Christopher Dykes [Sat, 15 Apr 2017 00:05:09 +0000 (17:05 -0700)]
Add portability support for PThread's TLS API
Summary:
This is the last piece needed to get Folly working on Windows without PThreads.
Updating Folly's test suite to support compiling without PThreads will come next.
Reviewed By: yfeldblum
Differential Revision:
D4894048
fbshipit-source-id:
6076317e1364aef82a5d3cb306bea7c2226b3cdc
Christopher Dykes [Sat, 15 Apr 2017 00:03:04 +0000 (17:03 -0700)]
Remove unused local in OpenSSLCertUtils
Summary: It's unused, and MSVC is warning about that fact.
Reviewed By: anirudhvr
Differential Revision:
D4894403
fbshipit-source-id:
767b944eb9fa0376c795555357c09bbf9179e24f
Andrew Krieger [Fri, 14 Apr 2017 21:36:30 +0000 (14:36 -0700)]
Suppress more warnings for MSVC
Summary:
Several other warnings that aren't reasonable to disable globally occur in folly headers.
- Wrap the unreachable code warnings in MSVC specific disable blocks to prevent problems for users.
- Use more careful bit twiddling instead of negating unsigned types
- Enable a simpler overload for bool->float conversion than one which attempts float->bool.
- Delete one unneeded undef.
Reviewed By: yfeldblum
Differential Revision:
D4891583
fbshipit-source-id:
4d2efda1fe720abcb083bf29b578c065127cda24
Christopher Dykes [Fri, 14 Apr 2017 21:09:23 +0000 (14:09 -0700)]
Use folly::getCurrentThreadId() in SSLContext
Summary: Rather than duplicating logic (and #ifdefs), just call the helper function instead.
Reviewed By: yfeldblum
Differential Revision:
D4888362
fbshipit-source-id:
f5096f6029cf05526aa74bd40235ac8014824789
Christopher Dykes [Fri, 14 Apr 2017 20:54:32 +0000 (13:54 -0700)]
Make folly::getCurrentThreadId() return a thread ID on OSX
Summary: It was previously returning a pointer masquerading as a `uint64_t`.
Reviewed By: yfeldblum
Differential Revision:
D4888325
fbshipit-source-id:
6cf6ed09f0e7a39a8a2f93d3fa14b06913c27805
Christopher Dykes [Fri, 14 Apr 2017 20:53:52 +0000 (13:53 -0700)]
Add FOLLY_HAVE_PTRHEAD
Summary:
Folly is gaining support for compiling without PThread support on Windows, but we need a way to know that at compile-time for certain APIs.
This also includes changes to guard the only API outside of portability/PThread.h that needs to be guarded on the existence of PThread.
Reviewed By: yfeldblum
Differential Revision:
D4889526
fbshipit-source-id:
21175ad90f60a47718c7e2775e3b429b2aad62e2
Christopher Dykes [Fri, 14 Apr 2017 20:51:43 +0000 (13:51 -0700)]
Add support for building without PThread to the CMake files
Summary: Because, soon it won't be required.
Reviewed By: yfeldblum
Differential Revision:
D4892076
fbshipit-source-id:
6bfbf5bf523262659c0af336f659398ae8545e50
Christopher Dykes [Fri, 14 Apr 2017 18:14:45 +0000 (11:14 -0700)]
Remove the old SpinLock implementations
Summary: They aren't actually needed as the primary implementation is supported on all platforms.
Reviewed By: yfeldblum
Differential Revision:
D4882687
fbshipit-source-id:
7208c5d3c1f35b29b0cabb6a20fe030fbf10b131
Mark Williams [Fri, 14 Apr 2017 16:00:08 +0000 (09:00 -0700)]
Revert
D4805628: [folly] Suppress more warnings for MSVC
Summary: This reverts commit
9574ea984e1e3daca01101259687f46effcf3a9f
Differential Revision:
D4805628
fbshipit-source-id:
3285317a4c3f2012da1f13e9c91b6c623726c9b7
Yedidya Feldblum [Fri, 14 Apr 2017 07:22:47 +0000 (00:22 -0700)]
Initialize the singleton-vault early in init()
Summary:
Initialize the singleton-vault early in `init()` to avoid ordering problems with other parts of `init()`.
Initialization is done by calling `SingletonVault::singleton()->registrationComplete()`. Do that before (almost) everything, in particular, before parsing args and initializing logging.
Note that this just marks it okay for singletons to be instantiated, but does not by itself instantiate any singletons. However, this is useful if, in some way, parsing args or initializing logging should happen to use singletons.
Reviewed By: andriigrynenko
Differential Revision:
D4889777
fbshipit-source-id:
6817db22d38a498cd20e361fc574a146029432ec
Andrew Krieger [Fri, 14 Apr 2017 07:06:53 +0000 (00:06 -0700)]
Suppress more warnings for MSVC
Summary:
Several other warnings that aren't reasonable to disable globally occur in folly headers.
- Wrap the unreachable code warnings in MSVC specific disable blocks to prevent problems for users.
- Cast to signed types to perform integral negation before casting back to unsigned for bit operations.
- Enable a simpler overload for bool->float conversion than one which attempts float->bool.
- Delete one unneeded undef.
(Note: this ignores all push blocking failures!)
Reviewed By: yfeldblum
Differential Revision:
D4805628
fbshipit-source-id:
9574ea984e1e3daca01101259687f46effcf3a9f
Andrew Krieger [Fri, 14 Apr 2017 05:57:20 +0000 (22:57 -0700)]
Remove pthread dependency from CacheLocality.h
Summary:
For getting a thread id for hashing, std::this_thread::get_id()
is just as good and also removes an unneeded pthreads dependency. This enables
using MPMCQueue on Windows without pthreads as well.
Reviewed By: Orvid, yfeldblum
Differential Revision:
D4879940
fbshipit-source-id:
c4fb5eea165eb450240f94904aa26a10aa394d1b
Christopher Dykes [Fri, 14 Apr 2017 04:25:29 +0000 (21:25 -0700)]
Switch TimedRWMutex implementation to use lock_guard where possible
Summary: As titled. There are some places where scoping prevents trivial conversion, so I've left those places with explicit lock/unlock calls.
Reviewed By: andriigrynenko, yfeldblum
Differential Revision:
D4888812
fbshipit-source-id:
c124306e0373ee9eb395cb98f88df63c91523d48
Christopher Dykes [Fri, 14 Apr 2017 04:19:08 +0000 (21:19 -0700)]
Use getCurrentThreadID() rather than pthread_self() in MemoryIdler
Summary: Ignoring the questionability of the approach it's being used in, it's better to not have the direct PThread dependency.
Reviewed By: yfeldblum
Differential Revision:
D4889245
fbshipit-source-id:
da099c6f938dbe98c1b9eeaf4de0a27a2c4d65f1
Andrew Krieger [Fri, 14 Apr 2017 02:23:50 +0000 (19:23 -0700)]
Explicitly use CreateFileA in readlink
Summary:
In Unicode enabled projects, this errors because CreateFile aliases
CreateFileW, which takes wchar_t* not char*.
Reviewed By: Orvid, yfeldblum
Differential Revision:
D4878424
fbshipit-source-id:
b44b369c0533e74163f68d95c2bf353584033731
Christopher Dykes [Fri, 14 Apr 2017 01:52:20 +0000 (18:52 -0700)]
Support folly::getCurrentThreadID() without PThread
Summary: Just use the native Windows API call instead, as it gives the same result.
Reviewed By: yfeldblum
Differential Revision:
D4887587
fbshipit-source-id:
adbfd288bea425f1aff46d0685083807490f7f09
Christopher Dykes [Fri, 14 Apr 2017 00:56:39 +0000 (17:56 -0700)]
Don't use pthread_spinlock_t in TimedRWMutex
Summary: `TimedMutex` was already using `folly::SpinLock`, so switch `TimedRWMutex` to do the same.
Reviewed By: andriigrynenko
Differential Revision:
D4888005
fbshipit-source-id:
6e782347bc22dc186ed41f2e77c6614b8444bae2
Christopher Dykes [Thu, 13 Apr 2017 20:57:26 +0000 (13:57 -0700)]
Switch includes of PThread to the portability header
Summary:
As part of the work towards getting Folly compiling without PThread, it needs to be treated as a non-portable include.
This switches the includes not already covered by other diffs.
Reviewed By: yfeldblum
Differential Revision:
D4882777
fbshipit-source-id:
d9521564c814f9bcff2fcb358cbb89b3777c327d
Christopher Dykes [Thu, 13 Apr 2017 19:45:06 +0000 (12:45 -0700)]
Fix 1/2 of exception_wrapper under MSVC
Summary:
MSVC didn't like referring to members of `exception_wrapper` in the initializer for a `static constexpr` field directly in `exception_wrapper`, so shift the initialization to the actual definition of the fields.
As the fields are only referred to via their address, dropping the `constexpr` loses nothing.
Reviewed By: ericniebler
Differential Revision:
D4873939
fbshipit-source-id:
30b690b1ab3f2f7a25b9dc4863b46f64c811797d
Neel Goyal [Thu, 13 Apr 2017 19:14:40 +0000 (12:14 -0700)]
Add method to check if SSL Lock is disabled
Summary:
Add a method where users can determine if a SSL lock is disabled.
This can help when it comes to making decisions about things like whether reusing SSL Contexts is safe in multithreaded programs.
Reviewed By: siyengar
Differential Revision:
D4875780
fbshipit-source-id:
91e9259fee25856be1b77823559d16d0679bde5b
Christopher Dykes [Thu, 13 Apr 2017 18:24:51 +0000 (11:24 -0700)]
Kill a couple of PThread includes
Summary: They aren't actually needed, and are in the way of being able to build Folly without PThread.
Reviewed By: yfeldblum
Differential Revision:
D4882410
fbshipit-source-id:
e4a48600f79c57bb965a35fab94962a29b8e06d4
Christopher Dykes [Thu, 13 Apr 2017 05:25:15 +0000 (22:25 -0700)]
Add CMake build system
Summary: This is for use on Windows only, and does not support Linux or OSX in the slightest. There are probably still a few hacks laying around in it, but it works, and that's good enough for now.
Reviewed By: yfeldblum
Differential Revision:
D4873498
fbshipit-source-id:
5952a927037dc1d56c51e71d51ba506d2cd17357
Nick Terrell [Thu, 13 Apr 2017 02:43:02 +0000 (19:43 -0700)]
Add bzip2 support
Summary:
Adds bzip2 support to `folly/io/Compression.h`.
Adds bzip2 to the default set of supported codecs for the `AutomaticCodec`.
Reviewed By: yfeldblum
Differential Revision:
D4873771
fbshipit-source-id:
d4f4861aef7e4b9efb67095e8892c265b5ae5557
Christopher Dykes [Thu, 13 Apr 2017 01:19:37 +0000 (18:19 -0700)]
Fix the last issues with exception_wrapper under MSVC
Summary:
Well, MSVC 2017 anyways.
MSVC 2017 wasn't able to handle the decltype in the type specialization, which was also entirely unnecesary.
Reviewed By: yfeldblum
Differential Revision:
D4880272
fbshipit-source-id:
b6deed72a12d4b58d6eca200e7287e536f94ca30
Alexey Spiridonov [Wed, 12 Apr 2017 21:43:04 +0000 (14:43 -0700)]
Replace Subprocess::pipe* syntax sugar with Subprocess::Options().pipe*
Summary:
This is a bit too magical -- it's not clear that the thing produces an Options object. If you do know that you can chain further option setters off this thing, it's nice, but otherwise, the first impression is "what just happened?".
So, let's have one good way for doing things.
Reviewed By: yfeldblum
Differential Revision:
D4863947
fbshipit-source-id:
3dfe83cfc077d47f604f47dcb21149fbaa2d2243
Alexey Spiridonov [Wed, 12 Apr 2017 21:43:03 +0000 (14:43 -0700)]
Delete | operator for Subprocess::Options
Summary:
This operator is WRONG. It has not worked correctly for years, e.g. it lacks support for chdir, and several other options.
The operator is not really useful after C++11. Usually, you should just chain setters, e.g. `Subprocess::Options().pipeStdout().pipeStderr()`. If you must repeatedly mutate options in a fixed way, in the C++11 world you can use a lambda instead.
Reviewed By: yfeldblum
Differential Revision:
D4862698
fbshipit-source-id:
a2d8ace53424b9232e178cf202cf51beb7b59b12
Nick Terrell [Wed, 12 Apr 2017 19:44:56 +0000 (12:44 -0700)]
Fix zlib + lzma memory usage inefficiency
Summary:
Since IOBuf rounds to a good malloc size, zlib and lzma waste space.
For a 4 MiB `length`, about 1 MiB of the IOBuf is wasted.
Reviewed By: yfeldblum
Differential Revision:
D4872830
fbshipit-source-id:
42fc83277b2dae22b75403e0e71c43e8f2b19f21
Jude Taylor [Wed, 12 Apr 2017 17:57:54 +0000 (10:57 -0700)]
codemod: ASN1_STRING_data -> ASN1_STRING_get0_data
Reviewed By: anirudhvr
Differential Revision:
D4830205
fbshipit-source-id:
55f34c3bddead8a73174c403f76572248c541a10
Christopher Dykes [Wed, 12 Apr 2017 17:44:00 +0000 (10:44 -0700)]
Make folly::StrictConjunction usable under MSVC
Summary:
Template pack expansion plays much nicer when being expanded into the template parameters of a base class than it does as the template parameters to a templated `using` when under MSVC.
This fixes the other half of `folly::exception_wrapper`.
Reviewed By: yfeldblum
Differential Revision:
D4874047
fbshipit-source-id:
6593d10f8737b940bd550353e54f43bc96402a48
Jude Taylor [Wed, 12 Apr 2017 16:40:55 +0000 (09:40 -0700)]
Provide a copy ctor for Digest that copies the current hash context
Summary: opensource changes for t16702532
Reviewed By: yfeldblum
Differential Revision:
D4846899
fbshipit-source-id:
67a610ff399e95c7cf1c9c8a5950f79bfc3aabb9
Artem Lantsev [Wed, 12 Apr 2017 09:16:51 +0000 (02:16 -0700)]
Add memory padding and alignment to prevent false sharing
Summary: This is a follow up commit for the thread https://github.com/facebook/folly/pull/378
Reviewed By: nbronson
Differential Revision:
D4860356
fbshipit-source-id:
f10a0d12a593c18b1abf94da5b477c524c04f4be
Aravind Anbudurai [Wed, 12 Apr 2017 08:36:29 +0000 (01:36 -0700)]
Support linux namespace clone flags
Summary:
This diffs adds supports on folly::Subprocess to be able to take clone flags and
use them to call clone(2) instead of the default vfork()
I checked that all tests pass when I replace vfork with fork on trunk. So there
isn't anything built assuming the parent is paused for an execve. Correct me if
I am wrong here.
(Note: this ignores all push blocking failures!)
Reviewed By: snarkmaster
Differential Revision:
D4853169
fbshipit-source-id:
7e5851df3a49996a4a5dc1945457686dd042e1f4
Angelo Failla [Wed, 12 Apr 2017 04:13:21 +0000 (21:13 -0700)]
IPAddressV(46) methods to convert IPs to their inverse in-addr.arpa / ip6.arpa representation
Summary: As per the title, this adds methods to convert IPv(46) to their inverse representation, this is useful for applications that needs to calculate the inverse arpa representation, typically used when making PTR DNS requests.
Reviewed By: Orvid
Differential Revision:
D4867502
fbshipit-source-id:
190e5c309b17a633e1c97b077f212ab38725860f
Eric Niebler [Tue, 11 Apr 2017 23:12:59 +0000 (16:12 -0700)]
non-throwing, non-allocating exception_wrapper
Summary:
The purpose of this reimplementation of `exception_wrapper` is threefold:
- Make `exception_wrapper` smaller. It goes from 48 bytes to 24.
- Give it `noexcept` ~~copy and~~ move
- Store small exception objects in an internal buffer; i.e., with zero allocations.
The ultimate goal is to change `folly::Try<T>` to a thin wrapper over `folly::Expected<T, exception_wrapper>`. (Currently, it stores the `exception_wrapper` on the heap.)
As part of this redesign, I:
- Remove `exception_wrapper::getCopied`. The user shouldn't care how the `exception_wrapper` stores the exception.
- Remove `exception_wrapper::operator==`. It was only used in 2 places in test code. The existing semantics (return true IFF two `exception_wrapper`s point to the //same// exception object) prevented the small-object optimization.
- Add new `handle()` API that behaves like cascading `catch` clauses. For instance:
```lang=c++
exception_wrapper ew = ...;
ew.handle(
[&](const SomeException& e) { /*...*/ },
[&](const AnotherException& e) { /*...*/ },
[&](...) { /* catch all*/ }, // yes, lambda with ellipses works!
```
- Add a `type()` member for accessing the `typeid` of the wrapped exception, if it's known or can be determined with a `catch(std::exception&)`.
This table shows the percent improvement for the exception_wrapper_benchmark test:
| Test | Percent improvement (gcc-5) | Percent improvement (gcc-4)
| ----- | ----- | -----
| exception_wrapper_create_and_test | 14.33% | -6.50%
| exception_wrapper_create_and_test_concurrent | 11.91% | 20.15%
| exception_wrapper_create_and_throw | -0.82% | -0.25%
| exception_wrapper_create_and_cast | 15.02% | 14.31%
| exception_wrapper_create_and_throw_concurrent | 18.37% | 8.03%
| exception_wrapper_create_and_cast_concurrent | 28.18% | -10.77%
The percent win for gcc-5 is 15% on average. The non-throwing tests show a greater win since the cost of actually throwing an exception drowns out the other improvements. (One of the reasons to use `exception_wrapper` is to not need to throw in the first place.) On gcc-4, there is roughly no change since the gcc-4 standard exceptions (`std::runtime_error`, std::logic_error`) are non-conforming since they have throwing copy operations.
Reviewed By: yfeldblum
Differential Revision:
D4385822
fbshipit-source-id:
63a8316c2923b29a79f8fa446126a8c37aa32989
Christopher Dykes [Tue, 11 Apr 2017 21:17:10 +0000 (14:17 -0700)]
Treat OpenSSL as a non-portable include
Summary: On Windows, the OpenSSL headers have to be included in a specific order relative to WinSock. Fixing the issues ad-hock is not a viable way to continue, so treat it as completely non-portable and require all includes of it in Folly to go through the portability header.
Reviewed By: yfeldblum
Differential Revision:
D4856858
fbshipit-source-id:
56aca1fb0fe095f41a8af12488c6c2080344603d
Christopher Dykes [Tue, 11 Apr 2017 19:26:53 +0000 (12:26 -0700)]
Implement clearenv in the portability headers
Summary:
It doesn't exist on OSX or Windows, so implement it.
Closes https://github.com/facebook/folly/issues/567
Reviewed By: yfeldblum
Differential Revision:
D4783463
fbshipit-source-id:
0a2586aced7123b797a8e55a3e86124b449634e4
Andrii Grynenko [Tue, 11 Apr 2017 06:30:46 +0000 (23:30 -0700)]
Fix TLRefCount::useGlobal to work with TSAN
Summary: TSAN has a limitation for the number of locks held concurrently - https://github.com/llvm-mirror/compiler-rt/blob/master/lib/sanitizer_common/sanitizer_deadlock_detector.h#L126, so we have to work around that.
Reviewed By: yfeldblum
Differential Revision:
D4866477
fbshipit-source-id:
958251e97d91c1c46ef4e907de2cf189fb04f88e
Andrii Grynenko [Tue, 11 Apr 2017 06:14:38 +0000 (23:14 -0700)]
Fix Fiber::LocalData leak for large fiber locals
Summary: facepaw
Reviewed By: yfeldblum
Differential Revision:
D4856227
fbshipit-source-id:
17a73d4d630c6efe1763bbd8aff8b7a2c7721c27
Christopher Dykes [Tue, 11 Apr 2017 02:23:54 +0000 (19:23 -0700)]
Revert
D4832473: [Folly] Disable EnvUtil::setAsCurrentEnvironment() on platforms without clearenv()
Summary: This reverts commit
a80aabb5a223264746ab45e3138d065bce5fe99c
Differential Revision:
D4832473
fbshipit-source-id:
1a78d61f3f09f1064876a786c15601a091f4402b
Alexander Shaposhnikov [Mon, 10 Apr 2017 22:24:47 +0000 (15:24 -0700)]
Remove superfluous std::move
Summary: Cleanup: remove superfluous std::move for return values.
Reviewed By: yfeldblum
Differential Revision:
D4856421
fbshipit-source-id:
bf2cb66f3b43eb8f3b183e22b0acd31e0a0c0958
Nick Terrell [Mon, 10 Apr 2017 20:44:04 +0000 (13:44 -0700)]
Fix std::max() call in Compression.cpp
Summary:
`std::max(uint64_t, size_t)` was called.
Fixes https://github.com/facebook/folly/issues/576.
Reviewed By: Orvid
Differential Revision:
D4861336
fbshipit-source-id:
1b6f67b291451048ba79d638d2c1184f9245dc0c
Christopher Dykes [Mon, 10 Apr 2017 19:14:41 +0000 (12:14 -0700)]
Fix the build... Whoops...
Summary: I landed a change to Folly that used a feature I had just re-written in Folly without updating the change. This has broken everything. Whoops.
Reviewed By: yfeldblum
Differential Revision:
D4861562
fbshipit-source-id:
05beadbe233d1fdf90fb4435e65b718cdec7f63e
Christopher Dykes [Mon, 10 Apr 2017 18:10:21 +0000 (11:10 -0700)]
Re-work the OpenSSL portability header to be a portability header
Summary:
It was defining the same functions as OpenSSL, but it was putting them in the `folly::ssl` namespace and requiring a manual `use namespace` everywhere to actually be able to use it. Proper portability headers require no interaction to function correctly other than including them in the build.
This adjusts it so including it is all that is required to get an OpenSSL 1.1.0 compatible API.
This also re-organizes the code to significantly lower the amount of nesting in the `#if`'s, and puts it in order from oldest APIs to newest, rather than haphazard.
Reviewed By: yfeldblum
Differential Revision:
D4856784
fbshipit-source-id:
7c00fdf4a3e51e37287d2b1ef294046f38a96a43
Jason Fried [Mon, 10 Apr 2017 18:05:00 +0000 (11:05 -0700)]
Client RequestChannels and folly/python/futures.h
Summary:
Use RequestChannel to construct Thrift Clients.
use folly/python/futures.h to marry up thrift client rpc futures to
asyncio futures. No gil exchanges
py3 clients are async context managers now.
Reviewed By: ambv, yfeldblum
Differential Revision:
D4639677
fbshipit-source-id:
6b400747507446882d3cfcb1cf7e85c3daa07a2e
Andrii Grynenko [Mon, 10 Apr 2017 17:44:31 +0000 (10:44 -0700)]
Fix folly/test:singleton_test - Singleton.SharedPtrUsage
Reviewed By: yfeldblum
Differential Revision:
D4856138
fbshipit-source-id:
19eee4f8c4108ef168d82d5e9b1f2d65ff9b03bd
Christopher Dykes [Mon, 10 Apr 2017 01:02:53 +0000 (18:02 -0700)]
Refactor FOLLY_GCC_DISABLE_WARNING to play nice with clang-format
Summary: Currently clang-format cannot be run on any file using the macro because it breaks it. By putting the warning name in a string clang-format no longer breaks it.
Reviewed By: yfeldblum
Differential Revision:
D4856945
fbshipit-source-id:
ebd2d223dd97f66224e9e2912610057e36287dae
Yedidya Feldblum [Sat, 8 Apr 2017 14:08:01 +0000 (07:08 -0700)]
All template params for PriorityMPMCQueue
Summary:
[Folly] All template params for `PriorityMPMCQueue`.
Mimic the suite of template params of `MPMCQueue`.
Reviewed By: magedm
Differential Revision:
D4857179
fbshipit-source-id:
3d441099c46ea1223155bd3f4cc819595814c47b
Christopher Dykes [Sat, 8 Apr 2017 04:22:45 +0000 (21:22 -0700)]
Fix include ordering for OpenSSLPtrTypes.h
Summary: When including OpenSSL on Windows, the portability header must be included before including any OpenSSL headers.
Reviewed By: yfeldblum
Differential Revision:
D4856376
fbshipit-source-id:
a8e6533b2ba3c003314ae2426605031fa67be05f
Christopher Dykes [Sat, 8 Apr 2017 04:15:44 +0000 (21:15 -0700)]
Qualify a couple local constexpr values as 'static' to allow use from within non-capturing lambdas
Summary: MSVC doesn't allow non-static locals to be accessed within a non-capturing lambda, and some were introduced into the json code.
Reviewed By: yfeldblum
Differential Revision:
D4856242
fbshipit-source-id:
b2db36ca37a79a41237d39cc7e7b839a5416922f
Giuseppe Ottaviano [Sat, 8 Apr 2017 02:50:22 +0000 (19:50 -0700)]
Fix leak in EliasFanoCodingTest
Summary: `EliasFanoCompressedList` requires explicit freeing.
Reviewed By: lbrandy, philippv, yfeldblum, luciang
Differential Revision:
D4855200
fbshipit-source-id:
ccee44b6371ae743ef15a72af9b691d5b56e37c2
Louis Brandy [Sat, 8 Apr 2017 02:38:33 +0000 (19:38 -0700)]
ReadMostlySharedPtrTest, subtle test bug caught by -fsanitize-address-use-after-scope
Summary: Because these `atomic<int>` are referenced in the destructor of the `ReadMostlyMainPtr`, they must be declared before it, otherwise they are destroyed before it i.
Reviewed By: yfeldblum
Differential Revision:
D4854964
fbshipit-source-id:
625a8f890bba166b682ca13558f8bcde86edcc75
Louis Brandy [Sat, 8 Apr 2017 02:25:07 +0000 (19:25 -0700)]
fix EvictingCacheMap test bug found w/ -fsanitize-address-use-after-scope
Summary: This test actually stores references to `sum` in objects it puts into the `EvictingCacheMap`. Those references get accessed in the destructor of the objects, which is the destructor of the `EvictingCacheMap`. That means the `sum` variable must outlive (and be declared -before-) the `EvictingCacheMap` it's testing.
Reviewed By: yfeldblum, meyering
Differential Revision:
D4854619
fbshipit-source-id:
b0a3109278f48d10b4cd0c52a12cb2064f4a00c5
Anirudh Ramachandran [Fri, 7 Apr 2017 18:58:27 +0000 (11:58 -0700)]
Fix leak in HMAC_CTX_free compat API
Summary:
D4406876 replaced EVP_MD_CTX and HMAC_CTX on the stack with heap-allocated
versions. We omitted the _cleanup functions thinking those didn't actually free
anything, but turns out HMAC_CTX_cleanup internally calls EVP_MD_CTX_cleanup
which OPENSSL_frees a bunch of stuff.
Reviewed By: yfeldblum
Differential Revision:
D4850388
fbshipit-source-id:
4ef1413a4105f3638140e4d12aeae64afe43f099
Phil Willoughby [Fri, 7 Apr 2017 14:11:55 +0000 (07:11 -0700)]
Fix a logic error in StringKeyedTest
Summary: Make sure that the last use of `s6` is sequenced before it is moved-from.
Reviewed By: nbronson
Differential Revision:
D4850480
fbshipit-source-id:
5272c14475ae2c62bc613cfd54f5df482b19d3dc
Alan Frindell [Thu, 6 Apr 2017 21:00:58 +0000 (14:00 -0700)]
Get rid of try/catch in messageAvailable, which is now noexcept
Summary: The problem with catching the exception here is stack information is lost. Just let it std::terminate if it throws
Reviewed By: ikobzar
Differential Revision:
D4831909
fbshipit-source-id:
42139bd7caee0fedff13328d52fa3be1c517e730
Andrii Grynenko [Thu, 6 Apr 2017 17:10:38 +0000 (10:10 -0700)]
Require registrationComplete() not only in dbg builds
Summary: Improve error message and change DFATAL to FATAL.
Reviewed By: yfeldblum, meyering
Differential Revision:
D4837840
fbshipit-source-id:
54ae53c9356fc2e85266b037996aa7b2bae3d4e4
Phil Willoughby [Thu, 6 Apr 2017 11:03:02 +0000 (04:03 -0700)]
Update StringKeyed... benchmarks to tell the whole story
Summary:
Benchmark each type with its native key type and with its non-native key type.
This allows you to make a more-informed decision about which to use based on
the types you have available at insertion/lookup time.
Reviewed By: yfeldblum
Differential Revision:
D4826527
fbshipit-source-id:
ee4ef5255d2bdd955e9deb30dc4c28be183cee07
Maged Michael [Thu, 6 Apr 2017 10:04:20 +0000 (03:04 -0700)]
IndexedMemPool: Fix race condition on size_ that can cause the destructor to access nonexistent slots
Summary:
Contention on allocation when there is exactly one slot remaining can cause size_ to exceed actualCapacity_.
Without taking the min of size_ and actualCapacity_ the destructor may try to destroy out-of-bound slots.
Added a test to test/IndexedMemPoolTest.cpp that failed before the fix and passes after the fix.
Reviewed By: nbronson
Differential Revision:
D4837251
fbshipit-source-id:
a887487727f17eaf2ba66345f40fc91d2fe3bc00
Alan Frindell [Wed, 5 Apr 2017 23:37:16 +0000 (16:37 -0700)]
Make NotificationQueue::Consumer::messageAvailable noexcept
Summary: There's a comment in that code that states that it *it* noexcept, and that is common for most folly async io callbacks.
Reviewed By: yfeldblum
Differential Revision:
D4831800
fbshipit-source-id:
78894ad72504b9dfe540c14b8d7bbb29247aaf87
Subodh Iyengar [Wed, 5 Apr 2017 22:15:49 +0000 (15:15 -0700)]
Add trimAtMost functions
Summary:
Add trimAtMost start and trimAtMost end.
These are useful functions to have if you
don't care about trimming an exact amount.
Reviewed By: yfeldblum
Differential Revision:
D4658087
fbshipit-source-id:
f7dd62c0f4837503e59cdec75fdad433be20306c
Christopher Dykes [Wed, 5 Apr 2017 19:50:28 +0000 (12:50 -0700)]
Disable EnvUtil::setAsCurrentEnvironment() on platforms without clearenv()
Summary:
Implementing `clearenv()` correctly on multiple OS's without relying on underlying implementation details is significantly more complicated, and this is preventing being able to build on OSX.
This function isn't actually used anywhere currently.
Reviewed By: yfeldblum
Differential Revision:
D4832473
fbshipit-source-id:
a80aabb5a223264746ab45e3138d065bce5fe99c
Eli Lindsey [Tue, 4 Apr 2017 19:49:07 +0000 (12:49 -0700)]
use folly singleton relaxed init on android
Summary: fixing
D4825412
Reviewed By: lbrandy
Differential Revision:
D4826815
fbshipit-source-id:
94691ff293dad53b16e7a2f139c1df490a61e565
Maxim Georgiev [Tue, 4 Apr 2017 19:16:07 +0000 (12:16 -0700)]
Make some AsyncTest methods virtual to allow mocking them using gtest/gmock
Summary: GMock can't mock class methods that are not virtual. This way classes relying on the non-virtual methods of AsyncSocket can't be efficiently tested using gtest/gmock.
Reviewed By: yfeldblum
Differential Revision:
D4806838
fbshipit-source-id:
fde852b75f7ac1d468c177e317c516cb0dc333b5
Sven Over [Tue, 4 Apr 2017 16:53:30 +0000 (09:53 -0700)]
Function.h: remove FOLLY_MSVC_DISABLE_WARNING
Summary:
We recently removed declarations of Function constructors and
assignment operators that took non-const Function references. They
were necessary at the time to make sure that the templated
constructor that can take any invokable object did not get
selected for copying Functions from non-const lvalues. This,
however, is now provided by some std::enable_if construct.
Reviewed By: ericniebler
Differential Revision:
D4818736
fbshipit-source-id:
08197721beef009f6ae482b576df78ec6999d132
Anirudh Ramachandran [Tue, 4 Apr 2017 06:30:01 +0000 (23:30 -0700)]
Support building with OpenSSL 1.1.0 and BoringSSL
Summary:
More work to get wangle compiling. wangle/facebook/http pulls in
proxygen libs and that's another pain altogether, so this only makes the rest of
wangle build with 1.1.0 and BoringSSL
Depends on
D4406876
Reviewed By: ngoyal
Differential Revision:
D4767060
fbshipit-source-id:
bd6bc6959d04028c84360e434f6bbdb2cde2faac
Anirudh Ramachandran [Tue, 4 Apr 2017 06:29:59 +0000 (23:29 -0700)]
OpenSSL 1.1.0 compatibility
Summary: Add more layers to support compiling with OpenSSL 1.1.0
Reviewed By: yfeldblum
Differential Revision:
D4406876
fbshipit-source-id:
8eff7e4f8e096605fc9e1093ef533b5afe0ee539
Andrii Grynenko [Tue, 4 Apr 2017 03:04:44 +0000 (20:04 -0700)]
Don't allow singleton use w/o registrationComplete()
Summary: This is same as
D4249032, but for now this is only enabled in dbg and dev builds. I'm planning to bump this to FATAL later, once we fix know problematic main()s.
Reviewed By: lbrandy, yfeldblum
Differential Revision:
D4821763
fbshipit-source-id:
5f7930f8cbcb10275d23a89848f1ec8ee34a8020
Nick Terrell [Mon, 3 Apr 2017 23:51:30 +0000 (16:51 -0700)]
Remove use of generic lambdas
Summary:
The use of generic lambdas caused compiler errors with gcc 4.8.2.
Fixes https://github.com/facebook/folly/issues/569.
Reviewed By: Orvid, yfeldblum
Differential Revision:
D4821115
fbshipit-source-id:
8372ee7695a3d0a1df0d033623618a923c261737
Harrison Klaperman [Mon, 3 Apr 2017 23:31:57 +0000 (16:31 -0700)]
Fix broken ManualExecutor
Summary:
The priority queue in the manual executor implementation is backwards. This means that scheduled things run in reverse order, and a later thing will block an earlier thing if you advance to a timestamp in between the two.
This diff fixes the problem and adds tests to confirm the fix. These tests fail on the old implementation.
Reviewed By: yfeldblum
Differential Revision:
D4739101
fbshipit-source-id:
6e429828460df5b3c656580568a9ae1eb4009527
Christopher Dykes [Mon, 3 Apr 2017 21:51:22 +0000 (14:51 -0700)]
Don't use macros for FBVector::insert
Summary: Macros are not the best solution anymore.
Reviewed By: yfeldblum
Differential Revision:
D4270626
fbshipit-source-id:
8acbd5b778d4effcfb587655b4397737bef0d379
Dan Melnic [Mon, 3 Apr 2017 21:09:47 +0000 (14:09 -0700)]
Remove/make private the default ***Holder constructor to allow compile time detection of ***Holder(mutexPtr_) constructs
Summary: Avoid unintended C++ stuff - this is not the same as the "fleeting rvalue" - this change might break the build but I will fix it
Reviewed By: yfeldblum
Differential Revision:
D4814104
fbshipit-source-id:
058a0eac44893c573062fcf5665d4fd022ee64a0
Dave Watson [Mon, 3 Apr 2017 17:51:19 +0000 (10:51 -0700)]
atomic_shared_ptr
Summary:
A (almost) lock-free atomic_shared_ptr, matching the proposed concurrency TS interface.
See notes at top of file
Reviewed By: davidtgoldblatt
Differential Revision:
D4716098
fbshipit-source-id:
b9ca2443ba9e227ebb6f40807128073c6e14222a
Giuseppe Ottaviano [Sat, 1 Apr 2017 05:17:57 +0000 (22:17 -0700)]
Optimize JSON escaping of ASCII strings
Summary:
`escapeString` is very slow even when there's very little to nothing to escape. This diff adds a fast path to copy sequences of bytes that don't need escaping.
It also optimizes appending escape sequences: `string::push_back` is slow because it has to do a capacity check for every character.
Before:
```
============================================================================
folly/test/JsonOtherTest.cpp relative time/iter iters/s
============================================================================
jsonSerialize 818.55ns 1.22M
jsonSerializeWithNonAsciiEncoding 1.35us 738.06K
jsonSerializeWithUtf8Validation 1.42us 705.60K
jsonSerializeAsciiWithUtf8Validation 3.27us 306.06K
parseSmallStringWithUtf 1.91us 522.38K
parseNormalString 1.51us 660.27K
parseBigString 384.44ns 2.60M
toJson 480.54ns 2.08M
============================================================================
```
After:
```
============================================================================
folly/test/JsonOtherTest.cpp relative time/iter iters/s
============================================================================
jsonSerialize 781.69ns 1.28M
jsonSerializeWithNonAsciiEncoding 847.68ns 1.18M
jsonSerializeWithUtf8Validation 928.68ns 1.08M
jsonSerializeAsciiWithUtf8Validation 199.85ns 5.00M
parseSmallStringWithUtf 1.93us 518.39K
parseNormalString 1.45us 689.11K
parseBigString 378.66ns 2.64M
toJson 446.38ns 2.24M
============================================================================
```
All string escaping benchmarks are slightly faster, and ASCII-only with no escapes is 8x faster.
Reviewed By: luciang, evilmucedin, yfeldblum
Differential Revision:
D4793233
fbshipit-source-id:
c40d07708bd787799c8c00f9f23a417b862ee9ae
Mainak Mandal [Sat, 1 Apr 2017 02:01:12 +0000 (19:01 -0700)]
fix memory leak in case of large number of retries
Summary: Infinite retries is something that is often needed for read-modify-write like workflows. The current implementation was creating a nested chain of implicit promises. This manifests as a memory leak after some time. Worse yet, even if it succeeds, it will take a long time to churn thru the chain of promises.
Reviewed By: yfeldblum
Differential Revision:
D4770335
fbshipit-source-id:
44b8d6df1084de4514b66919a9838cf2322d6dce
Wez Furlong [Fri, 31 Mar 2017 06:38:50 +0000 (23:38 -0700)]
avoid UB in StringKeyedCommon for default constructed StringPiece
Summary:
ubsan flags this problem:
```
buck-out/opt-ubsan/gen/folly/__default_headers__#header-mode-symlink-tree-with-header-map,headers/folly/experimental/StringKeyedCommon.h:31:18: runtime error: null pointer passed as argument 2, which is declared to never be null
third-party-buck/gcc-4.9-glibc-2.20-fb/build/glibc/include/string.h:47:28: note: nonnull attribute specified here
#0 0x5cb88f in std::pair<std::__detail::_Node_iterator<std::pair<folly::Range<char const*> const, facebook::eden::GitIgnore>, false, true>, bool> folly::StringKeyedUnorderedMap<facebook::eden::GitIgnore, folly::Hash, std::equal_to<folly::Range<char const*> >, std::allocator<std::pair<folly::Range<char const*> const, facebook::eden::GitIgnore> > >::emplace<facebook::eden::GitIgnore>(folly::Range<char const*>, facebook::eden::GitIgnore&&) buck-out/opt-ubsan/gen/folly/__default_headers__#header-mode-symlink-tree-with-header-map,headers/folly/experimental/StringKeyedCommon.h:31
#1 0x5c6652 in facebook::eden::(anonymous namespace)::IgnoreChecker::isIgnored(facebook::eden::detail::RelativePathBase<folly::Range<char const*> >) eden/fs/inodes/Dirstate.cpp:226
#2 0x5c6037 in facebook::eden::(anonymous namespace)::IgnoreChecker::isIgnored(facebook::eden::detail::RelativePathBase<folly::Range<char const*> >) eden/fs/inodes/Dirstate.cpp:180
#3 0x5c5173 in facebook::eden::Dirstate::getStatusForExistingDirectory(facebook::eden::detail::RelativePathBase<folly::Range<char const*> >) const eden/fs/inodes/Dirstate.cpp:372
#4 0x5c4509 in facebook::eden::Dirstate::getStatus() const eden/fs/inodes/Dirstate.cpp:272
#5 0x50a448 in verifyExpectedDirstate(facebook::eden::Dirstate const*, std::unordered_map<std::basic_fbstring<char, std::char_traits<char>, std::allocator<char>, std::fbstring_core<char> >, facebook::eden::StatusCode, std::hash<std::basic_fbstring<char, std::char_traits<char>, std::allocator<char>, std::fbstring_core<char> > >, std::equal_to<std::basic_fbstring<char, std::char_traits<char>, std::allocator<char>, std::fbstring_core<char> > >, std::allocator<std::pair<std::basic_fbstring<char, std::char_traits<char>, std::allocator<char>, std::fbstring_core<char> > const, facebook::eden::StatusCode> > >&&) eden/fs/inodes/test/DirstateTest.cpp:48
#6 0x50c70c in Dirstate_addDirectoriesWithMixOfFiles_Test::TestBody() eden/fs/inodes/test/DirstateTest.cpp:220
#7 0xb45857 in void testing::internal::HandleExceptionsInMethodIfSupported<testing::Test, void>(testing::Test*, void (testing::Test::*)(), char const*) /home/engshare/third-party2/gtest/1.7.0/src/gtest-1.7.0/./src/gtest.cc:2364
#8 0xb36784 in testing::Test::Run() /home/engshare/third-party2/gtest/1.7.0/src/gtest-1.7.0/./src/gtest.cc:2437
#9 0xb36957 in testing::TestInfo::Run() [clone .part.558] /home/engshare/third-party2/gtest/1.7.0/src/gtest-1.7.0/./src/gtest.cc:2612
#10 0xb36b74 in testing::TestCase::Run() [clone .part.559] /home/engshare/third-party2/gtest/1.7.0/src/gtest-1.7.0/./src/gtest.cc:2587
#11 0xb3806e in testing::internal::UnitTestImpl::RunAllTests() [clone .part.561] /home/engshare/third-party2/gtest/1.7.0/src/gtest-1.7.0/./src/gtest.cc:4571
#12 0xb382d9 in testing::UnitTest::Run() /home/engshare/third-party2/gtest/1.7.0/src/gtest-1.7.0/./src/gtest.cc:4519
#13 0x861ca7 in main third-party-buck/gcc-4.9-glibc-2.20-fb/build/gtest/include/gtest/gtest.h:2326
#14 0x7f20be0740f5 in __libc_start_main /home/engshare/third-party2/glibc/2.20/src/glibc-2.20/csu/libc-start.c:289
#15 0x4a552c in _start /home/engshare/third-party2/glibc/2.20/src/glibc-2.20/csu/../sysdeps/x86_64/start.S:122
UndefinedBehaviorSanitizer: undefined-behavior buck-out/opt-ubsan/gen/folly/__default_headers__#header-mode-symlink-tree-with-header-map,headers/folly/experimental/StringKeyedCommon.h:31:18
```
The issue is that `StringPiece` default constructs to `{nullptr, nullptr}` as a valid representation of an empty string.
This is tripping up UBSAN in this case.
The fix is a trivial nullptr check
Reviewed By: igorsugak
Differential Revision:
D4791015
fbshipit-source-id:
dec7484b29ecb29c17b8dd6a9b0e8093f07d63cb
Yedidya Feldblum [Fri, 31 Mar 2017 04:38:09 +0000 (21:38 -0700)]
Some constexpr symbols for whether sanitizers are enabled
Summary:
[Folly] Some constexpr symbols for whether sanitizers are enabled.
Useful where we need to do or not do some particular thing when sanitizers are enabled or disabled. Such cases do happen. It is better to use a C++ `constexpr` symbol rather than a preprocessor symbol in such cases, where possible.
Reviewed By: meyering
Differential Revision:
D4805110
fbshipit-source-id:
ca9ce074edef73dce3a04693dda102698c9cd29f
Yedidya Feldblum [Thu, 30 Mar 2017 22:43:00 +0000 (15:43 -0700)]
Fix double-delete in Singleton::make_mock
Summary:
[Folly] Fix double-`delete` in `Singleton::make_mock`.
In that function, we unconditionally destroy any existing singleton instance of the same singleton. Actually, we must conditionally destroy it - conditionally on one having been created and not yet destroyed.
This problem only rarely appears because it is masked by `delete static_cast<T*>(nullptr)` being a no-op. For example, if we call `make_mock` before actually instantiating the singleton, we call that no-op. The way to make it appear is to instantiate the singleton, and then to call `make_mock` twice in a row. The first call to `make_mock` unconditionally destroys the existing instance (it still should have checked) and the second call does it again, but because the existing instance is not `nullptr`, the second call is a double-delete of a non-`nullptr` instance and crashes. In the simple case, as reproduced in an attached test, the failure is observable with ASAN. In other cases, the double-`free` is tolerated, but the failure may only be observable depending on the singleton object's state - if running the dtor twice fails.
Reviewed By: ivmaykov
Differential Revision:
D4798163
fbshipit-source-id:
e7b65d030d89225dfdc2fad4c778d3653460806e
Eric Niebler [Thu, 30 Mar 2017 21:09:34 +0000 (14:09 -0700)]
Remove unneccessary test for &that==this in folly::Function's move assignment operator.
Summary: Self-moves are exceedingly rare and need not preserve the state of the object. They must only leave the object in a valid but unspecified state. By removing the branch, we make the common case (non-self move) faster.
Reviewed By: spacedentist, ot
Differential Revision:
D4803486
fbshipit-source-id:
3ef2e1e13cd08d9221ecb154bfb3338b16487717
Eric Niebler [Thu, 30 Mar 2017 21:09:32 +0000 (14:09 -0700)]
Properly constrain folly::Function's generic conversion constructor and fix its noexcept specification
Summary:
The generic conversion constructor for `folly::Function` was not checking that the source object could successfully be copy/move constructed, leading to some `is_constructible` false positives.
Also, the `noexcept` specification on the `Function` constructor wasn't taking into account that the source object might be copied into the Function, instead of moved. The copy could throw.
Reviewed By: yfeldblum
Differential Revision:
D4775037
fbshipit-source-id:
f337b41bf9ac431baa9457a501e63c18ca099e57
Nick Terrell [Thu, 30 Mar 2017 19:48:11 +0000 (12:48 -0700)]
Path variants for get_(ref_)default()
Summary: Useful for the same reason as the `get_ptr()` path function, but when you want to use a default value.
Reviewed By: luciang, yfeldblum
Differential Revision:
D4785728
fbshipit-source-id:
70fd56f9ffa7a9edd6740f6dd712d3a251bf9fb0
Igor Sugak [Thu, 30 Mar 2017 01:14:16 +0000 (18:14 -0700)]
Backed out changeset
d4e3fdfd5775
Summary:
https://our.intern.facebook.com/intern/sevmanager/view/s/145809/?hc_location=ufi
```
hg backout
d4e3fdfd57750c8dcabdb4c01ab3b528c4818624
```
Reviewed By: lbrandy
Differential Revision:
D4799568
fbshipit-source-id:
76366040702e68becc7e245d7ce15832f933e10e
Tom Jackson [Wed, 29 Mar 2017 22:17:25 +0000 (15:17 -0700)]
uncurry
Summary: Extending ApplyTuple to support this common functional construct, with good forwarding semantics.
Reviewed By: yfeldblum
Differential Revision:
D4787560
fbshipit-source-id:
2c740e448e0cb916abe948b79709d5ecd8ba54bb
Yedidya Feldblum [Wed, 29 Mar 2017 17:17:46 +0000 (10:17 -0700)]
Cut extraneous typename in StringKeyedCommon.h
Summary:
[Folly] Cut extraneous typename in `StringKeyedCommon.h`.
In this case, `typename` is optional, and, to avoid confusion, should be removed.
"In this case" means in a template definition, with a qualified non-dependent name.
Reviewed By: ericniebler
Differential Revision:
D4792087
fbshipit-source-id:
dbb9d2c747432a1f9bc8823e72a7668e78a2fbaa
Sven Over [Wed, 29 Mar 2017 10:48:52 +0000 (03:48 -0700)]
fix dead-lock in Future when executor discards function
Summary:
This diff adds two tests to futures/test/ViaTest.cpp:
viaDummyExecutorFutureSetValueFirst and
viaDummyExecutorFutureSetCallbackFirst. The latter resulted in a
dead-lock before the fix contained in this diff.
It is important that the callback function is destroyed after
it is executed, since it may capture objects (like a Promise)
that should be destroyed (so that e.g. a corresponding Future
throws BrokenPromise). When the callback is executed via an
executor, it is possible that the executor doesn't get around to
executing the task. We shouldn't rely on the task being executed
to do necessary clean-up. That clean-up should (also) happen
when the task (with its captured data) is destroyed (in the
spirit of RIAA).
Reviewed By: djwatson
Differential Revision:
D4779215
fbshipit-source-id:
d029cf8b8f7b55e1b03357749c5fb62d95986ca7
Nick Terrell [Wed, 29 Mar 2017 04:21:02 +0000 (21:21 -0700)]
AutomaticCodec
Summary:
Add codec that does automatic format detection for decompression.
The codec returned by `getCodec(CodecType::AUTOMATIC)` can decompress codecs `LZ4_FRAME`, `ZSTD`, `ZLIB`, `GZIP`, and `LZMA2`.
Additionally, the function `getAutomaticCodec()` allows users to pass their own custom codecs, but the 5 default codecs must also be supported.
We force the default codecs on users because having a common subset of formats is in general useful, and because we want to be able to add codecs to this set.
When compiled in debug mode, on construction it checks that the codecs are compatible.
This helps make sure that users don't accidentally add ambiguous codecs, and if we add a new default codec in the future, tests will fail on ambiguity.
Reviewed By: yfeldblum
Differential Revision:
D4760065
fbshipit-source-id:
9e65844aba9f1ae3d5c7b86643931782c06ab3eb
Ondrej Lehecka [Wed, 29 Mar 2017 01:58:12 +0000 (18:58 -0700)]
Cursor::read<T>() initializes return value
Summary:
If the compiler doesn't see thru the call to tryRead which always initializes the val variable, it will emit the following error.
I was getting error when compiling my android project:
error: 'val' may be used uninitialized in this function [-Werror=maybe-uninitialized]
This should not have a negative perf impact as smart compilers which inline call to tryRead can elliminate the initialization.
Reviewed By: djwatson
Differential Revision:
D4790864
fbshipit-source-id:
f353cfe54ca4d056b6ddfc075d00580c9f2d391e
Nick Terrell [Tue, 28 Mar 2017 23:30:48 +0000 (16:30 -0700)]
Optimize ZlibCodec::doUncompress()
Summary:
`ZlibCodec::doUncompress()` always allocates at least 4 MiB when the uncompressed size is unkown.
We can get a 10-20% speed boost for small inputs (as well as significant memory savings) if we dynamically choose the buffer size.
Benchmarks before the change:
============================================================================
experimental/terrelln/CompressionBenchmark.cpp relative time/iter iters/s
============================================================================
uncompress(ZLIB_100) 2.96us 338.19K
uncompress(ZLIB_1000) 7.22us 138.44K
uncompress(ZLIB_10000) 36.86us 27.13K
uncompress(ZLIB_100000) 299.70us 3.34K
uncompress(ZLIB_1000000) 2.74ms 365.32
uncompress(ZLIB_10000000) 26.91ms 37.16
uncompress(ZLIB_100000000) 273.92ms 3.65
============================================================================
After:
============================================================================
experimental/terrelln/CompressionBenchmark.cpp relative time/iter iters/s
============================================================================
uncompress(ZLIB_100) 2.34us 426.68K
uncompress(ZLIB_1000) 6.24us 160.38K
uncompress(ZLIB_10000) 35.52us 28.16K
uncompress(ZLIB_100000) 296.30us 3.37K
uncompress(ZLIB_1000000) 2.74ms 365.25
uncompress(ZLIB_10000000) 27.16ms 36.82
uncompress(ZLIB_100000000) 273.72ms 3.65
============================================================================
Reviewed By: yfeldblum
Differential Revision:
D4783871
fbshipit-source-id:
0fcce191ea520fd2117551db679c86f5ff734893
Christopher Dykes [Tue, 28 Mar 2017 18:21:24 +0000 (11:21 -0700)]
Implement S_ISDIR()
Summary: It's needed downstream in HHVM.
Reviewed By: yfeldblum
Differential Revision:
D4785204
fbshipit-source-id:
fe81ed8e23497fffeb44b2a366c457b788bcaada
Maged Michael [Tue, 28 Mar 2017 15:36:31 +0000 (08:36 -0700)]
Change trylock() to try_lock() in folly::SpinLock to conform to standard Lockable.
Reviewed By: yfeldblum
Differential Revision:
D4782707
fbshipit-source-id:
535b42b4f2558cadc78037268d6de81a8bb49840
Dylan Yudaken [Tue, 28 Mar 2017 15:02:42 +0000 (08:02 -0700)]
Allow custom sorting function for JSON serializer
Summary: In some situations it is useful to be able to sort the JSON keys according to some different scheme for readability, so allow the sorting function to be changed by the serializer_opts
Reviewed By: yfeldblum
Differential Revision:
D4782077
fbshipit-source-id:
032fa60a38804452bd1c22c67ba897521cb2cd1d
Nick Terrell [Mon, 27 Mar 2017 23:14:38 +0000 (16:14 -0700)]
Gate LZ4Frame behind version check
Summary:
D4715918 broke open source builds on Ubuntu 14.04, since it has lz4-r114, but the lz4 frame API was introduced in [r123](https://github.com/lz4/lz4/blob/r123/lz4frame.h).
Put the `LZ4FrameCodec` behind a lz4 version check.
Fixes https://github.com/facebook/fbthrift/issues/209.
Reviewed By: yfeldblum
Differential Revision:
D4780830
fbshipit-source-id:
19492a7e6bdd128e610c36b5778274e19eff9548
Nick Terrell [Mon, 27 Mar 2017 20:03:02 +0000 (13:03 -0700)]
Disallow temporary default values in get_ref_default()
Summary: Passing a temporary default value to `get_ref_default()` returns a dangling reference when it is used.
Reviewed By: lbrandy, yfeldblum
Differential Revision:
D4768769
fbshipit-source-id:
4c58a17dc7662ad553cf88a8544dae20016d2f6f
Arthur O'Dwyer [Sat, 25 Mar 2017 20:43:31 +0000 (13:43 -0700)]
Several fixes to the "SWMRList" example in experimental/hazptr.
Summary:
Fix a correctness bug in "SWMRList.h".
Thanks to Maged for the tip! The old code omitted setting a removed node's
"next" pointer to `nullptr`, which meant that if the writer removed node
A and then node B = A->next while the reader was looking at B, then the reader
might happily keep chasing pointers B->next, B->next->next,... without ever
noticing that it was now on a "dead branch" of the linked list.
(And then there's a bit of a trick: you really do have to remove the node
first and *then* set its "next" pointer to null, because if you do the null
assignment first, you'll truncate the list, which means that some readers
will see a truncated list and return the wrong answer. I've added comments
to try to explain this to future-me.)
Style nit: Avoid doing multiple atomic operations on the same line of code.
----
Modernize the parameter-passing conventions in SWMRList.h.
Pass `T`s by const reference, except in `add` where we're likely to want to
make a copy of the `T` anyway. In that case it would be more "STL-correct"
to supply two different overloads `add(T&&)` and `add(const T&)`, but this
is just an example so it makes sense to keep things simple.
----
Fix an undefined behavior in SWMRList example.
Searching an empty SWMRList<int> always invokes undefined behavior by
comparing an uninitialized `T elem` against `val` on the final line of
the `contains` function. Besides, this patch allows SWMRList to work
with `T`s that aren't default-constructible or even copy-constructible.
----
Closes https://github.com/facebook/folly/pull/566
Reviewed By: djwatson
Differential Revision:
D4772359
Pulled By: yfeldblum
fbshipit-source-id:
8f96573530800675cb56006aa91e7a5c5c1fb85d
Maged Michael [Sat, 25 Mar 2017 10:28:15 +0000 (03:28 -0700)]
(folly) Add lock holder interface to flat combining. Change an assert to a static_assert.
Summary:
Two changes:
- Add a lock holder interface to enable users to move exclusive access and implicitly release the held lock.
- Change an assert to a static_assert to check at compile time if a provided function fits in the Function structure without dynamic allocation.
Reviewed By: djwatson
Differential Revision:
D4770229
fbshipit-source-id:
89408164c08d7660231a6ca4e37637dd688356cd
Nick Terrell [Sat, 25 Mar 2017 01:12:06 +0000 (18:12 -0700)]
Add tryRead() and endian variants
Summary:
Add a `tryRead()`, and endian variants, which try to read into an arithmetic type, and if there isn't enough data they return false.
One use case is to quickly check if an IOBuf starts with a certain prefix, benchmarks show that using `tryReadLE()` is 6x faster than using `pullAtMost()` and `memcmp()`.
Reviewed By: yfeldblum
Differential Revision:
D4767855
fbshipit-source-id:
feb8c61092772933d4b8496b27d464559ff8b827
Nick Terrell [Fri, 24 Mar 2017 22:00:07 +0000 (15:00 -0700)]
Avoid passing temporary to get_ref_default()
Summary: `get_ref_default()` won't accept temporary default values after
D4768769.
Reviewed By: yfeldblum
Differential Revision:
D4769043
fbshipit-source-id:
46b039e13b80721922f69036c6cc02f5642a26f9
Nick Terrell [Fri, 24 Mar 2017 21:30:45 +0000 (14:30 -0700)]
Allow decompresion without uncompressedSize
Summary:
The `LZMA2` codec doesn't allow decompression without passing the uncompressed size.
Every other codec that supports streaming decompression without the uncompressed size by streaming into an IOBuf chain.
Benchmarks show that reducing the `defaultBufferLength` to 256 KiB doesn't slow down decompression of large files, but it speeds up decompression of small files (< 1000 bytes).
Reviewed By: yfeldblum
Differential Revision:
D4751571
fbshipit-source-id:
39dbe6754a1ecdc2b7ba3107e9face926d4c98ca
Nick Terrell [Fri, 24 Mar 2017 19:18:57 +0000 (12:18 -0700)]
Optimize ZSTDCodec::doUncompress()
Summary:
It is ~10% faster to call `ZSTD_decompress()` than use the
streaming API. The streaming API does some extra `memcpy`s that we
can avoid. We are working on improving the speed of the streaming
API in the case where all the data can be processed in one shot,
but that won't be available in the stable ZSTD api for a few versions.
Reviewed By: yfeldblum
Differential Revision:
D4731058
fbshipit-source-id:
39026c499c0f5002466097b5afe7e30f850e0ae8
Nick Terrell [Fri, 24 Mar 2017 18:08:17 +0000 (11:08 -0700)]
Add LZ4_FRAME codec
Summary:
The LZ4 Frame codec encodes data using the LZ4 frame format.
One advantage of the LZ4 frame format is that it has 4 magic bytes in the header, so users can transparently determine compression type.
It also allows the user to interop with the lz4 command line tool.
Reviewed By: yfeldblum
Differential Revision:
D4715918
fbshipit-source-id:
689833fef526b1cfe98685179e7b494380d49cba
Phil Willoughby [Fri, 24 Mar 2017 09:05:42 +0000 (02:05 -0700)]
Expand environment-handling in folly
Summary:
Extract the environment-as-STL-map logic out of `test::EnvVarSaver` and into a
separate class `experimental::EnvironmentState` so that other code which needs
to manipulate the environment can do so more easily.
Add routines to set the process environment from the state of an
`EnvironmentState`'s map, and to extract the environment in the forms required
by `Subprocess` and the UNIX routines `execve` etc.
Reviewed By: yfeldblum
Differential Revision:
D4713307
fbshipit-source-id:
6b1380dd29b9ba41c97b886814dd3eee91fc1c0f