Reviewed By: yfeldblum

Differential Revision: D16639497

fbshipit-source-id: ddefeb12e2f83bc86e410092233efb2cdeaacbd0

Summary: This can be useful to avoid the cost of re-throwing exceptions.

Reviewed By: yfeldblum, LeeHowes

Differential Revision: D16628722

fbshipit-source-id: 434842d99369d0293c4eeb894e6feac15ec16173

Summary: Defer uses inline continuations by default, if bound executors match.

Reviewed By: andriigrynenko

Differential Revision: D15300706

fbshipit-source-id: fc17c70ed9956442007a1cc2f6a082f75c555ad4

Summary:
After this diff, DeferredExecutor participates consistently in executor inline behaviour by being special cased in the core.

DeferredExecutor is no longer an executor, and is hence no longer special cased in the Future code. This is replaced with a variant of DeferredExecutor and Executor in Core.

Reviewed By: yfeldblum, andriigrynenko

Differential Revision: D15836529

fbshipit-source-id: 8324ba1de57e85fc757ecc3b431bf71858868a0d

Summary: Simple code reorg. DeferredExecutor class moved into Core as it is later becoming functionality of Core rather than of the Future.

Reviewed By: andriigrynenko

Differential Revision: D15836530

fbshipit-source-id: 5a14a6c9332666b275e39796cae5f32c96edacd5

Summary: addFrom on DeferredExecutor allows the DeferredExecutor to know what executor is running the work enqueuing to it, which it can check against its internal state to run the work inline if the executors match.

Reviewed By: andriigrynenko

Differential Revision: D15836532

fbshipit-source-id: 40c35d976b90072c88bb544360b059c1b151d4d0

Summary:
Allows adding directly to a KeepAlive if the KeepAlive is consumed in the process.

This allows simultaneous consumption and propagation of the KeepAlive into the callback, propagating information about the executor running the callback without the cost of reference counting the KeepAlive.

Reviewed By: andriigrynenko

Differential Revision: D15836528

fbshipit-source-id: 8f9e6f6ec47ad294391741d25fce68a85a429919

Summary: [Folly] `cacheline_align_t`, `cacheline_align_v` - since it is theoretically possible on some platforms for `alignas(hardware_destructive_interference_size)` to be a compile-time error.

Reviewed By: aary

Differential Revision: D16598497

fbshipit-source-id: d0dfe99b726cec5dba96e26147e73e494e62a895

Summary:
[Folly] Add unsafe-unlocked getter to `Synchronized`.

Provided as a backdoor for call-sites where it is known safe to be used. For example, when it is known that only one thread has access to the `Synchronized` instance.

To be used with care - this method explicitly overrides the normal safety guarantees provided by the rest of the `Synchronized` API.

Reviewed By: chadaustin, simpkins

Differential Revision: D8178205

fbshipit-source-id: 648ee392a43a06333452f72d992ec42ba4a1a73c

Summary: [Folly] Fix visibility attributes that gcc objects to in xlog by moving them from lambdas to proper functions.

Reviewed By: nbronson

Differential Revision: D16578895

fbshipit-source-id: 021fd10ef01bb610dc6f20d9a9a98a5a8e11df71

Summary: The TSX intrinsics should only be compiled and used when compilers and platforms support them. This diff wraps the intrinsics and adds checks on TSX support to server this purpose.

Reviewed By: nbronson

Differential Revision: D16585187

fbshipit-source-id: 6490c0da7f6221caaa4529bfcebb261cfd99cc7d

Summary:
The Python chained `IOBuf` has undesirable behavior on cyclic pattern data. For example, when we have a data chain with
```
chain = make_chain([IOBuf(b"aaa"), IOBuf(b"aaaa")])
```
`b"".join(chain)` will yield `b"aaa"` rather than `b"aaaaaaa"`.

The root cause to this bug is because in the `__iter__` method of the Python `IOBuf`, the original code checks whether the circular chain has been traversed by the `!=` operator (`self != next`), which has been overridden by `__richcmp__` function. The rich comparator then invokes the comparator in C++, which compares the underlying data in a `IOBuf` chain rather than their reference locations. In the above example, therefore, `chain == chain.next` would return `True`. However, in `__iter__`, in order to check whether the traversal is back to the head of the chain, we should compare by reference rather value.

Reviewed By: yfeldblum

Differential Revision: D16589600

fbshipit-source-id: 3b03c4d502bdc385edca3502949be03440543a21

Summary: [Folly] Extract Executor::KeepAlive bitmask constants to a non-template base class so that they can be used from elsewhere.

Reviewed By: LeeHowes

Differential Revision: D16617648

fbshipit-source-id: eb3ceddfa8c492bb4ddd7de0046bb5c2c6b4e762

Reviewed By: yfeldblum

Differential Revision: D16594071

fbshipit-source-id: 78755bc64f22b28f89958064184c1fdf35b1bb85

Summary: needed for openr

Reviewed By: simpkins

Differential Revision: D16010070

fbshipit-source-id: 6d485fa7e4e321e6cd23d9894b38c1ecc7574665

Summary: Don't constanly rebuild exceptions that are always the same.

Reviewed By: jrahman

Differential Revision: D16016383

fbshipit-source-id: b0b24ed39e6c41f95c61369c90a3ac83abfd1a66

Summary:
While developing on a project it is often convenient to be able to invoke its
build manually, rather than always needing to re-run `getdeps.py`.  This
updates the CMakeBuilder to also emit a script that can be used to manually
run CMake outside of `getdeps.py`.

The CMakeBuilder is the only builder that this really matters for right now,
as pretty much all of the projects where we do first-party development use
CMake for their build system.

Reviewed By: pkaush

Differential Revision: D16477399

fbshipit-source-id: c8a14af158af7b32d6c799ef685b037e68b748ff

Summary:
Move code that computes project hashes to ManifestLoader.  ManifestLoader is
the only class that has all of the information necessary to compute the
project hashes correctly.  The ManifestLoader object can also cache previously
computed hashes, so that we don't have to keep computing hashes for projects
over and over again.  Previously the `BuildOptions.compute_dirs()` function
would end up re-computing hashes for all dependencies each time it was called.

Reviewed By: strager

Differential Revision: D16477401

fbshipit-source-id: ce03642114f91ce4f859f612e6b2e747cf1653be

Summary:
The ManifestLoader contains all of the state needed to create a fetcher
object, so define a helper method on this object to create a fetcher.

Reviewed By: strager

Differential Revision: D16477395

fbshipit-source-id: 6de0942fe6b8de26c18c82bf99343f5467dc006a

Summary:
Add a new ManifestLoader class to handle loading manifests and computing
dependencies.

For now the main thing this class does is maintain the `manifest_by_name`
mapping.  In subsequent diffs we should be able to move some additional logic
into this class, which will help clean up the code and eliminate some redudant
work.  In particular, we can have this class cache project hashes, which will
avoid re-computing hashes over and over again for the same projects as we do
in many cases today.  We should also be able to save and re-use some of the
project dependency ordering computation in some cases as well.

Reviewed By: strager

Differential Revision: D16477400

fbshipit-source-id: f06f62f77d8443fccaa69fe4c1306e39c395b325

Summary:
Update `BuildOpts.compute_dirs()` to use the correct project-specific manifest
context when computing project hashes.  Previously it was incorrectly using
the initial project's context when evaluating all dependencies.  This would
result in some projects potentially seeing the wrong values for variables that
may change from project to project (like `test`).

Reviewed By: pkaush

Differential Revision: D16477398

fbshipit-source-id: 6c23f5e5e19b2402000a138b3920b79044446041

Summary:
Add a ContextGenerator class so that we actually use the correct per-project
context when loading projects and computing dependencies.

Previously commands like `build` and `test` would change the contexts for each
project as they iterated through and performed the build.  However, they did
not do this when first loading the projects.  This could cause them to use
different context values when loading dependencies than when performing the
build.  For instance, this could cause issues if a project depends on
`googletest` only when testing is enabled, as the code previously did not set
the "test" parameter when evaluating dependencies.

Reviewed By: pkaush

Differential Revision: D16477396

fbshipit-source-id: c1e055f07de1cb960861d19594e3bda20a2ccd87

Summary:
Check that all variable names are valid when loading manifest files.
This ensures that getdeps.py will error out if someone makes a typo in a
variable name, rather than treating it as never equal to anything.

Reviewed By: pkaush

Differential Revision: D16477397

fbshipit-source-id: 030e0642ff4a08db8eb74a0a0223e03d53e4880f

Summary:
This allows us to use a custom deleter to track the lifetime of exceptions in
the case of smart pointers which may be persisted outside catch blocks.

Reviewed By: yfeldblum

Differential Revision: D16572937

fbshipit-source-id: ad2f0a981a2b31ef4b17a902719b684f538c29c7

Reviewed By: JunqiWang

Differential Revision: D16497024

fbshipit-source-id: aa0e42b4bfba3c70769514131daa30df84851126

Summary: bring this in line with other facebook opensource projects

Reviewed By: saifhhasan

Differential Revision: D16577367

fbshipit-source-id: d762658505f824cc180c55ea4485cecf525b8fdc

Summary: Add convenience aliases for std::pmr versions of folly::sorted_vector_set and folly::sorted_vector_map.  I also add the traditional one-argument constructor taking the allocator.

Reviewed By: yfeldblum

Differential Revision: D16523576

fbshipit-source-id: 33514db62eb8dc1b603f45204977751091e528d9

Differential Revision:
D16568423

Original commit changeset: 533353ef623a

fbshipit-source-id: fa2a719d34ee422a3e20a6e69f3c5edfb0f00257

Summary: [Folly] Fix `-Wshadow` violations found by gcc.

Reviewed By: meyering

Differential Revision: D16578896

fbshipit-source-id: 3ffe6089ad7b78d9c01d085029b55ec94d6adcff

Reviewed By: yfeldblum, VitalyKalinkin

Differential Revision: D16568423

fbshipit-source-id: 533353ef623a4bd2629a3c1782b6228f598d6c01

Summary:
Simply linking `GMOCK_LIBS` into a binary was not telling CMake that the binary depends on `gmock` being built. So, let's add that dependency explicitly.

This wasn't breaking in production because we typically build with `-j 4`, and `gmock` was getting built before the first dependent binary would attempt to link.

Also, since `rsocket` bundles its own `gmock`, it is just a waste of time to compile a system-level gmock. It's not a real dependency.

NB: The change in `fbcode_builder.py` is needed because now that `rsocket` no longer depends on anything on Github, driver programs that were unconditionally setting `projects_dir` started to fail to build `rsocket`.

Reviewed By: simpkins

Differential Revision: D16461572

fbshipit-source-id: 1e95654e96256e7ed37d42e702b5433bf2fe5328

Summary:
This diff adds unit tests that print out the wall, user, and
system time for two backend implementations of CacheLocality gathering.

Reviewed By: yfeldblum

Differential Revision: D16547404

fbshipit-source-id: 5b5467965eb21bc50ad7248c0badd814edc71d68

Summary: [Folly] Fix non-portable annotation syntax in `F14TestUtil.h`, using portable `[[noreturn]]` v.s. non-portable `__annotation__((__noreturn__))`. The non-portable syntax works with GCC and Clang, but is not recognized by MSVC.

Reviewed By: jkedgar, terrelln, Orvid

Differential Revision: D16553166

fbshipit-source-id: a652d31d31eca773009885f648f44ea3798e1460

Summary: It's desirable to know if a given server socket is accepting or if a regular UDP socket is reading.

Reviewed By: dddmello

Differential Revision: D16529348

fbshipit-source-id: 2e9a2e505f97344d21449aed7c9effc298877e41

Summary:
Cache locality information under /sys is dispersed across a
very large number of files. This is a problem for short-lived processes
due to direct overheads and lock contention in the kernel. This diff
switches to a heuristic strategy that infers the interference pattern from
/proc/cpuinfo instead of computing it exactly. This doesn't necessarily
produce exactly the correct cache hierarchy info, but it yields the
correct topological sort for machines that have only core-local and
socket-local cache locality.

Differential Revision: D16459331

fbshipit-source-id: a322c126d1a4775d015bfb81451dbc6ad6fcc0fd

Summary:
flake8 (in Python 2.7 mode) complains that `typing` is mentioned in type annotations but is not defined:

```
fbcode_builder/getdeps/buildopts.py:251:21: F821 undefined name 'typing'
    subst_mapping,  # type: typing.Mapping[str, str]
                    ^
fbcode_builder/getdeps/buildopts.py:253:5: F821 undefined name 'typing'
    # type: (...) -> typing.Optional[str]
    ^
2     F821 undefined name 'typing'
2
```

Import `typing` explicitly to silence this warning.

Because `typing` may be unavailable, import it conditionally. (Because it's only referenced in comments, failing to import `typing` should have no effect at run time.)

Reviewed By: snarkmaster

Differential Revision: D16435696

fbshipit-source-id: 78a4a7b07acc46aa998f02b54b1a6e52c1daafde

Summary:
We need to wrap coro_.resume() with a RequestContextScopeGuard so that
if the task changes the request context, we restore it after the coroutine
suspends itself.

Reviewed By: lewissbaker

Differential Revision: D16497375

fbshipit-source-id: 439a6f41b2294e56d2ec1b0013a6013524b5ae6a

Summary:
Allow builds for 64-bit ARM (MSVC) to use the same implementation of `__builtin_clzll` as ARM and x86 architectures.
Pull Request resolved: https://github.com/facebook/folly/pull/1191

Test Plan: Imported from GitHub, without a `Test Plan:` line.

Reviewed By: yfeldblum

Differential Revision: D16350573

Pulled By: Orvid

fbshipit-source-id: f25087b3a53de09767f505ee8c7d971a6d957189

Summary:
Note that we only provide the strong exception guarantee for single key insertion and emplacement when `std::is_nothrow_constructible<value_type>::value == true`. We could `static_assert()` that in `sorted_vector_set`, since it seems to hold for all contbuilds. But for `sorted_vector_map` there are quite a few contbuilds broken by that static assert. So I've chosen not to static assert in `sorted_vector_set` for consistency.

`insert()`, `emplace*()` and `erase()` take `const_iterator` in compliance with C++11. `erase()` has an overload that takes an `iterator` in C++17, but since we are backed by a vector I don't think that is necessary.

Add `emplace_hint()` variants.

Reviewed By: yfeldblum

Differential Revision: D16427231

fbshipit-source-id: 4bcf6fd01d9b1320548d12b152d1cef9291c2dd2

Summary:
While wrapping my head around the TDigest code, I made a few small
drive-by clean-ups.

Reviewed By: yfeldblum

Differential Revision: D16478284

fbshipit-source-id: aaf289a57ef633f0f1527263beffb7c0edf596d1