PiperOrigin-RevId: 608947694
Change-Id: Ie53a91c4d78dcb80c57227616b488ec64b23c588

PiperOrigin-RevId: 608770171
Change-Id: Icca54086037e42826c272f04374aeb33d060ace5

Motivation: the implementation becomes more complicated with small object optimization.
PiperOrigin-RevId: 608742838
Change-Id: I55fc42321b1967f9c7bbee49817a2f2d4ee44b56

PiperOrigin-RevId: 608705933
Change-Id: I9c11d8d8b3d9057bdf215863210f450332cf56cb

PiperOrigin-RevId: 608661989
Change-Id: Ibfd94f8b2d23fd232bf93904ed68e11a400b3644

This often indicates a bug from adding synchronization logic but not using it.

PiperOrigin-RevId: 608600311
Change-Id: I48bc3ae18b1e9a00b5e2152f33b486cd6be324a1

Imported from GitHub PR https://github.com/abseil/abseil-cpp/pull/1625

Trivial relocatability also requires that the type not do anything weird with its assignment operator; update the type-trait to reflect this. (This is the definition used by BSL, Folly, HPX, Thrust, Parlay, Amadeus, and P1144.)

This is important if we want to use `absl::is_trivially_relocatable` as a gate for memcpy optimizations in `inlined_vector::erase` and/or `inlined_vector::swap`, because in those cases relocation is used to replace part of a sequence involving assignment; the optimization requires an assignment operator that behaves value-semantically.

Clang's builtin currently fails to check the assignment operator, so we stop using it entirely for now. We already refused to use it on Win32, Win64, and Apple, for various unrelated reasons. I'm working on giving Clang's builtin the behavior that would let us re-enable it here.

Assume that any compiler providing both `__cpp_impl_trivially_relocatable` and a builtin `__is_trivially_relocatable(T)` will use the appropriate (P1144) definition for its builtin. Right now there's only one such compiler (the P1144 reference implementation, which forks Clang), so this is largely a moot point, but I'm being optimistic.
Merge d943abdbabc1b7080aa5f0a2fff3e724135164dc into 34604d5b

Merging this change closes #1625

COPYBARA_INTEGRATE_REVIEW=https://github.com/abseil/abseil-cpp/pull/1625 from Quuxplusone:trivially-relocatable d943abdbabc1b7080aa5f0a2fff3e724135164dc
PiperOrigin-RevId: 607977323
Change-Id: I6436a60326c6d1064bdd71ec2e15b86b7a29efd4

It can be used instead of `MaskEmptyOrDeleted` in case of inserting to empty table.
`MaskNonFull` requires less operations, in particular it eliminates `_mm_set1_epi8` and `_mm_cmpgt_epi8` operations.

PiperOrigin-RevId: 607587394
Change-Id: Ia48f922d1ca6de38cc91e7ab0d608c45f8f2c446

These were described in https://github.com/itanium-cxx-abi/cxx-abi/pull/85 and implemented by LLVM.

PiperOrigin-RevId: 607555558
Change-Id: I9991ac88c1fcf63b25b93d93977a83ca343cdb5d

Imported from GitHub PR https://github.com/abseil/abseil-cpp/pull/1617

The intrinsics used aren't available on `x86_64` processors while running in 32-bit mode. See:

- list of 64-bit intrinsics (https://learn.microsoft.com/en-us/cpp/intrinsics/x64-amd64-intrinsics-list?view=msvc-170)
- list of 32-bit intrinsics (https://learn.microsoft.com/en-us/cpp/intrinsics/x86-intrinsics-list?view=msvc-170)
- list of predefined MSVC macros (https://learn.microsoft.com/en-us/cpp/preprocessor/predefined-macros?view=msvc-170)

The error message in question:

```console
F:\dev\opentrack-depends\onnxruntime-build\msvc\_deps\abseil_cpp-src\absl/crc/internal/crc32_x86_arm_combined_simd.h(145,32): error C3861: '_mm_crc32_u64': identifier not found
  return static_cast<uint32_t>(_mm_crc32_u64(crc, v));
                               ^
F:\dev\opentrack-depends\onnxruntime-build\msvc\_deps\abseil_cpp-src\absl/crc/internal/crc32_x86_arm_combined_simd.h(193,50): error C3861: '_mm_cvtsi128_si64': identifier not found
inline int64_t V128_Low64(const V128 l) { return _mm_cvtsi128_si64(l); }
```
Merge 06f5832108a2b01e0a900db51e1c870f7069a1f2 into 797501d1

Merging this change closes #1617

COPYBARA_INTEGRATE_REVIEW=https://github.com/abseil/abseil-cpp/pull/1617 from sthalik:pr/fix-msvc-32-bit-avx 06f5832108a2b01e0a900db51e1c870f7069a1f2
PiperOrigin-RevId: 607483370
Change-Id: Id2a6f6dd33c2707fe7ffe134e7335916f3fb9da3

PiperOrigin-RevId: 607095897
Change-Id: I077857f0b4c5cf8bbc62bfc8e51a20bb6845296c

Strings larger than 1 GiB on a platform with a 32-bit size_t could
potentially overflow size_t in `CEscapedLength()`, resulting in an
undersized allocation. The resulting write in
`CEscapeAndAppendInternal()` would then write beyond the bounds of the
output buffer.

A second overflow, where the calculated escaped length is added to the
size of the string being appended to, is also fixed.

In both cases the program will now abort prior to the overflow.

Credit goes to Ronald Crane (Zippenhop LLC) for reporting this issue.

PiperOrigin-RevId: 607019573
Change-Id: I97bf246cde96102a793d2db49446cccae08abf59

The optional single template argument for built-in types is not documented in LLVM's comments that list the grammar, but it is [intentionally implemented](https://github.com/llvm/llvm-project/commit/4bc4d51c18d5c087dfbdad1753c84bba8dbf3be0).

PiperOrigin-RevId: 606843797
Change-Id: Ib5ef1983812e9ea68a39fe64b87b6ad6ce81b93c

Imported from GitHub PR https://github.com/abseil/abseil-cpp/pull/1412

In Clang, an `-Xarch_` compiler flag indicates that its successor only applies to the specified platform (e.g., `-Xarch_x86_64 -maes`). This is used in `absl/copts/AbseilConfigureCopts.cmake` to selectively enable hardware AES support on Apple platforms.

However, when generating pkg-config files, those `-m` flags are filtered out, while the `-Xarch_` flags that precede them are left untouched. This led to the error reported in #1408.

Fix that by filtering out each `-Xarch_` flag with its successor at once.

Fixes #1408.
Merge 89d20ab816b7cead56f05d5a6bc5146d1c4f4335 into 34604d5b

Merging this change closes #1412

COPYBARA_INTEGRATE_REVIEW=https://github.com/abseil/abseil-cpp/pull/1412 from ZhongRuoyu:xarch-pkgconfig 89d20ab816b7cead56f05d5a6bc5146d1c4f4335
PiperOrigin-RevId: 606730193
Change-Id: I3e177a56721acd3145fd03c64102741898afd2a5

This function copies the parser's state onto the stack in order to perform backtracking.
Adding a complexity guard will ensure that it fails gracefully, instead of contributing to excessive consumption of stack space.

PiperOrigin-RevId: 606720511
Change-Id: I2f6f03e5f8bc4cc571a4159ecfc2af6f3e00fa68

absl::kuint128max should be replaced with absl::Uint128Max()
PiperOrigin-RevId: 606279299
Change-Id: Ib67b5431f14891f56bb706e62f183d0de75d335f

This allows us to annotate nontrivial types that should be flagged as unused
variables.  Compilers otherwise ignore nontrivial variables as their
constructors/destructors may be desired and intentional (for example, a scoped
lock).

PiperOrigin-RevId: 606266618
Change-Id: I64b5f6d32a3cec2f18e0aa9029905f5e836c11a9

For example, this covers the following:
```
template <typename T>
    requires std::integral<T>
int foo();
```

Refactor parsing of `Q <requires-clause expr>` into a single function that performs backtracking to avoid reimplementing `ParseTemplateArgs` in terms of nested if-else blocks.

PiperOrigin-RevId: 605785418
Change-Id: I118998a75e050dcf46af125b613b690312fd3cbe

The Apple implementation gives false positives starting with
Xcode 15.

PiperOrigin-RevId: 605726702
Change-Id: I2e5e574eca08071d24e97304f005cf9c78230913

These clauses show up in other places that end up mangled as parts of function signatures (e.g. on template struct params). We'll handle those in a follow-up.

PiperOrigin-RevId: 605694497
Change-Id: I1bfe4c0cfaa739fdd24548b1547482410c92b5a7

PiperOrigin-RevId: 605460827
Change-Id: I57007a7ad18829e7bfed27ba65871afbd227d012

`demangle`: Parse C++20-compatible template param declarations, except those with `requires` expressions

Support for `requires` expressions will be added in a follow-up.

PiperOrigin-RevId: 605418370
Change-Id: I2c84cdf0c4599e36683b3c94dcbb173ab4fc3ee8

This macro is used by a Google-internal service for automated
refactoring. It is being released so that Google projects that
use this service can also open-source their code that uses it.

PiperOrigin-RevId: 605322286
Change-Id: I16babcc56c454e07461690a8bb58be5da72d283f

PiperOrigin-RevId: 605288522
Change-Id: Id16ad96b9323c71bd60801a6c8c359f7ff3d320e

PiperOrigin-RevId: 605125821
Change-Id: I2ee260eaf283acafd80abfd2b7419a0e9f597a78

Avoid hash computation and `Group::Match` in small tables copy and use `IterateOverFullSlots` for iterating for all tables.

PiperOrigin-RevId: 605116090
Change-Id: Ia65c664421f7630225b00f1c45c636b4681121ce

Throughput of the 64 byte chunk loop inside `LowLevelHash` (or now in `LowLevelHashLenGt16`) gets limited by the loop carried dependency on `current_state`. By using 4 states instead of 2, we can reduce this duration by 1 cycle. On Skylake, it is reduced from 9 cycles to 8 cycles (12.5% faster asymptotically).

To see the reduction in a simplified version of `LowLevelHash` implementation on Skylake:
* Before: https://godbolt.org/z/Tcj9vsGax, llvm-mca (https://godbolt.org/z/3o78Msr63) shows 9 cycles / iteration.
* After: https://godbolt.org/z/q4GM4EjPr, llvm-mca (https://godbolt.org/z/W5d1KEMzq) shows 8 cycles / iteration.
* This CL is removing 1 xor (1 cycle) per iteration from the critical path.

A block for 32 byte chunk is also added.

Finally, just before returning, `Mix` is called 1 time instead of twice.

PiperOrigin-RevId: 605090653
Change-Id: Ib7517ebb8bef7484066cd14cf41a943953e93377

PiperOrigin-RevId: 605030710
Change-Id: Id273d0ff79ebff4f753a4ca9818e2b53661d6fee

`char_traits<>` implementation, recently removed from `libc++`.

PiperOrigin-RevId: 605023824
Change-Id: Ia708c91c24625afc4504fa77e959f8ed1e025589

SwissTable provides support for heterogeneous lookup in associative
containers through transparent Hash and Eq types. However, it is not
possible for user types to provide additional specializations to allow
their types to use this functionality.

This CL brings ability for user types to specify their own transparent
absl_container_hash and (optionally) absl_container_eq inner types to
achieve the same functionality.

PiperOrigin-RevId: 604994859
Change-Id: I302486d292c9a18b7d4c77033227008f5539e354

C++20 introduced constraints, where concepts such as `std::integral` can be used to restrict template parameters:

```
template <std::integral T>
int foo(T t);  // Only accepts integral types.
```

Clang mangles these starting with a prefix `Tk`, followed by the concept name, which mangles as a type.
For example, compare `foo` and `bar` here: https://godbolt.org/z/Kzbrrchde

Note that this implementation doesn't cover the more complex case where the concept is templated too. See [llvm's implementation](https://github.com/llvm/llvm-project/commit/4b163e343cfa54c8d55c9da73c70d58f55ea9df2) and its tests for a thorough version.

PiperOrigin-RevId: 604976260
Change-Id: Ic116c5f6f27c3f7714638bdee8de11dce871f0be

Roll forward 9d8588bf (which was rolled back in 146169f9) with fix.

PiperOrigin-RevId: 604534265
Change-Id: Ic70ae72d4d73ac8849196cb844c8de32790bd6c8

to validate that the input was actually valid hexadecimal data.

Mark the version of absl::HexStringToBytes() that does not validate
the input as deprecated.

Fixes #141

PiperOrigin-RevId: 604495678
Change-Id: Iac3020c33c9dbc6d8e31a43b746783fb345edaa7

PiperOrigin-RevId: 604369517
Change-Id: I6024a8828563c5a2487ba85ede91a88d7059f9c8

PiperOrigin-RevId: 604094890
Change-Id: If5634a15c7ea7c8ef2cc9b6d09636ee7836946d2

PiperOrigin-RevId: 603816996
Change-Id: Ifc7dc6299e65043697b4a0c6e9e8eef869297ce3

PiperOrigin-RevId: 603784442
Change-Id: I3d57e5f438b276c984f5d5416889b19e7ddb501a

PiperOrigin-RevId: 603696164
Change-Id: I876ae0ead5964b626c285c60d9a4b2bbe62e113d

PiperOrigin-RevId: 603576688
Change-Id: I5e7868de7016d76fb7a44fea93be72ab90f239d3

CycleClock::Now utilizes ABSL_INTERNAL_CYCLECLOCK_SHIFT to discourage reliance
on the raw CPU cycle counter values. As a side effect, it discourages
strictly-increasing assumption. Apply the idea to discourage over-reliance on
the precision of absl::Now/absl::GetCurrentTimeNanos.

Programs relying on a very high precision often exhibit portability issues on
machines with a lower cycle counter precision, or worse, race conditions. For
example, Apple M1 emulated x86 RDTSC only guarantees weakly-increasing RDTSC
values.

x86 clock speed is usually measured in GHz and is still precise after we drop 8
least significant bits.

PiperOrigin-RevId: 603493500
Change-Id: Ib1b00075109283f5dbcb39a43fe0ab722351a1e2