Summary:
add try_take_for to BlockingQueue, utilizing new LifoSem::try_wait_for method.
To be used to kill threads after an idle period in CPUThreadPoolExecutor.

Reviewed By: magedm

Differential Revision: D7477268

fbshipit-source-id: 76ea7fa1fc3b4b5f7a7529ce1424a91eef2ac7b1

Summary:
Linux's futex checks the waiter count atomically before grabbing the lock.  This results in much better perf if there are no waiters and you call futexWake().
Add the same improvement to ParkingLot.

Reviewed By: yfeldblum

Differential Revision: D6639337

fbshipit-source-id: 6e03eff07e4c9ce327bc66d0aabd8cec051a5b19

Summary:
According to headers, `pthread_getname_np` and `pthread_setname_np` are available since iOS 3.2 and macOS 10.6.
However, guarding just on min macOS version isn't going to work for iOS/watchOS/tvOS, because it's very rarely defined.
So add another case for when `__IPHONE_OS_VERSION_MIN_REQUIRED` is defined and is above iOS 3.2, which will also make watchOS and tvOS work.

Reviewed By: yfeldblum

Differential Revision: D7648204

fbshipit-source-id: 1e608cdad3f39242eff79fbb6da8e560eeee720a

Summary:
If the default inline executor is used for folly::window, and all the futures generated by the callback are also inline, then all spawn function calls (and their associated Futures) are added to the call stack and can overflow. This fixes that issue by using QueuedImmediateExecutor, which will put the executor callback into a thread-local queue, allowing the current function to exit and destruct all objects in its stack frame. Then, only the first call to QueuedImmediateExecutor in a thread is responsible for running each subsequent callback and will do so sequentially.

This will end up allocating dynamic queues that reach a max size of two. It is possible to avoid a dynamic queue by writing a new Executor that only stores a single function, if that is too much of a performance cost.

Reviewed By: yfeldblum

Differential Revision: D7597731

fbshipit-source-id: cd8899e6c086ecff812050f7241e80f39c546706

Summary: Avoid creating a loop in the ThreadEntryList in onThreadExit

Reviewed By: djwatson

Differential Revision: D7632840

fbshipit-source-id: 409358aae16a8aed4ae654ebfeca994385b43a9e

Summary:
Neither GCC or Clang are able to optimize a `memcpy` where
the length is known to be smaller than a word: a library call is
generated. This is a simple implementation of a subword
`loadUnaligned` to be used to read the tail of a buffer when reading
one word at a time.

Reviewed By: luciang

Differential Revision: D7631978

fbshipit-source-id: d5719580e6bef82aa8e3fbdce8f873b569527a3a

Summary: This function is no longer used anywhere.

Reviewed By: Orvid

Differential Revision: D7597741

fbshipit-source-id: 296acfceb59293bbafc6f6bf9c67d951cf41f7be

Summary:
Update `make_temp_dir()` to use `shutil.rmtree()` to clean up rather than just
`os.rmdir()`.  If the function using this context manager throws it may leave
behind files inside the temporary directory, and it is useful to always clean
those up.

Also update `read_fbcode_builder_config()` to work regardless of the current
working directory.  Previously this function only worked if the current working
directory was the directory containing the configuration file.  This makes sure
that `read_fbcode_builder_config()` calls are always treated as relative to the
directory containing the configuration file being read.

Reviewed By: yfeldblum

Differential Revision: D7597743

fbshipit-source-id: 817063104081336fcc0a9f825beb74d7d4377499

Summary: Creates a new SlidingWindow data structure for general purpose use in stats code.

Reviewed By: anakryiko

Differential Revision: D7597047

fbshipit-source-id: 227438055029c2bacb5edf2943ebc9cc66fdfe36

Summary: Use unbounded queue in NotificationQueue (prev changes: D7164130, D7251343, D7278275, D7359322, D7471153, D7606076)

Reviewed By: djwatson

Differential Revision: D7408628

fbshipit-source-id: 04ae235d9b3de603e28115e11fe98fe7e9aa1f8c

Summary:
fixes #819.
Closes https://github.com/facebook/folly/pull/820

Reviewed By: yfeldblum

Differential Revision: D7642932

Pulled By: Orvid

fbshipit-source-id: c49168ddf111c851d0218b74fbbdf405e9a68fc3

Summary:
- Expose an API to erase an item from `F14Set` and do something with it before it is destroyed
- This includes (but is not limited to) being able to `std::move` it to another container while avoiding a copy
- Change erase implementations to forward to `eraseInto(...args, emptyCallback)`

Reviewed By: shixiao

Differential Revision: D7588928

fbshipit-source-id: 5e9c2bda216c5c219c333282daadd2170e6a1eba

Summary:
Adds the ability for UDP sockets to optionally
listen to errors that happen on the socket,
for example ICMP errors.

In this iteration, we explicitly do not register for
read events if an error callback is set. So even
if you set an error callback to get error events
you must register a read callback.

Reviewed By: yfeldblum

Differential Revision: D7632877

fbshipit-source-id: 43c922d0145e1da97b993f5bf4058c20d3469deb

Summary:
Clang format files related to AsyncUDPSocket.
This helps keep changes consistent.

Reviewed By: yfeldblum, lnicco

Differential Revision: D7632865

fbshipit-source-id: 0761f0cfd06417334e680ca0880280d65480fe7e

Summary: [Folly] Tweak `SharedPromise` and `FutureSplitter` docs, in particular to reflection recent changes to execution semantics.

Reviewed By: LeeHowes

Differential Revision: D7630725

fbshipit-source-id: 68da70cd86d131aa09c1909d90a793868039e84f

Summary: [Folly] All comparison predicates for `IOBuf`, not just `EqualTo`.

Reviewed By: djwatson, stevegury

Differential Revision: D7596700

fbshipit-source-id: 2d5cef2797dc75d4fb76c2d13e4a938f738da163

Summary: [Folly] comparison-to-predicate adapters for comparisons which return `ordering`.

Reviewed By: stevegury

Differential Revision: D7596712

fbshipit-source-id: 1aeccf0d38eec27cfa88af2908f3114523d6b6b5

Summary: [Folly] `to_ordering` to convert from `int` to `ordering`.

Reviewed By: stevegury

Differential Revision: D7596709

fbshipit-source-id: cf1316a7560b9398b94d1c834258f09a797a98f8

Summary: [Folly] An `ordering` type, lifted from `FixedString`.

Reviewed By: stevegury

Differential Revision: D7596713

fbshipit-source-id: ba19e1e36e75b946e14f6db9c72ac1ac3c7270e9

Summary:
D7479058 changed the exception_tracer code to export their targets in the main
folly-targets.cmake file, but I forgot to remove the `install()` command
installing the old `folly-exception-tracer-targets.cmake` file.

This would cause `cmake configure` to exit with an error if it detected it
could built the exception_tracer code, but it would still generate all of the
necessary Makefiles correctly, so subsequent `make` commands would still work.

Reviewed By: yfeldblum

Differential Revision: D7626710

fbshipit-source-id: f015032c3c42cf2911fcaf9b355fc97616ff20e1

Summary:
The cmake globbing then excluding strategy implies that if the directory
in which we are compiling has a folder named `build` on its path like `/home/alice/project/build/folly/...`, everything breaks.
Note that it is the same for `tools` and `test`, I also fixed the tools one, but the test once has many files.... Using a file list would be the real solution.
Closes https://github.com/facebook/folly/pull/815

Reviewed By: simpkins

Differential Revision: D7624035

Pulled By: yfeldblum

fbshipit-source-id: a3f1d2f82e55e070d13acc5c394d8cfdfabeb7ed

Summary:
The existing histogram approach for estimating quantiles is inadequate for common use cases. This diff introduces TDigests for estimating quantiles of streaming values.

The current histogram approach suffers from two issues:
  - The user needs to know something about the universe of values, U, and the underlying distribution in order to get reasonably accurate values
  - Because the histogram is not biased, estimating tail quantiles such as p99.9 requires a large amount of memory.

t-digests (https://github.com/tdunning/t-digest/blob/master/docs/t-digest-paper/histo.pdf) are biased digests that allow for accurate estimates of tail quantiles, while using a small amount of memory and without requiring the user to know anything about U.

The scaling function in the paper is slow, because it uses arcsin. This approach uses a sqrt function instead. Details of why this works in the comments of the TDigest file.

Reviewed By: simpkins

Differential Revision: D7512432

fbshipit-source-id: 01460433ec7380fed88e5e942ef15f032dd794bb

Summary: [Folly] Fix `constexpr` predicate names, which have embarrassing typos.

Reviewed By: igorsugak

Differential Revision: D7605352

fbshipit-source-id: e04a87fd375f09472b020380bf446c7236a39e51

Summary:
Fix broken Android build

(Note: this ignores all push blocking failures!)

Reviewed By: michalgr

Differential Revision: D7606076

fbshipit-source-id: 0cf0426a34cf62b045acb932d42a8e4a3cd5da71

Summary:
When trying to compile `folly/experimental/exception_tracer` with gcc-7.2, I get the following error:

```
ExceptionCounterLib.cpp: In instantiation of '{anonymous}::throwHandler(void*, std::type_info*, void (*)(void*))::<lambda(auto:9&)> [with auto:9 = std::unordered_map<long unsigned int, folly::exception_tracer::ExceptionStats>]':
../../../folly/Synchronized.h:162:20:   required from 'auto folly::SynchronizedBase<Subclass, (folly::detail::MutexLevel)1>::withWLock(Function&&) [with Function = {anonymous}::throwHandler(void*, std::type_info*, void (*)(void*))::<lambda(auto:9&)>; Subclass = folly::Synchronized<std::unordered_map<long unsigned int, folly::exception_tracer::ExceptionStats>, folly::RWSpinLock>]'
ExceptionCounterLib.cpp:132:4:   required from here
ExceptionCounterLib.cpp:127:21: error: uninitialized variable 'info' in 'constexpr' function
       ExceptionInfo info;
                     ^~~~
In file included from ../../../folly/experimental/exception_tracer/ExceptionCounterLib.h:22:0,
                 from ExceptionCounterLib.cpp:16:
../../../folly/experimental/exception_tracer/ExceptionTracer.h:30:8: note: 'struct folly::exception_tracer::ExceptionInfo' has no user-provided default constructor
 struct ExceptionInfo {
        ^~~~~~~~~~~~~
../../../folly/experimental/exception_tracer/ExceptionTracer.h:31:25: note: and the implicitly-defined constructor does not initialize 'const std::type_info* folly::exception_tracer::ExceptionInfo::type'
   const std::type_info* type;
                         ^~~~
```

Fixed by initializing `ExceptionInfo::type`.

Compiling with gcc 7.2 now works.
Closes https://github.com/facebook/folly/pull/812

Reviewed By: Orvid

Differential Revision: D7598341

Pulled By: yfeldblum

fbshipit-source-id: 0a8859b6d5f174484c8f2148de59a2a592dc2b25

Summary: [Folly] Rename `IOBufEqual` to `IOBufEqualTo` to correspond with the standard naming of `std::equal_to`.

Reviewed By: Orvid

Differential Revision: D7596716

fbshipit-source-id: 2e87aa4b62a002520bdd53b5b7eaba7d9bd822da

Summary:
in Linux environment, `find_package(gflass)` searches `Findgflags.cmake`.
And it cannot find  https://github.com/facebook/folly/blob/master/CMake/FindGFlags.cmake , because linux is case-sensitive environment in filenames and cause error.

```
CMake Error at CMake/folly-deps.cmake:34 (message):
  Unable to determine the target name for the GFlags package.
Call Stack (most recent call first):
  CMakeLists.txt:65 (include)
```

For linux programmers, this line should find `FindGFlags.cmake` file which exists in the same directory.
Closes https://github.com/facebook/folly/pull/805

Reviewed By: yfeldblum

Differential Revision: D7505908

Pulled By: Orvid

fbshipit-source-id: c30097a25569d293df9a1bfecbee84cece260306

Summary:
[Folly] Function objects for performing `constexpr` comparisons for C++11 or gcc49.

After upgrading to C++14 and dropping support for C++11, these may be removed because the corresponding function objects in the C++14 standard library are `constexpr`-invocable.

Reviewed By: Orvid

Differential Revision: D7590351

fbshipit-source-id: a8702115a66836aff25761f374b8a2d7dbe12074

Summary: Delay threadEntry destruction until all instances have completed StaticMetaBase::onThreadExit

Reviewed By: djwatson

Differential Revision: D7471153

fbshipit-source-id: 709a54ed2d616c5dc906b923e9fdd862ac87de9c

Summary: The domain tests seem broken.  Digging in, it seems to be uninitialized ThreadCachedInts variables.  Initialize them.

Reviewed By: yfeldblum

Differential Revision: D7505520

fbshipit-source-id: aca8a6f5b38271bf06ea8a6377bfb692328f1e33

Reviewed By: djwatson

Differential Revision: D7571438

fbshipit-source-id: 1484e5e323c7d57a96ad66d46fd999ce51e69951

Summary: Similar to D7429494, this can be a CAS and save some latency for some readers.

Reviewed By: magedm

Differential Revision: D7505578

fbshipit-source-id: eeeb06c2969a74084edd3df26cf2d135b8b0e132

Summary:
If FOLLY_SETTING_DEFINE(...) is put into .cpp file right now you will get linker error about undefined references.

According to http://en.cppreference.com/w/cpp/language/inline
  2) The definition of an inline function or variable (since C++17) must be present in the translation unit where it is accessed (not necessarily before the point of access).

Hence the linker error. This diff fixes that.

Reviewed By: alikhtarov, mnv104

Differential Revision: D7567469

fbshipit-source-id: 02049e4881e50c414aeb506745e02eed1256daad

Summary:
Update `cmake_configure()`, `cmake_install()`, and `fb_github_cmake_install()`
to support specifying the directory where CMakeLists.txt is found, relative to
the directory where the build is being performed.  Previously these functions
where hardcoded to assume that CMakeLists.txt was always found at '..'

Reviewed By: snarkmaster

Differential Revision: D7540689

fbshipit-source-id: efd3d044345fadc0346e436c01d0a247e1b6fd70

Summary:
Add a macro to help users define the `getBaseLoggingConfig()` function.

While I would prefer to avoid macros if possible, this seems worthwhile.  This
saves 4 or 5 lines of boilerplate code in each program that sets a custom base
logger setting.  It also reduces the likelihood of a developer accidentally
having a typo in the function name, which would still build successfully but
not have the desired results.

Reviewed By: chadaustin

Differential Revision: D7457652

fbshipit-source-id: 1c316c7ea6949c16bd7b61c0440cc1ee69ecb83e

Summary:
Add folly::initLoggingOrDie(), which behaves like `initLogging()` but handles
errors by printing a message to stderr and then calling `exit(1)`.

Most programs are expected to initialize logging early on during program
start-up, and will likely use a command line argument or other configuration
value as the logging settings.  If an error occurs they usually want to print
the message to stderr and exit unsuccesfully. `initLoggingOrDie()` makes it
easy to perform these steps without requiring programs to write their own
exception handlers to handle errors from `initLogging()`.

Reviewed By: chadaustin

Differential Revision: D7164830

fbshipit-source-id: d688046f1b6240c948e4aaabfc9fb24ba3d9ff89

Summary:
- Let's enforce a single definition. Also makes the implementation much simpler.
- Collect all static info about a setting in a Metadata struct

Differential Revision: D7544210

fbshipit-source-id: d3f2ef617bd817d355a37e11fa9425b6a4259384

Summary:
Currently, the memory used by a map instance is available only via computeStats(),
which is quite expensive. A lightweight method is needed for getting the map memory
footprint on a hot code path.

Reviewed By: nbronson

Differential Revision: D7544180

fbshipit-source-id: 3cbc98bb9f1c5e7f75fd9eb7f89016580fbb6523

Summary: Which only supports C++11 style constepxr.

Reviewed By: simpkins

Differential Revision: D7546457

fbshipit-source-id: 52db98379096dd3b25f656e7e62238ce10537bf7

Summary:
According to http://en.cppreference.com/w/cpp/language/operators
  The overload of operator -> must either return a raw pointer or return an object (by reference or by value), for which operator -> is in turn overloaded.

Hence this change.

Reviewed By: mnv104

Differential Revision: D7548832

fbshipit-source-id: d98ff1f96bc00dc6e2e5cb084b96091a61421627