mono.git
10 years agoGetIdElement must look for "ID" and "id". Fixes #4938
Jonathan Gagnon [Fri, 12 Jul 2013 18:48:55 +0000 (14:48 -0400)]
GetIdElement must look for "ID" and "id". Fixes #4938

10 years agoPrefer user operators over predefined operators for binary enum operations too. Fixes...
Marek Safar [Fri, 12 Jul 2013 14:23:01 +0000 (16:23 +0200)]
Prefer user operators over predefined operators for binary enum operations too. Fixes #13202

10 years agoAllow async for explicitly implemented interface methods. Fixes #13197
Marek Safar [Fri, 12 Jul 2013 13:13:42 +0000 (15:13 +0200)]
Allow async for explicitly implemented interface methods. Fixes #13197

10 years ago.spec file, again.
Martin Baulig [Fri, 12 Jul 2013 08:42:12 +0000 (10:42 +0200)]
.spec file, again.

10 years agoMerge pull request #622 from killabytenow/master
Atsushi Eno [Fri, 12 Jul 2013 05:17:05 +0000 (22:17 -0700)]
Merge pull request #622 from killabytenow/master

UTF8 encoding/decoding improvements

10 years agoReplace more __ARM_ARCH_... checks with HAVE_ARMV<N>.
Alex Rønne Petersen [Thu, 11 Jul 2013 22:03:02 +0000 (00:03 +0200)]
Replace more __ARM_ARCH_... checks with HAVE_ARMV<N>.

10 years agoReverted changes of commit 6ad7f75a3777652a535ef4356e6d193eed4ed612
Gerardo García Peña [Thu, 11 Jul 2013 21:52:35 +0000 (23:52 +0200)]
Reverted changes of commit 6ad7f75a3777652a535ef4356e6d193eed4ed612

10 years agoBetter and more homogeneous error handling.
Gerardo García Peña [Thu, 11 Jul 2013 21:49:47 +0000 (23:49 +0200)]
Better and more homogeneous error handling.

10 years agoAdd ARM version detection to configure and use the resulting macros.
Alex Rønne Petersen [Thu, 11 Jul 2013 19:19:34 +0000 (21:19 +0200)]
Add ARM version detection to configure and use the resulting macros.

This defines ARM_VERSION_{5,6,7}. Several of these can be
defined at once. For example ARM_VERSION_5 and ARM_VERSION_6
will both be defined if the target is an ARM v6 CPU.

Previously, we defined HAVE_ARMV6 based on running a program.
This obviously didn't work for cross-compilation. This new
approach does, and detects more cases (v5, v6, v7) than the
previous check (v6).

10 years agoAdd a mono-hwcap module to utils.
Alex Rønne Petersen [Thu, 11 Jul 2013 18:12:01 +0000 (20:12 +0200)]
Add a mono-hwcap module to utils.

This is an attempt to unify all the ad-hoc CPU feature
detection we do all over the codebase (mostly in Mini
back ends). It will also be used to detect and use SMP
instructions on older ARM devices (we need this info in
utils/, not just in mini/).

Each architecture has a mono-hwcap-$TARGET.h header if
it has specified one. So for example, if you want to
find out if a VFP unit is available on ARM, you would
include mono-hwcap-arm.h and read mono_hwcap_arm_has_vfp.
Remember to guard inclusions of these files with a check
along the lines of:

    #ifdef TARGET_ARM
    #include "mono/utils/mono-hwcap-arm.h"
    /* ... */
    #endif

Architectures that don't specify a file simply get an
empty implementation with no feature variables. This
empty implementation is also used if Mono itself is
compiled as a cross compiler.

Initialization of the mono-hwcap module is done in the
mini_init () function in mini/mini.c, very early in the
startup process. So in general, calling it is not needed.

10 years agoFix the tests makefile.
Alex Rønne Petersen [Thu, 11 Jul 2013 17:54:35 +0000 (19:54 +0200)]
Fix the tests makefile.

10 years agoIf 'flush' flag is enabled when calling decoding/encoding function, flush buffer...
Gerardo García Peña [Thu, 11 Jul 2013 16:54:11 +0000 (18:54 +0200)]
If 'flush' flag is enabled when calling decoding/encoding function, flush buffer when input run out (satisfies test 'SufficientByteArray').

10 years agoFix (hopefully) the last architecture casing inconsistency.
Alex Rønne Petersen [Thu, 11 Jul 2013 16:42:18 +0000 (18:42 +0200)]
Fix (hopefully) the last architecture casing inconsistency.

10 years agoRemove an m68k check from configure.
Alex Rønne Petersen [Thu, 11 Jul 2013 16:38:36 +0000 (18:38 +0200)]
Remove an m68k check from configure.

10 years agoFix an s390x check broken by faff4654c8cbfbe0619e87808f65a22d2b39d413.
Alex Rønne Petersen [Thu, 11 Jul 2013 16:37:51 +0000 (18:37 +0200)]
Fix an s390x check broken by faff4654c8cbfbe0619e87808f65a22d2b39d413.

10 years agoAdd missing TARGET_* defines for SPARC and Itanium.
Alex Rønne Petersen [Thu, 11 Jul 2013 16:36:31 +0000 (18:36 +0200)]
Add missing TARGET_* defines for SPARC and Itanium.

10 years agoOops, remove extra %endif in .spec file.
Martin Baulig [Wed, 10 Jul 2013 16:09:14 +0000 (18:09 +0200)]
Oops, remove extra %endif in .spec file.

10 years ago[PCL]: Add System.Linq.Expressions type-forwarders.
Martin Baulig [Wed, 10 Jul 2013 14:51:53 +0000 (16:51 +0200)]
[PCL]: Add System.Linq.Expressions type-forwarders.

10 years agoAdd libmonoboehm package to the spec file.
Martin Baulig [Wed, 10 Jul 2013 14:46:02 +0000 (16:46 +0200)]
Add libmonoboehm package to the spec file.

10 years agoIgnore only local Makefile
Marek Safar [Wed, 10 Jul 2013 14:26:26 +0000 (16:26 +0200)]
Ignore only local Makefile

10 years agoAn #if has been negated (bugfix)
Gerardo García Peña [Wed, 10 Jul 2013 09:26:26 +0000 (11:26 +0200)]
An #if has been negated (bugfix)

10 years agoAdd atomic.c/.h to libmonoutils.vcxproj.
Alex Rønne Petersen [Wed, 10 Jul 2013 04:54:33 +0000 (06:54 +0200)]
Add atomic.c/.h to libmonoutils.vcxproj.

10 years agoconfigure: Set TARGET=S390X instead of S390x for consistency.
Alex Rønne Petersen [Wed, 10 Jul 2013 04:47:02 +0000 (06:47 +0200)]
configure: Set TARGET=S390X instead of S390x for consistency.

10 years agoMerge pull request #695 from jack-pappas/patch-4
Alex Rønne Petersen [Wed, 10 Jul 2013 04:27:40 +0000 (21:27 -0700)]
Merge pull request #695 from jack-pappas/patch-4

Fix preprocessor/compiler warning due to unbalanced apostophe (') character

10 years agoFix preprocessor/compiler warning due to unbalanced apostophe (') character
Jack Pappas [Wed, 10 Jul 2013 01:44:19 +0000 (21:44 -0400)]
Fix preprocessor/compiler warning due to unbalanced apostophe (') character

Reworded the warning message to be displayed when the fall-through case is taken by the preprocessor in `mono_thread_get_stack_bounds()`.

10 years agoEnsure the new FileStream unit tests does not throw in the non-main thread (which...
Sebastien Pouliot [Wed, 10 Jul 2013 00:17:37 +0000 (20:17 -0400)]
Ensure the new FileStream unit tests does not throw in the non-main thread (which nunit-console won't report but other tools, like Touch.Unit, will)

10 years agoUse a non rooted symlink and make it point to the right mono.
Rodrigo Kumpera [Tue, 9 Jul 2013 22:28:58 +0000 (18:28 -0400)]
Use a non rooted symlink and make it point to the right mono.

10 years ago[runtime]Revert the fast tls check as we always need it on mach. Add another iOS...
Rodrigo Kumpera [Tue, 9 Jul 2013 22:06:11 +0000 (18:06 -0400)]
[runtime]Revert the fast tls check as we always need it on mach. Add another iOS 7 offset.

10 years agoBump libmonosgen version in the .spec file.
Martin Baulig [Tue, 9 Jul 2013 22:05:28 +0000 (00:05 +0200)]
Bump libmonosgen version in the .spec file.

10 years ago[build]: Use a relative symlink from mono to mono-sgen.
Martin Baulig [Tue, 9 Jul 2013 20:44:16 +0000 (22:44 +0200)]
[build]: Use a relative symlink from mono to mono-sgen.

10 years agoMerge pull request #694 from madewokherd/pthreadinclude
Rodrigo Kumpera [Tue, 9 Jul 2013 20:19:54 +0000 (13:19 -0700)]
Merge pull request #694 from madewokherd/pthreadinclude

Don't include pthread.h when it's not needed.

10 years agoFix a #define style issue in mini.c.
Alex Rønne Petersen [Tue, 9 Jul 2013 19:41:12 +0000 (21:41 +0200)]
Fix a #define style issue in mini.c.

Introduced in d4016989729bd58274c03d7b8ed9495bfbfef82f.

10 years agoOnly require mono_arch_is_soft_float () to be defined when needed.
Alex Rønne Petersen [Tue, 9 Jul 2013 19:34:59 +0000 (21:34 +0200)]
Only require mono_arch_is_soft_float () to be defined when needed.

Now, only ports that define MONO_ARCH_SOFT_FLOAT_FALLBACK have to
provide this function. For ports that don't define this, a default
function is provided that returns FALSE.

10 years agoAdd an MSVC branch to MONO_ALWAYS_INLINE.
Alex Rønne Petersen [Tue, 9 Jul 2013 19:34:44 +0000 (21:34 +0200)]
Add an MSVC branch to MONO_ALWAYS_INLINE.

10 years ago[runtime] On mach-arm don't probe for tls offsets if the runtime was not built with...
Rodrigo Kumpera [Tue, 9 Jul 2013 19:49:51 +0000 (15:49 -0400)]
[runtime] On mach-arm don't probe for tls offsets if the runtime was not built with support for fast tls.

10 years agoChange the way we do soft float to a fallback mechanism.
Alex Rønne Petersen [Tue, 9 Jul 2013 16:56:38 +0000 (18:56 +0200)]
Change the way we do soft float to a fallback mechanism.

MONO_ARCH_SOFT_FLOAT previously meant that a back end was using
soft float unconditionally. It's now MONO_ARCH_SOFT_FLOAT_FALLBACK
and actually indicates whether the back end has a soft float
fallback compiled in - not necessarily that it's used.

A new port function has been added:

    gboolean mono_arch_is_soft_float (void) MONO_INTERNAL;

This function is what actually tells Mini whether the back end is
going to use soft float. If MONO_ARCH_SOFT_FLOAT_FALLBACK is not
defined, it must return FALSE. Otherwise, it should return TRUE or
FALSE depending on whether the back end determines that the hardware
supports floating point. Once mono_arch_init () has been called,
the value returned from this function must be constant for the rest
of the VM's lifetime.

For example, on ARM, we now try to detect VFP via /proc/self/auxv
or /proc/cpuinfo and, if found, return FALSE from the function. This
only applies to Linux and Android as iOS always has VFP.

It's important to note that if the runtime is built with the 'softfp'
or 'hard' ABIs, the soft float fallback is not compiled in, as the
runtime is assumed to only be executed on VFP-capable systems.

Finally, the ARM feature detection code has been moved from the
mono_arch_cpu_optimizations () function to mono_arch_init () so that
FPU detection/initialization is done in one place.

10 years ago[build] Add dep tracking between Makefile.am.in and Makefile.am when the extension...
Rodrigo Kumpera [Tue, 9 Jul 2013 17:21:54 +0000 (13:21 -0400)]
[build] Add dep tracking between Makefile.am.in and Makefile.am when the extension module is disabled.

10 years ago[runtime]On install, symlink mono to the right binary.
Rodrigo Kumpera [Tue, 9 Jul 2013 17:16:09 +0000 (13:16 -0400)]
[runtime]On install, symlink mono to the right binary.

10 years agoFix the spec file; we don't create any .mdb's for the PCL Facades anymore.
Martin Baulig [Tue, 9 Jul 2013 16:37:34 +0000 (18:37 +0200)]
Fix the spec file; we don't create any .mdb's for the PCL Facades anymore.

10 years agoDon't include pthread.h when it's not needed.
Vincent Povirk [Wed, 3 Jul 2013 20:15:41 +0000 (15:15 -0500)]
Don't include pthread.h when it's not needed.

Fixes mingw-w64 build.

This commit is licensed under MIT/X11.

10 years agoMerge pull request #692 from jack-pappas/patch-2
Alex Rønne Petersen [Tue, 9 Jul 2013 15:49:58 +0000 (08:49 -0700)]
Merge pull request #692 from jack-pappas/patch-2

Use the NI_MAXHOST constant to initialize the size of the hostname buffer.

10 years agoMIPS: Set USE_MUL to 0 by default (for now).
Alex Rønne Petersen [Tue, 9 Jul 2013 14:48:15 +0000 (16:48 +0200)]
MIPS: Set USE_MUL to 0 by default (for now).

Setting it to 1 breaks on Loongson CPUs which are the most common
desktop MIPS CPUs supported by e.g. Debian.

This change can be reverted when more investigation is done to find
out why exactly this breaks.

Thanks to Jo Shields (@directhex) for lots of testing and debugging
and spotting this define that turned out to be the culprit.

10 years agoAdded new test for checking EncoderFallbackException exception behaviour.
Gerardo García Peña [Tue, 9 Jul 2013 14:22:37 +0000 (16:22 +0200)]
Added new test for checking EncoderFallbackException exception behaviour.

10 years agoMerge pull request #693 from jack-pappas/patch-3
Marek Safar [Tue, 9 Jul 2013 12:42:18 +0000 (05:42 -0700)]
Merge pull request #693 from jack-pappas/patch-3

Print diagnostic messages for platform detection

10 years agoPrint diagnostic messages for platform detection
Jack Pappas [Tue, 9 Jul 2013 12:35:26 +0000 (08:35 -0400)]
Print diagnostic messages for platform detection

Print some diagnostic messages when running 'make' on this file, so it's clear when a platform has not been explicitly specified and we're going to compile with a platform value which has been automatically selected.

10 years agoUse the NI_MAXHOST constant to initialize the size of the hostname buffer.
Jack Pappas [Tue, 9 Jul 2013 12:12:41 +0000 (08:12 -0400)]
Use the NI_MAXHOST constant to initialize the size of the hostname buffer.

The NI_MAXHOST constant should be used when creating the hostname buffer to avoid potential buffer overflows.

10 years agoUpdate test name
Marek Safar [Tue, 9 Jul 2013 09:14:56 +0000 (11:14 +0200)]
Update test name

10 years agoEquality structural comparer needs to do recursive comparisons. Fixes #13110
Marek Safar [Tue, 9 Jul 2013 09:12:33 +0000 (11:12 +0200)]
Equality structural comparer needs to do recursive comparisons. Fixes #13110

10 years agoUpdate failing test
Marek Safar [Tue, 9 Jul 2013 08:47:29 +0000 (10:47 +0200)]
Update failing test

10 years agoUpdate member access completion after e40775d0dcc0512c444b995e631880412177ed21
Marek Safar [Tue, 9 Jul 2013 08:41:55 +0000 (10:41 +0200)]
Update member access completion after e40775d0dcc0512c444b995e631880412177ed21

10 years ago[runtime] Don't return success on invalid parameters, raise an exception instead.
Rodrigo Kumpera [Mon, 8 Jul 2013 23:20:10 +0000 (19:20 -0400)]
[runtime] Don't return success on invalid parameters, raise an exception instead.

10 years agoPlace the method_addresses symbol into the .text segment on mt, not rodata, since...
Zoltan Varga [Mon, 8 Jul 2013 22:26:39 +0000 (00:26 +0200)]
Place the method_addresses symbol into the .text segment on mt, not rodata, since it is actually code.

10 years agoBump mono to 3.1.1
Rodrigo Kumpera [Mon, 8 Jul 2013 21:47:00 +0000 (17:47 -0400)]
Bump mono to 3.1.1

10 years agoAdded new test for checking DecoderFallbackException exception behaviour.
Gerardo García Peña [Mon, 8 Jul 2013 21:31:06 +0000 (23:31 +0200)]
Added new test for checking DecoderFallbackException exception behaviour.

10 years agoUpdate after newline update
Marek Safar [Mon, 8 Jul 2013 21:11:00 +0000 (23:11 +0200)]
Update after newline update

10 years agoRemove the constraint checks from get_shared_inst (), it is no longer needed since...
Zoltan Varga [Mon, 8 Jul 2013 20:30:11 +0000 (22:30 +0200)]
Remove the constraint checks from get_shared_inst (), it is no longer needed since we replace gsharedvt arguments with a vtype now. Fixes #12727.

10 years agoMake the synchronized wrappers static if the original method was static. Fixes shared...
Zoltan Varga [Mon, 8 Jul 2013 16:24:39 +0000 (18:24 +0200)]
Make the synchronized wrappers static if the original method was static. Fixes shared-generic-synchronized.2.exe.

10 years agoUpdated gitignore
Marek Safar [Mon, 8 Jul 2013 16:12:54 +0000 (18:12 +0200)]
Updated gitignore

10 years agoAdd a test for #12110.
Zoltan Varga [Mon, 8 Jul 2013 15:13:43 +0000 (17:13 +0200)]
Add a test for #12110.

10 years agoRevert "Revert "Revert "Merge"""
Marek Safar [Mon, 8 Jul 2013 15:01:47 +0000 (17:01 +0200)]
Revert "Revert "Revert "Merge"""

This reverts commit 5705602f1b01fa05366c9c0328a9b5e3e4eecd89.

10 years agoRevert "Revert "Merge""
Marek Safar [Mon, 8 Jul 2013 15:00:36 +0000 (17:00 +0200)]
Revert "Revert "Merge""

This reverts commit ed1838b4e706dc1ba2d701c2f6b5fc71df03bfb3.

10 years agoRevert "Merge"
Marek Safar [Mon, 8 Jul 2013 14:58:33 +0000 (16:58 +0200)]
Revert "Merge"

This reverts commit cbe746f29f75cb72c0b0df1e8647d33238b5096f.

10 years agoMerge
Marek Safar [Mon, 8 Jul 2013 14:49:13 +0000 (16:49 +0200)]
Merge

10 years agoAdd test
Marek Safar [Mon, 8 Jul 2013 07:17:18 +0000 (09:17 +0200)]
Add test

10 years agoMerge pull request #686 from LHCGreg/12971-IsRedirectedAfterSet
Marek Safar [Mon, 8 Jul 2013 06:58:15 +0000 (23:58 -0700)]
Merge pull request #686 from LHCGreg/12971-IsRedirectedAfterSet

Fix behavior of Console.Is*Redirected.

10 years agoRemove some gcc-ism from mono_atomic_load_acquire/release, so it can be compiled...
Zoltan Varga [Sun, 7 Jul 2013 01:47:25 +0000 (03:47 +0200)]
Remove some gcc-ism from mono_atomic_load_acquire/release, so it can be compiled by msvc.

10 years agoMerge pull request #690 from akoeplinger/httpruntime-net45
Marek Safar [Sat, 6 Jul 2013 15:44:45 +0000 (08:44 -0700)]
Merge pull request #690 from akoeplinger/httpruntime-net45

Added handling of 'targetFramework' attribute to HttpRuntime

10 years agoAdded handling of 'targetFramework' attribute (new in .NET 4.5) to HttpRuntime and...
Alexander Köplinger [Sat, 6 Jul 2013 14:36:37 +0000 (16:36 +0200)]
Added handling of 'targetFramework' attribute (new in .NET 4.5) to HttpRuntime and HttpRuntimeSection

10 years agoFixes build
Marek Safar [Sat, 6 Jul 2013 10:13:18 +0000 (12:13 +0200)]
Fixes build

10 years agoMerge pull request #687 from miniBill/master
Marek Safar [Sat, 6 Jul 2013 07:16:29 +0000 (00:16 -0700)]
Merge pull request #687 from miniBill/master

Add IISVersion property to HttpRuntime

10 years agoAdd IISVersion property to HttpRuntime
Leonardo Taglialegne [Sat, 6 Jul 2013 07:01:00 +0000 (09:01 +0200)]
Add IISVersion property to HttpRuntime

10 years agoCorrectly reset buffer for nested asynchronous file stream writes. Fixes #12777
Marek Safar [Fri, 5 Jul 2013 16:47:22 +0000 (18:47 +0200)]
Correctly reset buffer for nested asynchronous file stream writes. Fixes #12777

10 years agoEmit additional sequence points for call entry when needed. Fixes #10782 and similars
Marek Safar [Fri, 5 Jul 2013 12:13:17 +0000 (14:13 +0200)]
Emit additional sequence points for call entry when needed. Fixes #10782 and similars

10 years agoCheck expression type of resolved expressions too
Marek Safar [Thu, 4 Jul 2013 16:43:31 +0000 (18:43 +0200)]
Check expression type of resolved expressions too

10 years agoFix behavior of Console.Is*Redirected.
Greg Najda [Thu, 4 Jul 2013 20:34:34 +0000 (16:34 -0400)]
Fix behavior of Console.Is*Redirected.

Setting Console.Out to a custom stream should not affect the value
of Console.IsOutputRedirected. This is how Microsoft's implementation
behaves.

10 years agoMake synchronized wrappers of generic methods generic the same way it is done with...
Zoltan Varga [Thu, 4 Jul 2013 23:02:10 +0000 (01:02 +0200)]
Make synchronized wrappers of generic methods generic the same way it is done with delegate invokes.

10 years agoAvoid passing class pointers to some marshal icalls since its not aot compatible...
Zoltan Varga [Thu, 4 Jul 2013 21:42:07 +0000 (23:42 +0200)]
Avoid passing class pointers to some marshal icalls since its not aot compatible. Fixes #12595.

10 years agoInterface members are hidden by class members if a type parameter has an effective...
Marek Safar [Thu, 4 Jul 2013 12:51:00 +0000 (14:51 +0200)]
Interface members  are hidden by class members if a type parameter has an effective base other than object

10 years agoMerge pull request #684 from jack-pappas/patch-7
Alex Rønne Petersen [Thu, 4 Jul 2013 01:03:55 +0000 (18:03 -0700)]
Merge pull request #684 from jack-pappas/patch-7

Clean up preprocessor ifdefs in mono_thread_get_stack_bounds and fix the way thread stack information is retrieved on BSD

10 years agoDisable a failing System.Core test for mt (#12987).
Zoltan Varga [Wed, 3 Jul 2013 23:23:24 +0000 (01:23 +0200)]
Disable a failing System.Core test for mt (#12987).

10 years agoClean up preprocessor ifdefs in mono_thread_get_stack_bounds and fix the way thread...
Jack Pappas [Wed, 3 Jul 2013 22:51:53 +0000 (18:51 -0400)]
Clean up preprocessor ifdefs in mono_thread_get_stack_bounds and fix the way thread stack information is retrieved on BSD

Cleaned up the heavily-nested preprocessor ifdefs. The code within the function was extremely difficult to understand because of the nested preprocessor directives, which has led to a number of avoidable configuration-related bugs. I've de-nested the ifdefs into a mostly-flat style, where we check for each platform just once and the entirety of the code for that platform falls under that single check. This improves the readability of the code and should fix the configuration-related issues on some platforms.

This code also fixes an issue where thread-stack information wasn't retrieved correctly on BSD-based systems; now the platform is detected correctly and the `pthread_attr_get_np` function used to retrieve the information.

This subsumes pull-request #553, which couldn't be merged with the current master branch due to several changes which were made to the `mono_thread_get_stack_bounds` function since that pull request was submitted. This commit also keeps the OpenBSD-related code for auditing purposes, though it may be removed later if testing shows the "generic" BSD handler works for OpenBSD.

10 years ago[xbuild] / [PCL]: Report an error for unsupported / missing PCL Profiles.
Martin Baulig [Wed, 3 Jul 2013 19:26:48 +0000 (21:26 +0200)]
[xbuild] / [PCL]: Report an error for unsupported / missing PCL Profiles.

We previously only reported a warning, then failed later with tons of
"The predefined type `System.Object' is not defined or imported", which
was very confusing to the user.

We now distinguish between missing Reference Assemblies (ie. none of
them are installed) and the requested profile not being available, reporting
a meaningful error message.

10 years agoUpdate the Mono LLVM EH Frame reading code to match llvm 5f3329c6d97bb765dc5459236708...
Zoltan Varga [Wed, 3 Jul 2013 17:45:16 +0000 (19:45 +0200)]
Update the Mono LLVM EH Frame reading code to match llvm 5f3329c6d97bb765dc54592367087d9ce8cd0040.

10 years agoUpdate ast visitor
Marek Safar [Wed, 3 Jul 2013 16:38:18 +0000 (18:38 +0200)]
Update ast visitor

10 years agoLoading of nested type of missing type needs to create another missing type. Fixes...
Marek Safar [Wed, 3 Jul 2013 15:38:13 +0000 (17:38 +0200)]
Loading of nested type of missing type needs to create another missing type. Fixes #12989

10 years agoRename the POSIX MonoThreadInfo::suspend_semaphore field to begin_suspend_semaphore.
Alex Rønne Petersen [Wed, 3 Jul 2013 14:21:35 +0000 (16:21 +0200)]
Rename the POSIX MonoThreadInfo::suspend_semaphore field to begin_suspend_semaphore.

Unbreaks the Linux build.

10 years agoRevert "Remove a duplicate field in mono-threads.h breaking the build on some systems."
Rodrigo Kumpera [Wed, 3 Jul 2013 13:33:52 +0000 (09:33 -0400)]
Revert "Remove a duplicate field in mono-threads.h breaking the build on some systems."

This reverts commit 3600480ced776859df5d48785039c977407a073a.

This the wrong fix.

10 years agoAdd await support in to evaluator and REPL
Marek Safar [Wed, 3 Jul 2013 12:56:23 +0000 (14:56 +0200)]
Add await support in to evaluator and REPL

10 years agoBetter exception for unmarked labels
Marek Safar [Wed, 3 Jul 2013 12:41:29 +0000 (14:41 +0200)]
Better exception for unmarked labels

10 years agoRemove a duplicate field in mono-threads.h breaking the build on some systems.
Alex Rønne Petersen [Wed, 3 Jul 2013 12:48:27 +0000 (14:48 +0200)]
Remove a duplicate field in mono-threads.h breaking the build on some systems.

10 years agoFix the implementation of sext.i4/zext.i4 for LLVM. Fixes test_0_regress_11058 (...
Zoltan Varga [Wed, 3 Jul 2013 01:43:58 +0000 (03:43 +0200)]
Fix the implementation of sext.i4/zext.i4 for LLVM. Fixes test_0_regress_11058 () on amd64.

10 years agoAdd regression tests for the last 3 fixes.
Rodrigo Kumpera [Tue, 2 Jul 2013 22:50:36 +0000 (18:50 -0400)]
Add regression tests for the last 3 fixes.

10 years ago[runtime]Replace suspend safe locks with semaphores as darwin's implementation is...
Rodrigo Kumpera [Tue, 2 Jul 2013 22:20:18 +0000 (18:20 -0400)]
[runtime]Replace suspend safe locks with semaphores as darwin's implementation is not reliable.

* mono-threads.c: On OSX, suspend safe locks are subject to a race condition
where the lock is unlocked but a thread is blocked waiting for it.

Semaphores are significantly slower as they always transition to the kernel, but given
the two affected locks are taken very sporadically, there's nothing to worry here.

10 years agoIntroduce MonoThreadInfo::inside_critical_region to avoid suspending threads that are
Rodrigo Kumpera [Mon, 1 Jul 2013 23:30:03 +0000 (19:30 -0400)]
Introduce MonoThreadInfo::inside_critical_region to avoid suspending threads that are
dealing with suspended threads. One cannot reliably suspend a thread in the middle
of inspecting another one without the risk of deadlocking.

It usually goes like this:

thread 2:
suspend thread 3, thread 3 is in the middle of some critical code.

thread 1:
suspend thread 2 while probing for state of 3;
try to suspend thread 3, fail to resume since it's on critical code.

10 years agoSuspend the world when cleaning up a domain as it's unsafe to walk the nursery with...
Rodrigo Kumpera [Mon, 1 Jul 2013 22:49:11 +0000 (18:49 -0400)]
Suspend the world when cleaning up a domain as it's unsafe to walk the nursery with the world running.

10 years agoDon't create any .mdb's for the PCL Facade Assemblies.
Martin Baulig [Tue, 2 Jul 2013 21:10:28 +0000 (23:10 +0200)]
Don't create any .mdb's for the PCL Facade Assemblies.

10 years agoRevert "Make synchronized wrappers of generic methods generic the same way it is...
Zoltan Varga [Tue, 2 Jul 2013 19:58:17 +0000 (21:58 +0200)]
Revert "Make synchronized wrappers of generic methods generic the same way it is done with delegate invokes. Fixes #12727."

This reverts commit afb16275b091bcde7f7ac8007f4e0e9d7edd52bf.

Revert this as it breaks tests in mini/.

10 years agoRevert "Add configure check for -Wno-tautological-compare when using clang."
Martin Baulig [Tue, 2 Jul 2013 19:12:51 +0000 (21:12 +0200)]
Revert "Add configure check for -Wno-tautological-compare when using clang."

This reverts commit 6015a30c7a3ba8d16740fbae7819585eb0b20d37.

10 years agoMake synchronized wrappers of generic methods generic the same way it is done with...
Zoltan Varga [Tue, 2 Jul 2013 18:48:52 +0000 (20:48 +0200)]
Make synchronized wrappers of generic methods generic the same way it is done with delegate invokes. Fixes #12727.

10 years agoAdd configure check for -Wno-tautological-compare when using clang.
Martin Baulig [Tue, 2 Jul 2013 18:23:57 +0000 (20:23 +0200)]
Add configure check for -Wno-tautological-compare when using clang.