Alexander Köplinger [Mon, 14 Nov 2016 12:43:34 +0000 (13:43 +0100)]
Merge pull request #3940 from akoeplinger/deprecate-nunit
[bcl] Deprecate old NUnit 2.4 framework shipped with Mono
Alexander Köplinger [Sun, 13 Nov 2016 23:00:05 +0000 (00:00 +0100)]
Merge pull request #3937 from akoeplinger/mph
[msvc] Add Mono_Unix_VersionString to monoposixhelper.def
Marek Habersack [Sun, 13 Nov 2016 21:40:08 +0000 (22:40 +0100)]
Revert "Install utils/{mono-dl.h,mono-compiler.h}"
This reverts commit
397430c7f9fe4fae487342db36860b0a08cd56c0.
Alexander Köplinger [Fri, 11 Nov 2016 13:38:35 +0000 (14:38 +0100)]
[msvc] Add Mono_Unix_VersionString to monoposixhelper.def
Fixes the EntryPointNotFoundException when running the Mono.Posix tests on Windows.
This check was introduced in
249ee0686bf87f33acf2efde76f3645ef06bb886 but the .def file wasn't updated.
With this the tests still crash so switching them off on PRs again for now.
Alexander Köplinger [Fri, 11 Nov 2016 01:55:05 +0000 (02:55 +0100)]
[bcl] Deprecate old NUnit 2.4 framework shipped with Mono
Print a message when running nunit-console about the deprecation.
Additionally, throw a compilation error if compiling against the
TestFixtureAttribute from old NUnit.
Needed changes in ms-test-suite as well since it was still using NUnit.
Marek Habersack [Sat, 12 Nov 2016 23:48:44 +0000 (00:48 +0100)]
Install utils/{mono-dl.h,mono-compiler.h}
These two headers are required by loader.h which is installed in
$prefix/include/mono-2.0/mono/metadata and their lack breaks compilation
of any software that includes loader.h
Niklas Therning [Sat, 12 Nov 2016 07:41:55 +0000 (08:41 +0100)]
Merge pull request #3938 from ntherning/UnicastIPAddressInformation.IPv4Mask-from-refsource
Change UnicastIPAddressInformation.IPv4Mask to work as in refsource on Windows
Zoltan Varga [Fri, 11 Nov 2016 20:41:26 +0000 (15:41 -0500)]
[runtime] Add back the locking to setup_interface_offsets () to work around crashes in roslyn.
Alexander Köplinger [Fri, 11 Nov 2016 18:55:36 +0000 (19:55 +0100)]
Merge pull request #3942 from monojenkins/update-csprojs
[msvc] Update csproj files
Alexander Köplinger [Fri, 11 Nov 2016 18:55:11 +0000 (19:55 +0100)]
Fix git submodule url
monojenkins [Fri, 11 Nov 2016 18:46:04 +0000 (18:46 +0000)]
[msvc] Update csproj files
Marek Safar [Fri, 11 Nov 2016 17:39:40 +0000 (18:39 +0100)]
Remove extracted linker sources and reuse them via submodule
Ludovic Henry [Fri, 11 Nov 2016 15:13:38 +0000 (10:13 -0500)]
[w32handle] Stop returning 0 in every cases for locking/unlocking (#3926)
* [w32handle] Stop returning 0 in every cases for locking/unlocking
* [w32handle] Ensure we lock the signal handle in SignalAndWait
Marek Safar [Fri, 11 Nov 2016 14:52:44 +0000 (15:52 +0100)]
[mcs] Update better conversion tie breaking rules to cover candidate with longer default parameters. Fixes #45950
Niklas Therning [Fri, 11 Nov 2016 13:47:57 +0000 (14:47 +0100)]
Change UnicastIPAddressInformation.IPv4Mask to work as in refsource on Windows
Mono returns null for disconnected interfaces with 169.254.X.X address while
refsources returns 255.255.0.0. This patch copies the relevant pieces from the
refsources to make Mono behave the same.
Joao Matos [Fri, 11 Nov 2016 12:57:56 +0000 (12:57 +0000)]
[offsets-tool] Fixed the Makefile to clone into the base directory instead.
Johan Lorensson [Fri, 11 Nov 2016 10:43:49 +0000 (11:43 +0100)]
Merge pull request #3809 from lateralusX/jlorenss/win-api-family-support-cleanup
Build mono runtime under none desktop Windows API family, adjustments and cleanup.
Niklas Therning [Fri, 11 Nov 2016 08:59:05 +0000 (09:59 +0100)]
Merge pull request #3930 from ntherning/fix-System.ServiceProcess
Fixes bugs in System.ServiceProcess on Windows
Niklas Therning [Thu, 10 Nov 2016 15:08:40 +0000 (16:08 +0100)]
Fixes bugs in System.ServiceProcess on Windows
There was a crash on Windows x64 in Win32ServiceController when enumerating
devices and services. Also fixed the ServiceControllerTest to succeed on .NET
as well Mono on Windows. Had to replace the services used by the test since
most of them have been removed from recent Windows versions.
This PR also enables the System.ServiceProcess tests in Windows PR builds.
Miguel de Icaza [Fri, 11 Nov 2016 04:14:03 +0000 (23:14 -0500)]
[eglib/mono] Aleksey caught a discrepancy between the semantics of Unix and Windows, our eglib version should always strdup (#3899)
Ankit Jain [Thu, 10 Nov 2016 18:26:27 +0000 (13:26 -0500)]
[xbuild] Don't reevaluate project when setting metadata in a dynamic ..
.. item.
If we have a project, that sets some properties or items at runtime,
like in a target:
```
<Target Name="Build">
<PropertyGroup>
<Bar>Bar01</Bar>
</PropertyGroup>
```
.. such a property can be accessed as `$(Bar)` subsequently.
Then at later point, a target tries to update metadata on an item group
with existing items, like:
```
<ItemGroup>
<FooItem Include="xyz" />
</ItemGroup>
<Target Name="Foo">
<ItemGroup>
<FooItem>
<SomeMetadata>MetadataValue</SomeMetadata>
</FooItem>
</ItemGroup>
```
.. then it is seen that the value for the earlier created `$(Bar)` disappears!
So, if we try to print the value of `$(Bar)` right after that item
group, it would appear as `''`!
The issue is that the metadata update in target `Foo` caused the project
to get re-evaluated, which meant that items/properites created after the
project load were lost! We should not be reevaluating the project when
setting metadata on the basis of dynamic items, like in the target
`Foo`.
These dynamic item groups in a target are represented as
`BuildItemTask`, and these create BuildItems which have their
`IsDynamic` property set. And this is used to avoid re-evaluating the
project in `BuildItem.SetMetadata`.
But when updating metadata (`BuildItem.UpdateMetadata`), we need to
update metadata on the *existing* items, which might not have been
created from such dynamic item groups! So, their `IsDynamic==false`.
Hence, trying to `SetMetadata` on such items would cause the project to
be reevaluated, and thus properties/items like `$(Bar)` would be lost!
Fixes: https://bugzilla.xamarin.com/show_bug.cgi?id=45137
In this particular case, `Bar` was the `DesignTimeBuild` property, which
was losing it's value when `ProjectReference`'s `AdditionalProperties`
metadata was set in:
```
<Target Name="BclBuildAddProjectReferenceProperties"
<ItemGroup>
<ProjectReference>
<AdditionalProperties>$(_BclBuildProjectReferenceProperties);%(ProjectReference.AdditionalProperties)</AdditionalProperties>
</ProjectReference>
</ItemGroup>
```
Bernhard Urban [Thu, 10 Nov 2016 16:34:12 +0000 (17:34 +0100)]
[arm64] always assume a cache line size of 4
otherwise we still have the theoretical chance of a race.
Niklas Therning [Thu, 10 Nov 2016 20:28:23 +0000 (21:28 +0100)]
Merge pull request #3931 from ntherning/enable-System.Core-tests-on-windows-pr-builds
Enable System.Core tests on Windows PR builds
João Matos [Thu, 10 Nov 2016 18:55:51 +0000 (18:55 +0000)]
Merge pull request #3927 from tritao/updated_offsets_tool
Updated offsets tool to latest CppSharp.
Marek Safar [Thu, 10 Nov 2016 17:41:15 +0000 (18:41 +0100)]
[System.Xml] Add XElement array handling to serialization interpreter. Fixes #12571
Marek Safar [Thu, 10 Nov 2016 17:30:19 +0000 (18:30 +0100)]
[system.xml] Enable more tests
Joao Matos [Thu, 10 Nov 2016 16:40:38 +0000 (16:40 +0000)]
[offsets-tool] Updated build scripts to handle multiple platforms.
Joao Matos [Thu, 10 Nov 2016 16:39:18 +0000 (16:39 +0000)]
[offsets-tool] Fixed whitespace inconsistency.
Joao Matos [Thu, 10 Nov 2016 12:09:21 +0000 (12:09 +0000)]
Updated offsets tool to latest CppSharp.
Niklas Therning [Thu, 10 Nov 2016 15:15:21 +0000 (16:15 +0100)]
Enable System.Core tests on Windows PR builds
System.Core is expected to be green now on Windows.
Bernhard Urban [Wed, 9 Nov 2016 15:03:52 +0000 (16:03 +0100)]
[ppc] clobber r0 in MONO_CONTEXT_GET_CURRENT
r0 can't be the base for a memory acces.
Bernhard Urban [Wed, 9 Nov 2016 14:40:41 +0000 (15:40 +0100)]
[ppc] some 32bit CPUs trap on std, use stw instead
Bernhard Urban [Wed, 9 Nov 2016 14:39:16 +0000 (15:39 +0100)]
[ppc] enable MONO_ARCH_HAVE_SIGCTX_TO_MONOCTX on every PowerPC configuration on Linux
Bernhard Urban [Wed, 9 Nov 2016 11:38:01 +0000 (12:38 +0100)]
[ppc] use proper calling convention when coming from a signal handler
ELF v2 ABI (aka. ppc64le) defines two entry points per function.
The global entry point (offset 0) sets up the TOC (r2) itself, while the
local entry point (offset 8) is a fast entry point if the TOC is
already set up properly.
When the global entry point is used, the address of the global entry
point is required in r12 according to the ABI.
Bernhard Urban [Wed, 9 Nov 2016 11:36:23 +0000 (12:36 +0100)]
[ppc] disable MONO_ARCH_HAVE_TLS_GET
`--with-tls=__thread` is broken on PowerPC.
`--with-tls=pthread` didn't work on PowerPC, but disabling
MONO_ARCH_HAVE_TLS_GET fixes it.
Instead of guarding the define with `HAVE_KW_THREAD`, let's just
disable it. We want to get rid of `--with-tls=__thread` anyway.
Johan Lorensson [Thu, 10 Nov 2016 08:37:55 +0000 (09:37 +0100)]
Merge pull request #3854 from lateralusX/jlorenss/win-clang-toolchain
Switch to Microsoft VS2015 MSVC toolchain for Windows x64 AOT compiler.
Zoltan Varga [Thu, 10 Nov 2016 03:54:29 +0000 (22:54 -0500)]
Method field count (#3907)
* [runtime] Add accessors for method.count and field.count.
* [runtime] Move method/field count to MonoClassDef.
Zoltan Varga [Thu, 10 Nov 2016 02:53:48 +0000 (21:53 -0500)]
[sdb] Avoid asserting when trying to single step on a thread with no managed frames. Fixes #34802.
Zoltan Varga [Wed, 9 Nov 2016 23:13:33 +0000 (18:13 -0500)]
[runtime] Fix another race in mono_class_setup_supertypes ().
Ludovic Henry [Wed, 9 Nov 2016 20:59:36 +0000 (15:59 -0500)]
[threads] Always use the `suspended` MonoOSEvent for self-suspend (#3915)
* [stw] Refactor to treat error/success cases first
* [threads] Always use the `suspended` MonoOSEvent for self-suspend
The previous self-suspend mechanism was inherently racy:
- T1 self suspend:
- T1 calls mono_thread_info_begin_self_suspend, but doesn't call mono_thread_info_end_self_suspend yet
- T2 suspends T1 <-- the suspend_count is > 0, trigerring the assertion at mono-threads.c:184
Ludovic Henry [Wed, 9 Nov 2016 20:59:24 +0000 (15:59 -0500)]
[threads] Destroy suspend MonoOSEvent when detaching (#3923)
* [threads] Destroy suspend MonoOSEvent when detaching
It could happen that we destroy the managed InternalThread object, before we actually detach the thread (in mono/tests/threadpool-exceptions1.exe for example).
Fixes https://bugzilla.xamarin.com/show_bug.cgi?id=46536
* [threads] Close thread handle once done with it
Zoltan Varga [Wed, 9 Nov 2016 19:51:11 +0000 (14:51 -0500)]
[runtime] Simplify the wait_for_tids () code. (#3920)
Ludovic Henry [Wed, 9 Nov 2016 19:44:29 +0000 (14:44 -0500)]
[os-event] Make them manual only (#3921)
We only use them as manual reset event in the runtime, and that greatly simplifies the code as well as the way they should be used.
Zoltan Varga [Wed, 9 Nov 2016 19:42:21 +0000 (14:42 -0500)]
[sdb] Avoid asserting when trying to single step on a thread with no managed frames. Fixes #34802.
Ludovic Henry [Mon, 31 Oct 2016 20:31:29 +0000 (16:31 -0400)]
[io-layer] Remove dead code
Alexander Köplinger [Wed, 9 Nov 2016 19:20:27 +0000 (20:20 +0100)]
[System] Fix a few type members on WatchOS
The .platformnotsupported.cs versions of some classes were
inconsistent with their "real" counterparts.
Ludovic Henry [Wed, 9 Nov 2016 13:00:11 +0000 (08:00 -0500)]
[process] Allocate a handle even for non-child processes (#3905)
* [w32handle] Inline only use of mono_w32handle_search
* [w32handle] Make special_wait return MonoW32HandleWaitRet
This is to avoid a unnecessary translation from WAIT_* to MonoW32HandleWaitRet
* [process] Use some more referencesource
* [process] Rename mono_processes to processes
* [process] Allocate a handle even for non-child processes
* [process] Create the handle only once the process has started
Ludovic Henry [Wed, 9 Nov 2016 12:50:12 +0000 (07:50 -0500)]
[os-event] Make unix version alertable (#3909)
This fixes `appdomain_unload.exe` test timeout. The following would happen:
- T1 unload the current domain
- T1 launch T2 to unload domain
- T1 waits on T2 to finish
- T2 tries to abort T1 <-- because the wait on T2 is not alertable, T1 is not going to be aborted
Zoltan Varga [Wed, 9 Nov 2016 07:00:09 +0000 (02:00 -0500)]
[runtime] Add locking to mono_class_setup_supertypes () since the 16 bit idepth field cannot be set atomically. Fix a race if mono_field_resolve_type () was called without locking. (#3919)
Zoltan Varga [Tue, 8 Nov 2016 23:55:51 +0000 (18:55 -0500)]
[x86] Fix max code size of throw trampoline.
Alexander Köplinger [Tue, 8 Nov 2016 20:24:39 +0000 (21:24 +0100)]
Bump nunitlite for display fixes
Alexis Christoforides [Tue, 8 Nov 2016 20:19:25 +0000 (15:19 -0500)]
Bump bockbuild
Alexander Köplinger [Tue, 8 Nov 2016 18:59:37 +0000 (19:59 +0100)]
[io-layer] Include sys/types.h header file in wapi.h
Fixes "error: unknown type name 'pid_t'" on Alpine/musl libc.
Alexander Köplinger [Tue, 8 Nov 2016 18:30:35 +0000 (19:30 +0100)]
[btls] Include stdint.h header file in btls-time64.c
Fixes "error: unknown type name 'int64_t'" on Alpine/musl libc.
Niklas Therning [Tue, 8 Nov 2016 17:52:55 +0000 (18:52 +0100)]
Merge pull request #3914 from ntherning/mono_context_get_current-on-windows-does-not-save-rsp-properly
mono_context_get_current() on Windows stores registers at incorrect indexes
Alexis Christoforides [Tue, 8 Nov 2016 16:10:26 +0000 (11:10 -0500)]
Merge pull request #3773 from mono/bockbuild-integration
Integration of Bockbuild's Mac SDK profile
Alexis Christoforides [Tue, 8 Nov 2016 16:10:05 +0000 (11:10 -0500)]
Bump bockbuild
Alexis Christoforides [Tue, 8 Nov 2016 16:09:38 +0000 (11:09 -0500)]
Move some more files back to bockbuild repo
Niklas Therning [Tue, 8 Nov 2016 14:21:36 +0000 (15:21 +0100)]
mono_context_get_current() on Windows stores registers at incorrect indexes
The indexes must correspond to the values in the AMD64_Reg_No enum in
amd64-codegen.h.
Alexander Köplinger [Tue, 8 Nov 2016 12:57:16 +0000 (13:57 +0100)]
Merge pull request #3913 from omwok/master
column name and ordinal fix.
lateralusX [Mon, 31 Oct 2016 08:45:30 +0000 (09:45 +0100)]
Switch to Microsoft VS2015 MSVC toolchain for Windows x64 AOT compiler.
In order to reduce dependencies on external toolchains when doing full AOT on Windows
targeting x64 Windows platforms we would like to use what's available in
VS 2015 MSVC toolchain. Mono's full AOT compiler need an assembler and a linker from
available toolchain in order to do different full AOT scenarios, emitting assembler, compile
objects or link complete shared library.
Since the current emitter emits GNU assembler it is beneficial to reuse that assembler
syntax without the need to adopt to a different syntax. In VS2015 Update 1 a new toolchain
option has been released utilizing the Clang parser combined with MSVC code gen. This will
produce object files fully compatible with MSVC linker. It turns out that VS2015 Clang version
also supports GNU assembler and since Mono just uses simple constructs in the assembler
(all code is already compiled and emitted as byte streams in assembler) it is possible to use Clang
for VS to compile our emitted assembler files.
This PR switches to Clang for Microsoft VS2015 and MSVC linker when doing full AOT targeting Windows x64
platforms hosted on a Windows platform. It requires an installation of at least Microsoft VS2015 Update 1
with Clang component selected. NOTE all other scenarios is currently left as is.
This PR also makes adjustments to AOT compiler in order for above scenario to work, supporting asm only output, static output
and regular shared libraries using MSVC linker.
Alexander Köplinger [Tue, 8 Nov 2016 11:46:44 +0000 (12:46 +0100)]
Merge pull request #3911 from akoeplinger/marshal-cleanuprcw
[corlib] Add Marshal.CleanupUnusedObjectsInCurrentContext() stub
lateralusX [Mon, 7 Nov 2016 15:58:07 +0000 (16:58 +0100)]
Additional none desktop API families triggered by changes.
* Changes to process.c triggered errors on none desktop API builds.
* Implementation of memory mapped files caused build errors on none desktop API builds.
* Adjustments to disable JIT caused build errors on none JIT builds.
lateralusX [Fri, 28 Oct 2016 13:24:53 +0000 (15:24 +0200)]
Build mono runtime under none desktop Windows API family, adjustments and cleanup.
Continuation work on previous none desktop Windows API family PR's adjusting all
changes to similar patterns used over previous PR's. This PR also move the logic
to silence a MS VS linker warning on empty source files into a macro. It also includes
the use of the macro in a couple of more source files causing linker warnings when build
on none desktop API families excluding JIT support.
Since other Windows API family work is still in progress it also disables build capabilities
for none desktop API families.
Marek Safar [Tue, 8 Nov 2016 07:55:14 +0000 (08:55 +0100)]
Merge pull request #3912 from akoeplinger/marshal-ptrtostrbstr
[corlib] Expose Marshal.PtrToStringBSTR on mobile
Marek Safar [Tue, 8 Nov 2016 07:50:50 +0000 (08:50 +0100)]
Merge pull request #3902 from henricm/fix-rename-lpstr-utf8
Rename fix for lpstr to utf8str on Windows
Alexander Köplinger [Tue, 8 Nov 2016 00:31:30 +0000 (01:31 +0100)]
[corlib] Expose Marshal.PtrToStringBSTR on mobile
The SecureStringHelper from referencesource I added in https://github.com/mono/mono/pull/3887
relies on this to convert from SecureString to plaintext.
It failed on mobile because we always throw a NotImplementedException there,
but we actually have an implementation on the runtime side (even in the DISABLE_COM case).
Alexander Köplinger [Mon, 7 Nov 2016 23:04:47 +0000 (00:04 +0100)]
[corlib] Add Marshal.CleanupUnusedObjectsInCurrentContext() stub
https://github.com/dotnet/standard/issues/94
Zoltan Varga [Mon, 7 Nov 2016 22:37:50 +0000 (17:37 -0500)]
[runtime] Fix the ios build.
Zoltan Varga [Mon, 7 Nov 2016 20:48:05 +0000 (15:48 -0500)]
[jit] Really fix the DISABLE_JIT build.
Aleksey Kliger (λgeek) [Mon, 7 Nov 2016 20:18:08 +0000 (15:18 -0500)]
Merge pull request #3896 from lambdageek/bug-46250
[runtime] Ensure generic type definition arity matches generic context arity
Ludovic Henry [Mon, 7 Nov 2016 20:11:22 +0000 (15:11 -0500)]
[threadpool] Fix hang on domain unloading (#3904)
Because we wouldn't return the correct ThreadPoolDomain for the requested domain, we would simply wait on the last inserted MonoDomain. This would lead to a hang when unloading the domain, as we wouldn't wait on the correct domain.
Alexander Köplinger [Mon, 7 Nov 2016 19:36:24 +0000 (20:36 +0100)]
Fix tarball build
Martin Baulig [Mon, 7 Nov 2016 17:23:42 +0000 (18:23 +0100)]
[BTLS]: Use `int64_t` instead of `long` in the native API. (#3857)
* [BTLS]: Use `int64_t` instead of `long` in the native API.
* Copy time64.c from Android's source code to get a 64-bit timegm64().
* Rename into btls_timegm64() and remove everything we don't need.
(cherry picked from commit
2365067bdae39f75979df30b65ae08f095cbd176)
Alexander Köplinger [Mon, 7 Nov 2016 17:24:14 +0000 (18:24 +0100)]
Merge pull request #3894 from akoeplinger/fix-actor
[System.ServiceModel] Recognize 'faultactor' in SOAP fault message
Zoltan Varga [Mon, 7 Nov 2016 16:37:59 +0000 (11:37 -0500)]
[jit] Fix the DISABLE_JIT build.
Rodrigo Kumpera [Mon, 7 Nov 2016 16:04:29 +0000 (08:04 -0800)]
Merge pull request #2408 from tastywheattasteslikechicken/MoreInterfaceSupport
Enhance maximum number of supported interfaces from 2^16.
Niklas Therning [Mon, 7 Nov 2016 14:04:54 +0000 (15:04 +0100)]
Merge pull request #3903 from ntherning/fix-NRE-in-Win32-IpcChannel.StopListening-when-no-server-started
Fix NRE in Win32 IpcChannel.StopListening() when no server started
Niklas Therning [Mon, 7 Nov 2016 14:01:39 +0000 (15:01 +0100)]
Merge pull request #3901 from ntherning/Process_GetModules_internal-crash-on-windows
Fix crash in Process.GetModules_internal() on Windows after refactorings
Niklas Therning [Mon, 7 Nov 2016 14:00:05 +0000 (15:00 +0100)]
Fix NRE in Win32 IpcChannel.StopListening() when no server started
This makes the System.Runtime.Remoting test suite fail in various places on
Windows due to the TestFixtureTearDown method in BaseCalls throwing an NRE.
Henric Müller [Mon, 7 Nov 2016 13:46:21 +0000 (14:46 +0100)]
Rename fix for lpstr to utf8str on Windows
PR https://github.com/mono/mono/pull/3776 renamed `mono_string_to_lpstr`
to `mono_string_to_utf8str` without updating the Windows version.
This fix renames `mono_string_to_lpstr` to `mono_string_to_utf8str` to
make sure the string is allocated using `CoTaskMemAlloc` on Windows
since it will be freed using `CoTaskMemFree`.
Niklas Therning [Mon, 7 Nov 2016 11:01:43 +0000 (12:01 +0100)]
Fix crash in Process.GetModules_internal() on Windows after refactorings
EnumProcessModules() is called incorrectly. Also fixes a missing curly braces
pair.
Alexander Köplinger [Mon, 7 Nov 2016 01:29:55 +0000 (02:29 +0100)]
Merge pull request #3900 from monojenkins/update-csprojs
[msvc] Update csproj files
Zoltan Varga [Sun, 6 Nov 2016 17:34:45 +0000 (12:34 -0500)]
[runtime] Move field.first and method.first to MonoClassDef. Add accessor functions for them. (#3898)
monojenkins [Sun, 6 Nov 2016 04:20:14 +0000 (04:20 +0000)]
[msvc] Update csproj files
Zoltan Varga [Sat, 5 Nov 2016 02:19:25 +0000 (22:19 -0400)]
[arm64] Set a flag in MonoContext if the fp regs are saved, and only restore them if they are. These registers are not saved when the context is created from a ucontext. (#3890)
Aleksey Kliger (λgeek) [Sat, 5 Nov 2016 01:02:23 +0000 (21:02 -0400)]
[mkbundle] Don't free bundled dylibrary directory (#3897)
mkdtemp modifies the template in-place, don't free it.
Rodrigo Kumpera [Fri, 4 Nov 2016 21:59:47 +0000 (14:59 -0700)]
Merge pull request #3796 from ntherning/windows-backend-for-MemoryMappedFile
MemoryMappedFile support on Windows
Marek Safar [Fri, 4 Nov 2016 17:34:24 +0000 (18:34 +0100)]
[mcs] Consider method candidates with misplaced named arguments not applicable. Fixes #46190
Marek Safar [Fri, 4 Nov 2016 17:33:44 +0000 (18:33 +0100)]
[mcs] Remove debug check
Alexis Christoforides [Fri, 4 Nov 2016 17:34:04 +0000 (13:34 -0400)]
Properly configure the bockbuild submodule
Zoltan Varga [Fri, 4 Nov 2016 16:34:13 +0000 (12:34 -0400)]
[runtime] Add a few mono mono_class_is_... accessors. (#3891)
Alexis Christoforides [Fri, 4 Nov 2016 15:31:07 +0000 (11:31 -0400)]
Fix setting the min. macOS version in the Xamarin profile
Alexis Christoforides [Fri, 4 Nov 2016 14:55:13 +0000 (10:55 -0400)]
Rename 'bockbuild' to 'packaging' so that it can be home to other packaging code
Alexis Christoforides [Fri, 4 Nov 2016 14:46:01 +0000 (10:46 -0400)]
Move the maintained common-use packages back to the bockbuild repo
Alexander Köplinger [Fri, 4 Nov 2016 14:59:18 +0000 (15:59 +0100)]
[System.ServiceModel] Recognize 'faultactor' in SOAP fault message
It's an optional element according to [1] and we were simply
throwing a NotImplementedException when encountering it.
Fixed by reading the element value instead. I noticed that we
had a 'details' variable that was never assigned, so the check
for `details == null` was always true and can be removed.
This makes inlining the body of the CreateFault() call easier
since we no longer need to choose different overloads and makes
the code consistent with the CreateFault12 method.
Also added a better exception in case of an unrecognized element.
The CreateFault12 method (handles SOAP 1.2) does not need a similar
fix since actor is not an element in that version [2].
Fixes https://bugzilla.xamarin.com/show_bug.cgi?id=42226
[1] https://www.w3.org/TR/soap11/#_Toc478383507
[2] https://www.w3.org/TR/soap12/#soapfault
Aleksey Kliger [Thu, 3 Nov 2016 21:07:19 +0000 (17:07 -0400)]
[runtime] Ensure generic type definition arity matches generic context arity
Don't instantiate generic type definitions (like Dictionary`2) with too few or too many
type arguments.
Fixes https://bugzilla.xamarin.com/show_bug.cgi?id=46250
Aleksey Kliger [Thu, 3 Nov 2016 21:05:53 +0000 (17:05 -0400)]
[tests] Test System.Type.GetType() on generics with wrong arity.
Marek Safar [Fri, 4 Nov 2016 15:03:52 +0000 (16:03 +0100)]
[mcs] More work on debug scopes declared out of order. Fixes #45774
Ludovic Henry [Fri, 4 Nov 2016 14:43:44 +0000 (10:43 -0400)]
[os-event] Fix set operation
This is a logic bug where we would only wake up one of the threads that are waiting for the manual MonoOSEvent to be set. This would be a problem only if more than one thread is waiting.