Export of internal Abseil changes.

--
15d7bcf28220750db46930f4d8c090b54e3ae5fe by Jon Cohen <cohenjon@google.com>:

Fix miscellaneous CMake change interleaving issues for the daily release:
  * add back the absl::container target
  * Add copts to absl_cc_library targets in absl/container/CMakeLists.txt
  * Add trailing newline to the end of AbseilConfigureCopts.cmake

PiperOrigin-RevId: 223057096

--
baac35470d75b6561477f688dc4eb021f604cf71 by Abseil Team <absl-team@google.com>:

Internal Cleanup.

PiperOrigin-RevId: 223051579

--
6791c2f2e35b030b5579f36d3c607c6ba92fa089 by Abseil Team <absl-team@google.com>:

Internal Change.

PiperOrigin-RevId: 223046855

--
5467ad987ea82aef77d2f1cc85aa9105e7d9c320 by Samuel Benzaquen <sbenza@google.com>:

Workaround for gcc bug https://gcc.gnu.org/PR88115

PiperOrigin-RevId: 223041901

--
36fa5cfd41df2b71d26487c45363901bbf6a2463 by Tom Manshreck <shreck@google.com>:

Clarify visit() constraints

PiperOrigin-RevId: 223032194

--
afdf4013de036b411db7f92cde8a2493e6665223 by Abseil Team <absl-team@google.com>:

Fix comment typos.

PiperOrigin-RevId: 223024090

--
e11c01927eb8b898f6633282824022104b258342 by Jon Cohen <cohenjon@google.com>:

Make absl::spinlock_test_common TESTONLY

This should fix https://github.com/abseil/abseil-cpp/issues/221

PiperOrigin-RevId: 222885323

--
5ccc576d1c68e4b92705aa8064f1e8d715e5415e by Abseil Team <absl-team@google.com>:

Internal change.

PiperOrigin-RevId: 222877017

--
96ff25bf78c4f4bca0d6e61faa4feeab91a2e73c by Jon Cohen <cohenjon@google.com>:

Align CMake and Bazel compile options.  This is the first step towards a single source of truth for Abseil compile options.  Also makes absl_test and absl_cc_test make binaries and targets with compatible names to each other to make testing easier.

PiperOrigin-RevId: 222858408

--
7dd3e2618ad5a5de5d918fc73e438ef0b98cec6a by Abseil Team <absl-team@google.com>:

Revert "absl: cap SpinLock backoff to 4ms"

PiperOrigin-RevId: 222656230

--
0d49538a3cab714156ed0a5651656c0aa098a1e5 by Abseil Team <absl-team@google.com>:

Update absl/container/CMakeLists.txt to use new functions
i.e. absl_cc_(library|test)

PiperOrigin-RevId: 222535766

--
92744e9d0e5c3bf9e1167a7bdf1a6777192531b1 by Abseil Team <absl-team@google.com>:

Disable header parsing for broken targets

PiperOrigin-RevId: 222257218

--
39a6c623601c44e02d91e412f126a813d719507b by Abseil Team <absl-team@google.com>:

absl: cap SpinLock backoff to 4ms

The current backoff logic has 3 problems:
1. It can produce too high values (up to 256ms), which can negatively
affect tail latency. The value was chosen long time ago and now it's
a good idea to reconsider it.
2. It does not have low bound, so on any iteration it can produce
a very small value that will lead to unnecessary cpu consumption.
3. It does not increase low bound with the number of iterations.
So if the SpinLock is actually somehow locked for a very prolonged time,
a waiter can still wake periodically.

Rework the logic to solve these problems.
Add lower bound of 128us, no code should rely on absence of episodic
delays in this range as they can occur everywhere.
Lower upper bound to 4ms. A thread sleeping for 4ms does not consume
significant cpu time (see below).
Grow lower bound with the number of iterations.

This is cpu consumption of a process doing usleep(x) in a loop
(sampled with ps):

    64us -> 4.0%
   128us -> 2.7%
   256us -> 3.5%
   512us -> 2.8%
  1024us -> 1.6%
  2048us -> 0.6%
  4096us -> 0.3%
  8192us -> 0.0%

Few millisecond sleeps do not consume significant time.

PiperOrigin-RevId: 222196086

--
17104a2396ddda61fb0faed0a72ff8c161ca17ea by Shahriar Rouf <nafi@google.com>:

Add benchmarks for hashing civil_times.

PiperOrigin-RevId: 222152108
GitOrigin-RevId: 15d7bcf28220750db46930f4d8c090b54e3ae5fe
Change-Id: I73b929feaf6ce72b70fdafd6108f53bbbeaf9738

1 files changed, 11 insertions, 7 deletions

diff --git a/absl/container/internal/layout.h b/absl/container/internal/layout.h
index 676c7d67ee68..a9c1244a4e79 100644
--- a/absl/container/internal/layout.h
+++ b/absl/container/internal/layout.h
@@ -232,13 +232,17 @@ struct SizeOf : NotAligned<T>, std::integral_constant<size_t, sizeof(T)> {};
 template <class T, size_t N>
 struct SizeOf<Aligned<T, N>> : std::integral_constant<size_t, sizeof(T)> {};
 
+// Note: workaround for https://gcc.gnu.org/PR88115
 template <class T>
-struct AlignOf : NotAligned<T>, std::integral_constant<size_t, alignof(T)> {};
+struct AlignOf : NotAligned<T> {
+  static constexpr size_t value = alignof(T);
+};
 
 template <class T, size_t N>
-struct AlignOf<Aligned<T, N>> : std::integral_constant<size_t, N> {
+struct AlignOf<Aligned<T, N>> {
   static_assert(N % alignof(T) == 0,
                 "Custom alignment can't be lower than the type's alignment");
+  static constexpr size_t value = N;
 };
 
 // Does `Ts...` contain `T`?
@@ -290,7 +294,7 @@ std::string TypeName() {
 #ifdef ABSL_INTERNAL_HAS_CXA_DEMANGLE
   demangled = abi::__cxa_demangle(typeid(T).name(), nullptr, nullptr, &status);
 #endif
-  if (status == 0 && demangled != nullptr) {  // Demangling succeeeded.
+  if (status == 0 && demangled != nullptr) {  // Demangling succeeded.
     absl::StrAppend(&out, "<", demangled, ">");
     free(demangled);
   } else {
@@ -396,7 +400,7 @@ class LayoutImpl<std::tuple<Elements...>, absl::index_sequence<SizeSeq...>,
     static_assert(N < NumOffsets, "Index out of bounds");
     return adl_barrier::Align(
         Offset<N - 1>() + SizeOf<ElementType<N - 1>>() * size_[N - 1],
-        ElementAlignment<N>());
+        ElementAlignment<N>::value);
   }
 
   // Offset in bytes of the array with the specified element type. There must
@@ -445,7 +449,7 @@ class LayoutImpl<std::tuple<Elements...>, absl::index_sequence<SizeSeq...>,
     return Size<ElementIndex<T>()>();
   }
 
-    // The number of elements of all arrays for which they are known.
+  // The number of elements of all arrays for which they are known.
   constexpr std::array<size_t, NumSizes> Sizes() const {
     return {{Size<SizeSeq>()...}};
   }
@@ -610,7 +614,7 @@ class LayoutImpl<std::tuple<Elements...>, absl::index_sequence<SizeSeq...>,
 #ifdef ADDRESS_SANITIZER
     PoisonPadding<Char, N - 1>(p);
     // The `if` is an optimization. It doesn't affect the observable behaviour.
-    if (ElementAlignment<N - 1>() % ElementAlignment<N>()) {
+    if (ElementAlignment<N - 1>::value % ElementAlignment<N>::value) {
       size_t start =
           Offset<N - 1>() + SizeOf<ElementType<N - 1>>() * size_[N - 1];
       ASAN_POISON_MEMORY_REGION(p + start, Offset<N>() - start);
@@ -690,7 +694,7 @@ class Layout : public internal_layout::LayoutType<sizeof...(Ts), Ts...> {
   //
   // It's allowed to pass fewer array sizes than the number of arrays. E.g.,
   // if all you need is to the offset of the second array, you only need to
-  // pass one argument -- the number of elements in the first arrays.
+  // pass one argument -- the number of elements in the first array.
   //
   //   // int[3] followed by 4 bytes of padding and an unknown number of
   //   // doubles.