Jon Purdy [Wed, 8 Mar 2017 23:44:13 +0000 (15:44 -0800)]
[exdoc] Enable formatting in deprecated annotations.
Jon Purdy [Wed, 8 Mar 2017 02:42:39 +0000 (18:42 -0800)]
[exdoc] Handle punctuation better in code formatting.
Jon Purdy [Tue, 28 Feb 2017 00:58:33 +0000 (16:58 -0800)]
[docs] Use Doxygen syntax for Monodoc.
Jon Purdy [Wed, 1 Mar 2017 02:34:06 +0000 (18:34 -0800)]
[exdoc] Update warnings.
Jon Purdy [Tue, 28 Feb 2017 00:58:09 +0000 (16:58 -0800)]
[exdoc] Support Doxygen syntax.
Jon Purdy [Thu, 23 Feb 2017 02:29:29 +0000 (18:29 -0800)]
[docs] Fix exdoc warnings.
Jon Purdy [Thu, 23 Feb 2017 02:27:54 +0000 (18:27 -0800)]
[exdoc] Refactor.
Avoid globals, use proper data structures instead of strings, and
produce some warnings for documentation errors.
Jon Purdy [Tue, 21 Feb 2017 23:41:15 +0000 (15:41 -0800)]
[exdoc] Add missing dependency.
Jon Purdy [Tue, 21 Feb 2017 23:40:59 +0000 (15:40 -0800)]
[exdoc] Enable 'warnings' and 'strict'.
Jon Purdy [Tue, 21 Feb 2017 23:15:20 +0000 (15:15 -0800)]
[exdoc] Fix whitespace.
Jon Purdy [Mon, 13 Feb 2017 23:42:55 +0000 (15:42 -0800)]
[docs] Remove redundant filename.
Jon Purdy [Mon, 13 Feb 2017 23:33:41 +0000 (15:33 -0800)]
[docs] Fix Doxygen errors.
Doxygen treats '@code' as the start of a '@code'...'@endcode' block, so
it raised spurious "reached end of file while inside code block" errors
when '@code' was used to denote a parameter. Change these instances to
'\param code' (declaration) and '\p code' (mention).
Jon Purdy [Mon, 13 Feb 2017 23:10:40 +0000 (15:10 -0800)]
[docs] Enable documentation for utils.
Jon Purdy [Sat, 11 Feb 2017 01:17:57 +0000 (17:17 -0800)]
[docs] Enable documentation for mini.
Jon Purdy [Fri, 10 Feb 2017 23:32:44 +0000 (15:32 -0800)]
[docs] Enable documentation for sgen.
Jon Purdy [Fri, 10 Feb 2017 01:43:01 +0000 (17:43 -0800)]
[docs] Enable documentation for metadata.
Jon Purdy [Fri, 10 Feb 2017 01:43:07 +0000 (17:43 -0800)]
[docs] Improve Doxygen config.
Marek Safar [Fri, 24 Mar 2017 19:04:55 +0000 (20:04 +0100)]
Revert "Update corefx"
This reverts commit
2a2c317d07e389088737b391e97853fdb6dbc390.
Marek Safar [Fri, 24 Mar 2017 19:04:03 +0000 (20:04 +0100)]
[tests] Remove duplicate property
Andi McClure [Fri, 24 Mar 2017 18:42:43 +0000 (14:42 -0400)]
Merge pull request #4590 from xmcclure/master
Improve "Could not resolve type with token" exception message
Andi McClure [Fri, 24 Mar 2017 18:37:26 +0000 (14:37 -0400)]
Improve "Could not resolve type with token" exception message
Marek Safar [Fri, 24 Mar 2017 18:36:23 +0000 (19:36 +0100)]
Update corefx
Marek Safar [Fri, 24 Mar 2017 14:36:31 +0000 (15:36 +0100)]
[mcs] Reset also all partial parts current-type
Vlad Brezae [Fri, 24 Mar 2017 14:21:30 +0000 (16:21 +0200)]
Merge pull request #4503 from BrzVlad/fix-appdomain-unload
[runtime] Fix abort issues
Miguel de Icaza [Fri, 24 Mar 2017 14:01:01 +0000 (10:01 -0400)]
A few fixes to the man page
Zoltan Varga [Thu, 23 Mar 2017 17:34:57 +0000 (13:34 -0400)]
[bcl] Add a finalizer to MemoryMappedFile.
Marek Safar [Fri, 24 Mar 2017 13:15:14 +0000 (14:15 +0100)]
[System.Data] Update tests for mobile profiles
Vlad Brezae [Thu, 23 Mar 2017 22:17:30 +0000 (00:17 +0200)]
[runtime] Don't reset abort exception in invoke wrapper
It leads to missing self aborts on full-aot.
Vlad Brezae [Mon, 20 Mar 2017 23:34:49 +0000 (01:34 +0200)]
[runtime] Improve debugging of thread interruptions
Vlad Brezae [Mon, 20 Mar 2017 23:00:17 +0000 (01:00 +0200)]
[runtime] Defer async requests during abort protected blocks
We avoid unnecessary checks for each interruption checkpoint.
Vlad Brezae [Fri, 17 Mar 2017 21:45:50 +0000 (23:45 +0200)]
[runtime] Fix detecting abort at end of abort protected block
If we have a pending abort then we know we will have the thread_state set and the thread_interruption_requested bumped. We used to consume this request and set the MonoError with the exception. The problems with this is that not all callsites handle the passed MonoError and handling it means converting it back to a pending exception anyway. We just leave the interruption request untouched now.
If the cctor self aborts it means that we consumed the abort exception so we do need to call mono_thread_resume_interruption to reactivate it. We need to make sure that we still don't throw a TypeInitializationException if we had a self abort.
Martin Baulig [Thu, 23 Mar 2017 20:29:47 +0000 (16:29 -0400)]
[AppleTls]: Use correct definition of `SecTrustResult`. (#4586)
* [AppleTls]: Use correct definition of `SecTrustResult`.
We need all the values in this enum because `SecTrustResult.Unspecified`
means success - which currently has an integer value of 4, not 0.
* Oops, make it build.
* Don't make it public.
* Revert whitespace change.
Andi McClure [Thu, 23 Mar 2017 20:29:06 +0000 (16:29 -0400)]
Bump version to 5.1.0
Filip Navara [Thu, 23 Mar 2017 11:36:58 +0000 (12:36 +0100)]
Add test for AutomaticDecompression behaviour
Test that HttpWebRequest.AutomaticDecompression actually performs the
decompression and that it removes the Content-Encoding header. Also fix
the value of Content-Encoding for other tests to follow the HTTP
specification.
Filip Navara [Thu, 23 Mar 2017 09:54:18 +0000 (10:54 +0100)]
Remove Content-Encoding header after doing automatic decompression
Match the .NET Framework behaviour by removing the Content-Encoding
header after doing automatic decompression.
monojenkins [Thu, 23 Mar 2017 16:21:08 +0000 (16:21 +0000)]
[msvc] Update csproj files
Aleksey Kliger [Wed, 22 Mar 2017 19:08:59 +0000 (15:08 -0400)]
[reflection] Set correct reflected type in MonoMethod.get_base_method
It can be the case that we have two different System.MonoMethod objects that both
point to the same native MonoMethod* but have two different reflected types.
(For example if we call derivedType.GetMethod("f") where f is virtual and
derivedType corresponds to a class that did not override a base definition. In
that case, the MonoMethod* corresponds to the parent type's f method, but the
reflected type is derivedType)
In this case, GetBaseDefinition will return the same MonoMethod* but the
reflected type of the result should be the base type, not the derived type.
Aleksey Kliger [Wed, 22 Mar 2017 20:19:25 +0000 (16:19 -0400)]
[tests] Additional tests of System.Reflection.MethodInfo.GetBaseDefinition
Ludovic Henry [Thu, 23 Mar 2017 14:09:33 +0000 (10:09 -0400)]
[threadpool-io] Ensure selector thread is running before waiting on it (#4572)
This could result in a unecessary wait on shutdown where the selector thread would have exited, and the finalizer thread would try to dispose a Socket, which would try to remove the socket from the IOSelector. This removal operation would wait on the selector thread to acknowledge the removal, but because the selector thread would have alreday exited, we would wait on an event that would never happen.
Andoni Morales Alastruey [Thu, 20 Oct 2016 09:17:38 +0000 (11:17 +0200)]
[mkbundle] Use correctly link paths when loading assembly deps
Brian Raderman [Wed, 8 Mar 2017 19:13:02 +0000 (14:13 -0500)]
invoke_array_extract_argument() will create objects and set them in the parameters array when it detects a null parameter, but when we return a by-value parameter may no longer be null, which is an error. This fix detects the change and sets the parameter back to null for by-value parameters before we return.
Adrian Cheng [Wed, 15 Mar 2017 12:57:53 +0000 (20:57 +0800)]
One brace missed
Adrian Cheng [Wed, 15 Mar 2017 12:47:58 +0000 (20:47 +0800)]
Clean unrelated changes left
Adrian Cheng [Wed, 15 Mar 2017 12:40:34 +0000 (20:40 +0800)]
Try hardly to fix the revert the style changes I made...
Adrian Cheng [Wed, 15 Mar 2017 10:53:57 +0000 (18:53 +0800)]
Put the spaces back before calling function and indexing to follow coding guidelins
Adrian Cheng [Wed, 15 Mar 2017 02:40:41 +0000 (10:40 +0800)]
Change brace style to follow coding guidelines
Adrian Cheng [Wed, 15 Mar 2017 02:32:46 +0000 (10:32 +0800)]
Replace whitespaces to 8-spaces tabs
Adrian Cheng [Tue, 14 Mar 2017 11:08:48 +0000 (19:08 +0800)]
Fix Bug #40969
Marek Safar [Tue, 21 Mar 2017 18:24:58 +0000 (19:24 +0100)]
Bump cecil
Aaron Bockover [Wed, 22 Mar 2017 14:30:36 +0000 (10:30 -0400)]
csharp: support command line arguments for scripts
This adds a command line argument preprocessor that splits command line
arguments into two groups, delimited by the "standard" `--` "stop
processing" argument, or by the `-s` argument which can be used with
Unix shebang syntax (where the path of the file being executed will
be appended to the command line specified on the shebang line).
- anything before `--` or `-s` will be passed to mcs/csharp as usual
- anything after `--` or `-s` will _not_ be processed by mcs/csharp,
and will be made available in an `Args` builtin, like `csi`
This _finally_ (after almost 10 years) allows `csharp` to be used for
proper script authoring!
Document the new arguments in the man page. Also remove references to
'gsharp' from the man page.
Marek Safar [Thu, 23 Mar 2017 09:14:56 +0000 (10:14 +0100)]
[System.Data] Fixes tests build with mobile profiles
Marek Safar [Wed, 22 Mar 2017 22:59:07 +0000 (23:59 +0100)]
[System.Configuration] Refactor InitForConfiguration to be readable
Vlad Brezae [Fri, 17 Mar 2017 20:41:04 +0000 (22:41 +0200)]
[runtime] Fix missing abort during abort protected regions
When aborting a thread we set the thread's interruption request and we also increment the global interruption request. If we were trying to abort a thread when it is in an abort protected region, it will just set an interruption deferred and not do anything else (in async_abort_critical). When the thread exited the abort protected region, it was missing the abort request since we didn't increment the global interruption request.
Refactor thread_interruption_requested modifications, making sure we always bump it when requesting an interruption and decrease it when we clear an interruption.
Vlad Brezae [Fri, 17 Mar 2017 13:05:10 +0000 (15:05 +0200)]
[runtime] Rework abort deferring
The problem with previous implementation was that we were not able to account for two pending exceptions at the same time, since when we execute an interruption request we clear all the state and there is no clear way to account for two exceptions at the same time. This means that, if we have a pending abort in a cctor, all exception handling inside the cctor will be broken (for the provided test case we were only throwing the cctor internal exception outside of it).
In order to fix this, we split into two types of interruptions SYNC and ASYNC (in order to separate mainly between abort and pending exceptions). Sync interruptions are interruptions requested by the thread itself (pending exceptions and self aborts) and they are processed as soon as possible. Async interruptions are requested by other threads (abort, interrupt, suspend) and we defer them as we were doing before (we have a deferred interruption if we have an async exception and abort block count is non zero).
Marek Safar [Wed, 22 Mar 2017 16:45:33 +0000 (17:45 +0100)]
[xbuild] Copy csc/mcs GC settings to xbuild script
Chris Hamons [Wed, 22 Mar 2017 18:15:02 +0000 (13:15 -0500)]
AppleTls Support (#4470)
monojenkins [Wed, 22 Mar 2017 16:30:36 +0000 (16:30 +0000)]
[msvc] Update csproj files
Egor Bogatov [Wed, 22 Mar 2017 12:51:57 +0000 (15:51 +0300)]
[System.Data] Integration tests (#4538)
EgorBo [Wed, 15 Mar 2017 22:11:11 +0000 (01:11 +0300)]
[System.Data] Fix bug-53217: use AddressFamily.InterNetworkV6 when needed
Zoltan Varga [Wed, 22 Mar 2017 08:20:28 +0000 (04:20 -0400)]
[runtime] Fix another typo in get_errno_from_g_file_error ().
Zoltan Varga [Wed, 22 Mar 2017 08:17:30 +0000 (04:17 -0400)]
[runtime] Fix a typo causing an assert in get_errno_from_g_file_error (). Fixes #53684.
Marek Safar [Tue, 21 Mar 2017 22:24:51 +0000 (23:24 +0100)]
[msvc] Fixes Cecil project update
Zoltan Varga [Tue, 21 Mar 2017 20:46:08 +0000 (16:46 -0400)]
[runtime] Add some documentation for the offsets stuff.
monojenkins [Tue, 21 Mar 2017 16:22:38 +0000 (16:22 +0000)]
[msvc] Update csproj files
Zoltan Varga [Tue, 21 Mar 2017 15:43:28 +0000 (11:43 -0400)]
[ppdb] Avoid asserting on ppdb images without a guid heap.
Alexander Köplinger [Tue, 21 Mar 2017 12:04:19 +0000 (13:04 +0100)]
[System.Web] Fix standalone-runner to return non-zero exit code on failure (#4562)
* [System.Web] Fix standalone-runner to return non-zero exit code on failure
We didn't notice that test runs with standalone-runner.exe failed
because it still exited with 0 in those cases.
Fixed that and the standalone tests that showed up as failed in
System.Web.Extensions (some differences likely from referencesource
code import).
* Fix replacement logic for Consts.cs.in to use Windows paths
Marek Safar [Tue, 21 Mar 2017 11:30:10 +0000 (12:30 +0100)]
Bump linker
Marek Safar [Mon, 20 Mar 2017 22:57:31 +0000 (23:57 +0100)]
[System] Fix Process tests on watch
Marek Safar [Mon, 20 Mar 2017 22:03:29 +0000 (23:03 +0100)]
[System.Core] Update mobile test
Vlad Brezae [Wed, 15 Mar 2017 23:15:58 +0000 (01:15 +0200)]
[runtime] Fix interruption deferring
We were accidentally clearing the interruption deferred flag when exiting from nested abort protected regions.
Alexander Kyte [Mon, 20 Mar 2017 18:52:42 +0000 (14:52 -0400)]
Merge pull request #4555 from alexanderkyte/finished_mangler
[runtime] Finish the AOT mangler
monojenkins [Mon, 20 Mar 2017 16:36:02 +0000 (16:36 +0000)]
[msvc] Update csproj files
Rodrigo Kumpera [Mon, 20 Mar 2017 15:33:43 +0000 (08:33 -0700)]
Merge pull request #4554 from kumpera/cleanup-random-stuff
Remove debug spew plus fix the code from PR #4536
Zoltan Varga [Mon, 20 Mar 2017 11:46:40 +0000 (07:46 -0400)]
[arm64] Fix the DISABLE_JIT build.
Vladimir Kazakov [Sat, 18 Mar 2017 21:26:59 +0000 (23:26 +0200)]
System.Security.Cryptography.X509Certificates enums from .NET Core.
TrustStatus and X509SelectionFlag are missing from .NET Core (https://github.com/dotnet/corefx/issues/17194), hence TrustStatus was taken from .NET Framework and X509SelectionFlag was left untouched.
Zoltan Varga [Mon, 20 Mar 2017 10:45:54 +0000 (06:45 -0400)]
[simd] Add some missing Vector<T> intrinsics. Add some code to print out missing intrinsics when running with -v -v. (#4560)
monojenkins [Sun, 19 Mar 2017 16:34:37 +0000 (16:34 +0000)]
[msvc] Update csproj files
Zoltan Varga [Mon, 20 Mar 2017 04:16:20 +0000 (00:16 -0400)]
[x86] Implement OP_EXPAND_I1 by lowering instead of directly to avoid register allocation problems. Increase stack size when simd types are used. (#4557)
Bernhard Urban [Fri, 17 Mar 2017 16:02:02 +0000 (17:02 +0100)]
[interp] allocate storage for value types in runtime_invoke
Bernhard Urban [Fri, 17 Mar 2017 16:02:02 +0000 (17:02 +0100)]
[interp] treat GENERICINST properly on runtime_invoke return
Bernhard Urban [Fri, 17 Mar 2017 16:02:02 +0000 (17:02 +0100)]
[interp] box IntPtr on return of runtime_invoke
Bernhard Urban [Fri, 17 Mar 2017 16:02:02 +0000 (17:02 +0100)]
[interp] add UnsafeMov intrinsic support
Bernhard Urban [Fri, 17 Mar 2017 16:02:02 +0000 (17:02 +0100)]
[interp] fix offset for starg_vt insn
fixes an issue in TestDriver, thus enabling some more tests.
Bernhard Urban [Fri, 17 Mar 2017 16:02:02 +0000 (17:02 +0100)]
[interp] remove register keyword hack.
we have 2017.
Bernhard Urban [Fri, 17 Mar 2017 16:02:02 +0000 (17:02 +0100)]
[interp] call mono_thread_abort after an exception couldn't be handled
Bernhard Urban [Fri, 17 Mar 2017 16:02:02 +0000 (17:02 +0100)]
[interp] don't assert on stack walking if we don't find a frame
e.g. `is_running_protected_wrapper` doesn't necessarily find a matching frame.
Bernhard Urban [Fri, 17 Mar 2017 15:58:49 +0000 (16:58 +0100)]
[lldb] update `monobt` so it understands interpreter frames.
```
(lldb) monobt
* thread #1
* frame #0: 0x00000001001b236e mono-sgen`interp_transform_call(td=0x00007fff5fbfd080, method=0x0000000100915a90, target_method=0x0000000000000000, domain=0x000000010090b741
frame #1: 0x00000001001a1c2e mono-sgen`generate(method=0x0000000100915a90, rtm=0x000000010382ac70, is_bb_start="\x01", generic_context=0x0000000100915ad0) + 9454 at tran8
transforming TestMonoAsyncGenerics::AsyncWithAwait || frame #2: 0x000000010019f553 mono-sgen`mono_interp_transform_method(runtime_method=0x000000010382ac70, context=0x004
TestMonoAsyncGenerics::AsyncWithAwait @ 0 || frame #3: 0x000000010018a178 mono-sgen`ves_exec_method_with_context(frame=0x00007fff5fbfe290, context=0x00007fff5fbfe3a8) +9
TestMonoAsyncGenerics::Main @ 12 "pop" || frame #4: 0x000000010018b4b1 mono-sgen`ves_exec_method_with_context(frame=0x00007fff5fbfe420, context=0x00007fff5fbfe3a8) + 5081
frame #5: 0x0000000100189e43 mono-sgen`mono_interp_runtime_invoke(method=0x000000010090ce38, obj=0x0000000000000000, params=0x00007fff5fbfea40, exc=0x0000000000000000, e0
frame #6: 0x00000001000164a2 mono-sgen`mono_jit_runtime_invoke(method=0x000000010090ce38, obj=0x0000000000000000, params=0x00007fff5fbfea40, exc=0x0000000000000000, erro1
frame #7: 0x000000010038b2b5 mono-sgen`do_runtime_invoke(method=0x000000010090ce38, obj=0x0000000000000000, params=0x00007fff5fbfea40, exc=0x0000000000000000, error=0x002
frame #8: 0x0000000100384e97 mono-sgen`mono_runtime_invoke_checked(method=0x000000010090ce38, obj=0x0000000000000000, params=0x00007fff5fbfea40, error=0x00007fff5fbfeb000
frame #9: 0x000000010038f335 mono-sgen`do_exec_main_checked(method=0x000000010090ce38, args=0x00000001020003c8, error=0x00007fff5fbfeb00) + 197 at object.c:4672
frame #10: 0x000000010038dd5c mono-sgen`mono_runtime_exec_main_checked(method=0x000000010090ce38, args=0x00000001020003c8, error=0x00007fff5fbfeb00) + 76 at object.c:4773
frame #11: 0x000000010038ddbf mono-sgen`mono_runtime_run_main_checked(method=0x000000010090ce38, argc=1, argv=0x00007fff5fbfef68, error=0x00007fff5fbfeb00) + 79 at objec2
frame #12: 0x00000001000d9a33 mono-sgen`mono_jit_exec(domain=0x000000010090b740, assembly=0x0000000100913610, argc=1, argv=0x00007fff5fbfef68) + 403 at driver.c:1029
frame #13: 0x00000001000dd9da mono-sgen`main_thread_handler(user_data=0x00007fff5fbfeea0) + 538 at driver.c:1098
frame #14: 0x00000001000dc21c mono-sgen`mono_main(argc=3, argv=0x00007fff5fbfef58) + 8636 at driver.c:2163
frame #15: 0x0000000100001b9e mono-sgen`mono_main_with_options(argc=3, argv=0x00007fff5fbfef58) + 46 at main.c:45
frame #16: 0x00000001000012dd mono-sgen`main(argc=3, argv=0x00007fff5fbfef58) + 77 at main.c:338
frame #17: 0x00007fffc2e66255 libdyld.dylib`start + 1
frame #18: 0x00007fffc2e66255 libdyld.dylib`start + 1
```
Zoltan Varga [Fri, 17 Mar 2017 23:05:39 +0000 (19:05 -0400)]
[arm] Fix the DISABLE_JIT build.
Frederik Carlier [Fri, 17 Mar 2017 22:30:35 +0000 (23:30 +0100)]
Make System.Drawing code compatible with .NET Core (#4548)
Alexander Kyte [Wed, 18 Jan 2017 15:50:40 +0000 (15:50 +0000)]
[runtime] Finish AOT method name mangler
Rodrigo Kumpera [Fri, 17 Mar 2017 20:30:09 +0000 (13:30 -0700)]
[runtime] Remove debug spew.
Rodrigo Kumpera [Fri, 17 Mar 2017 20:29:57 +0000 (13:29 -0700)]
[runtime] Remove debug leftover and have mono_assembly_is_problematic_version actually work.
su21 [Fri, 17 Mar 2017 19:35:41 +0000 (03:35 +0800)]
less allocation in Socket (#4516)
* less allocation in Socket
avoid allocate `System.Threading.QueueUserWorkItemCallback` in `SocketAsyncResult.Complete`
avoid allocate `Task` and anonymous function in `Socket.QueueIOSelectorJob`
* update: fix race and use replace static field with in line lambda
* cleanup code
* QueueIOSelectorJob keep async, and provie a fastpath
* SocketAsyncResult: simplify callback code
* add missing line `IOSelect.Add(handle, job)` :(
* fix typo
* fix compile error
Andi McClure [Fri, 17 Mar 2017 15:58:05 +0000 (11:58 -0400)]
Bump version number to 4.9.4 to support automated builders
Marek Safar [Sat, 11 Mar 2017 13:37:01 +0000 (14:37 +0100)]
Add support for xunit tests
Marek Safar [Fri, 17 Mar 2017 13:07:38 +0000 (14:07 +0100)]
[System] Add back empty MonoTouch namespace
Johan Lorensson [Fri, 17 Mar 2017 09:03:22 +0000 (10:03 +0100)]
Merge pull request #4542 from lateralusX/jlorenss/win-fix-unwind-tramp-reg-aot
mono_arch_unwindinfo_install_tramp_unwind_info can only be called for JIT:ed code.
Zoltan Varga [Fri, 17 Mar 2017 07:32:52 +0000 (03:32 -0400)]
[jit] Remove support for the separate mono_lmf variable in the JIT, use the mono_lmf_addr () variable instead. The former could be used to generate slighly faster code in managed-to-native wrappers, but its not worth all the complexity it caused. (#4549)
Frederik Carlier [Thu, 16 Mar 2017 23:10:44 +0000 (00:10 +0100)]
Remove CAS attributes from System.Drawing (#4544)
Aleksey Kliger (λgeek) [Thu, 16 Mar 2017 23:09:00 +0000 (19:09 -0400)]
Merge pull request #4453 from lambdageek/bug-49721
[loader] Check strong name when loading from application base. (Fixes #49721)