mono.git
13 years agoProperly handle compressed interface bitmaps on the array store wrapper.
Rodrigo Kumpera [Wed, 11 May 2011 21:20:13 +0000 (18:20 -0300)]
Properly handle compressed interface bitmaps on the array store wrapper.

* marshal.c (get_virtual_stelemref_kind): Don't emit optimized
stelemref wrapper if compressed interface bitmaps are enabled.

13 years agoFix windows build
Rodrigo Kumpera [Wed, 11 May 2011 21:14:16 +0000 (18:14 -0300)]
Fix windows build

13 years agoEnable the new interrupt machinery on x86 + linux|mach.
Rodrigo Kumpera [Wed, 11 May 2011 20:31:34 +0000 (17:31 -0300)]
Enable the new interrupt machinery on x86 + linux|mach.

* mono-threads.c (mono_thread_info_new_interrupt_enabled):
Enable it on the targets that have been ported and tested.

13 years agoImplement thread dump using new machinery.
Rodrigo Kumpera [Wed, 11 May 2011 20:09:55 +0000 (17:09 -0300)]
Implement thread dump using new machinery.

* gc.c: Thread dump now uses the finalizer thread
to be more reliable.

* threads.c (mono_threads_perform_thread_dump): New
function that dumps all threads using the new machinery.

* mini-posix.c (sigquit_signal_handler): Use new machinery
if available.

13 years agoMake sure the GC takes the appropriate locks during STW.
Rodrigo Kumpera [Wed, 11 May 2011 19:24:50 +0000 (16:24 -0300)]
Make sure the GC takes the appropriate locks during STW.

* sgen-gc.c:
* boehm-gc.c: The collector must takes the interruption
lock before starting STW to avoid deadlocks. This is specially
problematic in the case of hosts that are not signal based.

For example, say we have threads 1,2,3. Thread 1 is interrupting
thread 2 while thread 3 begins to Stop-The-World. Now given the
following ordered events:

-thread 1 suspends thread 2, which is in a gc unsafe area.

-thread 3 suspends thread 1 and 2.

-thread 3 tries to restart thread 2 since it is in a gc unsafe
area, but since its suspend count is 2, it fails to do so and
deadlock.

An alternative would be to make the suspended state not have
a counter, but that would make the STW code trickier as it would
need to restore the thread to its previous state and not simply
resume it.
This could work with sgen, but would require some surgery on
boehm, which is completely undesirable given how state it is.

13 years agoRevert previous XmlSerializerMessageContractImporter implementation.
Atsushi Eno [Wed, 11 May 2011 09:40:15 +0000 (18:40 +0900)]
Revert previous XmlSerializerMessageContractImporter implementation.

XsdDataContractImporter is not XmlSerializer-based, while this class is.

13 years agoAdd PipeException. fixed bug #692987 by Eberhard Beilharz.
Atsushi Eno [Wed, 11 May 2011 07:54:52 +0000 (16:54 +0900)]
Add PipeException. fixed bug #692987 by Eberhard Beilharz.

13 years agoFixed orders in XmlSchemaObjectTable and got some S.R.Serialization tests working.
Atsushi Eno [Wed, 11 May 2011 07:34:45 +0000 (16:34 +0900)]
Fixed orders in XmlSchemaObjectTable and got some S.R.Serialization tests working.

13 years agoImplement sync suspend based interruption
Rodrigo Kumpera [Tue, 10 May 2011 18:57:00 +0000 (15:57 -0300)]
Implement sync suspend based interruption

13 years agowork around full aot not working
Rodrigo Kumpera [Tue, 10 May 2011 17:28:39 +0000 (14:28 -0300)]
work around full aot not working

13 years agoRemove dead variable
Rodrigo Kumpera [Tue, 10 May 2011 17:28:21 +0000 (14:28 -0300)]
Remove dead variable

13 years agoMerge branch 'master' of github.com:mono/mono
Neale Ferguson [Tue, 10 May 2011 18:14:22 +0000 (14:14 -0400)]
Merge branch 'master' of github.com:mono/mono

13 years agoAdd MONO_ARCH_HAVE_CONTEXT_SET_INT_REG for s390x
Neale Ferguson [Tue, 10 May 2011 18:13:45 +0000 (14:13 -0400)]
Add MONO_ARCH_HAVE_CONTEXT_SET_INT_REG for s390x

13 years agoPass mono_install_handler_block_guard in the eh callback table
Rodrigo Kumpera [Tue, 10 May 2011 14:48:14 +0000 (11:48 -0300)]
Pass mono_install_handler_block_guard in the eh callback table

13 years agoCanImport() just returns false, without throwing error.
Atsushi Eno [Tue, 10 May 2011 13:02:20 +0000 (22:02 +0900)]
CanImport() just returns false, without throwing error.

13 years agosupport import of xs:anyType, as well as make couple of minor fixes.
Atsushi Eno [Tue, 10 May 2011 12:45:45 +0000 (21:45 +0900)]
support import of xs:anyType, as well as make couple of minor fixes.

13 years agoFix common issue in MetadataSet - call MoveToContent() while iterating nodes.
Atsushi Eno [Tue, 10 May 2011 10:58:17 +0000 (19:58 +0900)]
Fix common issue in MetadataSet - call MoveToContent() while iterating nodes.

13 years agoAdd new MetadataSetTest and fix couple of minor issues.
Atsushi Eno [Tue, 10 May 2011 10:33:52 +0000 (19:33 +0900)]
Add new MetadataSetTest and fix couple of minor issues.

13 years agoRevive this test by implementing mock web server.
Atsushi Eno [Tue, 10 May 2011 10:33:15 +0000 (19:33 +0900)]
Revive this test by implementing mock web server.

13 years agoAdd (not-working) callback instance context mode test and FIXME comment.
Atsushi Eno [Tue, 10 May 2011 10:32:25 +0000 (19:32 +0900)]
Add (not-working) callback instance context mode test and FIXME comment.

13 years agoUpdate XsdDataContractImporter.CanImport and implement ImportXmlType.
Atsushi Eno [Tue, 10 May 2011 08:40:35 +0000 (17:40 +0900)]
Update XsdDataContractImporter.CanImport and implement ImportXmlType.

13 years agoEnable async invoke on x86.
Rodrigo Kumpera [Tue, 10 May 2011 01:46:52 +0000 (22:46 -0300)]
Enable async invoke on x86.

* exceptions-x86.c (mono_arch_handle_exception): Extract
mono_arch_setup_async_callback which is used to setup
async calls.

* mini-x86.h: Enable async calls on x86.

13 years agoAdd (disabled) unit tests for sync use of CommunicationObject
Sebastien Pouliot [Mon, 9 May 2011 17:47:35 +0000 (13:47 -0400)]
Add (disabled) unit tests for sync use of CommunicationObject

* System.ServiceModel_test.dll.sources: Add new files to the tests
* Test/NUnitMoonHelper.cs: Helper to re-use Moonlight unit tests on NUnit
* Test/System.ServiceModel.Channels/CommunicationObjectSyncTest.cs: New.

13 years agoongoing XmlSerializerMessageContractImporter work.
Atsushi Eno [Mon, 9 May 2011 13:58:18 +0000 (22:58 +0900)]
ongoing XmlSerializerMessageContractImporter work.

It is part of bugfix #670945.

13 years agoMultiple schemas in wsdl could share same SourceUri, so check more identity.
Atsushi Eno [Mon, 9 May 2011 10:09:19 +0000 (19:09 +0900)]
Multiple schemas in wsdl could share same SourceUri, so check more identity.

Part of bugfix #670945.

13 years agoAllow xsi:nil for void. Fixed client side of bug #687580.
Atsushi Eno [Mon, 9 May 2011 04:50:38 +0000 (13:50 +0900)]
Allow xsi:nil for void. Fixed client side of bug #687580.

13 years agoNow our HttpListener can listen to wildcard prefix safely.
Atsushi Eno [Mon, 9 May 2011 04:40:16 +0000 (13:40 +0900)]
Now our HttpListener can listen to wildcard prefix safely.

Fixed server part of bug #687580.

13 years agoAvoid crash when setting the DialogResult in the FormClosing event method. Fixes...
Flose Maier [Sun, 8 May 2011 10:59:01 +0000 (12:59 +0200)]
Avoid crash when setting the DialogResult in the FormClosing event method. Fixes bug 686486.

13 years ago[utils] Fix lock-free data structures.
Mark Probst [Fri, 6 May 2011 23:39:00 +0000 (01:39 +0200)]
[utils] Fix lock-free data structures.

Hazard pointers didn't check for the third pointer.

Lots of memory barriers missing.

Small bug in LLS.

Comments and asserts.

13 years agoMore win32 fixes
Rodrigo Kumpera [Fri, 6 May 2011 19:43:17 +0000 (16:43 -0300)]
More win32 fixes

13 years agoMerge branch 'master' of github.com:mono/mono
Neale Ferguson [Fri, 6 May 2011 16:53:12 +0000 (12:53 -0400)]
Merge branch 'master' of github.com:mono/mono

13 years agoImplement soft debugger for s390x and fix context macro for s390x
Neale Ferguson [Fri, 6 May 2011 16:52:19 +0000 (12:52 -0400)]
Implement soft debugger for s390x and fix context macro for s390x

13 years ago[Web Services] Fix regression caused by previous commit
Lluis Sanchez [Fri, 6 May 2011 16:37:19 +0000 (18:37 +0200)]
[Web Services] Fix regression caused by previous commit

Fixes a regression caused by the implementation of support
for optional parameters. GlobalIndex was not being set
when using SOAP RPC protocol.

13 years agoUpdate S.X.S sources files to use S.XML sources
Sebastien Pouliot [Fri, 6 May 2011 12:40:11 +0000 (08:40 -0400)]
Update S.X.S sources files to use S.XML sources

13 years agoMake some S.X.S types compatible with the SL API (for SDK)
Sebastien Pouliot [Fri, 6 May 2011 12:02:21 +0000 (08:02 -0400)]
Make some S.X.S types compatible with the SL API (for SDK)

* XmlAnyElementAttributes.cs:
* XmlArrayItemAttributes.cs:
* XmlElementAttributes.cs:
Under Silverlight API this implement IList only (does not
inherit from CollectionBase).

* XmlCustomFormatter.cs:
* XmlSerializationWriterInterpreter.cs:
Use a Convert.ChangeType override that exists in SL API

* XmlReflectionImporter.cs: Provide an alternative for (missing in SL)
Enum.GetNames and select available (in SL) overloads for parsing and
formatting for enums

* XmlSerializationReader.cs: Fix ToEnum API to use IDictionary (in SL)
instead of Hashtable (even if the SL4 docs still states the later)

* XmlSerializationReaderInterpreter.cs:
* XmlTypeMapping.cs:
Use the default (identical to nonPublic being false) when using
Activator.CreateInstance under SL

* XmlSerializationWriter.cs: Internally use a Queue<T> instead a a Queue
(not available in SL). Also Namespaces (ArrayList) is replaced by
XmlNamespaces (IList) in SL API.

* XmlSerializer.cs: Do not use environment variables in MOONLIGHT (like
the rest of MOBILE). Use Encoding.UTF8 instead of ENcoding.Default since
the later does not exist in SL.

* XmlSerializerImplementation.cs: SL version use IDictionary instead of
Hashtable for properties.

* XmlSerializerNamespaces.cs: Internally use a generic Dictionary
* instead
of a ListDictionary (which is not available in SL)

13 years agoEnable some code to be included in Moonlight's S.X.S.dll SDK assembly
Sebastien Pouliot [Thu, 5 May 2011 23:23:52 +0000 (19:23 -0400)]
Enable some code to be included in Moonlight's S.X.S.dll SDK assembly

* System.Collections/ArrayList.cs:
* System.Collections/Hashtable.cs:
Do not include debugger-related attributes and some interfaces
when compiled outside corlib

* System.Runtime.Serialization/ObjectIDGenerator.cs: Ensure this can be
compiled outside corlib (needed for ML S.X.S.dll where even internals
cannot be used)

13 years agoFix the build for Windows (use int instead of ssize_t)
Jeffrey Stedfast [Fri, 6 May 2011 01:01:33 +0000 (21:01 -0400)]
Fix the build for Windows (use int instead of ssize_t)

13 years agoFixes for win32
Rodrigo Kumpera [Fri, 6 May 2011 00:33:03 +0000 (21:33 -0300)]
Fixes for win32

13 years agoFixed for Big Endian machines.
Jeffrey Stedfast [Thu, 5 May 2011 21:46:15 +0000 (17:46 -0400)]
Fixed for Big Endian machines.

13 years agoMake sure that implicit variant type conversion implies in a reference conversion.
Rodrigo Kumpera [Thu, 5 May 2011 19:25:22 +0000 (16:25 -0300)]
Make sure that implicit variant type conversion implies in a reference conversion.

* class.c (mono_class_is_variant_compatible): Now takes an aditional
parameter to tell if reference conversion should be checked.

* verify.c (verifier_class_is_assignable_from): Check that a variant
generic type convertion has a reference conversion.

Fixes #687193

13 years ago[ado.net]bugfix:688773 - Assign default value to columns that does not
Veerapuram Varadhan [Wed, 4 May 2011 21:58:53 +0000 (03:28 +0530)]
[ado.net]bugfix:688773 - Assign default value to columns that does not
exist in the row that is being imported.

13 years ago[ado.net]bugfix:681916 - Based on patch by Scott <srfcanada@hotmail.com>
Veerapuram Varadhan [Wed, 4 May 2011 20:17:50 +0000 (01:47 +0530)]
[ado.net]bugfix:681916 - Based on patch by Scott <srfcanada@hotmail.com>
- Handle null values for parameter types varbinary(max) and/or image

13 years ago[mono]bugfix:663287 - Validate the type of value against underlyingtype
Veerapuram Varadhan [Wed, 4 May 2011 18:57:42 +0000 (00:27 +0530)]
[mono]bugfix:663287 - Validate the type of value against underlyingtype
instead of nullableType.

13 years ago[ado.net]BugFix 663914 - Retain original version of a col when
Veerapuram Varadhan [Tue, 3 May 2011 18:04:25 +0000 (23:34 +0530)]
[ado.net]BugFix 663914 - Retain original version of a col when
(De)Serialization of DataSet and DataTables.

13 years ago[corlib] Do not cache disposed resource sets.
Marek Habersack [Wed, 4 May 2011 21:30:57 +0000 (23:30 +0200)]
[corlib] Do not cache disposed resource sets.

13 years ago[asp.net] Do not retrieve script's URL in RegisterScriptReference if script.Path...
Marek Habersack [Wed, 4 May 2011 21:27:29 +0000 (23:27 +0200)]
[asp.net] Do not retrieve script's URL in RegisterScriptReference if script.Path is set.

13 years agoMerge pull request #95 from XTZGZoReX/master.
Miguel de Icaza [Wed, 4 May 2011 14:40:03 +0000 (07:40 -0700)]
Merge pull request #95 from XTZGZoReX/master.

Fix spelling in Consts.cs.in

13 years ago[691531] Variance conversion requires type parameters to be reference types
Marek Safar [Wed, 4 May 2011 11:18:15 +0000 (12:18 +0100)]
[691531] Variance conversion requires type parameters to be reference types

13 years agoFix spelling in Consts.cs.in.
XTZGZoReX [Wed, 4 May 2011 10:49:03 +0000 (12:49 +0200)]
Fix spelling in Consts.cs.in.

13 years ago[json] revert part of 382a385a8f, it regresses moon drt 983
Andreia Gaita [Tue, 26 Apr 2011 18:58:44 +0000 (19:58 +0100)]
[json] revert part of 382a385a8f, it regresses moon drt 983

13 years agoMerge pull request #94 from XTZGZoReX/master.
Rodrigo Kumpera [Tue, 3 May 2011 20:40:10 +0000 (13:40 -0700)]
Merge pull request #94 from XTZGZoReX/master.

Remove dead NET_1_1 code in ILASM's Makefile

13 years agoilasm: Makefile: Remove dead NET_1_1 code.
XTZGZoReX [Tue, 3 May 2011 20:24:09 +0000 (22:24 +0200)]
ilasm: Makefile: Remove dead NET_1_1 code.

13 years agoAdd define to conditionally compile code using mono-context.
Rodrigo Kumpera [Tue, 3 May 2011 17:46:16 +0000 (14:46 -0300)]
Add define to conditionally compile code using mono-context.

13 years ago[sgen] Fix thread unregister deadlock race.
Mark Probst [Tue, 3 May 2011 13:54:02 +0000 (15:54 +0200)]
[sgen] Fix thread unregister deadlock race.

We had a race that happened now and then and led to a deadlock.  A
thread would be dying and at the same time the GC would kick in,
acquiring the GC lock.  The thread was still in the thread list, so
the GC would try to suspend it.  pthreads was already in the process
of destructing the thread's TLS, however, so its thread info was no
longer available and the thread could not reply to the suspend signal.
The thread's destructor, however, tried to acquire the GC lock, so no
progress would happen anymore.

The solution is to set a flag in the thread info indicating that the
GC wants the thread to suspend.  Instead of getting the GC lock in the
desctructor unconditionally we now just try to lock, and if we fail we
check the flag.  If it is set, we suspend the thread from within the
thread info destructor.

13 years ago[691417] Set array Packing/Size values without using custom attribute
Marek Safar [Tue, 3 May 2011 12:47:49 +0000 (13:47 +0100)]
[691417] Set array Packing/Size values without using custom attribute

13 years agoAdd socket timeouts to the soft debugger.
Martin Baulig [Wed, 27 Apr 2011 15:48:10 +0000 (17:48 +0200)]
Add socket timeouts to the soft debugger.

SDB now support setting a send/receive timeout on the client's side
of the communication socket.  To make this work, the server sends
a special 'keepalive' event when using receive timeouts.

There's a new method VirtualMachine.SetSocketTimeouts() to enable
this feature.

13 years agoAdd regression test for #691057
Rodrigo Kumpera [Tue, 3 May 2011 05:25:45 +0000 (02:25 -0300)]
Add regression test for #691057

13 years agoFix a couple of issues on Mono.Simd.
Rodrigo Kumpera [Tue, 3 May 2011 03:27:20 +0000 (00:27 -0300)]
Fix a couple of issues on Mono.Simd.

* local-propagation.c (mono_local_deadce): Don't perform
reverse copyprop on simd insert ops. The backend requires
dreg and sreg1 to match.

* mini-x86.c (mono_arch_output_basic_block): Fix insert_r8_slow
for the X element.

Fixes a few bugs due to #691057

13 years agoFix the ARM build.
Zoltan Varga [Tue, 3 May 2011 01:26:04 +0000 (03:26 +0200)]
Fix the ARM build.

13 years agoImplement BSTR icalls even when COM is disabled. Fixes #690255.
Zoltan Varga [Mon, 2 May 2011 23:02:27 +0000 (01:02 +0200)]
Implement BSTR icalls even when COM is disabled. Fixes #690255.

13 years agodon't init smr twice
Rodrigo Kumpera [Mon, 2 May 2011 21:33:30 +0000 (18:33 -0300)]
don't init smr twice

13 years agoFix resume_thread_internal return type
Rodrigo Kumpera [Mon, 2 May 2011 21:20:17 +0000 (18:20 -0300)]
Fix resume_thread_internal return type

13 years agoFix thread unregistration under sgen.
Rodrigo Kumpera [Mon, 2 May 2011 20:00:00 +0000 (17:00 -0300)]
Fix thread unregistration under sgen.

* mono-threads.c (unregister_thread): The only way to
remove the current thread from the thread list without
causing troubles to sgen is to do it with the gc lock
held, so from the unregister callback.

* sgen-gc.c (sgen_unregister_thread): Call into the
new function to remove the current thread from the thread
list.

13 years agoFix parameter name for Aes.Create
Sebastien Pouliot [Mon, 2 May 2011 16:18:43 +0000 (12:18 -0400)]
Fix parameter name for Aes.Create

* Aes.cs: Fix parameter name

13 years agoAdd new 4.0 method (stub) for X509Certificate
Sebastien Pouliot [Mon, 2 May 2011 16:17:11 +0000 (12:17 -0400)]
Add new 4.0 method (stub) for X509Certificate

* X509Certificate.cs: Add stub for FormatDate (4.0)

13 years agoMatch IDisposable with 4.0 changes in System.Security.Cryptography
Sebastien Pouliot [Mon, 2 May 2011 16:14:53 +0000 (12:14 -0400)]
Match IDisposable with 4.0 changes in System.Security.Cryptography

* AsymmetricAlgorithm.cs:
* CryptoAPITransform.cs:
* DeriveBytes.cs:
* FromBase64Transform.cs:
* HashAlgorithm.cs:
* PasswordDeriveBytes.cs:
* Rfc2898DeriveBytes.cs:
* RijndaelManagedTransform.cs:
* SymmetricAlgorithm.cs:
* ToBase64Transform.cs:
Adjust changes to IDisposable wrt 4.0

13 years agoAdd some new 4.0 features in System.Security.Cryptography
Sebastien Pouliot [Mon, 2 May 2011 16:13:23 +0000 (12:13 -0400)]
Add some new 4.0 features in System.Security.Cryptography

* CspProviderFlags.cs: Add new CreateEphemeralKey (128) value
* CryptoConfig.cs: Add 4.0 AllowOnlyFipsAlgorithms (always false) and
stubs for AddAlgorithm and AddOID
* CryptoStream.cs: Remove Close override (from 1.x days) and implement
HasFlushedFinalBlock (4.0)

13 years ago[runtime] Fix handling of ECMA key when parsing assembly name.
Marek Habersack [Mon, 2 May 2011 13:09:46 +0000 (15:09 +0200)]
[runtime] Fix handling of ECMA key when parsing assembly name.

This fixes mcs InternalsVisibleTo support for assemblies signed with the ECMA (ecma.pub)
key. Previously the key token would be returned as the assembly's key, which caused mcs
to fail to match friend assemblies.

13 years ago[mono-api-info] try to read local files before using the resolver
Jb Evain [Mon, 2 May 2011 13:38:16 +0000 (15:38 +0200)]
[mono-api-info] try to read local files before using the resolver

13 years agoFix the mobile build
Jb Evain [Mon, 2 May 2011 13:17:55 +0000 (15:17 +0200)]
Fix the mobile build

13 years ago[linker] Fix reading symbols of resolved assemblies
Jb Evain [Mon, 2 May 2011 11:57:36 +0000 (13:57 +0200)]
[linker] Fix reading symbols of resolved assemblies

13 years ago[linker] keep types in the module when sweeping them
Jb Evain [Mon, 2 May 2011 11:56:39 +0000 (13:56 +0200)]
[linker] keep types in the module when sweeping them

13 years agoFix non net_4_0 profiles build
Jb Evain [Mon, 2 May 2011 09:57:40 +0000 (11:57 +0200)]
Fix non net_4_0 profiles build

13 years agoProperly check the maxstacksize asked for the Thread
Jb Evain [Mon, 2 May 2011 09:39:21 +0000 (11:39 +0200)]
Properly check the maxstacksize asked for the Thread

13 years agoAdd test for Thread.ctor where the stacksize is not divisible by the system page...
Jb Evain [Mon, 2 May 2011 09:38:05 +0000 (11:38 +0200)]
Add test for Thread.ctor where the stacksize is not divisible by the system page size

13 years agoFix support for parameters with generic constraints in Type.IsAssignableFrom ()....
Zoltan Varga [Mon, 2 May 2011 07:39:49 +0000 (09:39 +0200)]
Fix support for parameters with generic constraints in Type.IsAssignableFrom (). Fixes #691119.

13 years agoMerge pull request #90 from nulltoken/patch-1.
Miguel de Icaza [Sun, 1 May 2011 21:59:29 +0000 (14:59 -0700)]
Merge pull request #90 from nulltoken/patch-1.

[XBuild) Make RemoveDir task able to recursively delete the content of a directory

13 years agoUpdate SGen sources to SGen license
Miguel de Icaza [Sun, 1 May 2011 21:18:47 +0000 (17:18 -0400)]
Update SGen sources to SGen license

13 years agoMerge branch 'sgen-lock-free-alloc'
Mark Probst [Sun, 1 May 2011 18:50:34 +0000 (20:50 +0200)]
Merge branch 'sgen-lock-free-alloc'

13 years ago[utils] Document the lock-free allocator.
Mark Probst [Sun, 1 May 2011 18:45:50 +0000 (20:45 +0200)]
[utils] Document the lock-free allocator.

13 years ago[sgen] Make sure remset clearing doesn't crash if we change thread unregistering.
Mark Probst [Sat, 23 Apr 2011 15:50:43 +0000 (17:50 +0200)]
[sgen] Make sure remset clearing doesn't crash if we change thread unregistering.

13 years ago[sgen] Attach thread info early so that hazard pointers are available.
Mark Probst [Sat, 23 Apr 2011 16:19:36 +0000 (18:19 +0200)]
[sgen] Attach thread info early so that hazard pointers are available.

13 years ago[sgen] Use the lock free allocator as the internal allocator for SGen.
Mark Probst [Sat, 23 Apr 2011 10:53:59 +0000 (12:53 +0200)]
[sgen] Use the lock free allocator as the internal allocator for SGen.

13 years ago[utils] Added lock-free queue and allocator.
Mark Probst [Fri, 22 Apr 2011 19:24:28 +0000 (21:24 +0200)]
[utils] Added lock-free queue and allocator.

13 years ago[eglib] make g_mkdir_with_parents work for paths not ending in /
Chris Toshok [Sun, 1 May 2011 06:11:32 +0000 (23:11 -0700)]
[eglib] make g_mkdir_with_parents work for paths not ending in /

13 years agoAvoid doubly declaring some types in the 2.1 profile for mobile
Geoff Norton [Sat, 30 Apr 2011 15:48:44 +0000 (11:48 -0400)]
Avoid doubly declaring some types in the 2.1 profile for mobile

13 years agoFix prefix selection
Gonzalo Paniagua Javier [Sat, 30 Apr 2011 09:33:00 +0000 (11:33 +0200)]
Fix prefix selection

We were picking the wrong prefix when the URL matched the path.
Fixes bug #688860.

13 years agoSupport FTP + RETR + proxy
Gonzalo Paniagua Javier [Sat, 30 Apr 2011 09:07:56 +0000 (11:07 +0200)]
Support FTP + RETR + proxy

Fixes bug #690887.

13 years agoDon't return values in functions returning void
Jeffrey Stedfast [Sat, 30 Apr 2011 00:25:58 +0000 (20:25 -0400)]
Don't return values in functions returning void

13 years agoIntroduce a couple of functions to abstract old and new interrupt machinery.
Rodrigo Kumpera [Fri, 29 Apr 2011 20:47:24 +0000 (17:47 -0300)]
Introduce a couple of functions to abstract old and new interrupt machinery.

* threads.c: Move code around to make it easier to have both
old and new machinery around.

13 years ago[boehm] Avoid a ENOMEM when allocating across an unallocated page
Geoff Norton [Fri, 29 Apr 2011 20:23:15 +0000 (16:23 -0400)]
[boehm] Avoid a ENOMEM when allocating across an unallocated page

A very rare allocation pattern could cause the boehm free list to call
the GC_unmap_gap function, which actually calls down into munmap().
Darwins virtual memory manager will return a KERN_INVALID_ADDRESS
which is translated into a ENOMEM, if mprotect attempts to change
the protection of a range which includes an unallocated page.
We address this by just mmap() back to ANON, instead of actually
unallocating the page.

13 years agoMake SMR available during the whole registration/unregister.
Rodrigo Kumpera [Fri, 29 Apr 2011 20:01:23 +0000 (17:01 -0300)]
Make SMR available during the whole registration/unregister.

* hazard-pointer.c (mono_hazard_pointer_get): Use
mono_thread_info_get_small_id to get the small id
instead of reading it from the thread info since it
might be gone during unregister.

* mono-threads.c: Add new TLS key to to hold the small
id so we can have it available during unregister. This is
needed since pthreads cleans up the thread_info TLS before
calling it's cleanup function which, in turn, renders SMR
unavailable.

The new lock-free allocator requires it so we need to make
SMR available for all runtime callbacks.

13 years agoMake thread unregister more robust.
Rodrigo Kumpera [Fri, 29 Apr 2011 15:59:49 +0000 (12:59 -0300)]
Make thread unregister more robust.

* mono-threads.c (unregister_thread): Remove
the dying thread from the thread list before
doing the gc callback. This avoid some potential
trouble if a GC happens in the meanwhile.

13 years agoAllow external memory management of nodes.
Rodrigo Kumpera [Fri, 29 Apr 2011 15:58:25 +0000 (12:58 -0300)]
Allow external memory management of nodes.

* mono-linked-list-set.c (mono_lls_init): If
free_node_func is null, the caller of remove is reponsible
for reclaiming memory.

13 years ago[GCHandle] Refactor so == and != do not box unnecessarily.
Alan McGovern [Thu, 28 Apr 2011 10:14:55 +0000 (11:14 +0100)]
[GCHandle] Refactor so == and != do not box unnecessarily.

13 years ago[webclient] Process uri.UserInfo when present
Gonzalo Paniagua Javier [Fri, 29 Apr 2011 09:53:19 +0000 (11:53 +0200)]
[webclient] Process uri.UserInfo when present

If there are no crendetials, the Uri might have them.
Based on a patch by QuickJack.
Fixes bug #688007.

13 years agoAvoid a needless buffer copy that is susceptible to overflow
Geoff Norton [Fri, 29 Apr 2011 03:05:15 +0000 (23:05 -0400)]
Avoid a needless buffer copy that is susceptible to overflow

13 years ago[eglib] add some more MAX/MIN macros
Chris Toshok [Thu, 28 Apr 2011 22:07:51 +0000 (15:07 -0700)]
[eglib] add some more MAX/MIN macros

13 years agoFix a typo for Rodrigo
Geoff Norton [Thu, 28 Apr 2011 22:07:17 +0000 (18:07 -0400)]
Fix a typo for Rodrigo

13 years agoFix loading from volatile base addresses in the LLVM backend. Fixes #690020.
Zoltan Varga [Thu, 28 Apr 2011 21:45:43 +0000 (23:45 +0200)]
Fix loading from volatile base addresses in the LLVM backend. Fixes #690020.