mono.git
10 years ago[MSBuild] Fix the ${level} and ${number} capture.
Jonathan Pryor [Tue, 24 Sep 2013 20:16:32 +0000 (16:16 -0400)]
[MSBuild] Fix the ${level} and ${number} capture.

Fixes: https://bugzilla.xamarin.com/show_bug.cgi?id=14767

Some people, when confronted with a problem, think
“I know, I'll use regular expressions.”   Now they have two problems.
- http://regex.info/blog/2006-09-15/247

ToolTask uses a regular expression to extract various bits of
information from the warning/error message. Unfortunately it was
inadequate, and didn't correctly match on some errors.

The error message in this case is:

class1.cs(16,4): error CS0152: The label `case 1:' already occurs in this switch statement.

The problem was that while the previous regular expression would
succeed in matching the above string, it was horribly inaccurate in
what it captured:

${file}   = "error CS0152"
${line}   = "16"
${column} = "4"
${level}  = "The"
${number} = "label `case 1"
${message}= "' already occurs in this switch statement."

Did I mention hilariously wrong?

There were two problems with the previous pattern:

 1. It allowed filename+(line,column) to be _repeated_, which is
    clearly insane. Filename/etc. shouldn't be repeated.

 2. ${number} captured ".*\d", i.e. anything ending with a number, so
    when the message had a number embedded within it, it was captured.

For good measure, use RegexOptions.IgnorePatternWhitespace so that
we don't have a horrendously long regex pattern to compare for diffs
the next time this needs fixing, and add NUnit tests.

10 years ago[sgen] Fix a thread startup deadlock on POSIX.
Mark Probst [Tue, 24 Sep 2013 17:55:22 +0000 (10:55 -0700)]
[sgen] Fix a thread startup deadlock on POSIX.

If the GC stops the world and a thread is in the process of starting
up, but has not yet registered (it's not in the thread_list), it is
possible that the thread does register while the world is stopped.
When restarting the GC will then try to restart said thread, but since
it never got the suspend signal, it cannot answer the restart signal,
so a deadlock results.

This is triggered by `tests/sgen-bridge.cs`, about once every 10 runs.

We solve this by using a negative value in the `signal` field in the
thread info to indicate threads that have never been suspended.

10 years agoBetter checks for inaccessible member used via dynamic proxy. Fixes #14762
Marek Safar [Tue, 24 Sep 2013 17:17:06 +0000 (19:17 +0200)]
Better checks for inaccessible member used via dynamic proxy. Fixes #14762

10 years agoReenable accidentally disabled test
Marek Safar [Tue, 24 Sep 2013 14:04:07 +0000 (16:04 +0200)]
Reenable accidentally disabled test

10 years agoFixes compilation error on Mavericks
Marek Safar [Tue, 24 Sep 2013 14:00:00 +0000 (16:00 +0200)]
Fixes compilation error on Mavericks

10 years agoInitialize cookie container for default requests. Fixes #14783
Marek Safar [Tue, 24 Sep 2013 13:56:45 +0000 (15:56 +0200)]
Initialize cookie container for default requests. Fixes #14783

10 years agoOptmizes one string comparison.
Marek Safar [Tue, 24 Sep 2013 13:53:25 +0000 (15:53 +0200)]
Optmizes one string comparison.
Remove useless GC.SuppressFinalize

10 years agoAllow workitems to startup faster and reduce polling thread interval in the threadpool.
Paolo Molaro [Tue, 24 Sep 2013 13:25:53 +0000 (15:25 +0200)]
Allow workitems to startup faster and reduce polling thread interval in the threadpool.

Deals with xambug #14644 and similar issues.

10 years agoUpdate cecil sources to latest
Sebastien Pouliot [Tue, 24 Sep 2013 12:13:05 +0000 (08:13 -0400)]
Update cecil sources to latest

10 years ago[linker] Resolve every type references if we're to save an assembly that is not linke...
Sebastien Pouliot [Sat, 21 Sep 2013 02:26:54 +0000 (22:26 -0400)]
[linker] Resolve every type references if we're to save an assembly that is not linked (so the scope won't point to assemblies that were removed)

10 years agoCheck modifying members of readonly value-type field. Fixes #14802
Marek Safar [Tue, 24 Sep 2013 07:44:44 +0000 (09:44 +0200)]
Check modifying members of readonly value-type field. Fixes #14802

10 years agoFix the windows build.
Zoltan Varga [Tue, 24 Sep 2013 04:11:50 +0000 (06:11 +0200)]
Fix the windows build.

10 years agoFix the windows build.
Rodrigo Kumpera [Mon, 23 Sep 2013 23:14:05 +0000 (19:14 -0400)]
Fix the windows build.

10 years agoAdd test case for xref behavior for the sgen bridge.
Rodrigo Kumpera [Mon, 23 Sep 2013 23:07:42 +0000 (19:07 -0400)]
Add test case for xref behavior for the sgen bridge.

10 years agoMake sure we properly account for transitive liveness when processing bridge objects...
Rodrigo Kumpera [Mon, 23 Sep 2013 22:51:34 +0000 (18:51 -0400)]
Make sure we properly account for transitive liveness when processing bridge objects. Fixes #13858

We depend on the latent liveness state of objects at the time we collect bridged objects. To
preserve this we must perform the dfs1 step right after collecting them and before we drain the
gray stack or all objects pointed by candidate bridge objects will be marked as alive.

10 years agoWrap the code in MONO_BEGIN/END_DECLS
Miguel de Icaza [Mon, 23 Sep 2013 17:03:05 +0000 (13:03 -0400)]
Wrap the code in MONO_BEGIN/END_DECLS

10 years agoUpdate message
Miguel de Icaza [Mon, 16 Sep 2013 20:28:29 +0000 (16:28 -0400)]
Update message

10 years agoPass cancellation token to CopyAsync read block
Marek Safar [Sun, 22 Sep 2013 17:23:17 +0000 (19:23 +0200)]
Pass cancellation token to CopyAsync read block

10 years agoImplements content loading cancellation. Fixes #13200
Marek Safar [Sun, 22 Sep 2013 17:15:08 +0000 (19:15 +0200)]
Implements content loading cancellation. Fixes #13200

10 years agoTest update
Marek Safar [Sun, 22 Sep 2013 15:43:41 +0000 (17:43 +0200)]
Test update

10 years agoBump ikvm-reflection
Marek Safar [Sun, 22 Sep 2013 15:42:47 +0000 (17:42 +0200)]
Bump ikvm-reflection

10 years ago[runtime] Ignore --llvm when LLVM is not enabled in configure.
Zoltan Varga [Sat, 21 Sep 2013 12:56:58 +0000 (14:56 +0200)]
[runtime] Ignore --llvm when LLVM is not enabled in configure.

10 years agoAdd test for native thiscall methods.
Vincent Povirk [Mon, 22 Apr 2013 21:06:10 +0000 (16:06 -0500)]
Add test for native thiscall methods.

Licensed under MIT/X11.

10 years agoAdd support for thiscall pinvokes on x86.
Vincent Povirk [Fri, 26 Jul 2013 20:50:11 +0000 (15:50 -0500)]
Add support for thiscall pinvokes on x86.

Licensed under MIT/X11.

10 years ago[aot] Pass correct architecture to llc on x86-64.
Zoltan Varga [Sat, 21 Sep 2013 03:03:48 +0000 (05:03 +0200)]
[aot] Pass correct architecture to llc on x86-64.

10 years agoMake the test-aot test less verbose.
Zoltan Varga [Sat, 21 Sep 2013 00:45:09 +0000 (02:45 +0200)]
Make the test-aot test less verbose.

10 years agoMerge pull request #765 from knocte/runtime_fixage
Zoltan Varga [Fri, 20 Sep 2013 22:51:35 +0000 (15:51 -0700)]
Merge pull request #765 from knocte/runtime_fixage

[sgen] Improve fix for BXC#14834 to avoid redundant code

10 years ago[sgen] Improve fix for BXC#14834 to avoid redundant code
Andrés G. Aragoneses [Fri, 20 Sep 2013 20:15:42 +0000 (22:15 +0200)]
[sgen] Improve fix for BXC#14834 to avoid redundant code

Improve https://github.com/mono/mono/commit/d2cc22580898df5d4a15e0f99ab513e1570a6082
by making OBJ_BITMAP_FOREACH be less redundant thanks to new macro
GNUC_BUILTIN_CTZ

10 years agoFixed TaskFactory::ContinueWhenAny with func result. Fixes #14839
Marek Safar [Fri, 20 Sep 2013 17:39:12 +0000 (19:39 +0200)]
Fixed TaskFactory::ContinueWhenAny with func result. Fixes #14839

10 years ago[sgen] Use __builtin_ctzl () in OBJ_BITMAP_FOREACH_PTR () on 64 bit platforms. Fixes...
Zoltan Varga [Fri, 20 Sep 2013 17:06:34 +0000 (19:06 +0200)]
[sgen] Use __builtin_ctzl () in OBJ_BITMAP_FOREACH_PTR () on 64 bit platforms. Fixes #14834.

10 years agoFix the build.
Mark Probst [Fri, 20 Sep 2013 16:34:44 +0000 (09:34 -0700)]
Fix the build.

10 years agoAggregateException::GetBaseException uses only 1-based inner exceptions. Fixes #14824
Marek Safar [Fri, 20 Sep 2013 15:59:28 +0000 (17:59 +0200)]
AggregateException::GetBaseException uses only 1-based inner exceptions. Fixes #14824

10 years agoMerge pull request #760 from mayerwin/patch-2
Marek Safar [Fri, 20 Sep 2013 15:39:51 +0000 (08:39 -0700)]
Merge pull request #760 from mayerwin/patch-2

Fix: bug https://bugzilla.xamarin.com/show_bug.cgi?id=14811 (to be reviewed)

10 years ago[xbuild]: Actually add the $(TargetFrameworkMonikerAssemblyAttributesPath) to @(Compile).
Martin Baulig [Thu, 19 Sep 2013 22:56:30 +0000 (00:56 +0200)]
[xbuild]: Actually add the $(TargetFrameworkMonikerAssemblyAttributesPath) to @(Compile).

10 years agoMerge pull request #758 from echampet/capabilities
Marek Habersack [Fri, 20 Sep 2013 12:21:52 +0000 (05:21 -0700)]
Merge pull request #758 from echampet/capabilities

Fix crashes with browser capabilities

10 years ago[System.Web] Re-enable AppBrowsersTest.cs tests for net_2_0
Etienne CHAMPETIER [Wed, 18 Sep 2013 11:37:45 +0000 (13:37 +0200)]
[System.Web] Re-enable AppBrowsersTest.cs tests for net_2_0

For now mcs/class/System.Web/Test/mainsoft/NunitWebResources/adapters.browser
depends on System.Web_test_net_2_0.dll

Signed-off-by: Etienne CHAMPETIER <etienne.champetier@fiducial.net>
10 years ago[Install] Add Compat.browser files for net_4_0 and net_4_5
Etienne CHAMPETIER [Thu, 19 Sep 2013 11:48:06 +0000 (13:48 +0200)]
[Install] Add Compat.browser files for net_4_0 and net_4_5

Without that referencing browser 'Default' (parentID="Default")
in a .browser file make the app throw exceptions here
mcs/class/System.Web/System.Web.Configuration_2.0/nBrowser/Build.cs#L206

Signed-off-by: Etienne CHAMPETIER <etienne.champetier@fiducial.net>
10 years ago[System.Web] Do not crash on duplicate keys in browser capabilities
Frederic Mestayer [Mon, 16 Sep 2013 16:18:04 +0000 (18:18 +0200)]
[System.Web] Do not crash on duplicate keys in browser capabilities

Correct https://bugzilla.xamarin.com/show_bug.cgi?id=14720 and
probably https://bugzilla.xamarin.com/show_bug.cgi?id=13014

Signed-off-by: Etienne CHAMPETIER <etienne.champetier@fiducial.net>
Signed-off-by: Frederic Mestayer <f.mestayer@fiducial.net>
10 years ago[System.Net] Fix CFNetworkCopyProxiesForAutoConfigurationScript signature. Fixes...
Rolf Bjarne Kvinge [Fri, 20 Sep 2013 08:33:24 +0000 (10:33 +0200)]
[System.Net] Fix CFNetworkCopyProxiesForAutoConfigurationScript signature. Fixes #7923.

10 years ago[System.Net.Http] Add a way to add extra build arguments.
Rolf Bjarne Kvinge [Thu, 19 Sep 2013 22:18:22 +0000 (00:18 +0200)]
[System.Net.Http] Add a way to add extra build arguments.

This is used by monotouch to add a reference to monotouch.dll.

Also add a profile-specific sources file for monotouch so the
extensions module can add even more files.

10 years agoMerge pull request #763 from madewokherd/mingwbuild
Rodrigo Kumpera [Thu, 19 Sep 2013 22:05:15 +0000 (15:05 -0700)]
Merge pull request #763 from madewokherd/mingwbuild

Fix the mingw-w64 build

10 years agoFix undeclared variables in exceptions-amd64.c on win64.
Vincent Povirk [Thu, 19 Sep 2013 21:16:04 +0000 (16:16 -0500)]
Fix undeclared variables in exceptions-amd64.c on win64.

This commit is licensed under MIT/X11.

10 years agoTest for InterlockedCompareExchange64 in the configure script.
Vincent Povirk [Thu, 19 Sep 2013 21:14:52 +0000 (16:14 -0500)]
Test for InterlockedCompareExchange64 in the configure script.

mingw.org doesn't have this function but mingw-w64 does.

This commit is licensed under MIT/X11.

10 years agoDisable the LLVM version check for now, it doesn't work if llvm was not compiled...
Zoltan Varga [Thu, 19 Sep 2013 20:33:16 +0000 (22:33 +0200)]
Disable the LLVM version check for now, it doesn't work if llvm was not compiled in a git repo.

10 years agoCheck the version of LLVM at configure time to make sure it has a matching version.
Zoltan Varga [Thu, 19 Sep 2013 20:04:49 +0000 (22:04 +0200)]
Check the version of LLVM at configure time to make sure it has a matching version.

10 years agoAdd test-aot target.
Rodrigo Kumpera [Thu, 19 Sep 2013 20:03:12 +0000 (16:03 -0400)]
Add test-aot target.

10 years ago[runtime]Fix typo in detection of the need of 64bits atomic intrinsics.
Rodrigo Kumpera [Thu, 19 Sep 2013 19:13:49 +0000 (15:13 -0400)]
[runtime]Fix typo in detection of the need of 64bits atomic intrinsics.

10 years agoChange Random implementation to JKISS which is faster and has a longer period. Also...
Sebastien Pouliot [Thu, 19 Sep 2013 19:28:40 +0000 (15:28 -0400)]
Change Random implementation to JKISS which is faster and has a longer period. Also match .NET 2.0 behaviour wrt calls to protected Sample. Add unit tests (and removed the commented ones). Ref: www0.cs.ucl.ac.uk/staff/d.jones/GoodPracticeRNG.pdf

10 years ago[runtime] Improve error checking for temporary file creation in the AOT compiler.
Zoltan Varga [Thu, 19 Sep 2013 19:23:24 +0000 (21:23 +0200)]
[runtime] Improve error checking for temporary file creation in the AOT compiler.

10 years agoRevert "Add System.Net.Http.dll to the monotouch build."
Rolf Bjarne Kvinge [Thu, 19 Sep 2013 17:47:29 +0000 (19:47 +0200)]
Revert "Add System.Net.Http.dll to the monotouch build."

This reverts commit 9df6eebc346a066811e26128b9b500a2cd46170c.

Not quite ready for this yet, more changes are needed.

10 years agoAdd System.Net.Http.dll to the monotouch build.
Rolf Bjarne Kvinge [Thu, 19 Sep 2013 16:50:34 +0000 (18:50 +0200)]
Add System.Net.Http.dll to the monotouch build.

10 years ago[llvm] On OSX we need to search in /Libraries in addition to /lib.
Rodrigo Kumpera [Thu, 19 Sep 2013 16:50:27 +0000 (12:50 -0400)]
[llvm] On OSX we need to search in /Libraries in addition to /lib.

10 years ago[runtime] Fix a bug in the local register allocator when an fp vreg wasn't clobbered...
Zoltan Varga [Wed, 18 Sep 2013 22:19:41 +0000 (00:19 +0200)]
[runtime] Fix a bug in the local register allocator when an fp vreg wasn't clobbered by a call if its href happened to be equal to the hreg of the call instruction. Fixes #12510.

10 years agoFix the Makefile addition of a new test.
Rodrigo Kumpera [Wed, 18 Sep 2013 20:21:49 +0000 (16:21 -0400)]
Fix the Makefile addition of a new test.

10 years ago[runtime]Do an alignment check before using the atomic op.
Rodrigo Kumpera [Wed, 18 Sep 2013 20:12:00 +0000 (16:12 -0400)]
[runtime]Do an alignment check before using the atomic op.

10 years agoFix the DISABLE_JIT build.
Zoltan Varga [Wed, 18 Sep 2013 20:09:23 +0000 (22:09 +0200)]
Fix the DISABLE_JIT build.

10 years agoAdd a sgen test for threads joining/leaving at random places.
Rodrigo Kumpera [Tue, 17 Sep 2013 23:25:27 +0000 (19:25 -0400)]
Add a sgen test for threads joining/leaving at random places.

10 years ago[sgen] Kill SgenThreadInfo::thread_is_dying in favor of the thread-info state informa...
Rodrigo Kumpera [Tue, 17 Sep 2013 23:23:08 +0000 (19:23 -0400)]
[sgen] Kill SgenThreadInfo::thread_is_dying in favor of the thread-info state information.

10 years ago[runtime]Properly set mono-thread state as when starting/running/shutting down/dead.
Rodrigo Kumpera [Tue, 17 Sep 2013 23:16:03 +0000 (19:16 -0400)]
[runtime]Properly set mono-thread state as when starting/running/shutting down/dead.

10 years ago[sgen]Remove doing_handshake field since the race condition it fixes is now handled...
Rodrigo Kumpera [Tue, 17 Sep 2013 22:24:10 +0000 (18:24 -0400)]
[sgen]Remove doing_handshake field since the race condition it fixes is now handled by mono-threads.

10 years ago[io-layer] Fix a warning.
Rodrigo Kumpera [Tue, 17 Sep 2013 21:04:53 +0000 (17:04 -0400)]
[io-layer] Fix a warning.

10 years ago[runtime] Small optimizations and cleanup in ves_icall_System_Array_FastCopy.
Rodrigo Kumpera [Tue, 17 Sep 2013 21:04:11 +0000 (17:04 -0400)]
[runtime] Small optimizations and cleanup in ves_icall_System_Array_FastCopy.

10 years ago[runtime]Add more fast variants of the mono_array functions as those shown up in...
Rodrigo Kumpera [Tue, 17 Sep 2013 21:01:30 +0000 (17:01 -0400)]
[runtime]Add more fast variants of the mono_array functions as those shown up in profiling.

10 years ago[sgen] Force inlining of some perf-critical functions.
Rodrigo Kumpera [Tue, 17 Sep 2013 20:59:26 +0000 (16:59 -0400)]
[sgen] Force inlining of some perf-critical functions.

10 years ago[sgen]We no longer need joined_stw as new threads can't join once the suspend lock...
Rodrigo Kumpera [Tue, 17 Sep 2013 20:43:09 +0000 (16:43 -0400)]
[sgen]We no longer need joined_stw as new threads can't join once the suspend lock has been acquired.

Since no new thread can join until the collection is over, FOREACH_THREAD won't miss any thread.

10 years ago[runtime]Provide a consistent view of registered threads during suspend operations.
Rodrigo Kumpera [Tue, 17 Sep 2013 19:29:54 +0000 (15:29 -0400)]
[runtime]Provide a consistent view of registered threads during suspend operations.

We previously failed to provide a consistent view of the registered threads set. This
would should up in the following ways:

-New threads would join in between two FOREACH_THREAD passes. The recently joined threads
lead to broken invariants that one loop carried to the other.

-Existing threads would leave during a FOREACH_THREAD. The now defunct threads lead
to code waiting them to ack back and deadlock in the process.

-It was possible to witness an in-progress cleanup. This would show up by
mono_thread_info_current returning null when it should not.

To amend this we provide a a consistent view of all registered threads once the global
suspend lock is acquired. This is done by acquiring that lock when registering/unregistering.

This allows us to remove some of the work-around in the sgen thread registration
code and avoid taking the GC lock.

10 years ago[sgen] Don't stack overflow when there are too many binary protocol buffers.
Mark Probst [Tue, 17 Sep 2013 23:07:16 +0000 (16:07 -0700)]
[sgen] Don't stack overflow when there are too many binary protocol buffers.

Binary protocol buffers are linearly linked from newest to oldest, but need to be
written out from oldest to newest.  This was sloppily implemented with a recursive
call, which caused a stack overflow when there were too many buffers.  We now
allocate an array and store pointers to all the buffers in it, then iterate over it in
reverse.

10 years agoFix: bug https://bugzilla.xamarin.com/show_bug.cgi?id=14811 (to be reviewed)
mayerwin [Tue, 17 Sep 2013 17:14:04 +0000 (01:14 +0800)]
Fix: bug https://bugzilla.xamarin.com/show_bug.cgi?id=14811 (to be reviewed)

10 years agoMerge pull request #756 from akoeplinger/fix-xmlreader
Atsushi Eno [Tue, 17 Sep 2013 17:09:28 +0000 (10:09 -0700)]
Merge pull request #756 from akoeplinger/fix-xmlreader

Fixing inconsistency in XmlSchemaValidatingReader between Mono and MS.NET

10 years agoMerge pull request #754 from blackfusion/master
Rodrigo Kumpera [Tue, 17 Sep 2013 13:25:00 +0000 (06:25 -0700)]
Merge pull request #754 from blackfusion/master

[Scripts] Fix a Shebang in mono-test-install for freeBSD

10 years ago[runtime] Fix a regression introduced by 55039913729e26c839e45ff05f17de1f2aa1c751...
Zoltan Varga [Tue, 17 Sep 2013 11:41:28 +0000 (13:41 +0200)]
[runtime] Fix a regression introduced by 55039913729e26c839e45ff05f17de1f2aa1c751. Use __builtin_popcountl () on 64 bit machines.

10 years agoDisable a profiler test which fails on osx.
Zoltan Varga [Mon, 16 Sep 2013 21:26:24 +0000 (23:26 +0200)]
Disable a profiler test which fails on osx.

10 years agoFix the profiler tests.
Zoltan Varga [Mon, 16 Sep 2013 21:09:56 +0000 (23:09 +0200)]
Fix the profiler tests.

10 years agoRevert "Set tls->context.valid to FALSE if a thread is suspended in native code....
Zoltan Varga [Mon, 16 Sep 2013 18:45:26 +0000 (20:45 +0200)]
Revert "Set tls->context.valid to FALSE if a thread is suspended in native code. Fixes #1540."

This reverts commit 5ddbe31ee5c42f5e379cb077e3473667791bc15a.

Revert this as its incorrect, and causes a regression (#14058).

10 years ago[mdoc] Switched from swallowing error to guarded execution
Larry O'Brien [Mon, 16 Sep 2013 17:19:00 +0000 (07:19 -1000)]
[mdoc] Switched from swallowing error to guarded execution

Improves a60ff0fd1e181cfde5b862cc491c682c11b9660a

Tests for file existence and outputs diagnostic rather than treating it as a local recoverable exception.

Leaving the question of general mdoc error-reporting for another day.

10 years ago[runtime] Avoid constructing vtables in a few places during AOT.
Zoltan Varga [Mon, 16 Sep 2013 13:02:06 +0000 (15:02 +0200)]
[runtime] Avoid constructing vtables in a few places during AOT.

10 years ago[runtime] Avoid calling mono_resolve_patch_target () while AOTing, its not needed.
Zoltan Varga [Sun, 15 Sep 2013 23:44:02 +0000 (01:44 +0200)]
[runtime] Avoid calling mono_resolve_patch_target () while AOTing, its not needed.

10 years ago[runtime] Add support for constrained gsharedvt calls to normal virtual methods.
Zoltan Varga [Sun, 15 Sep 2013 23:21:06 +0000 (01:21 +0200)]
[runtime] Add support for constrained gsharedvt calls to normal virtual methods.

10 years ago[runtime] Avoid looking up pinvoke addresses when AOT compiling.
Zoltan Varga [Sun, 15 Sep 2013 16:52:38 +0000 (18:52 +0200)]
[runtime] Avoid looking up pinvoke addresses when AOT compiling.

10 years agoBugzilla 14672 (mdoc namespace-level summary and remarks elements weren't being included)
Larry O'Brien [Sun, 15 Sep 2013 21:12:27 +0000 (11:12 -1000)]
Bugzilla 14672 (mdoc namespace-level summary and remarks elements weren't being included)

commit 6ec0ddc7e5c4d5640da7420600898c10578a7974
Author: Larry O'Brien <lobrien@knowing.net>
Date:   Sun Sep 15 11:03:32 2013 -1000

    Coding style

commit f034514fca8298a8ed2a167f3ab288f32bc7c74f
Author: Larry O'Brien <lobrien@knowing.net>
Date:   Sun Sep 15 10:28:47 2013 -1000

    Bugzilla 14672

    Added namespace file ("ns-{namespace}.xml") summary and remarks into the nsSummaries, which results in displaying the data prior to the namespace type-list when the user searches at the namespace level.

    Appears to only be necessary for native viewer (Web viewer for some reason "knows" about the summaries).

10 years ago[runtime] Fix a few more places which accessed jinfo->method directly.
Zoltan Varga [Sun, 15 Sep 2013 01:21:44 +0000 (03:21 +0200)]
[runtime] Fix a few more places which accessed jinfo->method directly.

10 years agoFix a problem introduced by c6989ab9f905ae27773140c8bf8c8e27ca4a6a1e. Avoid foreach...
Zoltan Varga [Sun, 15 Sep 2013 00:42:28 +0000 (02:42 +0200)]
Fix a problem introduced by c6989ab9f905ae27773140c8bf8c8e27ca4a6a1e. Avoid foreach on _items since that includes empty entries too.

10 years ago[sdb] Simplify tests.
Zoltan Varga [Sat, 14 Sep 2013 23:51:25 +0000 (01:51 +0200)]
[sdb] Simplify tests.

10 years agoUpdate AOT after changes to SeqPoint in 488888ce612491aca1c97f79ddbad685302d43cc.
Zoltan Varga [Sat, 14 Sep 2013 23:39:05 +0000 (01:39 +0200)]
Update AOT after changes to SeqPoint in 488888ce612491aca1c97f79ddbad685302d43cc.

10 years ago[sdb] Add an implicit seq point after every nonvoid call and make step over skip...
Zoltan Varga [Sat, 14 Sep 2013 23:36:38 +0000 (01:36 +0200)]
[sdb] Add an implicit seq point after every nonvoid call and make step over skip them. Fixes #10782.

10 years agoFix the ENABLE_JIT_MAP build. Fixes #14752.
Zoltan Varga [Sat, 14 Sep 2013 19:10:02 +0000 (21:10 +0200)]
Fix the ENABLE_JIT_MAP build. Fixes #14752.

10 years ago[runtime] Make mono_install_handler_block_guard () async-safe. Fixes #14073.
Zoltan Varga [Tue, 10 Sep 2013 14:39:46 +0000 (16:39 +0200)]
[runtime] Make mono_install_handler_block_guard () async-safe. Fixes #14073.

10 years ago[runtime] Make the stack walking code async safe when run in an async context.
Zoltan Varga [Tue, 10 Sep 2013 14:28:54 +0000 (16:28 +0200)]
[runtime] Make the stack walking code async safe when run in an async context.

10 years ago[runtime] Make mono_aot_find_jit_info () async-safe by allocating memory from the...
Zoltan Varga [Tue, 10 Sep 2013 14:26:37 +0000 (16:26 +0200)]
[runtime] Make mono_aot_find_jit_info () async-safe by allocating memory from the lock-free domain mempool, avoiding loading metadata, and caching jit info entries in a lock-free array.

10 years ago[runtime] Add a 'is_async_context' property to MonoThreadInfo, allowing functions...
Zoltan Varga [Tue, 10 Sep 2013 14:12:47 +0000 (16:12 +0200)]
[runtime] Add a 'is_async_context' property to MonoThreadInfo, allowing functions to detect whenever they are in async context without having to pass an 'async' parameter around.

10 years ago[runtime] Add a lock-free mempool allocator to MonoDomain.
Zoltan Varga [Tue, 10 Sep 2013 14:07:19 +0000 (16:07 +0200)]
[runtime] Add a lock-free mempool allocator to MonoDomain.

10 years agoMove dynamic handling out of Binary::DoResolve to reduce stack size
Marek Safar [Fri, 13 Sep 2013 16:55:09 +0000 (18:55 +0200)]
Move dynamic handling out of Binary::DoResolve to reduce stack size

10 years agoAdd new test
Marek Safar [Fri, 13 Sep 2013 15:49:49 +0000 (17:49 +0200)]
Add new test

10 years agoFix a warning.
Zoltan Varga [Fri, 13 Sep 2013 16:46:26 +0000 (18:46 +0200)]
Fix a warning.

10 years ago[runtime] Make the address->aot module lookup in mono_jit_info_table_find () async...
Zoltan Varga [Tue, 3 Sep 2013 03:04:58 +0000 (05:04 +0200)]
[runtime] Make the address->aot module lookup in mono_jit_info_table_find () async-safe by storing the mapping in another MonoJitInfoTable.

10 years agoExtract the jit info table code into separate functions so we it can be used in more...
Zoltan Varga [Tue, 3 Sep 2013 00:47:56 +0000 (02:47 +0200)]
Extract the jit info table code into separate functions so we it can be used in more places.

10 years ago[runtime] Avoid direct access to MonoJitInfo.method, use an accessor function instead.
Zoltan Varga [Tue, 10 Sep 2013 14:00:19 +0000 (16:00 +0200)]
[runtime] Avoid direct access to MonoJitInfo.method, use an accessor function instead.

10 years ago[sgen] Fix a toggleref array resize bug.
Mark Probst [Fri, 13 Sep 2013 16:21:32 +0000 (09:21 -0700)]
[sgen] Fix a toggleref array resize bug.

10 years agoUTF8Encoding.GetMaxByteCount() was returning too big number.
Atsushi Eno [Fri, 13 Sep 2013 15:56:48 +0000 (00:56 +0900)]
UTF8Encoding.GetMaxByteCount() was returning too big number.

The thing is, unlike the standards describe in codepoints, .NET String
characters beyond BMP are represented as surrogate pairs and they can be
converted all in 3 bytes.

The implementation didn't take care of encoder fallbacks, or byte order
marks either, so take them into considertion too.

10 years agoDon't crash when unreachable code contains unreachable constant switch. Fixes #14587
Marek Safar [Fri, 13 Sep 2013 14:52:30 +0000 (16:52 +0200)]
Don't crash when unreachable code contains unreachable constant switch. Fixes #14587