Summary: AsyncSocketSendmmsgIntegrationTest.PingPongRequest fails during stress runs with multiple jobs due to the 5ms timeout in `UDPClient::sendPing()`. Increasing the timeout fixes the issue.

Reviewed By: yfeldblum

Differential Revision: D29623870

fbshipit-source-id: ad6ee0356001fbded9ce01ddc099b708d1477899

Summary: `RequestContext::StaticContextAccessor` acts as a guard, preventing all threads with a `StaticContext` from being destroyed (or created).

Reviewed By: yfeldblum

Differential Revision: D29536635

fbshipit-source-id: 008a612ec65aa7095450f4d0ab522e8ecd864548

Summary:
Codegen for copy and move assignment is suboptimal when the vectors involved are inline, in particular if the destination is default-constructed (which is very common) and even more so if the value type is trivially copyable.

The main optimization is that if the storage is inline and the type is trivially-copyable, we can just copy the whole storage, regardless of the size of the container. This avoids a branchy loop, and it's usually just a handful of MOVs.

While at it, optimized all the related code paths by avoiding delegating to `swap()` and `assign()` when possible, which introduce further branches that are hard to optimize. Also move and copy in place when the vector is non-empty, avoiding unnecessary destructions.

Also fixed a couple of bugs:
- The move constructor did not clear its argument when inline, inconsistently with the move assignment and `std::vector`
- The move assignment operator passed its heap storage to the argument instead of freeing it.

Reviewed By: yfeldblum

Differential Revision: D29592519

fbshipit-source-id: 6281cdda775568d28619afea8b7ca2fb168c7e5d

Summary: It is unused. An upcoming Concurrency TS will have a replacement.

Reviewed By: mshneer

Differential Revision: D29597486

fbshipit-source-id: a108b945ce32eb17cedefad89630c9171fc5c9c2

Summary:
BSD readlink doesn't have -f.

I'm not using TravisCI, however docker is still convenient for reproducing builds locally.

Reviewed By: yns88

Differential Revision: D29523333

fbshipit-source-id: e01169f3eabca7b8baec95bc70fe119cad201b35

Summary: Implement C++20 `contains()`.

Reviewed By: yfeldblum

Differential Revision: D29556436

fbshipit-source-id: 1c3a8e6e9a07f6e47491b063f88df6bf9da2d87b

Reviewed By: yfeldblum, mshneer

Differential Revision: D29578762

fbshipit-source-id: 2aa9af8f68825997db8b71fc97dda5477fa57413

Summary:
[Forward iterators are required to be default constructible](https://en.cppreference.com/w/cpp/named_req/ForwardIterator). So as of today, `IteratorAdaptor` is lying!

We should default construct the wrapped iterator into a [singular iterator](https://eel.is/c++draft/iterator.requirements#general-7):
> Iterators can also have singular values that are not associated with any sequence. Results of most expressions are undefined for singular values; the only exceptions are destroying an iterator that holds a singular value, the assignment of a non-singular value to an iterator that holds a singular value, and, for iterators that meet the `Cpp17DefaultConstructible` requirements, using a value-initialized iterator as the source of a copy or move operation.

Reviewed By: yfeldblum, mshneer

Differential Revision: D29578765

fbshipit-source-id: 7f2f0762b17f0b1a056532fc5db2abdd76cca3ea

Summary: Let's give them better names than `first` and `second`.

Reviewed By: yfeldblum, mshneer

Differential Revision: D29536636

fbshipit-source-id: bbb37f2e1c0c51dd1ce2a0cfeca85399f409c415

Summary: Fix BatchSemaphore to post baton for a multiple waiter in a list as long as there are token available.

Reviewed By: yfeldblum

Differential Revision: D29430799

fbshipit-source-id: 98b0a616d0ce863108dcf331e491fd2cc12429d1

Summary:
Define a single implementation type `lock_base` which handles all cases, including unique/shared/upgrade locks and including sans-state/with-state locks. Add `unique_lock_base`, `shared_lock_base`, and `upgrade_lock_base`.

Revise `upgrade_lock` simply to derive `upgrade_lock_base`. We may use `upgrade_lock` as an example for specializing `unique_lock` and `shared_lock`

Remove `ProxyLockableUniqueLock` since `unique_lock_base` absorbs it. Let the `unique_lock` specializations for `DistributedMutex` inherit `unique_lock_base` instead.

Add lock invokers for every lock, try-lock, unlock, and lock-transition member. Were these used only internally to implement the lock-policy types and the lock-transition functions they might be left in detail but there may be broader use-cases for at least some of them. Putting the invokers in this header is consistent with proper placement since this header is intended to own all lock primitives and facilities.

Revise the lock-transition functions to handle cases where the from and to lock types are sans-state/with-state. Expand the set of lock-transition functions for completeness: lock-transitions include x->s, x->u, u->s, u->x; while try-lock-transitions include s->x, s->u, u->x.

Reviewed By: aary

Differential Revision: D28767313

fbshipit-source-id: 153adc8270f0f4338db6acf544b8d358556d6f49

Summary:
Reenables dynamicconfig loading with eden backingstore. Previously it
broke edenfs-windows-release, but we believe the
opensource/fbcode_builder/manifests/eden tweak has fixed it.

Reviewed By: xavierd

Differential Revision: D29561192

fbshipit-source-id: 775dd21d177f3baa09b0192e7d3f7231008c3766

Differential Revision: D29497151

fbshipit-source-id: 64ad1adbd68d10066fc65ddc41e9cff5ef3c6b53

Summary:
For whatever reason, sc_testpilot default to --return-zero-on-failures, which
means that test failures are simply not reported properly to the signal box.
Fix that by explicitely passing --return-nonzero-on-failures

Reviewed By: fanzeyi

Differential Revision: D29509158

fbshipit-source-id: ef991f91df48e99769f96ffd8d7015cdf507c723

Summary: Import or backport `std::reinterpret_pointer_cast` into folly. The relevant guard is the use of the C++17 language and, if feature-test macros are exported, `__cpp_lib_shared_ptr_arrays >= 201611`.

Reviewed By: yfeldblum

Differential Revision: D29433489

fbshipit-source-id: 92596d05f5057cff4c65283711b4ed6778d20758

Summary: Pull Request resolved: https://github.com/facebook/folly/pull/1614

Reviewed By: yfeldblum

Differential Revision: D29496725

fbshipit-source-id: 7c726e3e310eb28cf33603ebd83df6d832488369

Summary:
`guard()` is ignoring the return from `handler()`, thereby breaking `take()`.

(Note: this ignores all push blocking failures!)

Reviewed By: yfeldblum

Differential Revision: D29416513

fbshipit-source-id: 1e84d3ba53cb68a1664a349e71398c782b1df751

Summary: A channel is a sender and receiver pair that allows one component to send values to another. A sender and receiver pair is similar to an AsyncPipe and AsyncGenerator pair. However, unlike AsyncPipe/AsyncGenerator, senders and receivers can be used by higher level transformation abstractions that are much more memory-efficient than using AsyncGenerator directly. These higher level abstractions do not require long-lived coroutine frames to wait on incoming values.

Reviewed By: aary

Differential Revision: D29158424

fbshipit-source-id: 88c51d0f9d73677a04906197f4c44fe84ac01cdb

Summary: Pull Request resolved: https://github.com/facebook/folly/pull/1610

Reviewed By: yfeldblum

Differential Revision: D29434206

Pulled By: Orvid

fbshipit-source-id: e20d29d0572307e917f8673bb161e3af1a6c55c3

Summary:
Added options to enforce ALPN when both client and support support ALPN for
folly/openssl.

Reviewed By: knekritz

Differential Revision: D29298491

fbshipit-source-id: acdd6001fea89606e2438640a4434cc56454f1aa

Summary: `FixedString` should be constructable from a `std::string_view`. This allows passing `std::string_view` as key to `.emplace` in heterogeneous maps where the key is a `FixedString`.

Reviewed By: yfeldblum

Differential Revision: D29456948

fbshipit-source-id: 4d2e428c7de05c1dc34b50727f8107aec915b632

Summary:
The approach taken to fixing the lock inversion is not to use locks. Instead, we take a page from the existing futures playbook and use an atomic acyclic finite state machine! In this case, we have a single atomic pointer which may own either a raised object of type `exception_wrapper` or an interrupt-handler. Both are at least 4-byte aligned (8-byte aligned on 64-bit architectures) so both have the bottom two bits free. We use the bottom two bits as the state machine to identify whether `0x0u` initial where nothing is owned, `0x1u` where an interrupt-handler is owned, `0x2u` where a raised object is owned, or `0x3u` terminal where both an interrupt-handler and a raised object have been set and the handler has been invoked on the object.

As choices, we forbid repeated calls to set the interrupt handler but permit repeated calls to raise an object. Calls after the first to set an interrupt handler terminate while calls to raise an object after the first are ignored. Existing tests demonstrate raising twice so we may like to be cautious about breaking that behavior.

Some semantics are changed. Raised objects and interrupt handlers are destroyed earlier than before: they are now destroyed immediately after invoking handlers on objects.

The lock order inversion is observed by thread sanitizer (tsan) in the new shared-promise test.

Differential Revision: D29250149

fbshipit-source-id: 63e257d03cebbf8dba95a514b7e89680cae569a7

Summary:
Make `Skipper` accept the `shared_ptr` by value, like `Accessor` does, so we can avoid the copy.

Also expose the accessor, so after we handed over the `shared_ptr` we can still retain access to the list.

Reviewed By: philippv

Differential Revision: D29464861

fbshipit-source-id: 75e93f0e72c4b8a2635d25e5e7860f27aec2a0a5

Summary: Hack to use parexec and the par directly.

Reviewed By: chadaustin

Differential Revision: D29461444

fbshipit-source-id: 9ac6c1aa43728782b8de0a71446109f7fd5dab1d

Summary: It's already not async-signal-safe so just use the fast version of getting the unsymbolized stack traces.

Reviewed By: chadaustin

Differential Revision: D29439597

fbshipit-source-id: 5a727d6a1c37a1c29bce84d866bf4863774c5ff1

Reviewed By: Mizuchi

Differential Revision: D29329759

fbshipit-source-id: 588aa9ce11db2c1aea51dcf96e0de3b50633fa29

Summary:
This diff adds a computeChainCapacity method to IOBuf.

This method will loop through each chain of an IOBuf and return the total capacity of all IOBuf chains. This method is used very similarly to computeChainDataLength(), except rather than returning total data length, it returns total capacity.

Reviewed By: yfeldblum

Differential Revision: D28875393

fbshipit-source-id: 7e5f4a93916fa6c30eef8f8ee8c346a4537302af

Summary: If a callback associated with an observer depends on yet another observer, the callback may be called more than once per distinct snapshot of the associated observer. This diff fixes this problem.

Reviewed By: andriigrynenko

Differential Revision: D29313796

fbshipit-source-id: 345bfa0a1ac9dfaee806be21256088d4667aef15

Summary:
To support "clearing" a ring-buffer without violating TurnSequencer's
invariants, allow storing a Cursor and comparing them in a loop.

Differential Revision: D29345250

fbshipit-source-id: d1d3bbd0b38489690334bdd63a7f40d9f6bad6c6

Summary: The CancellationSource already provides this synchronization

Reviewed By: yfeldblum

Differential Revision: D29368728

fbshipit-source-id: 085467d1fcb70d6fc059174ec673f7825025c076

Summary: The empty size of UnboundedQueue is hundreds of bytes, which is a problem for many users of AsyncPipe. Switch the default to a smaller queue, adding a template parameter so high-throughput users can switch back.

Differential Revision: D29375068

fbshipit-source-id: 9fb2dad81697eeb70d58929e6b9c1c64102aa4a8

Summary: A smaller version of coro::UnboundedQueue to make AsyncPipe usable in memory-constrained programs.

Reviewed By: yfeldblum

Differential Revision: D29375067

fbshipit-source-id: 5d27e4514312d2748055be764af88c0688d46065

Summary: Cycle detection can be very expensive, so it's better to disable it in opt mode. Because of that we need to make sure we catch such cycles in dbg builds, so we have to replace exceptions with LOG(FATAL).

Reviewed By: joshkirstein

Differential Revision: D29367695

fbshipit-source-id: 9c2038eb5b42f98f4ab997f963b6a131b8d26cf9

Summary: There is a golden image of the `Core` layout so that tests will catch accidental increases of the `Core` size. Updated it to the latest `Core` layout.

Differential Revision: D29251632

fbshipit-source-id: d16086390548848f4302678e0b86d9841be1140b

Summary:
- `std::is_pod_v` is only available in C++17, shifting to `std::is_pod`
- `std::errc` needs the `system_error` header; missing header wasn't causing an error on most platforms because it is included indirectly

Differential Revision: D29355681

fbshipit-source-id: e035c3f4ffac9d2c6f0d8ec511f7e0ea8544ba80

Summary: Remove dependency on `folly/FixedString.h`

Differential Revision: D29317987

fbshipit-source-id: dbce91f117776a1dcd966230d9eed616b2a95613

Summary:
D28945040 added `collectAny()` which early-returns when any of the SemiAwaitables produces a value (or exception).  There's the potential for discarded results, however - multiple SemiAwaitables can produce results depending on whether they're at a cancellation point and when cancellation is signaled.

This diff adds a variant `collectAnyNoDiscard()` that signals cancellation when any SemiAwaitable finishes and returns *all* results that completed.  It produces a tuple of optional values from the SemiAwaitables (or `std::nullopt` if it was canceled).

Reviewed By: iahs

Differential Revision: D29240725

fbshipit-source-id: 3e664339e8692cbb9114138a96345cf9f9d5cb0b

Summary:
Bring Watchman closer to the EdenFS file name convention by moving
source files and includes into watchman/.

Reviewed By: fanzeyi

Differential Revision: D29242789

fbshipit-source-id: 6e29a4a50e7202dbf6b603ccc7e4c8184afeb115

Summary: Move some of the test utilities into `TcpInfoTestUtil.h` to enable use by other components.

Differential Revision: D29307490

fbshipit-source-id: 978947ff57ed02e438addf6190a4ea9955596333

Summary:
## An Overview of ConcurrentSkipList Synchronization
folly::ConcurrentSkipList, at a high level, is a normal skip list, except:
* Access to the nodes' "next" pointers are atomic. (The implementation used release/consume, which this diff changes to release/acquire. It's not clear the original use for consume was correct, and consume is very complicated without any practical benefit, so we should just avoid it.)
* All accesses (read/write) must go through an Accessor, which basically is nothing except an RAII object that calls addRef() on construction and releaseRef() on destruction.
* Deleting a node will defer the deletion to a "recycler", which is just a vector of nodes to be recycled. When releaseRef() drops the refcount to zero, the nodes in the recycler are deleted.

Intuitively speaking, when refcount drops to zero, it is safe to delete the nodes in the recycler because nobody holds any Accessors. It's a very simple way to manage the lifetime of the nodes, without having to worry about *which* nodes are accessed or to be deleted.

However, this refcount/recycling behavior is very hard to get right when using atomics as the main synchronization mechanism. In the buggy implementation before this diff, I'll highlight three relevant parts:
* To find an element in the skip list (either to fetch, insert, or delete), we start from the head node and skips by following successor pointers at the appropriate levels until we arrives at the node in question. Rough pseudocode:
  ```
  def find(val):
    node = head
    while read(node) < val:
      node = skip(node)  # read the node to find the successor at some level
    return node
  ```

* To delete an element from the skip list, after finding the element, we modify the predecessor at each level by changing their successors to point to the deleted element's successors, and place the deleted element into the recycler.
  ```
  def delete(node):
    for level in range(...):
      node->pred->setSkip(node->succ)
    recycle(node)

  def recycle(node):
    lock(RECYCLER_LOCK)
    recycler.add(node)
    unlock(RECYCLER_LOCK)
  ```

* releaseRef() and addRef():
  ```
  def releaseRef():
    if refcount > 1:
      refcount--
      return

    lock(RECYCLER_LOCK)
    if refcount > 1:
      refcount--
      return
    to_recycle, recycler = recycler, [] # swap out nodes to recycle
    unlock(RECYCLER_LOCK)

    for node in to_recycle:
      free(node)
    refcount--

  def addRef():
    recount++
  ```

## The Safety Argument
The Accessor/deletion mechanism is safe if we can ensure the following:
* If for a particular node X, a thread performs read(X) and a different thread performs free(X), the read(X) happens-before free(X).

### Problem 1: Relaxed decrement
The buggy implementation used relaxed memory order when doing `refcount--`. Let's see why this is a problem.

Let thread 1 be the one performing read(X) and let thread 2 be the one performing free(X). The intuitive argument is that free(X) can only happen after the refcount drops to zero, which cannot be true while read(X) is happening. The somewhat more formal argument is that read(X) happens before refcount-- in thread 1, which happens before refcount--, which happens before free(X). But because we use relaxed memory order, the two refcount-- operations do not synchronize.

### Problem 2: Relaxed increment
The implementation also used relaxed memory order for addRef(). Normally, for a refcount, it is OK to use relaxed increments, but this refcount is different: the count can go back up once it reaches zero. When reusing refcounts this way, we can no longer use relaxed increment.

To see why, suppose thread 2 performs the following in this order:
```
setSkip(P, not X)  # step before deleting X; P is a predecessor at some level
recycle(X)
refcount-- # acq_rel, gets 0
free(X)
```
and thread 2 performs the following in this order:
```
refcount++ # relaxed
skip(P) # gets X
read(X)
```
See Appendix A below; it's possible for the refcount to reach 0, and thread 2 to get X when reading the successor from P. This means that free(X) might theoretically still race with read(X), as we failed to show that once we delete something from the skip list, another accessor can't possibly reach X again.

This might feel like an unnecessary argument, but without this reasoning, we would not have found a problem even if we just modified releaseRef() to instead delete some random nodes from the list. That will certainly cause trouble when someone else later tries to read something.

### Problem 3: No release operation on refcount before freeing nodes
This is much more subtle. Suppose thread 1 performs the following:
```
refcount--  # end of some previous accessor
refcount++  # begin of new accessor
skip(P)
```
and thread 2 does this:
```
setSkip(P, not X)  # step before deleting X; P is a predecessor
recycle(X)
read(refcount) # gets 1, proceeds to lock
lock(RECYCLER_LOCK)
read(refcount) # gets 1  ***
unlock(RECYCLER_LOCK)
free(X)
refcount--
```

The interleaving to make this possible is:
```
thread 1 refcount--
thread 2 everything until free(X)
thread 1 refcount++
thread 2 free(X)
thread 2 refcount--
```

We wish to show that `setSkip(P, not X)` happens before `skip(P)`, because this will allow us to prove that `skip(P) != X` (otherwise, we would not be able to show that a subsequent read(X) in thread 1 happens before free(X) - it might legitimately not).

The intuitive argument is that `setSkip(P, not X)` happened before we decrement the refcount, which happens before the increment of the refcount in thread 1, which happens before `skip(P)`. However, because thread 2 actually decremented refcount quite late, it might be the case that thread 1's `refcount++` happened before thread 2's `refcount--` (and the increment synchronized with its own decrement earlier). There's nothing else in the middle that provided a synchronizes-with relationship (in particular, the read(refcount) operations do not provide synchronization because those are *loads* - wrong direction!).

### Correct implementation
In addition to using acq_rel memory order on all operations on refcount, this diff modifies releaseRef() like this:
```
def releaseRef():
  if refcount > 1: # optimization
    refcount--
    return

  lock(RECYCLER_LOCK)
  if --refcount == 0:  # "GUARD"
    to_recycle, recycler = recycler, [] # swap out nodes to recycle
  unlock(RECYCLER_LOCK)

  for node in to_recycle:
    free(node)
```

I'll use "GUARD" to denote the event that the --refcount within the lock *returns 0*. I'll use this for the correctness proof.

### Correct implementation proof
The proof will still be to show that if thread 1 performs read(X) and thread 2 performs free(X), read(X) happens-before free(X).

Proof: thread 1 must have grabbed an accessor while reading X, so its sequence of actions look like this:
```
refcount++
skip(P)  # gets X
read(X)
refcount--
```
thread 2 performs:
```
GUARD
free(X)
```

Now, all writes on refcount are RMW operations and they all use acq_rel ordering, so all the RMW operations on refcount form a total order where successive operations have a synchronizes-with relationship. We'll look at where GUARD might stand in this total order.

* Case 1: GUARD is after refcount-- from thread 1 in the total order.
  * In this case, read(X) happens before refcount-- in thread 1, which happens before GUARD, which happens before free(X).
* Case 2: GUARD is between refcount++ and refcount-- from thread 1 in the total order.
  * In this case, observe (by looking at the total ordering on refcount RMW) that we have at least two threads (1 and 2) that contribute 1 to the refcount, right before GUARD. In other words, GUARD could not possibly have returned 0, which is a contradiction.
* Case 3: GUARD is before refcount++ from thread 1 in the total order.
  * Let `setSkip(P, not X)` be a predecessor write operation before X is added to the recycler (this can happen on any thread). We will prove in the below lemma that `setSkip(P, not X)` happens before GUARD. Once we have that, then `setSkip(P, not X)` happens before GUARD which happens before thread 1's refcount++ which happens before `skip(P)`, and that renders it impossible for `skip(P)` to return X, making it a contradiction.

It remains to prove that `setSkip(P, not X)` happens before GUARD.

In the thread that performs an `setSkip(P, not X)` operation, it subsequently performs `recycle(X)`, which adds X to the recycler within RECYCLER_LOCK. In thread 2, GUARD happens within the RECYCLER_LOCK, and the subsequent swapping of the recycler vector contained X (which is shown by the fact that we free(X) after GUARD), so the lock must have been grabbed *after* the lock that added X to the recycler. In other words, we have the relationship that `setSkip(P, not X)` happens before `recycler.add(X)` which happens before `unlock(RECYCLER_LOCK)` which happens before `lock(RECYCLER_LOCK)` which happens before GUARD.

Note that just like the original implementation, the optimization on top of releaseRef() is not a perfect optimization; it may delay the deletion of otherwise safe-to-delete nodes. However, that does not affect our correctness argument because it's always at least as safe to *delay* deletions (this hand-wavy argument is not part of the proof).

## Appendix A
Consider the following two threads:
```
std::atomic<int> x{0}, y{1};
// Thread 1:
x.store(1, std::memory_order_release);  // A
int y1 = y.fetch_add(-1, std::memory_order_acq_rel);  // B

// Thread 2:
y.fetch_add(1, std::memory_order_relaxed);  // C
int x2 = x.load(std::memory_order_acquire);  // D
```
Intuitively, if y1 = 1, then thread 1's fetch_add was executed first, so thread 2 should get x2 = 1. Otherwise, if thread 2's fetch_add happened first, then y1 = 2, and x2 could be either 0 or 1.

But, could it happen that y1 = 1 and x2 = 0? Let's look at the happens-before relationships between these operations. For intra-thread (sequenced-before), we have A < B and C < D (I'm using < to denote happens-before). Now, for inter-thread (synchronizes-with), the only pair that could establish a synchronizes-with relationship is A and D. (B and C is not eligible because C uses relaxed ordering.) For y1 = 1 and x2 = 0 to happen, we must not have D read the result of A, so it must be the case that they A does not synchronize-with D. But that's as far as we can go; there's nothing that really enforces much of an ordering between the two threads.

We can also think about this in terms of reordering of memory operations. Thread 1 is pretty safe from reordering because of the acq_release, but in thread 2, an "acquire" ordering typically means no memory operations after D may be reordered before D, but it doesn't prevent C from being reordered after D. C itself does not prevent reordering due to it being an relaxed operation. So if thread 2 executed D and then C, then it would be trivially possible to have y1 = 1 and x2 = 0.

The point of this is to highlight that just by having a release/acquire pair does not magically *order* them. The pair merely provides a synchronizes-with relationship *if* the read happens to obtain the value written by the write, but not any guarantees of *which* value would be read.

## Appendix B
Problem 1 is detected by TSAN, but problems 2 and 3 are not. Why?

TSAN detects data races by deriving synchronization relationships from the *actual* interleaving of atomics at runtime. If the actual interleaving would always happen but is not *guaranteed* by the standard, there may be a real undetected data race.

For example, it is well known that the following code will be detected by TSAN as a data race on int "x":
```
int x = 1;
std::atomic<bool> y{false};

// Thread 1
x = 2;
y.store(true, memory_order_relaxed);

// Thread 2
while (!y.load(memory_order_acquire)) {  // acquire is useless because writer used relaxed
}
std::cout << x << std::endl;
```
TSAN reports a data race on x because `y` failed to provide proper synchronizes-with relationship between the two threads due to incorrect memory ordering. However, when compiling on x86, most likely we will end up with a binary that always guarantees the intuitively desired behavior anyway.

So now consider the following code:
```
std::atomic<int> x = 1;
std::atomic<bool> y{false};
int z = 8;

// Thread 1
z = 9;
x.store(2, memory_order_release);
y.store(true, memory_order_relaxed);

// Thread 2
while (!y.load(memory_order_acquire)) {  // acquire is useless because writer used relaxed
}
x.load(memory_order_acquire);
std::cout << z << std::endl;
```
There is a data race on the access to z, because the happens-before chain of `write z -> x.store -> y.store -> y.load -> x.load -> read z` is broken on the `y.store -> y.load` link. However, TSAN will not report a data race, because it sees the chain as `write z -> x.store -> x.load -> read z`. It sees x.store as synchronizing with x.load because it *observed* that the x.load obtained the value written by the x.write *at runtime*, so it inferred that it was valid synchronization. This isn't guaranteed, though, because it's possible in some execution (in theory) that x.load does not get the value written by x.store (similar to Appendix A).

Reviewed By: yfeldblum

Differential Revision: D29248955

fbshipit-source-id: 2a3c9379c7c3a6469183df64582ca9cf763c0890