Summary: A simpler manual executor that can be executed by a python thread, and release the GIL while waiting.

Reviewed By: andriigrynenko

Differential Revision: D8087779

fbshipit-source-id: a75ecbda3804a53f1b705aa04546b74368738b7b

Summary:
AsyncioLoopController is totally generic.
We were able, after all, to put all the python specifics in the Executor, and keep it simple.
I'm using it with ManualExecutor succesfully in my project.

Reviewed By: andriigrynenko

Differential Revision: D8085689

fbshipit-source-id: 007f65a72e41aa98113b7ee4e188719d9bc20340

Summary: We use proxy to wrap the server side of TTLS connection, thus we can safely disable TTLS option on server side. This helps with cases where people create server socket, then pass the FD to child, and that FD is a UNIX socket because of TTLS.

Reviewed By: djwatson

Differential Revision: D8096569

fbshipit-source-id: 657c3cee3a9df49cbb80fad771a897264d5277f4

Summary:
Diff exposes a more convenient constructor interface
in CPUThreadPoolExecutor that is analogous to the existing constructor
interface with the only exception that it takes a pair of min and max
number of threads that drives the executor.

Reviewed By: djwatson

Differential Revision: D8088545

fbshipit-source-id: 3ee92070335267948332ef75d67817d25436262f

Summary:
`std::vector<bool>` is a bit specific ([1], [2] and [3]), so element access doesn't return bool, but `std::vector<bool>::reference`  instead. Folly did not know how to cast it to bool.
[1]: http://en.cppreference.com/w/cpp/container/vector_bool
[2]: http://en.cppreference.com/w/cpp/container/vector_bool/reference
[3]: https://en.wikipedia.org/w/index.php?title=Sequence_container_(C%2B%2B)&oldid=767869909#Specialization_for_bool

Reviewed By: yfeldblum, ot

Differential Revision: D7986013

fbshipit-source-id: 01c1350b3f6815e9da4ce6305f25e4b9b1ca9d6d

Summary: This is blocking D8022229

Reviewed By: yfeldblum

Differential Revision: D8086230

fbshipit-source-id: 3d6f2c32190b405b9089c9d41378827e8cb2bb80

Summary: [Folly] Update some outdated fields in Futures Core test - the callback type and the interrupt lock type have been changed.

Reviewed By: Orvid

Differential Revision: D8084531

fbshipit-source-id: 3991360052c9309b531b7be64ef8e92d5ce6200c

Summary:
[Folly] Introduce makePromiseContract, for producing a paired promise-and-semi-future or a paired promise-and-future, depending on which overload is invoked.

It is not ideal that a future is gotten from a promise, but only once. Rather, it is ideal that they be created together.

C++17 code may use structured bindings:
```lang=c++
auto [p, f] = makePromiseContract(); // promise with semi-future
auto [p, f] = makePromiseContract(&e); // promise with future, given executor e
```

Reviewed By: LeeHowes

Differential Revision: D7984050

fbshipit-source-id: 52ee5624fde7c48aa0e81243564ad067c6c6919f

Summary: This will allow incremental migration of `collectAll` to `collectAllSemiFuture`. Once everything is migrated to `collectAllSemiFuture`, `collectAll` will change to be the exact same as `collectAllSemiFuture`, then all uses will be moved back to `collectAll` and `collectAllSemiFuture` will be removed.

Reviewed By: yfeldblum

Differential Revision: D8015664

fbshipit-source-id: 07a154dc733fd7e1e003f4c103355e7b5d1ad1b5

Summary:
Add hazptr_domain* member to hazptr_rec that never changes.
Remove hazptr_domain* member from hazptr_holder.

The primary optimization is reducing the sizes of hazptr_holder and hazptr_array from two pointers to one pointer per owned hazard pointer.

A secondary advantage is allowing swap of an individual holder in hazptr_array with a hazptr_holder object or an element of another hazptr_array under the conditions: both are either empty or nonempty and if nonempty both hazptrs belong to the same domain.

Performance effects on hazptr_holder:
+ Half the steps for empty construction, move and swap.
+ One less step in typical construction.
- Indirect instead of direct read of pointer to domain, but indirection is likely to be a cache hit if the hazptr was used recently.

Measurements:
- Faster construction/destruction of hazptr_array<3>. 7.5 ns. instead of 8.5 ns.
- No regression in other microbenchmarks.

Reviewed By: davidtgoldblatt

Differential Revision: D8010395

fbshipit-source-id: 01ffd26f675e57547b1506b777c79d18052ccc65

hazptr: Link counting. Protecting descendants of protected objects. Mutable and immutable links. Automatic retirement.

Summary:
The class hazptr_obj_linked includes all the reference counting and automatic retirement mechanisms. It supports:
- Mutable and Immutable links.
- Certain and uncertain removal. With certain removal users call retire. With uncertain users call unlink when an object is unlinked. The library determines automatically when to retire such object.
- E.g., UnboundedQueue uses immutable links with certain removal.
- E.g., ConcurrentHashMap uses mutable links with uncertain removal.

Reviewed By: djwatson

Differential Revision: D7674658

fbshipit-source-id: 948f5c3690367deaa0e2023a2def3bed2c22b9f0

Summary: This makes keep-alive join also happen if stop/join are called directly.

Reviewed By: yfeldblum

Differential Revision: D8060479

fbshipit-source-id: 1d1c0cdbcaac2e439aee9ad14f00fa7e4fda2339

Summary:
This is useful to fix legacy tests that use loop(), as keep-alive tokens are used more widely.

(Note: this ignores all push blocking failures!)

Reviewed By: yfeldblum

Differential Revision: D8065959

fbshipit-source-id: 2e7a27f538ac9d36eeea7ba3a882633a193141d1

Summary:
[Folly] Split Futures exceptions by `Promise` vs `Future`.

* Move `Promise` exceptions into the `Promise` header.
* Move `Future` exceptions into the `Future` header.
* Split `NoState` into `PromiseInvalid` and `FutureInvalid`.
* Remove the newly-empty exceptions header.

Reviewed By: andriigrynenko

Differential Revision: D7966499

fbshipit-source-id: 2dc6d2a941493979ebf47b3e70e5cf6a6fbd33cf

Summary: Trivial change to allow different services to turn this flag off at runtime

Reviewed By: djwatson

Differential Revision: D8060492

fbshipit-source-id: 8a62065ee1913fc67b6ae6983d03625849e139b3

Summary:
[Folly] `like`, `like_t`, `forward_like`, as per p0847r0.

See: http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2018/p0847r0.html.

Reviewed By: nbronson

Differential Revision: D8040722

fbshipit-source-id: 4c8a64037ec7447c8afa7b086c36db0f836f66f2

Summary: Make erase(iterator) return an iterator, like std containers.

Reviewed By: yfeldblum

Differential Revision: D8047159

fbshipit-source-id: 3cc94782a7afca0bf9fedefdc7296b9395ca9859

Summary:
This fix aims to prevent cases (pointed out by David Goldblatt) where cleanup returns while a concurrent slower bulk_reclaim has not completely reclaimed objects that are supposed to be reclaimed before the return from reclaim.

This fix uses an atomic count of the number of ongoing  bulk_reclaim-s. Cleanup returns only if it observes that that count is zero.

Reviewed By: djwatson

Differential Revision: D7586325

fbshipit-source-id: 341af641d20a7a07206c7238ea0356a9eabdbab5

Summary:
Remove Armed: it added an avoidable extra state transition, plus avoidable semantic complexity in the behavior/contract of setResult() & setCallback(), i.e., if either of those transitioned to Armed, that thread may or may not be the one that made the transition to Done with the possibly-synchronous call to the callback.

After this change there is only one transition after Only* - to Done, and the new contract/behavior of setResult/setCallback is still asynchronous (the caller cannot know which thread will call the callback) but at least has only one asynchrony instead of two (the thread that transitions to Done is always the thread that calls doCallback()).

Reviewed By: yfeldblum

Differential Revision: D8049396

fbshipit-source-id: 89a1e40ab8c09db4c8ab0ac3a1fd2cc5e092830d

Summary: Use synchronization/hazptr instead of experimental/hazptr

Reviewed By: djwatson

Differential Revision: D7565384

fbshipit-source-id: 26790eec80f344cb4b227672e6d51d97c9170dee

Reviewed By: Orvid

Differential Revision: D8053447

fbshipit-source-id: 4550e4fabffb7e6be7d7927292aed2b33c10dee5

Summary: Kerberos has issues with include order on Windows, so give it it's own portability header to force the correct include order and undefine the mess it `#define`'s.

Reviewed By: yfeldblum

Differential Revision: D8050238

fbshipit-source-id: 6c718f8f0db9f039734bead5f90cd289ea1dfd78

Summary:
This is blocking D8022229

(Note: this ignores all push blocking failures!)

Reviewed By: yfeldblum

Differential Revision: D8038878

fbshipit-source-id: 4b5f085b86f065df0612a132ec26c3d4980edbd1

Summary:
Because it defines `HANDLE_FLAG_INHERIT`

(Note: this ignores all push blocking failures!)

Reviewed By: yfeldblum

Differential Revision: D8050035

fbshipit-source-id: a67d19aa7882f312273e0761123d6827fe07bd81

Summary: Avoid overhead of std::function (virtual fn, possible malloc) when the interrupt msg reaches the core before the interrupt handler (the former in future.raise(), the latter in promise.setInterruptHandler()).

Reviewed By: yfeldblum

Differential Revision: D7994756

fbshipit-source-id: 58827e7ac320b6585f1a1c3df08932bbe38f91ee

Summary: For greater clarity, return a BlockingQueueResult struct from add instead of a raw bool

Reviewed By: yfeldblum, magedm

Differential Revision: D7709300

fbshipit-source-id: 36fce6b1006ccf29985f431515c1c61deb84b6d3

Summary:
For history of the experimental version see folly/experimental/hazptr

Added:
- Support for deterministic schedule testing.

Fixes:
- Eliminates performance pathological cases for hazptr_array and hazptr_local construction when a thread cache can only partially satisfy the needed number of hazptr_holder-s.
- try_protect sets ptr to the value read from src on success. Otherwise, ptr may be used by the client code when it is invalid.

Code improvements including:
- Using separate files for different components.
- Eliminated all macro definitions except one.
- Simplified thread local structures.

Reviewed By: davidtgoldblatt

Differential Revision: D7565357

fbshipit-source-id: fb00fad7395148e66d6231a374b5b717574ed270

Summary: Core::detachFuture() & detachPromise() are called from dtors; make them 'noexcept'

Reviewed By: yfeldblum

Differential Revision: D7995163

fbshipit-source-id: 5c15e0516be2559a2ce1d283dc6fa75c2f506cd8

Summary:
Duplicate of line 22. Since the included header contains #pragma once it was doing nothing.

If it hadn't been a duplicate it would have been a problem because it was included within either a `namespace folly` or `namespace std...` block.

Reviewed By: yfeldblum

Differential Revision: D8031245

fbshipit-source-id: 4486859a11939a61d634c5603243785e6ec5a36c

Summary:
This method ensures the singleton exists with less overhead than
calling `try_get` or `try_get_fast`.

Reviewed By: yfeldblum

Differential Revision: D8020292

fbshipit-source-id: 470ebea9fdb07e46a469c820f9149d6fc5af5d2f

Summary:
There is a race with stop() and timeouts.  stop() may race with a thread timeout, and then block indefinitely on
joinStoppedThreads(), waiting to join the already joined timed-out thread.

To fix, just put everything behind a write lock, simplifying the code at the same time.

Reviewed By: davidtgoldblatt

Differential Revision: D7971779

fbshipit-source-id: eb4a898ca266ddda0f95033950ac041d1cc812ba

Summary:
A lot of functions in AsyncUDPSocket just use a CHECK for fd_ and fail
hard if ths socket hasn't bound to a fd. This adds an API for user to actually
find out if socket has bound without crash.

(Note: this ignores all push blocking failures!)

Reviewed By: siyengar

Differential Revision: D8026035

fbshipit-source-id: fc7bf9c1d6d7f045cd4dba94e3278d20d80de887

Summary:
`folly/test/DynamicTest.cpp` contained a duplicate definition of
`dynamic::print_as_pseudo_json()`.  This caused link errors due if this test
was linked against `json.cpp`.

Part of the problem here is that the `//folly:dynamic` library is somewhat of
a lie: the `dynamic::print_as_pseudo_json()` method is provided by the
`//folly:json` library rather than by `//folly:dynamic`.  This violates
assumptions made by autodeps.  Ultimately the right thing to do might be to
combine `//folly:json` and `//folly:dynamic` into a single rule.

Reviewed By: yangchi

Differential Revision: D8020841

fbshipit-source-id: e563611309c122680a5fc69c3f0da1ad5d4337b0

Fix "multiple definition of 'folly::dynamic::print_as_pseudo_json(std::ostream&) const'" linker error

Summary: Fix "multiple definition of 'folly::dynamic::print_as_pseudo_json(std::ostream&) const'" linker error

Reviewed By: yfeldblum

Differential Revision: D8016705

fbshipit-source-id: 7ef8306852afd8e765f22e657f44fbe67c66f6f5

Summary:
constexpr_clamp_cast<> provides sane numeric conversions from float point numbers to
integral numbers, and between different types of integral numbers.

When doing clamp_cast<DstT>(value), if `value` is in valid range of DstT, it will give correct result in DstT, equal to `value`.

If `value` is outside the representable range of DstT, it will be clamped to MAX or MIN in DstT.

Float NaNs are converted to 0 in integral type.

For comparison, static_cast<> will return garbage results when SrcT is float point number and it's outside DstT valid range. This is because float-cast-overflow is undefined behavior.

Reviewed By: yfeldblum

Differential Revision: D7716736

fbshipit-source-id: 5899e4b20cdd997d0c7b121ddaababbc8c748301

Summary: As per title.

Reviewed By: yfeldblum

Differential Revision: D8014022

fbshipit-source-id: df24a839804e9bb6c063994adbbde3544535b66c

Summary: [Folly] Use `throw_exception` in `folly/futures/`, replacing the one-off throw functions.

Reviewed By: Orvid

Differential Revision: D7965947

fbshipit-source-id: ed5855aaccf4aa07ecb40489db9fa92090df9016

Summary: [Folly] Let `Future` handle throwing `FutureNotReady`, since it is logically the thing that knows about specific exceptions in its contract.

Reviewed By: LeeHowes

Differential Revision: D7964855

fbshipit-source-id: 49f3220a6a41ebe87f577814c0940e00497092c0

Summary:
Update the folly::Init code to define a `--logging` command line flag, and call
`folly::initLoggingOrDie()` with the value of this command line during
initialization.

This is similar to the existing code that initializes the glog library.
(Programs can use both glog and folly logging together in the same program, and
I expect that many programs will do so as parts get converted to folly::logging
and parts remain using glog.)

Reviewed By: yfeldblum

Differential Revision: D7827344

fbshipit-source-id: 8aa239fbad43bc0b551cbe40cad7b92fa97fcdde

Summary: Currently, when we add or delete a `RequestData` instance to `RequestContext` for the first time, we do not call the `RequestData`'s `onSet` or `onUnset` callback. This means if the `RequestData` is doing tracking-related work, it may miss the beginning or the request until the first switch of `RequestContext` has happend, and in the end as well. This Diff fixes the issue by calling `onSet` and `onUnset` when adding or deleting.

Reviewed By: anakryiko

Differential Revision: D7978623

fbshipit-source-id: a524286371f890598bc773e957ef609e90086cea