Summary: Always using 10000 is no good for concurrent tests.

Reviewed By: Orvid

Differential Revision: D9656798

fbshipit-source-id: 4fb7e48239e42c3b2acd2cc976cb74535c573ef6

Summary: [Folly] Do only single var decl per statement in `folly/test/DynamicConverterTest.cpp`.

Reviewed By: igorsugak

Differential Revision: D9653672

fbshipit-source-id: 08bb529cc90761c69dabd2e00f58f1767f07146a

Summary:
[Folly] Cut the extra braces in `folly/test/ExceptionTest.cpp`.

And reformat.

Reviewed By: igorsugak

Differential Revision: D9653533

fbshipit-source-id: dfeec84d322b32921a375721669ef5ac7a77099f

Summary: Forgot to include the paper number

Reviewed By: yfeldblum

Differential Revision: D9650597

fbshipit-source-id: 05c0720a18ee39a8085739a72fea269589e3c0fa

Summary:
There are efforts as part of the OlympicQos initiative to ensure QOS configuration for all network transfers. As part of these efforts, one of the requirements is the ensure TOS marking for server-->client traffic. Since the server does not have a way of knowing the priority of the response, the server needs to reflect the TOS settings from the client. The idea is the cache the client's SYN, derive the TOS from the same and set hte same TOS for the responses. Made changes to

1) setsockopt(TCP_SAVE_SYN) in AsyncServerSocket() for all listen sockets
2) Upon an accepted connection, getsockopt(TCP_SAVED_SYN) to get the TOS from IPv6 header and setsockopt to set TOS on the connected socket. Do this conditionallly (based on tosReflect_ setting and only for family IPv6)
3) Added API setTosReflect() to enable the feature and included flag on thrift server to control the same

Reviewed By: yfeldblum

Differential Revision: D9566136

fbshipit-source-id: ad61fed4e205893e2b7c5654e1b4f21ed5472c06

Summary: [Folly] `FOLLY_DECLARE_REUSED`.

Reviewed By: marksantaniello

Differential Revision: D9626462

fbshipit-source-id: 23be21e2c6a314b3f65bbfca660958e596915b53

Summary:
This diff allows `folly::dynamic` to lookup and mutate using any type that is
convertible to `StringPiece` (i.e., `const char*`, `std::string`, `fbstring`)
without having to construct a `dynamic` object (which requires copying the
string).

Previously (D8299863), we SFINAE'd and took `StringPiece` by reference. This
caused linker errors for non-member constexpr string constants like the
following which are odr-used.
  struct S {
    static constexpr auto kXXX = "xxx";
  };
  d.at(kXXX); // linker error
Hence in this diff, we template on types that are convertible to dynamic but
NOT to StringPiece, and add a separate overload which takes StringPiece by
value.

Reviewed By: yfeldblum

Differential Revision: D9540388

fbshipit-source-id: 529de31585200947b8b92fcf3126442a8ed51a73

Summary: [Folly] Apply `clang-format` to `folly/compression/`.

Reviewed By: igorsugak

Differential Revision: D9625742

fbshipit-source-id: 5098aa62a3f5d9cc12da0932ab4f33d5a11111dc

Summary: [Folly] Apply `clang-format` to `folly/system/`.

Reviewed By: igorsugak

Differential Revision: D9625730

fbshipit-source-id: 88f93231f77497d933233d4267c6983558baeb25

Summary: [Folly] Apply `clang-format` to `folly/fibers/`.

Reviewed By: igorsugak

Differential Revision: D9625725

fbshipit-source-id: e00f6406db994a1a3f17c1d23b415bea13495e99

Summary: [Folly] Apply `clang-format` to `folly/lang/`.

Reviewed By: igorsugak

Differential Revision: D9625707

fbshipit-source-id: 345dcfbeb67981e928093b03f75125089c4210c1

Summary: [Folly] Apply `clang-format` to `folly/portability/`.

Reviewed By: igorsugak

Differential Revision: D9625664

fbshipit-source-id: fa63f3e3a719c22ea88cd14172d34e360353145f

Summary: [Folly] Do conditional compilation for `MPMCPipeline` using C++ via selecting a base class v.s. using the preprocessor.

Reviewed By: Orvid

Differential Revision: D9625568

fbshipit-source-id: 2e0d7195bb7fa4f44cf8b99ed3d7564b18393b73

Summary: Remove stop() comment since it's not accurate

Reviewed By: yfeldblum

Differential Revision: D9600621

fbshipit-source-id: 599551efc775a622d5a5b4a45ead1f3dcbf10abb

Summary:
It was missing things.

Closes https://github.com/facebook/folly/issues/925

Reviewed By: yfeldblum

Differential Revision: D9631683

fbshipit-source-id: 3ea572ecf5b4d640b6cd2f376cd4c6916e43ed1d

Summary:
[Folly] Make `ThreadPoolExecutorTest.DynamicThreadsTest` not timing-sensitive.

The timing-sensitive version of the test sometimes fails under load, and is observed in some internal `folly-oss` jobs.

Rewrite the test to avoid using long sleeps to try to approximate synchronization. Use a barrier and spin-yield loop instead.

Differential Revision: D9628339

fbshipit-source-id: fef752ad15dbff853cfa7cbaac1081580007754c

Summary: [Folly] Remove some `[[noreturn]]` functions - many of them can be replaced with `throw_exception`.

Reviewed By: Orvid

Differential Revision: D9624966

fbshipit-source-id: e5e472c660677112edf99e30e36d78c1b1718b52

Summary: [Folly] Let `StaticMeta` have a deleted dtor - a compile-time enforcement that the dtor will not be invoked, to replace the existing runtime non-enforcement.

Reviewed By: elsteveogrande

Differential Revision: D9624686

fbshipit-source-id: 9185defbea343e4123e66c8d0d38f20c561aebea

Summary: [Folly] Check the behavior of ThreadPoolExecutor destruction regarding execution or dropping of outstanding tasks.

Differential Revision: D9627237

fbshipit-source-id: 4bb3c1480cc94ceeb3e550f52c298e3d2dcf71c5

Summary:
Contains a subset of the functions described in p1135r0, with some additions
for timed waiting that essentially extends the futex interface to work with
non-standard futex widths

In the regular 32 bit case, we fall back to the existing folly futex()
implementation.  In all other cases, we use folly::ParkingLot to mimic futex()

Reviewed By: djwatson

Differential Revision: D9381922

fbshipit-source-id: faf84e105e1d44a6dd6034e25440fcb3eb664846

Summary: [Folly] Write `/* static */ if` as `if /* constexpr */`. Once the time comes to jump to C++17, all that remains is to remove the comments.

Reviewed By: aary

Differential Revision: D9627826

fbshipit-source-id: 56078e727a3a4420851f1c441783b80da55a8e6c

Summary:
Folly warning correction -- the author clearly intended unsigned math

Fixes #923.

Reviewed By: yfeldblum

Differential Revision: D9508962

fbshipit-source-id: 0ffa1b25f7c427030987cc5ba03c6e7f160a335b

Summary: [Folly] Less duplication in `AtomicStruct` test.

Reviewed By: aary

Differential Revision: D9601770

fbshipit-source-id: c11fe49e081079f4c33b9d80c9b44437202390d9

Summary: [Folly] Cut explicit `decltype`-based return types in `folly/gen/File-inl.h` - they only required pre-gcc49.

Reviewed By: Orvid

Differential Revision: D9601484

fbshipit-source-id: 8750091a9ef3e4ba8b8f4e26a96b28fe8d242e7a

Summary:
[Folly] Pull out a local variable in `MMapAlloc`.

Moves the conditional compilation out of a compilcated expression and into a separate full-statement by itself.

Reviewed By: Orvid

Differential Revision: D9598437

fbshipit-source-id: 770c835f729e72b880aadcefb8aae7ead2754bd9

Summary: Because.

Reviewed By: yfeldblum

Differential Revision: D9568442

fbshipit-source-id: 0a1ecf1904499afa9069d345d70279a527e31b1a

Summary: As a mitigation for an `folly::dynamic` iteration order dependency on iOS.

Reviewed By: yfeldblum

Differential Revision: D9614129

fbshipit-source-id: 09a822a32327fd5389e8e4d169ab3af4f428426c

Summary:
I want to access the list of deps without the install
instructions attached later in this stack of diffs in `shell_builder.py`.

Refs: https://github.com/facebook/watchman/pull/639

Reviewed By: snarkmaster

Differential Revision: D9552419

fbshipit-source-id: 333cfa0c33bc1aa623de7ac3546a8f1785869a4f

Summary:
This is a bit of a tricky issue; here's the bug in python 2.7
that prevents using the function-like syntax:
https://bugs.python.org/issue21591

The nature of this bug is that scoped exec only works with nested functions
when using the `exec` statement in 2.7.   That is a syntax error in python 3,
which only allows using the `exec` function.

To make the function syntax work in 2.7 we therefore need to avoid using
lambdas or nested functions and have to resort to using a global variable
to act as the "closure".

Refs: https://github.com/facebook/watchman/pull/639

Reviewed By: snarkmaster

Differential Revision: D9552413

fbshipit-source-id: a2cb9d325e18fb6fb06b9e284a84f0a0c80cfe7e

Summary: Per T33416893, D9441402 caused a regression in FB4A. Goal: reenable lvalue-qual Future::onError(). Make it deprecated to minimize the chances of anyone creating new callsites.

Reviewed By: yfeldblum

Differential Revision: D9597490

fbshipit-source-id: 9c6a34c37c8f576e2497164bd440cdbc3565920d

Summary:
[Folly] Use `static_cast` v.s. C cast in `folly/Expected.h`.

And don't have empty namespaces - at least have a comment.

Reviewed By: igorsugak

Differential Revision: D9596449

fbshipit-source-id: 6b5b712d7d46eba4a83b7f62c0628a8bcda1f722

Summary: [Folly] Formatting of `FOR_EACH` etc uses.

Reviewed By: Orvid

Differential Revision: D9595505

fbshipit-source-id: bdb506f8904de2577c42ada6755a2af4f3efb095

Summary:
In `folly::symbolizer::ElfFile`, we `mmap` the file and access the file content via `mmap`-ed pointer directly. This is nice and efficient, but causes the whole program crashes on `SIGBUS` is the underlying frile is broken.

This Diff changes the first read of the file, for ELF magic number, to use `read` syscall and check return result, instead of accessing the `mmap`-ed pointer directly. This way APIs like `ElfFile::openNoThrow` would return error instead of crashing if the underlying file is broken.

It still doesn't solve the problem if the file becomes broken between we read the ELF magic number and we read file content with the `mmap`-ed file pointer later. But this is the best effort to check for the case with minimal performance cost.

Reviewed By: yfeldblum

Differential Revision: D9474328

fbshipit-source-id: 969e0323fcb4011e6a080d914452c9215d2fd13c

disable on android, fix getAllocatedMemorySize test to account for implementation-defined bucket_count for default containers

Summary:
This serves as a mitigation until we look further into the android crashes we
discovered in android instrumentation tests.

The bucket_count of the default constructed `unordered_map` is
implementation-defined. This diff takes that into account in the
`getAllocatedMemorySize` test.

`unordered_map|set` impl in Android NDK does not match the assumptions made in
`visitAllocationClasses`. Temporarily disable the tests with that method.

Reviewed By: mzlee

Differential Revision: D9578626

fbshipit-source-id: 355940b026f00f783bf65c2f1e10725d5aa920d3

Summary:
I noticed this while testing some changes; the comment
doesn't seem to match up to the `raise` statement, and attempting
to throw a list fails in python.

Reviewed By: strager

Differential Revision: D9566018

fbshipit-source-id: 563e43dfd30867993cc2a0f816db3b835a52e6df

Summary:
When running on macOS there is no /etc/issue.  When running
with `set -e` enabled, we don't want this to break the build.

Neither of these is possible today, but will be later in this diff stack.

While I'm in here, also print out the cmake version.

Refs: https://github.com/facebook/watchman/pull/639

Reviewed By: snarkmaster

Differential Revision: D9552416

fbshipit-source-id: 432b5b70678908de9f79787d37c3615a7f8649d3

Summary: Support for building Folly with autoconf is going away; this isn't going to keep working, so remove it.

Reviewed By: yfeldblum, philipjameson

Differential Revision: D9569019

fbshipit-source-id: 1fe99453c741ff9fc8632bbcfdd69df6b1f86d01

Summary: This means that Windows builds no longer require pthreads in order to run tests. (previously we were relying on the pthreads implementation we were using to provide semaphore.h for us)

Reviewed By: yfeldblum

Differential Revision: D9568125

fbshipit-source-id: 712f542c9a29627f72b41e1def2b79d591a33317

Summary: It's not a perfect mapping, but it's good enough.

Reviewed By: yfeldblum

Differential Revision: D9569405

fbshipit-source-id: 7559fb94f8a57717b5a594be0c05a99ac88c6c6e

Summary: The `(get|set)rlimit` stub implementations don't support setting memory limits, and I have no intention of implementing it, so disable it, as there are no meaningful memory limits on Windows to begin with.

Reviewed By: yfeldblum

Differential Revision: D9569424

fbshipit-source-id: 738d73f90eb33957ca4a80cb9b3c568034ef141e