Summary:
This fixes various include and dependency issues found when building
`folly/compression/test/facebook:compression_fuzz_test.so` with modules.

Reviewed By: yfeldblum

Differential Revision: D9271280

fbshipit-source-id: e0ae46c2a6d95c8f4d0a66e3dc20e1d045ffc6a4

Summary: This reverts my previous change "RequestContext: replace F14FastMap by std::unordred_map"

Reviewed By: ot

Differential Revision: D9280469

fbshipit-source-id: ad3f4d5f520e88ab0e0fab0506d87bab2cd10d41

Summary:
[Folly] Epochs for `BufferedRandomDevice`.

To maintain the security property that the thread-local object is reset after fork in the child process, but in an async-signal-safe way.

This approach has the downside that it makes the fast path larger by 3 instruction (`mov`, `cmp`, `jne`).

Reviewed By: djwatson, ot

Differential Revision: D9256536

fbshipit-source-id: dc34e1aeb75e1009610d47fe4ef7030a4a0e2a90

Summary:
The following code won't compile with g++ 8.2 when `-Wall -Werror` is turned on (`-Werror=class-memaccess` implied):

    std::memcpy(dst, src, n * sizeof(Value))

where in `VectorContainerPolicy`'s case, `Value` is typically an `std::pair`, which  indeed has no trivial copy-assignment.
Pull Request resolved: https://github.com/facebook/folly/pull/907

Reviewed By: yfeldblum

Differential Revision: D9244951

Pulled By: Orvid

fbshipit-source-id: 26787d43dab7d4212c5975a509dfd8703b0a157c

Summary:
Overall plan to modify Future<T>::then to be r-value qualified and use Future<T>::thenTry or Future<T>::thenValue.

The goal is to disambiguate folly::Future and to improve type and lifetime safety of Future and its methods.

Reviewed By: yfeldblum

Differential Revision: D9273440

fbshipit-source-id: f1e1b91fd8fe02020a5919b9193644e256413982

Summary:
[Folly] Cut `FOLLY_HAS_INCLUDE` now that all supported platforms have it.

MSVC2015 and gcc4.9.1 don't support it, but folly no longer targets them.

Reviewed By: simpkins, Orvid

Differential Revision: D9262258

fbshipit-source-id: e518492acd7c11374ab971b26adb642bce7b19c3

Summary: Fix various dependency issues in folly for modular build.

Reviewed By: yfeldblum

Differential Revision: D9246772

fbshipit-source-id: 32afa48db3222b906077376c54e1301c8366d97d

Summary: Move Password in file from wangle to folly

Reviewed By: ngoyal, yfeldblum

Differential Revision: D9192652

fbshipit-source-id: b883ddaf2d815b37677d9c5e924b3dd06aa0ec6b

Summary:
Move the definition of `HAZPTR_DEBUG` from code to the target to fix modular
build. This is needed because `HAZPTR_DEBUG` is effectively a configuration
macro affecting the API.

Reviewed By: magedm

Differential Revision: D9256127

fbshipit-source-id: 44deb0e915373ea88206098544b03fa335b31a22

Summary:
F14 varies its inline code according to the flags passed to the
compiler. This can cause problems if libraries override compilation flags
on their own, since it won't be safe to exchange F14 instances between
compilation domains.  This diff introduces a template specialization
that varies according to the intrinsics mode and that has only a single
definition (in F14Table.cpp).  The member is called on a couple of cold
paths (exception handling for copy construction and rehash).  This makes
it very likely that a compilation unit that uses F14 and is compiled
with different compiler flags than F14Table.cpp will get a linker error.

Reviewed By: yfeldblum

Differential Revision: D9200315

fbshipit-source-id: 9cbca18eef0ddd6efcf6d9d6057eda2400f8653c

Summary:
As indicated at https://github.com/krb5/krb5/blob/master/src/include/krb5.h,
this header is deprecated, so update to use the new location (and fix lint).

Reviewed By: luciang

Differential Revision: D9259770

fbshipit-source-id: 6832206b8c890184f5d3369b0305048ae9c377d9

Summary:
Make the file name consistent with its class name
Add benchmark code
Attach performance data

Reviewed By: magedm

Differential Revision: D9250954

fbshipit-source-id: b9f9b90cb83ca319d91fe65667d0072511be5e26

Summary:
Update the CMake build files to install a libfolly.pc file so downstream
projects can use pkg-config to more easily depend on folly.

Reviewed By: yfeldblum

Differential Revision: D9213429

fbshipit-source-id: 849c0fce01d4903d3f8a7b6d196037bf950e5921

Summary:
folly, and therefore most of our other open source projects, no longer support
gcc-4.9.  Drop Ubuntu 14.04 and Debian 8.6 from our CI platforms list, since
the these images only support gcc 4.9.

We should ideally add support for Ubuntu 18.04 soon, but I'll leave that for a
later diff.  Removing the older platforms is somewhat higher priority for now
to get our Travis CI builds green again.

Reviewed By: snarkmaster

Differential Revision: D9258797

fbshipit-source-id: 3cab47a6c51b2dbe63214034240f844c85963c3d

Summary:
Fork test tries to test that we don't reuse the SecureRandom buffer on fork, but it only tests one byte.

This gives us probability one in 256*256 (or something like that) that we spuriously fail.

Check the whole buffer instead before failing.

Reviewed By: fredemmott

Differential Revision: D9242788

fbshipit-source-id: b9fb809518797c439d7fe16084a04f9964a1f597

Summary:
experimental/symbolizer/ElfUtil.cpp was deleted in D9218592 so it no longer
needs to be explicitly excluded in CMakeLists.txt

Reviewed By: Orvid, calebmarchent

Differential Revision: D9229945

fbshipit-source-id: 7d708b265573e70e4e6de88edefa2fa5b41af200

Summary:
The code in folly/experimental/symbolizer/* is already included in the main
libfolly library if it is supported by the build configuration.  Remove the
separate libfolly_symbolizer target added in D9216312 that includes a second
copy of these objects.

This change effectively reverts D9216312.

Reviewed By: Orvid, calebmarchent

Differential Revision: D9229946

fbshipit-source-id: d8e6874eeab8c9c1aa519441bd29ae107aef03ac

Summary:
Previously it was only capable of handling 512 byte strings under MSVC due to the constexpr recursion depth limit being 512 by default under MSVC.
In the future once we have C++14 constexpr everywhere, this should just be a for loop, which should remove the limit.

Reviewed By: yfeldblum

Differential Revision: D9230717

fbshipit-source-id: 88bcf0ef94b200301eb41392482e910b2e8de008

Summary:
Separates l-value and r-value versions of parameterless then.

Deprecates l-value version.

Reviewed By: yfeldblum

Differential Revision: D9222345

fbshipit-source-id: 3dd95b3e35bf1b06d426334c30150b049a5e0b70

Summary:
Bug fix: call `reserve()` trying to allocate more memory than available
Pull Request resolved: https://github.com/facebook/folly/pull/888

Reviewed By: ot

Differential Revision: D9234444

Pulled By: yfeldblum

fbshipit-source-id: 4d12236aee745ad663d95a370d273f619f92b1bd

Summary:
Update all of the RequestContext tests to use a test fixture that helps ensure
they all start from a consistent state.

Otherwise tests can end up running with RequestContxt settings left over from
previous test functions, which can affect their results.

In particular, several tests leave behind data for the "test" key.  This
causes subsequent calls to `setContextData()` to not actually set new data for
this key, but to instead reset its value to null.  This would cause many tests
to segfault since they did not check if `getContextData()` returned null
before dereferencing the result.

Reviewed By: yfeldblum, A5he

Differential Revision: D9233617

fbshipit-source-id: ce7a7c738592305f01b16fec7796505ea192ede4

Summary:
- Fix the variable name passed to FIND_PACKAGE_HANDLE_STANDARD_ARGS() so that
  it matches the module name (`FindGMock.cmake`).  This makes sure that the
  failure is detected correctly if GMock is not found.
- googletest depends on pthreads, so use `find_package(Threads)` and add
  `Threads::Threads` to the list of required library dependencies.

Reviewed By: yfeldblum

Differential Revision: D9212536

fbshipit-source-id: eae6220a2dbfe777bff1ac0ee0fb4df53de16955

Summary:
Use CMake's built-in FindThreads module rather than having our own custom
FindPThread.cmake module.

Reviewed By: yfeldblum

Differential Revision: D9212535

fbshipit-source-id: 9a652c510cb310c18547d31bd8215dedd8532a4c

Summary:
Ubuntu 18.04 (and presumably other Debian-based distributions) install
dwarf.h at /usr/include/libdwarf/dwarf.h.  However other platforms like Mac
typically install dwarf.h directly inside include/ without a libdwarf
subdirectory.

Commit e66fb34c previously changed the code to look only for "dwarf.h" to
support Homebrew installations of dwarfutils on Mac, but accidentally broke
finding dwarf.h on Ubuntu in the process.  This change should hopefully work
on both platforms.

Fixes: #905

Reviewed By: yfeldblum

Differential Revision: D9212428

fbshipit-source-id: a4520a47718a2da25e597ac7350344931ba21634

Summary: [Folly] Import `Policy` members into `F14Table` to deduplicate repeated uses of `Policy::`.

Reviewed By: shixiao

Differential Revision: D9176546

fbshipit-source-id: bbd32a17cca7d38b3e5d55c1840353821f4e2ddc

Summary: Add thread cached ints benchmark

Reviewed By: yfeldblum

Differential Revision: D9212949

fbshipit-source-id: 552aab15e7421f74b5cec3ce4abd211ab3141153

Summary:
Given an exception_wrapper `ew` holding a `std::runtime_error`, the following assertion fails:

```
try {
  ew.handle();
} catch( std::runtime_error const & ) {
} catch( std::exception const & ) {
  assert(false);
}
```

Reviewed By: simpkins

Differential Revision: D9199924

fbshipit-source-id: b6f90ccf269a1b58f0dd07723451ff34cde1d529

Summary: Move the definition of the `ZSTD_STATIC_LINKING_ONLY` and `ZDICT_STATIC_LINKING_ONLY` macros from code to the target to fix the modular build. It is needed because these macros affect the public API of `zstd.h`.

Reviewed By: terrelln

Differential Revision: D9189720

fbshipit-source-id: efacc1456803900fd65159a43cf1fba2daf15673

Summary: On fork, flush the secureRandom buffer, so that we don't share entropy between the parent and child.

Reviewed By: ricklavoie

Differential Revision: D9196474

fbshipit-source-id: 12ff8488d814466186df61328a5f1d4000beb27f

Summary:
Print benchmark results as and when they complete instead of outputting the
whole thing once at the end

Reviewed By: djwatson, ot

Differential Revision: D9177038

fbshipit-source-id: 54898b303771f214c21163cd87b11f8e2083cdf4

Summary:
When passing a callable to `folly::via` that returns a SemiFuture,
via should return a Future. The implementation does the correct thing
but the declaration's return type is not correct. This diff fixes that.

Reviewed By: yfeldblum

Differential Revision: D9219041

fbshipit-source-id: e61544231e7846c713ff43394c3bf5d1aae18233

Summary: Late change in D9175807 removed necessary temporaries. This caused a lifetime management problem for GCC.

Reviewed By: Orvid

Differential Revision: D9227937

fbshipit-source-id: d5fea8a98855c6302189a7dee8cd5862d13a63a3

Summary: folly::Future's map() functions create a std::vector<Future<...>>, then use push_back to populate that vector. This could result in reallocations. Avoid that by reserving proper space ahead of the push_backs.

Reviewed By: yfeldblum

Differential Revision: D9204713

fbshipit-source-id: 51d9154f483a47c6196a4d4d9d4af14e4eeab6e0

Summary: autotools-based builds are being deprecated, but they lack the symbolizer library target.

Reviewed By: yfeldblum

Differential Revision: D9216312

fbshipit-source-id: 585ce3c43a0800823c7de4ece99b56a1af82bae4

Summary:
Overall plan to modify Future<T>::then to be r-value qualified and use Future<T>::thenTry or Future<T>::thenValue.

The goal is to disambiguate folly::Future and to improve type and lifetime safety of Future and its methods.

 * Deprecate l-value qualified calls to Future<T>::then.
 * Make appropriate changes to folly to make this correct without local deprecation warnings.

Reviewed By: yfeldblum

Differential Revision: D9175807

fbshipit-source-id: 9dc9812270ca3fb800f3a56d3aef8ef3d1d209e6

Summary: [Folly] Dead Code: `folly/experimental/symbolizer/ElfUtil.cpp`.

Reviewed By: calebmarchent

Differential Revision: D9218592

fbshipit-source-id: e231852c068eb797c7c3a4a3955fefc38e4be2fc

Summary:
Original commit changeset: 6b5faf000e5f

Breaks builds on Ubuntu Bionic, effectivly undoing the work of D3878961.
Note: both autotools and CMake builds incorrect on Bionic, its just that
autotools build fails explicitly where CMake fails silently; failing to
detect a correctly installed libdwarf headers. The difference being that
CMake does not build symbolizer.

Reviewed By: yfeldblum

Differential Revision: D9194639

fbshipit-source-id: c6798a0672195a70ced4898535f2f932018224d5

Summary: When merging centroids together, we recalculate the mean and weight. This can sometimes lead to inaccuracy due to floating point arithmetic. In a bad scenario, this can actually lead to the digest being unsorted, which triggers assertions in the code. To fix this, sort the digest at the end of merge. Since this is bounded to 100 elements that are mostly sorted, this appears to be cheap according to benchmarks.

Reviewed By: yfeldblum

Differential Revision: D9213936

fbshipit-source-id: 5a2978a1b759b70206aef57245bef193aed65efd

Summary:
Blocking algorithm:
 Push:
1.  push
2.  increase the PTicket number
3.  put (PTicket << 1) to the corresponding place in futex array.
4.  if the original  futex value & 1  == 1,  wake up ()

Pop:
1.  increase the CTicket number
2.  spin if the corresponding CTicket futex is not ready
3.  set the right bit of the CTicket futex  to 1, wait on it.
4. pop

Add additional control to improve the accuracy:
For the top several levels (top 0 - 3 levels by default), it does not arbitrarily insert any node in the middle of the list. Thus guaranteeing no later coming lower priorities can be insert at the top levels (pruning strategy still valid for the top four levels). For the bottom levels (start from level 7 by default), if insertion requires to hold parent + child locks, it traverses the parent list to check whether the parent list has lower priority nodes than the new node. If the condition holds, insert the new node to the parent and get the last node in the parent list and insert it to the child list.

Reviewed By: magedm

Differential Revision: D9094903

fbshipit-source-id: fa4d31fb0c21da746570029e0e3b4b3fbeaa2024

Summary: This code is hot enough that this might matter. Though likely over-optimization.

Reviewed By: djwatson

Differential Revision: D9197636

fbshipit-source-id: d9e2a90410c03a73d1754d99c72967f5dcd96c0f