
>>> +      field = get_field_from_struct (__processor_model_type,
>>> +                                    M_AMDFAM10H_ISTANBUL);
>>> +      which_struct = __cpu_model_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1:
>>> +      field = get_field_from_struct (__processor_model_type,
>>> +                                    M_AMDFAM15H_BDVER1);
>>> +      which_struct = __cpu_model_var;
>>> +      break;
>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2:
>>> +      field = get_field_from_struct (__processor_model_type,
>>> +                                    M_AMDFAM15H_BDVER2);
>>> +      which_struct = __cpu_model_var;
>>> +      break;
>>> +    default:
>>> +      return NULL_TREE;
>>> +    }
>>> +
>>> +  return build3 (COMPONENT_REF, TREE_TYPE (field), which_struct, field, NULL_TREE);
>>> +}
>>> +
>>> +static tree
>>> +ix86_fold_builtin (tree fndecl, int n_args ATTRIBUTE_UNUSED,
>>> +                  tree *args ATTRIBUTE_UNUSED, bool ignore ATTRIBUTE_UNUSED)
>>> +{
>>> +  const char* decl_name = IDENTIFIER_POINTER (DECL_NAME (fndecl));
>>> +  if (DECL_BUILT_IN_CLASS (fndecl) == BUILT_IN_MD
>>> +      && strstr(decl_name, "__builtin_cpu") != NULL)
>>> +    {
>>> +      enum ix86_builtins code = (enum ix86_builtins)
>>> +                               DECL_FUNCTION_CODE (fndecl);
>>> +      return fold_builtin_cpu (code);
>>> +    }
>>> +  return NULL_TREE;
>>> +}
>>> +
>>> +/* A builtin to init/return the cpu type or feature.  Returns an
>>> +   integer and the type is a const if IS_CONST is set. */
>>> +
>>> +static void
>>> +make_platform_builtin (const char* name, int code, int is_const)
>>> +{
>>> +  tree decl;
>>> +  tree type;
>>> +
>>> +  type = ix86_get_builtin_func_type (INT_FTYPE_VOID);
>>> +  decl = add_builtin_function (name, type, code, BUILT_IN_MD,
>>> +                              NULL, NULL_TREE);
>>> +  gcc_assert (decl != NULL_TREE);
>>> +  ix86_builtins[(int) code] = decl;
>>> +  if (is_const)
>>> +    TREE_READONLY (decl) = 1;
>>> +}
>>> +
>>> +/* Builtins to get CPU type and features supported. */
>>> +
>>> +static void
>>> +ix86_init_platform_type_builtins (void)
>>> +{
>>> +  make_platform_builtin ("__builtin_cpu_init",
>>> +                        IX86_BUILTIN_CPU_INIT, 0);
>>> +  make_platform_builtin ("__builtin_cpu_supports_cmov",
>>> +                        IX86_BUILTIN_CPU_SUPPORTS_CMOV, 1);
>>> +  make_platform_builtin ("__builtin_cpu_supports_mmx",
>>> +                        IX86_BUILTIN_CPU_SUPPORTS_MMX, 1);
>>> +  make_platform_builtin ("__builtin_cpu_supports_popcount",
>>> +                        IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT, 1);
>>> +  make_platform_builtin ("__builtin_cpu_supports_sse",
>>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE, 1);
>>> +  make_platform_builtin ("__builtin_cpu_supports_sse2",
>>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE2, 1);
>>> +  make_platform_builtin ("__builtin_cpu_supports_sse3",
>>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE3, 1);
>>> +  make_platform_builtin ("__builtin_cpu_supports_ssse3",
>>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSSE3, 1);
>>> +  make_platform_builtin ("__builtin_cpu_supports_sse4_1",
>>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE4_1, 1);
>>> +  make_platform_builtin ("__builtin_cpu_supports_sse4_2",
>>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE4_2, 1);
>>> +  make_platform_builtin ("__builtin_cpu_is_amd",
>>> +                        IX86_BUILTIN_CPU_IS_AMD, 1);
>>> +  make_platform_builtin ("__builtin_cpu_is_intel_atom",
>>> +                        IX86_BUILTIN_CPU_IS_INTEL_ATOM, 1);
>>> +  make_platform_builtin ("__builtin_cpu_is_intel_core2",
>>> +                        IX86_BUILTIN_CPU_IS_INTEL_CORE2, 1);
>>> +  make_platform_builtin ("__builtin_cpu_is_intel",
>>> +                        IX86_BUILTIN_CPU_IS_INTEL, 1);
>>> +  make_platform_builtin ("__builtin_cpu_is_intel_corei7",
>>> +                        IX86_BUILTIN_CPU_IS_INTEL_COREI7, 1);
>>> +  make_platform_builtin ("__builtin_cpu_is_intel_corei7_nehalem",
>>> +                        IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM, 1);
>>> +  make_platform_builtin ("__builtin_cpu_is_intel_corei7_westmere",
>>> +                        IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE, 1);
>>> +  make_platform_builtin ("__builtin_cpu_is_intel_corei7_sandybridge",
>>> +                        IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE, 1);
>>> +  make_platform_builtin ("__builtin_cpu_is_amdfam10",
>>> +                        IX86_BUILTIN_CPU_IS_AMDFAM10H, 1);
>>> +  make_platform_builtin ("__builtin_cpu_is_amdfam10_barcelona",
>>> +                        IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA, 1);
>>> +  make_platform_builtin ("__builtin_cpu_is_amdfam10_shanghai",
>>> +                        IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI, 1);
>>> +  make_platform_builtin ("__builtin_cpu_is_amdfam10_istanbul",
>>> +                        IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL, 1);
>>> +  make_platform_builtin ("__builtin_cpu_is_amdfam15_bdver1",
>>> +                        IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1, 1);
>>> +  make_platform_builtin ("__builtin_cpu_is_amdfam15_bdver2",
>>> +                        IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2, 1);
>>> +}
>>> +
>>>  /* Internal method for ix86_init_builtins.  */
>>>
>>>  static void
>>> @@ -27529,6 +28143,9 @@ ix86_init_builtins (void)
>>>
>>>   ix86_init_builtin_types ();
>>>
>>> +  /* Builtins to get CPU type and features. */
>>> +  ix86_init_platform_type_builtins ();
>>> +
>>>   /* TFmode support builtins.  */
>>>   def_builtin_const (0, "__builtin_infq",
>>>                     FLOAT128_FTYPE_VOID, IX86_BUILTIN_INFQ);
>>> @@ -29145,6 +29762,48 @@ ix86_expand_builtin (tree exp, rtx target, rtx sub
>>>   enum machine_mode mode0, mode1, mode2, mode3, mode4;
>>>   unsigned int fcode = DECL_FUNCTION_CODE (fndecl);
>>>
>>> +  /* For CPU builtins that can be folded, fold first and expand the fold.  */
>>> +  switch (fcode)
>>> +    {
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_CMOV:
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_MMX:
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT:
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE:
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE2:
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE3:
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSSE3:
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_1:
>>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_2:
>>> +    case IX86_BUILTIN_CPU_IS_AMD:
>>> +    case IX86_BUILTIN_CPU_IS_INTEL:
>>> +    case IX86_BUILTIN_CPU_IS_INTEL_ATOM:
>>> +    case IX86_BUILTIN_CPU_IS_INTEL_CORE2:
>>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7:
>>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM:
>>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE:
>>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE:
>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H:
>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA:
>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI:
>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL:
>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1:
>>> +    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2:
>>> +      {
>>> +        tree fold_expr = fold_builtin_cpu ((enum ix86_builtins) fcode);
>>> +       gcc_assert (fold_expr != NULL_TREE);
>>> +        return expand_expr (fold_expr, target, mode, EXPAND_NORMAL);
>>> +      }
>>> +    case IX86_BUILTIN_CPU_INIT:
>>> +      {
>>> +       /* Make it call __cpu_indicator_init in libgcc. */
>>> +       tree call_expr, fndecl, type;
>>> +        type = build_function_type_list (integer_type_node, NULL_TREE);
>>> +       fndecl = build_fn_decl ("__cpu_indicator_init", type);
>>> +       call_expr = build_call_expr (fndecl, 0);
>>> +       return expand_expr (call_expr, target, mode, EXPAND_NORMAL);
>>> +      }
>>> +    }
>>> +
>>>   /* Determine whether the builtin function is available under the current ISA.
>>>      Originally the builtin was not created if it wasn't applicable to the
>>>      current ISA based on the command line switches.  With function specific
>>> @@ -38610,6 +39269,12 @@ ix86_autovectorize_vector_sizes (void)
>>>  #undef TARGET_BUILD_BUILTIN_VA_LIST
>>>  #define TARGET_BUILD_BUILTIN_VA_LIST ix86_build_builtin_va_list
>>>
>>> +#undef TARGET_FOLD_BUILTIN
>>> +#define TARGET_FOLD_BUILTIN ix86_fold_builtin
>>> +
>>>  #undef TARGET_ENUM_VA_LIST_P
>>>  #define TARGET_ENUM_VA_LIST_P ix86_enum_va_list
>>> Index: gcc/testsuite/gcc.target/i386/builtin_target.c
>>> ===================================================================
>>> --- gcc/testsuite/gcc.target/i386/builtin_target.c      (revision 0)
>>> +++ gcc/testsuite/gcc.target/i386/builtin_target.c      (revision 0)
>>> @@ -0,0 +1,61 @@
>>> +/* This test checks if the __builtin_cpu_* calls are recognized. */
>>> +
>>> +/* { dg-do run } */
>>> +
>>> +int
>>> +fn1 ()
>>> +{
>>> +  if (__builtin_cpu_supports_cmov () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_supports_mmx () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_supports_popcount () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_supports_sse () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_supports_sse2 () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_supports_sse3 () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_supports_ssse3 () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_supports_sse4_1 () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_supports_sse4_2 () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_is_amd () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_is_intel () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_is_intel_atom () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_is_intel_core2 () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_is_intel_corei7 () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_is_intel_corei7_nehalem () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_is_intel_corei7_westmere () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_is_intel_corei7_sandybridge () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_is_amdfam10 () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_is_amdfam10_barcelona () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_is_amdfam10_shanghai () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_is_amdfam10_istanbul () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_is_amdfam15_bdver1 () < 0)
>>> +    return -1;
>>> +  if (__builtin_cpu_is_amdfam15_bdver2 () < 0)
>>> +    return -1;
>>> +
>>> +  return 0;
>>> +}
>>> +
>>> +int main ()
>>> +{
>>> +  return fn1 ();
>>> +}
>>>
>>> --
>>> This patch is available for review at http://codereview.appspot.com/5754058

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Xinliang David Li]

On Wed, Mar 7, 2012 at 5:51 AM, Richard Guenther
<richard.guenther@gmail.com> wrote:
> On Wed, Mar 7, 2012 at 1:49 AM, Sriraman Tallam <tmsriram@google.com> wrote:
>> Patch for CPU detection at run-time.
>> ===================================
>>
>> Patch for CPU detection at run-time, to be used in dispatching of
>> multi-versioned functions.   Please see this discussion:
>> http://gcc.gnu.org/ml/gcc-patches/2011-08/msg01355.html
>> when this patch for reviewed the last time.
>>
>> For more detailed description:
>> http://gcc.gnu.org/ml/gcc/2012-03/msg00074.html
>>
>> One of the main concerns was about making CPU detection initialization a
>> constructor. The main point raised was about constructor ordering. I have
>> added a priority value to the CPU detection constructor to make it very high
>> priority so that it is guaranteed to fire before every constructor without
>> an explicitly marked priority value of 101.  However, IFUNC initializers
>> will still fire before this constructor, so the cpu initialization routine
>> has to be explicitly called in such initializers for which I have added a
>> builtin: __builtin_cpu_init ().
>>
>> This patch adds the following new builtins:
>>
>> * __builtin_cpu_init
>> * __builtin_cpu_supports_cmov
>> * __builtin_cpu_supports_mmx
>> * __builtin_cpu_supports_popcount
>> * __builtin_cpu_supports_sse
>> * __builtin_cpu_supports_sse2
>> * __builtin_cpu_supports_sse3
>> * __builtin_cpu_supports_ssse3
>> * __builtin_cpu_supports_sse4_1
>> * __builtin_cpu_supports_sse4_2
>> * __builtin_cpu_is_amd
>> * __builtin_cpu_is_intel_atom
>> * __builtin_cpu_is_intel_core2
>> * __builtin_cpu_is_intel
>> * __builtin_cpu_is_intel_corei7
>> * __builtin_cpu_is_intel_corei7_nehalem
>> * __builtin_cpu_is_intel_corei7_westmere
>> * __builtin_cpu_is_intel_corei7_sandybridge
>> * __builtin_cpu_is_amdfam10
>> * __builtin_cpu_is_amdfam10_barcelona
>> * __builtin_cpu_is_amdfam10_shanghai
>> * __builtin_cpu_is_amdfam10_istanbul
>> * __builtin_cpu_is_amdfam15_bdver1
>> * __builtin_cpu_is_amdfam15_bdver2
>
> I think the non-feature detection functions are not necessary at all.

They are useful if compiler needs to do auto versioning based on cpu model.

> Builtin functions are not exactly cheap, nor is the scheme you invent
> backward/forward compatible.  Instead, why not add a single builtin
> function, __builtin_cpu_supports(const char *), and decode from
> a comma-separated list of features?  Unknown features are simply
> "not present".  So I can write code with only a single configure check,

This is a good idea.

__builtin_is_cpu (const char* );
__builtin_cpu_supports (char char*);

thanks,

David


> for __builtin_cpu_supports, and cater for future features or older compilers.
>
> And of course that builtin would be even cross-platform.
>
> Implementation-wise I'll leave this to x86 maintainers to comment on.
>
> Richard.
>
>>
>>        * config/i386/i386.c (build_struct_with_one_bit_fields): New function.
>>        (make_var_decl): New function.
>>        (get_field_from_struct): New function.
>>        (fold_builtin_target): New function.
>>        (ix86_fold_builtin): New function.
>>        (ix86_expand_builtin): Expand new builtins by folding them.
>>        (make_platform_builtin): New functions.
>>        (ix86_init_platform_type_builtins): Make the new builtins.
>>        (ix86_init_builtins): Make new builtins to detect CPU type.
>>        (TARGET_FOLD_BUILTIN): New macro.
>>        (IX86_BUILTIN_CPU_SUPPORTS_CMOV): New enum value.
>>        (IX86_BUILTIN_CPU_SUPPORTS_MMX): New enum value.
>>        (IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT): New enum value.
>>        (IX86_BUILTIN_CPU_SUPPORTS_SSE): New enum value.
>>        (IX86_BUILTIN_CPU_SUPPORTS_SSE2): New enum value.
>>        (IX86_BUILTIN_CPU_SUPPORTS_SSE3): New enum value.
>>        (IX86_BUILTIN_CPU_SUPPORTS_SSSE3): New enum value.
>>        (IX86_BUILTIN_CPU_SUPPORTS_SSE4_1): New enum value.
>>        (IX86_BUILTIN_CPU_SUPPORTS_SSE4_2): New enum value.
>>        (IX86_BUILTIN_CPU_INIT): New enum value.
>>        (IX86_BUILTIN_CPU_IS_AMD): New enum value.
>>        (IX86_BUILTIN_CPU_IS_INTEL): New enum value.
>>        (IX86_BUILTIN_CPU_IS_INTEL_ATOM): New enum value.
>>        (IX86_BUILTIN_CPU_IS_INTEL_CORE2): New enum value.
>>        (IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM): New enum value.
>>        (IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE): New enum value.
>>        (IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE): New enum value.
>>        (IX86_BUILTIN_CPU_IS_AMDFAM10_BARCELONA): New enum value.
>>        (IX86_BUILTIN_CPU_IS_AMDFAM10_SHANGHAI): New enum value.
>>        (IX86_BUILTIN_CPU_IS_AMDFAM10_ISTANBUL): New enum value.
>>        (IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1): New enum value.
>>        (IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2): New enum value.
>>        * config/i386/i386-builtin-types.def: New function type.
>>        * testsuite/gcc.target/builtin_target.c: New testcase.
>>
>>        * libgcc/config/i386/i386-cpuinfo.c: New file.
>>        * libgcc/config/i386/t-cpuinfo: New file.
>>        * libgcc/config.host: Include t-cpuinfo.
>>        * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
>>        and __cpu_features.
>>
>> Index: libgcc/config.host
>> ===================================================================
>> --- libgcc/config.host  (revision 184971)
>> +++ libgcc/config.host  (working copy)
>> @@ -1142,7 +1142,7 @@ i[34567]86-*-linux* | x86_64-*-linux* | \
>>   i[34567]86-*-kfreebsd*-gnu | x86_64-*-kfreebsd*-gnu | \
>>   i[34567]86-*-knetbsd*-gnu | \
>>   i[34567]86-*-gnu*)
>> -       tmake_file="${tmake_file} t-tls i386/t-linux"
>> +       tmake_file="${tmake_file} t-tls i386/t-linux i386/t-cpuinfo"
>>        if test "$libgcc_cv_cfi" = "yes"; then
>>                tmake_file="${tmake_file} t-stack i386/t-stack-i386"
>>        fi
>> Index: libgcc/config/i386/t-cpuinfo
>> ===================================================================
>> --- libgcc/config/i386/t-cpuinfo        (revision 0)
>> +++ libgcc/config/i386/t-cpuinfo        (revision 0)
>> @@ -0,0 +1 @@
>> +LIB2ADD += $(srcdir)/config/i386/i386-cpuinfo.c
>> Index: libgcc/config/i386/i386-cpuinfo.c
>> ===================================================================
>> --- libgcc/config/i386/i386-cpuinfo.c   (revision 0)
>> +++ libgcc/config/i386/i386-cpuinfo.c   (revision 0)
>> @@ -0,0 +1,306 @@
>> +/* Get CPU type and Features for x86 processors.
>> +   Copyright (C) 2011 Free Software Foundation, Inc.
>> +   Contributed by Sriraman Tallam (tmsriram@google.com)
>> +
>> +This file is part of GCC.
>> +
>> +GCC is free software; you can redistribute it and/or modify it under
>> +the terms of the GNU General Public License as published by the Free
>> +Software Foundation; either version 3, or (at your option) any later
>> +version.
>> +
>> +GCC is distributed in the hope that it will be useful, but WITHOUT ANY
>> +WARRANTY; without even the implied warranty of MERCHANTABILITY or
>> +FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
>> +for more details.
>> +
>> +You should have received a copy of the GNU General Public License
>> +along with GCC; see the file COPYING3.  If not see
>> +<http://www.gnu.org/licenses/>. */
>> +
>> +#include "cpuid.h"
>> +#include "tsystem.h"
>> +
>> +int __cpu_indicator_init (void) __attribute__ ((constructor (101)));
>> +
>> +enum vendor_signatures
>> +{
>> +  SIG_INTEL =  0x756e6547 /* Genu */,
>> +  SIG_AMD =    0x68747541 /* Auth */
>> +};
>> +
>> +/* ISA Features supported. */
>> +
>> +struct __processor_features
>> +{
>> +  unsigned int __cpu_cmov : 1;
>> +  unsigned int __cpu_mmx : 1;
>> +  unsigned int __cpu_popcnt : 1;
>> +  unsigned int __cpu_sse : 1;
>> +  unsigned int __cpu_sse2 : 1;
>> +  unsigned int __cpu_sse3 : 1;
>> +  unsigned int __cpu_ssse3 : 1;
>> +  unsigned int __cpu_sse4_1 : 1;
>> +  unsigned int __cpu_sse4_2 : 1;
>> +} __cpu_features;
>> +
>> +/* Processor Model. */
>> +
>> +struct __processor_model
>> +{
>> +  /* Vendor. */
>> +  unsigned int __cpu_is_amd : 1;
>> +  unsigned int __cpu_is_intel : 1;
>> +  /* CPU type. */
>> +  unsigned int __cpu_is_intel_atom : 1;
>> +  unsigned int __cpu_is_intel_core2 : 1;
>> +  unsigned int __cpu_is_intel_corei7 : 1;
>> +  unsigned int __cpu_is_intel_corei7_nehalem : 1;
>> +  unsigned int __cpu_is_intel_corei7_westmere : 1;
>> +  unsigned int __cpu_is_intel_corei7_sandybridge : 1;
>> +  unsigned int __cpu_is_amdfam10h : 1;
>> +  unsigned int __cpu_is_amdfam10h_barcelona : 1;
>> +  unsigned int __cpu_is_amdfam10h_shanghai : 1;
>> +  unsigned int __cpu_is_amdfam10h_istanbul : 1;
>> +  unsigned int __cpu_is_amdfam15h_bdver1 : 1;
>> +  unsigned int __cpu_is_amdfam15h_bdver2 : 1;
>> +} __cpu_model;
>> +
>> +/* Get the specific type of AMD CPU.  */
>> +
>> +static void
>> +get_amd_cpu (unsigned int family, unsigned int model)
>> +{
>> +  switch (family)
>> +    {
>> +    /* AMD Family 10h.  */
>> +    case 0x10:
>> +      switch (model)
>> +       {
>> +       case 0x2:
>> +         /* Barcelona.  */
>> +         __cpu_model.__cpu_is_amdfam10h = 1;
>> +         __cpu_model.__cpu_is_amdfam10h_barcelona = 1;
>> +         break;
>> +       case 0x4:
>> +         /* Shanghai.  */
>> +         __cpu_model.__cpu_is_amdfam10h = 1;
>> +         __cpu_model.__cpu_is_amdfam10h_shanghai = 1;
>> +         break;
>> +       case 0x8:
>> +         /* Istanbul.  */
>> +         __cpu_model.__cpu_is_amdfam10h = 1;
>> +         __cpu_model.__cpu_is_amdfam10h_istanbul = 1;
>> +         break;
>> +       default:
>> +         break;
>> +       }
>> +      break;
>> +    /* AMD Family 15h.  */
>> +    case 0x15:
>> +      /* Bulldozer version 1.  */
>> +      if (model >= 0 && model <= 0xf)
>> +        __cpu_model.__cpu_is_amdfam15h_bdver1 = 1;
>> +      /* Bulldozer version 2.  */
>> +      if (model >= 0x10 && model <= 0x1f)
>> +        __cpu_model.__cpu_is_amdfam15h_bdver2 = 1;
>> +      break;
>> +    default:
>> +      break;
>> +    }
>> +}
>> +
>> +/* Get the specific type of Intel CPU.  */
>> +
>> +static void
>> +get_intel_cpu (unsigned int family, unsigned int model, unsigned int brand_id)
>> +{
>> +  /* Parse family and model only if brand ID is 0. */
>> +  if (brand_id == 0)
>> +    {
>> +      switch (family)
>> +       {
>> +       case 0x5:
>> +         /* Pentium.  */
>> +         break;
>> +       case 0x6:
>> +         switch (model)
>> +           {
>> +           case 0x1c:
>> +           case 0x26:
>> +             /* Atom.  */
>> +             __cpu_model.__cpu_is_intel_atom = 1;
>> +             break;
>> +           case 0x1a:
>> +           case 0x1e:
>> +           case 0x1f:
>> +           case 0x2e:
>> +             /* Nehalem.  */
>> +             __cpu_model.__cpu_is_intel_corei7 = 1;
>> +             __cpu_model.__cpu_is_intel_corei7_nehalem = 1;
>> +             break;
>> +           case 0x25:
>> +           case 0x2c:
>> +           case 0x2f:
>> +             /* Westmere.  */
>> +             __cpu_model.__cpu_is_intel_corei7 = 1;
>> +             __cpu_model.__cpu_is_intel_corei7_westmere = 1;
>> +             break;
>> +           case 0x2a:
>> +             /* Sandy Bridge.  */
>> +             __cpu_model.__cpu_is_intel_corei7 = 1;
>> +             __cpu_model.__cpu_is_intel_corei7_sandybridge = 1;
>> +             break;
>> +           case 0x17:
>> +           case 0x1d:
>> +             /* Penryn.  */
>> +           case 0x0f:
>> +             /* Merom.  */
>> +             __cpu_model.__cpu_is_intel_core2 = 1;
>> +             break;
>> +           default:
>> +             break;
>> +           }
>> +         break;
>> +       default:
>> +         /* We have no idea.  */
>> +         break;
>> +       }
>> +    }
>> +}
>> +
>> +static void
>> +get_available_features (unsigned int ecx, unsigned int edx)
>> +{
>> +  __cpu_features.__cpu_cmov = (edx & bit_CMOV) ? 1 : 0;
>> +  __cpu_features.__cpu_mmx = (edx & bit_MMX) ? 1 : 0;
>> +  __cpu_features.__cpu_sse = (edx & bit_SSE) ? 1 : 0;
>> +  __cpu_features.__cpu_sse2 = (edx & bit_SSE2) ? 1 : 0;
>> +  __cpu_features.__cpu_popcnt = (ecx & bit_POPCNT) ? 1 : 0;
>> +  __cpu_features.__cpu_sse3 = (ecx & bit_SSE3) ? 1 : 0;
>> +  __cpu_features.__cpu_ssse3 = (ecx & bit_SSSE3) ? 1 : 0;
>> +  __cpu_features.__cpu_sse4_1 = (ecx & bit_SSE4_1) ? 1 : 0;
>> +  __cpu_features.__cpu_sse4_2 = (ecx & bit_SSE4_2) ? 1 : 0;
>> +}
>> +
>> +
>> +/* Sanity check for the vendor and cpu type flags.  */
>> +
>> +static int
>> +sanity_check (void)
>> +{
>> +  unsigned int one_type = 0;
>> +
>> +  /* Vendor cannot be Intel and AMD. */
>> +  gcc_assert((__cpu_model.__cpu_is_intel == 0)
>> +             || (__cpu_model.__cpu_is_amd == 0));
>> +
>> +  /* Only one CPU type can be set. */
>> +  one_type = (__cpu_model.__cpu_is_intel_atom
>> +             + __cpu_model.__cpu_is_intel_core2
>> +             + __cpu_model.__cpu_is_intel_corei7_nehalem
>> +             + __cpu_model.__cpu_is_intel_corei7_westmere
>> +             + __cpu_model.__cpu_is_intel_corei7_sandybridge
>> +             + __cpu_model.__cpu_is_amdfam10h_barcelona
>> +             + __cpu_model.__cpu_is_amdfam10h_shanghai
>> +             + __cpu_model.__cpu_is_amdfam10h_istanbul
>> +             + __cpu_model.__cpu_is_amdfam15h_bdver1
>> +             + __cpu_model.__cpu_is_amdfam15h_bdver2);
>> +
>> +  gcc_assert (one_type <= 1);
>> +  return 0;
>> +}
>> +
>> +/* A noinline function calling __get_cpuid. Having many calls to
>> +   cpuid in one function in 32-bit mode causes GCC to complain:
>> +   "can’t find a register in class ‘CLOBBERED_REGS’".  This is
>> +   related to PR rtl-optimization 44174. */
>> +
>> +static int __attribute__ ((noinline))
>> +__get_cpuid_output (unsigned int __level,
>> +                   unsigned int *__eax, unsigned int *__ebx,
>> +                   unsigned int *__ecx, unsigned int *__edx)
>> +{
>> +  return __get_cpuid (__level, __eax, __ebx, __ecx, __edx);
>> +}
>> +
>> +
>> +/* A constructor function that is sets __cpu_model and __cpu_features with
>> +   the right values.  This needs to run only once.  This constructor is
>> +   given the highest priority and it should run before constructors without
>> +   the priority set.  However, it still runs after ifunc initializers and
>> +   needs to be called explicitly there.  */
>> +
>> +int __attribute__ ((constructor (101)))
>> +__cpu_indicator_init (void)
>> +{
>> +  unsigned int eax, ebx, ecx, edx;
>> +
>> +  int max_level = 5;
>> +  unsigned int vendor;
>> +  unsigned int model, family, brand_id;
>> +  unsigned int extended_model, extended_family;
>> +  static int called = 0;
>> +
>> +  /* This function needs to run just once.  */
>> +  if (called)
>> +    return 0;
>> +  else
>> +    called = 1;
>> +
>> +  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
>> +  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
>> +    return -1;
>> +
>> +  vendor = ebx;
>> +  max_level = eax;
>> +
>> +  if (max_level < 1)
>> +    return -1;
>> +
>> +  if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
>> +    return -1;
>> +
>> +  model = (eax >> 4) & 0x0f;
>> +  family = (eax >> 8) & 0x0f;
>> +  brand_id = ebx & 0xff;
>> +  extended_model = (eax >> 12) & 0xf0;
>> +  extended_family = (eax >> 20) & 0xff;
>> +
>> +  if (vendor == SIG_INTEL)
>> +    {
>> +      /* Adjust model and family for Intel CPUS. */
>> +      if (family == 0x0f)
>> +       {
>> +         family += extended_family;
>> +         model += extended_model;
>> +       }
>> +      else if (family == 0x06)
>> +       model += extended_model;
>> +
>> +      /* Get CPU type.  */
>> +      __cpu_model.__cpu_is_intel = 1;
>> +      get_intel_cpu (family, model, brand_id);
>> +    }
>> +
>> +  if (vendor == SIG_AMD)
>> +    {
>> +      /* Adjust model and family for AMD CPUS. */
>> +      if (family == 0x0f)
>> +       {
>> +         family += extended_family;
>> +         model += (extended_model << 4);
>> +       }
>> +
>> +      /* Get CPU type.  */
>> +      __cpu_model.__cpu_is_amd = 1;
>> +      get_amd_cpu (family, model);
>> +    }
>> +
>> +  /* Find available features. */
>> +  get_available_features (ecx, edx);
>> +
>> +  sanity_check ();
>> +
>> +  return 0;
>> +}
>> Index: libgcc/config/i386/libgcc-glibc.ver
>> ===================================================================
>> --- libgcc/config/i386/libgcc-glibc.ver (revision 184971)
>> +++ libgcc/config/i386/libgcc-glibc.ver (working copy)
>> @@ -147,6 +147,11 @@ GCC_4.3.0 {
>>   __trunctfxf2
>>   __unordtf2
>>  }
>> +
>> +GCC_4.8.0 {
>> +  __cpu_model
>> +  __cpu_features
>> +}
>>  %else
>>  GCC_4.4.0 {
>>   __addtf3
>> @@ -183,4 +188,9 @@ GCC_4.4.0 {
>>  GCC_4.5.0 {
>>   __extendxftf2
>>  }
>> +
>> +GCC_4.8.0 {
>> +  __cpu_model
>> +  __cpu_features
>> +}
>>  %endif
>> Index: gcc/config/i386/i386-builtin-types.def
>> ===================================================================
>> --- gcc/config/i386/i386-builtin-types.def      (revision 184971)
>> +++ gcc/config/i386/i386-builtin-types.def      (working copy)
>> @@ -143,6 +143,7 @@ DEF_FUNCTION_TYPE (UINT64)
>>  DEF_FUNCTION_TYPE (UNSIGNED)
>>  DEF_FUNCTION_TYPE (VOID)
>>  DEF_FUNCTION_TYPE (PVOID)
>> +DEF_FUNCTION_TYPE (INT)
>>
>>  DEF_FUNCTION_TYPE (FLOAT, FLOAT)
>>  DEF_FUNCTION_TYPE (FLOAT128, FLOAT128)
>> Index: gcc/config/i386/i386.c
>> ===================================================================
>> --- gcc/config/i386/i386.c      (revision 184971)
>> +++ gcc/config/i386/i386.c      (working copy)
>> @@ -25637,6 +25637,33 @@ enum ix86_builtins
>>   /* CFString built-in for darwin */
>>   IX86_BUILTIN_CFSTRING,
>>
>> +  /* Builtins to get CPU features. */
>> +  IX86_BUILTIN_CPU_SUPPORTS_CMOV,
>> +  IX86_BUILTIN_CPU_SUPPORTS_MMX,
>> +  IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT,
>> +  IX86_BUILTIN_CPU_SUPPORTS_SSE,
>> +  IX86_BUILTIN_CPU_SUPPORTS_SSE2,
>> +  IX86_BUILTIN_CPU_SUPPORTS_SSE3,
>> +  IX86_BUILTIN_CPU_SUPPORTS_SSSE3,
>> +  IX86_BUILTIN_CPU_SUPPORTS_SSE4_1,
>> +  IX86_BUILTIN_CPU_SUPPORTS_SSE4_2,
>> +  /* Builtins to get CPU type. */
>> +  IX86_BUILTIN_CPU_INIT,
>> +  IX86_BUILTIN_CPU_IS_AMD,
>> +  IX86_BUILTIN_CPU_IS_INTEL,
>> +  IX86_BUILTIN_CPU_IS_INTEL_ATOM,
>> +  IX86_BUILTIN_CPU_IS_INTEL_CORE2,
>> +  IX86_BUILTIN_CPU_IS_INTEL_COREI7,
>> +  IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM,
>> +  IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE,
>> +  IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE,
>> +  IX86_BUILTIN_CPU_IS_AMDFAM10H,
>> +  IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA,
>> +  IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI,
>> +  IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL,
>> +  IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1,
>> +  IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2,
>> +
>>   IX86_BUILTIN_MAX
>>  };
>>
>> @@ -27446,6 +27473,593 @@ ix86_init_mmx_sse_builtins (void)
>>     }
>>  }
>>
>> +/* Returns a struct type with name NAME and number of fields equal to
>> +   NUM_FIELDS.  Each field is a unsigned int bit field of length 1 bit. */
>> +
>> +static tree
>> +build_struct_with_one_bit_fields (int num_fields, const char *name)
>> +{
>> +  int i;
>> +  char field_name [10];
>> +  tree field = NULL_TREE, field_chain = NULL_TREE;
>> +  tree type = make_node (RECORD_TYPE);
>> +
>> +  strcpy (field_name, "k_field");
>> +
>> +  for (i = 0; i < num_fields; i++)
>> +    {
>> +      /* Name the fields, 0_field, 1_field, ... */
>> +      field_name [0] = '0' + i;
>> +      field = build_decl (UNKNOWN_LOCATION, FIELD_DECL,
>> +                         get_identifier (field_name), unsigned_type_node);
>> +      DECL_BIT_FIELD (field) = 1;
>> +      DECL_SIZE (field) = bitsize_one_node;
>> +      if (field_chain != NULL_TREE)
>> +       DECL_CHAIN (field) = field_chain;
>> +      field_chain = field;
>> +    }
>> +  finish_builtin_struct (type, name, field_chain, NULL_TREE);
>> +  return type;
>> +}
>> +
>> +/* Returns a extern, comdat VAR_DECL of type TYPE and name NAME. */
>> +
>> +static tree
>> +make_var_decl (tree type, const char *name)
>> +{
>> +  tree new_decl;
>> +  struct varpool_node *vnode;
>> +
>> +  new_decl = build_decl (UNKNOWN_LOCATION,
>> +                        VAR_DECL,
>> +                        get_identifier(name),
>> +                        type);
>> +
>> +  DECL_EXTERNAL (new_decl) = 1;
>> +  TREE_STATIC (new_decl) = 1;
>> +  TREE_PUBLIC (new_decl) = 1;
>> +  DECL_INITIAL (new_decl) = 0;
>> +  DECL_ARTIFICIAL (new_decl) = 0;
>> +  DECL_PRESERVE_P (new_decl) = 1;
>> +
>> +  make_decl_one_only (new_decl, DECL_ASSEMBLER_NAME (new_decl));
>> +  assemble_variable (new_decl, 0, 0, 0);
>> +
>> +  vnode = varpool_node (new_decl);
>> +  gcc_assert (vnode != NULL);
>> +  /* Set finalized to 1, otherwise it asserts in function "write_symbol" in
>> +     lto-streamer-out.c. */
>> +  vnode->finalized = 1;
>> +
>> +  return new_decl;
>> +}
>> +
>> +/* Traverses the chain of fields in STRUCT_TYPE and returns the FIELD_NUM
>> +   numbered field. */
>> +
>> +static tree
>> +get_field_from_struct (tree struct_type, int field_num)
>> +{
>> +  int i;
>> +  tree field = TYPE_FIELDS (struct_type);
>> +
>> +  for (i = 0; i < field_num; i++, field = DECL_CHAIN(field))
>> +    {
>> +      gcc_assert (field != NULL_TREE);
>> +    }
>> +
>> +  return field;
>> +}
>> +
>> +/* FNDECL is a __builtin_cpu_* call that is folded into an integer defined
>> +   in libgcc/config/i386/i386-cpuinfo.c */
>> +
>> +static tree
>> +fold_builtin_cpu (enum ix86_builtins fn_code)
>> +{
>> +  /* This is the order of bit-fields in __processor_features in
>> +     i386-cpuinfo.c */
>> +  enum processor_features
>> +  {
>> +    F_CMOV = 0,
>> +    F_MMX,
>> +    F_POPCNT,
>> +    F_SSE,
>> +    F_SSE2,
>> +    F_SSE3,
>> +    F_SSSE3,
>> +    F_SSE4_1,
>> +    F_SSE4_2,
>> +    F_MAX
>> +  };
>> +
>> +  /* This is the order of bit-fields in __processor_model in
>> +     i386-cpuinfo.c */
>> +  enum processor_model
>> +  {
>> +    M_AMD = 0,
>> +    M_INTEL,
>> +    M_INTEL_ATOM,
>> +    M_INTEL_CORE2,
>> +    M_INTEL_COREI7,
>> +    M_INTEL_COREI7_NEHALEM,
>> +    M_INTEL_COREI7_WESTMERE,
>> +    M_INTEL_COREI7_SANDYBRIDGE,
>> +    M_AMDFAM10H,
>> +    M_AMDFAM10H_BARCELONA,
>> +    M_AMDFAM10H_SHANGHAI,
>> +    M_AMDFAM10H_ISTANBUL,
>> +    M_AMDFAM15H_BDVER1,
>> +    M_AMDFAM15H_BDVER2,
>> +    M_MAX
>> +  };
>> +
>> +  static tree __processor_features_type = NULL_TREE;
>> +  static tree __cpu_features_var = NULL_TREE;
>> +  static tree __processor_model_type = NULL_TREE;
>> +  static tree __cpu_model_var = NULL_TREE;
>> +  static tree field;
>> +  static tree which_struct;
>> +
>> +  if (__processor_features_type == NULL_TREE)
>> +    __processor_features_type = build_struct_with_one_bit_fields (F_MAX,
>> +                                 "__processor_features");
>> +
>> +  if (__processor_model_type == NULL_TREE)
>> +    __processor_model_type = build_struct_with_one_bit_fields (M_MAX,
>> +                                 "__processor_model");
>> +
>> +  if (__cpu_features_var == NULL_TREE)
>> +    __cpu_features_var = make_var_decl (__processor_features_type,
>> +                                       "__cpu_features");
>> +
>> +  if (__cpu_model_var == NULL_TREE)
>> +    __cpu_model_var = make_var_decl (__processor_model_type,
>> +                                    "__cpu_model");
>> +
>> +  /* Look at the code to identify the field requested. */
>> +  switch (fn_code)
>> +    {
>> +    case IX86_BUILTIN_CPU_SUPPORTS_CMOV:
>> +      field = get_field_from_struct (__processor_features_type, F_CMOV);
>> +      which_struct = __cpu_features_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_SUPPORTS_MMX:
>> +      field = get_field_from_struct (__processor_features_type, F_MMX);
>> +      which_struct = __cpu_features_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT:
>> +      field = get_field_from_struct (__processor_features_type, F_POPCNT);
>> +      which_struct = __cpu_features_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE:
>> +      field = get_field_from_struct (__processor_features_type, F_SSE);
>> +      which_struct = __cpu_features_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE2:
>> +      field = get_field_from_struct (__processor_features_type, F_SSE2);
>> +      which_struct = __cpu_features_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE3:
>> +      field = get_field_from_struct (__processor_features_type, F_SSE3);
>> +      which_struct = __cpu_features_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSSE3:
>> +      field = get_field_from_struct (__processor_features_type, F_SSSE3);
>> +      which_struct = __cpu_features_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_1:
>> +      field = get_field_from_struct (__processor_features_type, F_SSE4_1);
>> +      which_struct = __cpu_features_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_2:
>> +      field = get_field_from_struct (__processor_features_type, F_SSE4_2);
>> +      which_struct = __cpu_features_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_IS_AMD:
>> +      field = get_field_from_struct (__processor_model_type, M_AMD);
>> +      which_struct = __cpu_model_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_IS_INTEL:
>> +      field = get_field_from_struct (__processor_model_type, M_INTEL);
>> +      which_struct = __cpu_model_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_IS_INTEL_ATOM:
>> +      field = get_field_from_struct (__processor_model_type, M_INTEL_ATOM);
>> +      which_struct = __cpu_model_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_IS_INTEL_CORE2:
>> +      field = get_field_from_struct (__processor_model_type, M_INTEL_CORE2);
>> +      which_struct = __cpu_model_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7:
>> +      field = get_field_from_struct (__processor_model_type,
>> +                                    M_INTEL_COREI7);
>> +      which_struct = __cpu_model_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM:
>> +      field = get_field_from_struct (__processor_model_type,
>> +                                    M_INTEL_COREI7_NEHALEM);
>> +      which_struct = __cpu_model_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE:
>> +      field = get_field_from_struct (__processor_model_type,
>> +                                    M_INTEL_COREI7_WESTMERE);
>> +      which_struct = __cpu_model_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE:
>> +      field = get_field_from_struct (__processor_model_type,
>> +                                    M_INTEL_COREI7_SANDYBRIDGE);
>> +      which_struct = __cpu_model_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H:
>> +      field = get_field_from_struct (__processor_model_type,
>> +                                    M_AMDFAM10H);
>> +      which_struct = __cpu_model_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA:
>> +      field = get_field_from_struct (__processor_model_type,
>> +                                    M_AMDFAM10H_BARCELONA);
>> +      which_struct = __cpu_model_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI:
>> +      field = get_field_from_struct (__processor_model_type,
>> +                                    M_AMDFAM10H_SHANGHAI);
>> +      which_struct = __cpu_model_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL:
>> +      field = get_field_from_struct (__processor_model_type,
>> +                                    M_AMDFAM10H_ISTANBUL);
>> +      which_struct = __cpu_model_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1:
>> +      field = get_field_from_struct (__processor_model_type,
>> +                                    M_AMDFAM15H_BDVER1);
>> +      which_struct = __cpu_model_var;
>> +      break;
>> +    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2:
>> +      field = get_field_from_struct (__processor_model_type,
>> +                                    M_AMDFAM15H_BDVER2);
>> +      which_struct = __cpu_model_var;
>> +      break;
>> +    default:
>> +      return NULL_TREE;
>> +    }
>> +
>> +  return build3 (COMPONENT_REF, TREE_TYPE (field), which_struct, field, NULL_TREE);
>> +}
>> +
>> +static tree
>> +ix86_fold_builtin (tree fndecl, int n_args ATTRIBUTE_UNUSED,
>> +                  tree *args ATTRIBUTE_UNUSED, bool ignore ATTRIBUTE_UNUSED)
>> +{
>> +  const char* decl_name = IDENTIFIER_POINTER (DECL_NAME (fndecl));
>> +  if (DECL_BUILT_IN_CLASS (fndecl) == BUILT_IN_MD
>> +      && strstr(decl_name, "__builtin_cpu") != NULL)
>> +    {
>> +      enum ix86_builtins code = (enum ix86_builtins)
>> +                               DECL_FUNCTION_CODE (fndecl);
>> +      return fold_builtin_cpu (code);
>> +    }
>> +  return NULL_TREE;
>> +}
>> +
>> +/* A builtin to init/return the cpu type or feature.  Returns an
>> +   integer and the type is a const if IS_CONST is set. */
>> +
>> +static void
>> +make_platform_builtin (const char* name, int code, int is_const)
>> +{
>> +  tree decl;
>> +  tree type;
>> +
>> +  type = ix86_get_builtin_func_type (INT_FTYPE_VOID);
>> +  decl = add_builtin_function (name, type, code, BUILT_IN_MD,
>> +                              NULL, NULL_TREE);
>> +  gcc_assert (decl != NULL_TREE);
>> +  ix86_builtins[(int) code] = decl;
>> +  if (is_const)
>> +    TREE_READONLY (decl) = 1;
>> +}
>> +
>> +/* Builtins to get CPU type and features supported. */
>> +
>> +static void
>> +ix86_init_platform_type_builtins (void)
>> +{
>> +  make_platform_builtin ("__builtin_cpu_init",
>> +                        IX86_BUILTIN_CPU_INIT, 0);
>> +  make_platform_builtin ("__builtin_cpu_supports_cmov",
>> +                        IX86_BUILTIN_CPU_SUPPORTS_CMOV, 1);
>> +  make_platform_builtin ("__builtin_cpu_supports_mmx",
>> +                        IX86_BUILTIN_CPU_SUPPORTS_MMX, 1);
>> +  make_platform_builtin ("__builtin_cpu_supports_popcount",
>> +                        IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT, 1);
>> +  make_platform_builtin ("__builtin_cpu_supports_sse",
>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE, 1);
>> +  make_platform_builtin ("__builtin_cpu_supports_sse2",
>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE2, 1);
>> +  make_platform_builtin ("__builtin_cpu_supports_sse3",
>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE3, 1);
>> +  make_platform_builtin ("__builtin_cpu_supports_ssse3",
>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSSE3, 1);
>> +  make_platform_builtin ("__builtin_cpu_supports_sse4_1",
>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE4_1, 1);
>> +  make_platform_builtin ("__builtin_cpu_supports_sse4_2",
>> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE4_2, 1);
>> +  make_platform_builtin ("__builtin_cpu_is_amd",
>> +                        IX86_BUILTIN_CPU_IS_AMD, 1);
>> +  make_platform_builtin ("__builtin_cpu_is_intel_atom",
>> +                        IX86_BUILTIN_CPU_IS_INTEL_ATOM, 1);
>> +  make_platform_builtin ("__builtin_cpu_is_intel_core2",
>> +                        IX86_BUILTIN_CPU_IS_INTEL_CORE2, 1);
>> +  make_platform_builtin ("__builtin_cpu_is_intel",
>> +                        IX86_BUILTIN_CPU_IS_INTEL, 1);
>> +  make_platform_builtin ("__builtin_cpu_is_intel_corei7",
>> +                        IX86_BUILTIN_CPU_IS_INTEL_COREI7, 1);
>> +  make_platform_builtin ("__builtin_cpu_is_intel_corei7_nehalem",
>> +                        IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM, 1);
>> +  make_platform_builtin ("__builtin_cpu_is_intel_corei7_westmere",
>> +                        IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE, 1);
>> +  make_platform_builtin ("__builtin_cpu_is_intel_corei7_sandybridge",
>> +                        IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE, 1);
>> +  make_platform_builtin ("__builtin_cpu_is_amdfam10",
>> +                        IX86_BUILTIN_CPU_IS_AMDFAM10H, 1);
>> +  make_platform_builtin ("__builtin_cpu_is_amdfam10_barcelona",
>> +                        IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA, 1);
>> +  make_platform_builtin ("__builtin_cpu_is_amdfam10_shanghai",
>> +                        IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI, 1);
>> +  make_platform_builtin ("__builtin_cpu_is_amdfam10_istanbul",
>> +                        IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL, 1);
>> +  make_platform_builtin ("__builtin_cpu_is_amdfam15_bdver1",
>> +                        IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1, 1);
>> +  make_platform_builtin ("__builtin_cpu_is_amdfam15_bdver2",
>> +                        IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2, 1);
>> +}
>> +
>>  /* Internal method for ix86_init_builtins.  */
>>
>>  static void
>> @@ -27529,6 +28143,9 @@ ix86_init_builtins (void)
>>
>>   ix86_init_builtin_types ();
>>
>> +  /* Builtins to get CPU type and features. */
>> +  ix86_init_platform_type_builtins ();
>> +
>>   /* TFmode support builtins.  */
>>   def_builtin_const (0, "__builtin_infq",
>>                     FLOAT128_FTYPE_VOID, IX86_BUILTIN_INFQ);
>> @@ -29145,6 +29762,48 @@ ix86_expand_builtin (tree exp, rtx target, rtx sub
>>   enum machine_mode mode0, mode1, mode2, mode3, mode4;
>>   unsigned int fcode = DECL_FUNCTION_CODE (fndecl);
>>
>> +  /* For CPU builtins that can be folded, fold first and expand the fold.  */
>> +  switch (fcode)
>> +    {
>> +    case IX86_BUILTIN_CPU_SUPPORTS_CMOV:
>> +    case IX86_BUILTIN_CPU_SUPPORTS_MMX:
>> +    case IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT:
>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE:
>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE2:
>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE3:
>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSSE3:
>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_1:
>> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_2:
>> +    case IX86_BUILTIN_CPU_IS_AMD:
>> +    case IX86_BUILTIN_CPU_IS_INTEL:
>> +    case IX86_BUILTIN_CPU_IS_INTEL_ATOM:
>> +    case IX86_BUILTIN_CPU_IS_INTEL_CORE2:
>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7:
>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM:
>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE:
>> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE:
>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H:
>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA:
>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI:
>> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL:
>> +    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1:
>> +    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2:
>> +      {
>> +        tree fold_expr = fold_builtin_cpu ((enum ix86_builtins) fcode);
>> +       gcc_assert (fold_expr != NULL_TREE);
>> +        return expand_expr (fold_expr, target, mode, EXPAND_NORMAL);
>> +      }
>> +    case IX86_BUILTIN_CPU_INIT:
>> +      {
>> +       /* Make it call __cpu_indicator_init in libgcc. */
>> +       tree call_expr, fndecl, type;
>> +        type = build_function_type_list (integer_type_node, NULL_TREE);
>> +       fndecl = build_fn_decl ("__cpu_indicator_init", type);
>> +       call_expr = build_call_expr (fndecl, 0);
>> +       return expand_expr (call_expr, target, mode, EXPAND_NORMAL);
>> +      }
>> +    }
>> +
>>   /* Determine whether the builtin function is available under the current ISA.
>>      Originally the builtin was not created if it wasn't applicable to the
>>      current ISA based on the command line switches.  With function specific
>> @@ -38610,6 +39269,12 @@ ix86_autovectorize_vector_sizes (void)
>>  #undef TARGET_BUILD_BUILTIN_VA_LIST
>>  #define TARGET_BUILD_BUILTIN_VA_LIST ix86_build_builtin_va_list
>>
>> +#undef TARGET_FOLD_BUILTIN
>> +#define TARGET_FOLD_BUILTIN ix86_fold_builtin
>> +
>>  #undef TARGET_ENUM_VA_LIST_P
>>  #define TARGET_ENUM_VA_LIST_P ix86_enum_va_list
>> Index: gcc/testsuite/gcc.target/i386/builtin_target.c
>> ===================================================================
>> --- gcc/testsuite/gcc.target/i386/builtin_target.c      (revision 0)
>> +++ gcc/testsuite/gcc.target/i386/builtin_target.c      (revision 0)
>> @@ -0,0 +1,61 @@
>> +/* This test checks if the __builtin_cpu_* calls are recognized. */
>> +
>> +/* { dg-do run } */
>> +
>> +int
>> +fn1 ()
>> +{
>> +  if (__builtin_cpu_supports_cmov () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_supports_mmx () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_supports_popcount () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_supports_sse () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_supports_sse2 () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_supports_sse3 () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_supports_ssse3 () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_supports_sse4_1 () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_supports_sse4_2 () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_is_amd () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_is_intel () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_is_intel_atom () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_is_intel_core2 () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_is_intel_corei7 () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_is_intel_corei7_nehalem () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_is_intel_corei7_westmere () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_is_intel_corei7_sandybridge () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_is_amdfam10 () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_is_amdfam10_barcelona () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_is_amdfam10_shanghai () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_is_amdfam10_istanbul () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_is_amdfam15_bdver1 () < 0)
>> +    return -1;
>> +  if (__builtin_cpu_is_amdfam15_bdver2 () < 0)
>> +    return -1;
>> +
>> +  return 0;
>> +}
>> +
>> +int main ()
>> +{
>> +  return fn1 ();
>> +}
>>
>> --
>> This patch is available for review at http://codereview.appspot.com/5754058

Re: Support for Runtime CPU type detection via builtins (issue5754058)

[Richard Guenther]

On Wed, Mar 7, 2012 at 1:49 AM, Sriraman Tallam <tmsriram@google.com> wrote:
> Patch for CPU detection at run-time.
> ===================================
>
> Patch for CPU detection at run-time, to be used in dispatching of
> multi-versioned functions.   Please see this discussion:
> http://gcc.gnu.org/ml/gcc-patches/2011-08/msg01355.html
> when this patch for reviewed the last time.
>
> For more detailed description:
> http://gcc.gnu.org/ml/gcc/2012-03/msg00074.html
>
> One of the main concerns was about making CPU detection initialization a
> constructor. The main point raised was about constructor ordering. I have
> added a priority value to the CPU detection constructor to make it very high
> priority so that it is guaranteed to fire before every constructor without
> an explicitly marked priority value of 101.  However, IFUNC initializers
> will still fire before this constructor, so the cpu initialization routine
> has to be explicitly called in such initializers for which I have added a
> builtin: __builtin_cpu_init ().
>
> This patch adds the following new builtins:
>
> * __builtin_cpu_init
> * __builtin_cpu_supports_cmov
> * __builtin_cpu_supports_mmx
> * __builtin_cpu_supports_popcount
> * __builtin_cpu_supports_sse
> * __builtin_cpu_supports_sse2
> * __builtin_cpu_supports_sse3
> * __builtin_cpu_supports_ssse3
> * __builtin_cpu_supports_sse4_1
> * __builtin_cpu_supports_sse4_2
> * __builtin_cpu_is_amd
> * __builtin_cpu_is_intel_atom
> * __builtin_cpu_is_intel_core2
> * __builtin_cpu_is_intel
> * __builtin_cpu_is_intel_corei7
> * __builtin_cpu_is_intel_corei7_nehalem
> * __builtin_cpu_is_intel_corei7_westmere
> * __builtin_cpu_is_intel_corei7_sandybridge
> * __builtin_cpu_is_amdfam10
> * __builtin_cpu_is_amdfam10_barcelona
> * __builtin_cpu_is_amdfam10_shanghai
> * __builtin_cpu_is_amdfam10_istanbul
> * __builtin_cpu_is_amdfam15_bdver1
> * __builtin_cpu_is_amdfam15_bdver2

I think the non-feature detection functions are not necessary at all.
Builtin functions are not exactly cheap, nor is the scheme you invent
backward/forward compatible.  Instead, why not add a single builtin
function, __builtin_cpu_supports(const char *), and decode from
a comma-separated list of features?  Unknown features are simply
"not present".  So I can write code with only a single configure check,
for __builtin_cpu_supports, and cater for future features or older compilers.

And of course that builtin would be even cross-platform.

Implementation-wise I'll leave this to x86 maintainers to comment on.

Richard.

>
>        * config/i386/i386.c (build_struct_with_one_bit_fields): New function.
>        (make_var_decl): New function.
>        (get_field_from_struct): New function.
>        (fold_builtin_target): New function.
>        (ix86_fold_builtin): New function.
>        (ix86_expand_builtin): Expand new builtins by folding them.
>        (make_platform_builtin): New functions.
>        (ix86_init_platform_type_builtins): Make the new builtins.
>        (ix86_init_builtins): Make new builtins to detect CPU type.
>        (TARGET_FOLD_BUILTIN): New macro.
>        (IX86_BUILTIN_CPU_SUPPORTS_CMOV): New enum value.
>        (IX86_BUILTIN_CPU_SUPPORTS_MMX): New enum value.
>        (IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT): New enum value.
>        (IX86_BUILTIN_CPU_SUPPORTS_SSE): New enum value.
>        (IX86_BUILTIN_CPU_SUPPORTS_SSE2): New enum value.
>        (IX86_BUILTIN_CPU_SUPPORTS_SSE3): New enum value.
>        (IX86_BUILTIN_CPU_SUPPORTS_SSSE3): New enum value.
>        (IX86_BUILTIN_CPU_SUPPORTS_SSE4_1): New enum value.
>        (IX86_BUILTIN_CPU_SUPPORTS_SSE4_2): New enum value.
>        (IX86_BUILTIN_CPU_INIT): New enum value.
>        (IX86_BUILTIN_CPU_IS_AMD): New enum value.
>        (IX86_BUILTIN_CPU_IS_INTEL): New enum value.
>        (IX86_BUILTIN_CPU_IS_INTEL_ATOM): New enum value.
>        (IX86_BUILTIN_CPU_IS_INTEL_CORE2): New enum value.
>        (IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM): New enum value.
>        (IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE): New enum value.
>        (IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE): New enum value.
>        (IX86_BUILTIN_CPU_IS_AMDFAM10_BARCELONA): New enum value.
>        (IX86_BUILTIN_CPU_IS_AMDFAM10_SHANGHAI): New enum value.
>        (IX86_BUILTIN_CPU_IS_AMDFAM10_ISTANBUL): New enum value.
>        (IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1): New enum value.
>        (IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2): New enum value.
>        * config/i386/i386-builtin-types.def: New function type.
>        * testsuite/gcc.target/builtin_target.c: New testcase.
>
>        * libgcc/config/i386/i386-cpuinfo.c: New file.
>        * libgcc/config/i386/t-cpuinfo: New file.
>        * libgcc/config.host: Include t-cpuinfo.
>        * libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
>        and __cpu_features.
>
> Index: libgcc/config.host
> ===================================================================
> --- libgcc/config.host  (revision 184971)
> +++ libgcc/config.host  (working copy)
> @@ -1142,7 +1142,7 @@ i[34567]86-*-linux* | x86_64-*-linux* | \
>   i[34567]86-*-kfreebsd*-gnu | x86_64-*-kfreebsd*-gnu | \
>   i[34567]86-*-knetbsd*-gnu | \
>   i[34567]86-*-gnu*)
> -       tmake_file="${tmake_file} t-tls i386/t-linux"
> +       tmake_file="${tmake_file} t-tls i386/t-linux i386/t-cpuinfo"
>        if test "$libgcc_cv_cfi" = "yes"; then
>                tmake_file="${tmake_file} t-stack i386/t-stack-i386"
>        fi
> Index: libgcc/config/i386/t-cpuinfo
> ===================================================================
> --- libgcc/config/i386/t-cpuinfo        (revision 0)
> +++ libgcc/config/i386/t-cpuinfo        (revision 0)
> @@ -0,0 +1 @@
> +LIB2ADD += $(srcdir)/config/i386/i386-cpuinfo.c
> Index: libgcc/config/i386/i386-cpuinfo.c
> ===================================================================
> --- libgcc/config/i386/i386-cpuinfo.c   (revision 0)
> +++ libgcc/config/i386/i386-cpuinfo.c   (revision 0)
> @@ -0,0 +1,306 @@
> +/* Get CPU type and Features for x86 processors.
> +   Copyright (C) 2011 Free Software Foundation, Inc.
> +   Contributed by Sriraman Tallam (tmsriram@google.com)
> +
> +This file is part of GCC.
> +
> +GCC is free software; you can redistribute it and/or modify it under
> +the terms of the GNU General Public License as published by the Free
> +Software Foundation; either version 3, or (at your option) any later
> +version.
> +
> +GCC is distributed in the hope that it will be useful, but WITHOUT ANY
> +WARRANTY; without even the implied warranty of MERCHANTABILITY or
> +FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
> +for more details.
> +
> +You should have received a copy of the GNU General Public License
> +along with GCC; see the file COPYING3.  If not see
> +<http://www.gnu.org/licenses/>. */
> +
> +#include "cpuid.h"
> +#include "tsystem.h"
> +
> +int __cpu_indicator_init (void) __attribute__ ((constructor (101)));
> +
> +enum vendor_signatures
> +{
> +  SIG_INTEL =  0x756e6547 /* Genu */,
> +  SIG_AMD =    0x68747541 /* Auth */
> +};
> +
> +/* ISA Features supported. */
> +
> +struct __processor_features
> +{
> +  unsigned int __cpu_cmov : 1;
> +  unsigned int __cpu_mmx : 1;
> +  unsigned int __cpu_popcnt : 1;
> +  unsigned int __cpu_sse : 1;
> +  unsigned int __cpu_sse2 : 1;
> +  unsigned int __cpu_sse3 : 1;
> +  unsigned int __cpu_ssse3 : 1;
> +  unsigned int __cpu_sse4_1 : 1;
> +  unsigned int __cpu_sse4_2 : 1;
> +} __cpu_features;
> +
> +/* Processor Model. */
> +
> +struct __processor_model
> +{
> +  /* Vendor. */
> +  unsigned int __cpu_is_amd : 1;
> +  unsigned int __cpu_is_intel : 1;
> +  /* CPU type. */
> +  unsigned int __cpu_is_intel_atom : 1;
> +  unsigned int __cpu_is_intel_core2 : 1;
> +  unsigned int __cpu_is_intel_corei7 : 1;
> +  unsigned int __cpu_is_intel_corei7_nehalem : 1;
> +  unsigned int __cpu_is_intel_corei7_westmere : 1;
> +  unsigned int __cpu_is_intel_corei7_sandybridge : 1;
> +  unsigned int __cpu_is_amdfam10h : 1;
> +  unsigned int __cpu_is_amdfam10h_barcelona : 1;
> +  unsigned int __cpu_is_amdfam10h_shanghai : 1;
> +  unsigned int __cpu_is_amdfam10h_istanbul : 1;
> +  unsigned int __cpu_is_amdfam15h_bdver1 : 1;
> +  unsigned int __cpu_is_amdfam15h_bdver2 : 1;
> +} __cpu_model;
> +
> +/* Get the specific type of AMD CPU.  */
> +
> +static void
> +get_amd_cpu (unsigned int family, unsigned int model)
> +{
> +  switch (family)
> +    {
> +    /* AMD Family 10h.  */
> +    case 0x10:
> +      switch (model)
> +       {
> +       case 0x2:
> +         /* Barcelona.  */
> +         __cpu_model.__cpu_is_amdfam10h = 1;
> +         __cpu_model.__cpu_is_amdfam10h_barcelona = 1;
> +         break;
> +       case 0x4:
> +         /* Shanghai.  */
> +         __cpu_model.__cpu_is_amdfam10h = 1;
> +         __cpu_model.__cpu_is_amdfam10h_shanghai = 1;
> +         break;
> +       case 0x8:
> +         /* Istanbul.  */
> +         __cpu_model.__cpu_is_amdfam10h = 1;
> +         __cpu_model.__cpu_is_amdfam10h_istanbul = 1;
> +         break;
> +       default:
> +         break;
> +       }
> +      break;
> +    /* AMD Family 15h.  */
> +    case 0x15:
> +      /* Bulldozer version 1.  */
> +      if (model >= 0 && model <= 0xf)
> +        __cpu_model.__cpu_is_amdfam15h_bdver1 = 1;
> +      /* Bulldozer version 2.  */
> +      if (model >= 0x10 && model <= 0x1f)
> +        __cpu_model.__cpu_is_amdfam15h_bdver2 = 1;
> +      break;
> +    default:
> +      break;
> +    }
> +}
> +
> +/* Get the specific type of Intel CPU.  */
> +
> +static void
> +get_intel_cpu (unsigned int family, unsigned int model, unsigned int brand_id)
> +{
> +  /* Parse family and model only if brand ID is 0. */
> +  if (brand_id == 0)
> +    {
> +      switch (family)
> +       {
> +       case 0x5:
> +         /* Pentium.  */
> +         break;
> +       case 0x6:
> +         switch (model)
> +           {
> +           case 0x1c:
> +           case 0x26:
> +             /* Atom.  */
> +             __cpu_model.__cpu_is_intel_atom = 1;
> +             break;
> +           case 0x1a:
> +           case 0x1e:
> +           case 0x1f:
> +           case 0x2e:
> +             /* Nehalem.  */
> +             __cpu_model.__cpu_is_intel_corei7 = 1;
> +             __cpu_model.__cpu_is_intel_corei7_nehalem = 1;
> +             break;
> +           case 0x25:
> +           case 0x2c:
> +           case 0x2f:
> +             /* Westmere.  */
> +             __cpu_model.__cpu_is_intel_corei7 = 1;
> +             __cpu_model.__cpu_is_intel_corei7_westmere = 1;
> +             break;
> +           case 0x2a:
> +             /* Sandy Bridge.  */
> +             __cpu_model.__cpu_is_intel_corei7 = 1;
> +             __cpu_model.__cpu_is_intel_corei7_sandybridge = 1;
> +             break;
> +           case 0x17:
> +           case 0x1d:
> +             /* Penryn.  */
> +           case 0x0f:
> +             /* Merom.  */
> +             __cpu_model.__cpu_is_intel_core2 = 1;
> +             break;
> +           default:
> +             break;
> +           }
> +         break;
> +       default:
> +         /* We have no idea.  */
> +         break;
> +       }
> +    }
> +}
> +
> +static void
> +get_available_features (unsigned int ecx, unsigned int edx)
> +{
> +  __cpu_features.__cpu_cmov = (edx & bit_CMOV) ? 1 : 0;
> +  __cpu_features.__cpu_mmx = (edx & bit_MMX) ? 1 : 0;
> +  __cpu_features.__cpu_sse = (edx & bit_SSE) ? 1 : 0;
> +  __cpu_features.__cpu_sse2 = (edx & bit_SSE2) ? 1 : 0;
> +  __cpu_features.__cpu_popcnt = (ecx & bit_POPCNT) ? 1 : 0;
> +  __cpu_features.__cpu_sse3 = (ecx & bit_SSE3) ? 1 : 0;
> +  __cpu_features.__cpu_ssse3 = (ecx & bit_SSSE3) ? 1 : 0;
> +  __cpu_features.__cpu_sse4_1 = (ecx & bit_SSE4_1) ? 1 : 0;
> +  __cpu_features.__cpu_sse4_2 = (ecx & bit_SSE4_2) ? 1 : 0;
> +}
> +
> +
> +/* Sanity check for the vendor and cpu type flags.  */
> +
> +static int
> +sanity_check (void)
> +{
> +  unsigned int one_type = 0;
> +
> +  /* Vendor cannot be Intel and AMD. */
> +  gcc_assert((__cpu_model.__cpu_is_intel == 0)
> +             || (__cpu_model.__cpu_is_amd == 0));
> +
> +  /* Only one CPU type can be set. */
> +  one_type = (__cpu_model.__cpu_is_intel_atom
> +             + __cpu_model.__cpu_is_intel_core2
> +             + __cpu_model.__cpu_is_intel_corei7_nehalem
> +             + __cpu_model.__cpu_is_intel_corei7_westmere
> +             + __cpu_model.__cpu_is_intel_corei7_sandybridge
> +             + __cpu_model.__cpu_is_amdfam10h_barcelona
> +             + __cpu_model.__cpu_is_amdfam10h_shanghai
> +             + __cpu_model.__cpu_is_amdfam10h_istanbul
> +             + __cpu_model.__cpu_is_amdfam15h_bdver1
> +             + __cpu_model.__cpu_is_amdfam15h_bdver2);
> +
> +  gcc_assert (one_type <= 1);
> +  return 0;
> +}
> +
> +/* A noinline function calling __get_cpuid. Having many calls to
> +   cpuid in one function in 32-bit mode causes GCC to complain:
> +   "can’t find a register in class ‘CLOBBERED_REGS’".  This is
> +   related to PR rtl-optimization 44174. */
> +
> +static int __attribute__ ((noinline))
> +__get_cpuid_output (unsigned int __level,
> +                   unsigned int *__eax, unsigned int *__ebx,
> +                   unsigned int *__ecx, unsigned int *__edx)
> +{
> +  return __get_cpuid (__level, __eax, __ebx, __ecx, __edx);
> +}
> +
> +
> +/* A constructor function that is sets __cpu_model and __cpu_features with
> +   the right values.  This needs to run only once.  This constructor is
> +   given the highest priority and it should run before constructors without
> +   the priority set.  However, it still runs after ifunc initializers and
> +   needs to be called explicitly there.  */
> +
> +int __attribute__ ((constructor (101)))
> +__cpu_indicator_init (void)
> +{
> +  unsigned int eax, ebx, ecx, edx;
> +
> +  int max_level = 5;
> +  unsigned int vendor;
> +  unsigned int model, family, brand_id;
> +  unsigned int extended_model, extended_family;
> +  static int called = 0;
> +
> +  /* This function needs to run just once.  */
> +  if (called)
> +    return 0;
> +  else
> +    called = 1;
> +
> +  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
> +  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
> +    return -1;
> +
> +  vendor = ebx;
> +  max_level = eax;
> +
> +  if (max_level < 1)
> +    return -1;
> +
> +  if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
> +    return -1;
> +
> +  model = (eax >> 4) & 0x0f;
> +  family = (eax >> 8) & 0x0f;
> +  brand_id = ebx & 0xff;
> +  extended_model = (eax >> 12) & 0xf0;
> +  extended_family = (eax >> 20) & 0xff;
> +
> +  if (vendor == SIG_INTEL)
> +    {
> +      /* Adjust model and family for Intel CPUS. */
> +      if (family == 0x0f)
> +       {
> +         family += extended_family;
> +         model += extended_model;
> +       }
> +      else if (family == 0x06)
> +       model += extended_model;
> +
> +      /* Get CPU type.  */
> +      __cpu_model.__cpu_is_intel = 1;
> +      get_intel_cpu (family, model, brand_id);
> +    }
> +
> +  if (vendor == SIG_AMD)
> +    {
> +      /* Adjust model and family for AMD CPUS. */
> +      if (family == 0x0f)
> +       {
> +         family += extended_family;
> +         model += (extended_model << 4);
> +       }
> +
> +      /* Get CPU type.  */
> +      __cpu_model.__cpu_is_amd = 1;
> +      get_amd_cpu (family, model);
> +    }
> +
> +  /* Find available features. */
> +  get_available_features (ecx, edx);
> +
> +  sanity_check ();
> +
> +  return 0;
> +}
> Index: libgcc/config/i386/libgcc-glibc.ver
> ===================================================================
> --- libgcc/config/i386/libgcc-glibc.ver (revision 184971)
> +++ libgcc/config/i386/libgcc-glibc.ver (working copy)
> @@ -147,6 +147,11 @@ GCC_4.3.0 {
>   __trunctfxf2
>   __unordtf2
>  }
> +
> +GCC_4.8.0 {
> +  __cpu_model
> +  __cpu_features
> +}
>  %else
>  GCC_4.4.0 {
>   __addtf3
> @@ -183,4 +188,9 @@ GCC_4.4.0 {
>  GCC_4.5.0 {
>   __extendxftf2
>  }
> +
> +GCC_4.8.0 {
> +  __cpu_model
> +  __cpu_features
> +}
>  %endif
> Index: gcc/config/i386/i386-builtin-types.def
> ===================================================================
> --- gcc/config/i386/i386-builtin-types.def      (revision 184971)
> +++ gcc/config/i386/i386-builtin-types.def      (working copy)
> @@ -143,6 +143,7 @@ DEF_FUNCTION_TYPE (UINT64)
>  DEF_FUNCTION_TYPE (UNSIGNED)
>  DEF_FUNCTION_TYPE (VOID)
>  DEF_FUNCTION_TYPE (PVOID)
> +DEF_FUNCTION_TYPE (INT)
>
>  DEF_FUNCTION_TYPE (FLOAT, FLOAT)
>  DEF_FUNCTION_TYPE (FLOAT128, FLOAT128)
> Index: gcc/config/i386/i386.c
> ===================================================================
> --- gcc/config/i386/i386.c      (revision 184971)
> +++ gcc/config/i386/i386.c      (working copy)
> @@ -25637,6 +25637,33 @@ enum ix86_builtins
>   /* CFString built-in for darwin */
>   IX86_BUILTIN_CFSTRING,
>
> +  /* Builtins to get CPU features. */
> +  IX86_BUILTIN_CPU_SUPPORTS_CMOV,
> +  IX86_BUILTIN_CPU_SUPPORTS_MMX,
> +  IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT,
> +  IX86_BUILTIN_CPU_SUPPORTS_SSE,
> +  IX86_BUILTIN_CPU_SUPPORTS_SSE2,
> +  IX86_BUILTIN_CPU_SUPPORTS_SSE3,
> +  IX86_BUILTIN_CPU_SUPPORTS_SSSE3,
> +  IX86_BUILTIN_CPU_SUPPORTS_SSE4_1,
> +  IX86_BUILTIN_CPU_SUPPORTS_SSE4_2,
> +  /* Builtins to get CPU type. */
> +  IX86_BUILTIN_CPU_INIT,
> +  IX86_BUILTIN_CPU_IS_AMD,
> +  IX86_BUILTIN_CPU_IS_INTEL,
> +  IX86_BUILTIN_CPU_IS_INTEL_ATOM,
> +  IX86_BUILTIN_CPU_IS_INTEL_CORE2,
> +  IX86_BUILTIN_CPU_IS_INTEL_COREI7,
> +  IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM,
> +  IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE,
> +  IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE,
> +  IX86_BUILTIN_CPU_IS_AMDFAM10H,
> +  IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA,
> +  IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI,
> +  IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL,
> +  IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1,
> +  IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2,
> +
>   IX86_BUILTIN_MAX
>  };
>
> @@ -27446,6 +27473,593 @@ ix86_init_mmx_sse_builtins (void)
>     }
>  }
>
> +/* Returns a struct type with name NAME and number of fields equal to
> +   NUM_FIELDS.  Each field is a unsigned int bit field of length 1 bit. */
> +
> +static tree
> +build_struct_with_one_bit_fields (int num_fields, const char *name)
> +{
> +  int i;
> +  char field_name [10];
> +  tree field = NULL_TREE, field_chain = NULL_TREE;
> +  tree type = make_node (RECORD_TYPE);
> +
> +  strcpy (field_name, "k_field");
> +
> +  for (i = 0; i < num_fields; i++)
> +    {
> +      /* Name the fields, 0_field, 1_field, ... */
> +      field_name [0] = '0' + i;
> +      field = build_decl (UNKNOWN_LOCATION, FIELD_DECL,
> +                         get_identifier (field_name), unsigned_type_node);
> +      DECL_BIT_FIELD (field) = 1;
> +      DECL_SIZE (field) = bitsize_one_node;
> +      if (field_chain != NULL_TREE)
> +       DECL_CHAIN (field) = field_chain;
> +      field_chain = field;
> +    }
> +  finish_builtin_struct (type, name, field_chain, NULL_TREE);
> +  return type;
> +}
> +
> +/* Returns a extern, comdat VAR_DECL of type TYPE and name NAME. */
> +
> +static tree
> +make_var_decl (tree type, const char *name)
> +{
> +  tree new_decl;
> +  struct varpool_node *vnode;
> +
> +  new_decl = build_decl (UNKNOWN_LOCATION,
> +                        VAR_DECL,
> +                        get_identifier(name),
> +                        type);
> +
> +  DECL_EXTERNAL (new_decl) = 1;
> +  TREE_STATIC (new_decl) = 1;
> +  TREE_PUBLIC (new_decl) = 1;
> +  DECL_INITIAL (new_decl) = 0;
> +  DECL_ARTIFICIAL (new_decl) = 0;
> +  DECL_PRESERVE_P (new_decl) = 1;
> +
> +  make_decl_one_only (new_decl, DECL_ASSEMBLER_NAME (new_decl));
> +  assemble_variable (new_decl, 0, 0, 0);
> +
> +  vnode = varpool_node (new_decl);
> +  gcc_assert (vnode != NULL);
> +  /* Set finalized to 1, otherwise it asserts in function "write_symbol" in
> +     lto-streamer-out.c. */
> +  vnode->finalized = 1;
> +
> +  return new_decl;
> +}
> +
> +/* Traverses the chain of fields in STRUCT_TYPE and returns the FIELD_NUM
> +   numbered field. */
> +
> +static tree
> +get_field_from_struct (tree struct_type, int field_num)
> +{
> +  int i;
> +  tree field = TYPE_FIELDS (struct_type);
> +
> +  for (i = 0; i < field_num; i++, field = DECL_CHAIN(field))
> +    {
> +      gcc_assert (field != NULL_TREE);
> +    }
> +
> +  return field;
> +}
> +
> +/* FNDECL is a __builtin_cpu_* call that is folded into an integer defined
> +   in libgcc/config/i386/i386-cpuinfo.c */
> +
> +static tree
> +fold_builtin_cpu (enum ix86_builtins fn_code)
> +{
> +  /* This is the order of bit-fields in __processor_features in
> +     i386-cpuinfo.c */
> +  enum processor_features
> +  {
> +    F_CMOV = 0,
> +    F_MMX,
> +    F_POPCNT,
> +    F_SSE,
> +    F_SSE2,
> +    F_SSE3,
> +    F_SSSE3,
> +    F_SSE4_1,
> +    F_SSE4_2,
> +    F_MAX
> +  };
> +
> +  /* This is the order of bit-fields in __processor_model in
> +     i386-cpuinfo.c */
> +  enum processor_model
> +  {
> +    M_AMD = 0,
> +    M_INTEL,
> +    M_INTEL_ATOM,
> +    M_INTEL_CORE2,
> +    M_INTEL_COREI7,
> +    M_INTEL_COREI7_NEHALEM,
> +    M_INTEL_COREI7_WESTMERE,
> +    M_INTEL_COREI7_SANDYBRIDGE,
> +    M_AMDFAM10H,
> +    M_AMDFAM10H_BARCELONA,
> +    M_AMDFAM10H_SHANGHAI,
> +    M_AMDFAM10H_ISTANBUL,
> +    M_AMDFAM15H_BDVER1,
> +    M_AMDFAM15H_BDVER2,
> +    M_MAX
> +  };
> +
> +  static tree __processor_features_type = NULL_TREE;
> +  static tree __cpu_features_var = NULL_TREE;
> +  static tree __processor_model_type = NULL_TREE;
> +  static tree __cpu_model_var = NULL_TREE;
> +  static tree field;
> +  static tree which_struct;
> +
> +  if (__processor_features_type == NULL_TREE)
> +    __processor_features_type = build_struct_with_one_bit_fields (F_MAX,
> +                                 "__processor_features");
> +
> +  if (__processor_model_type == NULL_TREE)
> +    __processor_model_type = build_struct_with_one_bit_fields (M_MAX,
> +                                 "__processor_model");
> +
> +  if (__cpu_features_var == NULL_TREE)
> +    __cpu_features_var = make_var_decl (__processor_features_type,
> +                                       "__cpu_features");
> +
> +  if (__cpu_model_var == NULL_TREE)
> +    __cpu_model_var = make_var_decl (__processor_model_type,
> +                                    "__cpu_model");
> +
> +  /* Look at the code to identify the field requested. */
> +  switch (fn_code)
> +    {
> +    case IX86_BUILTIN_CPU_SUPPORTS_CMOV:
> +      field = get_field_from_struct (__processor_features_type, F_CMOV);
> +      which_struct = __cpu_features_var;
> +      break;
> +    case IX86_BUILTIN_CPU_SUPPORTS_MMX:
> +      field = get_field_from_struct (__processor_features_type, F_MMX);
> +      which_struct = __cpu_features_var;
> +      break;
> +    case IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT:
> +      field = get_field_from_struct (__processor_features_type, F_POPCNT);
> +      which_struct = __cpu_features_var;
> +      break;
> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE:
> +      field = get_field_from_struct (__processor_features_type, F_SSE);
> +      which_struct = __cpu_features_var;
> +      break;
> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE2:
> +      field = get_field_from_struct (__processor_features_type, F_SSE2);
> +      which_struct = __cpu_features_var;
> +      break;
> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE3:
> +      field = get_field_from_struct (__processor_features_type, F_SSE3);
> +      which_struct = __cpu_features_var;
> +      break;
> +    case IX86_BUILTIN_CPU_SUPPORTS_SSSE3:
> +      field = get_field_from_struct (__processor_features_type, F_SSSE3);
> +      which_struct = __cpu_features_var;
> +      break;
> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_1:
> +      field = get_field_from_struct (__processor_features_type, F_SSE4_1);
> +      which_struct = __cpu_features_var;
> +      break;
> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_2:
> +      field = get_field_from_struct (__processor_features_type, F_SSE4_2);
> +      which_struct = __cpu_features_var;
> +      break;
> +    case IX86_BUILTIN_CPU_IS_AMD:
> +      field = get_field_from_struct (__processor_model_type, M_AMD);
> +      which_struct = __cpu_model_var;
> +      break;
> +    case IX86_BUILTIN_CPU_IS_INTEL:
> +      field = get_field_from_struct (__processor_model_type, M_INTEL);
> +      which_struct = __cpu_model_var;
> +      break;
> +    case IX86_BUILTIN_CPU_IS_INTEL_ATOM:
> +      field = get_field_from_struct (__processor_model_type, M_INTEL_ATOM);
> +      which_struct = __cpu_model_var;
> +      break;
> +    case IX86_BUILTIN_CPU_IS_INTEL_CORE2:
> +      field = get_field_from_struct (__processor_model_type, M_INTEL_CORE2);
> +      which_struct = __cpu_model_var;
> +      break;
> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7:
> +      field = get_field_from_struct (__processor_model_type,
> +                                    M_INTEL_COREI7);
> +      which_struct = __cpu_model_var;
> +      break;
> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM:
> +      field = get_field_from_struct (__processor_model_type,
> +                                    M_INTEL_COREI7_NEHALEM);
> +      which_struct = __cpu_model_var;
> +      break;
> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE:
> +      field = get_field_from_struct (__processor_model_type,
> +                                    M_INTEL_COREI7_WESTMERE);
> +      which_struct = __cpu_model_var;
> +      break;
> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE:
> +      field = get_field_from_struct (__processor_model_type,
> +                                    M_INTEL_COREI7_SANDYBRIDGE);
> +      which_struct = __cpu_model_var;
> +      break;
> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H:
> +      field = get_field_from_struct (__processor_model_type,
> +                                    M_AMDFAM10H);
> +      which_struct = __cpu_model_var;
> +      break;
> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA:
> +      field = get_field_from_struct (__processor_model_type,
> +                                    M_AMDFAM10H_BARCELONA);
> +      which_struct = __cpu_model_var;
> +      break;
> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI:
> +      field = get_field_from_struct (__processor_model_type,
> +                                    M_AMDFAM10H_SHANGHAI);
> +      which_struct = __cpu_model_var;
> +      break;
> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL:
> +      field = get_field_from_struct (__processor_model_type,
> +                                    M_AMDFAM10H_ISTANBUL);
> +      which_struct = __cpu_model_var;
> +      break;
> +    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1:
> +      field = get_field_from_struct (__processor_model_type,
> +                                    M_AMDFAM15H_BDVER1);
> +      which_struct = __cpu_model_var;
> +      break;
> +    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2:
> +      field = get_field_from_struct (__processor_model_type,
> +                                    M_AMDFAM15H_BDVER2);
> +      which_struct = __cpu_model_var;
> +      break;
> +    default:
> +      return NULL_TREE;
> +    }
> +
> +  return build3 (COMPONENT_REF, TREE_TYPE (field), which_struct, field, NULL_TREE);
> +}
> +
> +static tree
> +ix86_fold_builtin (tree fndecl, int n_args ATTRIBUTE_UNUSED,
> +                  tree *args ATTRIBUTE_UNUSED, bool ignore ATTRIBUTE_UNUSED)
> +{
> +  const char* decl_name = IDENTIFIER_POINTER (DECL_NAME (fndecl));
> +  if (DECL_BUILT_IN_CLASS (fndecl) == BUILT_IN_MD
> +      && strstr(decl_name, "__builtin_cpu") != NULL)
> +    {
> +      enum ix86_builtins code = (enum ix86_builtins)
> +                               DECL_FUNCTION_CODE (fndecl);
> +      return fold_builtin_cpu (code);
> +    }
> +  return NULL_TREE;
> +}
> +
> +/* A builtin to init/return the cpu type or feature.  Returns an
> +   integer and the type is a const if IS_CONST is set. */
> +
> +static void
> +make_platform_builtin (const char* name, int code, int is_const)
> +{
> +  tree decl;
> +  tree type;
> +
> +  type = ix86_get_builtin_func_type (INT_FTYPE_VOID);
> +  decl = add_builtin_function (name, type, code, BUILT_IN_MD,
> +                              NULL, NULL_TREE);
> +  gcc_assert (decl != NULL_TREE);
> +  ix86_builtins[(int) code] = decl;
> +  if (is_const)
> +    TREE_READONLY (decl) = 1;
> +}
> +
> +/* Builtins to get CPU type and features supported. */
> +
> +static void
> +ix86_init_platform_type_builtins (void)
> +{
> +  make_platform_builtin ("__builtin_cpu_init",
> +                        IX86_BUILTIN_CPU_INIT, 0);
> +  make_platform_builtin ("__builtin_cpu_supports_cmov",
> +                        IX86_BUILTIN_CPU_SUPPORTS_CMOV, 1);
> +  make_platform_builtin ("__builtin_cpu_supports_mmx",
> +                        IX86_BUILTIN_CPU_SUPPORTS_MMX, 1);
> +  make_platform_builtin ("__builtin_cpu_supports_popcount",
> +                        IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT, 1);
> +  make_platform_builtin ("__builtin_cpu_supports_sse",
> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE, 1);
> +  make_platform_builtin ("__builtin_cpu_supports_sse2",
> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE2, 1);
> +  make_platform_builtin ("__builtin_cpu_supports_sse3",
> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE3, 1);
> +  make_platform_builtin ("__builtin_cpu_supports_ssse3",
> +                        IX86_BUILTIN_CPU_SUPPORTS_SSSE3, 1);
> +  make_platform_builtin ("__builtin_cpu_supports_sse4_1",
> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE4_1, 1);
> +  make_platform_builtin ("__builtin_cpu_supports_sse4_2",
> +                        IX86_BUILTIN_CPU_SUPPORTS_SSE4_2, 1);
> +  make_platform_builtin ("__builtin_cpu_is_amd",
> +                        IX86_BUILTIN_CPU_IS_AMD, 1);
> +  make_platform_builtin ("__builtin_cpu_is_intel_atom",
> +                        IX86_BUILTIN_CPU_IS_INTEL_ATOM, 1);
> +  make_platform_builtin ("__builtin_cpu_is_intel_core2",
> +                        IX86_BUILTIN_CPU_IS_INTEL_CORE2, 1);
> +  make_platform_builtin ("__builtin_cpu_is_intel",
> +                        IX86_BUILTIN_CPU_IS_INTEL, 1);
> +  make_platform_builtin ("__builtin_cpu_is_intel_corei7",
> +                        IX86_BUILTIN_CPU_IS_INTEL_COREI7, 1);
> +  make_platform_builtin ("__builtin_cpu_is_intel_corei7_nehalem",
> +                        IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM, 1);
> +  make_platform_builtin ("__builtin_cpu_is_intel_corei7_westmere",
> +                        IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE, 1);
> +  make_platform_builtin ("__builtin_cpu_is_intel_corei7_sandybridge",
> +                        IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE, 1);
> +  make_platform_builtin ("__builtin_cpu_is_amdfam10",
> +                        IX86_BUILTIN_CPU_IS_AMDFAM10H, 1);
> +  make_platform_builtin ("__builtin_cpu_is_amdfam10_barcelona",
> +                        IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA, 1);
> +  make_platform_builtin ("__builtin_cpu_is_amdfam10_shanghai",
> +                        IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI, 1);
> +  make_platform_builtin ("__builtin_cpu_is_amdfam10_istanbul",
> +                        IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL, 1);
> +  make_platform_builtin ("__builtin_cpu_is_amdfam15_bdver1",
> +                        IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1, 1);
> +  make_platform_builtin ("__builtin_cpu_is_amdfam15_bdver2",
> +                        IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2, 1);
> +}
> +
>  /* Internal method for ix86_init_builtins.  */
>
>  static void
> @@ -27529,6 +28143,9 @@ ix86_init_builtins (void)
>
>   ix86_init_builtin_types ();
>
> +  /* Builtins to get CPU type and features. */
> +  ix86_init_platform_type_builtins ();
> +
>   /* TFmode support builtins.  */
>   def_builtin_const (0, "__builtin_infq",
>                     FLOAT128_FTYPE_VOID, IX86_BUILTIN_INFQ);
> @@ -29145,6 +29762,48 @@ ix86_expand_builtin (tree exp, rtx target, rtx sub
>   enum machine_mode mode0, mode1, mode2, mode3, mode4;
>   unsigned int fcode = DECL_FUNCTION_CODE (fndecl);
>
> +  /* For CPU builtins that can be folded, fold first and expand the fold.  */
> +  switch (fcode)
> +    {
> +    case IX86_BUILTIN_CPU_SUPPORTS_CMOV:
> +    case IX86_BUILTIN_CPU_SUPPORTS_MMX:
> +    case IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT:
> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE:
> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE2:
> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE3:
> +    case IX86_BUILTIN_CPU_SUPPORTS_SSSE3:
> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_1:
> +    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_2:
> +    case IX86_BUILTIN_CPU_IS_AMD:
> +    case IX86_BUILTIN_CPU_IS_INTEL:
> +    case IX86_BUILTIN_CPU_IS_INTEL_ATOM:
> +    case IX86_BUILTIN_CPU_IS_INTEL_CORE2:
> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7:
> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM:
> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE:
> +    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE:
> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H:
> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA:
> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI:
> +    case IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL:
> +    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1:
> +    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2:
> +      {
> +        tree fold_expr = fold_builtin_cpu ((enum ix86_builtins) fcode);
> +       gcc_assert (fold_expr != NULL_TREE);
> +        return expand_expr (fold_expr, target, mode, EXPAND_NORMAL);
> +      }
> +    case IX86_BUILTIN_CPU_INIT:
> +      {
> +       /* Make it call __cpu_indicator_init in libgcc. */
> +       tree call_expr, fndecl, type;
> +        type = build_function_type_list (integer_type_node, NULL_TREE);
> +       fndecl = build_fn_decl ("__cpu_indicator_init", type);
> +       call_expr = build_call_expr (fndecl, 0);
> +       return expand_expr (call_expr, target, mode, EXPAND_NORMAL);
> +      }
> +    }
> +
>   /* Determine whether the builtin function is available under the current ISA.
>      Originally the builtin was not created if it wasn't applicable to the
>      current ISA based on the command line switches.  With function specific
> @@ -38610,6 +39269,12 @@ ix86_autovectorize_vector_sizes (void)
>  #undef TARGET_BUILD_BUILTIN_VA_LIST
>  #define TARGET_BUILD_BUILTIN_VA_LIST ix86_build_builtin_va_list
>
> +#undef TARGET_FOLD_BUILTIN
> +#define TARGET_FOLD_BUILTIN ix86_fold_builtin
> +
>  #undef TARGET_ENUM_VA_LIST_P
>  #define TARGET_ENUM_VA_LIST_P ix86_enum_va_list
> Index: gcc/testsuite/gcc.target/i386/builtin_target.c
> ===================================================================
> --- gcc/testsuite/gcc.target/i386/builtin_target.c      (revision 0)
> +++ gcc/testsuite/gcc.target/i386/builtin_target.c      (revision 0)
> @@ -0,0 +1,61 @@
> +/* This test checks if the __builtin_cpu_* calls are recognized. */
> +
> +/* { dg-do run } */
> +
> +int
> +fn1 ()
> +{
> +  if (__builtin_cpu_supports_cmov () < 0)
> +    return -1;
> +  if (__builtin_cpu_supports_mmx () < 0)
> +    return -1;
> +  if (__builtin_cpu_supports_popcount () < 0)
> +    return -1;
> +  if (__builtin_cpu_supports_sse () < 0)
> +    return -1;
> +  if (__builtin_cpu_supports_sse2 () < 0)
> +    return -1;
> +  if (__builtin_cpu_supports_sse3 () < 0)
> +    return -1;
> +  if (__builtin_cpu_supports_ssse3 () < 0)
> +    return -1;
> +  if (__builtin_cpu_supports_sse4_1 () < 0)
> +    return -1;
> +  if (__builtin_cpu_supports_sse4_2 () < 0)
> +    return -1;
> +  if (__builtin_cpu_is_amd () < 0)
> +    return -1;
> +  if (__builtin_cpu_is_intel () < 0)
> +    return -1;
> +  if (__builtin_cpu_is_intel_atom () < 0)
> +    return -1;
> +  if (__builtin_cpu_is_intel_core2 () < 0)
> +    return -1;
> +  if (__builtin_cpu_is_intel_corei7 () < 0)
> +    return -1;
> +  if (__builtin_cpu_is_intel_corei7_nehalem () < 0)
> +    return -1;
> +  if (__builtin_cpu_is_intel_corei7_westmere () < 0)
> +    return -1;
> +  if (__builtin_cpu_is_intel_corei7_sandybridge () < 0)
> +    return -1;
> +  if (__builtin_cpu_is_amdfam10 () < 0)
> +    return -1;
> +  if (__builtin_cpu_is_amdfam10_barcelona () < 0)
> +    return -1;
> +  if (__builtin_cpu_is_amdfam10_shanghai () < 0)
> +    return -1;
> +  if (__builtin_cpu_is_amdfam10_istanbul () < 0)
> +    return -1;
> +  if (__builtin_cpu_is_amdfam15_bdver1 () < 0)
> +    return -1;
> +  if (__builtin_cpu_is_amdfam15_bdver2 () < 0)
> +    return -1;
> +
> +  return 0;
> +}
> +
> +int main ()
> +{
> +  return fn1 ();
> +}
>
> --
> This patch is available for review at http://codereview.appspot.com/5754058

Support for Runtime CPU type detection via builtins (issue5754058)

[Sriraman Tallam]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 32174 bytes --]

Patch for CPU detection at run-time.
===================================

Patch for CPU detection at run-time, to be used in dispatching of
multi-versioned functions.   Please see this discussion:
http://gcc.gnu.org/ml/gcc-patches/2011-08/msg01355.html
when this patch for reviewed the last time.

For more detailed description:
http://gcc.gnu.org/ml/gcc/2012-03/msg00074.html

One of the main concerns was about making CPU detection initialization a
constructor. The main point raised was about constructor ordering. I have
added a priority value to the CPU detection constructor to make it very high
priority so that it is guaranteed to fire before every constructor without
an explicitly marked priority value of 101.  However, IFUNC initializers
will still fire before this constructor, so the cpu initialization routine
has to be explicitly called in such initializers for which I have added a
builtin: __builtin_cpu_init ().

This patch adds the following new builtins:

* __builtin_cpu_init
* __builtin_cpu_supports_cmov
* __builtin_cpu_supports_mmx
* __builtin_cpu_supports_popcount
* __builtin_cpu_supports_sse
* __builtin_cpu_supports_sse2
* __builtin_cpu_supports_sse3
* __builtin_cpu_supports_ssse3
* __builtin_cpu_supports_sse4_1
* __builtin_cpu_supports_sse4_2
* __builtin_cpu_is_amd
* __builtin_cpu_is_intel_atom
* __builtin_cpu_is_intel_core2
* __builtin_cpu_is_intel
* __builtin_cpu_is_intel_corei7
* __builtin_cpu_is_intel_corei7_nehalem
* __builtin_cpu_is_intel_corei7_westmere
* __builtin_cpu_is_intel_corei7_sandybridge
* __builtin_cpu_is_amdfam10
* __builtin_cpu_is_amdfam10_barcelona
* __builtin_cpu_is_amdfam10_shanghai
* __builtin_cpu_is_amdfam10_istanbul
* __builtin_cpu_is_amdfam15_bdver1
* __builtin_cpu_is_amdfam15_bdver2


	* config/i386/i386.c (build_struct_with_one_bit_fields): New function.
	(make_var_decl): New function.
	(get_field_from_struct): New function.
	(fold_builtin_target): New function.
	(ix86_fold_builtin): New function.
	(ix86_expand_builtin): Expand new builtins by folding them.
	(make_platform_builtin): New functions.
	(ix86_init_platform_type_builtins): Make the new builtins.
	(ix86_init_builtins): Make new builtins to detect CPU type.
	(TARGET_FOLD_BUILTIN): New macro.
	(IX86_BUILTIN_CPU_SUPPORTS_CMOV): New enum value.
	(IX86_BUILTIN_CPU_SUPPORTS_MMX): New enum value.
	(IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT): New enum value.
	(IX86_BUILTIN_CPU_SUPPORTS_SSE): New enum value.
	(IX86_BUILTIN_CPU_SUPPORTS_SSE2): New enum value.
	(IX86_BUILTIN_CPU_SUPPORTS_SSE3): New enum value.
	(IX86_BUILTIN_CPU_SUPPORTS_SSSE3): New enum value.
	(IX86_BUILTIN_CPU_SUPPORTS_SSE4_1): New enum value.
	(IX86_BUILTIN_CPU_SUPPORTS_SSE4_2): New enum value.
	(IX86_BUILTIN_CPU_INIT): New enum value.
	(IX86_BUILTIN_CPU_IS_AMD): New enum value.
	(IX86_BUILTIN_CPU_IS_INTEL): New enum value.
	(IX86_BUILTIN_CPU_IS_INTEL_ATOM): New enum value.
	(IX86_BUILTIN_CPU_IS_INTEL_CORE2): New enum value.
	(IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM): New enum value.
	(IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE): New enum value.
	(IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE): New enum value.
	(IX86_BUILTIN_CPU_IS_AMDFAM10_BARCELONA): New enum value.
	(IX86_BUILTIN_CPU_IS_AMDFAM10_SHANGHAI): New enum value.
	(IX86_BUILTIN_CPU_IS_AMDFAM10_ISTANBUL): New enum value.
	(IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1): New enum value.
	(IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2): New enum value.
	* config/i386/i386-builtin-types.def: New function type.
	* testsuite/gcc.target/builtin_target.c: New testcase.
	
	* libgcc/config/i386/i386-cpuinfo.c: New file.
	* libgcc/config/i386/t-cpuinfo: New file.
	* libgcc/config.host: Include t-cpuinfo.
	* libgcc/config/i386/libgcc-glibc.ver: Version symbols __cpu_model
	and __cpu_features.

Index: libgcc/config.host
===================================================================
--- libgcc/config.host	(revision 184971)
+++ libgcc/config.host	(working copy)
@@ -1142,7 +1142,7 @@ i[34567]86-*-linux* | x86_64-*-linux* | \
   i[34567]86-*-kfreebsd*-gnu | x86_64-*-kfreebsd*-gnu | \
   i[34567]86-*-knetbsd*-gnu | \
   i[34567]86-*-gnu*)
-	tmake_file="${tmake_file} t-tls i386/t-linux"
+	tmake_file="${tmake_file} t-tls i386/t-linux i386/t-cpuinfo"
 	if test "$libgcc_cv_cfi" = "yes"; then
 		tmake_file="${tmake_file} t-stack i386/t-stack-i386"
 	fi
Index: libgcc/config/i386/t-cpuinfo
===================================================================
--- libgcc/config/i386/t-cpuinfo	(revision 0)
+++ libgcc/config/i386/t-cpuinfo	(revision 0)
@@ -0,0 +1 @@
+LIB2ADD += $(srcdir)/config/i386/i386-cpuinfo.c
Index: libgcc/config/i386/i386-cpuinfo.c
===================================================================
--- libgcc/config/i386/i386-cpuinfo.c	(revision 0)
+++ libgcc/config/i386/i386-cpuinfo.c	(revision 0)
@@ -0,0 +1,306 @@
+/* Get CPU type and Features for x86 processors.
+   Copyright (C) 2011 Free Software Foundation, Inc.
+   Contributed by Sriraman Tallam (tmsriram@google.com)
+
+This file is part of GCC.
+
+GCC is free software; you can redistribute it and/or modify it under
+the terms of the GNU General Public License as published by the Free
+Software Foundation; either version 3, or (at your option) any later
+version.
+
+GCC is distributed in the hope that it will be useful, but WITHOUT ANY
+WARRANTY; without even the implied warranty of MERCHANTABILITY or
+FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+for more details.
+
+You should have received a copy of the GNU General Public License
+along with GCC; see the file COPYING3.  If not see
+<http://www.gnu.org/licenses/>. */
+
+#include "cpuid.h"
+#include "tsystem.h"
+
+int __cpu_indicator_init (void) __attribute__ ((constructor (101)));
+
+enum vendor_signatures
+{
+  SIG_INTEL =	0x756e6547 /* Genu */,
+  SIG_AMD =	0x68747541 /* Auth */
+};
+
+/* ISA Features supported. */
+
+struct __processor_features
+{
+  unsigned int __cpu_cmov : 1;
+  unsigned int __cpu_mmx : 1;
+  unsigned int __cpu_popcnt : 1;
+  unsigned int __cpu_sse : 1;
+  unsigned int __cpu_sse2 : 1;
+  unsigned int __cpu_sse3 : 1;
+  unsigned int __cpu_ssse3 : 1;
+  unsigned int __cpu_sse4_1 : 1;
+  unsigned int __cpu_sse4_2 : 1;
+} __cpu_features;
+
+/* Processor Model. */
+
+struct __processor_model
+{
+  /* Vendor. */
+  unsigned int __cpu_is_amd : 1;
+  unsigned int __cpu_is_intel : 1;
+  /* CPU type. */
+  unsigned int __cpu_is_intel_atom : 1;
+  unsigned int __cpu_is_intel_core2 : 1;
+  unsigned int __cpu_is_intel_corei7 : 1;
+  unsigned int __cpu_is_intel_corei7_nehalem : 1;
+  unsigned int __cpu_is_intel_corei7_westmere : 1;
+  unsigned int __cpu_is_intel_corei7_sandybridge : 1;
+  unsigned int __cpu_is_amdfam10h : 1;
+  unsigned int __cpu_is_amdfam10h_barcelona : 1;
+  unsigned int __cpu_is_amdfam10h_shanghai : 1;
+  unsigned int __cpu_is_amdfam10h_istanbul : 1;
+  unsigned int __cpu_is_amdfam15h_bdver1 : 1;
+  unsigned int __cpu_is_amdfam15h_bdver2 : 1;
+} __cpu_model;
+
+/* Get the specific type of AMD CPU.  */
+
+static void
+get_amd_cpu (unsigned int family, unsigned int model)
+{
+  switch (family)
+    {
+    /* AMD Family 10h.  */
+    case 0x10:
+      switch (model)
+	{
+	case 0x2:
+	  /* Barcelona.  */
+	  __cpu_model.__cpu_is_amdfam10h = 1;
+	  __cpu_model.__cpu_is_amdfam10h_barcelona = 1;
+	  break;
+	case 0x4:
+	  /* Shanghai.  */
+	  __cpu_model.__cpu_is_amdfam10h = 1;
+	  __cpu_model.__cpu_is_amdfam10h_shanghai = 1;
+	  break;
+	case 0x8:
+	  /* Istanbul.  */
+	  __cpu_model.__cpu_is_amdfam10h = 1;
+	  __cpu_model.__cpu_is_amdfam10h_istanbul = 1;
+	  break;
+	default:
+	  break;
+	}
+      break;
+    /* AMD Family 15h.  */
+    case 0x15:
+      /* Bulldozer version 1.  */
+      if (model >= 0 && model <= 0xf)
+        __cpu_model.__cpu_is_amdfam15h_bdver1 = 1;
+      /* Bulldozer version 2.  */
+      if (model >= 0x10 && model <= 0x1f)
+        __cpu_model.__cpu_is_amdfam15h_bdver2 = 1;
+      break;
+    default:
+      break;
+    }
+}
+
+/* Get the specific type of Intel CPU.  */
+
+static void
+get_intel_cpu (unsigned int family, unsigned int model, unsigned int brand_id)
+{
+  /* Parse family and model only if brand ID is 0. */
+  if (brand_id == 0)
+    {
+      switch (family)
+	{
+	case 0x5:
+	  /* Pentium.  */
+	  break;
+	case 0x6:
+	  switch (model)
+	    {
+	    case 0x1c:
+	    case 0x26:
+	      /* Atom.  */
+	      __cpu_model.__cpu_is_intel_atom = 1;
+	      break;
+	    case 0x1a:
+	    case 0x1e:
+	    case 0x1f:
+	    case 0x2e:
+	      /* Nehalem.  */
+	      __cpu_model.__cpu_is_intel_corei7 = 1;
+	      __cpu_model.__cpu_is_intel_corei7_nehalem = 1;
+	      break;
+	    case 0x25:
+	    case 0x2c:
+	    case 0x2f:
+	      /* Westmere.  */
+	      __cpu_model.__cpu_is_intel_corei7 = 1;
+	      __cpu_model.__cpu_is_intel_corei7_westmere = 1;
+	      break;
+	    case 0x2a:
+	      /* Sandy Bridge.  */
+	      __cpu_model.__cpu_is_intel_corei7 = 1;
+	      __cpu_model.__cpu_is_intel_corei7_sandybridge = 1;
+	      break;
+	    case 0x17:
+	    case 0x1d:
+	      /* Penryn.  */
+	    case 0x0f:
+	      /* Merom.  */
+	      __cpu_model.__cpu_is_intel_core2 = 1;
+	      break;
+	    default:
+	      break;
+	    }
+	  break;
+	default:
+	  /* We have no idea.  */
+	  break;
+	}
+    }
+}	             	
+
+static void
+get_available_features (unsigned int ecx, unsigned int edx)
+{
+  __cpu_features.__cpu_cmov = (edx & bit_CMOV) ? 1 : 0;
+  __cpu_features.__cpu_mmx = (edx & bit_MMX) ? 1 : 0;
+  __cpu_features.__cpu_sse = (edx & bit_SSE) ? 1 : 0;
+  __cpu_features.__cpu_sse2 = (edx & bit_SSE2) ? 1 : 0;
+  __cpu_features.__cpu_popcnt = (ecx & bit_POPCNT) ? 1 : 0;
+  __cpu_features.__cpu_sse3 = (ecx & bit_SSE3) ? 1 : 0;
+  __cpu_features.__cpu_ssse3 = (ecx & bit_SSSE3) ? 1 : 0;
+  __cpu_features.__cpu_sse4_1 = (ecx & bit_SSE4_1) ? 1 : 0;
+  __cpu_features.__cpu_sse4_2 = (ecx & bit_SSE4_2) ? 1 : 0;
+}
+
+
+/* Sanity check for the vendor and cpu type flags.  */
+
+static int
+sanity_check (void)
+{
+  unsigned int one_type = 0;
+
+  /* Vendor cannot be Intel and AMD. */
+  gcc_assert((__cpu_model.__cpu_is_intel == 0)
+             || (__cpu_model.__cpu_is_amd == 0));
+ 
+  /* Only one CPU type can be set. */
+  one_type = (__cpu_model.__cpu_is_intel_atom
+	      + __cpu_model.__cpu_is_intel_core2
+	      + __cpu_model.__cpu_is_intel_corei7_nehalem
+	      + __cpu_model.__cpu_is_intel_corei7_westmere
+	      + __cpu_model.__cpu_is_intel_corei7_sandybridge
+	      + __cpu_model.__cpu_is_amdfam10h_barcelona
+	      + __cpu_model.__cpu_is_amdfam10h_shanghai
+	      + __cpu_model.__cpu_is_amdfam10h_istanbul
+	      + __cpu_model.__cpu_is_amdfam15h_bdver1
+	      + __cpu_model.__cpu_is_amdfam15h_bdver2);
+
+  gcc_assert (one_type <= 1);
+  return 0;
+}
+
+/* A noinline function calling __get_cpuid. Having many calls to
+   cpuid in one function in 32-bit mode causes GCC to complain:
+   "can’t find a register in class ‘CLOBBERED_REGS’".  This is
+   related to PR rtl-optimization 44174. */
+
+static int __attribute__ ((noinline))
+__get_cpuid_output (unsigned int __level,
+		    unsigned int *__eax, unsigned int *__ebx,
+		    unsigned int *__ecx, unsigned int *__edx)
+{
+  return __get_cpuid (__level, __eax, __ebx, __ecx, __edx);
+}
+
+
+/* A constructor function that is sets __cpu_model and __cpu_features with
+   the right values.  This needs to run only once.  This constructor is
+   given the highest priority and it should run before constructors without
+   the priority set.  However, it still runs after ifunc initializers and
+   needs to be called explicitly there.  */
+
+int __attribute__ ((constructor (101)))
+__cpu_indicator_init (void)
+{
+  unsigned int eax, ebx, ecx, edx;
+
+  int max_level = 5;
+  unsigned int vendor;
+  unsigned int model, family, brand_id;
+  unsigned int extended_model, extended_family;
+  static int called = 0;
+
+  /* This function needs to run just once.  */
+  if (called)
+    return 0;
+  else
+    called = 1;
+
+  /* Assume cpuid insn present. Run in level 0 to get vendor id. */
+  if (!__get_cpuid_output (0, &eax, &ebx, &ecx, &edx))
+    return -1;
+
+  vendor = ebx;
+  max_level = eax;
+
+  if (max_level < 1)
+    return -1;
+
+  if (!__get_cpuid_output (1, &eax, &ebx, &ecx, &edx))
+    return -1;
+
+  model = (eax >> 4) & 0x0f;
+  family = (eax >> 8) & 0x0f;
+  brand_id = ebx & 0xff;
+  extended_model = (eax >> 12) & 0xf0;
+  extended_family = (eax >> 20) & 0xff;
+
+  if (vendor == SIG_INTEL)
+    {
+      /* Adjust model and family for Intel CPUS. */
+      if (family == 0x0f)
+	{
+	  family += extended_family;
+	  model += extended_model;
+	}
+      else if (family == 0x06)
+	model += extended_model;
+
+      /* Get CPU type.  */
+      __cpu_model.__cpu_is_intel = 1;
+      get_intel_cpu (family, model, brand_id);
+    }
+
+  if (vendor == SIG_AMD)
+    {
+      /* Adjust model and family for AMD CPUS. */
+      if (family == 0x0f)
+	{
+	  family += extended_family;
+	  model += (extended_model << 4);
+	}
+
+      /* Get CPU type.  */
+      __cpu_model.__cpu_is_amd = 1;
+      get_amd_cpu (family, model);
+    }
+
+  /* Find available features. */
+  get_available_features (ecx, edx);
+
+  sanity_check ();
+
+  return 0;
+}
Index: libgcc/config/i386/libgcc-glibc.ver
===================================================================
--- libgcc/config/i386/libgcc-glibc.ver	(revision 184971)
+++ libgcc/config/i386/libgcc-glibc.ver	(working copy)
@@ -147,6 +147,11 @@ GCC_4.3.0 {
   __trunctfxf2
   __unordtf2
 }
+
+GCC_4.8.0 {
+  __cpu_model
+  __cpu_features
+}
 %else
 GCC_4.4.0 {
   __addtf3
@@ -183,4 +188,9 @@ GCC_4.4.0 {
 GCC_4.5.0 {
   __extendxftf2
 }
+
+GCC_4.8.0 {
+  __cpu_model
+  __cpu_features
+}
 %endif
Index: gcc/config/i386/i386-builtin-types.def
===================================================================
--- gcc/config/i386/i386-builtin-types.def	(revision 184971)
+++ gcc/config/i386/i386-builtin-types.def	(working copy)
@@ -143,6 +143,7 @@ DEF_FUNCTION_TYPE (UINT64)
 DEF_FUNCTION_TYPE (UNSIGNED)
 DEF_FUNCTION_TYPE (VOID)
 DEF_FUNCTION_TYPE (PVOID)
+DEF_FUNCTION_TYPE (INT)
 
 DEF_FUNCTION_TYPE (FLOAT, FLOAT)
 DEF_FUNCTION_TYPE (FLOAT128, FLOAT128)
Index: gcc/config/i386/i386.c
===================================================================
--- gcc/config/i386/i386.c	(revision 184971)
+++ gcc/config/i386/i386.c	(working copy)
@@ -25637,6 +25637,33 @@ enum ix86_builtins
   /* CFString built-in for darwin */
   IX86_BUILTIN_CFSTRING,
 
+  /* Builtins to get CPU features. */
+  IX86_BUILTIN_CPU_SUPPORTS_CMOV,
+  IX86_BUILTIN_CPU_SUPPORTS_MMX,
+  IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT,
+  IX86_BUILTIN_CPU_SUPPORTS_SSE,
+  IX86_BUILTIN_CPU_SUPPORTS_SSE2,
+  IX86_BUILTIN_CPU_SUPPORTS_SSE3,
+  IX86_BUILTIN_CPU_SUPPORTS_SSSE3,
+  IX86_BUILTIN_CPU_SUPPORTS_SSE4_1,
+  IX86_BUILTIN_CPU_SUPPORTS_SSE4_2,
+  /* Builtins to get CPU type. */
+  IX86_BUILTIN_CPU_INIT,
+  IX86_BUILTIN_CPU_IS_AMD,
+  IX86_BUILTIN_CPU_IS_INTEL,
+  IX86_BUILTIN_CPU_IS_INTEL_ATOM,
+  IX86_BUILTIN_CPU_IS_INTEL_CORE2,
+  IX86_BUILTIN_CPU_IS_INTEL_COREI7,
+  IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM,
+  IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE,
+  IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE,
+  IX86_BUILTIN_CPU_IS_AMDFAM10H,
+  IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA,
+  IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI,
+  IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL,
+  IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1,
+  IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2,
+
   IX86_BUILTIN_MAX
 };
 
@@ -27446,6 +27473,593 @@ ix86_init_mmx_sse_builtins (void)
     }
 }
 
+/* Returns a struct type with name NAME and number of fields equal to
+   NUM_FIELDS.  Each field is a unsigned int bit field of length 1 bit. */
+
+static tree
+build_struct_with_one_bit_fields (int num_fields, const char *name)
+{
+  int i;
+  char field_name [10];
+  tree field = NULL_TREE, field_chain = NULL_TREE;
+  tree type = make_node (RECORD_TYPE);
+
+  strcpy (field_name, "k_field");
+
+  for (i = 0; i < num_fields; i++)
+    {
+      /* Name the fields, 0_field, 1_field, ... */
+      field_name [0] = '0' + i;
+      field = build_decl (UNKNOWN_LOCATION, FIELD_DECL,
+			  get_identifier (field_name), unsigned_type_node);
+      DECL_BIT_FIELD (field) = 1;
+      DECL_SIZE (field) = bitsize_one_node;
+      if (field_chain != NULL_TREE)
+	DECL_CHAIN (field) = field_chain;
+      field_chain = field;
+    }
+  finish_builtin_struct (type, name, field_chain, NULL_TREE);
+  return type;
+}
+
+/* Returns a extern, comdat VAR_DECL of type TYPE and name NAME. */
+
+static tree
+make_var_decl (tree type, const char *name)
+{
+  tree new_decl;
+  struct varpool_node *vnode;
+
+  new_decl = build_decl (UNKNOWN_LOCATION,
+	                 VAR_DECL,
+	  	         get_identifier(name),
+		         type);
+
+  DECL_EXTERNAL (new_decl) = 1;
+  TREE_STATIC (new_decl) = 1;
+  TREE_PUBLIC (new_decl) = 1;
+  DECL_INITIAL (new_decl) = 0;
+  DECL_ARTIFICIAL (new_decl) = 0;
+  DECL_PRESERVE_P (new_decl) = 1;
+
+  make_decl_one_only (new_decl, DECL_ASSEMBLER_NAME (new_decl));
+  assemble_variable (new_decl, 0, 0, 0);
+
+  vnode = varpool_node (new_decl);
+  gcc_assert (vnode != NULL);
+  /* Set finalized to 1, otherwise it asserts in function "write_symbol" in
+     lto-streamer-out.c. */
+  vnode->finalized = 1;
+
+  return new_decl;
+}
+
+/* Traverses the chain of fields in STRUCT_TYPE and returns the FIELD_NUM
+   numbered field. */
+
+static tree
+get_field_from_struct (tree struct_type, int field_num)
+{
+  int i;
+  tree field = TYPE_FIELDS (struct_type);
+
+  for (i = 0; i < field_num; i++, field = DECL_CHAIN(field))
+    {
+      gcc_assert (field != NULL_TREE);
+    }
+
+  return field;
+}
+
+/* FNDECL is a __builtin_cpu_* call that is folded into an integer defined
+   in libgcc/config/i386/i386-cpuinfo.c */
+
+static tree 
+fold_builtin_cpu (enum ix86_builtins fn_code)
+{
+  /* This is the order of bit-fields in __processor_features in
+     i386-cpuinfo.c */
+  enum processor_features
+  {
+    F_CMOV = 0,
+    F_MMX,
+    F_POPCNT,
+    F_SSE,
+    F_SSE2,
+    F_SSE3,
+    F_SSSE3,
+    F_SSE4_1,
+    F_SSE4_2,
+    F_MAX
+  };
+
+  /* This is the order of bit-fields in __processor_model in
+     i386-cpuinfo.c */
+  enum processor_model
+  {
+    M_AMD = 0,
+    M_INTEL,
+    M_INTEL_ATOM,
+    M_INTEL_CORE2,
+    M_INTEL_COREI7,
+    M_INTEL_COREI7_NEHALEM,
+    M_INTEL_COREI7_WESTMERE,
+    M_INTEL_COREI7_SANDYBRIDGE,
+    M_AMDFAM10H,
+    M_AMDFAM10H_BARCELONA,
+    M_AMDFAM10H_SHANGHAI,
+    M_AMDFAM10H_ISTANBUL,
+    M_AMDFAM15H_BDVER1,
+    M_AMDFAM15H_BDVER2,
+    M_MAX
+  };
+
+  static tree __processor_features_type = NULL_TREE;
+  static tree __cpu_features_var = NULL_TREE;
+  static tree __processor_model_type = NULL_TREE;
+  static tree __cpu_model_var = NULL_TREE;
+  static tree field;
+  static tree which_struct;
+
+  if (__processor_features_type == NULL_TREE)
+    __processor_features_type = build_struct_with_one_bit_fields (F_MAX,
+ 			          "__processor_features");
+
+  if (__processor_model_type == NULL_TREE)
+    __processor_model_type = build_struct_with_one_bit_fields (M_MAX,
+ 			          "__processor_model");
+
+  if (__cpu_features_var == NULL_TREE)
+    __cpu_features_var = make_var_decl (__processor_features_type,
+					"__cpu_features");
+
+  if (__cpu_model_var == NULL_TREE)
+    __cpu_model_var = make_var_decl (__processor_model_type,
+				     "__cpu_model");
+
+  /* Look at the code to identify the field requested. */ 
+  switch (fn_code)
+    {
+    case IX86_BUILTIN_CPU_SUPPORTS_CMOV:
+      field = get_field_from_struct (__processor_features_type, F_CMOV);
+      which_struct = __cpu_features_var;
+      break;
+    case IX86_BUILTIN_CPU_SUPPORTS_MMX:
+      field = get_field_from_struct (__processor_features_type, F_MMX);
+      which_struct = __cpu_features_var;
+      break;
+    case IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT:
+      field = get_field_from_struct (__processor_features_type, F_POPCNT);
+      which_struct = __cpu_features_var;
+      break;
+    case IX86_BUILTIN_CPU_SUPPORTS_SSE:
+      field = get_field_from_struct (__processor_features_type, F_SSE);
+      which_struct = __cpu_features_var;
+      break;
+    case IX86_BUILTIN_CPU_SUPPORTS_SSE2:
+      field = get_field_from_struct (__processor_features_type, F_SSE2);
+      which_struct = __cpu_features_var;
+      break;
+    case IX86_BUILTIN_CPU_SUPPORTS_SSE3:
+      field = get_field_from_struct (__processor_features_type, F_SSE3);
+      which_struct = __cpu_features_var;
+      break;
+    case IX86_BUILTIN_CPU_SUPPORTS_SSSE3:
+      field = get_field_from_struct (__processor_features_type, F_SSSE3);
+      which_struct = __cpu_features_var;
+      break;
+    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_1:
+      field = get_field_from_struct (__processor_features_type, F_SSE4_1);
+      which_struct = __cpu_features_var;
+      break;
+    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_2:
+      field = get_field_from_struct (__processor_features_type, F_SSE4_2);
+      which_struct = __cpu_features_var;
+      break;
+    case IX86_BUILTIN_CPU_IS_AMD:
+      field = get_field_from_struct (__processor_model_type, M_AMD);
+      which_struct = __cpu_model_var;
+      break;
+    case IX86_BUILTIN_CPU_IS_INTEL:
+      field = get_field_from_struct (__processor_model_type, M_INTEL);
+      which_struct = __cpu_model_var;
+      break;
+    case IX86_BUILTIN_CPU_IS_INTEL_ATOM:
+      field = get_field_from_struct (__processor_model_type, M_INTEL_ATOM);
+      which_struct = __cpu_model_var;
+      break;
+    case IX86_BUILTIN_CPU_IS_INTEL_CORE2:
+      field = get_field_from_struct (__processor_model_type, M_INTEL_CORE2);
+      which_struct = __cpu_model_var;
+      break;
+    case IX86_BUILTIN_CPU_IS_INTEL_COREI7:
+      field = get_field_from_struct (__processor_model_type,
+				     M_INTEL_COREI7);
+      which_struct = __cpu_model_var;
+      break;
+    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM:
+      field = get_field_from_struct (__processor_model_type,
+				     M_INTEL_COREI7_NEHALEM);
+      which_struct = __cpu_model_var;
+      break;
+    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE:
+      field = get_field_from_struct (__processor_model_type,
+				     M_INTEL_COREI7_WESTMERE);
+      which_struct = __cpu_model_var;
+      break;
+    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE:
+      field = get_field_from_struct (__processor_model_type,
+				     M_INTEL_COREI7_SANDYBRIDGE);
+      which_struct = __cpu_model_var;
+      break;
+    case IX86_BUILTIN_CPU_IS_AMDFAM10H:
+      field = get_field_from_struct (__processor_model_type,
+				     M_AMDFAM10H);
+      which_struct = __cpu_model_var;
+      break;
+    case IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA:
+      field = get_field_from_struct (__processor_model_type,
+				     M_AMDFAM10H_BARCELONA);
+      which_struct = __cpu_model_var;
+      break;
+    case IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI:
+      field = get_field_from_struct (__processor_model_type,
+				     M_AMDFAM10H_SHANGHAI);
+      which_struct = __cpu_model_var;
+      break;
+    case IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL:
+      field = get_field_from_struct (__processor_model_type,
+				     M_AMDFAM10H_ISTANBUL);
+      which_struct = __cpu_model_var;
+      break;
+    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1:
+      field = get_field_from_struct (__processor_model_type,
+				     M_AMDFAM15H_BDVER1);
+      which_struct = __cpu_model_var;
+      break;
+    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2:
+      field = get_field_from_struct (__processor_model_type,
+				     M_AMDFAM15H_BDVER2);
+      which_struct = __cpu_model_var;
+      break;
+    default:
+      return NULL_TREE;
+    }
+
+  return build3 (COMPONENT_REF, TREE_TYPE (field), which_struct, field, NULL_TREE);
+}
+
+static tree
+ix86_fold_builtin (tree fndecl, int n_args ATTRIBUTE_UNUSED,
+		   tree *args ATTRIBUTE_UNUSED, bool ignore ATTRIBUTE_UNUSED)
+{
+  const char* decl_name = IDENTIFIER_POINTER (DECL_NAME (fndecl));
+  if (DECL_BUILT_IN_CLASS (fndecl) == BUILT_IN_MD
+      && strstr(decl_name, "__builtin_cpu") != NULL)
+    {
+      enum ix86_builtins code = (enum ix86_builtins)
+				DECL_FUNCTION_CODE (fndecl);
+      return fold_builtin_cpu (code);
+    }
+  return NULL_TREE;
+}
+
+/* A builtin to init/return the cpu type or feature.  Returns an
+   integer and the type is a const if IS_CONST is set. */
+
+static void
+make_platform_builtin (const char* name, int code, int is_const)
+{
+  tree decl;
+  tree type;
+
+  type = ix86_get_builtin_func_type (INT_FTYPE_VOID);
+  decl = add_builtin_function (name, type, code, BUILT_IN_MD,
+			       NULL, NULL_TREE);
+  gcc_assert (decl != NULL_TREE);
+  ix86_builtins[(int) code] = decl;
+  if (is_const)
+    TREE_READONLY (decl) = 1;
+} 
+
+/* Builtins to get CPU type and features supported. */
+
+static void
+ix86_init_platform_type_builtins (void)
+{
+  make_platform_builtin ("__builtin_cpu_init",
+			 IX86_BUILTIN_CPU_INIT, 0);
+  make_platform_builtin ("__builtin_cpu_supports_cmov",
+			 IX86_BUILTIN_CPU_SUPPORTS_CMOV, 1);
+  make_platform_builtin ("__builtin_cpu_supports_mmx",
+			 IX86_BUILTIN_CPU_SUPPORTS_MMX, 1);
+  make_platform_builtin ("__builtin_cpu_supports_popcount",
+			 IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT, 1);
+  make_platform_builtin ("__builtin_cpu_supports_sse",
+			 IX86_BUILTIN_CPU_SUPPORTS_SSE, 1);
+  make_platform_builtin ("__builtin_cpu_supports_sse2",
+			 IX86_BUILTIN_CPU_SUPPORTS_SSE2, 1);
+  make_platform_builtin ("__builtin_cpu_supports_sse3",
+			 IX86_BUILTIN_CPU_SUPPORTS_SSE3, 1);
+  make_platform_builtin ("__builtin_cpu_supports_ssse3",
+			 IX86_BUILTIN_CPU_SUPPORTS_SSSE3, 1);
+  make_platform_builtin ("__builtin_cpu_supports_sse4_1",
+			 IX86_BUILTIN_CPU_SUPPORTS_SSE4_1, 1);
+  make_platform_builtin ("__builtin_cpu_supports_sse4_2",
+			 IX86_BUILTIN_CPU_SUPPORTS_SSE4_2, 1);
+  make_platform_builtin ("__builtin_cpu_is_amd",
+			 IX86_BUILTIN_CPU_IS_AMD, 1);
+  make_platform_builtin ("__builtin_cpu_is_intel_atom",
+			 IX86_BUILTIN_CPU_IS_INTEL_ATOM, 1);
+  make_platform_builtin ("__builtin_cpu_is_intel_core2",
+			 IX86_BUILTIN_CPU_IS_INTEL_CORE2, 1);
+  make_platform_builtin ("__builtin_cpu_is_intel",
+			 IX86_BUILTIN_CPU_IS_INTEL, 1);
+  make_platform_builtin ("__builtin_cpu_is_intel_corei7",
+			 IX86_BUILTIN_CPU_IS_INTEL_COREI7, 1);
+  make_platform_builtin ("__builtin_cpu_is_intel_corei7_nehalem",
+			 IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM, 1);
+  make_platform_builtin ("__builtin_cpu_is_intel_corei7_westmere",
+			 IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE, 1);
+  make_platform_builtin ("__builtin_cpu_is_intel_corei7_sandybridge",
+			 IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE, 1);
+  make_platform_builtin ("__builtin_cpu_is_amdfam10",
+			 IX86_BUILTIN_CPU_IS_AMDFAM10H, 1);
+  make_platform_builtin ("__builtin_cpu_is_amdfam10_barcelona",
+			 IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA, 1);
+  make_platform_builtin ("__builtin_cpu_is_amdfam10_shanghai",
+			 IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI, 1);
+  make_platform_builtin ("__builtin_cpu_is_amdfam10_istanbul",
+			 IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL, 1);
+  make_platform_builtin ("__builtin_cpu_is_amdfam15_bdver1",
+			 IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1, 1);
+  make_platform_builtin ("__builtin_cpu_is_amdfam15_bdver2",
+			 IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2, 1);
+}
+
 /* Internal method for ix86_init_builtins.  */
 
 static void
@@ -27529,6 +28143,9 @@ ix86_init_builtins (void)
 
   ix86_init_builtin_types ();
 
+  /* Builtins to get CPU type and features. */
+  ix86_init_platform_type_builtins ();
+
   /* TFmode support builtins.  */
   def_builtin_const (0, "__builtin_infq",
 		     FLOAT128_FTYPE_VOID, IX86_BUILTIN_INFQ);
@@ -29145,6 +29762,48 @@ ix86_expand_builtin (tree exp, rtx target, rtx sub
   enum machine_mode mode0, mode1, mode2, mode3, mode4;
   unsigned int fcode = DECL_FUNCTION_CODE (fndecl);
 
+  /* For CPU builtins that can be folded, fold first and expand the fold.  */
+  switch (fcode)
+    {
+    case IX86_BUILTIN_CPU_SUPPORTS_CMOV:
+    case IX86_BUILTIN_CPU_SUPPORTS_MMX:
+    case IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT:
+    case IX86_BUILTIN_CPU_SUPPORTS_SSE:
+    case IX86_BUILTIN_CPU_SUPPORTS_SSE2:
+    case IX86_BUILTIN_CPU_SUPPORTS_SSE3:
+    case IX86_BUILTIN_CPU_SUPPORTS_SSSE3:
+    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_1:
+    case IX86_BUILTIN_CPU_SUPPORTS_SSE4_2:
+    case IX86_BUILTIN_CPU_IS_AMD:
+    case IX86_BUILTIN_CPU_IS_INTEL:
+    case IX86_BUILTIN_CPU_IS_INTEL_ATOM:
+    case IX86_BUILTIN_CPU_IS_INTEL_CORE2:
+    case IX86_BUILTIN_CPU_IS_INTEL_COREI7:
+    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM:
+    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE:
+    case IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE:
+    case IX86_BUILTIN_CPU_IS_AMDFAM10H:
+    case IX86_BUILTIN_CPU_IS_AMDFAM10H_BARCELONA:
+    case IX86_BUILTIN_CPU_IS_AMDFAM10H_SHANGHAI:
+    case IX86_BUILTIN_CPU_IS_AMDFAM10H_ISTANBUL:
+    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1:
+    case IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2:
+      {
+        tree fold_expr = fold_builtin_cpu ((enum ix86_builtins) fcode);
+	gcc_assert (fold_expr != NULL_TREE);
+        return expand_expr (fold_expr, target, mode, EXPAND_NORMAL);
+      }
+    case IX86_BUILTIN_CPU_INIT:
+      {
+	/* Make it call __cpu_indicator_init in libgcc. */
+	tree call_expr, fndecl, type;
+        type = build_function_type_list (integer_type_node, NULL_TREE); 
+	fndecl = build_fn_decl ("__cpu_indicator_init", type);
+	call_expr = build_call_expr (fndecl, 0); 
+	return expand_expr (call_expr, target, mode, EXPAND_NORMAL);
+      }
+    }
+
   /* Determine whether the builtin function is available under the current ISA.
      Originally the builtin was not created if it wasn't applicable to the
      current ISA based on the command line switches.  With function specific
@@ -38610,6 +39269,12 @@ ix86_autovectorize_vector_sizes (void)
 #undef TARGET_BUILD_BUILTIN_VA_LIST
 #define TARGET_BUILD_BUILTIN_VA_LIST ix86_build_builtin_va_list
 
+#undef TARGET_FOLD_BUILTIN
+#define TARGET_FOLD_BUILTIN ix86_fold_builtin
+
 #undef TARGET_ENUM_VA_LIST_P
 #define TARGET_ENUM_VA_LIST_P ix86_enum_va_list
Index: gcc/testsuite/gcc.target/i386/builtin_target.c
===================================================================
--- gcc/testsuite/gcc.target/i386/builtin_target.c	(revision 0)
+++ gcc/testsuite/gcc.target/i386/builtin_target.c	(revision 0)
@@ -0,0 +1,61 @@
+/* This test checks if the __builtin_cpu_* calls are recognized. */
+
+/* { dg-do run } */
+
+int
+fn1 ()
+{
+  if (__builtin_cpu_supports_cmov () < 0)
+    return -1;
+  if (__builtin_cpu_supports_mmx () < 0)
+    return -1;
+  if (__builtin_cpu_supports_popcount () < 0)
+    return -1;
+  if (__builtin_cpu_supports_sse () < 0)
+    return -1;
+  if (__builtin_cpu_supports_sse2 () < 0)
+    return -1;
+  if (__builtin_cpu_supports_sse3 () < 0)
+    return -1;
+  if (__builtin_cpu_supports_ssse3 () < 0)
+    return -1;
+  if (__builtin_cpu_supports_sse4_1 () < 0)
+    return -1;
+  if (__builtin_cpu_supports_sse4_2 () < 0)
+    return -1;
+  if (__builtin_cpu_is_amd () < 0)
+    return -1;
+  if (__builtin_cpu_is_intel () < 0)
+    return -1;
+  if (__builtin_cpu_is_intel_atom () < 0)
+    return -1;
+  if (__builtin_cpu_is_intel_core2 () < 0)
+    return -1;
+  if (__builtin_cpu_is_intel_corei7 () < 0)
+    return -1;
+  if (__builtin_cpu_is_intel_corei7_nehalem () < 0)
+    return -1;
+  if (__builtin_cpu_is_intel_corei7_westmere () < 0)
+    return -1;
+  if (__builtin_cpu_is_intel_corei7_sandybridge () < 0)
+    return -1;
+  if (__builtin_cpu_is_amdfam10 () < 0)
+    return -1;
+  if (__builtin_cpu_is_amdfam10_barcelona () < 0)
+    return -1;
+  if (__builtin_cpu_is_amdfam10_shanghai () < 0)
+    return -1;
+  if (__builtin_cpu_is_amdfam10_istanbul () < 0)
+    return -1;
+  if (__builtin_cpu_is_amdfam15_bdver1 () < 0)
+    return -1;
+  if (__builtin_cpu_is_amdfam15_bdver2 () < 0)
+    return -1;
+
+  return 0;
+}
+
+int main ()
+{
+  return fn1 ();
+}

--
This patch is available for review at http://codereview.appspot.com/5754058