Summary:
[Folly] Cut `SingletonThreadLocal::getWrapperInline`.

It would not be useful in any scenario, given that the generated code is very large.

Reviewed By: djwatson

Differential Revision: D7315501

fbshipit-source-id: 1ebc15ca0eaf13adaf52fae3971e7f9914bfe10e

Reviewed By: yfeldblum

Differential Revision: D7272553

fbshipit-source-id: f0aaa278e6a81a53f78ae29b68f1bf9f1ba02899

Use unbounded queue in NotificationQueue (same as in D7164130 but with the hazptr fix from D7256905 and the singleton one from D7283390)

Summary: Use unbounded queue in NotificationQueue (same as in D7164130 but with the hazptr fix from D7256905 and the singleton one from D7283390)

Reviewed By: yfeldblum

Differential Revision: D7278275

fbshipit-source-id: 78ea94e1c9d492febd33106b06ce275d4a90a56a

Summary:
F14VectorMap and F14VectorSet use ItemIter to interact with F14Table,
but they don't actually need to be able to identify the begin ItemIter.
This diff removes both the storage for packedBegin_ and the code to
update it (present in all insert and erase functions) for that policy.
This is a small CPU optimization, and reduces sizeof(F14Vector{Map,Set})
from 32 bytes to 24.

Reviewed By: shixiao

Differential Revision: D7181769

fbshipit-source-id: 318c63584543fca339a002d8bcb5872a7bd0bdd3

Summary:
Fiber suspends are incompatible with C++ exception handling. We should enforce that callers are not suspending fibers while handling exceptions.
See https://fb.facebook.com/groups/560979627394613/permalink/1119586164867287/

NOTE: This check currently fires not just for baton.wait() (and all things based on baton), but also for calling runInMainContext. I'm unsure of the safety of this.

Reviewed By: yfeldblum

Differential Revision: D6695110

fbshipit-source-id: cf36edbab9f0573cbf639bf79d0c267681ac5914

Summary: Rename set/getServiceIdentity to set/getSessionKey.  This is a more representative description since verified information is in the cert.

Reviewed By: yfeldblum

Differential Revision: D7184169

fbshipit-source-id: c9404419cb37d2b5ab2fcd8299975e514e3f495a

Summary:
[Folly] Move `folly/UncaughtExceptions.h` to `folly/lang/`.

(Note: this ignores all push blocking failures!)

Reviewed By: andrewjcg

Differential Revision: D7319259

fbshipit-source-id: 38d47503e150e01b5e4bcded6c8207484048bf84

Summary:
Makefile does not install detail/Singleton.h, which is needed by proxygen
Closes https://github.com/facebook/folly/pull/804

Reviewed By: andrewjcg

Differential Revision: D7320496

Pulled By: yfeldblum

fbshipit-source-id: 8c576fe53be8292fa51a3db50dcfaa1d1cb13b6a

Summary:
[Folly] Tweaks to `folly/TokenBucket.h`.

* No custom clock wrapper - just use the clock interface directly.
* Rename from `Parameterized` to `Basic`.

Reviewed By: andrewjcg

Differential Revision: D7319361

fbshipit-source-id: 1a7d8761e58f5f9b06dd8fdf5f7916913354c3b9

Summary:
[Folly] Tests for `throw_exception`.

Currently, only for platforms with exceptions.

Reviewed By: andrewjcg

Differential Revision: D7319127

fbshipit-source-id: 689cf68d45619eddbbda6a4c36226fc7695c1243

Summary:
[Folly] Let `SingletonThreadLocal` no longer depend on `Singleton`.

After recent changes, the dependency was only for `DefaultTag` and `DefaultMake`, not for `Singleton` or `LeakySingleton` itself. So extract the two small types to a small detail header, and split.

Reviewed By: ot

Differential Revision: D7300727

fbshipit-source-id: 5ccfc0abff7dbb8c2c9ea83fb9bb3a085904ce58

Summary: [Folly][coro] Include `Portability.h` before using `FOLLY_HAS_COROUTINES`.

Reviewed By: Orvid

Differential Revision: D7315956

fbshipit-source-id: 65490ced2573652e346c54aca4e67350fe3579fe

Summary:
Boost handles ensuring destruction order is correct far better than the implementation previously being used.
This gets the Windows build working again, as it's currently hitting a thread-local static destruction order issue. Specifically, the thread_local containing the backing map for the portability implementation was being destroyed before a thread_local in rcu, which caused attempts to access thread local values that had already been freed.

Reviewed By: yfeldblum

Differential Revision: D7312978

fbshipit-source-id: 4d7f1f4d8bb0b92e0c52366078ee7ee34c4248f2

Summary: [Folly][coro] Disable CoroTest for gcc builds.

Reviewed By: yfeldblum

Differential Revision: D7314274

fbshipit-source-id: 06747135350001b55d6ce95e1ce4bf738fca0227

Summary: Add a generic Identity class to AsyncTransport that applications are free to implement to convey identity information for both self and peer.

Reviewed By: yfeldblum

Differential Revision: D7177085

fbshipit-source-id: f1dcd9bb2dbf57751fe7ef0608663ddc1d527b92

Summary: [Folly][coro] Enable tests and fix build.

Reviewed By: yfeldblum, modocache

Differential Revision: D7304801

fbshipit-source-id: 0975201a0521cf8275d82ee6aaebfa02986c5b79

Summary:
This check is useful for other platforms and in other context where F14 is
included. For the sake of fast compilation, separate this macro out so that one
doesn't have to pull in the entire F14Map/Set headers to use this macro.

Reviewed By: yfeldblum

Differential Revision: D7300144

fbshipit-source-id: d1aeb7e3c5d2f15703dd356fc54622751e3fa71a

Summary: Unneeded header in frequently-used file; drop it

Reviewed By: yfeldblum

Differential Revision: D7165040

fbshipit-source-id: b60a3cdf924ab93961a7ead3ac0653a2339a4faf

Summary: Removing expensive, unneeded includes (esp. Boost includes) gradually.

Reviewed By: yfeldblum

Differential Revision: D7144471

fbshipit-source-id: b568ec75251d360a72e7fa2d656ab9463b5a4728

Summary: Add getTaskQueueSize() method in CPUThreadPoolExecutor

Reviewed By: yfeldblum

Differential Revision: D7294802

fbshipit-source-id: 1f37558a1f14fcc2888c654248d649365c323432

Summary: Windows defines `IP_MTU_DISCOVER`, but does it differently from everywhere else, so disable it unless everything required is defined.

Reviewed By: yfeldblum

Differential Revision: D7298640

fbshipit-source-id: 8ecc2dfd9ba83f31e9d72e0f42fb4fbf8f3e50b5

Summary:
assert(x) doesn't mark x as used in prod builds, causing problems
for builds that warn on unused variables and promote warnings to errors.
This diff changes assert calls to FOLLY_SAFE_DCHECK, which is better in
this regard.

Reviewed By: phoad, shixiao

Differential Revision: D7291084

fbshipit-source-id: 1ae48a271f30182ac8a1c2cf126af0e6b4fd0a25

Summary:
[Folly] Alias `std::launder` on `MSVC` to avoid ambiguity.

Fixes #802.

Reviewed By: Orvid

Differential Revision: D7286052

fbshipit-source-id: 7cd89c9ad236a433f53fbdb9e2e024b85105827e

Summary:
[Folly] Fix `SingletonThreadLocal` use from headers included by separate dynamically-loaded modules.

There is no need in `SingletonThreadLocal` specifically to check for double-registration and `SingletonThreadLocal` does not support mocks, so `LeakySingleton` may be bypassed entirely.

Reviewed By: djwatson

Differential Revision: D7283390

fbshipit-source-id: 45295c4dab233b9373ed32a47d4c80832b0f9eff

Summary:
MEMBARRIER_CMD_SHARED is too slow, use EXPEDITED (or the mprotect hack if kernel is too old).

hazptr has been using EXPEDITED for a while, the only other users are RCU and read_mostly_shared_ptr

Reviewed By: yfeldblum, osandov

Differential Revision: D7280190

fbshipit-source-id: 2a9aeab035f8996566b557d7f406686c361a37d7

Summary: Provide UBSAN-clean add/sub function on integral types. Helps to avoid overflow.

Reviewed By: yfeldblum

Differential Revision: D7230723

fbshipit-source-id: 10fa54b405f9643dfbbb1940cf100516fd1341e1

Summary:
This makes it easier to use EventBase in cases where
classes require ScheduledExecutor instead of Executor. Previously
the client would have to implement a wrapper around EventBase.

Reviewed By: yfeldblum

Differential Revision: D7179869

fbshipit-source-id: 21730a56ed5b71fd731ffd272f9f3752b70d63ce

Summary: Expose Default{Hasher, KeyEqual, Alloc} to platforms without intrinsics available

Reviewed By: yfeldblum

Differential Revision: D7277896

fbshipit-source-id: 92f46aef9ab0926c32c85e29e368f7c0b58c2b3e

Summary: Add missing WIN32 compat builtins

Reviewed By: Orvid

Differential Revision: D7276240

fbshipit-source-id: 094c0861126b3477e9112987f033edacbde4aca4

Reviewed By: yfeldblum

Differential Revision: D7257019

fbshipit-source-id: 2fe54fe1d42bfe81b2f6f8e43ae0f7e738e93edd

Summary:
hazptr_priv_list needs to be reinitialized on fork, as well as ODR violations.

This broke stuff when trying to land D7164130

TL;DR the thread_locals need to be in a .cpp file, the templates don't always get merged correctly, especially in the face of dlopen.

You can, however, use a thread_local * pointer cache to the object.  This is basically what folly::SingletonThreadLocal does.

Folly's ThreadLocal runs destructors when forking even too, so nothing special is required.

Reviewed By: yfeldblum

Differential Revision: D7256905

fbshipit-source-id: c817167b5c3db27fa929feaa39295fd939c1cb4c

Summary:
Expose the default structs in `folly::f14` namespace so that users _could_
potentially refer to them.

Reviewed By: nbronson

Differential Revision: D7271846

fbshipit-source-id: 81a193209e59ed80260098ed97c99a7b6f2ad151

Summary: [Folly] Parameterize `AlignedSysAllocator` by a policy.

Reviewed By: ot

Differential Revision: D7246946

fbshipit-source-id: 5b6a89707c242e49bdbafba9dec6df4c65ef385c

Summary:
For some reason, compilation of certain Observable code fails in opt mode with the error that Observer::core_ is private.

It seems that ObserverCreator is not recognized as a friend of Observer. This forward declaration of ObserverCreator seems to fix this issue.

Reviewed By: yfeldblum

Differential Revision: D7268318

fbshipit-source-id: 1ea76efb0f50cb8a6b28147305ec7547f55338b8

Summary: [Folly] Enable `-Wunused-value`.

Reviewed By: Orvid

Differential Revision: D7266246

fbshipit-source-id: dc9b85a5f8ce67802cc2fa94746a26529012ec22

Summary:
F14 was missing the MSVC portability shims, and also needs to
use an MSVC-specific intrinsic to get the high word of a 64-bit multiply.

Reviewed By: yfeldblum

Differential Revision: D7263349

fbshipit-source-id: f47ceb2b2370d9bef1409e4790b236752c9f6ac4

Summary:
Currently, an error is thrown if a user tries to hash an instance of `folly::dynamic` which is of type `object` or `array`. Primitive underlying types are fine.

There doesn't seem to be a strict reason for this, as objects and arrays are just as hashable as primitive types. Additionally, the inability to hash here causes some unnecessary drawbacks (e.g. interacting with `unordered_map`s).

This diff uses the existing dependency on `folly/hash/Hash.h` to start allowing `folly::dynamic` objects/arrays to be hashed as well.

Reviewed By: ot

Differential Revision: D7219760

fbshipit-source-id: 1dff3f99bdacb66f719c614040cbd668ce51a33e

Summary: EventBase meets the definition of IOExecutor (it is an executor that "manages" at least one EventBase) and is already claimed to be one in the docstring.

Reviewed By: yfeldblum

Differential Revision: D7064481

fbshipit-source-id: 026e0e4c24d6fb15a32bc4772b35689f830ed944

Summary: deferError adds an error handling callback to SemiFuture, enqueued on the internal deferred executor, and thus will run on the chained executor or inline with .get() as for defer.

Reviewed By: yfeldblum

Differential Revision: D7243193

fbshipit-source-id: 9fe7dab9fbc8b236d717b058c81a303d87fa9253

Summary:
F14 is a 14-way probing hash table that resolves collisions by double
hashing.  Up to 14 keys are stored in a chunk at a single hash table
position.  SSE2 vector instructions are used to filter within a chunk;
intra-chunk search takes only a handful of instructions.  "F14" refers
to the fact that the algorithm "F"ilters up to "14" keys at a time.
This strategy allows the hash table to be operated at a high maximum
load factor (12/14) while still keeping probe chains very short.

F14 provides compelling replacements for most of the hash tables we use in
production at Facebook.  Switching to it can improve memory efficiency
and performance at the same time.  The hash table implementations
widely deployed in C++ at Facebook exist along a spectrum of space/time
tradeoffs.  The fastest is the least memory efficient, and the most
memory efficient is much slower than the rest.  F14 moves the curve,
simultaneously improving memory efficiency and performance when compared
to the existing algorithms, especially for complex keys and large maps.

Reviewed By: yfeldblum

Differential Revision: D7154343

fbshipit-source-id: 42ebd11b353285855c0fed5dd4b3af4620d39e98