Граф коммитов

50 Коммитов

Автор SHA1 Сообщение Дата
Jeff Walden a3d9df23d5 Bug 1407137 - Implement Swap(Atomic<T, Order>&, Atomic<T, Order>&) = delete; to remove an actually-non-atomic footgun. r=froydnj
--HG--
extra : rebase_source : 6afed8e85fcf8801144a1a34ea07f590baa0f5e0
2017-10-10 10:19:31 -07:00
Nathan Froyd e5fbad82a1 Bug 1363426 - part 3 - minor cleanups to Atomics.h; r=erahm
We can move the <atomic> #include to its normal place in the file now,
as well as merging some namespace blocks at the toplevel.
2017-05-10 09:58:28 -04:00
Nathan Froyd c0a40e46f7 Bug 1363426 - part 2 - remove unneeded MOZ_HAVE_CXX1_ATOMICS; r=erahm
Now that we can use <atomic> everywhere, we don't need to test whether
the standard library we're using has <atomic> support.
2017-05-10 09:58:28 -04:00
Nathan Froyd fe42aa0b9d Bug 1363426 - part 1 - remove #ifdeffery in Atomics.h; r=erahm
Every platform where we use GCC has <atomic>, so there's no need to use
GCC-specific __sync* intrinsics anymore.  The <atomic> header may
generate better code for several operations, as well.
2017-05-10 09:58:28 -04:00
Chris Peterson 43c2748f66 Bug 1277775 - Replace MOZ_CONSTEXPR{_VAR,_TMPL} with constexpr. r=froydnj 2016-07-08 14:39:53 -07:00
Nathan Froyd 6a5930b454 Bug 1247338 - really make Atomic's constructor constexpr; r=Waldo
Atomic's constructor is marked as constexpr, but it calls a
non-constexpr function, ToStorageTypeArgument::convert.  For compilers
which require constexpr-ness on constructors to inline away the actual
constructor call, the call to ToStorageTypeArgument::convert completely
disables the constexpr-ness of the constructor.  Let's fix this by
marking all relevant instances of ToStorageTypeArgument::convert as
MOZ_CONSTEXPR, thus satisfying the compiler once again.
2016-02-10 11:28:36 -05:00
Jeff Walden fe5445d030 Bug 1220693 - Make mozilla::Atomic<enum class> work even on compilers that don't have <atomic>. r=froydnj
--HG--
extra : rebase_source : 2eb167a1b86a1a93527761a9541ea30ad213785a
2015-11-03 13:03:26 -08:00
Nathan Froyd 78b4b30d08 Bug 1189894 - remove Atomics.h IntrinsicAddSub hack; r=erahm
This hack was only required for broken headers in GCC 4.6.  Since we
only support GCC 4.7+ now, this hack is no longer necessary.
2015-07-31 22:25:21 -04:00
David Major ec86960f3c Bug 1185686: Fix Atomics.h ifdefs for clang-cl. r=froydnj 2015-07-20 17:04:46 -04:00
Ehsan Akhgari 399276d5fc Bug 1153348 - Add an analysis to prohibit operator bools which aren't marked as either explicit or MOZ_IMPLICIT; r=jrmuizel
This is the counterpart to the existing analysis to catch
constructors which aren't marked as either explicit or
MOZ_IMPLICIT.
2015-04-21 21:40:49 -04:00
Wes Kocher 638b28b8e0 Backed out changeset 02e6a50741a9 (bug 1153348) to hopefully fix the static bustage CLOSED TREE 2015-04-21 15:47:40 -07:00
Ehsan Akhgari 3d21a05904 Bug 1153348 - Add an analysis to prohibit operator bools which aren't marked as either explicit or MOZ_IMPLICIT; r=jrmuizel
This is the counterpart to the existing analysis to catch
constructors which aren't marked as either explicit or
MOZ_IMPLICIT.
2015-04-21 16:31:58 -04:00
Seth Fowler 36709ff42c Bug 1155864 - Rip out Windows intrinsics from Atomics.h and allow Atomic to hold 64-bit types everywhere. r=Waldo 2015-04-17 18:40:52 -07:00
David Major 13d9def98b Bug 1154104 - Clean up atomics code for old Visual Studio versions. r=nfroyd
--HG--
extra : rebase_source : 83ed35150a55c26c66994f3f7a6c4ae7ae14ba66
2015-04-13 17:13:00 +02:00
Trevor Saunders 5b86ecf80c bug 1077549 - remove useless gcc version checks r=glandium 2015-01-15 21:30:05 -05:00
Nicholas Nethercote e147e4d209 Bug 1057134 - Remove std::_Debug_message linking hack that's no longer needed. r=dmajor.
--HG--
extra : rebase_source : 53ee0cebc42269dcc228cfc5a18b0288e13716f3
2015-01-08 21:59:43 -08:00
Ehsan Akhgari 4354953b4f Bug 1118486 - Part 1: Use `= delete` instead of MOZ_DELETE directly; r=Waldo
Most of this patch (with the exception of dom/bindings/Codegen.py) was
generated by the following bash script:

#!/bin/bash

function convert() {
echo "Converting $1 to $2..."
find . ! -wholename "*nsprpub*" \
       ! -wholename "*security/nss*" \
       ! -wholename "*/.hg*" \
       ! -wholename "*/.git*" \
       ! -wholename "obj-*" \
         -type f \
      \( -iname "*.cpp" \
         -o -iname "*.h" \
         -o -iname "*.cc" \
         -o -iname "*.idl" \
         -o -iname "*.ipdl" \
         -o -iname "*.ipdlh" \
         -o -iname "*.mm" \) | \
    xargs -n 1 sed -i -e "s/\b$1\b/$2/g"
}

convert MOZ_DELETE '= delete'
2015-01-08 23:19:05 -05:00
Ehsan Akhgari 039292de5d Bug 1119071 - Clean up some code used for older unsupported MSVC versions in MFBT; r=froydnj 2015-01-08 10:40:15 -05:00
Jan Beich 2e6d8524b0 Bug 1073709 - Disable C++11 <atomic> when using GCC with libc++. r=waldo 2014-09-26 15:15:00 +02:00
Hannes Verschore e729b5e319 Bug 1061764 - Atomic<bool, Relaxed> should mean no synchronization on windows x86, r=froydnj 2014-09-04 11:08:00 +02:00
Nicholas Nethercote 2a0942be41 Bug 1036789 - Convert the third quarter of MFBT to Gecko style. r=Ms2ger.
--HG--
extra : rebase_source : 668cd394806203ddfa34bd4f226335ff26c846b5
2014-07-10 19:10:17 -07:00
Ehsan Akhgari c2cbd699a1 Bug 1025900 - #include intrin.h directly in Atomics.h instead of declaring functions manually from it; r=froydnj
This is basically more of bug 1022049.

--HG--
extra : rebase_source : 39cf2745d47d90ba1684c2c26a38f9781a4d179e
2014-06-16 18:19:49 -04:00
Ehsan Akhgari ac9615fa0d Bug 1021290 - More Atomics.h build fixes for clang-cl. r=froydnj
This is basically more of bug 932454
2014-06-09 14:49:34 -04:00
Ehsan Akhgari 8b30e40485 Bug 1013662 - Fix bad implicit conversion constructors in MFBT; r=froydnj,Waldo 2014-06-06 23:17:06 -04:00
Nathan Froyd 002c131055 Bug 1020118 - fix atomic add/sub for pointer types on Win64; r=jcranmer 2014-06-04 15:19:51 -04:00
Wes Kocher 8ae9f3fb4a Backed out changeset 34f371b1024b (bug 1021290) for Windows build bustage on a CLOSED TREE 2014-06-05 17:35:47 -07:00
Jeff Muizelaar c50b5c14aa Bug 1021290. More Atomics.h build fixes for clang-cl. r=froydnj
This is basically more of bug 932454

--HG--
extra : rebase_source : 0f9a5d43164efb6940d17e406f41691cb3d26435
2014-06-05 19:48:52 -04:00
Nicholas Nethercote 7e98d1c248 Bug 1014377 - Convert the first quarter of MFBT to Gecko style. r=froydnj.
--HG--
extra : rebase_source : b3b2da775e2c0e8a6ecbed70e7bd0c8f7af67b47
2014-05-29 22:40:33 -07:00
David Major 926fd8dee8 Bug 982310 - Hack around problems linking std::_Debug_message; r=ehsan
--HG--
extra : amend_source : 2e0f6f26823ee5fcce16dfaf8791ae57d8ac62cb
2014-03-21 12:53:00 -07:00
Viktor Stanchev 529b872af2 Bug 965837 - Add bool support to mozilla::Atomic. r=froydnj 2014-02-06 14:57:30 -05:00
Jeff Muizelaar 3ebdf849ba Bug 932454. Make Atomics.h compilable with clang on Windows. r=nfroyd,jwalden
MSVC is sloppy with typedefs leaking. Clang is more strict.

--HG--
extra : rebase_source : e8000f43391f8a32fbde96d54ff11e09e49345f9
2013-10-29 17:55:38 -04:00
Joshua Cranmer 7cc0542786 Bug 906783 - Attempt to detect the standard C++ library version, r=Waldo 2013-08-29 22:44:23 -05:00
Trevor Saunders 523a468b97 bug 905243 - make the constructors for mozilla::Atomic constexpr r=froydnj 2013-08-14 16:28:17 -04:00
Nathan Froyd 23a4a49409 Bug 900965 - compile failures with gcc 4.6.3 on std::atomic enum support from bug 888548 r=waldo 2013-08-02 18:18:41 -07:00
Birunthan Mohanathas 1baab48cb7 Bug 888548 - Part 3: Add enum support to mozilla::Atomic<T>. r=froydnj
Due to a bug in GCC, the compareExchange function is not available with enum types.
2013-08-01 21:21:32 -04:00
Birunthan Mohanathas e58014264e Bug 888548 - Part 2: Refactor and cleanup mozilla::Atomic<T> implementation. r=froydnj
This moves the increment and decrement operators from detail::AtomicBase to
detail::AtomicBaseIncDec and moves the implementation of the assignment
operator into detail::AtomicBase. Additionally, this changes the integral
implementation to use mozilla::EnableIf for its specialization.
2013-08-01 21:21:32 -04:00
Daniel Holbert f8d5639ac1 backout f607ac59de19 (Bug 888548 part 2) for B2G build bustage 2013-07-31 19:05:05 -07:00
Daniel Holbert 819cbd5af1 backout fc98067f0aa4 (Bug 888548 part 3) for B2G build bustage 2013-07-31 19:04:25 -07:00
Birunthan Mohanathas 6f4e72b203 Bug 888548 - Part 3: Add enum support to mozilla::Atomic<T>. r=froydnj
Due to a bug in GCC, the compareExchange function is not available with enum types.
2013-07-31 21:15:25 -04:00
Birunthan Mohanathas 19404c8ec4 Bug 888548 - Part 2: Refactor and cleanup mozilla::Atomic<T> implementation. r=froydnj
This moves the increment and decrement operators from detail::AtomicBase to
detail::AtomicBaseIncDec and moves the implementation of the assignment
operator into detail::AtomicBase. Additionally, this changes the integral
implementation to use mozilla::EnableIf for its specialization.
2013-07-31 21:15:25 -04:00
Ehsan Akhgari 2824b29025 Bug 895322 - Part 1: Replace the usages of MOZ_STATIC_ASSERT with C++11 static_assert; r=Waldo
This patch was mostly generated by running the following scripts on the codebase, with some
manual changes made afterwards:

# static_assert.sh
#!/bin/bash
# Command to convert an NSPR integer type to the equivalent standard integer type

function convert() {
echo "Converting $1 to $2..."
find . ! -wholename "*nsprpub*" \
       ! -wholename "*security/nss*" \
       ! -wholename "*/.hg*" \
       ! -wholename "obj-ff-dbg*" \
       ! -name nsXPCOMCID.h \
       ! -name prtypes.h \
         -type f \
      \( -iname "*.cpp" \
         -o -iname "*.h" \
         -o -iname "*.cc" \
         -o -iname "*.mm" \) | \
    xargs -n 1 `dirname $0`/assert_replacer.py #sed -i -e "s/\b$1\b/$2/g"
}

convert MOZ_STATIC_ASSERT static_assert
hg rev --no-backup mfbt/Assertions.h \
                   media/webrtc/signaling/src/sipcc/core/includes/ccapi.h \
                   modules/libmar/src/mar_private.h \
                   modules/libmar/src/mar.h


# assert_replacer.py
#!/usr/bin/python

import sys
import re

pattern = re.compile(r"\bMOZ_STATIC_ASSERT\b")

def replaceInPlace(fname):
  print fname
  f = open(fname, "rw+")
  lines = f.readlines()
  for i in range(0, len(lines)):
    while True:
      index = re.search(pattern, lines[i])
      if index != None:
        index = index.start()
        lines[i] = lines[i][0:index] + "static_assert" + lines[i][index+len("MOZ_STATIC_ASSERT"):]
        for j in range(i + 1, len(lines)):
          if lines[j].find("                 ", index) == index:
            lines[j] = lines[j][0:index] + lines[j][index+4:]
          else:
            break
      else:
        break
  f.seek(0, 0)
  f.truncate()
  f.write("".join(lines))
  f.close()

argc = len(sys.argv)
for i in range(1, argc):
  replaceInPlace(sys.argv[i])

--HG--
extra : rebase_source : 4b4a4047d82f2c205b9fad8d56dfc3f1afc0b045
2013-07-18 13:59:53 -04:00
Nathan Froyd 86df43c75a Bug 898491 - use the four argument form of compare_exchange_strong in Atomics.h; r=Waldo
The C++ standard (29.6p20-22 in N3337) specifies limitations on the failure ordering
for atomic compare-and-exchange.  Specifically, you can't pass memory_order_acq_rel or
memory_order_release.  For the (T&, T, std::memory_order) version, which we use, the
standard specifies that the provided argument should be "lowered" to comply with the
above restrictions on the failure ordering (29.6p21).

However, it seems that some versions of GCC's <atomic> header don't follow the spec
for the generic versions of std::atomic<>, though they do follow the spec with the
appropriate specializations (bool, integer, and pointer) of std::atomic<>.  This
results in mysterious failures when using atomic enums, as bug 888548 purports to
do, and ReleaseAcquire ordering.

Happily, we can work around this by using the more explicit version of
compare-and-exchange.  I've chosen to add another member to AtomicOrderConstraints,
even though it'd be the same as LoadOrder.  I feel explicitness is to be preferred
here.
2013-07-26 12:31:19 -04:00
Ms2ger 51f391870b Bug 896341 - Update include guards and modelines in MFBT; r=Waldo 2013-07-24 09:41:39 +02:00
Timothy B. Terriberry a5f847a092 Bug 881163 - Require gcc 4.5.2 for C++11 atomics, r=froydnj 2013-06-10 09:17:00 -07:00
Landry Breuil fae08ebf77 Bug 876156: Don't try using <atomic> on OpenBSD which uses an old libstdc++ r=nfroyd 2013-06-01 00:23:01 +02:00
Nathan Froyd 443c507139 Bug 873649 - don't assume <atomic> is usable with clang on a Linux system; r=Waldo 2013-05-17 17:04:30 -04:00
Justin Lebar 7729e7ff63 Bug 876691 - Unbreak mfbt/Atomics.h for Windows x64. r=me 2013-05-28 15:26:25 -04:00
Justin Lebar c002d5a3a3 Bug 873800 - Add compare-and-exchange to mfbt/Atomics.h. r=froydnj
--HG--
extra : rebase_source : c9e455ed249e6b3cd0f532d240abfb602e2e88a8
2013-05-24 13:10:47 -04:00
Justin Lebar d0980f2231 Bug 873799 - Add an explicit barrier before __sync_lock_test_and_set in mfbt/Atomics.h r=froydnj
--HG--
extra : rebase_source : 64d23f1df2fef8719413800ebb0afe6f51d1e3af
2013-05-24 13:10:47 -04:00
Nathan Froyd 58d13e374c Bug 732043 - part 1 - add mfbt/Atomics.h; r=Waldo; original-author=jcranmer 2013-04-22 14:12:03 -04:00