date:20170101

r290808 - DR1391: Check for implicit conversion sequences for non-dependent function

2017-01-01 Thread Richard Smith via cfe-commits

Author: rsmith
Date: Sun Jan  1 20:42:17 2017
New Revision: 290808

URL: http://llvm.org/viewvc/llvm-project?rev=290808=rev
Log:
DR1391: Check for implicit conversion sequences for non-dependent function
template parameters between deduction and substitution. The idea is to accept
as many cases as possible, on the basis that substitution failure outside
the immediate context is much more common during substitution than during
implicit conversion sequence formation.

This does not implement the partial ordering portion of DR1391, which so
far appears to be misguided.

Modified:
cfe/trunk/include/clang/Sema/Overload.h
cfe/trunk/include/clang/Sema/Sema.h
cfe/trunk/lib/Sema/SemaOverload.cpp
cfe/trunk/lib/Sema/SemaTemplateDeduction.cpp
cfe/trunk/test/CXX/drs/dr13xx.cpp
cfe/trunk/test/Misc/diag-template-diffing.cpp
cfe/trunk/test/SemaCXX/overload-call.cpp
cfe/trunk/test/SemaCXX/overload-member-call.cpp
cfe/trunk/www/cxx_dr_status.html

Modified: cfe/trunk/include/clang/Sema/Overload.h
URL: 
http://llvm.org/viewvc/llvm-project/cfe/trunk/include/clang/Sema/Overload.h?rev=290808=290807=290808=diff
==
--- cfe/trunk/include/clang/Sema/Overload.h (original)
+++ cfe/trunk/include/clang/Sema/Overload.h Sun Jan  1 20:42:17 2017
@@ -531,6 +531,13 @@ namespace clang {
   Ambiguous.construct();
 }
 
+void setAsIdentityConversion(QualType T) {
+  setStandard();
+  Standard.setAsIdentityConversion();
+  Standard.setFromType(T);
+  Standard.setAllToTypes(T);
+}
+
 /// \brief Whether the target is really a std::initializer_list, and the
 /// sequence only represents the worst element conversion.
 bool isStdInitializerListElement() const {
@@ -603,6 +610,11 @@ namespace clang {
 ovl_fail_ext_disabled,
   };
 
+  /// A list of implicit conversion sequences for the arguments of an
+  /// OverloadCandidate.
+  typedef llvm::MutableArrayRef
+  ConversionSequenceList;
+
   /// OverloadCandidate - A single candidate in an overload set (C++ 13.3).
   struct OverloadCandidate {
 /// Function - The actual function that this candidate
@@ -627,18 +639,13 @@ namespace clang {
 /// is a surrogate, but only if IsSurrogate is true.
 CXXConversionDecl *Surrogate;
 
-/// Conversions - The conversion sequences used to convert the
-/// function arguments to the function parameters, the pointer points to a
-/// fixed size array with NumConversions elements. The memory is owned by
-/// the OverloadCandidateSet.
-ImplicitConversionSequence *Conversions;
+/// The conversion sequences used to convert the function arguments
+/// to the function parameters.
+ConversionSequenceList Conversions;
 
 /// The FixIt hints which can be used to fix the Bad candidate.
 ConversionFixItGenerator Fix;
 
-/// NumConversions - The number of elements in the Conversions array.
-unsigned NumConversions;
-
 /// Viable - True to indicate that this overload candidate is viable.
 bool Viable;
 
@@ -677,9 +684,9 @@ namespace clang {
 /// hasAmbiguousConversion - Returns whether this overload
 /// candidate requires an ambiguous conversion or not.
 bool hasAmbiguousConversion() const {
-  for (unsigned i = 0, e = NumConversions; i != e; ++i) {
-if (!Conversions[i].isInitialized()) return false;
-if (Conversions[i].isAmbiguous()) return true;
+  for (auto  : Conversions) {
+if (!C.isInitialized()) return false;
+if (C.isAmbiguous()) return true;
   }
   return false;
 }
@@ -728,7 +735,7 @@ namespace clang {
 SmallVector Candidates;
 llvm::SmallPtrSet Functions;
 
-// Allocator for OverloadCandidate::Conversions. We store the first few
+// Allocator for ConversionSequenceLists. We store the first few
 // elements inline to avoid allocation for small sets.
 llvm::BumpPtrAllocator ConversionSequenceAllocator;
 
@@ -769,30 +776,45 @@ namespace clang {
 size_t size() const { return Candidates.size(); }
 bool empty() const { return Candidates.empty(); }
 
-/// \brief Add a new candidate with NumConversions conversion sequence 
slots
-/// to the overload set.
-OverloadCandidate (unsigned NumConversions = 0) {
-  Candidates.push_back(OverloadCandidate());
-  OverloadCandidate  = Candidates.back();
+/// \brief Allocate storage for conversion sequences for NumConversions
+/// conversions.
+ConversionSequenceList
+allocateConversionSequences(unsigned NumConversions) {
+  ImplicitConversionSequence *Conversions;
 
   // Assign space from the inline array if there are enough free slots
   // available.
   if (NumConversions + NumInlineSequences <= 16) {
 ImplicitConversionSequence *I =
 (ImplicitConversionSequence *)InlineSpace.buffer;
-C.Conversions =

Re: r290792 - [c++17] Implement P0522R0 as written. This allows a template template argument

2017-01-01 Thread Richard Smith via cfe-commits

Thanks both of you, r290807.

On 1 January 2017 at 07:28, Erik Pilkington 
wrote:

>
> > On Dec 31, 2016, at 4:41 PM, Richard Smith via cfe-commits <
> cfe-commits@lists.llvm.org> wrote:
> >
> > Author: rsmith
> > Date: Sat Dec 31 15:41:23 2016
> > New Revision: 290792
> >
> > URL: http://llvm.org/viewvc/llvm-project?rev=290792=rev
> > Log:
> > [c++17] Implement P0522R0 as written. This allows a template template
> argument
> > to be specified for a template template parameter whenever the parameter
> is at
> > least as specialized as the argument (when there's an obvious and correct
> > mapping from uses of the parameter to uses of the argument). For
> example, a
> > template with more parameters can be passed to a template template
> parameter
> > with fewer, if those trailing parameters have default arguments.
> >
> > This is disabled by default, despite being a DR resolution, as it's
> fairly
> > broken in its current state: there are no partial ordering rules to cope
> with
> > template template parameters that have different parameter lists,
> meaning that
> > code that attempts to decompose template-ids based on arity can hit
> unavoidable
> > ambiguity issues.
> >
> > The diagnostics produced on a non-matching argument are also pretty bad
> right
> > now, but I aim to improve them in a subsequent commit.
> >
> > Added:
> >cfe/trunk/test/SemaTemplate/temp_arg_template_cxx1z.cpp
> > Modified:
> >cfe/trunk/include/clang/Basic/LangOptions.def
> >cfe/trunk/include/clang/Driver/Options.td
> >cfe/trunk/include/clang/Sema/Sema.h
> >cfe/trunk/lib/Driver/Tools.cpp
> >cfe/trunk/lib/Frontend/CompilerInvocation.cpp
> >cfe/trunk/lib/Sema/SemaTemplate.cpp
> >cfe/trunk/lib/Sema/SemaTemplateDeduction.cpp
> >cfe/trunk/test/SemaTemplate/temp_arg_template.cpp
> >cfe/trunk/www/cxx_status.html
> >
> > Modified: cfe/trunk/include/clang/Basic/LangOptions.def
> > URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/include/
> clang/Basic/LangOptions.def?rev=290792=290791=290792=diff
> > 
> ==
> > --- cfe/trunk/include/clang/Basic/LangOptions.def (original)
> > +++ cfe/trunk/include/clang/Basic/LangOptions.def Sat Dec 31 15:41:23
> 2016
> > @@ -134,6 +134,7 @@ LANGOPT(NoBuiltin , 1, 0, "disab
> > LANGOPT(NoMathBuiltin , 1, 0, "disable math builtin functions")
> > LANGOPT(GNUAsm, 1, 1, "GNU-style inline assembly")
> > LANGOPT(CoroutinesTS  , 1, 0, "C++ coroutines TS")
> > +LANGOPT(RelaxedTemplateTemplateArgs, 1, 0, "C++17 relaxed matching of
> tempalte template arguments")
> >
> > BENIGN_LANGOPT(ThreadsafeStatics , 1, 1, "thread-safe static
> initializers")
> > LANGOPT(POSIXThreads  , 1, 0, "POSIX thread support")
> >
> > Modified: cfe/trunk/include/clang/Driver/Options.td
> > URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/include/
> clang/Driver/Options.td?rev=290792=290791=290792=diff
> > 
> ==
> > --- cfe/trunk/include/clang/Driver/Options.td (original)
> > +++ cfe/trunk/include/clang/Driver/Options.td Sat Dec 31 15:41:23 2016
> > @@ -1088,6 +1088,11 @@ def fapplication_extension : Flag<["-"],
> >   HelpText<"Restrict code to those available for App Extensions">;
> > def fno_application_extension : Flag<["-"], "fno-application-extension">,
> >   Group;
> > +def frelaxed_template_template_args : Flag<["-"],
> "frelaxed-template-template-args">,
> > +  Flags<[CC1Option]>, HelpText<"Enable C++17 relaxed template template
> argument matching">,
> > +  Group;
> > +def fno_relaxed_template_template_args : Flag<["-"],
> "fno-relaxed-template-template-args">,
> > +  Group;
> > def fsized_deallocation : Flag<["-"], "fsized-deallocation">,
> Flags<[CC1Option]>,
> >   HelpText<"Enable C++14 sized global deallocation functions">,
> Group;
> > def fno_sized_deallocation: Flag<["-"], "fno-sized-deallocation">,
> Group;
> >
> > Modified: cfe/trunk/include/clang/Sema/Sema.h
> > URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/include/
> clang/Sema/Sema.h?rev=290792=290791=290792=diff
> > 
> ==
> > --- cfe/trunk/include/clang/Sema/Sema.h (original)
> > +++ cfe/trunk/include/clang/Sema/Sema.h Sat Dec 31 15:41:23 2016
> > @@ -6719,6 +6719,9 @@ public:
> >   bool isMoreSpecializedThanPrimary(VarTemplatePartialSpecializationDecl
> *T,
> > sema::TemplateDeductionInfo );
> >
> > +  bool isTemplateTemplateParameterAtLeastAsSpecializedAs(
> > +  TemplateParameterList *P, TemplateDecl *AArg, SourceLocation Loc);
> > +
> >   void MarkUsedTemplateParameters(const TemplateArgumentList
> ,
> >   bool OnlyDeduced,
> >   unsigned Depth,
> >
> > Modified: cfe/trunk/lib/Driver/Tools.cpp
> > URL:

r290807 - Address post-commit review comments.

2017-01-01 Thread Richard Smith via cfe-commits

Author: rsmith
Date: Sun Jan  1 20:38:22 2017
New Revision: 290807

URL: http://llvm.org/viewvc/llvm-project?rev=290807=rev
Log:
Address post-commit review comments.

Modified:
cfe/trunk/include/clang/Basic/LangOptions.def
cfe/trunk/lib/Sema/SemaTemplateDeduction.cpp

Modified: cfe/trunk/include/clang/Basic/LangOptions.def
URL: 
http://llvm.org/viewvc/llvm-project/cfe/trunk/include/clang/Basic/LangOptions.def?rev=290807=290806=290807=diff
==
--- cfe/trunk/include/clang/Basic/LangOptions.def (original)
+++ cfe/trunk/include/clang/Basic/LangOptions.def Sun Jan  1 20:38:22 2017
@@ -134,7 +134,7 @@ LANGOPT(NoBuiltin , 1, 0, "disab
 LANGOPT(NoMathBuiltin , 1, 0, "disable math builtin functions")
 LANGOPT(GNUAsm, 1, 1, "GNU-style inline assembly")
 LANGOPT(CoroutinesTS  , 1, 0, "C++ coroutines TS")
-LANGOPT(RelaxedTemplateTemplateArgs, 1, 0, "C++17 relaxed matching of tempalte 
template arguments")
+LANGOPT(RelaxedTemplateTemplateArgs, 1, 0, "C++17 relaxed matching of template 
template arguments")
 
 BENIGN_LANGOPT(ThreadsafeStatics , 1, 1, "thread-safe static initializers")
 LANGOPT(POSIXThreads  , 1, 0, "POSIX thread support")

Modified: cfe/trunk/lib/Sema/SemaTemplateDeduction.cpp
URL: 
http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Sema/SemaTemplateDeduction.cpp?rev=290807=290806=290807=diff
==
--- cfe/trunk/lib/Sema/SemaTemplateDeduction.cpp (original)
+++ cfe/trunk/lib/Sema/SemaTemplateDeduction.cpp Sun Jan  1 20:38:22 2017
@@ -4786,9 +4786,6 @@ bool Sema::isTemplateTemplateParameterAt
   QualType AType = Context.getTemplateSpecializationType(X, AArgs);
   QualType PType = Context.getTemplateSpecializationType(X, PArgs);
 
-  SmallVector Deduced;
-  Deduced.resize(A->size());
-
   //   ... the function template corresponding to P is at least as specialized
   //   as the function template corresponding to A according to the partial
   //   ordering rules for function templates.


___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D28212: typeinfo: provide a partial implementation for Win32

2017-01-01 Thread David Majnemer via Phabricator via cfe-commits

majnemer added inline comments.



Comment at: src/typeinfo.cpp:28-29
+#if defined(_WIN64)
+  static constexpr const size_t fnv_offset_basis = 14695981039346656037;
+  static constexpr const size_t fnv_prime = 10995116282110;
+#else

majnemer wrote:
> Why make these static? Seems strange to use that storage duration.
These literals are ill-formed, I think you need a ULL suffix here.



Comment at: src/typeinfo.cpp:28-32
+  static constexpr const size_t fnv_offset_basis = 14695981039346656037;
+  static constexpr const size_t fnv_prime = 10995116282110;
+#else
+  static constexpr const size_t fnv_offset_basis = 2166136261;
+  static constexpr const size_t fnv_prime = 16777619;

Why make these static? Seems strange to use that storage duration.


Repository:
  rL LLVM

https://reviews.llvm.org/D28212



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D28213: [Frontend] Correct values of ATOMIC_*_LOCK_FREE to match builtin

2017-01-01 Thread Michał Górny via Phabricator via cfe-commits

mgorny created this revision.
mgorny added reviewers: rsmith, t.p.northover, hfinkel, EricWF.
mgorny added a subscriber: cfe-commits.

Correct the logic used to set `ATOMIC_*_LOCK_FREE` preprocessor macros not
to rely on the ABI alignment of types. Instead, just assume all those
types are aligned correctly by default since clang uses safe alignment
for `_Atomic` types even if the underlying types are aligned to a lower
boundary by default.

For example, the `long long` and `double` types on x86 are aligned to
32-bit boundary by default. However, `_Atomic long long` and `_Atomic
double` are aligned to 64-bit boundary, therefore satisfying
the requirements of lock-free atomic operations.

This fixes PR #19355 by correcting the value of
`__GCC_ATOMIC_LLONG_LOCK_FREE` on x86, and therefore also fixing
the assumption made in libc++ tests. This also fixes PR #30581 by
applying a consistent logic between the functions used to implement
both interfaces.


https://reviews.llvm.org/D28213

Files:
  lib/Frontend/InitPreprocessor.cpp
  test/Sema/atomic-ops.c


Index: test/Sema/atomic-ops.c
===
--- test/Sema/atomic-ops.c
+++ test/Sema/atomic-ops.c
@@ -14,11 +14,7 @@
 _Static_assert(__GCC_ATOMIC_SHORT_LOCK_FREE == 2, "");
 _Static_assert(__GCC_ATOMIC_INT_LOCK_FREE == 2, "");
 _Static_assert(__GCC_ATOMIC_LONG_LOCK_FREE == 2, "");
-#ifdef __i386__
-_Static_assert(__GCC_ATOMIC_LLONG_LOCK_FREE == 1, "");
-#else
 _Static_assert(__GCC_ATOMIC_LLONG_LOCK_FREE == 2, "");
-#endif
 _Static_assert(__GCC_ATOMIC_POINTER_LOCK_FREE == 2, "");
 
 _Static_assert(__c11_atomic_is_lock_free(1), "");
Index: lib/Frontend/InitPreprocessor.cpp
===
--- lib/Frontend/InitPreprocessor.cpp
+++ lib/Frontend/InitPreprocessor.cpp
@@ -286,12 +286,12 @@
 
 /// Get the value the ATOMIC_*_LOCK_FREE macro should have for a type with
 /// the specified properties.
-static const char *getLockFreeValue(unsigned TypeWidth, unsigned TypeAlign,
-unsigned InlineWidth) {
+static const char *getLockFreeValue(unsigned TypeWidth, unsigned InlineWidth) {
   // Fully-aligned, power-of-2 sizes no larger than the inline
   // width will be inlined as lock-free operations.
-  if (TypeWidth == TypeAlign && (TypeWidth & (TypeWidth - 1)) == 0 &&
-  TypeWidth <= InlineWidth)
+  // Note: we do not need to check alignment since _Atomic(T) is always
+  // appropriately-aligned in clang.
+  if ((TypeWidth & (TypeWidth - 1)) == 0 && TypeWidth <= InlineWidth)
 return "2"; // "always lock free"
   // We cannot be certain what operations the lib calls might be
   // able to implement as lock-free on future processors.
@@ -881,7 +881,6 @@
 #define DEFINE_LOCK_FREE_MACRO(TYPE, Type) \
 Builder.defineMacro("__GCC_ATOMIC_" #TYPE "_LOCK_FREE", \
 getLockFreeValue(TI.get##Type##Width(), \
- TI.get##Type##Align(), \
  InlineWidthBits));
 DEFINE_LOCK_FREE_MACRO(BOOL, Bool);
 DEFINE_LOCK_FREE_MACRO(CHAR, Char);
@@ -894,7 +893,6 @@
 DEFINE_LOCK_FREE_MACRO(LLONG, LongLong);
 Builder.defineMacro("__GCC_ATOMIC_POINTER_LOCK_FREE",
 getLockFreeValue(TI.getPointerWidth(0),
- TI.getPointerAlign(0),
  InlineWidthBits));
 #undef DEFINE_LOCK_FREE_MACRO
   }


Index: test/Sema/atomic-ops.c
===
--- test/Sema/atomic-ops.c
+++ test/Sema/atomic-ops.c
@@ -14,11 +14,7 @@
 _Static_assert(__GCC_ATOMIC_SHORT_LOCK_FREE == 2, "");
 _Static_assert(__GCC_ATOMIC_INT_LOCK_FREE == 2, "");
 _Static_assert(__GCC_ATOMIC_LONG_LOCK_FREE == 2, "");
-#ifdef __i386__
-_Static_assert(__GCC_ATOMIC_LLONG_LOCK_FREE == 1, "");
-#else
 _Static_assert(__GCC_ATOMIC_LLONG_LOCK_FREE == 2, "");
-#endif
 _Static_assert(__GCC_ATOMIC_POINTER_LOCK_FREE == 2, "");
 
 _Static_assert(__c11_atomic_is_lock_free(1), "");
Index: lib/Frontend/InitPreprocessor.cpp
===
--- lib/Frontend/InitPreprocessor.cpp
+++ lib/Frontend/InitPreprocessor.cpp
@@ -286,12 +286,12 @@
 
 /// Get the value the ATOMIC_*_LOCK_FREE macro should have for a type with
 /// the specified properties.
-static const char *getLockFreeValue(unsigned TypeWidth, unsigned TypeAlign,
-unsigned InlineWidth) {
+static const char *getLockFreeValue(unsigned TypeWidth, unsigned InlineWidth) {
   // Fully-aligned, power-of-2 sizes no larger than the inline
   // width will be inlined as lock-free operations.
-  if (TypeWidth == TypeAlign && (TypeWidth & (TypeWidth - 1)) == 0 &&
-  TypeWidth <= InlineWidth)
+  // Note: we do not need to check alignment since _Atomic(T) is always
+  // appropriately-aligned in clang.
+  if ((TypeWidth &

[PATCH] D28212: typeinfo: provide a partial implementation for Win32

2017-01-01 Thread Saleem Abdulrasool via Phabricator via cfe-commits

compnerd created this revision.
compnerd added reviewers: EricWF, mclow.lists, majnemer, rnk.
compnerd added subscribers: smeenai, kastiglione, cfe-commits.
compnerd set the repository for this revision to rL LLVM.

The RTTI structure is different on Windows when building under MS ABI.
Update the definition to reflect this.  The structure itself contains an
area for caching the undecorated name (which is 0-initialized).  The
decorated name has a bitfield followed by the linkage name.  When
std::type_info::name is invoked for the first time, the runtime should
undecorate the name, cache it, and return the undecorated name.  This
requires access to an implementation of `__unDName`.  For now, return
the raw name.

This uses the fnv-1a hash to hash the name of the RTTI.  We could use an 
alternate hash (murmur?  city?), but, this was the quickest to throw together.


Repository:
  rL LLVM

https://reviews.llvm.org/D28212

Files:
  include/typeinfo
  src/typeinfo.cpp

Index: src/typeinfo.cpp
===
--- src/typeinfo.cpp
+++ src/typeinfo.cpp
@@ -15,6 +15,44 @@
 
 #include "typeinfo"
 
+#if defined(_WIN32)
+#include 
+
+const char *std::type_info::name(std::type_info::data *__data) {
+  // TODO(compnerd) cache demangled &__data.__decorated_name[1]
+  return &__data->__decorated_name[1];
+}
+
+size_t std::type_info::hash(const std::type_info::data *__data) {
+#if defined(_WIN64)
+  static constexpr const size_t fnv_offset_basis = 14695981039346656037;
+  static constexpr const size_t fnv_prime = 10995116282110;
+#else
+  static constexpr const size_t fnv_offset_basis = 2166136261;
+  static constexpr const size_t fnv_prime = 16777619;
+#endif
+
+  size_t value = fnv_offset_basis;
+  for (const char *c = &__data->__decorated_name[1]; *c; ++c) {
+value ^= static_cast(static_cast(*c));
+value *= fnv_prime;
+  }
+
+#if defined(_WIN64)
+  value ^= value >> 32;
+#endif
+
+  return value;
+}
+
+int std::type_info::compare(const std::type_info::data *__lhs,
+const std::type_info::data *__rhs) {
+  if (__lhs == __rhs)
+return 0;
+  return strcmp(&__lhs->__decorated_name[1], &__rhs->__decorated_name[1]);
+}
+#endif
+
 #if !defined(LIBCXXRT) && !defined(_LIBCPPABI_VERSION)
 
 std::type_info::~type_info()
Index: include/typeinfo
===
--- include/typeinfo
+++ include/typeinfo
@@ -72,6 +72,49 @@
 namespace std  // purposefully not using versioning namespace
 {
 
+#if defined(_WIN32)
+class _LIBCPP_EXCEPTION_ABI type_info
+{
+  type_info & operator=(const type_info &) _LIBCPP_EQUAL_DELETE;
+  type_info (const type_info &) _LIBCPP_EQUAL_DELETE;
+
+  mutable struct data {
+const char *__undecorated_name;
+const char __decorated_name[1];
+  } __data;
+
+  static const char *name(type_info::data *__data);
+  static size_t hash(const type_info::data *__data);
+  static int compare(const type_info::data *__l, const type_info::data *__r);
+
+public:
+  virtual ~type_info() _NOEXCEPT;
+
+  _LIBCPP_INLINE_VISIBILITY
+  const char *name() const _NOEXCEPT {
+return name(&__data);
+  }
+
+  _LIBCPP_INLINE_VISIBILITY
+  bool before(const type_info &__arg) const _NOEXCEPT {
+return compare(&__data, &__arg.__data) < 0;
+  }
+
+  _LIBCPP_INLINE_VISIBILITY
+  size_t hash_code() const _NOEXCEPT {
+return hash(&__data);
+  }
+
+  _LIBCPP_INLINE_VISIBILITY
+  bool operator==(const type_info &__arg) const _NOEXCEPT {
+return compare(&__data, &__arg.__data) == 0;
+  }
+
+  bool operator!=(const type_info &__arg) const _NOEXCEPT {
+return compare(&__data, &__arg.__data) != 0;
+  }
+};
+#else
 class _LIBCPP_EXCEPTION_ABI type_info
 {
 type_info& operator=(const type_info&);
@@ -148,6 +191,7 @@
 {return __builtin_strcmp(name(), __arg.name());}
 #endif
 };
+#endif
 
 class _LIBCPP_EXCEPTION_ABI bad_cast
 : public exception
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Re: [libcxx] r290806 - chrono: address post commit comments from Howard

2017-01-01 Thread Howard Hinnant via cfe-commits

Very close!  Comment below:

On Jan 1, 2017, at 5:04 PM, Saleem Abdulrasool via cfe-commits 
 wrote:
> 
> Author: compnerd
> Date: Sun Jan  1 16:04:38 2017
> New Revision: 290806
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=290806=rev
> Log:
> chrono: address post commit comments from Howard
> 
> Drawing some inspiration from code from Bill O'Neal as pointed out by
> Howard, rework the code to avoid an overflow in the duration.  Adjust
> the style to match libc++ style as well.
> 
> Create a local typedef for the FILETIME duration (100-ns units).  Use
> this to define the difference between the NT and the UNIX epochs (which
> previously overflowed due to the representation limits due to the
> bouncing to ns).  Return the FILETIME duration biased by the NT-to-UNIX
> epoch conversion.
> 
> Use of the custom duration makes it easier to read and reason about the
> code.
> 
> Modified:
>libcxx/trunk/src/chrono.cpp
> 
> Modified: libcxx/trunk/src/chrono.cpp
> URL: 
> http://llvm.org/viewvc/llvm-project/libcxx/trunk/src/chrono.cpp?rev=290806=290805=290806=diff
> ==
> --- libcxx/trunk/src/chrono.cpp (original)
> +++ libcxx/trunk/src/chrono.cpp Sun Jan  1 16:04:38 2017
> @@ -46,26 +46,30 @@ system_clock::time_point
> system_clock::now() _NOEXCEPT
> {
> #if defined(_WIN32)
> -  // The Windows epoch is Jan 1 1601, the Unix epoch Jan 1 1970.  The 
> difference
> -  // in nanoseconds is the windows epoch offset.
> -  static const constexpr __int64 kWindowsEpochOffset = 0x19db1ded53e8000;
> +  // FILETIME is in 100ns units
> +  using filetime_duration =
> +  _VSTD::chrono::duration<__int64,
> +  _VSTD::ratio_multiply<_VSTD::ratio<100, 1>,
> +nanoseconds::period>>;
> 
> -  FILETIME ftSystemTime;
> +  // The Windows epoch is Jan 1 1601, the Unix epoch Jan 1 1970.
> +  static _LIBCPP_CONSTEXPR const filetime_duration
> +  nt_to_unix_epoch{11644473600};

If you give nt_to_unix_epoch units of seconds, instead of filetime_duration, 
then I _believe_ this will be correct.  But I have not tested it.

I like this direction much better as all of your computations have units now.

Howard

> +
> +  FILETIME ft;
> #if _WIN32_WINNT >= _WIN32_WINNT_WIN8
> #if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
> -  GetSystemTimePreciseAsFileTime();
> +  GetSystemTimePreciseAsFileTime();
> #else
> -  GetSystemTimeAsFileTime();
> +  GetSystemTimeAsFileTime();
> #endif
> #else
> -  GetSystemTimeAsFileTime();
> +  GetSystemTimeAsFileTime();
> #endif
> -  __int64 llWinTimeNS =
> -  ((static_cast<__int64>(ftSystemTime.dwHighDateTime) << 32) |
> -   static_cast<__int64>(ftSystemTime.dwLowDateTime)) *
> -  100;
> -  return time_point(duration_cast(
> -  (nanoseconds(llWinTimeNS - kWindowsEpochOffset;
> +
> +  filetime_duration d{(static_cast<__int64>(ft.dwHighDateTime) << 32) |
> +   static_cast<__int64>(ft.dwLowDateTime)};
> +  return time_point(duration_cast(d - nt_to_unix_epoch));
> #else
> #ifdef CLOCK_REALTIME
> struct timespec tp;
> 
> 
> ___
> cfe-commits mailing list
> cfe-commits@lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D28211: typeinfo: provide a destructor for type_info

2017-01-01 Thread Saleem Abdulrasool via Phabricator via cfe-commits

compnerd created this revision.
compnerd added reviewers: EricWF, mclow.lists.
compnerd added subscribers: smeenai, cfe-commits.
compnerd set the repository for this revision to rL LLVM.

When building without libc++abi and libcxxrt, provide the definition for
the std::type_info destructor.


Repository:
  rL LLVM

https://reviews.llvm.org/D28211

Files:
  src/typeinfo.cpp


Index: src/typeinfo.cpp
===
--- src/typeinfo.cpp
+++ src/typeinfo.cpp
@@ -17,6 +17,10 @@

 #if !defined(LIBCXXRT) && !defined(_LIBCPPABI_VERSION)

+std::type_info::~type_info()
+{
+}
+
 std::bad_cast::bad_cast() _NOEXCEPT
 {
 }


Index: src/typeinfo.cpp
===
--- src/typeinfo.cpp
+++ src/typeinfo.cpp
@@ -17,6 +17,10 @@

 #if !defined(LIBCXXRT) && !defined(_LIBCPPABI_VERSION)

+std::type_info::~type_info()
+{
+}
+
 std::bad_cast::bad_cast() _NOEXCEPT
 {
 }
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D28210: system_error: inline error_category ctor on Win32

2017-01-01 Thread Saleem Abdulrasool via Phabricator via cfe-commits

compnerd created this revision.
compnerd added reviewers: EricWF, mclow.lists, majnemer.
compnerd added subscribers: smeenai, cfe-commits.
compnerd set the repository for this revision to rL LLVM.

Because the class is marked as `dllexport` and instantiated in
`algorithm.cpp` and is strongly defined in `system_error.cpp`, we end up
with multiple definitions.  By keeping the definition as an inline
definition, the `dllexport` will instantiate it.  Being inlined, they
will be marked as COMDAT(SELECT_ANY) and will still be exported from the
dll.


Repository:
  rL LLVM

https://reviews.llvm.org/D28210

Files:
  src/system_error.cpp


Index: src/system_error.cpp
===
--- src/system_error.cpp
+++ src/system_error.cpp
@@ -9,7 +9,9 @@

 #include "__config"

+#if !defined(_WIN32)
 #define _LIBCPP_BUILDING_SYSTEM_ERROR
+#endif
 #include "system_error"

 #include "include/config_elast.h"
@@ -29,9 +31,11 @@

 // class error_category

+#if !defined(_WIN32)
 error_category::error_category() _NOEXCEPT
 {
 }
+#endif

 error_category::~error_category() _NOEXCEPT
 {


Index: src/system_error.cpp
===
--- src/system_error.cpp
+++ src/system_error.cpp
@@ -9,7 +9,9 @@

 #include "__config"

+#if !defined(_WIN32)
 #define _LIBCPP_BUILDING_SYSTEM_ERROR
+#endif
 #include "system_error"

 #include "include/config_elast.h"
@@ -29,9 +31,11 @@

 // class error_category

+#if !defined(_WIN32)
 error_category::error_category() _NOEXCEPT
 {
 }
+#endif

 error_category::~error_category() _NOEXCEPT
 {
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Re: [libcxx] r290803 - chrono: implement a Windows version of system_clock::now

2017-01-01 Thread Saleem Abdulrasool via cfe-commits

Thanks for the review on this!  I hope you wont mind taking a look at SVN
r290806 as well which tries to address the comments.

On Sun, Jan 1, 2017 at 1:09 PM, Howard Hinnant 
wrote:

> On Jan 1, 2017, at 3:20 PM, Saleem Abdulrasool via cfe-commits <
> cfe-commits@lists.llvm.org> wrote:
> >
> > Author: compnerd
> > Date: Sun Jan  1 14:20:41 2017
> > New Revision: 290803
> >
> > URL: http://llvm.org/viewvc/llvm-project?rev=290803=rev
> > Log:
> > chrono: implement a Windows version of system_clock::now
> >
> > system_clock::now is not entirely straight forward on Windows, which
> > does not have a clock_gettime function.
> >
> > GetSystemTimeAsFileTime gives us the value relative to the NT epoch (Jan
> > 1 1601) rather than the Unix epoch (Jan 1 1970).  However, this function
> > has a low resolution (~10ms).  Newer versions of Windows provide
> > GetSystemTimePreciseAsFileTime which gives us a much more accurate time
> > (<1us).  Unfortunately, the latter is only available on Windows 8+ when
> > targeting desktop apps.
> >
> > Modified:
> >libcxx/trunk/src/chrono.cpp
> >
> > Modified: libcxx/trunk/src/chrono.cpp
> > URL: http://llvm.org/viewvc/llvm-project/libcxx/trunk/src/
> chrono.cpp?rev=290803=290802=290803=diff
> > 
> ==
> > --- libcxx/trunk/src/chrono.cpp (original)
> > +++ libcxx/trunk/src/chrono.cpp Sun Jan  1 14:20:41 2017
> > @@ -12,9 +12,18 @@
> > #include "system_error"  // __throw_system_error
> > #include // clock_gettime, CLOCK_MONOTONIC and
> CLOCK_REALTIME
> >
> > +#if defined(_WIN32)
> > +#define WIN32_LEAN_AND_MEAN
> > +#define VC_EXTRA_LEAN
> > +#include 
> > +#if _WIN32_WINNT >= _WIN32_WINNT_WIN8
> > +#include 
> > +#endif
> > +#else
> > #if !defined(CLOCK_REALTIME)
> > #include // for gettimeofday and timeval
> > #endif
> > +#endif
> >
> > #if !defined(_LIBCPP_HAS_NO_MONOTONIC_CLOCK) &&
> !defined(CLOCK_MONOTONIC)
> > #if __APPLE__
> > @@ -36,6 +45,28 @@ const bool system_clock::is_steady;
> > system_clock::time_point
> > system_clock::now() _NOEXCEPT
> > {
> > +#if defined(_WIN32)
> > +  // The Windows epoch is Jan 1 1601, the Unix epoch Jan 1 1970.  The
> difference
> > +  // in nanoseconds is the windows epoch offset.
> > +  static const constexpr __int64 kWindowsEpochOffset =
> 0x19db1ded53e8000;
>
> This is the correct difference, but the comment is wrong.  It isn’t
> nanoseconds.  It is 100ns units.
>
> Maybe I got the math wrong, but this was supposed to be ns, which is why I
bounced between 100ns and ns.


> > +
> > +  FILETIME ftSystemTime;
> > +#if _WIN32_WINNT >= _WIN32_WINNT_WIN8
> > +#if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
> > +  GetSystemTimePreciseAsFileTime();
> > +#else
> > +  GetSystemTimeAsFileTime();
> > +#endif
> > +#else
> > +  GetSystemTimeAsFileTime();
> > +#endif
> > +  __int64 llWinTimeNS =
> > +  ((static_cast<__int64>(ftSystemTime.dwHighDateTime) << 32) |
> > +   static_cast<__int64>(ftSystemTime.dwLowDateTime)) *
> > +  100;
>
> The * 100 will overflow.  The * 100 converts 100ns to ns.  ns has a signed
> 64bit range of +/- 292 years, which is greater than the difference between
> the two epochs.
>
>
This is what I get for trying to be clever and not use durations.


> > +  return time_point(duration_cast(
> > +  (nanoseconds(llWinTimeNS - kWindowsEpochOffset;
>
> It really isn’t desirable to bounce this off of nanoseconds because of the
> overflow.


Yeap;  This now becomes a duration_cast(filetime_duration) ...
letting durations figure out how to do the conversion from 100ns
(filetime_duration) to us (sytem_clock::duration) units.


> > +#else
> > #ifdef CLOCK_REALTIME
> > struct timespec tp;
> > if (0 != clock_gettime(CLOCK_REALTIME, ))
> > @@ -46,6 +77,7 @@ system_clock::now() _NOEXCEPT
> > gettimeofday(, 0);
> > return time_point(seconds(tv.tv_sec) + microseconds(tv.tv_usec));
> > #endif  // CLOCK_REALTIME
> > +#endif
> > }
>
> Thank you for preserving the 1970 epoch.  Though not currently specified
> by the standard, it is very important.  And hopefully will be specified by
> a future standard.
>
> Here is code written by Billy O’Neal (MS) doing the conversion you’re
> looking for:
>
> https://github.com/HowardHinnant/date/wiki/Examples-and-Recipes#FILETIME
>
> It has no intermediate results with precision finer than 100ns.  The range
> of the 64bit 100ns unit goes out to +/- 29 thousand years.


Thanks for the pointer!


>
> Howard
>
>


-- 
Saleem Abdulrasool
compnerd (at) compnerd (dot) org
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[libcxx] r290805 - chrono: address post-commit comments from majnemer

2017-01-01 Thread Saleem Abdulrasool via cfe-commits

Author: compnerd
Date: Sun Jan  1 16:04:36 2017
New Revision: 290805

URL: http://llvm.org/viewvc/llvm-project?rev=290805=rev
Log:
chrono: address post-commit comments from majnemer

Correct style to match libc++ style as pointed out by David Majnemer on
IRC.  NFC.

Modified:
libcxx/trunk/src/chrono.cpp

Modified: libcxx/trunk/src/chrono.cpp
URL: 
http://llvm.org/viewvc/llvm-project/libcxx/trunk/src/chrono.cpp?rev=290805=290804=290805=diff
==
--- libcxx/trunk/src/chrono.cpp (original)
+++ libcxx/trunk/src/chrono.cpp Sun Jan  1 16:04:36 2017
@@ -160,14 +160,14 @@ steady_clock::now() _NOEXCEPT
 steady_clock::time_point
 steady_clock::now() _NOEXCEPT
 {
-  static LARGE_INTEGER liFreq;
-  static BOOL bQPFRun = FALSE;
-  if (bQPFRun == FALSE)
-bQPFRun = QueryPerformanceFrequency();
+  static LARGE_INTEGER freq;
+  static BOOL initialized = FALSE;
+  if (!initialized)
+initialized = QueryPerformanceFrequency(); // always succceeds
 
-  LARGE_INTEGER liCntr;
-  QueryPerformanceCounter();
-  return time_point(duration(liCntr.QuadPart * nano::den / liFreq.QuadPart));
+  LARGE_INTEGER counter;
+  QueryPerformanceCounter();
+  return time_point(duration(counter.QuadPart * nano::den / freq.QuadPart));
 }
 
 #elif defined(CLOCK_MONOTONIC)


___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[libcxx] r290806 - chrono: address post commit comments from Howard

2017-01-01 Thread Saleem Abdulrasool via cfe-commits

Author: compnerd
Date: Sun Jan  1 16:04:38 2017
New Revision: 290806

URL: http://llvm.org/viewvc/llvm-project?rev=290806=rev
Log:
chrono: address post commit comments from Howard

Drawing some inspiration from code from Bill O'Neal as pointed out by
Howard, rework the code to avoid an overflow in the duration.  Adjust
the style to match libc++ style as well.

Create a local typedef for the FILETIME duration (100-ns units).  Use
this to define the difference between the NT and the UNIX epochs (which
previously overflowed due to the representation limits due to the
bouncing to ns).  Return the FILETIME duration biased by the NT-to-UNIX
epoch conversion.

Use of the custom duration makes it easier to read and reason about the
code.

Modified:
libcxx/trunk/src/chrono.cpp

Modified: libcxx/trunk/src/chrono.cpp
URL: 
http://llvm.org/viewvc/llvm-project/libcxx/trunk/src/chrono.cpp?rev=290806=290805=290806=diff
==
--- libcxx/trunk/src/chrono.cpp (original)
+++ libcxx/trunk/src/chrono.cpp Sun Jan  1 16:04:38 2017
@@ -46,26 +46,30 @@ system_clock::time_point
 system_clock::now() _NOEXCEPT
 {
 #if defined(_WIN32)
-  // The Windows epoch is Jan 1 1601, the Unix epoch Jan 1 1970.  The 
difference
-  // in nanoseconds is the windows epoch offset.
-  static const constexpr __int64 kWindowsEpochOffset = 0x19db1ded53e8000;
+  // FILETIME is in 100ns units
+  using filetime_duration =
+  _VSTD::chrono::duration<__int64,
+  _VSTD::ratio_multiply<_VSTD::ratio<100, 1>,
+nanoseconds::period>>;
 
-  FILETIME ftSystemTime;
+  // The Windows epoch is Jan 1 1601, the Unix epoch Jan 1 1970.
+  static _LIBCPP_CONSTEXPR const filetime_duration
+  nt_to_unix_epoch{11644473600};
+
+  FILETIME ft;
 #if _WIN32_WINNT >= _WIN32_WINNT_WIN8
 #if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
-  GetSystemTimePreciseAsFileTime();
+  GetSystemTimePreciseAsFileTime();
 #else
-  GetSystemTimeAsFileTime();
+  GetSystemTimeAsFileTime();
 #endif
 #else
-  GetSystemTimeAsFileTime();
+  GetSystemTimeAsFileTime();
 #endif
-  __int64 llWinTimeNS =
-  ((static_cast<__int64>(ftSystemTime.dwHighDateTime) << 32) |
-   static_cast<__int64>(ftSystemTime.dwLowDateTime)) *
-  100;
-  return time_point(duration_cast(
-  (nanoseconds(llWinTimeNS - kWindowsEpochOffset;
+
+  filetime_duration d{(static_cast<__int64>(ft.dwHighDateTime) << 32) |
+   static_cast<__int64>(ft.dwLowDateTime)};
+  return time_point(duration_cast(d - nt_to_unix_epoch));
 #else
 #ifdef CLOCK_REALTIME
 struct timespec tp;


___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Re: [libcxx] r290803 - chrono: implement a Windows version of system_clock::now

2017-01-01 Thread Howard Hinnant via cfe-commits

On Jan 1, 2017, at 3:20 PM, Saleem Abdulrasool via cfe-commits 
 wrote:
> 
> Author: compnerd
> Date: Sun Jan  1 14:20:41 2017
> New Revision: 290803
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=290803=rev
> Log:
> chrono: implement a Windows version of system_clock::now
> 
> system_clock::now is not entirely straight forward on Windows, which
> does not have a clock_gettime function.
> 
> GetSystemTimeAsFileTime gives us the value relative to the NT epoch (Jan
> 1 1601) rather than the Unix epoch (Jan 1 1970).  However, this function
> has a low resolution (~10ms).  Newer versions of Windows provide
> GetSystemTimePreciseAsFileTime which gives us a much more accurate time
> (<1us).  Unfortunately, the latter is only available on Windows 8+ when
> targeting desktop apps.
> 
> Modified:
>libcxx/trunk/src/chrono.cpp
> 
> Modified: libcxx/trunk/src/chrono.cpp
> URL: 
> http://llvm.org/viewvc/llvm-project/libcxx/trunk/src/chrono.cpp?rev=290803=290802=290803=diff
> ==
> --- libcxx/trunk/src/chrono.cpp (original)
> +++ libcxx/trunk/src/chrono.cpp Sun Jan  1 14:20:41 2017
> @@ -12,9 +12,18 @@
> #include "system_error"  // __throw_system_error
> #include // clock_gettime, CLOCK_MONOTONIC and CLOCK_REALTIME
> 
> +#if defined(_WIN32)
> +#define WIN32_LEAN_AND_MEAN
> +#define VC_EXTRA_LEAN
> +#include 
> +#if _WIN32_WINNT >= _WIN32_WINNT_WIN8
> +#include 
> +#endif
> +#else
> #if !defined(CLOCK_REALTIME)
> #include // for gettimeofday and timeval
> #endif
> +#endif
> 
> #if !defined(_LIBCPP_HAS_NO_MONOTONIC_CLOCK) && !defined(CLOCK_MONOTONIC)
> #if __APPLE__
> @@ -36,6 +45,28 @@ const bool system_clock::is_steady;
> system_clock::time_point
> system_clock::now() _NOEXCEPT
> {
> +#if defined(_WIN32)
> +  // The Windows epoch is Jan 1 1601, the Unix epoch Jan 1 1970.  The 
> difference
> +  // in nanoseconds is the windows epoch offset.
> +  static const constexpr __int64 kWindowsEpochOffset = 0x19db1ded53e8000;

This is the correct difference, but the comment is wrong.  It isn’t 
nanoseconds.  It is 100ns units.

> +
> +  FILETIME ftSystemTime;
> +#if _WIN32_WINNT >= _WIN32_WINNT_WIN8
> +#if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
> +  GetSystemTimePreciseAsFileTime();
> +#else
> +  GetSystemTimeAsFileTime();
> +#endif
> +#else
> +  GetSystemTimeAsFileTime();
> +#endif
> +  __int64 llWinTimeNS =
> +  ((static_cast<__int64>(ftSystemTime.dwHighDateTime) << 32) |
> +   static_cast<__int64>(ftSystemTime.dwLowDateTime)) *
> +  100;

The * 100 will overflow.  The * 100 converts 100ns to ns.  ns has a signed 
64bit range of +/- 292 years, which is greater than the difference between the 
two epochs.

> +  return time_point(duration_cast(
> +  (nanoseconds(llWinTimeNS - kWindowsEpochOffset;

It really isn’t desirable to bounce this off of nanoseconds because of the 
overflow.

> +#else
> #ifdef CLOCK_REALTIME
> struct timespec tp;
> if (0 != clock_gettime(CLOCK_REALTIME, ))
> @@ -46,6 +77,7 @@ system_clock::now() _NOEXCEPT
> gettimeofday(, 0);
> return time_point(seconds(tv.tv_sec) + microseconds(tv.tv_usec));
> #endif  // CLOCK_REALTIME
> +#endif
> }

Thank you for preserving the 1970 epoch.  Though not currently specified by the 
standard, it is very important.  And hopefully will be specified by a future 
standard.

Here is code written by Billy O’Neal (MS) doing the conversion you’re looking 
for:

https://github.com/HowardHinnant/date/wiki/Examples-and-Recipes#FILETIME

It has no intermediate results with precision finer than 100ns.  The range of 
the 64bit 100ns unit goes out to +/- 29 thousand years.

Howard

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[libcxx] r290803 - chrono: implement a Windows version of system_clock::now

2017-01-01 Thread Saleem Abdulrasool via cfe-commits

Author: compnerd
Date: Sun Jan  1 14:20:41 2017
New Revision: 290803

URL: http://llvm.org/viewvc/llvm-project?rev=290803=rev
Log:
chrono: implement a Windows version of system_clock::now

system_clock::now is not entirely straight forward on Windows, which
does not have a clock_gettime function.

GetSystemTimeAsFileTime gives us the value relative to the NT epoch (Jan
1 1601) rather than the Unix epoch (Jan 1 1970).  However, this function
has a low resolution (~10ms).  Newer versions of Windows provide
GetSystemTimePreciseAsFileTime which gives us a much more accurate time
(<1us).  Unfortunately, the latter is only available on Windows 8+ when
targeting desktop apps.

Modified:
libcxx/trunk/src/chrono.cpp

Modified: libcxx/trunk/src/chrono.cpp
URL: 
http://llvm.org/viewvc/llvm-project/libcxx/trunk/src/chrono.cpp?rev=290803=290802=290803=diff
==
--- libcxx/trunk/src/chrono.cpp (original)
+++ libcxx/trunk/src/chrono.cpp Sun Jan  1 14:20:41 2017
@@ -12,9 +12,18 @@
 #include "system_error"  // __throw_system_error
 #include // clock_gettime, CLOCK_MONOTONIC and CLOCK_REALTIME
 
+#if defined(_WIN32)
+#define WIN32_LEAN_AND_MEAN
+#define VC_EXTRA_LEAN
+#include 
+#if _WIN32_WINNT >= _WIN32_WINNT_WIN8
+#include 
+#endif
+#else
 #if !defined(CLOCK_REALTIME)
 #include // for gettimeofday and timeval
 #endif
+#endif
 
 #if !defined(_LIBCPP_HAS_NO_MONOTONIC_CLOCK) && !defined(CLOCK_MONOTONIC)
 #if __APPLE__
@@ -36,6 +45,28 @@ const bool system_clock::is_steady;
 system_clock::time_point
 system_clock::now() _NOEXCEPT
 {
+#if defined(_WIN32)
+  // The Windows epoch is Jan 1 1601, the Unix epoch Jan 1 1970.  The 
difference
+  // in nanoseconds is the windows epoch offset.
+  static const constexpr __int64 kWindowsEpochOffset = 0x19db1ded53e8000;
+
+  FILETIME ftSystemTime;
+#if _WIN32_WINNT >= _WIN32_WINNT_WIN8
+#if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
+  GetSystemTimePreciseAsFileTime();
+#else
+  GetSystemTimeAsFileTime();
+#endif
+#else
+  GetSystemTimeAsFileTime();
+#endif
+  __int64 llWinTimeNS =
+  ((static_cast<__int64>(ftSystemTime.dwHighDateTime) << 32) |
+   static_cast<__int64>(ftSystemTime.dwLowDateTime)) *
+  100;
+  return time_point(duration_cast(
+  (nanoseconds(llWinTimeNS - kWindowsEpochOffset;
+#else
 #ifdef CLOCK_REALTIME
 struct timespec tp;
 if (0 != clock_gettime(CLOCK_REALTIME, ))
@@ -46,6 +77,7 @@ system_clock::now() _NOEXCEPT
 gettimeofday(, 0);
 return time_point(seconds(tv.tv_sec) + microseconds(tv.tv_usec));
 #endif  // CLOCK_REALTIME
+#endif
 }
 
 time_t


___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[libcxx] r290801 - build: dont detect libraries for Windows

2017-01-01 Thread Saleem Abdulrasool via cfe-commits

Author: compnerd
Date: Sun Jan  1 14:20:38 2017
New Revision: 290801

URL: http://llvm.org/viewvc/llvm-project?rev=290801=rev
Log:
build: dont detect libraries for Windows

Hard code the defaults for Windows for the time being.  The checks
really are always going to return the same value.  Technically, the
pthread linkage is possible, however, it seems better to use the Win32
threading along with the external threading support that we have added.

Modified:
libcxx/trunk/cmake/config-ix.cmake

Modified: libcxx/trunk/cmake/config-ix.cmake
URL: 
http://llvm.org/viewvc/llvm-project/libcxx/trunk/cmake/config-ix.cmake?rev=290801=290800=290801=diff
==
--- libcxx/trunk/cmake/config-ix.cmake (original)
+++ libcxx/trunk/cmake/config-ix.cmake Sun Jan  1 14:20:38 2017
@@ -1,9 +1,20 @@
 include(CheckLibraryExists)
 include(CheckCXXCompilerFlag)
 
-check_library_exists(c fopen "" LIBCXX_HAS_C_LIB)
+if(WIN32 AND NOT MINGW)
+  # NOTE(compnerd) this is technically a lie, there is msvcrt, but for now, 
lets
+  # let the default linking take care of that.
+  set(LIBCXX_HAS_C_LIB NO)
+else()
+  check_library_exists(c fopen "" LIBCXX_HAS_C_LIB)
+endif()
+
 if (NOT LIBCXX_USE_COMPILER_RT)
-  check_library_exists(gcc_s __gcc_personality_v0 "" LIBCXX_HAS_GCC_S_LIB)
+  if(WIN32 AND NOT MINGW)
+set(LIBCXX_HAS_GCC_S_LIB NO)
+  else()
+check_library_exists(gcc_s __gcc_personality_v0 "" LIBCXX_HAS_GCC_S_LIB)
+  endif()
 endif()
 
 # libc++ is built with -nodefaultlibs, so we want all our checks to also
@@ -32,7 +43,9 @@ if (LIBCXX_SUPPORTS_NODEFAULTLIBS_FLAG)
   endif ()
 endif ()
 
-include(CheckLibcxxAtomic)
+if(NOT WIN32 OR MINGW)
+  include(CheckLibcxxAtomic)
+endif()
 
 # Check compiler flags
 
@@ -45,6 +58,14 @@ check_cxx_compiler_flag(/GR-
 
 
 # Check libraries
-check_library_exists(pthread pthread_create "" LIBCXX_HAS_PTHREAD_LIB)
-check_library_exists(m ccos "" LIBCXX_HAS_M_LIB)
-check_library_exists(rt clock_gettime "" LIBCXX_HAS_RT_LIB)
+if(WIN32 AND NOT MINGW)
+  # TODO(compnerd) do we want to support an emulation layer that allows for the
+  # use of pthread-win32 or similar libraries to emulate pthreads on Windows?
+  set(LIBCXX_HAS_PTHREAD_LIB NO)
+  set(LIBCXX_HAS_M_LIB NO)
+  set(LIBCXX_HAS_RT_LIB NO)
+else()
+  check_library_exists(pthread pthread_create "" LIBCXX_HAS_PTHREAD_LIB)
+  check_library_exists(m ccos "" LIBCXX_HAS_M_LIB)
+  check_library_exists(rt clock_gettime "" LIBCXX_HAS_RT_LIB)
+endif()


___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[libcxx] r290800 - build: differentiate between building for and on Windows

2017-01-01 Thread Saleem Abdulrasool via cfe-commits

Author: compnerd
Date: Sun Jan  1 14:20:36 2017
New Revision: 290800

URL: http://llvm.org/viewvc/llvm-project?rev=290800=rev
Log:
build: differentiate between building for and on Windows

This is necessary to support cross-compiling a Windows libc++ from
Linux.  The CMAKE_SYSTEM_HOST_NAME tells you what, in autotools
parlance, is known as the build as opposed to WIN32 which maps to, in
autotools parlance, host.

Modified:
libcxx/trunk/include/CMakeLists.txt

Modified: libcxx/trunk/include/CMakeLists.txt
URL: 
http://llvm.org/viewvc/llvm-project/libcxx/trunk/include/CMakeLists.txt?rev=290800=290799=290800=diff
==
--- libcxx/trunk/include/CMakeLists.txt (original)
+++ libcxx/trunk/include/CMakeLists.txt Sun Jan  1 14:20:36 2017
@@ -29,7 +29,7 @@ if (LIBCXX_INSTALL_HEADERS)
 
   if (LIBCXX_NEEDS_SITE_CONFIG)
 set(UNIX_CAT cat)
-if (WIN32)
+if("${CMAKE_HOST_SYSTEM_NAME}" STREQUAL "Windows")
   set(UNIX_CAT type)
 endif()
 # Generate and install a custom __config header. The new header is created


___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[libcxx] r290804 - chrono: give Windows a steady_clock

2017-01-01 Thread Saleem Abdulrasool via cfe-commits

Author: compnerd
Date: Sun Jan  1 14:20:43 2017
New Revision: 290804

URL: http://llvm.org/viewvc/llvm-project?rev=290804=rev
Log:
chrono: give Windows a steady_clock

Provide a definition for a steady monotonic clock by wrapping
QueryPerformanceCounter.

Modified:
libcxx/trunk/src/chrono.cpp

Modified: libcxx/trunk/src/chrono.cpp
URL: 
http://llvm.org/viewvc/llvm-project/libcxx/trunk/src/chrono.cpp?rev=290804=290803=290804=diff
==
--- libcxx/trunk/src/chrono.cpp (original)
+++ libcxx/trunk/src/chrono.cpp Sun Jan  1 14:20:43 2017
@@ -25,10 +25,10 @@
 #endif
 #endif
 
-#if !defined(_LIBCPP_HAS_NO_MONOTONIC_CLOCK) && !defined(CLOCK_MONOTONIC)
+#if !defined(_LIBCPP_HAS_NO_MONOTONIC_CLOCK)
 #if __APPLE__
 #include   // mach_absolute_time, mach_timebase_info_data_t
-#else
+#elif !defined(_WIN32) && !defined(CLOCK_MONOTONIC)
 #error "Monotonic clock not implemented"
 #endif
 #endif
@@ -101,18 +101,7 @@ system_clock::from_time_t(time_t t) _NOE
 
 const bool steady_clock::is_steady;
 
-#ifdef CLOCK_MONOTONIC
-
-steady_clock::time_point
-steady_clock::now() _NOEXCEPT
-{
-struct timespec tp;
-if (0 != clock_gettime(CLOCK_MONOTONIC, ))
-__throw_system_error(errno, "clock_gettime(CLOCK_MONOTONIC) failed");
-return time_point(seconds(tp.tv_sec) + nanoseconds(tp.tv_nsec));
-}
-
-#elif defined(__APPLE__)
+#if defined(__APPLE__)
 
 //   mach_absolute_time() * MachInfo.numer / MachInfo.denom is the number of
 //   nanoseconds since the computer booted up.  MachInfo.numer and 
MachInfo.denom
@@ -166,6 +155,32 @@ steady_clock::now() _NOEXCEPT
 return time_point(duration(fp()));
 }
 
+#elif defined(_WIN32)
+
+steady_clock::time_point
+steady_clock::now() _NOEXCEPT
+{
+  static LARGE_INTEGER liFreq;
+  static BOOL bQPFRun = FALSE;
+  if (bQPFRun == FALSE)
+bQPFRun = QueryPerformanceFrequency();
+
+  LARGE_INTEGER liCntr;
+  QueryPerformanceCounter();
+  return time_point(duration(liCntr.QuadPart * nano::den / liFreq.QuadPart));
+}
+
+#elif defined(CLOCK_MONOTONIC)
+
+steady_clock::time_point
+steady_clock::now() _NOEXCEPT
+{
+struct timespec tp;
+if (0 != clock_gettime(CLOCK_MONOTONIC, ))
+__throw_system_error(errno, "clock_gettime(CLOCK_MONOTONIC) failed");
+return time_point(seconds(tp.tv_sec) + nanoseconds(tp.tv_nsec));
+}
+
 #else
 #error "Monotonic clock not implemented"
 #endif


___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[libcxx] r290802 - build: further improve flags handling for cl

2017-01-01 Thread Saleem Abdulrasool via cfe-commits

Author: compnerd
Date: Sun Jan  1 14:20:40 2017
New Revision: 290802

URL: http://llvm.org/viewvc/llvm-project?rev=290802=rev
Log:
build: further improve flags handling for cl

This allows us to build with cl (or rather clang-cl) by using the
correct spelling for `-include` (`/FI` for cl).  clang-cl and cl default
to C++11/C++14 as they support it rather than permitting an explicit
language standard.

Modified:
libcxx/trunk/CMakeLists.txt

Modified: libcxx/trunk/CMakeLists.txt
URL: 
http://llvm.org/viewvc/llvm-project/libcxx/trunk/CMakeLists.txt?rev=290802=290801=290802=diff
==
--- libcxx/trunk/CMakeLists.txt (original)
+++ libcxx/trunk/CMakeLists.txt Sun Jan  1 14:20:40 2017
@@ -354,8 +354,10 @@ if (LIBCXX_HAS_MUSL_LIBC)
 endif()
 add_compile_flags_if_supported(-std=${LIBCXX_STANDARD_VER})
 mangle_name("LIBCXX_SUPPORTS_STD_EQ_${LIBCXX_STANDARD_VER}_FLAG" 
SUPPORTS_DIALECT_NAME)
-if (NOT MSVC AND NOT ${SUPPORTS_DIALECT_NAME})
-  message(FATAL_ERROR "C++11 or greater is required but the compiler does not 
support ${LIBCXX_STANDARD_VER}")
+if(NOT ${SUPPORTS_DIALECT_NAME})
+  if(NOT "${CMAKE_CXX_COMPILER_ID}" STREQUAL "MSVC" AND NOT 
"${CMAKE_CXX_SIMULATE_ID}" STREQUAL "MSVC")
+message(FATAL_ERROR "C++11 or greater is required but the compiler does 
not support ${LIBCXX_STANDARD_VER}")
+  endif()
 endif()
 
 # On all systems the system c++ standard library headers need to be excluded.
@@ -515,13 +517,17 @@ if (DEFINED WIN32 AND LIBCXX_ENABLE_STAT
 endif()
 
 if (LIBCXX_NEEDS_SITE_CONFIG)
-  configure_file(
-include/__config_site.in
-${LIBCXX_BINARY_DIR}/__config_site
-@ONLY)
+  configure_file("include/__config_site.in"
+ "${LIBCXX_BINARY_DIR}/__config_site"
+ @ONLY)
+
   # Provide the config definitions by included the generated __config_site
   # file at compile time.
-  add_compile_flags("-include ${LIBCXX_BINARY_DIR}/__config_site")
+  if("${CMAKE_CXX_COMPILER_ID}" STREQUAL "MSVC" OR "${CMAKE_CXX_SIMULATE_ID}" 
STREQUAL "MSVC")
+add_compile_flags("/FI\"${LIBCXX_BINARY_DIR}/__config_site\"")
+  else()
+add_compile_flags("-include ${LIBCXX_BINARY_DIR}/__config_site")
+  endif()
 endif()
 
 
#===


___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

r290799 - CodeGen: update comment about RTTI field

2017-01-01 Thread Saleem Abdulrasool via cfe-commits

Author: compnerd
Date: Sun Jan  1 13:16:02 2017
New Revision: 290799

URL: http://llvm.org/viewvc/llvm-project?rev=290799=rev
Log:
CodeGen: update comment about RTTI field

The MS ABI RTTI has a reserved field which is used as a cache for the
demangled name.  It must be zero-initialized, which is used as a hint by
the runtime to say that the cache has not been populated.  Since this
field is populated at runtime, the RTTI structures must be placed in the
.data section rather than .rdata.  NFC

Modified:
cfe/trunk/lib/CodeGen/MicrosoftCXXABI.cpp

Modified: cfe/trunk/lib/CodeGen/MicrosoftCXXABI.cpp
URL: 
http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/MicrosoftCXXABI.cpp?rev=290799=290798=290799=diff
==
--- cfe/trunk/lib/CodeGen/MicrosoftCXXABI.cpp (original)
+++ cfe/trunk/lib/CodeGen/MicrosoftCXXABI.cpp Sun Jan  1 13:16:02 2017
@@ -3536,7 +3536,7 @@ llvm::GlobalVariable *MSRTTIBuilder::get
 
   // Initialize the base class ClassHierarchyDescriptor.
   llvm::Constant *Fields[] = {
-  llvm::ConstantInt::get(CGM.IntTy, 0), // Unknown
+  llvm::ConstantInt::get(CGM.IntTy, 0), // reserved by the runtime
   llvm::ConstantInt::get(CGM.IntTy, Flags),
   llvm::ConstantInt::get(CGM.IntTy, Classes.size()),
   
ABI.getImageRelativeConstant(llvm::ConstantExpr::getInBoundsGetElementPtr(


___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D28081: Make GetStyle return Expected instead of FormatStyle

2017-01-01 Thread Antonio Maiorano via Phabricator via cfe-commits

amaiorano added inline comments.



Comment at: lib/Format/Format.cpp:1900
 
-  if (!getPredefinedStyle(FallbackStyle, Style.Language, )) {
-llvm::errs() << "Invalid fallback style \"" << FallbackStyle
- << "\" using LLVM style\n";
-return Style;
-  }
+  FormatStyle FallbackStyle = getNoStyle();
+  if (!getPredefinedStyle(FallbackStyleName, Style.Language, ))

amaiorano wrote:
> Going over my diff, I realize that I didn't revert this local FormatStyle 
> instance to make use of the top-level declared "FormatStyle Style" local 
> variable. I will revert this one too.
I wrote the above comment before realizing there was a bug when fallback style 
is "none". As I explained in my long comment, we may actually want to keep this 
as a separate variable to fix the bug and ensure fallback style is only ever 
returned when no config is found, which is exactly what my change currently 
does.


https://reviews.llvm.org/D28081



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Re: r290792 - [c++17] Implement P0522R0 as written. This allows a template template argument

2017-01-01 Thread Erik Pilkington via cfe-commits


> On Dec 31, 2016, at 4:41 PM, Richard Smith via cfe-commits 
>  wrote:
> 
> Author: rsmith
> Date: Sat Dec 31 15:41:23 2016
> New Revision: 290792
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=290792=rev
> Log:
> [c++17] Implement P0522R0 as written. This allows a template template argument
> to be specified for a template template parameter whenever the parameter is at
> least as specialized as the argument (when there's an obvious and correct
> mapping from uses of the parameter to uses of the argument). For example, a
> template with more parameters can be passed to a template template parameter
> with fewer, if those trailing parameters have default arguments.
> 
> This is disabled by default, despite being a DR resolution, as it's fairly
> broken in its current state: there are no partial ordering rules to cope with
> template template parameters that have different parameter lists, meaning that
> code that attempts to decompose template-ids based on arity can hit 
> unavoidable
> ambiguity issues.
> 
> The diagnostics produced on a non-matching argument are also pretty bad right
> now, but I aim to improve them in a subsequent commit.
> 
> Added:
>cfe/trunk/test/SemaTemplate/temp_arg_template_cxx1z.cpp
> Modified:
>cfe/trunk/include/clang/Basic/LangOptions.def
>cfe/trunk/include/clang/Driver/Options.td
>cfe/trunk/include/clang/Sema/Sema.h
>cfe/trunk/lib/Driver/Tools.cpp
>cfe/trunk/lib/Frontend/CompilerInvocation.cpp
>cfe/trunk/lib/Sema/SemaTemplate.cpp
>cfe/trunk/lib/Sema/SemaTemplateDeduction.cpp
>cfe/trunk/test/SemaTemplate/temp_arg_template.cpp
>cfe/trunk/www/cxx_status.html
> 
> Modified: cfe/trunk/include/clang/Basic/LangOptions.def
> URL: 
> http://llvm.org/viewvc/llvm-project/cfe/trunk/include/clang/Basic/LangOptions.def?rev=290792=290791=290792=diff
> ==
> --- cfe/trunk/include/clang/Basic/LangOptions.def (original)
> +++ cfe/trunk/include/clang/Basic/LangOptions.def Sat Dec 31 15:41:23 2016
> @@ -134,6 +134,7 @@ LANGOPT(NoBuiltin , 1, 0, "disab
> LANGOPT(NoMathBuiltin , 1, 0, "disable math builtin functions")
> LANGOPT(GNUAsm, 1, 1, "GNU-style inline assembly")
> LANGOPT(CoroutinesTS  , 1, 0, "C++ coroutines TS")
> +LANGOPT(RelaxedTemplateTemplateArgs, 1, 0, "C++17 relaxed matching of 
> tempalte template arguments")
> 
> BENIGN_LANGOPT(ThreadsafeStatics , 1, 1, "thread-safe static initializers")
> LANGOPT(POSIXThreads  , 1, 0, "POSIX thread support")
> 
> Modified: cfe/trunk/include/clang/Driver/Options.td
> URL: 
> http://llvm.org/viewvc/llvm-project/cfe/trunk/include/clang/Driver/Options.td?rev=290792=290791=290792=diff
> ==
> --- cfe/trunk/include/clang/Driver/Options.td (original)
> +++ cfe/trunk/include/clang/Driver/Options.td Sat Dec 31 15:41:23 2016
> @@ -1088,6 +1088,11 @@ def fapplication_extension : Flag<["-"],
>   HelpText<"Restrict code to those available for App Extensions">;
> def fno_application_extension : Flag<["-"], "fno-application-extension">,
>   Group;
> +def frelaxed_template_template_args : Flag<["-"], 
> "frelaxed-template-template-args">,
> +  Flags<[CC1Option]>, HelpText<"Enable C++17 relaxed template template 
> argument matching">,
> +  Group;
> +def fno_relaxed_template_template_args : Flag<["-"], 
> "fno-relaxed-template-template-args">,
> +  Group;
> def fsized_deallocation : Flag<["-"], "fsized-deallocation">, 
> Flags<[CC1Option]>,
>   HelpText<"Enable C++14 sized global deallocation functions">, 
> Group;
> def fno_sized_deallocation: Flag<["-"], "fno-sized-deallocation">, 
> Group;
> 
> Modified: cfe/trunk/include/clang/Sema/Sema.h
> URL: 
> http://llvm.org/viewvc/llvm-project/cfe/trunk/include/clang/Sema/Sema.h?rev=290792=290791=290792=diff
> ==
> --- cfe/trunk/include/clang/Sema/Sema.h (original)
> +++ cfe/trunk/include/clang/Sema/Sema.h Sat Dec 31 15:41:23 2016
> @@ -6719,6 +6719,9 @@ public:
>   bool isMoreSpecializedThanPrimary(VarTemplatePartialSpecializationDecl *T,
> sema::TemplateDeductionInfo );
> 
> +  bool isTemplateTemplateParameterAtLeastAsSpecializedAs(
> +  TemplateParameterList *P, TemplateDecl *AArg, SourceLocation Loc);
> +
>   void MarkUsedTemplateParameters(const TemplateArgumentList ,
>   bool OnlyDeduced,
>   unsigned Depth,
> 
> Modified: cfe/trunk/lib/Driver/Tools.cpp
> URL: 
> http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Driver/Tools.cpp?rev=290792=290791=290792=diff
> ==
> --- cfe/trunk/lib/Driver/Tools.cpp (original)
> +++ cfe/trunk/lib/Driver/Tools.cpp Sat Dec 31 15:41:23 2016
> @@ -6020,6 +6020,13 @@ void

[PATCH] D28208: Remove isIgnored()-test that is more expensive than the analysis behind it

2017-01-01 Thread Daniel Jasper via Phabricator via cfe-commits

djasper created this revision.
djasper added a reviewer: rsmith.
djasper added a subscriber: cfe-commits.

In many translation units I have tried, the calls to isIgnored() removed in 
this patch are more expensive than doing the analysis that is behind it. The 
speed-up in translation units I have tried is between 10 and 20%.


https://reviews.llvm.org/D28208

Files:
  lib/Sema/SemaTemplate.cpp


Index: lib/Sema/SemaTemplate.cpp
===
--- lib/Sema/SemaTemplate.cpp
+++ lib/Sema/SemaTemplate.cpp
@@ -4244,7 +4244,7 @@
 UnnamedLocalNoLinkageFinder(Sema , SourceRange SR) : S(S), SR(SR) { }
 
 bool Visit(QualType T) {
-  return inherited::Visit(T.getTypePtr());
+  return T.isNull() ? false : inherited::Visit(T.getTypePtr());
 }
 
 #define TYPE(Class, Parent) \
@@ -4497,17 +4497,7 @@
   //
   // C++11 allows these, and even in C++03 we allow them as an extension with
   // a warning.
-  bool NeedsCheck;
-  if (LangOpts.CPlusPlus11)
-NeedsCheck =
-!Diags.isIgnored(diag::warn_cxx98_compat_template_arg_unnamed_type,
- SR.getBegin()) ||
-!Diags.isIgnored(diag::warn_cxx98_compat_template_arg_local_type,
- SR.getBegin());
-  else
-NeedsCheck = Arg->hasUnnamedOrLocalType();
-
-  if (NeedsCheck) {
+  if (LangOpts.CPlusPlus11 || Arg->hasUnnamedOrLocalType()) {
 UnnamedLocalNoLinkageFinder Finder(*this, SR);
 (void)Finder.Visit(Context.getCanonicalType(Arg));
   }


Index: lib/Sema/SemaTemplate.cpp
===
--- lib/Sema/SemaTemplate.cpp
+++ lib/Sema/SemaTemplate.cpp
@@ -4244,7 +4244,7 @@
 UnnamedLocalNoLinkageFinder(Sema , SourceRange SR) : S(S), SR(SR) { }
 
 bool Visit(QualType T) {
-  return inherited::Visit(T.getTypePtr());
+  return T.isNull() ? false : inherited::Visit(T.getTypePtr());
 }
 
 #define TYPE(Class, Parent) \
@@ -4497,17 +4497,7 @@
   //
   // C++11 allows these, and even in C++03 we allow them as an extension with
   // a warning.
-  bool NeedsCheck;
-  if (LangOpts.CPlusPlus11)
-NeedsCheck =
-!Diags.isIgnored(diag::warn_cxx98_compat_template_arg_unnamed_type,
- SR.getBegin()) ||
-!Diags.isIgnored(diag::warn_cxx98_compat_template_arg_local_type,
- SR.getBegin());
-  else
-NeedsCheck = Arg->hasUnnamedOrLocalType();
-
-  if (NeedsCheck) {
+  if (LangOpts.CPlusPlus11 || Arg->hasUnnamedOrLocalType()) {
 UnnamedLocalNoLinkageFinder Finder(*this, SR);
 (void)Finder.Visit(Context.getCanonicalType(Arg));
   }
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D28207: Add second fast path for DiagnosticsEngine::GetDiagStatePointForLoc

2017-01-01 Thread Daniel Jasper via Phabricator via cfe-commits

djasper created this revision.
djasper added a reviewer: rsmith.
djasper added a subscriber: cfe-commits.
Herald added a subscriber: sanjoy.

In many translation units I have tested, many of the calls to 
DiagnosticsEngine::GetDiagStatePointForLoc are for source locations before the 
first diag state point. AFAICT, this happens frequently for locations in the 
STL and other base libraries. This patch adds a fast path for this case to 
avoid the costly binary search. This binary search is costly, because the 
relatively slow SourceManager::isBeforeInTranslationUnit() is used to compare 
DiagStatePoints.


https://reviews.llvm.org/D28207

Files:
  lib/Basic/Diagnostic.cpp


Index: lib/Basic/Diagnostic.cpp
===
--- lib/Basic/Diagnostic.cpp
+++ lib/Basic/Diagnostic.cpp
@@ -163,21 +163,28 @@
   assert(DiagStatePoints.front().Loc.isInvalid() &&
  "Should have created a DiagStatePoint for command-line");
 
-  if (!SourceMgr)
+  if (!SourceMgr || DiagStatePoints.size() == 1)
 return DiagStatePoints.end() - 1;
 
   FullSourceLoc Loc(L, *SourceMgr);
   if (Loc.isInvalid())
 return DiagStatePoints.end() - 1;
 
-  DiagStatePointsTy::iterator Pos = DiagStatePoints.end();
+  // Most frequent case: L is after the last diag state change.
   FullSourceLoc LastStateChangePos = DiagStatePoints.back().Loc;
-  if (LastStateChangePos.isValid() &&
-  Loc.isBeforeInTranslationUnitThan(LastStateChangePos))
-Pos = std::upper_bound(DiagStatePoints.begin(), DiagStatePoints.end(),
-   DiagStatePoint(nullptr, Loc));
-  --Pos;
-  return Pos;
+  assert(LastStateChangePos.isValid());
+  if (!Loc.isBeforeInTranslationUnitThan(LastStateChangePos))
+return DiagStatePoints.end() - 1;
+
+  // 2nd most frequent case: L is before the first diag state change.
+  FullSourceLoc FirstStateChangePos = DiagStatePoints[1].Loc;
+  assert(FirstStateChangePos.isValid());
+  if (Loc.isBeforeInTranslationUnitThan(FirstStateChangePos))
+return DiagStatePoints.begin();
+
+  // Fall back to binary search.
+  return std::upper_bound(DiagStatePoints.begin(), DiagStatePoints.end(),
+  DiagStatePoint(nullptr, Loc)) - 1;
 }
 
 void DiagnosticsEngine::setSeverity(diag::kind Diag, diag::Severity Map,


Index: lib/Basic/Diagnostic.cpp
===
--- lib/Basic/Diagnostic.cpp
+++ lib/Basic/Diagnostic.cpp
@@ -163,21 +163,28 @@
   assert(DiagStatePoints.front().Loc.isInvalid() &&
  "Should have created a DiagStatePoint for command-line");
 
-  if (!SourceMgr)
+  if (!SourceMgr || DiagStatePoints.size() == 1)
 return DiagStatePoints.end() - 1;
 
   FullSourceLoc Loc(L, *SourceMgr);
   if (Loc.isInvalid())
 return DiagStatePoints.end() - 1;
 
-  DiagStatePointsTy::iterator Pos = DiagStatePoints.end();
+  // Most frequent case: L is after the last diag state change.
   FullSourceLoc LastStateChangePos = DiagStatePoints.back().Loc;
-  if (LastStateChangePos.isValid() &&
-  Loc.isBeforeInTranslationUnitThan(LastStateChangePos))
-Pos = std::upper_bound(DiagStatePoints.begin(), DiagStatePoints.end(),
-   DiagStatePoint(nullptr, Loc));
-  --Pos;
-  return Pos;
+  assert(LastStateChangePos.isValid());
+  if (!Loc.isBeforeInTranslationUnitThan(LastStateChangePos))
+return DiagStatePoints.end() - 1;
+
+  // 2nd most frequent case: L is before the first diag state change.
+  FullSourceLoc FirstStateChangePos = DiagStatePoints[1].Loc;
+  assert(FirstStateChangePos.isValid());
+  if (Loc.isBeforeInTranslationUnitThan(FirstStateChangePos))
+return DiagStatePoints.begin();
+
+  // Fall back to binary search.
+  return std::upper_bound(DiagStatePoints.begin(), DiagStatePoints.end(),
+  DiagStatePoint(nullptr, Loc)) - 1;
 }
 
 void DiagnosticsEngine::setSeverity(diag::kind Diag, diag::Severity Map,
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

r290808 - DR1391: Check for implicit conversion sequences for non-dependent function

Re: r290792 - [c++17] Implement P0522R0 as written. This allows a template template argument

r290807 - Address post-commit review comments.

[PATCH] D28212: typeinfo: provide a partial implementation for Win32

[PATCH] D28213: [Frontend] Correct values of ATOMIC_*_LOCK_FREE to match builtin

[PATCH] D28212: typeinfo: provide a partial implementation for Win32

Re: [libcxx] r290806 - chrono: address post commit comments from Howard

[PATCH] D28211: typeinfo: provide a destructor for type_info

[PATCH] D28210: system_error: inline error_category ctor on Win32

Re: [libcxx] r290803 - chrono: implement a Windows version of system_clock::now

[libcxx] r290805 - chrono: address post-commit comments from majnemer

[libcxx] r290806 - chrono: address post commit comments from Howard

Re: [libcxx] r290803 - chrono: implement a Windows version of system_clock::now

[libcxx] r290803 - chrono: implement a Windows version of system_clock::now

[libcxx] r290801 - build: dont detect libraries for Windows

[libcxx] r290800 - build: differentiate between building for and on Windows

[libcxx] r290804 - chrono: give Windows a steady_clock

[libcxx] r290802 - build: further improve flags handling for cl

r290799 - CodeGen: update comment about RTTI field

[PATCH] D28081: Make GetStyle return Expected instead of FormatStyle

Re: r290792 - [c++17] Implement P0522R0 as written. This allows a template template argument

[PATCH] D28208: Remove isIgnored()-test that is more expensive than the analysis behind it

[PATCH] D28207: Add second fast path for DiagnosticsEngine::GetDiagStatePointForLoc

23 matches

Site Navigation

Mail list logo

Footer information