Summary: This fixes folly::ThreadLocal perf after several refactorings which touched the code.

Reviewed By: ericniebler

Differential Revision: D3170761

fb-gh-sync-id: d44198f19aebc9f9f2588a65950efba0694e2a11
fbshipit-source-id: d44198f19aebc9f9f2588a65950efba0694e2a11

Summary:Currently GuargPageAllocator mmap's a memory block enough for 100 fiber stacks and then protects a bottom page for each stack. If FiberManager is used to process a single task, protecting 100 stacks doesn't make much sense, but is costly.

This change makes sure we protect a bottom page of every stack the first time given stack is requested.

Reviewed By: alikhtarov

Differential Revision: D3139944

fb-gh-sync-id: d9d724eaa0e65a227eac1d09a33018e6cb098aae
fbshipit-source-id: d9d724eaa0e65a227eac1d09a33018e6cb098aae

Reviewed By: yfeldblum

Differential Revision: D3174112

fb-gh-sync-id: 00a01f64fb43d2333818097bb2212e26e4dd5775
fbshipit-source-id: 00a01f64fb43d2333818097bb2212e26e4dd5775

Summary:We use `size_t` but didn't include `<stdlib.h>`.
Fix #385

Reviewed By: markisaa

Differential Revision: D3171643

fb-gh-sync-id: 1f915f9602c4be59a21cdf77b31fe18816f35f81
fbshipit-source-id: 1f915f9602c4be59a21cdf77b31fe18816f35f81

Summary: Have the AsyncSocket keep track of the timeout used for connecting and add a getter to retrieve it.

Reviewed By: hiteshk

Differential Revision: D3170625

fb-gh-sync-id: 61d0ecd8d975c49978a1cf222671aa16a2160499
fbshipit-source-id: 61d0ecd8d975c49978a1cf222671aa16a2160499

Summary: Add it to the Makefile.am

Reviewed By: Orvid

Differential Revision: D3168833

fb-gh-sync-id: 7184acc565ffebde0f93bc8bdce4d38ba1a79923
fbshipit-source-id: 7184acc565ffebde0f93bc8bdce4d38ba1a79923

Summary:Change the order of EXPECT_EQ parameters to
EXPECT_EQ(expected, actual). It will make gtest print
the correct error message if a test fails.

Reviewed By: yfeldblum

Differential Revision: D3161878

fb-gh-sync-id: 69e4536c6f396858d189a9d25904ef4df639ad49
fbshipit-source-id: 69e4536c6f396858d189a9d25904ef4df639ad49

Summary:[Folly] Some cleanups to `folly::EventBase` after converting to `folly::Function`.

* Fix up some comments referring to `std::function`.
* Remove the `SmallFunctor` bits - `folly::Function` takes over for that.
* Remove `runFunctionPtr` - it's unused.

Reviewed By: spacedentist

Differential Revision: D3155511

fb-gh-sync-id: 38c75d1254993f59c8eaa7826dc8d9facb50a3a1
fbshipit-source-id: 38c75d1254993f59c8eaa7826dc8d9facb50a3a1

Summary: This header is necessary for the placement new operator used in the construct method (line 265)

Reviewed By: yfeldblum

Differential Revision: D3156741

fb-gh-sync-id: 84a64821553b42d46fb70b5ee267a401028c94f2
fbshipit-source-id: 84a64821553b42d46fb70b5ee267a401028c94f2

Summary: Primarily to support slightly more flexible implementations of `split()`;

Reviewed By: ot

Differential Revision: D3116763

fb-gh-sync-id: 69023c0f26058516f25b9c1f9824055efc7021f9
fbshipit-source-id: 69023c0f26058516f25b9c1f9824055efc7021f9

Summary:We have a bunch of code that manually parses `folly::dynamic`s into program structures. I can be quite hard to get this parsing to be good, user-friendly, and concise. This diff was primarily motivated by the mass of JSON-parsing done by Bistro, but this pattern recurs in other pieces of internal code that parse dynamics.

This diff **not** meant to replace using Thrift structs with Thrift's JSON serialization / deserialization. When all you have to deal with is correct, structured plain-old-data objects produced by another program -- **not** manually entered user input -- Thrift + JSON is perfect. Go use that.

However, sometimes you need to parse human-edited configuration. The input JSON might have complex semantics, and require validation beyond type-checking. The UI for editing your configs can easily enforce correct JSON syntax. Perhaps, you can use `folly/experimental/JSONSchema.h` to have your edit UI provide type correctness. Despite all this, people can still make semantic errors, and those can be impossible to detect until you interpret the config at runtime. Also, as your system evolves, sometimes you need to break semantic backwards-compatibility for the sake of moving forward ? thus making previously valid configurations invalid, and requiring them to be fixed up manually.

So, people end up needing to write manual parsers for `dynamic`s. These all have very similar recurring issues:

 - Verbose: to get an int field out of an object, typical code: (i) tests if the field is present, (ii) checks if the field is an integer, (iii) extracts the integer. Sometimes, you also want to handle exceptions, and compose helpful error messages. This makes the code far longer than its intent, and encourages people to write bad parsers.

 - Unsystematic: sometimes, we use `if (const auto* p = dyn_obj.get_ptr("key")) { ... }`, other times we use `dyn_obj.getDefault()` or `if (dyn_obj.count())`, and so on. The patterns differ subtly in meaning. Exceptions sometimes get thrown, leading to error messages that cannot be understood by the user.

 - Imperative parses: a typical parse proceeds step by step, and throws at the earliest error. This is bad because (i) errors have to be fixed one-by-one, instead of getting a full list upfront, (ii) even if 99% of the config is parseable, the imperative code has no way of recording the information it would have parsed after the first error.

 `DynamicParser` fixes all of the above, and makes your parsing so clean that you might not even bother with `JSONSchema` as your first line of defense -- type-coercing, type-enforcing, friendly-error-generating C++ ends up being more concise. Besides all the sweet syntax sugar, `DynamicParser` lets you parse **all** the valid data in your config, while recording  *all* the errors in a way that does not lose the original, buggy config. This means your code can parse a config that has errors, and still be able to meaningfully export it back to JSON. As a result, stateless clients (think REST APIs) can provide a far better user experience than just discarding the user?s input, and returning a cryptic error message.

For the details, read the docs (and see the example) in `DynamicParser.h`. Here are the principles of `DynamicParser::RECORD` mode in a nutshell:
 - Pre-populate your program struct with meaningful defaults **before** you parse.
 - Any config part that fails to parse will keep the default.
 - Any config part that parses successfully will get to update the program struct.
 - Any errors will be recorded with a helpful error message, the portion of the dynamic that caused the error, and the path through the dynamic to that portion.

 I ported Bistro to use this in D3136954. I looked at using this for JSONSchema's parsing of schemas, but it seemed like too much trouble for the gain, since it would require major surgery on the code.

Reviewed By: yfeldblum

Differential Revision: D2906819

fb-gh-sync-id: aa997b0399b17725f38712111715191ffe7f27aa
fbshipit-source-id: aa997b0399b17725f38712111715191ffe7f27aa

Summary:[Folly] Use `folly::Function` in `folly::EventBase`.

`folly::Function` is moveable but noncopyable and therefore supports wrapping moveable but noncopyable lambdas - like the kind that arises when move-capturing a `std::unique_ptr`.

`std::function` is copyable - therefore it does not support wrapping such noncopyable lambdas.

Switching `folly::EventBase` to use it will allow callers to pass such noncopyable lambdas, allowing, e.g.:

```
auto numptr = folly::make_unique<int>(7); // unique_ptr is noncopyable
folly::EventBase eb;
eb.runInLoop([numptr = std::move(numptr)] { // therefore lambda is noncopyable
  int num = *numptr;
});
eb.loop();
```

This allows us to move away from the `folly::MoveWrapper` hack, which worked like:

```
auto numptr = folly::make_unique<int>(7); // unique_ptr is noncopyable
auto numptrw = folly::makeMoveWrapper(std::move(numptr)); // MoveWrapper is "copyable" - hacky
folly::EventBase eb;
eb.runInLoop([numptrw] { // therefore lambda is "copyable" - hacky
  int num = **numptrw;
});
```

We needed to do that hack while:

But neither condition is true anymore.

Reviewed By: spacedentist

Differential Revision: D3143931

fb-gh-sync-id: 4fbdf5fb77eb5848ed1c6de942b022382141577f
fbshipit-source-id: 4fbdf5fb77eb5848ed1c6de942b022382141577f

Reviewed By: luciang

Differential Revision: D3145033

fb-gh-sync-id: f37d0a2c6e22c1866134159b78722ff43d70918b
fbshipit-source-id: f37d0a2c6e22c1866134159b78722ff43d70918b

Summary: Use malloc/malloc.h there instead

Reviewed By: Orvid

Differential Revision: D3143768

fb-gh-sync-id: 3258b68df49cf7fbff28a3853663d39c147aee8b
fbshipit-source-id: 3258b68df49cf7fbff28a3853663d39c147aee8b

Summary: Sometimes variable shadowing is fine, sometimes it can cause subtle mistakes.

Reviewed By: yfeldblum

Differential Revision: D3140450

fb-gh-sync-id: acf7195ad65614d4f012bef8ce7dccb3a3038456
fbshipit-source-id: acf7195ad65614d4f012bef8ce7dccb3a3038456

Summary:We are jumping through some ugly hoops in the implementation of
fibers to allow for non-copyable functors/lambdas. We can get rid of
those by using folly::Function instead of std::function to store
functions in folly::fibers::Fiber.

This involves one observable interface change: the virtual function
folly::fibers::InlineFunctionRunner::run must take a
folly::Function<void()> argument instead of a std::function<void()>.

Reviewed By: andriigrynenko

Differential Revision: D3102711

fb-gh-sync-id: 56705b972dc24cc0da551109ed44732b97cb6e13
fbshipit-source-id: 56705b972dc24cc0da551109ed44732b97cb6e13

Summary: For exceptions aggregation conveniece

Reviewed By: philippv

Differential Revision: D3128132

fb-gh-sync-id: 48c72df364228c1d2c8f29161c2b85c131b4fea8
fbshipit-source-id: 48c72df364228c1d2c8f29161c2b85c131b4fea8

Summary:This is an ugly but occassionally useful hack to accomodate cases
where we want to propagate folly::Function but run into the brick wall that is
std::function

@override-unit-failures

Reviewed By: spacedentist

Differential Revision: D3118432

fb-gh-sync-id: 80ed56494dfcf60e0f266013d880107acabc7dcf
fbshipit-source-id: 80ed56494dfcf60e0f266013d880107acabc7dcf

Summary: Allow ASan to detect access to invalidated strings even when they are small (in-situ). See D3114022 for details.

Reviewed By: luciang

Differential Revision: D3118161

fb-gh-sync-id: 3b83e3309d510ee9721dd505d176b09bdb7fd42d
fbshipit-source-id: 3b83e3309d510ee9721dd505d176b09bdb7fd42d

Summary:Now that folly::Future uses folly::Function, we can use non-copyable
callbacks. That allows us to get rid of folly::MoveWrapper in the
implementaion.

This diff also enforces perfect forwarding in the implementation of
folly::Future, thereby reducing the number of times that a callable
that is passed to Future::then et al. gets passed by value.

Before folly::Function, Future::then(callback) has invoked the move
constructor of the callback type 5 times for small callback objects
(fitting into the in-place storage inside folly::detail::Core) and
6 times for large callback objects. This has been reduced to 5 times
in all cases with the switch to UniqueFunction. This diff reduces it
to 2 times.

Reviewed By: yfeldblum

Differential Revision: D2976647

fb-gh-sync-id: 9da470d7e9130bd7ad8af762fd238ef9a3ac5892
fbshipit-source-id: 9da470d7e9130bd7ad8af762fd238ef9a3ac5892

Summary:By using folly::Function instead of std::function to internally store
functions in FunctionScheduler, this diff allows users to pass
non-copyable lambdas to FunctionScheduler::addFunction.

All exisiting unit tests still pass. Also, passing std::function to
addFunction still works, as the std::function will be implicitly
converted (i.e. wrapped) in a folly::Function. However, this implies
a performance penalty.

Reviewed By: fugalh

Differential Revision: D3092944

fb-gh-sync-id: 1e8081e70d4717025f2eadbb6b6da173460d4373
fbshipit-source-id: 1e8081e70d4717025f2eadbb6b6da173460d4373

Summary:Instead of using preprocessor to disable SSO, use a default argument. Also
reimplement `insertImplDiscr` to mirror `insertImpl`.

Reviewed By: philippv

Differential Revision: D3130901

fb-gh-sync-id: a5b0ba97b3c7d91323be01ab617ca9b09dbb0fd2
fbshipit-source-id: a5b0ba97b3c7d91323be01ab617ca9b09dbb0fd2

Summary: The only thing that matters in it is dirname, and Windows doesn't have it.

Reviewed By: yfeldblum

Differential Revision: D2977655

fb-gh-sync-id: eb5703485089f5c66882fbc9cc142686214c1148
fbshipit-source-id: eb5703485089f5c66882fbc9cc142686214c1148

Summary: Windows doesn't have it, so stub it out for now, because not getting output to the system log is a minor inconvience, not a breaking issue.

Reviewed By: yfeldblum

Differential Revision: D2977537

fb-gh-sync-id: b05d9d3c722ad4f3953a979f91024602d17d3ed0
fbshipit-source-id: b05d9d3c722ad4f3953a979f91024602d17d3ed0

Summary: MSVC doesn't like this and complains about ambigious overloads.

Reviewed By: ericniebler

Differential Revision: D3107438

fb-gh-sync-id: 597ba85fdb7640b0a0fda352bf53ce31cf0e38b7
fbshipit-source-id: 597ba85fdb7640b0a0fda352bf53ce31cf0e38b7

Summary:The swap function belongs in the same namespace as Function: folly.
Also, to avoid amibiguities with a generic swap function in
<algorithm>, we need two variants: one for identical types of
folly::Function, and one for folly::Functions with different
configurations.

Reviewed By: ericniebler

Differential Revision: D3106429

fb-gh-sync-id: 11b04e9bc709d52016ac94c078278410f5ee43c6
fbshipit-source-id: 11b04e9bc709d52016ac94c078278410f5ee43c6

Summary:Update setFromPath() to accept a StringPiece rather than just std::string
or a plain const char*.

Also fix two other minor issues:
- Leave the old address untouched on failure.  Previously it could leave the
  SocketAddress in a partially updated state.
- Don't assume the input is nul terminated.  Previously the input code read
  one past the specified input length, and copied this into the address,
  assuming it was a nul terminator.  The new code explicitly writes a 0 byte.

Reviewed By: yfeldblum

Differential Revision: D3119882

fb-gh-sync-id: 3e2258f42034b4f470ade0a23ea085e132a3dd0f
fbshipit-source-id: 3e2258f42034b4f470ade0a23ea085e132a3dd0f

Summary: Because we don't have these builtins under MSVC, and not having to deal with the differences in API in the places that use these builtins is good.

Reviewed By: yfeldblum

Differential Revision: D2984842

fb-gh-sync-id: 34db7455debf81e4abffe57c154eb731ae097ff6
fbshipit-source-id: 34db7455debf81e4abffe57c154eb731ae097ff6

Summary: We don't have dirent.h on Windows, but we can emulate its behavior.

Reviewed By: yfeldblum

Differential Revision: D2978570

fb-gh-sync-id: af5ade0ea64ba22900440250e7125aa039a77f62
fbshipit-source-id: af5ade0ea64ba22900440250e7125aa039a77f62

Summary: The primary issue is that the pthread implementation we use for Windows defines `pthread_t` as a struct (yes, it is allowed to do this), which breaks a lot of things.

Reviewed By: yfeldblum

Differential Revision: D2862671

fb-gh-sync-id: 551569b6a9e2e374cf77e186e148b6b397f8f25f
fbshipit-source-id: 551569b6a9e2e374cf77e186e148b6b397f8f25f

Summary: `strndup` is defined on modern versions of OSX

Reviewed By: Orvid, grp

Differential Revision: D3122635

fb-gh-sync-id: 678792765addf2fb226e835d3b7a67e155ed6dc5
fbshipit-source-id: 678792765addf2fb226e835d3b7a67e155ed6dc5

Summary:Most header files in folly already used "#pragma once" to
protect against multiple inclusion. This diff removes old-style
ifndef/define/endif header guards and replaces them with
pragma once.

In some cases the defined symbol is tested in other header
files. In those cases the "#define" is kept.

Reviewed By: igorsugak

Differential Revision: D3054492

fb-gh-sync-id: 20aa0b9b926a30dd021e4b8f5440e8888874681c
fbshipit-source-id: 20aa0b9b926a30dd021e4b8f5440e8888874681c

Summary: This diff adds folly/docs/Function.md

Reviewed By: yfeldblum

Differential Revision: D3120617

fb-gh-sync-id: fecc0e507e05016aaac43ba981eab49431229bd7
fbshipit-source-id: fecc0e507e05016aaac43ba981eab49431229bd7

Summary:This adds basic print functions for FiberManager, Fiber and FiberManager map.
It also adds a global list of fibers to FiberManager. Fibers are only removed from that list on Fiber object destruction, so it shouldn't have any perf impact.

Inspired by tao/server/scripts/fiber_bt.gdb

FiberManager map example:
  (gdb) print_folly_fiber_manager_map
    Global FiberManager map has 2 entries.
      (folly::EventBase*)0x7fffffffdb60 -> (folly::fibers::FiberManager*)0x7ffff5b58480
      (folly::EventBase*)0x7fffffffd930 -> (folly::fibers::FiberManager*)0x7ffff5b58300

FiberManager example:
  (gdb) print_folly_fiber_manager &manager
    (folly::fibers::FiberManager*)0x7fffffffdbe0

    Fibers active: 3
    Fibers allocated: 3
    Fibers pool size: 0
    Active fiber: (folly::fibers::Fiber*)(nil)
    Current fiber: (folly::fibers::Fiber*)(nil)

    Active fibers:
      (folly::fibers::Fiber*)0x7ffff5b5b000   State: Awaiting
      (folly::fibers::Fiber*)0x7ffff5b5b300   State: Awaiting
      (folly::fibers::Fiber*)0x7ffff5b5b600   State: Awaiting

Fiber example: P56244621

Reviewed By: yfeldblum

Differential Revision: D3119616

fb-gh-sync-id: defa27b84aafbd4333b2ca301f07c226f0386f44
fbshipit-source-id: defa27b84aafbd4333b2ca301f07c226f0386f44

Summary:gcc-5.0 introduced non-relocatable strings in libgcc. Only treat gcc < 5 strings as relocatable. Enable relocation for fbstrings via typedef.

Re https://github.com/facebook/folly/issues/316, thanks tomhughes for reporting this.

Reviewed By: snarkmaster, ot

Differential Revision: D3115580

fb-gh-sync-id: d8aff947d55a0a0c57f6b997f651a190e05f2779
fbshipit-source-id: d8aff947d55a0a0c57f6b997f651a190e05f2779

Summary:The current implementation of `insertImpl` assumes that
`expand_noinit` does not reallocate if the `size() + delta <=
capacity()`, but D3114022 makes this assumption invalid when compiling
with ASan. It also doesn't guarantee exponential growth, so repeated
inserting at the end could trigger quadratic behavior.

The new implementation fixes the problems above, and it's much
simpler.

Reviewed By: yfeldblum, Orvid

Differential Revision: D3119813

fb-gh-sync-id: 946ebeeaf924a531f7f03fb7e79c75e352a50c58
fbshipit-source-id: 946ebeeaf924a531f7f03fb7e79c75e352a50c58

Summary: Alerts may be sent by clients, potentially letting us know why connections fail.

Reviewed By: siyengar

Differential Revision: D3117395

fb-gh-sync-id: bddf51f2399eb9e7e397981d5440adb3e815d6d2
fbshipit-source-id: bddf51f2399eb9e7e397981d5440adb3e815d6d2

Summary: Windows doesn't have it, so be nice and at least stub it out.

Reviewed By: yfeldblum

Differential Revision: D2984232

fb-gh-sync-id: 3e8871ab78c5d7c8785a52af24548245f842f19b
fbshipit-source-id: 3e8871ab78c5d7c8785a52af24548245f842f19b

Summary:Let's break OSX!

Alright, maybe not. Neither OSX nor Windows define malloc_usable_size, so we implement them based on what is available on the respective platforms.
This moves the implementation for OSX out of Portability.h and into the new header, so it likely breaks something on OSX, although I'm not sure what.

Reviewed By: yfeldblum

Differential Revision: D3019938

fb-gh-sync-id: df95faef09535098fb73b7b3479d7a73f6b49712
fbshipit-source-id: df95faef09535098fb73b7b3479d7a73f6b49712

Summary: Adding a benchmark for one the most common scenarios (used in Futures, EventBase, fibers etc).

Reviewed By: yfeldblum

Differential Revision: D3106365

fb-gh-sync-id: 8cb55959b3803e8836ab5a5bdf43fdfc1db02d4c
fbshipit-source-id: 8cb55959b3803e8836ab5a5bdf43fdfc1db02d4c