Summary: This patch adds support for `std::variant` to `folly::variant_match` but without including `<variant>`.

Reviewed By: yfeldblum

Differential Revision: D23196572

fbshipit-source-id: 9d5b519958b04c13cc024d393247c73cec00ac67

Summary:
We should only be capturing `folly::RequestContext` in HHWheelTimer's callback
objects (already  happens). And avoid capturing it in underlying AsyncTimeout.

Notice that `AsyncTimeout::scheduleTimeout`/`scheduleTimeoutHighRes` APIs capture
current `folly::RequestContext` and `AsyncTimeout::cancelTimeout`
releases the captured `folly::RequestContext`

Now, the way HHWheelTimer works, it re-uses one instance of AsyncTimeout for
multiple timeouts. This opens up possibility for leaking lifetime of RCTX.
Imagine first  request A and then B schedule timeouts using same HHWheelTimer
instnace. A's RCTX is captured by underlying AsyncTimeout. Next, request A is
complete and we need to cancel its timeout. Notice, however that A's RCTX will
not be destroyed as long as B is still running. That's the lifetime management
leak.

Reviewed By: jordalgo, andriigrynenko

Differential Revision: D23260733

fbshipit-source-id: 0bdf79db40d34d6682e89e87ce30811ffccc8cbf

Summary: Remove implementation based on read locking (replaced by implementation based on hazard pointers).

Reviewed By: yfeldblum

Differential Revision: D20881677

fbshipit-source-id: da89eee6a92fa34fa1c111a421bf5368873ccd57

Summary: Pull Request resolved: https://github.com/facebook/folly/pull/1421

Reviewed By: yfeldblum

Differential Revision: D23023218

Pulled By: Orvid

fbshipit-source-id: 7cc182319dece37b9f2db42958717fa70b526025

Reviewed By: yfeldblum, lewissbaker

Differential Revision: D23240860

fbshipit-source-id: 74aab6151c9c44159a2d3fc4e041b475b69e29e7

Summary: Do not call scheduleImmediateRead() if the EVB has changed

Reviewed By: yfeldblum

Differential Revision: D23234912

fbshipit-source-id: 8394700a53b4e2fa925519271c1c27fc0ce11a85

Summary:
Changed the way how enums generated in thrift-py3:

1. Use C++ map to store name to unique Enum Python instances cache
2. Dynamically find instances from cache (or create one and add to cache) instead of populating all instances at module load time
3. use `__getattr__` and `__getitem__` instead of code-gen all the getters

Reviewed By: asp2insp

Differential Revision: D23182622

fbshipit-source-id: 75baef997ee2d604b4ef7606f282cfc3306208ac

Reviewed By: yfeldblum

Differential Revision: D23223918

fbshipit-source-id: 72fb70ccabb57229298515f55334807ad2dfff36

Summary: As titled.

Differential Revision: D23216535

fbshipit-source-id: 4d278cd1f5d4c661eddda265d92f51d391192404

Summary:
[Folly] Re-port memrchr, avoiding build-system detection.

Proper detection at build-system time is challenging.
* Some platforms do not define `memrchr`.
* Some platforms define it as a C function.
* Some platforms define it as a pair of C++ functions.

Proper porting requires proper build-system detection.

Rather than going down that rabbit-hole, we can wrap `memrchr` and do compile-time detection using SFINAE.

This change adds a new strategy. A subsequent change may remove the old strategy.

Reviewed By: Orvid

Differential Revision: D22125127

fbshipit-source-id: 2da7cca571de335b2cff43ff7846ab04d90136ef

Reviewed By: yfeldblum

Differential Revision: D23178843

fbshipit-source-id: d9fade6f7a5f8472bffb7a40d285eaf7646ac0e8

Summary: [Folly] Minimize conditional compilation in `AsyncUDPSocket`.

Differential Revision: D23172282

fbshipit-source-id: 6ead5cf2232b9db0f709080666262ccc4b5984da

Summary:
The examples showing `thenValue(executor, ...)` is outdated since the executor-taking form of `then` have been `delete`d.

The example showing racy execution of continuations of `Future`s is no longer relevant. It used to be a valid example when the `Future`s used to be "activated" on destruction, but they no longer have this behavior.

Fixes #1422.

Reviewed By: LeeHowes

Differential Revision: D23181958

fbshipit-source-id: 46d2bedd4fef344734d3eb15eb295b9a98969665

Summary:
This came from a [request from Fred Emmott](https://fb.workplace.com/groups/osssupport/permalink/4116491528399431/), but also is something we should do due to the GitHub UI redesign:

Nowadays on GitHub, tags are not shown in the sidebar (just the count) and instead GitHub Releases are prioritized and the most recent *release* is shown.

Secondly, the Source Code zip archives for tags/releases are created on-the-fly on GitHub. This can cause problems for downstream projects that need consistent content hashes of the archive for security/reproducability reasons. This action also creates its own archives (zip and tar.gz) and attaches them to the release.

Reviewed By: fredemmott

Differential Revision: D23167073

fbshipit-source-id: 463e9d93a2c4af260ac3c4dbc1ceab7894add714

Summary:
- Added a commandline flag to ignore testpilot even when it's available
- Fixed an oversight that caused getdeps to return 0 even if ctest tests have failed.

Reviewed By: wez

Differential Revision: D23161362

fbshipit-source-id: 1ed97c481315e2b39f1128680386390930194970

Summary: This diff allows the user to specify an optional onClosed callback for AsyncPipe, analogous to the callback accepted by thrift's ServerStreamPublisher.

Reviewed By: andriigrynenko

Differential Revision: D23024437

fbshipit-source-id: 546cc7e0ea1099cadfa085f5dd2fb9085b88f695

Summary: folly/experimental/symbolizer requires libunwind. Do not enable it unless libunwind is available.

Reviewed By: yfeldblum, luciang

Differential Revision: D22964401

fbshipit-source-id: d71991346927fbf18167637770b2143c03b16476

Summary:
Task -> semi has to lift void to unit because semifutures don't support void.
Semi -> task (via co_await) does not drop unit. It probably should because roundtrip properties are nice, but on the off chance someone actually wants to pass a unit around that would become impossible.
But for the result type the Try<void> is already implicitly convertible to Try<Unit>, as those types aren't meaningfully different (and not at all in this context where empty Try is not possible). So we can return Try<void> from co_awaitTry instead of Try<Unit>, and then generic code that wants `co_yield co_result(co_await co_awaitTry(SemiFuture<T|Unit>))` from a `Task<T|void>` will work.

Reviewed By: andriigrynenko

Differential Revision: D23004090

fbshipit-source-id: 5254e2930ac02952abd92fd2856c2a7d3e952e20

Differential Revision:
D22499183 (https://github.com/facebook/folly/commit/24364356023d04f966a0ef6fc7536363932bdc82)

Original commit changeset: b2af8d94024c

fbshipit-source-id: a6bd56cdb4d4b73618a9374528371c5b18a0df10

Summary:
basic_fbstring provides an allocator template paramater for
compatibility with basic_string, but ignores it. Assert at compile
time that a custom allocator isn't used.

This would have prevented the regression introduced in D22970418 (https://github.com/facebook/folly/commit/abb4ece62120ff32aaf28dc4bbfdc6a2f2b41f78) and
fixed in D23035304 (https://github.com/facebook/folly/commit/8b650246220717f27046593b5819fbaa60427bcb).

Reviewed By: luciang

Differential Revision: D23144806

fbshipit-source-id: c541f274e97736b9408651a66088bcbb78025539

Summary:
Fixes a few documentation errors that get surfaced when compiling with the -Wdocumentation flag.

Full explanation of errors with examples here:
https://clang.llvm.org/docs/DiagnosticsReference.html#wdocumentation

Reviewed By: yfeldblum

Differential Revision: D23119100

fbshipit-source-id: 4e84f624e4d81914b78211af4860540498f8110c

Summary: try_dequeue() used to consume from the queue but not from the semaphore, causing the two to diverge and breaking subsequent dequeue() calls.

Reviewed By: yfeldblum

Differential Revision: D23160324

fbshipit-source-id: f709733d984f51abc53595d38dbef3daffbfa6d9

Summary: [Folly] Move `DeterministicSchedule` helper alias `TLState` to the source, and outline all uses into the source.

Reviewed By: nbronson, luciang

Differential Revision: D23157071

fbshipit-source-id: 957576a61a796f46da781e02419a1229e681d4c1

Reviewed By: mzlee

Differential Revision: D23066827

fbshipit-source-id: 6362acb7a6680a24871859d7874a9caec0a1538e

Summary:
Changed the way how enums generated in thrift-py3:

1. Use C++ map to store name to unique Enum Python instances cache
2. Dynamically find instances from cache (or create one and add to cache) instead of populating all instances at module load time
3. use `__getattr__` and `__getitem__` instead of code-gen all the getters

Reviewed By: yfeldblum

Differential Revision: D22499183

fbshipit-source-id: b2af8d94024c3591c3d7faaa15cbd026a134cc78

Summary: [Folly] Remove wrong use of `errno` in `AsyncUDPSocket`.

Reviewed By: markisaa

Differential Revision: D23153435

fbshipit-source-id: f567aa97ebe243ff5b6ad99e300212df290800bb

Summary:
Previously, there was a race between exitingSems_[...] = tls.sem (in beforeThreadExit) and exitingSems_[...].post() (in joinAll). The race exists because exitingSems_ is a map that is not protected by the "shared access" synchronization mechanism used by DeterministicSchedule (i.e. it does not sit between a beforeSharedAccess and afterSharedAccess pair). Specifically, the main thread can be reading exitingSems_ while a child thread is writing to it.

This diff fixes that: after all threads are ready to exit (this part isn't changed), we first wait for the joiner thread itself to be scheduled, and then, it will post the semaphore to wake up the child thread to run its thread-local destructors, and at the same time call .join() on the child. All this runs in a "critical section" between beforeSharedAccess and afterSharedAccess in the joiner thread.

For clarity also changed the child thread to use descheduleCurrentThread, and to make the DeterministicSchedule destructor no longer call beforeThreadExit since it makes little sense.

Reviewed By: yfeldblum

Differential Revision: D23119075

fbshipit-source-id: 22bc6ffd9a9d888cc09bdee066dcac7f05b29fa7

Summary: [Folly] Reimplement `SignalSafeElfCache::Path` minimally so that it meets the expectations of `std::basic_string` as used by the rest of `SignalSafeElfCache`.

Reviewed By: chadaustin, luciang

Differential Revision: D23152224

fbshipit-source-id: 48c7a7277d00a928d3de2937699ad112d6d560c2

Summary: [Folly] Cut an extra `rdtsc` in `DistributedMutex` spin.

Reviewed By: aary

Differential Revision: D21947987

fbshipit-source-id: 8d6dca75f5290581a886385cefe994c19af839e1

Summary:
It's difficult to use C++ allocators if they are not default-constructible.  The non-default-constructibility "infects" the container, and any class that holds the container (transitively).

In D21967640 (https://github.com/facebook/folly/commit/f0471228b89be4a399afcf0e9692c84754d06875), I made `folly::SysArenaAllocator` default constructible, and adopted the behavior of merely throwing `std::bad_alloc` in the event that we didn't *actually* initialize it with a `folly::SysArena`.  In hindsight, it's not clear this was an improvement.  All it did was move a compile-time problem to run-time.

What I do here is a bit better.  I enhance `CxxAllocatorAdaptor` such that it has two modes:
- **Normal / classic mode:** non-default constructible.
  - Guaranteed at compile-time to have a non-null "inner allocator"
- **`FallbackToStdAlloc` mode:** default-constructible.
  - Passes allocations through to `std::allocator` if no "inner allocator" provided

Using this support, I introduce a new allocator, `folly::FallbackSysArenaAllocator`, which is similar to `SysArenaAllocator` *except* that it uses the fallback mode and is therefore default-constructible.

This is useful in a large codebase where we want to use arena-allocation on a specific *instance* of a container type -- and not every single instance.

Reviewed By: yfeldblum

Differential Revision: D22938039

fbshipit-source-id: feaba12d76564a386a03e759eedf06a77bbc8942

Summary:
Now that the subprocess output is no longer piped, we can see a weird situation
where the command being run is displayed after the compilation step:
  + cd /data/users/xavierd/scratch/dataZusersZxavierdZfbsource/fbcode_builder_getdeps/build/eden && \
  [1/13] rust_job_pool
      Finished release [optimized] target(s) in 0.75s
  [12/13] Install the project...
  -- Install configuration: "RelWithDebInfo"
  -- Installing: /data/users/xavierd/scratch/dataZusersZxavierdZfbsource/fbcode_builder_getdeps/installed/eden/bin/edenfs
  -- Set runtime path of "/data/users/xavierd/scratch/dataZusersZxavierdZfbsource/fbcode_builder_getdeps/installed/eden/bin/edenfs" to ""
  -- Up-to-date: /data/users/xavierd/scratch/dataZusersZxavierdZfbsource/fbcode_builder_getdeps/installed/eden/bin/edenfsctl
  -- Up-to-date: /data/users/xavierd/scratch/dataZusersZxavierdZfbsource/fbcode_builder_getdeps/installed/eden/lib/libbackingstore_rs.a
  -- Up-to-date: /data/users/xavierd/scratch/dataZusersZxavierdZfbsource/fbcode_builder_getdeps/installed/eden/lib/libbackingstore.a
  -- Up-to-date: /data/users/xavierd/scratch/dataZusersZxavierdZfbsource/fbcode_builder_getdeps/installed/eden/include/eden/scm/lib/backingstore/c_api/HgNativeBackingStore.h
  -- Up-to-date: /data/users/xavierd/scratch/dataZusersZxavierdZfbsource/fbcode_builder_getdeps/installed/eden/include/eden/scm/lib/backingstore/c_api/RustBackingStore.h
  + /data/users/xavierd/scratch/dataZusersZxavierdZfbsource/fbcode_builder_getdeps/installed/cmake-Ncng4tsJb6gdOu40ggy14-YtgNQD43 (https://github.com/facebook/folly/commit/f47621ec247c8009d8a8aae8823f0de7c27a70fe)k5ev0n-FXq99I/bin/cmake \
  +      --build \
  +      /data/users/xavierd/scratch/dataZusersZxavierdZfbsource/fbcode_builder_getdeps/build/eden \
  +      --target \
  +      install \
  +      --config \
  +      Release \
  +      -j \
  +      24

This is a bit awkward. Flushing stdout's buffer allows for the ordering to be
correct.

Reviewed By: wez

Differential Revision: D23079405

fbshipit-source-id: e2bf25b098d6ab4a788a5ec07deb635a42cae18c

Summary: Original commit changeset: 2d4ad67a4a01

Reviewed By: yfeldblum

Differential Revision: D23064000

fbshipit-source-id: 88d8d550772c2045f61d5a355cf33c9d8df7d046

Summary: Found because I saw the error message

Reviewed By: magedm

Differential Revision: D23019298

fbshipit-source-id: 6744e9e6d0c21cf76db756f68b3e948a044e6888

Summary: [Folly] Fix TSAN-reported race in `ShutdownSocketSet` test where the shutdown does not wait for the server.

Reviewed By: yfeldblum

Differential Revision: D23057822

fbshipit-source-id: 4b682e6e80cff503e85cb2b663f8399b69265a77

Summary:
The folly::fibers GDB extension, specifically the `info fibers` command,
depended on a specific format of the std::unique_ptr pretty printer
output. This pretty printer format changed between GCC 7 and GCC 9, with
the latter omitting the target address. However, the new pretty printer
does expose an iterator for the child target address that we can use
instead.

Differential Revision: D23063364

fbshipit-source-id: 6125edaba9abcfd0b7a9c0741a18854637c97804

Summary:
`std::allocator<void>::allocate` is not invocable as policy. Let folly allocators behave the same.

Switch existing sites to parameterizing the allocators over `char` instead, since `std::allocator<char>::allocate` is invocable.

Reviewed By: yfeldblum

Differential Revision: D22986960

fbshipit-source-id: ca1ff3cb1785029b6b6f59ddc4046b1b105d8bd6

Summary:
The hinted forms of insert_or_assign were forwarding with
std::move rather than std::forward, which has the potential to elevate
a non-const lvalue ref to an rvalue ref.

(Note: this ignores all push blocking failures!)

Reviewed By: yfeldblum

Differential Revision: D21475865

fbshipit-source-id: e91e759762cf6f045c843cd7de1402953e6afb61

Summary: Original commit changeset: 8d5e96feea4d

Reviewed By: yfeldblum

Differential Revision: D23046629

fbshipit-source-id: e24c4a0ba5f40a584646a602f618e881dbf0f344

Summary:
basic_fbstring unfortunately just ignores the allocator. Switching it to use std::basic_string.

This is necessary because ElfCache must be async-signal-safe, but basic_fbstring ignores the allocator and always uses standard malloc/free, which are not safe async-signal-safe.

Reviewed By: yfeldblum

Differential Revision: D23035304

fbshipit-source-id: 7b4e77a3764701366556a5fe5a352f952b052ed4

Summary:
These are all test setup issues:
 - EventBase needs to be destructed before objects that callbacks may refer to.
 - A socket should not be operated on by a thread that is different from the event base it is associated with (in this case, the socket was being closed on the RSA computation thread).

Reviewed By: yfeldblum

Differential Revision: D23043485

fbshipit-source-id: 733868317aef09e32e79169aedb84a988de3bb41