* [merp] Put thread into async context before running summarizer
[merp] Put thread into async context before running summarizer
Followup work for https://github.com/mono/mono/pull/17090
In order to get managed method info for unmanaged stacktraces, we call
`mini_jit_info_table_find_ext` which calls `decode_exception_debug_info` which may
allocate if it's not in async context.
Do the switch in `mono_threads_summarize_execute` so that non-supervising threads
all switch when they run the `sigterm_signal_handler`. `mono_threads_summarize`
already does it for the supervisor thread.
Fixes https://github.com/mono/mono/issues/17180
Backport of #17194.
/cc @lambdageek
* [merp] Don't call install_state_summarizer in EnableMicrosoftTelemetry
install_state_summarizer installs a SIGTERM handler. Unfortunately
sigterm_signal_handler assumes that it will only be called when a crash is
already in progress (it assumes it won't be the supervisor, and so it doesn't
setup the memory for capturing a crash report among other things).
* [merp] Register SIGTERM handler in dump_native_stacktrace
Previously it was installed by EnableMicrosoftTelemetry, which is too early.
Now we get here without a handler installed even when mono_merp_enabled returns true.
There is an issue with jumping into DST for some time zones when the incorrect date-time offset is returned for date-time in UTC (which comes from DateTime.Now). The fix is to just check if the incoming date-time is in UTC
Also I added set of tests for some time zones but they verify jumping into DST in general.
Fixes https://github.com/mono/mono/issues/16395
Backport of https://github.com/mono/mono/pull/16430
* [2019-06] Bump msbuild and sdk versions to 3.0.1xx latest
* Bump msbuild again to pull in the build updates
* pull in roslyn-binaries fix
* One more bump
* Fix the roslyn hash
* [msbuild] Bump to pick up a build fix from https://github.com/mono/msbuild/pull/138
* Pull in final update from msbuild of nuget and sdk
NuGet 5.3.0-rtm.6251
Sdk 3.0.100-rc2.19467.3
* Pick up packaging fix in https://github.com/mono/msbuild/pull/141
We sometimes saw the following build error in Mono SDKs builds on Jenkins:
```
[2019-09-16T01:03:30.709Z] CSC [xammac] nunit-lite-console.exe
[2019-09-16T01:03:30.709Z] error CS0009: Metadata file '/Users/builder/jenkins/workspace/archive-mono/2019-08/mac/release/mcs/class/lib/xammac/nunitlite.dll' could not be opened -- PE image doesn't contain managed metadata.
```
Taking a closer look we found out that nunit-lite was trying to be built twice for the same profile which caused a race where the file was corrupted.
The reason for this is that the corlib Makefile defines two test assemblies: the normal PROFILE_corlib_test.dll and BinarySerializationOverVersionsTest.dll and
integrating the BinarySerializationOverVersionsTest.dll via the `test-vts` target incorrectly overwrote the `test` target instead of using `test-local`.
Backport of #16894.
Callers of mono_reflection_create_custom_attr_data_args_noalloc were leaking some of the returned information. Accessed attributes are FixedBufferAttribute and UnmanagedFunctionPointerAttribute.
Add RenamedEvent* to FSW sources from CoreFX
Fixes#16486 - This was only partially fixed until now
* [csproj] Update project files
* Add test files which test these code paths on corefx
* Revert "[aot] partial revert of #14043 (#14135)"
This reverts commit 223e0bd78d.
* Revert "[aot] change method table generation so that it works for larger assemblies on ARM (#14043)"
This reverts commit 9ff3b0d65e.
* Use the right FileSystemWatcher sources
The corefx code for FileSystemEventArgs has some changes to support
constructing one with a null filename.
Fixes#16486
* Remove the mono sources for the newly replaced files
* [csproj] Update project files
Resolves https://github.com/mono/mono/issues/16122
Since we put in the corefx version of GZipStream, it is expected that
our version of DeflateStream actually implements Write+ReadAsyncMemory.
This change does that by just calling the plain WriteAsync & ReadAsync
methods until we decide (or not) to pull in the corefx DeflateStream.
Backport of https://github.com/mono/mono/pull/16523.
* [w32socket] Translate some errno codes to WSA_ errors imprecisely.
These aren't a perfect match, but better than falling into the `default` case
where Mono asserts and crashes.
Addresses part of https://github.com/mono/mono/issues/16024
* [w32error] Add WSA_INVALID_PARAMETER and WSA_INVALID_HANDLE
Backport of #16331.
As of XCode 11 beta6 the MacOSX SDK defines the MAP_32BIT symbol.
This causes the `mono_valloc()` function to try to use the `MAP_32BIT` flag for `mmap()`.
However as mentioned in the comment for the symbol in mman.h it seems to require a special entitlement which isn't available/documented anywhere yet.
This in turn causes the mmap call to fail presumably because we're missing that entitlement.
Instead we now skip setting this flag on Apple platforms to make `mmap()` behave like it did before.
Backport of https://github.com/mono/mono/pull/16441
* Bump msbuild to track mono-2019-06
.. to get changes from https://github.com/mono/msbuild/pull/130
which tracks dotnet `release/3.0.100-preview9`.
Also, bump Roslyn to 3.3.1-beta3-19421-04 .
* Bump the roslyn reference
Using the CoreCLR version of this line fixed the problem and my investigations
also revealed why.
The `SafePipeHandle` contains a reference to both the Socket as well as the
`SafeSocketHandle`, taking ownership of them. If there are no additional
references to either of them, so that both are collected in the same GC pass,
their finalizer may be invoked in any order.
If the Socket gets finalized first, it's `SafeSocketHandle` will have a `_mstate`
of 8 (two references) at the beginning of `DangerousReleaseInternal()` - which
will set it to 6 (one reference plus disposed flag).
Then when the `SafePipeHandle` gets collected, it will call `DangerousRelease ()`
on it's `SafeSocketHandle`. As mentioned above, it's `_mstate` will be 6.
Now we own the handle, so we'll decrease it's reference count by one and set the
`Closed` flag - the new `_mstate` should thus be 3.
However, using
new_state = (old_state & RefCount_Mask) - RefCount_One;
would set it to 1 - thus effectively clearing the `Disposed' flag. And since it's
reference count is now also zero, the next finalize pass will then throw.
The CoreCLR version of this class uses
new_state = old_state - RefCount_One;
which fixes this.
Backport of #16244.
The intention of calling `g_file_test (file, G_FILE_TEST_EXISTS)` where file is
the name of a shared library we want to open is to speed up probing for
non-existent libraries.
See https://github.com/mono/mono/pull/12074
The problem is that if file is just a simple "libdl.so" then `dlopen (file)`
doesn't just look for it in the current working directory, it will probe some
other paths too. (For example on desktop linux you'd also look in all the
directories in LD_LIBRARY_PATH). So the g_file_test() call is not a robust way
to avoid calling dlopen if the filename is relative.
But it actually broke more things: https://github.com/xamarin/xamarin-android/issues/3388
When probing for "libdl.so" on Android mono_lookup_pinvoke_call will first try
prepending some paths that it knows about and we end up calling
`dlopen ("/system/lib/libdl.so")` which will fail because Bionic has security
restrictions on what code can dlopen something from /system/lib with an
absolute path. Eventually mono_lookup_pinvoke_call will go back to trying the
bare "libdl.so" which hits `g_file_test` and returns NULL.
The new code only does the file test if we pass it an absolute path, which
gives Bionic's dlopen a chance to deal with relative paths however it needs to.
* Add missing `Dispose()` call on error in `MonoTlsStream.CreateStream()`.
* Remove the `finally` block to make it more easy to read.
Backport of #16233.
Fixes https://github.com/mono/mono/issues/16032
When we switched to corefx File.Copy on unix, the behavior changed where it opened a source and destination stream in order to pass the safe handles to Interop.Sys.CopyFile. This exposed a flaw in our sharing validation rules in MonoIO.Open, which threw an exception in this scenario:
new FileStream("source.txt", FileMode.OpenOrCreate, FileAccess.ReadWrite, FileShare.ReadWrite, 2048, FileOptions.DeleteOnClose);
new FileStream("source.txt", FileMode.Open, FileAccess.Read, FileShare.Read, 2048, FileOptions.None);
Simplified the FileShare check down to:
1. If the caller specifies FileShare.None, then it's an exclusive lock and no
sharing is allowed.
2. If the caller specifies some other FileShare option, then FileShare.None
cannot be specified.
This will allow the interpreter to use System.Reflection.Emit.
Note: There is a small size increase (about 50kb) for an application
built in release mode with the LLVM backend (even if the interpreter
is now used).
Backport of https://github.com/mono/mono/pull/16151
* Fixed RuntimeType.GetConstructorCandidates("") to return 0 elements.
https://github.com/mono/mono/issues/16010
The current behavior for GetConstructorCandidates when the name is null, "",
or "*" is to return the set of constructors for the type. For "", .NET Core and
.NET Framework return 0 elements. This PR adds a check for "" to match that
behavior.
* Removed unncessary changes
* Unnecessary spaces
* Made the if check a bit more readable
Prompted by mono/msbuild#123 :
This includes:
Update to upstream vs16.3 branch
SDK update to track dotnet release/3.0.100-preview8
Roslyn updated to 3.3.0-beta2-19381-14
* Work around remoting1.exe test issue with new Roslyn
* [marshal] Always use gchandles in delegate_hash_table
Makes the code easier to follow and it also fixes race from https://github.com/mono/mono/commit/caa4a753ca8e15d43baaa01adb0f56f374b74a2b with boehm.
* [marshal] Free delegates with target that are passed to native code.
For static method delegates, we have a unique delegate_trampoline that is shared among all delegates. We always keep alive the first static method delegate passed to native, by creating a normal gchandle to it and storing it in delegate_hash_table. For instance methods, each delegate will create a separate wrapper and these wrappers are never shared (which was wrongly assumed in https://github.com/mono/mono/commit/caa4a753ca8e15d43baaa01adb0f56f374b74a2b). We shuldn't keep the delegate alive and this commit reverts the behavior for delegate with instance methods introduced in the mentioned commit.
Fixes https://github.com/mono/mono/issues/15751
Backport of #15935.
We need to clearly distinguish between synchronous and asynchronous completions
and avoid any ambiguities and race conditions when it comes to error handling.
On synchronous completion, we must not invoke the async completion delegate.
In `System.Net.Sockets/Socket.cs`, the internal `BeginSConnect()` and `BeginMConnect()`
now return a `bool` indicating whether or not an async operation is pending.
If any error happens prior to starting the async operation, then we shall call
`SocketAsyncResult.Complete(..., true)` and return `false`. And since this is
a synchronous completion, it's `AsyncCallback` will not be invoked.
In `referencesource/System/net/System/Net/Sockets/_MultipleConnectAsync.cs`,
I compared the entire file against the CoreFX version and then copied the
`AttemptConnection()` method from their implementation to adapt their behavior.
The difference is that on synchronous completion, the `InternalConnectCallback(null, args)`
should be invoked.
Backport of #15947.
[2019-06] [threads] Enter GC Unsafe before pumping HP queue in unregister_thread
If the thread info TLS key dtor runs `unregister_thread` from a foreign thread and we have work in the hazard pointer queue, we need to first switch to GC Unsafe mode because some of the free methods passed to `mono_thread_hazardous_try_free` need to be coop-aware.
Also, added checked mode assertions that `conc_table_free` and `jit_info_table_free` are called from GC Unsafe mode.
Fixes https://github.com/mono/mono/issues/15878
Backport of #15890.
/cc @lambdageek
[2019-06][sre] Make creation of dynamic method synchronized
Before, reflection_create_dynamic_method could end up being called by two threads simultaneously. At the end of the method the ilgen field of the MonoReflectionDynamicMethod object is cleared, to enable its collection since it is no longer needed. The problem with this is that it can confuse the other thread compiling the dynamic method, making it see that there is no body.
We could either bother making this clearing work with multiple threads (and I'm not 100% convinced that the rest of the code is thread safe), disable the clearing of ilgen (which would make us use more memory) or go with the simple approach of locking on this DynamicMethod so only one thread is compiling it at one time.
Backport of https://github.com/mono/mono/pull/15803
The C# version doesn't work on 64bit. It would require to update its dependencies, but Zoltan already added a Python version for WebAssembly which works fine.
Some notes:
* changed the semantics of `--xcode-path` in order to match the output of `xcode-select -p`
* switched to python3/pip3
* removed hack for arm64_32, it works properly now because the Python version uses stock `clang`
* removed some leftover usages of `XCODE32_DIR`.
Manually verified by comparing output with the C# version.
This will unblock the iOS team when building on Catalina/Xcode11, as they can't consume SDK archives there. See xamarin/xamarin-macios#6603 (comment)
Contributes to #9621
master (#15744)
In some jobs we use throttle() logic which would mean only one job can execute on master, blocking other jobs in the meantime.
We don't really care which node executes this so let Jenkins pick one.
(cherry picked from commit 7784f20a26)
Before the code
(a, b) = (b, a);
was the equivalent of
a = b;
b = a;
now, it evaluates it to so something like
tmp1 = a;
tmp2 = b;
a = tmp2;
b = tmp1;
Signed-off-by: Bijan Tabatabai <[email protected]>
This is to handle the following example. The issue is that in Check() we have
a constrained call where the called method is IAdder`1<!!U>::PlusPlus() which
is a default interface method, and we need to determine at runtime whether to
box the this argument.
```
using System;
public interface IAdder<T> {
int Add (int x);
int PlusPlus () {
return Add (1);
}
}
interface IGen3<T> { }
struct Adder<T> : IAdder<IGen3<T>> {
int _field;
public int Add (int x) {
_field = x + _field;
return _field;
}
}
public class P {
public static int Check<T, U>(T t) where T : IAdder<U> {
return t.PlusPlus () + t.PlusPlus ();
}
public static void Main () {
var x = new Adder<object> ();
int y = Check<Adder<object>, IGen3<object>> (x);
Console.WriteLine ("expect 2, got {0}", y);
}
}
```
1) Copied changes from corefx for System.Data (also mono/corefx#310)
2) New Tracing api is left empty (because the current API is empty too)
3) New AES types throw PNSE
Bump netstandard.dll AssemblyVersion to 2.1 (fixes#13412)
Backport of #15696.
This reverts commit d732cef5fc.
The github commit status setter step requires a node context so we now just use on on master instead.
(cherry picked from commit 0509ceccb6)
For some reason when using a separate file and loading the function
from there Jenkins still requires us to have a node() context when
calling the function which is not what we want.
In preparation for notarizing our .pkg with Apple we need to bump the timeout
since it can take up to an hour for Apple's server to return a result.
(cherry picked from commit 7480830b36)
* The example that reproduces the bug does this: var thread = vm.GetThreads()[0]; greetingValue = program.InvokeMethod(thread, greeting, new Value[0]) as StringMirror;
But sometimes the Thread[0] was the MainThread of the program and it works, and other times was the GC Finalizer, but when it tries to run something on GC Finalizer thread, the thread is not suspended and throws an exception.
In the fix I removed the Finalizer thread of the list that is returned when CMD_VM_ALL_THREADS is called.
Fix#13311
* The example that reproduces the bug does this: var thread = vm.GetThreads()[0]; greetingValue = program.InvokeMethod(thread, greeting, new Value[0]) as StringMirror;
But sometimes the Thread[0] was the MainThread of the program and it works, and other times was the GC Finalizer, but when it tries to run something on GC Finalizer thread, the thread is not suspended and throws an exception.
In the fix I removed the Finalizer thread if ti's not executing a managed code of the list that is returned when CMD_VM_ALL_THREADS is called.
Fix#13311
* Changing what was suggested by Aleksey.
* Removing unused attribute.
Fixes#15366 and #15306.
- Fixes NULL termination of `StringBuilder` to UTF-16 marshalling. Additionally an extra NULL is added as defense-in-depth protection for incorrect handling in native calls. This extra NULL adds protection when trying to marshal the NULL-terminated string back and the native function doesn't null-terminate at the size of the buffer.
- Fixes marshalling back from UTF-16/UTF-8 buffers into chunked `StringBuilder`.
- Fixes missing `mono_marshal_free` call causing a memory leak when marshalling `StringBuilder` to UTF-16 and back.
- Skips marshalling back to `StringBuilder` when the parameter doesn't have `[Out]` attribute.
Unit tests are still TBD. Not sure where to put them and how to structure them. Any help would be appreciated. /cc @vargaz @EgorBo
Backport of #15392.
* Fixes https://github.com/mono/mono/issues/14839
The local timezone name was not known and defaulted to "Local".
This change adds an sdk function that returns the local name.
Backport of #15667.
* Permit DYLD mangling, so p/invokes work from non-system paths
* Add com.apple.security.cs.disable-library-validation
This is required for the .dylibs used in p/invokes to work.
* Use adhoc signing
(cherry picked from commit 202dca075b)
Android NDK r20 contains a header file (`ioctl.h`) with this code:
int ioctl(int __fd, int __request, ...);
#if !defined(BIONIC_IOCTL_NO_SIGNEDNESS_OVERLOAD)
/* enable_if(1) just exists to break overloading ties. */
int ioctl(int __fd, unsigned __request, ...) __overloadable __enable_if(1, "") __RENAME(ioctl);
#endif
The `_RENAME` macro is defined as `__asm__(x)` and serves the purpose of
replacing an existing symbol with the new one, decorated with the keyword.
Unfortunately, Mono's `offsets-tool` doesn't parse the header properly as
`CppSharp` doesn't understand `__asm__`:
Processing triple: armv7-none-linux-androideabi
Error parsing '/Users/rodo/git/xa2/external/mono/mono/metadata/metadata-cross-helpers.c'
/Users/rodo/android-toolchain/ndk/sysroot/usr/include/bits/ioctl.h(60,5): error: conflicting types for 'ioctl'
Error parsing '/Users/rodo/git/xa2/external/mono/mono/mini/mini-cross-helpers.c'
/Users/rodo/android-toolchain/ndk/sysroot/usr/include/bits/ioctl.h(60,5): error: conflicting types for 'ioctl'
make: *** [/Users/rodo/git/xa2/external/mono/sdks/builds/android-cross-arm-release/armv7-linux-android.h] Error 1
The workaround, implemented in this commit, is to define the
`BIONIC_IOCTL_NO_SIGNEDNESS_OVERLOAD` macro thus removing the "conflicting"
declaration from the picture, which changes nothing from the offset tool's point
of view and lets us process the headers in NDK r20 properly.
* Add TypeForwards for:
* System.IO.Enumeration.FileSystemName
* System.Security.Cryptography.CryptographicOperations
Since these were removed in NS 2.1
* Bump API snapshot submodule
[System]: Allow reading fully buffered `HttpWebResponse` body after close.
If we have already read the entire response body, then we should allow the
user to read it even after the request has been aborted.
* `WebConnectionStream.Read ()`: Move `Operation.ThrowIfClosedOrDisposed ()`
call down after argument checks and call a new protected abstract method
called `TryReadFromBufferedContent ()` prior to it.
* `WebResponseStream.TryReadFromBufferedContent ()`: Implement it here.
If the entire response body has been buffered, then `bufferedEntireContent`
is true and `innerStream is BufferedReadStream` and we can call the new
`BufferedReadStream.TryReadFromBuffer ()`.
* `WebResponseStream.ReadAllAsync ()`: Set `bufferedEntireContent = true`
after reading the entire response body.
* `BufferedReadStream.TryReadFromBuffer ()`: New internal method.
Fixes#10645.
Backport of #15220.
* [sdks] Use @loader_path for mono and libmonosgen
This way we are able to dynamically load libmonosgen and also use host
mono binary.
The library dependencies look like this after the change:
```
otool -L /Users/rodo/git/mono/sdks/out/android-host-Darwin-release/bin/mono
/Users/rodo/git/mono/sdks/out/android-host-Darwin-release/bin/mono:
@loader_path/libmonosgen-2.0.dylib (compatibility version 2.0.0, current version 2.0.0)
...
otool -L /Users/rodo/git/mono/sdks/out/android-host-Darwin-release/lib/libmonosgen-2.0.dylib
/Users/rodo/git/mono/sdks/out/android-host-Darwin-release/lib/libmonosgen-2.0.dylib:
@loader_path/libmonosgen-2.0.dylib (compatibility version 2.0.0, current version 2.0.0)
...
```
This test was checking a broken Mono-specific behavior, which has been
removed after we switched to the new corefx-based handler. It also does
not make much sense to use a `Transfer-Encoding` header with a GET request.
In managed, these functions are used as `static external IntPtr`. This means that previously on arm64 the top bits of the return value were garbage. A comparison with 64-bit -1 like in LinuxNetworkChange.EnsureSocket would never be true, which was causing us to hit other assertions in the runtime.
Co-authored-by: Aleksey Kliger <[email protected]>
* Time zone DST out fix
* Add unit tests for Europe/Vatican, Iran and Europe/Guernsey time zones
* Disable the tests on WINAOT; fixed Iran time zone id
CADMessages are passed directly between appdomains. When using contained data, we need to make sure we don't create any cross domain references.
When domains are unloaded, all threads executing code in that appdomain are aborted and the entire heap is traversed and objects in that appdomain are cleared. If any objects in other domains are directly referencing objects in the unloaded domain, they will reference junk. If any, cross domain refs should only be allowed when the referenced object is in the root domain which is never unloaded.
We have seen sporadic failures in thread pool tests on CI related to
test case:
System.Threading.ThreadPools.Tests.ThreadPoolTests.QueueRegisterPositiveAndFlowTest [FAIL]
Assert.Equal() Failure
Expected: 0
Actual: 1
Stack Trace:
at System.Threading.ThreadPools.Tests.ThreadPoolTests.QueueRegisterPositiveAndFlowTest () [0x00143] in D:\j\workspace\z3\label\w64\external\corefx\src\System.Threading.ThreadPool\tests\ThreadPoolTests.cs:337
at (wrapper managed-to-native) System.Reflection.RuntimeMethodInfo.InternalInvoke(System.Reflection.RuntimeMethodInfo,object,object[],System.Exception&)
at System.Reflection.RuntimeMethodInfo.Invoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) [0x0006a] in D:\j\workspace\z3\label\w64\mcs\class\corlib\System.Reflection\RuntimeMethodInfo.cs:395
Finally managed to track down the issue to a race when using the same final
event in sequential calls to RegisterWaitForSingleObject. The event could, depending on
timing, get signaled multiple times for the same RegisterWaitHandle instance
causing a thread waiting for completion on another RegisterWaitHandle using same event
to be signaled before that callback was completed. That will cause the race seen in the
failing test above.
Fix make sure we only signal the final event one time per RegisterWaitHandle
instance.
* [jit] Throw exception if casting fails for special array interface
After the standard interface check (mini_emit_iface_cast), for special interfaces we need to do additional checks. We were skipping these checks if the object was not an array, without throwing a cast exception. This code path is reached for configurations where remoting is disabled.
Fixes https://github.com/mono/mono/issues/14729
* [tests] Add test for interface casting case
This test would have failed on fullaot lane for example.
We recently switched the AOT compiler for iOS armv7 from 32bit host to 64bit host. At the same time we _also_ switched from LLVM 3.6 to the newer LLVM 6.0 fork we support.
In a Xamarin.iOS Release build for iOS armv7 exception handling would crash. Specifically the problem was that the exception object wasn't properly injected into the handler block. The way it works is, that the exception handler machinery passes the pointer to the exception object via `r0`.
With the newer LLVM the generated code of the exception handling block
looks like this:
```
LBB1_15: @ %EH_CLAUSE0_BB3
Ltmp8:
ldr r0, [sp, #24]
str r0, [sp, #4]
ldr r0, [sp, #28]
ldr r0, [sp, #12]
ldr r0, [r0]
cmp r0, #0
beq LBB1_17
```
thus overwriting `r0`, so that the pointer to the exception object is lost. With the older LLVM 3.6 the generated code looks like this:
```
LBB1_8: @ %EH_CLAUSE0_BB3
Ltmp8:
str r0, [sp, #4]
ldr r0, [r6]
cmp r0, #0
beq LBB1_10
```
correctly storing the exception object into a stack slot for later usage.
After some time I figured out that there are _different_ exception handling models. Depending on the model, the exception pointer is passed in `r0` or not:
https://github.com/mono/llvm/blob/2bd2f1db1803f7b36687e5abf912c69baa848305/lib/Target/ARM/ARMISelLowering.cpp#L14416-L14428
`SjLj` stands for "SetJump / LongJump" and means that exception handling is implemented with this. On iOS this is the default model that is used. We want `dwarf` instead, so we tell this `llc` now and the generated code handles it correctly.
So why was it working before? In our older LLVM 3.6 fork we hardcoded it:
https://github.com/mono/llvm/blob/f80899cb3eb75f7f5640b4519e83bd96991bffb8/lib/Target/ARM/ARMISelLowering.cpp#L756-L762
The `-exception-model=` option is also not available in LLVM 3.6.
Contributes to https://github.com/mono/mono/issues/15058 and https://github.com/mono/mono/issues/9621
Homebrew Git 2.22.0 misparses the submodule command and passes arguments like --hard and -xdff to git-submodule instead of to git-reset or git-clean. Passing the entire command as a single argument seems to help.
Git mailing list bug report: https://public-inbox.org/git/[email protected]/
* When the catch found is in a MONO_WRAPPER_RUNTIME_INVOKE the exception should be treated as an UNHANDLED by the debugger.
* Adding unit test.
* trying to fix side effect.
* I've tested the MonoTests.DebuggerTests.UnhandledException_2 test on VsWin using .net Framework and I think the test is wrong because the first break is in throw and the next break will be on breakpoint. In our implementation we were ignoring the unhandled exception and stopping only on breakpoint.
Alias analysis attempts to remove indirection when using OP_LDADDR (which fetches the address of a variable). For example :
ldaddr r1 <- r0
load_mem r2 <- r1
is equivalent to
mov r2 <- r0
The remembered r1 alias needs to be forgotten in case r1 is changed. We were handling this for direct assignment with MOV, but not for the more subtle case where an alias to r1 would be passed to a call, which could lead to changing r1.
https://github.com/mono/mono/issues/14872
We have several issues (e.g. https://github.com/mono/mono/issues/14181) where the installer occasionally fails in the postinstall script for unknown causes, on some hosts.
This seems to be sometimes the culpable section and this patch should help trace the issue or alleviate the common symptoms of it (see https://github.com/mono/mono/issues/14181)
Xamarin.Mac can call back here from a native thread that isn't attached to the
runtime. SGen gchandle code is lockfree, and the profiler event can fire
without involving Mono.
Some profiler events (for example gc_handle_deleted) could be raised on threads
that aren't attached to the runtime. Since the thread won't participate in
Mono's stop the world (since the runtime can't see it), there's no need to do
the transition to GC Safe while the thread is trying to grab the profiler
buffer lock.
* Implementing reabstraction of default interface methods.
* Fixing side effect on unit tests. Tested with another example and changed the implementation: using System; public interface I1 { void M1(); /*void M2(){ System.Console.WriteLine("I1.M2"); }*/ }
public interface I2 : I1
{
abstract void I1.M1();
/* void M2(){
System.Console.WriteLine("I2.M2");
}*/
}
class Test1 : I2
{
static void Main()
{
I1 i1 = new Test1();
i1.M1();
//i1.M2();
}
void I1.M1()
{
System.Console.WriteLine("Test1.M1");
}
}
* Fixing side effect on unit tests.
* Fixing side effect on unit tests.
* Implementing unit test. Fixing side effect.
* Fix unit test.
* Fixing reabstraction.il coreclr test.
* Changing comments.
* Changing things pointed by Aleksey.
* Changing things pointed by Aleksey. Implementing generics test.
* Changing things pointed by Aleksey.
The thread_stopped profiler event can be raised by the thread_info_key_dtor tls
key destructor when the thread is already doesn't have a domain set. In that
case, don't call process_profiler_event since it cannot handle a thread with
null TLS values.
Addresses https://github.com/xamarin/xamarin-android/issues/2920
with the following stack trace
```
* thread #20, name = 'Filter', stop reason = signal SIGSEGV: invalid address (fault address: 0xbc)
* frame #0: libmonosgen-2.0.so`mono_class_vtable_checked(domain=0x0000000000000000, klass=0x0000007200230648, error=0x00000071e92f9178) at object.c:1890
frame #1: libmonosgen-2.0.so`get_current_thread_ptr_for_domain(domain=0x0000000000000000, thread=0x00000071ebfec508) at threads.c:595
frame #2: libmonosgen-2.0.so`mono_thread_current at threads.c:1939
frame #3: libmonosgen-2.0.so`process_event(event=<unavailable>, arg=<unavailable>, il_offset=<unavailable>, ctx=<unavailable>, events=<unavailable>, suspend_policy=<unavailable>) at debugger-agent.c:3715
frame #4: libmonosgen-2.0.so`thread_end [inlined] process_profiler_event(event=EVENT_KIND_THREAD_DEATH, arg=0x00000071ebfec508) at debugger-agent.c:3875
frame #5: libmonosgen-2.0.so`thread_end(prof=<unavailable>, tid=<unavailable>) at debugger-agent.c:3991
frame #6: libmonosgen-2.0.so`mono_profiler_raise_thread_stopped(tid=<unavailable>) at profiler-events.h:105
frame #7: libmonosgen-2.0.so`mono_thread_detach_internal(thread=<unavailable>) at threads.c:979
frame #8: libmonosgen-2.0.so`thread_detach(info=0x00000071e949a000) at threads.c:3215
frame #9: libmonosgen-2.0.so`unregister_thread(arg=<unavailable>) at mono-threads.c:544
frame #10: libmonosgen-2.0.so`thread_info_key_dtor(arg=0x00000071e949a000) at mono-threads.c:774
frame #11: 0x00000072899c58e8 libc.so`pthread_key_clean_all() + 124
frame #12: 0x00000072899c5374 libc.so`pthread_exit + 76
frame #13: 0x00000072899c5264 libc.so`__pthread_start(void*) + 44
frame #14: 0x000000728996617c libc.so`__start_thread + 72
```
This is unfortunate.
The general pattern of:
if initialized
use the data
else:
initialize the data, possibly with lots of locks and barriers
mark initialized
is racy, because "use the data" can be scheduled ahead of "if initialized".
"Barriers come in pairs" generally, and this was missing one.
It depends somewhat. If the data is all pointers, initialized in the else path,
from null to non-null, and use includes dereferencing, which is a common
but not universal case, then it is ok on all non-Alpha processors, due
to "data dependency".
But if the data includes reading globals, then there is a race.
This is a bit of a slow down on fast paths on arm, and possibly
other architectures.
There are barrier-free ways to solve this, involving a thread local,
that should seriously be considered, and applied throughout the runtime.
The runtime has a lot of on-demand initialization and a lot of looks racy.
See www.open-std.org/jtc1/sc22/wg21/docs/papers/2008/n2660.htm.
Which would have to be adopted to be coop-friendly which should not be difficult.
Notice that the existing "lazy" mechanism is also not coop-friendly.
That is not changed by this PR.
* [System.Net.Http]Add hack to the legacy HttpClient to pass down Timeout property. #12577.
* `HttpClientHandler`, `IMonoHttpClientHandler`: add new internal `MonoSetTimeout (TimeSpan)` function.
* `HttpClient.SendAsyncWorker()`: if `handler` is `HttpClientHandler`, call the new internal
`MonoSetTimeout()` function to pass down the timeout.
* `MonoWebRequestHandler.SendAsync()`: this is only used here to pass the timeout to the `HttpWebRequest`.
* Address feedback.
* Update HttpMessageInvoker.cs
* Make it build.
* Really make it build (forgot monotouch_watch).
* [llvm] Enable disabling inlining on llvm
* [llvm] Fix disabling direct calls on llvm
* [llvm] Rename mono_inited even when doing self_init (generated init functions refer to it)
Prevent the SDKs from making dangling symlinks when packaging the MXE
cross-compilers: should hav
mono.exe -> mono-sgen.exe
rather than
mono -> mono-sgen (dangling).
* RSACertificateExtensions.CopyWithPrivateKey(): implement.
Since `RSACertificateExtensions` lives in `System.Core` and only `mscorlib`
internals (but not `System` internals) are visible to it, we need to use a
little trick here:
* System.Security.Cryptography.X509Certificates.X509CertificateImpl:
add new abstract methods CopyWithPrivateKey() and CreateCertificate().
* System.Security.Cryptography.X509Certificates.X509Certificate2Impl:
sealed implement them here.
The `httpcfg` tool, only checks whether the key is valid, so we don't really
need to call `CopyWithPrivateKey()` (which requires us to take a dependency
on `System.Core`).
* [AppDomain] Cross-appdomain wrapper throws NIE for methods with ref struct arguments
If we have a ref struct and a method that takes it as an argument:
```
[Serializable]
public ref struct R {
public int i;
}
public interface I {
public void M1 (R r, SomeClass o);
}
public class M : MarshalByRefObject, I {
public void M1 (R r, SomeClass o) { ... }
}
```
If we create an instance of M in another domain and try to call it:
```
I obj = (I)otherDomain.CreateInstanceAndUnwrap (typeof(M).Assembly.FullName,
nameof(M));
R r;
r.i = 42;
SomeClass o = ...;
obj.M1 (r, o);
```
Mono will need to create a wrapper that can invoke M.M1 in the other domain.
The way the wrapper works in `mono_marshal_get_xappdomain_invoke` is by
creating an array of all the arguments that need to be serialized (see
`mono_get_xdomain_marshal_type`) and then serializing the whole array (by
calling out to `System.Runtime.Remoting.RemotingServices.SerializeCallData`).
In the example it would make a two element array and try to serialize `r` and
`o` into it. For valuetypes it normally does this by boxing the argument.
However since `R` is a ref-struct, boxing it is not allowed.
This works on .NET Framework.
However for Mono we would need to serilize `R` without boxing it, which seems
challenging with our current setup.
So for now we generate a wrapper that just throws a NotImplementedException if
it is ever called.
This fixes https://github.com/mono/mono/issues/14809 (in that marshalling a
TextWriter across domains works again) but in an unsatisfactory way (because
you can't call any of the ReadOnlySpan<char> overloads on the transparent proxy
object).
* [corlib] Add back Serializable attribute on some reflection classes
Tests that use https://www.nuget.org/packages/NUnit.ApplicationDomain/
need to serialize MethodInfo (and hence MethodBase).
* [corlib] Add test for transparent proxies for classes with methods that take ref struct args
Regression test for https://github.com/mono/mono/issues/14809
* Bump API snapshot submodule
Contributes to https://github.com/mono/mono/issues/9621
Fixes https://github.com/mono/mono/issues/14841
Note that it breaks iOS 32bit on devices. The Apple Watch _might_ still works (only tested on `2019-02`). It unblocks some build related issues on the Xamarin.iOS `2019-06` integration, and we have to do it anyway. I'll come back to it to fix device related issues.
It is important to provide this, otherwise the `opt` command will make modifications of every GC safepoint load. For the SmartHotel360 Android app, this can result in `opt` taking 19 minutes and 20 seconds to finish the optimization of the Mono.Android.dll AOT modules.
### IR Example:
Before Opt:
```
> %GC_SAFE_POINT_FLAG_7 = load i64*, i64** getelementptr inbounds ([166068 x i64*], [166068 x i64*]* @mono_aot_Mono_Android_llvm_got, i32 0, i32 7), !invariant.load !139597
> %4 = load volatile i64, i64* %GC_SAFE_POINT_FLAG_7
> %5 = icmp eq i64 %4, 0
> br i1 %5, label %CONT_BB2, label %POLL_BB1
```
After Opt:
```
< %GC_SAFE_POINT_FLAG_7 = load i64*, i64** getelementptr inbounds ([166068 x i64*], [166068 x i64*]* @mono_aot_Mono_Android_llvm_got, i64 0, i64 7), align 8, !invariant.load !139597
< %3 = load volatile i64, i64* %GC_SAFE_POINT_FLAG_7, align 4
< %4 = icmp eq i64 %3, 0
< br i1 %4, label %BB2, label %POLL_BB1
```
This has a huge impact on build time, giving InstCombine a huge slowdown. This change removes the need for opt to rewrite all of these basic blocks.
Backport of #14714.
1) CC_MD5_Init, CC_MD5_Update and CC_MD5_Final are deprecated.
```c
mono-md5.c:45:2: error: 'CC_MD5_Init' is deprecated: first deprecated in macOS 10.15 - This function is
cryptographically broken and should not be used in security contexts. Clients should migrate to SHA256
(or stronger). [-Werror,-Wdeprecated-declarations]
CC_MD5_Init (ctx);
^
```
2) objc_msgSend is deprecated
```
mono-threads-mach-helper.c:95:17: error: too many arguments to function call, expected 0, have 2
objc_msgSend (value, release);
~~~~~~~~~~~~ ^~~~~~~~~~~~~~
```
XCode has a special switch `Enable Strict Checking of objc_msgSend Calls` which is now `true` by default. It's fixed by `-DOBJC_OLD_DISPATCH_PROTOTYPES=1` flag to clang or the corresponding define in the source file.
3) Fixes https://github.com/mono/mono/issues/14793
It worked on pre 10.15 without `-framework GSS`, because it got pulled in through the dependencies. You can see it by running DYLD_PRINT_LIBRARIES=1 csharp and then trace it back. On Mojave /System/Library/Frameworks/GSS.framework/Versions/A/GSS got loaded as dependency of /usr/lib/libcups.2.dylib and so on. Unlike Windows, macOS will resolve the symbols to whatever is already available so it was just happy that someone loaded GSS before libmono-native[-compat/-unified].
Due to few recent updates to dotnet/standard (NETStandard 2.1) we had 40 apicompat issues.
Mostly copy-pasted from dotnet/corefx.
Didn't implement Cryptography stuff yet.
Checked all profiles.
Implements #14609
Backport of #14679.
[2019-06] [arm64] Correct exception filter stack frame size
Fixes: https://github.com/mono/mono/issues/14170
Fixes: https://github.com/xamarin/xamarin-android/issues/3112
The `call_filter()` function generated by `mono_arch_get_call_filter()`
was overwriting a part of the previous stack frame because it was not
creating a large enough new stack frame for itself. This had been
working by luck before the switch to building the runtime with Android
NDK r19.
I used a local build of [xamarin/xamarin-android/d16-1@87a80b][0] to
verify that this change resolved both of the issues linked above. I
also confirmed that I was able to reintroduce the issues in my local
environment by removing the change and rebuilding.
After this change, the generated `call_filter()` function now starts by
reserving sufficient space on the stack to hold a 64-bit value at the
`ctx_offset` location. The 64-bit `ctx` value is saved to the
`ctx_offset` location (offset 344 in the new stack frame) shortly
afterwards:
stp x29, x30, [sp,#-352]!
mov x29, sp
str x0, [x29,#344]
As expected, the invocation of `call_filter()` now no longer modifies
the top of the previous stack frame.
Top of the stack before `call_filter()`:
(gdb) x/8x $sp
0x7fcd2774a0: 0x7144d250 0x00000000 0x2724b700 0x00000000
0x7fcd2774b0: 0x1ee19ff0 0x00000071 0x0cc00300 0x00000071
Top of the stack after `call_filter()` (unchanged):
(gdb) x/8x $sp
0x7fcd2774a0: 0x7144d250 0x00000000 0x2724b700 0x00000000
0x7fcd2774b0: 0x1ee19ff0 0x00000071 0x0cc00300 0x00000071
<details>
<summary>Additional background information</summary>
### The original lucky, "good" behavior for [mono/mono/2018-08@725ba2a built with Android NDK r14][1]
1. The `resume` parameter for `mono_handle_exception_internal()` is
held in register `w23`.
2. That register is saved into the current stack frame at offset 20 by
a `str` instruction:
0x00000000000bc1bc <+3012>: str w23, [sp,#20]
3. `handle_exception_first_pass()` invokes `call_filter()` via a `blr`
instruction:
2279 filtered = call_filter (ctx, ei->data.filter);
0x00000000000bc60c <+4116>: add x9, x22, x24, lsl #6
0x00000000000bc610 <+4120>: ldr x8, [x8,#3120]
0x00000000000bc614 <+4124>: ldr x1, [x9,#112]
0x00000000000bc618 <+4128>: add x0, sp, #0x110
0x00000000000bc61c <+4132>: blr x8
Before the `blr` instruction, the top of the stack looks like this:
(gdb) x/8x $sp
0x7fcd2774a0: 0x7144d250 0x00000000 0x0a8b31d8 0x00000071
0x7fcd2774b0: 0x00000000 0x00000000 0x1ef51980 0x00000071
4. `call_filter()` runs. This function is generated by
`mono_arch_get_call_filter()` and starts with:
stp x29, x30, [sp,#-336]!
mov x29, sp
str x0, [x29,#344]
Note in particular how the first line subtracts 336 from `sp` to
start a new stack frame, but the third line writes to a position
that is 344 bytes beyond that, back within the *previous* stack
frame.
5. After the invocations of `call_filter()` and
`handle_exception_first_pass()`, `w23` is restored from the stack:
0x00000000000bc820 <+4648>: ldr w23, [sp,#20]
At this step, the top of the stack looks like this:
(gdb) x/8x $sp
0x7fcd2774a0: 0x7144d250 0x00000000 0xcd2775b0 0x0000007f
0x7fcd2774b0: 0x00000000 0x00000000 0x1ef51980 0x00000071
Notice how `call_filter()` has overwritten bytes 8 through 15 of the
stack frame. In this original lucky scenario, this does not affect
the value restored into register `w23` because that value starts at
byte 20.
6. `mono_handle_exception_internal()` tests the value of `w23` to
decide how to set the `ji` local variable:
2574 if (resume) {
0x00000000000bb960 <+872>: cbz w23, 0xbb9c4 <mono_handle_exception_internal+972>
Since `w23` is `0`, `mono_handle_exception_internal()` correctly
continues into the `else` branch.
### The bad behavior for [mono/mono/2018-08@725ba2a built with Android NDK r19][2] works slightly differently
1. As before, the local `resume` parameter starts in register `w23`.
2. This time, the register is saved into the stack frame at offset 12
(instead of 20):
0x00000000000bed7c <+3200>: str w23, [sp,#12]
3. As before, `handle_exception_first_pass()` invokes `call_filter()`
via a `blr` instruction.
At this step, the top of the stack looks like this:
(gdb) x/8x $sp
0x7fcd2774a0: 0x7144d250 0x00000000 0x2724b700 0x00000000
0x7fcd2774b0: 0x1ee19ff0 0x00000071 0x0cc00300 0x00000071
4. `call_filter()` runs as before. And the first few instructions of
that function are the same as before:
stp x29, x30, [sp,#-336]!
mov x29, sp
str x0, [x29,#344]
5. `w23` is again restored from the stack, but this time from offset 12:
0x00000000000bf2c0 <+4548>: ldr w23, [sp,#12]
The problem is that `call_filter()` has again overwritten bytes 8
through 15 of the stack frame:
(gdb) x/8x $sp
0x7fcd2774a0: 0x7144d250 0x00000000 0xcd2775b0 0x0000007f
0x7fcd2774b0: 0x1ee19ff0 0x00000071 0x0cc00300 0x00000071
So after the `ldr` instruction, `w23` has an incorrect value of
`0x7f` rather than the correct value `0`.
6. As before, `mono_handle_exception_internal()` tests the value of `w23` to
decide how to set the `ji` local variable:
2574 if (resume) {
0x00000000000be430 <+820>: cbz w23, 0xbe834 <mono_handle_exception_internal+1848>
But this time because `w23` is *not* zero, execution continues
*incorrectly* into the `if` branch rather than the `else` branch.
The result is that `ji` is set to `0` rather than a valid
`(MonoJitInfo *)` value. This incorrect value for `ji` leads to a
null pointer dereference, as seen in the bug reports.
[0]: https://github.com/xamarin/xamarin-android/tree/xamarin-android-9.3.0.22
[1]: https://jenkins.xamarin.com/view/Xamarin.Android/job/xamarin-android-freestyle/1432/Azure/
[2]: https://jenkins.xamarin.com/view/Xamarin.Android/job/xamarin-android-freestyle/1436/Azure/
</details>
Backport of #14757.
/cc @lambdageek @brendanzagaeski
When passing a delegate to native code, we pass the delegate_trampoline which can be used to invoke the delegate from native code and we store this mapping (delegate_trampoline->delegate) in a hashtable so we can get a delegate for the delegate_trampoline, when marshalling the opposite way. If we are trying to marshal a ftnptr (for managed code) to a delegate and we don't find this mapping we can crash. This could happen if the finalization for some other delegate that uses the same ftnptr removes this entry from the table. We solve this by never freeing these entries. The first delegate-ftnptr pair will always be alive. This is easier, faster and potentially uses less memory than tracking in the hashtable every pair. There should be a small and limited set of methods marshalled to native.
Fixes https://github.com/mono/mono/issues/13231
* Add a warning to the gensources Makefile so it's not confusing when it fails to compile in non-build profiles
* Add CI test script that verifies the compiler server functions
* Fix bugs
* Add compiler server tag to jenkins script
* Add +x
* Rename compiler server test tag
The test case CopyFileWithData_MemberData with input of size 1024 * 1024 * 10
causes the Xamarin.Android.Bcl_Tests to crash for some reason. Temporarily
exclude this entire set of tests from running.
Also disable System.IO.Tests.FileInfo_CopyTo_str because it derives from the
above test class.
* [corlib]: Move sre dependency into their own method in `System.RuntimeType`.
Move call to `TypeBuilderInstantiation.MakeGenericType` into a new
`System.RuntimeType.MakeTypeBuilderInstantiation` method to allow the
linker to remove it.
Requires linker PR https://github.com/mono/linker/pull/590.
* Bump linker.
* Bump cecil as well.
* Fix the build.
* Really fix the build.
* Bump API snapshot submodule
* [csproj] Update project files
when the allocation does not fit, i.e. when it is >= MONO_MEMPOOL_PREFER_INDIVIDUAL_ALLOCATION_SIZE.
Or at least make the code clearer and smaller.
"fixes" https://github.com/mono/mono/issues/14669.
It's a private helper that's always directly awaited. No need to allocate a task each time it's called and data is synchronously available.
Signed-off-by: dotnet-bot <[email protected]>
* Just use `new T[]` when elements are not pointer-free
* reduce zeroing out when not necessary.
* use AllocateUninitializedArray in ArrayPool
Signed-off-by: dotnet-bot <[email protected]>
We didn't set all the categories that the products used in our profiles.
Now that the products can consume the exclude categories from mono we
need to unify the values.
Removed ValueAdd since it's not used anymore and combined MacNotWorking
into NotOnMac.
It makes sense to pass --interp (--interp=-inline) while having another execution mode previously set up (like --full-aot-interp).
Cleanup code a little bit by killing mono_enable_interp which doesn't do much.
* Fix enabling the compiler server using autogen, and force it off if mcs is being used
* autogen updates
* Formatting fixes
* Fix vbcs getting turned on if unspecified even if mcs was selected
* Dynamically evaluate ENABLE_COMPILER_SERVER each time we compile instead of evaluating it once at some arbitrary point during makefile evaluation
* More detailed comment based on testing
* Handle the /shared option in genproj so it doesn't fail
* Raise basic profile check requirement to 6.2 and hopefully assert that named pipes work
* Default compiler server to off for now.
During XI integration we noticed that just using the verbatim commandline
-exclude or -notrait is not ideal since XI uses the nunit/xunit APIs to
pass that information and not the command line parser.
Instead we now write the exclude categories/traits as a newline-separated file.
[interp] Share more wrappers for different interp in signatures
The passed arguments are already considered to be gpointer in interp_entry and the real value is obtained using stackval_from_data. For types that are passed by value in a register the wrappers will be identical and we can share them.
This commit reduces the size used by interp_in wrappers (in fullaot-interp) by half. These wrappers still take a fair amount of size, which comes mainly from signatures including value types, which aren't shared.
We used to use the static MXE toolchain, but after switching to Homebrew, we
have to explicitly pass flags to build binaries that don't depend on the
MinGW32 GCC support libraries (libstdc++-6.dll, libwinpthread-1.dll and others)
when building the LLVM and Mono binaries.
The simplest approach is just to pass CMAKE_EXE_LINKER_FLAGS=-static to cmake
and LDFLAGS=-all-static to libtool. The rest of the changes are just to make
it easier to do that by adding a configure flag and updating the SDKs
infrastructure.
We will also need an LLVM bump to actually build a new llvm archive with the
new build flags; until then, building Mono will provision an old build of LLVM
with the old build flags that still depends on libstdc++-6.dll.
---
An alternative approach instead of passing -all-static and -static would be to
probe for the assorted -static-libXYZ flags to GCC at configure time. (For the
LLVM build we would need to hardcode them as that build doesn't go through
autoconf). Right now Mono doesn't have any other DLL dependencies (other than
the core windows DLLs), so -all-static works just as well and is simpler.
Otherwise the Xcode linker tries to guess the version from the path,
and since we use Xcode101.app it'd use sdk=101 as the version which is wrong.
By explicitly specifying e.g. `$(XCODE_DIR)/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS12.1.sdk`
as the sysroot it correctly guesses sdk=12.1 instead.
See also https://github.com/xamarin/xamarin-macios/commit/834c0a9fd7062c0c25011721fd7d524b8a60b41e
Reduce stack size used by interpreter frame on Windows MSVC.
Interpreter frame was 1248 bytes in size under Windows MSVC. This was almost double the size of other platforms ~670 bytes. After investigating different constructs in interpreter switch statement in interp_exec_method_full, it turns out that local declared errors couldn't be reused over different case statements and added ~100 bytes each. Switching local declared errors to shared one, reduce the stack size down to ~670 bytes under Windows MSVC.
ifdef __cplusplus
define g_static_assert(x) static_assert (x, "")
else
define g_static_assert(x) g_assert (x)
endif
It is, granted, C++11, not C++98, and some C variations.
I myself am bumping for ODBC fixes. This includes a change to SslProtocols to include the Tls13 entry required by the CoreFX changes. (It doesn't wire it up to anything, including BTLS.)
* Fix CollectionTest.UnusableSyncLockTest test
The test checked whether the SyncRoot is of type object, but the corefx implementation switched to returning `this`.
* Bump API snapshot submodule
SemaphoreSlim.Dispose nulls out its lock object, and that's then used as an indication in other methods whether they should throw ObjectDisposedException. But nulling out an object used for synchronization in other methods is a bad practice, and while SemaphoreSlim.Dispose is not thread-safe to be used concurrently with other usage of the instance, it's still confusing when such usage leads to NullReferenceExceptions due to trying to lock on a null lock object. This change just changes the lock to be readonly, always set, and then whether the instance has been disposed is just a value set on that object (such that it's no larger than it used to be).
Signed-off-by: dotnet-bot <[email protected]>
* [System.Net.NetworkInformation] Fork FreeBSD from MacOs
They are related, but not identical. In particular, MacOS AF_INET6 is wrong
on FreeBSD and breaks IPv6 networking on FreeBSD. Establish an inheritence
relationship with overrides for now; it's possible more differences will be
needed in the future.
This change is released under the MIT license.
* [csproj] Update project files
* [jit] Also allow System.Runtime.CompilerServices.Unsafe intrinsics from the System.Runtime.CompilerServices assembly, not just corlib.
* [jit] Fix Unsafe.RefIsAddressLessThan()/RefIsAddressGreaterThan().
Relevant issue: #14376
There are potentially more functions this can apply to, but this covers the ones listed in the issue and an extra one I found in the process.
Functions were either broken off into an external-only and internal version, or such a pair already existed and the internal version was moved from the source file to a separate `-internals` header, similarly to in 3c74df5667.
This gets slightly complicated with `mono_domain_set` and `mono_domain_set_internal`, both of which are currently public (thanks to 4beada2892) and have different signatures. Additionally, the latter does not actually include the gc-unsafe transition despite being public.
I've opted to introduce `mono_domain_set_fast` as the 'internal' version of `mono_domain_set`, to add the proper gc-unsafe transition to `mono_domain_set_internal`, and then to migrate internal usage to `mono_domain_set_internal_with_options`. As part of that I've marked `mono_domain_set_internal` as external only, which is unfortunate naming-wise but I believe the best option here without introducing API breakage.
[llvm] Use CoreCLR GC interop configuration
Since they're identical right now, we can use the coreclr configuration
to make our IR more easily analyzed by tools built around LLVM.
```
class CoreCLRGC : public GCStrategy {
public:
CoreCLRGC() {
UseStatepoints = true;
// These options are all gc.root specific, we specify them so that the
// gc.root lowering code doesn't run.
InitRoots = false;
NeededSafePoints = 0;
UsesMetadata = false;
CustomRoots = false;
}
Optional<bool> isGCManagedPointer(const Type *Ty) const override {
// Method is only valid on pointer typed values.
const PointerType *PT = cast<PointerType>(Ty);
// We pick addrspace(1) as our GC managed heap.
return (1 == PT->getAddressSpace());
}
};
/// A GCStrategy for the Mono Runtime.
class MonoGC : public GCStrategy {
public:
MonoGC() {
UseStatepoints = true;
// These options are all gc.root specific, we specify them so that the
// gc.root lowering code doesn't run.
InitRoots = false;
NeededSafePoints = 0;
UsesMetadata = false;
CustomRoots = false;
}
};
```
Source: https://github.com/mono/llvm/blob/64c0343537016c153894dc60316bd7b316b812c8/lib/CodeGen/BuiltinGCs.cpp#L105-L137
On the stage of the initialization of the SqliteMembershipProvider the existed code always produced an error "Failed to map path '/'" in case if the ASP.NET application is placed not at the root path. Detailed step by step description of the problem is provided in the issue description.
The bug was that was implemented the call of the function with wrong parameters list. So the fix was in changing the function call. The same function call is also used in SqlMembershipProvider.cs.
Fixes#14234
[merp] Don't leak g_getenv return value
Coverity said:
```
CID 1445100: Resource leaks (RESOURCE_LEAK)
Failing to save or free storage allocated by "monoeg_g_getenv("MONO_CRASH_NOFILE")" leaks it.
```
First set of fixes to get the following tests working under interpreter
on Windows x64:
* mono/mini richeck
* mono/mini mixedcheck
* mono/mini fullaotmixedcheck
* mono/tests winx64structs.exe
Fixes are primarily related to Windows x64 ABI support in interpreter. Fixes
also makes sure our interpreter trampolines get unwind info making it possible
to debug through the trampolines on Windows x64.
Several fixes related to value types was needed, also fixed support for return
value types not fitting into registers, current implementation didn't return
passed in value type address in return register as mandated by both
System V and Win x64 ABI's.
Identified and fixed memory leak on every pinvoke from interpreter that needs stack.
Since Windows x64 ABI always use 32 bytes of stack allocated by caller, the
memory leak was hit on every pinvoke call.
It was using an unescaped makefile variable in the template, i.e. ended up empty which meant the build just took the host version as the min version.
Fixed the variable and also moved it to CFLAGS/CXXFLAGS instead of CPPFLAGS to be consistent with ios.mk
* Throw ObjectDisposedException instead of crashing when Cairo.Context.handle == IntPtr.Zero
We are seeing instances where VS for Mac is crashing with a SIGSEGV when
calling into certain Cairo.Context functions. Looking at the code, it's clear
that it's possible to have Context.handle == IntPtr.Zero after disposing the
object.
From research, it looks like passing a nullptr to raw cairo functions can
result in a SIGSEGV:
https://bugs.launchpad.net/ubuntu/+source/compiz/+bug/93807
This change adds check to see if Conext.handle == IntPtr.Zero in Context
painting, font, and path methods. This will turn uncatchable SIGSEGVs
into ObjectDisposedException.
* Update Context.cs
* Ensure SetTarget results in a valid handle
* [llvm] Remove unused default_mono_llvm_unhandled_exception.
It references mono_gchandle_get_target_internal which is not marked MONO_LLVM_INTERNAL
so attempts to run it will fail.
This is caught by linking wit -z now (https://github.com/mono/mono/pull/14562).
* [loaded LLVM] mono_debug_get_seq_points and mono_class_enum_basetype_internak MONO_LLVM_INTERNAL
* Mark mono_debug_method_lookup_location MONO_LLVM_INTERNAL.
Found by linking with -z now (https://github.com/mono/mono/pull/14562)
which we should probably do for this and other reasons.
* #include mono-compiler.h for MONO_LLVM_INTERNAL.
* [loaded LLVM] Mark mini_is_gsharedvt_variable_signature as MONO_LLVM_INTERNAL.
Found by https://github.com/mono/mono/pull/14562.
* Mark mono_print_ins MONO_LLVM_INTERNAL.
* mono_rgctx_info_type_to_str MONO_LLVM_INTERNAL
* [loaded LLVM] mono_compile_create_var MONO_LLVM_INTERNAL
* ifdef around use of MONO_LLVM_INTERNAL and hope
* remove include mono-compiler.h from public header to internal header
* MONO_API_NO_EXTERN_C mono_debug_method_lookup_location() instead of mono_debug_method_lookup_location() MONO_LLVM_INTERNAL to workaround possible lack of MONO_LLVM_INTERNAL.
Rework use of marshalling for RuntimeTypeHandle, RuntimeModule, RuntimeAssembly, and IRuntimeMethodInfo as used by QCalls
- Remove special QCall only used marshallers for RuntimeAssembly, RuntimeModule and IRuntimeMethodInfo
- Following the pattern of ObjectHandleOnStack, implement QCall handle types for RuntimeAssembly/Module/TypeHandle. Use these in all QCalls that once passed the types directly.
- For uses of IRuntimeMethodInfo, follow the existing RuntimeMethodHandleInternal pattern
Also perform some replacement of bool marshalling with use of Interop.BOOL, and a few cases of using pointers instead of byref arguments.
Fix delivers a relatively small win on startup, and small throughput gains around reflection as IL stubs are no longer necessary for many functions in reflection that once needed them.
Reduces methods jitted on powershell startup from 422 to 399, (About 5%) but performance win is only about 5ms on ~400ms as the methods removed are simple.
Signed-off-by: dotnet-bot <[email protected]>
* [System.Core] Ensure monotouch[_runtime] API are identical
Ensure the API surface for `System.Core.dll` is identical between the
`monotouch` and `monotouch_runtime` (used for REPL and the interpreter).
This ensure code compiled against one will be fine when building against
the other (e.g. avoiding linker errors).
Removed methods:
```csharp
public void CompileToMethod (System.Reflection.Emit.MethodBuilder method);
public void CompileToMethod (System.Reflection.Emit.MethodBuilder method, System.Runtime.CompilerServices.DebugInfoGenerator debugInfoGenerator);
```
Removed methods:
```csharp
[Obsolete ("do not use this method")]
public static IRuntimeVariables MergeRuntimeVariables (IRuntimeVariables first, IRuntimeVariables second, int[] indexes);
[Obsolete ("do not use this method")]
public static System.Linq.Expressions.Expression Quote (System.Linq.Expressions.Expression expression, object hoistedLocals, object[] locals);
```
* [csproj] Update project files
* Bump API snapshot submodule
[interp] set correct stack type on op_Implicit/op_Explicit intrinsics for native types
Needs some additional conversions which happen to be included by accident previously.
This fixes those warnings on amd64:
```
Test run:
image=/Users/lewurm/work/mono-dim-call/mono/mini/builtin-types.exe
BuiltinTests.test_0_nuint_fieldload: Store local stack type mismatch 0 1
BuiltinTests.test_0_much_casting: Store local stack type mismatch 0 1
Results: total tests: 84, all pass
Elapsed time: 0.026546 secs (0.026546, 0.000000)
```
and those warnings on armv7:
```
Test run: image=/mono/mono/mini/builtin-types.exe
BuiltinTests.test_0_nint_inc: 0011 arith type mismatch add.i4 0 1
BuiltinTests.test_0_nint_dec: 0011 arith type mismatch sub.i4 0 1
BuiltinTests.test_0_nint_implicit_decimal: 0034 arith type mismatch
add.i4 0 1
BuiltinTests.test_0_nuint_inc: 0011 arith type mismatch add.i4 0 1
BuiltinTests.test_0_nuint_dec: 0011 arith type mismatch sub.i4 0 1
BuiltinTests.test_0_nuint_implicit_decimal: 0034 arith type mismatch
add.i4 0 1
Results: total tests: 84, all pass
Elapsed time: 0.039214 secs (0.039214, 0.000000)
```
[Android] Fix runtime loading of DSOs for 64-bit processes
Context: https://github.com/xamarin/xamarin-android/issues/2780
On Android the executable for the application is going to be
`/system/bin/app_process{32,64}` depending on the application's architecture.
However, libraries for the different architectures live in different
subdirectories of `/system`: `lib` for 32-bit apps and `lib64` for 64-bit ones.
Thus appending `/lib` below will fail to load the DSO for a 64-bit app, even if
it exists there, because it will have a different architecture. The quickest fix
is to use `lib64` explicitly for 64-bit Android apps.
<!--
Thank you for your Pull Request!
If you are new to contributing to Mono, please try to do your best at conforming to our coding guidelines http://www.mono-project.com/community/contributing/coding-guidelines/ but don't worry if you get something wrong. One of the project members will help you to get things landed.
Does your pull request fix any of the existing issues? Please use the following format: Fixes #issue-number
-->
Backport of #13281.
/cc @lambdageek @grendello
Use example.com and example.org instead.
Fixes https://github.com/mono/mono/issues/14585
Note that not all of the things I replaced make an actual network request but I thought it'd be nice to be consistent.
* Plug leak of CustomAttribute initializer name
All the CustomAttribute initializer names (fields, property) except the last were leaked. Ensure all names are freed.
Fixes#14543
* Apply suggestions from code review
Co-Authored-By: Martin Baulig <[email protected]>
* [driver] add --version-number
Useful for third parties consuming the version number of mono.
* s/--version-number/--version=number/g
* add man page entry
* Implement AppDomain.MonitoringTotalAllocatedMemorySize
* Change precise:true back to precise:false in GetTotalAllocatedBytes
Signed-off-by: dotnet-bot <[email protected]>
* Modified format logic so that braces are not allowed in custom format specifiers.
* Disabled failing test temporarily and fixed code style issue.
PR needed to pass test: https://github.com/dotnet/corefx/pull/35820
* Slight changes to logic to better align the diff.
* Fixed counting error.
* Fix nullable
* Move exclusion near related test
* Add skip to new issue file
* Delete CoreFX.issues.json
Signed-off-by: dotnet-bot <[email protected]>
This affects the case where SemaphoreSlim.WaitAsync is invoked, the operation completes asynchronously because there's no count available in the semaphore, and where a timeout and/or a cancellation token is passed to WaitAsync. In that case, we need to wait for the underlying wait task to complete, for cancellation to be requested, or for the timeout to elapse.
There's currently one code path that handles this. This change improves that slightly by avoiding a defensive array copy. However, we can do much better when we get a cancelable token but no timeout, in which case we can avoid operations like Task.Delay, can register with the original cancellation token rather than a new one (which means we avoid forcing a new CancellationTokenSource's lazily-allocated partitions into existence), etc.
Signed-off-by: dotnet-bot <[email protected]>
The most important point of the "temporary" aspect, esp. given
the complete conversion, is renaming MONO_PATCH_INFO_JIT_ICALL to MONO_PATCH_INFO_JIT_ICALL_ID.
Any accidental or merged MONO_PATCH_INFO_JIT_ICALL will fail to compile.
* Copy icall-def.h to icall-def-core.h.
* Manually preprocess icall-def.h vs. icall-def-core.h.
The only common part in this construction is the JIT icalls at the end.
There much overlap, now duplicated.
* Remove JIT icalls fro icall-def-core.h.
* Merge icall-def-core.h with master.
* Merge icall-def.h with master.
* Rename icall-def-core.h to icall-def-netcore.h.
* Remove more unused icalls
Fixes#24663 by calling `Write(ReadOnlySpan<char>)` instead of `Write(object)` with a boxed `Memory<char>`.
Signed-off-by: dotnet-bot <[email protected]>
Currently if calling GetSystemTimeZones, will return a list which in include Utc object created from the enumeration. But when calling FindSystemTimeZoneById("UTC") will return the cached Utc object which is different than the one returned from the enumeration. This in general could be not a problem but it is better to ensure returned the same object from both APIs. The reason is in our internal code, we use some reference equality for the perf reason. Ensure the usage of Utc object will make it more reliable.
Signed-off-by: dotnet-bot <[email protected]>
We encountered some cultures have a time patterns with single quotes which we should handle keeping the parts between quotes as fixed text inside the pattern and not treat it as any of the time specifiers
Signed-off-by: dotnet-bot <[email protected]>
Build / profile changes:
------------------------
* `build/profiles/testing_aot_common.make`: define `DISABLE_REMOTING = yes` and
`NO_MULTIPLE_APPDOMAINS = yes`.
* `mcs/class/corlib/Makefile`: check `DISABLE_REMOTING` when setting
`REFERENCE_SOURCES_FLAGS` (and don't define `FEATURE_REMOTING`, `MONO_COM`
and `FEATURE_COMINTEROP`).
* added new `testing_aot_common_corlib.dll.exclude.sources`. Several types in
the `System.Runtime.Remoting.Messaging` namespace are used by the runtime in
non-remoting scenarios, so those need to be preserved.
For all other remoting namespaces, we only need a few stripped down stub versions
of the some of the types; those have been added to a new `legacy` directory, so
we can wildcard-exclude all the files here.
* added new `testing_aot_common_corlib.dll.sources`; this currently only contains
`legacy/*.cs` to list above mentioned stub versions.
* added new `testing_aot_common_corlib_test.dll.exclude.sources` as a common test
excludes file.
API Changes (conditional to `DISABLE_REMOTING`):
------------------------------------------------
* `mcs/class/corlib/legacy/`: add stub-versions for some of the remoting-related
classes that are used by the runtime (so we can't completely remove those types):
- `System.MarshalByRefObject`.
- `System.Runtime.Remoting.IRemotingTypeInfo`.
- `System.Runtime.Remoting.ObjectHandle`.
- `System.Runtime.Remoting.RemotingServices`.
- `System.Runtime.Remoting.Activation.ActivationServices`.
- `System.Runtime.Remoting.Contexts.Context`.
- `System.Runtime.Remoting.Messaging.ILogicalThreadAffinative`.
- `System.Runtime.Remoting.Messaging.LogicalCallContext`.
- `System.Runtime.Remoting.Proxies.TransparentProxy`.
- `System.Runtime.Remoting.Proxies.RealProxy`.
some of these have fields that need to be preserved; these are marked with
appropriate `#region` pragmas.
All changes below are conditional to `DISABLE_REMOTING`.
* `System.Runtime.Remoting.Messaging.AsyncResult`: this class is used by some of
the async code, so we can't completely remove it. However, we can remove the
`MonoMethodMessage call_message` field when `DISABLE_REMOTING`.
* `System.Runtime.Remoting.Messaging.MonoMethodMessage`: this class is also used
by some runtime code and thus needs to be preserved. However, when `DISABLE_REMOTING`
is defined, then we make some changes to it:
- remove all iterfaces (`IMethodCallMessage`, `IMethodReturnMessage`, and `IInternalMessage`).
- `Properties` throws `PlatformNotSupportedException`.
* `System.Runtime.InteropServices.Marshal`: disable some COM-related code.
* `System.Threading.Thread`: remove `CurrentContext` property.
* `System.Threading.ExecutionContext`: adjust conditional logic; don't define
`FEATURE_REMOTING` if `DISABLE_REMOTING`.
* `System.Activator`: add `DISABLE_REMOTING` to the existing
`#if FEATURE_REMOTING || MOBILE_LEGACY` conditional in this class.
* `System.AppDomain`: since we only have one domain, `DefaultDomain` will always
return the root domain; comment out the call to `RemotingServices.GetDomainProxy`
as well as some internal methods.
* `System.__ComObject`: stub out on `DISABLE_REMOTING` as well as `!FULL_AOT_RUNTIME`.
* `System.Runtime.Serialization.Formatters.Binary.BinaryFormatter`: adjust `IFormatter`
versus `IRemotingFormatter` interface logic to use the former when `DISABLE_REMOTING`.
* `System.Runtime.Serialization.Formatters.Binary.BinaryObjectReader`: adjust existing
`#if FEATURE_REMOTING || MOBILE_LEGACY` conditional to add `DISABLE_REMOTING` to it.
* `System.Diagnostics.CorrelationManager`: comment out some `CallContext.LogicalGetData()`
and `CallContext.LogicalSetData`.
* `System.Diagnostics.TraceEventCache`: likewise.
Test changes:
-------------
* `mono/tests/Makefile.am`: conditionally disable some remoting tests.
* `corlib/Test/System.Runtime.Serialization/SerializationTest.cs`: conditionally
disable all tests requiring remoting.
* `corlib/Test/System/ActivatorTest.cs`: same; trying to preserve as much as possible
in this test.
* added some `DISABLE_REMOTING` and `DISABLE_SECURITY` conditionals to some of the tests.
Fix leak in mono_parse_options
When MONO_ENV_OPTIONS is null, we would leak the array and buffer
<!--
Thank you for your Pull Request!
If you are new to contributing to Mono, please try to do your best at conforming to our coding guidelines http://www.mono-project.com/community/contributing/coding-guidelines/ but don't worry if you get something wrong. One of the project members will help you to get things landed.
Does your pull request fix any of the existing issues? Please use the following format: Fixes #issue-number
-->
* [runtime] Fix mono_icall_get_wrapper_full () to always compile the wrapper even if a trampoline was created previously.
* [jit] Add a mini_patch_llvm_jit_callees () function to patch callsite variables for LLVM JIT.
* [llvm] Patch LLVM JIT callees in trampolines as well. Patching after a method is compiled is not enough, since new callsites can be created after a method has been compiled.
* [llvm] Cleanup calls from JITted code. Compile referenced jit icall wrappers immediately, since the calls will not be patched.
* Add a missing MONO_LLVM_INTERNAL.
* [arm] one more attempt to fix slotsize issue on llvmonly
Regression of https://github.com/mono/mono/pull/12992 & https://github.com/mono/mono/pull/14362
In the end, the actual fix boils down to:
```patch
--- a/mono/mini/mini-arm.c
+++ b/mono/mini/mini-arm.c
@@ -2374,7 +2374,7 @@ mono_arch_get_llvm_call_info (MonoCompile *cfg, MonoMethodSignature *sig)
lainfo->nslots = ALIGN_TO (ainfo->struct_size, 8) / 8;
lainfo->esize = 8;
} else {
- lainfo->nslots = ainfo->struct_size / sizeof (target_mgreg_t);
+ lainfo->nslots = ALIGN_TO (ainfo->struct_size, sizeof (target_mgreg_t)) / sizeof (target_mgreg_t);
lainfo->esize = 4;
}
break;
```
Tested on `xamarin-macios/arm64_32-v3` branch:
* mscorlib
* mini
* dont link
* monotouch tests
* and a modified version of https://github.com/mono/mono/issues/8486#issuecomment-414365860 (added larger structs too)
with each device, Watch Series 3 (`armv7k`) and Watch Series 4 (`arm64_32`).
I hope that is the last iteration on this.
* remove arm64_32_abi
* Reduce JIT icall hash use.
* PR: Rename mono_jit_icall_info_at to mono_find_jit_icall_info.
* Use enum instead of int, and update comment to reality (the assert is arguably leniently off by one.)
* Cast integer to enum for C++.
* More int vs. enum, and use two variables, and const.
* Undo a little fishing for CI failure.
* [netcore] Use bash explicitly for Arcade things
* [netcore] Work around https://github.com/dotnet/cli/issues/10014
* [netcore] Try forcing powershell Arcade on Windows
* [netcore] Switch build properties from /p: to -p: due to MSYS2/dotnet bug
* [netcore] Switch from nuget pack to dotnet pack
* [netcore] Add Windows x64 build
* [monodroid] Introduce 'win32 host' BCL build
* [sdks] Add _bcl_$(1)_BUILD_FLAGS variable to BclTemplate build step
Allows Android SDK on Windows to pass PROFILE_PLATFORM=win32
* [sdks] Pass _bcl_$(1)_BUILD_FLAGS when building the tests, too
* Copy test source exclusion files from the general monodroid profile
Apparently we don't fall back to the general exclusion sources for tests if the platorm-specific exclusion file does not exist.
* Also add win32 copies of the nunit exclude files:
files=$(find . -name "monodroid*_test*exclude.sources")
for file in $files
do
name=$(basename $file)
pushd $(dirname $file)
cp $name win32_$name
git add win32_$name
popd
done
* [csproj] Update project files
* [netcore] Exclude unused native locale icalls
* Fix builds with mono_feature_disable_sockets, currently not enabled by configure script
* [netcore] Don't compile in System.IO icall in netcore profile
* Make genmdesc output better typed and more readable -- almost hand-maintainable
but no clear route to actually hand-maintainable.
* Replace != with < and undo debug code.
* Changing EnableDefaultCompileItems=false to EnableDefaultItems=false
* Taking the VS auto-fixes for the solution and corelib csproj
* Removing the duplicated T4 file includes in the main csproj
* Regenerating the included T4 templates
* Fixing S.P.Corelib to properly list available platforms/configurations
* Add a comment on the T4 template service
Signed-off-by: dotnet-bot <[email protected]>
- Assembly.ToString() won't return null
- Several attributes have the intent that their arguments are non-nullable, even though they don't check and throw; the annotations should match the intent in such cases.
Signed-off-by: dotnet-bot <[email protected]>
Since commit d8af775, c++ compiler is required to compile.
This patch checks if the c++ compiler is available as soon as possible
and returns an error in case it is not found.
Fixes: https://github.com/mono/mono/issues/14195
Signed-off-by: Angelo Compagnucci <[email protected]>
Modify the sdks/builds/ infrastructure to support compiling the Android SDK on Windows.
The initial goal of this commit is to build only the monodroid BCL on Windows. Compilation of the host and cross-compiler mono binaries is future work.
There's one kind of hacky bit in here. Had to make sure that scripts/ci/run-jenkins.sh does _not_ set MONO_EXECUTABLE. Normally windows builds use the msvc in-tree compiled mono binary to build the BCL. In the case of the SDKs, we don't build that msvc binary, and we just use the usual (cygwin) mono to build the BCL.
* [sdks] allow building on cygwin
* [sdks] Start adding stub templates to llvm.mk
* [sdks] provision-android stubs for win32
* [sdks] Pass cygwin host triple to calls to configure on win32
* [sdks] Add runtime stub templates
* [sdks] Add android stub templates for win32
This is enough to get configure-android to work on cygwin x86_64
* [sdks] On win32, the HostMxe build is the host buildwe want the x86_64-w64-mingw32 toolchain, not the cygwin toolchain
* [sdks] For android on cygwin depend on the mingw64-{x86_64,i686}-zlib
* [sdks] Add android SDK build artifacts to .gitignore
* [sdks] Add AndroidCrossMXETemplate stub to android.mk for Windows builds
* [sdks] Fixup comments in android.mk
* [sdks] use UNAME=Windows for Cygwin and WSL
Don't assume Cygwin is the build platform for Windows Android SDK builds.
Keep using UNAME for coarse platform descriptions.
Switch to HOST_ARCH_MINGW32 instead of CYGWIN_BUILD_MACHINE for variable holding
the machine part of the host triplet for Windows.
* [sdks] Dummy provision-mxe target for windows
* [ci] Don't provision-android on Windows
It doesn't work yet.
* [sdks] temporarily turn off android-host-mxe-Win64-release
Needs mingw64-x86_64-zlib on all the CI builds bots.
* FIXME: hack - don't set MONO_EXECUTABLE to the msvc mono for Android SDK build
* [sdks] drop LLVMProvisionTemplateStub
[arm64] set MONO_ARCH_HAVE_UNWIND_BACKTRACE for watch4
Fixes a couple of mscorlib test failures, failing with
> Stack walks are not supported on this platform
* Revert "[System] Add FSEvent FileSystemWatcher to 'monotouch' BCL profile"
This reverts commit 46fc5352f3dfe0027e684a0f61e71130fccd559f.
While iOS FileSystemWatcher works, we now find that it is using unofficially-available symbols and cannot publish apps using those.
Fixes https://github.com/mono/mono/issues/14290
* [csproj] Update project files
* Enable nullable at the project level
* Remove `#nullable enable` from individual files
Removes `#nullable enable` from almost all .cs files in System.Private.CoreLib. I left it only in the ~30 files (out of ~1480 that had it) that are mirrored to corefx, that are built into projects by corefx that don't yet set NullableContextOptions at the project level, and that use nullable annotations; otherwise, they'd break the corefx build.
Signed-off-by: dotnet-bot <[email protected]>
* Enable MSVC builds of Mono w/ ENABLE_NETCORE
* Fix non-netcore build
* Move the ENABLE_NETCORE definition to config.h where it belongs
* Replace .def with linker #pragmas
* TARGET_WIN32 -> HOST_WIN32 && HOST_X86
* Pick up MONO_ENABLE_NETCORE from cygconfig.h
* Make `./configure --with-core=only` MSVC builds working
* Include netcore sources unconditionally, move the condition into the .c file
* Fix MSVC C++ netcore build on Win32
* [corlib] Change the RuntimeModule.GetGuid and RuntimeAssembly.GetAotId icalls to return a byte array instead of a string, so it doesn't have to be parsed by the Guid ctor. Also allocate the arrays in the c# code and pass them to the icalls.
* Do not allocate handles in ReleaseMutex.
Take advantage of thread/internal_thread maturity/pinning.
There are arguments for and against this.
It is arguably better to use handles everywhere.
But it is dubious to require an allocation to release a lock.
Releasing a lock should be infallible.
And surviving low memory shouldn't be a lost cause, imho..
This diff could also be viewed as an opportunistic cycle/size optimization
and be placed under ifdef -- ifdef THREAD_ALWAYS_PINNED orsuch.
This is an alternative to https://github.com/mono/mono/pull/14397.
* PR: Move comments.
Several fixes to the log profiler in order to work on Windows.
* Add new dynamic library project to build log profiler on Windows MSVC.
* Fix dependecy between log profiler and Mono.
* Rewrite command pipe logic on Windows since it's not supported to select
on both sockets and pipes.
* Several additional socket fixes needed for log profiler to work on Windows.
The less mechical parts:
- spell out monoeg_g_free
- register_dyn_icall not used, remove
- register_icall_no_wrapper only used once, spell it out instead
This is nice cosmetically on its own and will help remove JIT icall hashing.
* Introduce enum and static data for JIT icalls, that are currently hashed
by string and address and use dynamic allocation.
These will be used shortly.
* PR: Move MONO_JIT_ICALLS content from class-internals.h to jit-icall-reg.h.
Not clear what is the best name but moving along.
[gitignore] Add crash dump files, .dylib.dSYM directories, and test-path
Wasn't sure where in the file to best put the .dylib.dSYM stuff, so suggestions are welcome
* Mechanism strings to register_dyn_icall in mini-runtime.c.
That is both, generate them kinda, and make them available for some later manipulation.
* Mechanize strings for JIT icall registration in mini-runtime.c.
Generate them slightly and make them amenable to crude preprocessor transforms later.
* Try to improve github diff presentation.
The constructor for mono_error_set_argument lacks a `const char *msg_format`
so clearly we shouldn't be applying formatting from an entirely
unrelated parameter.
This was flagged by `-Werror=format-security`
During GC we allocate a gray queue protected by a mutex, sgen_gray_object_queue_init.
This mutex is never destroyed when clearing up the object containing it,
sgen_gray_object_queue_dispose, so it will leak. Overtime this could cause
OOM on some systems depending on how memory related to system primitives are
handled and partitioned.
* Make mono_fmod extern "C and not static, to simplify later
removal of JIT icall hashing. Making it more consistent with other icalls.
Costing maybe one instruction in some slow paths.
* Move mono_fmod from mini-runtime.c to jit-icalls.c.
* Fix declaratoin of mono_fmod.
Convert managed product binary to use SDK project system.
- Uses Arcade for versions strings
- Overrides Arcade defined output paths - should change in the future
Signed-off-by: dotnet-bot <[email protected]>
* [WinForms] If DataGridViewCheckBoxCell not in threeState mode
EditingCellFormattedValue must return true or false, instead of
value of CheckState type
* [WinForms] DataGrid: fix type check in EditingCellFormattedValue setter.
When set bool value to EditingCellFormattedValue there will be exception.
Type equality must be done through IsAssignableFrom not direct
* [WinForms] DataGrid: Add test for EditingCellFormattedValue getter and setter.
* match the parameter
* Removed special handling.
Removed special handling of TypedReference and RuntimeArgumentHandle
* Remove the whole special casing block
* Revert part of last commit
Restored the check for IsByRefLikeAttribute
* Corrected the function name
I accidentally left these MICROSOFT_BCL_ASYNCINTERFACES_PACKAGE references in as part of some recent churn. This compilation constant isn't set anywhere.
And the CORERT ifdef is no longer needed now that we've consolidated the code to shared.
Signed-off-by: dotnet-bot <[email protected]>
* Add ManualResetValueTaskSourceCore / AsyncIterateMethodBuilder to Microsoft.Bcl.AsyncInterfaces
These two types needed modifications to target .NET Standard 2.0 and are the necessary pieces to enable the compiler to compile async iterators.
- Copied ManualResetValueTaskSourceCore.cs from coreclr and tweaked it. I opted to do this rather than ifdef because the changes are not localized and I didn't want to significantly perturb the primary implementation.
- Added a few ifdefs to the shared AsyncIteratorMethodBuilder. It already had ifdefs, so I just added to it.
- Added a test project, and included the existing ManualResetValueTaskSourceCore tests. I had to disable two of the tests because of some of the optimization differences.
- Augmented those tests to validate that the compiler is able to successfully generate iterators and await foreach them.
* Address PR feedback
Signed-off-by: dotnet-bot <[email protected]>
* Implement GC.RegisterMemoryLoadChangeNotification
This allows users to register a callback to be called whenever the
memory load is between a low and high thresholds. The idea is to allow
them to tweak their memory usage based on the available memory.
It is only internal for now because we are not confident on the shape of
the API and if it's the best we can provide users. This needs to go
through a few iterations with first party users like ASP.NET or corefx.
WIP for https://github.com/dotnet/coreclr/issues/18619
* fixup! Implement GC.RegisterMemoryLoadChangeNotification
* fixup! Implement GC.RegisterMemoryLoadChangeNotification
* fixup! Implement GC.RegisterMemoryLoadChangeNotification
Signed-off-by: dotnet-bot <[email protected]>
* Added overloads to Path.Join
* Refactor Path.Join and Path.Combine to use ValueStringBuilder
* Use proper defined const for empty string
* Fix Windows-specific const
Signed-off-by: dotnet-bot <[email protected]>
We decided after all to put these on a different type, TaskAsyncEnumerableExtensions. This commit adds the new type. We can delete the methods in the old location once corefx consumes an updated build and switches over to using the new ones.
Signed-off-by: dotnet-bot <[email protected]>
The previous attempt ( https://github.com/mono/mono/pull/14078 ) only fixed it for ABIs where the stack pointer alignment is 8 bytes. ABIs, such as watchOS, require 16 byte alignment.
* [llvm] Widen vector equality (OP_PCMPEQx)
Fixes https://github.com/mono/mono/issues/14261
Fixes https://github.com/mono/mono/issues/14143
This was causing an issue when trying to inline
System.Numerics.Vectors.Vector`1<byte>:Equals into System.Span.IndexOf
from System.Memory.dll as provided by nuget. I believe that it is
comparing the result of a SIMD operation on two SIMD registers there with a
SIMD register full of zero bytes.
When we produce the Equals call (no inline), we spill the SIMD <4 x i32>
vector to the stack in the caller, and we then later load it
with the widened type used here <16 x i8>.
When we inline, we end up making the args into XMOV opcodes,
and then doing the "pcmpeqb" mini opcode with the return values
from those operations. This becomes a register to register no-op
copy, and the original <4 x i32> gets compared against the <16 x i8>
type:
```
%116 = icmp eq <16 x i8> zeroinitializer, <4 x i32> %115
```
This does not go the way we want it to. It reliably leads to IndexOf
breaking and only finding the desired byte in the portion not processed
by the SIMD code (in the last few elements of the array).
This fix checks if the input dimensions differ, and does a conversion.
This produces the following IR:
```
%114 = bitcast <4 x i32> %99 to <16 x i8>
%115 = icmp eq <16 x i8> %114, %113
%116 = sext <16 x i1> %115 to <16 x i8>
%117 = icmp eq <16 x i8> zeroinitializer, %116
```
* [llvm] Widen vector greater-than
Here we do the same trick used in the above fix, but for the parallel greater-than operation.
Technologies like SharpDX uses a minimal set of COM support from runtime
and System.Runtime.InteropServices.Marshal:
AddRef
Release
QueryInterface
All these are low level wrappers around methods in IUknown interface and
doesn't need any of the additional runtime COM support enabled to work.
Technologies like CoreRT includes these methods in their full AOT
profile, meaning that SharpDX can run on CoreRT on Windows platforms. Mono's
WinAOT profile and runtime currently don't, making it impossible to use
SharpDX.
This PR adds the needed methods into WinAOT profile and also make
sure corresponding icalls always gets included on Windows build Mono runtime.
I'm taking the LibraryNameVariation helper from System.Runtime.Loader for the ride as well because it's a general purpose probing logic that is useful in a managed implementation of NativeLibrary.
Signed-off-by: dotnet-bot <[email protected]>
Applies changes to `Array.IndexOf()` and `Array.LastIndexOf()` similar
to the changes made in #20855, so that types other than `byte` and
`char` can use use the fast vectorized path.
Also allows 32-bit and 64-bit types for which
`RuntimeHelpers.IsBitwiseEquatable<T>()` returns `true` to use the
faster implementation of `IndexOf` and `LastIndexOf` from
`MemoryExtensions`.
Signed-off-by: dotnet-bot <[email protected]>
* Workaround memset alignment sensitivity
memset is up to 2x slower on misaligned block on some types of hardware. The problem is uneven performance of "rep stosb"
used to implement the memset in some cases. The exact matrix on when it is slower and by how much is very complex.
This change workarounds the issue by aligning the memory block before it is passed to memset and filling in the potential misaligned
part manually. This workaround will regress performance by a few percent (<10%) in some cases, but we will gain up to 2x improvement
in other cases.
Fixes#24300
Signed-off-by: dotnet-bot <[email protected]>
* Increasing precision properly and leaving the rounding of to casting operator
* disabling corefx tests in coreclr and addressing feedback
* correcting test project name
* Fix json
Signed-off-by: dotnet-bot <[email protected]>
The --pinvoke-libs=x argument to packager.exe can be used to generate pinvoke linking tables.
'x' is a list of library names used in DllImport declarations, i.e. --pinvoke-libs=libtest is used
to link with [DllImport ("libtest")].
* Fix Satellite Assembly loading
When loading satellite assemblies, we should probe next to the parent
assembly and load into the same AssemblyLoadContext as the parent
assembly.
Disable fallback probing for satellite assemblies.
Add AssemblyLoadContext.Resolving handler to probe for satellite
assemblies next to parent
Fixes#20979
* Call ResolveSatelliteAssembly from native
Only call ResolveSatelliteAssembly from native when
resolving a satellite assembly
* PR Feedback
Minimize string creation
Remove unnecessary if null checks
Eliminate corner cases by only allowing one case insensitive matching directory.
* ResolveSatelliteAssembly should ...
ResolveSatelliteAssembly should always be called on the ALC which loaded parentAssembly
Simplify code.
Add Debug.Assert
* Remove case insensitive culture search
* PR Feedback
* Fix parentAssembly logic
* Fixes from initial testing
* Add probe for lower case culture name
* PR feedback
Signed-off-by: dotnet-bot <[email protected]>
Don't use popcount intrinsic.
Alternative but not exactly contradictory to https://github.com/mono/mono/pull/14248.
Note that this is a single inlinable instruction on x86,
if you assume at least SSE4 from circa 2007, which I am avoiding here.
* [interp] Optimize enum.HasFlag for csc pattern
Avoid boxing and slow call. Makes it 100x faster.
* [interp] Optimize enum.HasFlag for mcs pattern
* [interp] Transform ldc.i4 + conv.i8 into ldc.i8
Simplifies code, making enum.HasFlag optimization work for long enums.
* Refactor WebAssembly.Core object instantiation.
- Move the supported Core object identification into javascript.
- Runtime the core object type is now passed into the Runtime from javascript binding support instead of being housed in managed code.
* Address review comments.
* Fix indentation
* Implement suggestion by Katelyn Gadd to use Symbol instead of a Map or WeakMap.
- This removes the previous mono_wasm_core_map used for type mapping.
- Attach a special named symbol for `wasm type` to the prototype property for the Core objects that we have bound.
- The Managed Type is stored in the symbol property that will be passed to the managed side.
* [interp] Optimize special static field access
3x faster. For value types 10x.
* [interp] Remove duplicated code
* [interp] Further optimize thread static field access
Provides an additional 50% perf gain.
This field is used by the internal hash table to link nodes and it can be changed during hash table insertion of other imethods. Copying this field back was leading to random hangs in hash table lookup.
* [runtime] Add a python version of the offsets tool.
This version only depends on python and libclang. It currently only
works on osx.
* [runtime] Add a dummy structure needed by the python version of the offsets tool.
* [offsets-tool-py] Change some argument names for compatibility.
* [wasm] Use the python offsets tool on osx, the c# version is 32 bit, and the 64 bit version doesn't work.
* [wasm] Fix the ordering of mono libraries on the link line.
- The routine specifically will escape path separators when needed for instance on windows instead of using the simplified `.Replace ("\\", "\\\\")` as previously implemented.
- The routine will also add quoting when necessary.
The version number in version.h only shows e.g. "6.3.0", and we
want full version numbers ideally, to match what our Mac pkg files
show. This logic calculates that version number.
* Revert "[Tailcall] [arm] [jit] Fix moving tailcall parameters. (#12701)"
This reverts commit 257efc6374.
* When moving parameters for tailcall, and having pushed ip,
and adjusting the offset to account for that, split the adjustment
between sp and frame pointer, as frame pointer may or may not be sp.
This fixes https://github.com/mono/mono/issues/11489.
* Restore Roslyn compilers from NuGet to allow up-to-date version to be used
* Move the Roslyn version to makefile, ensure that it is updated before building System.Private.CoreLib
* Pass the whole path to Roslyn props file instead of just Roslyn version to dotnet build
* Make the Roslyn restore compatible with .NET Core 2.2 SDK
Under hybrid suspend a thread can be cooperative suspended but still
checked for a platform critical region under wow64. Since thread is
not preemptive suspended we can't reliably call GetThreadContext and
doing so could also (even correctly) return a context flagged as critical
triggering an assert in stw for cooperative suspended threads.
We were inadvertently running the roslyn tests against the system mono in CI.
Instead we now install mono to a custom prefix, prepend that to PATH and run
the roslyn tests against the correct mono we just built.
* [netcore] Convert netcore/Makefile.am to a normal makefile, get the configure variables from config.make.
* [netcore] Convert mcs/class/System.Private.CoreLib/Makefile.am to a normal makefile.
We need to emit a real call here. On arm/android we would emit this now:
```asm
ldr pc, =label
.ltorg # place address of `label` here.
```
which is a jump to `label`, not a call.
We need a different solution to make larger assemblies work, but for now
I revert it so the `2019-02` intergration is unblocked.
Partial revert of #14043
* Bump CoreFX version, include few more packages
* Add exclusions for System.Collections.NonGeneric.Tests
* Exclude tests that crash runtime
* Add more exclusions for known failures
* Exclude System.Utf8String.Experimental.Tests, it is CoreCLR specific
Exclude System.ComponentModel.Composition.Registration.Tests, the published version throws PNSE for all methods
Please enter the commit message for your changes. Lines starting
* Make a list of known non-passing tests and establish baseline for 'make xtestpass'
* Add exclusions
* Reenable System.Net.NameResolution.Functional.Tests
* Reenable System.Net.Sockets.Tests
* Reenable System.Diagnostics.Tracing.Tests
* Reenable System.Net.Http.Functional.Tests
* Reenable System.Net.HttpListener.Tests
* Reenable System.CodeDom.Tests
* Reenable System.Linq.Expressions.Tests
* Reenable System.Threading.ThreadPool.Tests
* Reenable System.Runtime.Tests
* Reenable System.Collections.Tests
* Reenable System.Threading.ThreadPool.Tests
* Reenable System.Memory.Tests
* Reenable System.Linq.Parallel.Tests
* Update exclusions for System.Diagnostics.Process.Tests
* Reenable System.Dynamic.Runtime.Tests
* Reenable System.Runtime.Numerics.Tests
* Sort excluded tests
* Reenable System.Reflection.Tests
* Reenable System.Reflection.Emit.Tests
* Reenable System.Linq.Parallel.Tests
* Reenable System.Diagnostics.TraceSource.Tests
* Reenable System.Threading.Tests
* Disable tests that trigger UI on macOS
* Update System.Runtime.Tests exclusion list
* Update line numbers
e.g. whenever underlying representation is nul terminated string.
MONO_PATCH_INFO_LDSTR_LIT
MONO_PATCH_INFO_OBJC_SELECTOR_REF
almost the same as
MONO_PATCH_INFO_JIT_ICALL
MONO_PATCH_INFO_JIT_ICALL_ADDR
MONO_PATCH_INFO_JIT_ICALL_ADDR_NOCALL
MONO_PATCH_INFO_LDSTR_LIT no longer allocate
and copy the string from the AOT image into image_pool.
g_str_equal check pointer equality first.
While obviously correct, not obviously faster,
as it adds cost when strings are unequal.
Uninline g_str_equal, seems cleaner, but perhaps less efficient.
Combine the switch-exit and default: paths,
as only one case other than default exited the switch.
Lift out other common subexpressions.
* Package the wasm and wasm Facades correctly.
* Move the build props and targets file to netstandard2.0 build directory.
* Place framework .props file properly for build
* Place runtime .props file properly for build
* A little cleanup and add escaping windows path separator.
- There are problems executing mononlinker.exe on windows if the windows path separators are not escaped.
* [image] Create MonoImageStorage to own the image raw data
Create a new data structure: MonoImageStorage.
It is an object that will have the responsibility for the raw data of a
MonoImage. It has a string key and a refcount that is used to share a
MonoImageStorage between multiple MonoImage objects.
The reason we need this is because the current MonoAssembly/MonoImage design is
broken for multiple domains and (more evidently) it will be broken when we add
AssemblyLoadContext support. The issue is that a MonoImage may be shared
between multiple domains (or ALCs), but it has a 'references' field which
points to a single other MonoAssembly.
This is a problem because the references of an image may be resolved
differently in different domains (or ALCs).
The eventual solution will be to stop sharing MonoImages based on path name (or
in the case of in-memory images based on a name made up from the address of the
byte blob).
However we still don't want to open the same data more than once (if we're on a
config where we don't have mmap, we malloc some memory and dump the data in
there - we don't want to do that multiple times for the same file).
So the solution is to create this MonoImageStorage object and make it possible
to share it based on path name, but don't give it any responsibilities except
for owning the memory. It doesn't know anything about metadata or assemblies
or any of that stuff - it just owns a chunk of memory and knows when and how to
free it.
This commit just adds the MonoImageStorage object and wires it up in MonoImage
loading. There should be no observable behavioral changes from this commit.
* [win32] call FreeLibrary from MonoImageStorage dtor, not mono_image_close_except_pools
If we're sharing a MonoImageStorage, only call FreeLibrary when the storage is
finally destroyed, not every time one of the images is closed.
* Update dependencies from https://github.com/dotnet/corefx build 20190419.5
- Microsoft.NETCore.Platforms - 3.0.0-preview5.19219.5
- Microsoft.Private.CoreFx.NETCoreApp - 4.6.0-preview5.19219.5
* Removing the Sse2.StoreLow(long/ulong*, ...) methods that have been replaced
* Moving the Sse2.StoreLow(long/ulong*, ...) tests to be Sse2.StoreScalar tests
* Fixing some existing usages of Sse2.StoreLow(long/ulong*, ...)
Signed-off-by: dotnet-bot <[email protected]>
* removing using systems and adding nullability for System.Diagnostics.SymbolStore
* annotating the class that implements the internal interface
* adding bang to fix failure
* correcting the byte array declarations
Signed-off-by: dotnet-bot <[email protected]>
* Nullable: shared\System (most of it)
And some other things it touches.
* Nullable: src\System (most of it)
* Address PR feedback
Signed-off-by: dotnet-bot <[email protected]>
* Nullable: System.IO
Everything exception StreamReader, StreamWriter, BinaryReader, BinaryWriter, which are waiting for some changes in master to propagate over to the feature branch.
* Address PR feedback
Signed-off-by: dotnet-bot <[email protected]>
* Nullable: System.Runtime.InteropServices.Marshal
* Fixing the nullable annotations for BindToMoniker and GetObjectForIUnknown
* Fixing the nullability of PtrToStringAuto(IntPtr, int)
Signed-off-by: dotnet-bot <[email protected]>
Ultimate goal is to significantly reduce or maybe completely
eliminate the hashing of JIT icalls.
It is not a small change.
This is a small digestable albeit tedious step toward that.
The "names" of opcode emulation icalls are needed for
token pasting and/or struct fields. They are needed without quotes.
This PR removes the quotes.
And then for now, just always restores them.
This is extracted from https://github.com/mono/mono/pull/14047
as a step toward making it easier to review.
* [loader] Dont't store through a null ptr
If 'status' is null, don't try to write to it.
Fixes https://github.com/mono/mono/issues/13941
* [metadata] Protect writes to MonoImageOpenStatus* in a few places.
When possible assert that `MonoImageOpenStatus *status` arguments to various
functions in the runtime aren't null. There are a few places (MONO_API) where
it's possible that callers are passing null. In that case, use a local
MonoImageOpenStatus, or check for null before doing assignments.
* Cut down on strlen calls.
But keep the shared memory format unchanged to match .NET. (i.e. don't
put string lengths in it).
* PR: MONO_ENTER_NO_SAFEPOINTS instead of gchandle.
Push/pop frames in loops and mind the return/goto/break/continue carefully, instead of using assignment.
*PR: Reduce gsize to adequate int, move multiple int declarations to same line.
* PR: Fix prototypes in disabled code.
* PR: Spell out `int` as `gint32`.
Use an uppercase guid to workaround Windows problem.
* A bit more true to the previous version.
* Reduce diff a little.
* if (is_ok()) around one line instead of goto around it.
* Restore some old code.
* Restore more verbosity, that used to have a point but no longer does.
* [aot] Add a depfile= argument which causes the AOT compiler to write out a gcc style dep file listing the dependencies between the output file and all referenced assemblies.
* Add a manpage entry.
* Update after comments.
When a static method is called the obj is not used, so it can be null and it doesn't need to be boxed.
Adding an assert in the method to prevent other crash in this method if a NULL is received.
[interp] Simplify cpobj.vt
[interp] Optimize stelem.vt/ldelem.vt
[interp] Avoid unnecessary write barrier for ldobj.vt
Write barriers are only required for stores in the major heap/los.
[interp] Simplify ldfld.vt/stfld.vt
* [interp] Remove redundant check
It can't be an enum if it is stored on the vtstack
* [interp] Optimize boxing
Makes boxing of primitives 4x faster and boxing of vt 3x faster.
* [interp] Remove unneccessary check in unbox
We do the same check as jit. Makes unboxing 10x faster.
* [interp] Extract some duplicate code for boxing
* [interp] Fix boxing of nullable with constrained calls
Before the box opcode optimization, it was handling nullable. When normally boxing a nullable we call the Box method, but this is awkward when doing a constrained call, so we just add a separate opcode. This is slightly slower than calling Box, but much faster than the old code using the generic mono_value_box_checked.
Now all custom attribute related tests pass in System.Runtime.Tests.
Also this PR fixes broken acceptance tests for regular mono introduced in https://github.com/mono/mono/pull/13809
* Move most of EventSource to shared CoreLib partition
* Reenable some code that can be shared with Microsoft.Diagnostics.Tracing.EventSource.Redist in CoreFX
* Remove unused functions
* Fix placement of #endregion
* Clean up P/Invoke prototypes
* Move remaining code from EventSource_CoreCLR to shared partition (guarded by !CORERT)
* Remove more unnecessary marshalling attributes
* Move remaining CoreRT/ProjectN code to shared partition
* Avoid using Windows errors on non-Windows platforms
Signed-off-by: dotnet-bot <[email protected]>
* [netcore] Add linker descriptor. This is just a copy of the mscorlib descriptor.
* [netcore] Update linker descriptor for netcore.
* Remove com/remoting related classes.
* Other smaller changes.
* [netcore] Allow loading corlib from MONO_PATH.
* Update System.Private.CoreLib.xml
[aot] change method table generation so that it works for larger assemblies on ARM
At the end of the AOT compilation, we dump a table containing all method
entrypoints using the toolchain linker to resolve the addresses. Previously
we used `bl` for that, however on larger assemblies the offset might
overflow the 24bit immediate (thus, the linker bailing out with
"Error: branch out of range"). Instead, using `ldr pc,=<label>` followed
by `.ltorg` emits the full 32bit address to a method. That takes 8 bytes
for each method instead of 4 bytes.
( I couldn't reopen https://github.com/mono/mono/pull/3486 anymore)
* Check for lower bounds on non-zero-based arrays
* Grab correct version of System.Reflection.Context
* Move Attribute type validation to fix checking for custom attributes using interface types
* Fix parameter checks
* [interp] Further optimize static vt field access
We don't require any write barrier checking since we always copy between interpreter vt_stack and static data. We can also include the vt size in the code stream. Makes it over 1.5x times faster.
* [interp] Fix static field access using ldfld/stfld
Static field access using the ldfld/stfld instructions was not updated when stsfld/ldsfld were optimized.
* [interp] Fix special static vtype field access
* Switch to msbuild nuget which is available for all platforms
* Remove check-env rule dependency
* Add missing reference to System.Composition.TypedParts
* Sort references by name
* Fix Windows builds of CLR hosting API
* Make System.Private.CoreLib compilable on Windows
* Remove accidental change
* Fix Unix build
* Update GlobalizationMode.cs
* Update GlobalizationMode.Unix.cs
* Move DateTime Windows implementation to shared CoreLib
* Make all the BOOL/struct marshalling explicit
* internal -> private
* Use unsafe code on all layers, remove unnecessary fixed blocks
Signed-off-by: dotnet-bot <[email protected]>
This compat quirk is increasingly more broken since the framework is generally not compatible with, and we have not heard anybody actually using it.
Changed Environment.HasShutdownStarted to unconditionally return false. It does not make sense for it to ever return true with shutdown finalization disabled.
Signed-off-by: dotnet-bot <[email protected]>
* Update dependencies from https://github.com/dotnet/corefx build 20190411.2
- Microsoft.NETCore.Platforms - 3.0.0-preview5.19211.2
- Microsoft.Private.CoreFx.NETCoreApp - 4.6.0-preview5.19211.2
* Update dependencies from https://github.com/dotnet/corefx build 20190411.22
- Microsoft.NETCore.Platforms - 3.0.0-preview5.19211.22
- Microsoft.Private.CoreFx.NETCoreApp - 4.6.0-preview5.19211.22
* Removing renamed HWIntrinsic APIs from S.P.Corelib
* Updating the hwintrinsiclistxarch to match the new intrinsic API names
* Updating the HWIntrinsic generated tests to use the new API names
Signed-off-by: dotnet-bot <[email protected]>
* Make GC.GetGCMemoryInfo public
This is to be used to allow users to optimize memory consumption based
on what's available to the GC and on the system.
This is based on https://github.com/dotnet/corefx/issues/34631
* Address reviews
- Pass GCMemoryInfo struct to native in place of many arguments
- Make GCMemoryInfo.HeapSize and GCMemoryInfo.Fragmentation long in place of IntPtr
* Address reviews
- Mark GCMemoryInfo readonly
* Rearrange fields and match managed/native names
* Fix compilation
* Use getter shorthand
* Address API Design Review
https://github.com/dotnet/corefx/issues/34631#issuecomment-481358549
* Fix comments
* Fix comments
* Do everything in managed
* Address review
- Remove unecessary [StructLayout.Sequential]
- Remove "_" prefix for parameters variables
Signed-off-by: dotnet-bot <[email protected]>
* Update Range.GetOffsetAndLength
Make it return offset and length tuple instead of the old Type Range.OffsetAndLength
* Fix the return Tuple fields names
Signed-off-by: dotnet-bot <[email protected]>
* Updating the Avx2.ConvertToVector256UInt APIs to return a signed type (matching the native signature and Sse41 APIs).
* Adding explicit overloads for `Sse41.ConvertToVector128Int*` and `Avx2.ConvertToVector256Int*` that take a pointer
* Updating the word ordering for the Compare/CompareScalar methods.
Signed-off-by: dotnet-bot <[email protected]>
* rename BaseCounter to DiagnosticCounter
* Change MetaData->Metadata
* Make EventSource and Name a property for counter classes
* Make the counter APIs public
* fix build errors
* Change float to double
* Few cleanups, fix test
* fix GetMetadataString
* PR feedback
* More PR feedback
Signed-off-by: dotnet-bot <[email protected]>
* [interp] Remove dead code
It doesn't make sense to dereference a value type
* [interp] Emit faster instructions for ldobj
* [interp] Emit faster instructions for stobj
* [interp] Simply ldobj vt
This makes it 10x faster
* [interp] Make ldind.r8 unaligned friendly
Assembly.LoadFromPartialName(string) should not throw FileNotFoundException, but
should rather return null.
ArgumentException should use nameof(partialName)
Fixes#19817
Signed-off-by: dotnet-bot <[email protected]>
Previously for `<`/`<=` it would throw if `v1` argument was null, but for `>`/`>=` would throw if the `v1` argument was null, albeit using `v1` in the exception parameter name. It's also not expected for comparisons like this to throw, in general.
Signed-off-by: dotnet-bot <[email protected]>
Symlinks on TVOS run into sandboxing restrictions. We've already disabled a similar test.
This test was added recently as part of 26e0b1deec.
(cherry picked from commit b85fdbdb10)
There are many minor arguments here.
Where this matters most -- iOS -- the pool is already dynamic.
Other platforms could use a dynamic pool or JIT (despite FullAOT).
Using JIT would defeat the test coverage.
Mitigates https://github.com/mono/mono/issues/13888.
If this feels too tenative, go bigger.
`make dist` stopped erroring in #13961 but didn't actually work, because
the relevant Makefile.am was not distributed, causing configure to fail
(because the Makefile.am was cited as an input, unconditionally).
We can't rely on the dist-recusrive from library.make or build.make because
this lib is in a weirdo format which doesn't comply with the requirements
of those helpers, so craft a simple handwritten replacement for our special
case
Cleaning exceptions during static constructor loading
Reset the stack_trace and trace_ips because the type initialization exception is reused.
Fixes#13714
* Use a custom prefix instead of trying to infer it, which doesn't work under bockbuild.
The following line is problematic in the scope of building the Mac SDK package (using bockbuild):
https://github.com/fsharp/fsharp/blob/ccfccbacfdb16b2ab143cf83f5ece20f738fd864/mono/config.make#L5
It's problematic because where `mono` lives in this case would be something like `/macsdk-staging-dir/bin/mono` which is not the actual prefix (prefix here being the final destination, `/Library/Frameworks/Mono.framework/...`) but a staging area that bockbuild uses (this is how we build on system prefixes without installing on CI)
For the F# build, bockbuild expects things to end up in `/fsharp-staging-dir/$(prefix)` (because that's what most `make install`s do when given `DESTDIR=/fsharp-staging-dir/`). If things end up in _any other_ directory in `/fsharp-staging-dir/` (like `/fsharp-staging-dir/macsdk-staging-dir` in our case) that would cause us to miss packaging those files (this has happened before, with packages that have broken or different use of `DESTDIR`). So there is an added write-protection for anywhere in `/fsharp-staging-dir` that *isn't* `/fsharp-staging-dir/$(prefix)` , because a write-protection error, while obtuse, is much preferable to missing the fact that some files are not packaged (as would have happened here)
So the fix is to stop trying to infer the prefix, and return to providing it explicitly to the build system (this used to be the case via the `configure` step)
* Also, if we are doing an override of the 'make' command, we need to take care of passing DESTDIR ourselves too
* Since now we perform 'make install' on the make step, make the 'install' step into a no-operation.
* Look for sn where mono is, not in the final destination (which again, may not be the same)
Currently we have two nullable API's that will have value in the public
embedding API (already requestsed by users), eliminating the needs to go
from native->managed and back to get nullable information about a type.
mono_class_is_nullable
mono_class_get_nullable_param
This PR moves them from class-internals.h into class.h and mark them
with MONO_API and when needed include GC UNSAFE/SAFE transition.
GitHub provides a way to download a .zip/.tar.gz of the repository content in their UI, however the archive doesn't contain git submodules.
The mono build then fails with confusing "missing file" errors.
Since GitHub doesn't let us disable the archive download we should at least add some logic to detect this broken case and bail out.
uses_handles parameter is a gboolean -> int32_t in Mono but was uint8_t
in WASM implementation. This triggered a problem when a uninitialized
gboolean was used and only parts of it got set and method still returns
an address, indicating success. This is a different behavior compared to
Mono's default implementation. Fix to this particular problem was to init
the gboolean variable before calling, but good to make sure the signatures are in
sync anyways.
icalls registered with mono_add_internal_call, mono_add_internal_call_with_flags
or mono_add_internal_call_internal ended up in two different hash tables
depending if they were "foreign" or not. This caused double lookups in
mono_lookup_internal_call_full. There are also use cases where it could be
interesting to keep additional flags related to the icall, that could be
used by codegen for various optimization.
This PR merge the two hash tables into one and add a flag describing icall
details. The flag could later be extended with additional behaviors going
forward if needed.
* Improve RuntimeHelpers.GetSubArray
This change does three things.
First, it fixes `GetSubArray` to work when the supplied array is actually a `U[]` where `U : T`. Currently this case ends up throwing an exception inside of span, which doesn't like working with arrays covariantly.
Second, it fixes argument validation so that we throw an ArgumentNullException if the input array is null rather than NullReferenceException.
Third, it improves the performance of `GetSubArray` for the 95% common case where either `T` is a value type or the type of the array matches the `T` type specified.
* Only use `Array.Empty<T>` when `typeof(T[]) == array.GetType()`
Signed-off-by: dotnet-bot <[email protected]>
* Match PtrToStringUTF8 to the other PtrToString* families.
* Make PtrToString*(IntPtr,int) exceptions consistent.
* Exclude out-of-date tests in CoreFX.
* Update exclusions to cover tests that are now throwing ArgumentOutOfRangeExceptions instead of ArgumentExceptions.
* Fix parameter order for exception.
* Pass value to exception.
Signed-off-by: dotnet-bot <[email protected]>
* Moving some JITDUMP calls into Compiler::lookupNamedIntrinsic
* Marking the IsSupported methods as Intrinsic for all HWIntrinsic ISAs
* Updating the hwintrinsic importation to more generally handle IsSupported and ThrowPNSE
* Applying formatting patch.
* Adding using System.Runtime.CompilerServices to the various x86 PlatformNotSupported HWIntrinsic files
Signed-off-by: dotnet-bot <[email protected]>
These types null out some fields on Dispose, and some of those fields are accessible directly via exposed properties, which means that nulling them out would either force us to annotate those properties as being nullable or would require us to lie about them being nullable. Nulling them out also makes it easier to accidentally null ref in the implementation, and prevents us from making the fields readonly.
It should be exceedingly rare that you dispose of one of these objects and then keep it alive while also being concerned about keeping alive the other helper objects the type uses. So this change just uses a separate flag to track disposal, and avoids nulling out the fields, which is not a common practice elsewhere in the framework.
This will be a visible change in that after Dispose, properties like StreamWriter.BaseStream/Encoding will now return the stream/encoding rather than null. In theory that could break someone if they were using those properties to determine whether the writer had been disposed, but that seems unlikely / similar fallout from most any bug fix, and objects generally aren't supposed to be touched after Dispose, anyway.
Signed-off-by: dotnet-bot <[email protected]>
[aot] use g_malloc () instead of malloc ()
Unbreaks jemalloc support as it hooks into `g_malloc ()` instead of replacing `malloc ()`.
```
%4 0x000055ee6af0d27b in sigabrt_signal_handler (_dummy=6, _info=0x7ffc7b0c4ff0, context=0x7ffc7b0c4ec0) at mini-posix.c:232
%5 <signal handler called>
%6 __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:51
%7 0x00007f4ae0d73801 in __GI_abort () at abort.c:79
%8 0x000055ee6b2ab6ed in rtree_child_leaf_tryread (elm=0x55ee6b75b028 <je_extents_rtree+704104>, dependent=true) at src/rtree.c:205
%9 0x000055ee6b2ab95f in je_rtree_leaf_elm_lookup_hard (tsdn=0x7f4ae1f05738, rtree=0x55ee6b6af1c0 <je_extents_rtree>, rtree_ctx=0x7f4ae1f05758, key=94482510772944, dependent=true, init_missing=false) at src/rtree.c:292
%10 0x000055ee6b242f70 in rtree_leaf_elm_lookup (tsdn=0x7f4ae1f05738, rtree=0x55ee6b6af1c0 <je_extents_rtree>, rtree_ctx=0x7f4ae1f05758, key=94482510772944, dependent=true, init_missing=false) at include/jemalloc/internal/rtree.h:369
%11 0x000055ee6b242fb5 in rtree_read (tsdn=0x7f4ae1f05738, rtree=0x55ee6b6af1c0 <je_extents_rtree>, rtree_ctx=0x7f4ae1f05758, key=94482510772944, dependent=true) at include/jemalloc/internal/rtree.h:394
%12 0x000055ee6b2431ee in rtree_szind_slab_read (tsdn=0x7f4ae1f05738, rtree=0x55ee6b6af1c0 <je_extents_rtree>, rtree_ctx=0x7f4ae1f05758, key=94482510772944, dependent=true, r_szind=0x7ffc7b0c6000, r_slab=0x7ffc7b0c6004) at include/jemalloc/internal/rtree.h:446
%13 0x000055ee6b24d468 in ifree (tsd=0x7f4ae1f05738, ptr=0x55ee6c7dfad0, tcache=0x7f4ae1f058f0, slow_path=true) at src/jemalloc.c:2126
%14 0x000055ee6b24de06 in mono_jefree (ptr=0x55ee6c7dfad0) at src/jemalloc.c:2290
%15 0x000055ee6b2c6c68 in monoeg_g_free (ptr=0x55ee6c7dfad0) at gmem.c:83
%16 0x000055ee6adefafa in emit_method_code (acfg=0x7f4ae0986000, cfg=0x7f4ae0ad4e00) at aot-compiler.c:6243
%17 0x000055ee6adf9534 in emit_code (acfg=0x7f4ae0986000) at aot-compiler.c:9704
%18 0x000055ee6ae04a9d in emit_aot_image (acfg=0x7f4ae0986000) at aot-compiler.c:13535
%19 0x000055ee6ae03b7c in mono_compile_assembly (ass=0x7f4ae09a9300, opts=374434303, aot_options=0x7f4ae093c000 "bind-to-runtime-version,outfile=./../../class/lib/build-linux//Microsoft.CodeAnalysis.CSharp.dll.so", global_aot_state=0x7ffc7b0c6438) at aot-compiler.c:13357
```
* Nullable: remainder of globalization
* annotate windows specific files
* add TODO-NULLABLE next to TODO suggesting comments
* fixes after merging other globalization annotations, apply feedback
* fix windows
* address feedback
* fix windows
* fixes
* apply feedback
* remove workaround
* s_Invariant?, couple of Debug.Asserts
Signed-off-by: dotnet-bot <[email protected]>
* Nullable: NumberFormatInfo, CultureInfo
* fix build
* fix windows
* fix APPX issue
* another appx failure
* get rid of line split
* mark retVal as nullable
Signed-off-by: dotnet-bot <[email protected]>
We can now build `--with-core=only` on Linux or Mac with the same makefile, and have a functional `make nupkg` rule which attempts to produce a .nupkg of the runtime in the format consumed by core-setup.
* [corlib] Make NSLogWriter thread-safe.
* [corlib] Simplify newline checking in NSLogWriter.
Checking the StringBuilder buffer after writing to it instead of the input
string/char[] makes the code somewhat simpler.
* [corlib] Fix NSLogWriter.Write to take into account the index and count parameters.
CoreCLR allows reflection on methods with ByRef return types under certain conditions. The PR implementing it is https://github.com/dotnet/coreclr/pull/17639.
Mono already implemented most of it in the marshalling code but artificially blocked it in `ves_icall_InternalInvoke`. I changed the condition for the exception to match the condition in `emit_invoke_call` (marshal-ilgen.c). It's similar to the condition in CoreCLR, but it may not handle byref-like types correctly. It would have to fixed on both places if it turns out to be the case.
The remaining issue is that null references returned in ByRef are not checked and converted to correct exception. It should result in `NullReferenceException` with special message that doesn't get wrapped in `TargetInvocationException`. I put a TODO marker at a place where I think it should be handled, but it's a bit over my head to implement it.
/cc @vargaz
An icall declaration can now use the following attribute:
MonoDirectICallSymbolNameAttribute("MyNativeCFunction")
and if not already defined in icall-def.h and direct-icalls are used when
AOT compile the assembly, a direct call will use symbol defined in attribute
as the symbol for direct call. Using this attribute will also enable inlining
for the specific icall wrapper generating inlined direct icalls to external
defined native methods.
This feature is valuable when embedding the runtime and extend with icalls
added using mono_add_internal_call. By applying the above attribute, it is
possible to do AOT with direct-icalls and get static linked icalls without
the need to extend and rebuild cross compiler in the process.
* Add support for core SharedArrayBuffer.
- Add SharedArrayBuffer implementation
- Add SharedArrayBuffer constructors to TypedArray implemenations
* Add tests for SharedArrayBuffer.
- Refactored the test code to add specific test files for Core Bindings. This separates the tests into specific test regions.
- Moved bindings tests from the http tests to the core bindings tests.
* Cleanup debugging code.
It contained DllImports to some Windows APIs which causes issues on xamarin-macios which expects all pinvokes to exist.
Turns out we don't even need that file.
* Add JavaScript Core support for Function
* Fix compile
* Fix bindings tests
* Add Core Function tests
- TestFunctionSum
- Create a new Function with a body that returns the sum of two numbers
- call method `Call` on the sum Function
- TestFunctionApply
- Return the Math.min function from the Global Math space
- Call method `Apply` on the min Function to return the minimum value from an array.
* Use Core Function to determine streaming instead of calling InvokeJS which uses javascript Eval.
* Move nested array checking to managed code as per review comments.
- This removes the icall as per review suggestion.
* [wasm] Generate the pinvoke tables.
Put the default non-linked version of the table into version control, usable by the interpreter only build.
* Add a comment to the generated file.
Create a new data structure: MonoImageStorage.
It is an object that will have the responsibility for the raw data of a
MonoImage. It has a string key and a refcount that is used to share a
MonoImageStorage between multiple MonoImage objects.
The reason we need this is because the current MonoAssembly/MonoImage design is
broken for multiple domains and (more evidently) it will be broken when we add
AssemblyLoadContext support. The issue is that a MonoImage may be shared
between multiple domains (or ALCs), but it has a 'references' field which
points to a single other MonoAssembly.
This is a problem because the references of an image may be resolved
differently in different domains (or ALCs).
The eventual solution will be to stop sharing MonoImages based on path name (or
in the case of in-memory images based on a name made up from the address of the
byte blob).
However we still don't want to open the same data more than once (if we're on a
config where we don't have mmap, we malloc some memory and dump the data in
there - we don't want to do that multiple times for the same file).
So the solution is to create this MonoImageStorage object and make it possible
to share it based on path name, but don't give it any responsibilities except
for owning the memory. It doesn't know anything about metadata or assemblies
or any of that stuff - it just owns a chunk of memory and knows when and how to
free it.
This commit just adds the MonoImageStorage object and wires it up in MonoImage
loading. There should be no observable behavioral changes from this commit.
* [netcore] Implement RuntimeImports.RhBulkMoveWithWriteBarrier
* Add the actual write barrier to RhBulkMoveWithWriteBarrier
* Rename ves_icall_System_Runtime_RuntimeImports_Memmove_wbarrier to ves_icall_System_Runtime_RuntimeImports_RhBulkMoveWithWriteBarrier
Fixes 15 failures in System.Memory.Tests.
From: https://llvm.org/docs/LangRef.html
```
Furthermore, the semantics of the noalias attribute on return values are stronger than the semantics of the attribute when used on function arguments. On function return values, the noalias attribute indicates that the function acts like a system memory allocation function, returning a pointer to allocated storage disjoint from the storage for any other object accessible to the caller.
```
Think this probably helps LLVM reorder the method calls, telling it that allocating won't return a pointer to something else on the stack.
The aggregate modifiers container is owned by a MonoImageSet. The aggregate
modifiers may be shared by more than one MonoType that happen to need the same
modifiers. (Similar to how generic instantiations are shared)
When the image set is cleaned up the aggregate modifier cleanup will free the
types in the aggregate modifiers container (the call to
free_aggregate_modifiers in mono_metadata_clean_for_image).
With this extra call to free_aggregate_modifiers in mono_metadata_free_type may
cause the same memory to be freed twice.
Fixes https://github.com/mono/mono/issues/13775
* Add MONO_PATCH_INFO_LAZY_FETCH_ADDR and MONO_PATCH_INFO_TRAMP_ADDR.
These are int-based alternatives to MONO_PATCH_INFO_JIT_ICALL_ADDR + sprintf.
This clarifies work to remove the JIT icall hash tables that are used with:
MONO_PATCH_INFO_JIT_ICALL
MONO_PATCH_INFO_JIT_ICALL_ADDR
MONO_PATCH_INFO_JIT_ICALL_ADDR_NOCALL
making it easier to make those enum-based, indexing a fixed size dense array.
As a related side effect, that could be achieved otherwise,
this fixes apparent leaks.
And clarify MONO_PATCH_INFO_JIT_ICALL_ADDR_NOCALL slightly, by making
searches for it find one more hit.
That is, those three enums are mostly used with a fixed set of strings.
Except for as in this PR where they are used with indefinite strings.
The removal of the indefinite strings eases replacement of the fixed
set of strings with an enum.
* Bump MONO_AOT_FILE_VERSION.
* PR feedback: Remove unused code and historical commentary.
Also updated PR title with more verbose actual names of the new symbols:
`MONO_PATCH_INFO_TRAMPOLINE_FUNC_ADDR`
`MONO_PATCH_INFO_SPECIFIC_TRAMPOLINE_LAZY_FETCH_ADDR`
* Add missing break.
* [netcore] Fix Marshal test failures.
* [netcore] Add exclusions for System.Runtime.InteropServices.Tests.
* [netcore] Make an icall internal.
* Add more test exclusions.
* Address review comments.
1. it doesn't touch the managed runtime
2. it may be called (arguably it ought to be called only) before the runtime is
initialized (and in particular before the current thread is attached).
This method returns a function pointer that can be called with a calli instruction. On interpreter we use a pointer to InterpMethod while on jit we use the native code address. Normally, GetFunctionPointer should return the InterpMethod pointer if called from interp or the native code address if called from jit. Since we don't have any information about the execution engine of the caller, we solve this by intrinsifying all these calls, that happen in the interpreter.
Passing such function pointers between jitted and interp code is probably still unreliable.
Fixes https://github.com/mono/mono/issues/13654
This test https://github.com/dotnet/coreclr/blob/1649da12db73c8c07513c9168cb30ec70c310063/tests/src/Regressions/coreclr/20452/twopassvariance.il
On mono we were ordering the list of implemented interfaces and shouldn't do this as it's described on ECMA-335 Section II.12.2.1 and II.12.2.2.
The ordering was removed on ilasm and some changes were needed on metadata either because it's not ordered anymore.
There were two bugs, one on ilasm and other on class initialization on mono.
### ILAsm ###
If we get the .exe generated by .net Framework it doesn't execute on mono without modifications because we ordered the implemented interfaces list by id. And doing that the order defined on source code is ignored.
If we get the .exe generated by Mono it doesn't execute on .net Framework because on IlAsm we order the list and loose the order defined on source code.
The interfaces on metadata before these changes on IlAsm were like this:
```
1: Fooer1 implements class IFoo<class Derived>
2: Fooer1 implements class IFoo<class Base>
3: Fooer2 implements class IFoo<class Base>
4: Fooer2 implements class IFoo<class Derived>
5: Fooer3 implements class IFoo<object>
6: Fooer3 implements class IBar<class Derived>
7: Fooer3 implements class IBaz<class Base>
8: IBar implements class IFoo<!0>
9: IBaz implements class IFoo<!0>
10: Fooer4 implements IQux
11: Fooer4 implements class IBar<class Derived>
12: IQux implements class IFoo<class Base>
13: IQux implements class IFoo<class Derived>
14: Fooer5 implements IQux
15: Fooer5 implements class IBar<class Derived>
16: Fooer6 implements class IBar<class Base>
17: Fooer6 implements class IFoo<class Base>
18: Fooer7 implements class IFoo<class Base>
19: Fooer7 implements class IBar<class Base>
```
And After the changes they follow the definition order which is the same order of IlAsm for .net framework and .net core.
```
1: IBar implements class IFoo<!0>
2: IBaz implements class IFoo<!0>
3: IQux implements class IFoo<class Base>
4: IQux implements class IFoo<class Derived>
5: Fooer1 implements class IFoo<class Base>
6: Fooer1 implements class IFoo<class Derived>
7: Fooer2 implements class IFoo<class Derived>
8: Fooer2 implements class IFoo<class Base>
9: Fooer3 implements class IBaz<class Base>
10: Fooer3 implements class IBar<class Derived>
11: Fooer3 implements class IFoo<object>
12: Fooer4 implements IQux
13: Fooer4 implements class IBar<class Derived>
14: Fooer5 implements class IBar<class Derived>
15: Fooer5 implements IQux
16: Fooer6 implements class IFoo<class Base>
17: Fooer6 implements class IBar<class Base>
18: Fooer7 implements class IBar<class Base>
19: Fooer7 implements class IFoo<class Base>
```
### Mono vtable layout ###
Internally when Mono computes the interface table for each class, it would previously sort the interfaces by `MonoClass:interface_id` which is a unique id assigned (roughly) in load order. This was incorrect because the order of the interfaces determines the final vtable for the class. This PR changes the interface table computation so that the interfaces of a single class are ordered in declaration order, following ECMA.
One consequence is that we can no longer use binary search with the interface id as the key to find out if a class implements an interface. We have to do a linear scan. On the other hand, we expect that the number of interfaces implemented by a single class is small in practice.
This is how the Vtable for Fooer1 and Fooer2 looks like after the changes:
```
*** Vtable for class 'Fooer1' at "FINALLY" (size 6)
[O][000][INDEX 000] object:ToString () [0x7fd78681db20]
[O][001][INDEX 001] object:GetHashCode () [0x7fd78681daf8]
[O][002][INDEX 002] object:Finalize () [0x7fd78681dad0]
[O][003][INDEX 003] object:Equals (object) [0x7fd78681daa8]
[I][004][INDEX 000] IFoo:Gimme () [0x7fd786503800]
[I][005][INDEX 000] IFoo:Gimme () [0x7fd7865038c0]
Packed interface table for class Fooer1 has size 2
[000][UUID 078][SLOT 004][SIZE 001] interface IFoo[Base]
[001][UUID 079][SLOT 005][SIZE 001] interface IFoo[Derived]
```
and
```
*** Vtable for class 'Fooer2' at "FINALLY" (size 6)
[O][000][INDEX 000] object:ToString () [0x7fd78681db20]
[O][001][INDEX 001] object:GetHashCode () [0x7fd78681daf8]
[O][002][INDEX 002] object:Finalize () [0x7fd78681dad0]
[O][003][INDEX 003] object:Equals (object) [0x7fd78681daa8]
[I][004][INDEX 000] IFoo:Gimme () [0x7fd7865038c0]
[I][005][INDEX 000] IFoo:Gimme () [0x7fd786503800]
Packed interface table for class Fooer2 has size 2
[000][UUID 079][SLOT 004][SIZE 001] interface IFoo[Derived]
[001][UUID 078][SLOT 005][SIZE 001] interface IFoo[Base]
```
This is how the VTable for Fooer1 and Fooer2 looks like before the changes, as you can see Packed interface table are the same in Fooer1 and Fooer2, so the behavior of both classes was the same:
```
Vtable for class 'Fooer1' at "FINALLY" (size 6)
[O][000][INDEX 000] object:ToString () [0x7fcf3801d920]
[O][001][INDEX 001] object:GetHashCode () [0x7fcf3801d8f8]
[O][002][INDEX 002] object:Finalize () [0x7fcf3801d8d0]
[O][003][INDEX 003] object:Equals (object) [0x7fcf3801d8a8]
[I][004][INDEX 000] IFoo:Gimme () [0x7fcf37c29880]
[I][005][INDEX 000] IFoo:Gimme () [0x7fcf37c29940]
Packed interface table for class Fooer1 has size 2
[000][UUID 076][SLOT 004][SIZE 001] interface IFoo[Base]
[001][UUID 077][SLOT 005][SIZE 001] interface IFoo[Derived]
```
```
Vtable for class 'Fooer2' at "FINALLY" (size 6)
[O][000][INDEX 000] object:ToString () [0x7fcf3801d920]
[O][001][INDEX 001] object:GetHashCode () [0x7fcf3801d8f8]
[O][002][INDEX 002] object:Finalize () [0x7fcf3801d8d0]
[O][003][INDEX 003] object:Equals (object) [0x7fcf3801d8a8]
[I][004][INDEX 000] IFoo:Gimme () [0x7fcf37c29940]
[I][005][INDEX 000] IFoo:Gimme () [0x7fcf37c29880]
Packed interface table for class Fooer2 has size 2
[000][UUID 076][SLOT 005][SIZE 001] interface IFoo[Base]
[001][UUID 077][SLOT 004][SIZE 001] interface IFoo[Derived]
```
[llvm] Fix llvm JIT.
<!--
Thank you for your Pull Request!
If you are new to contributing to Mono, please try to do your best at conforming to our coding guidelines http://www.mono-project.com/community/contributing/coding-guidelines/ but don't worry if you get something wrong. One of the project members will help you to get things landed.
Does your pull request fix any of the existing issues? Please use the following format: Fixes #issue-number
-->
* Add cpu counter
* Fix windows build
* Make the counter just return current CPU usage as %
* Add Unix
* Fix unix build
* Some cleanup
* rename
* fixing some build errors
* some cleanup
* remove unused using
* more cleanup
* newline
* Address PR feedback
* more pr feedback
* More feedback
Signed-off-by: dotnet-bot <[email protected]>
Match Mono's behavior by changing the Auto character set to mean UTF-8 on non-Windows platforms (new behavior) and UCS-2/UTF-16 on Windows (current behavior).
Fixes#23464Fixesdotnet/corefx#32442
Impact of breaking change: It is highly unlikely that anyone is actively using current behavior since it is inconsistent with Mono and doesn't match any native system APIs on non-Windows platforms (they're all UTF-8 based).
We will need to update our documentation to reflect this updated behavior.
Signed-off-by: dotnet-bot <[email protected]>
The OnProcessExit was iterating over the s_contextsToUnload and calling
Unload on each of the contexts. However the s_contextsToUnload is a
misnomer, it contains all the contexts, both collectible and
non-collectible ones. So we were calling Unload on non-collectible
contexts too there. That resulted in InvalidOperationException being
thrown from the OnProcessExit.
This change fixes that by calling unload only on ALCs whose
IsCollectible returns true. It also renames the s_contextsToUnload to
s_allContexts.
Close#22902
Signed-off-by: dotnet-bot <[email protected]>
We want to avoid calling stackval_to/from methods. Also include the field address in the instruction stream. This makes static vtype field access 10x faster.
Refactor the code a bit.
[gsharedvt] Fix sizeof opcode
We were setting the stack type by using another global local named ins.
This lead to missing methods in the aot image due to invalid IL. Fixes https://github.com/mono/mono/issues/13479.
Adds 5 quick checks, on prefix-tests tag:
* Build a basic file with mcs
* Build a basic file with Roslyn
* AOT an assembly
* LLVM AOT an assembly (loadedllvm)
* LLVM JIT an assembly (loadedllvm)
This should let us spot regressions affecting installed prefixes sooner
These console PInvokes are used in nearly every app. Omitting unnecessary SetLastError
will save us from generating marshalling stub for them.
Signed-off-by: dotnet-bot <[email protected]>
Also make cimag/creal macros for guaranteed inlining -- instead of passing structs by value.
And add spaces before parens to follow coding convention (which is designed
to ease low tech searches for entire words).
This was agreed to when the code was first ported to C++ but got dropped.
The compromise though was to put the defines relatively late to limit their impact.
In this case we relied on nunitlite to be compiled when the parent profile is compiled. That doesn't work however if we compile the child profile first (e.g. during a parallel make) since it'll
complain about missing nunitlite.dll.
We now compile nunitlite in the child profiles too even though we don't use it.
This is an alternative solution to what I tried in https://github.com/mono/mono/pull/5357 since I couldn't get it to work and this should be the easier fix.
It allows us to finally remove the `-i` (ignore errors) from the CI scripts and catch test assembly breakages in mobile profiles which we ignored before.
* fix test
* fix tests
* Fix ArgumentNullException empty ParamName
* System.Text.Tests.StringBuilderTests are fine
* fix BlockCopy test
* enable more tests
* Enable more tests
* address feedback
* [System] Clear MONO_CFG_DIR before starting mcs in CodeDOM CSharpCodeGenerator
This showed up in the xamarin-macios 2019-02 integration.
When you used Xslt with a custom C# script in a Xamarin.Mac Full application
we'd try to launch mcs via CodeDOM to compile the script.
This started failing in 2019-02 because System.Native (mono-native) wasn't found,
the reason is that XM sets MONO_CFG_DIR to point to the Contents/MonoBundle
in the app bundle where it packages the global config file with dllmaps.
However the started Mono wasn't able to find mono-native since the config
there uses a different name for the file.
To prevent that we now clear MONO_CFG_DIR so it uses the default value for the
mono process we're starting (note that setting it to String.Empty doesn't work)
* Make it specific to XM 4.5
* [netcore] Add error checking for assembly version parsing, pass the full 32 bit value to managed code.
* [netcore] Implement RuntimeType.StructLayoutAttribute.
* [runtime] Implement RuntimeFieldHandle.GetValueDirect (), convert SetValueDirect () to handles.
* [netcore] Implement Assembly.GetFiles ()/GetForwardedTypes ().
* [runtime] Handle 'ARM' as a valid architecture in assembly name parsing.
* [netcore] Implement ParameterInfo.MarshalAsAttribute, AssemblyName.EscapedCodeBase.
* [netcore] Add exclusions for System.Reflection.Tests.
* [netcore] Update test exclusions.
* [netcore] Fix CustomAttributeExtensions.
* [netcore] Add a list of passing test suites.
* [netcore] Add a missing argument check.
* Update Object.cs
* Fix a few deficiencies in mono_path_filename_in_basedir.
The main one is that strstr is too loose.
Add tests.
Make it case insensitive on Win32 and allow some denormal input.
Note that case sensitivity is a function of file system, not
operating system, but a lot of code gets it wrong, so here does too.
Add more data to asserts.
Account for test reality -- filename is not absolute.
The Win32 code paths can be tested on Unix and vice versa
w/o compromise. See https://github.com/mono/mono/pull/13707
* Reduce testing per PR feedback.
* Edit/add comments.
This is an attempt to clarify and hopefully doesn't backfire.
* Try to make tests clearer, and stop skipping some oops.
* PR: Restore case insensitivity testing on Windows. Tweak a little bit.
* Include the .c file instead of .h to fix some builds.
SUBDIRS=$(build_with_msvc) mk po $(libgc_dir) llvm mono $(ikvm_native_dir) support data runtime scripts man samples $(tools_dir)$(build_without_msvc)$(docs_dir) acceptance-tests
# Keep in sync with SUBDIRS
DIST_SUBDIRS=$(build_with_msvc) m4 mk po $(libgc_dir) llvm mono ikvm-native support data runtime scripts man samples tools $(build_without_msvc) docs acceptance-tests
DIST_SUBDIRS=$(build_with_msvc) m4 mk po $(libgc_dir) llvm mono ikvm-native support data runtime scripts man samples tools $(build_without_msvc) docs acceptance-tests netcore
all:update_submodules
all:$(update_submodules)
SUBMODULE_ERROR='Could not recursively update all git submodules. You may experience compilation problems if some submodules are out of date'
<!-- Licensed to the .NET Foundation under one or more agreements. The .NET Foundation licenses this file to you under the MIT license. See the LICENSE file in the project root for more information. -->
<Project Sdk="Microsoft.NET.Sdk">
<!--
This MSBuild file is intended to be used as the body of the default
publishing release pipeline. The release pipeline will use this file
to invoke the PushToStaticFeed task that will read the build asset
manifest and publish the assets described in the manifest to
<!-- Licensed to the .NET Foundation under one or more agreements. The .NET Foundation licenses this file to you under the MIT license. See the LICENSE file in the project root for more information. -->
<Project Sdk="Microsoft.NET.Sdk">
<!--
This MSBuild file is intended to be used as the body of the default
publishing release pipeline. The release pipeline will use this file
to invoke the PublishSymbols tasks to publish symbols to MSDL and SymWeb.
Parameters:
- PDBArtifactsDirectory : Full path to directory containing PDB files to be published.
- BlobBasePath : Full path containing *.symbols.nupkg packages to be published.
- DotNetSymbolServerTokenMsdl : PAT to access MSDL.
- DotNetSymbolServerTokenSymWeb : PAT to access SymWeb.
- DotNetSymbolExpirationInDays : Expiration days for published packages. Default is 3650.
!!! Changes made in this directory are subject to being overwritten by automation !!!
The files in this directory are shared by all Arcade repos and managed by automation. If you need to make changes to these files, open an issue or submit a pull request to https://github.com/dotnet/arcade first.
<!-- Licensed to the .NET Foundation under one or more agreements. The .NET Foundation licenses this file to you under the MIT license. See the LICENSE file in the project root for more information. -->
<Project Sdk="Microsoft.NET.Sdk">
<!--
This MSBuild file is intended to be used as the body of the default
publishing release pipeline. The release pipeline will use this file
to invoke the the SignCheck tool to validate that packages about to
be published are correctly signed.
Parameters:
- PackageBasePath : Directory containing all files that need to be validated.
- SignCheckVersion : Version of SignCheck package to be used.
- SignValidationExclusionList : ItemGroup containing exclusion list to be forwarded to SignCheck.
- EnableJarSigningCheck : Whether .jar files should be validated.
- EnableStrongNameCheck : Whether strong name check should be performed.
-->
<PropertyGroup>
<TargetFramework>netcoreapp2.1</TargetFramework>
</PropertyGroup>
<!--
From 'Signing.props' we import $(SignValidationExclusionList)
echo"Creates a toolchain and sysroot used for cross-compiling for Android."
echo.
echo"Usage: $0 [BuildArch] [ApiLevel]"
echo.
echo"BuildArch is the target architecture of Android. Currently only arm64 is supported."
echo"ApiLevel is the target Android API level. API levels usually match to Android releases. See https://source.android.com/source/build-numbers.html"
echo.
echo"By default, the toolchain and sysroot will be generated in cross/android-rootfs/toolchain/[BuildArch]. You can change this behavior"
echo"by setting the TOOLCHAIN_DIR environment variable"
echo.
echo"By default, the NDK will be downloaded into the cross/android-rootfs/android-ndk-$__NDK_Version directory. If you already have an NDK installation,"
echo"you can set the NDK_DIR environment variable to have this script use that installation of the NDK."
echo"By default, this script will generate a file, android_platform, in the root of the ROOTFS_DIR directory that contains the RID for the supported and tested Android build: android.21-arm64. This file is to replace '/etc/os-release', which is not available for Android."
exit1
}
__ApiLevel=21# The minimum platform for arm64 is API level 21
echo"BuildArch can be: arm(default), armel, arm64, x86"
echo"LinuxCodeName - optional, Code name for Linux, can be: trusty, xenial(default), zesty, bionic, alpine. If BuildArch is armel, LinuxCodeName is jessie(default) or tizen."
echo"lldbx.y - optional, LLDB version, can be: lldb3.9(default), lldb4.0, lldb5.0, lldb6.0 no-lldb. Ignored for alpine"
echo"--skipunmount - optional, will skip the unmount of rootfs folder."
<!-- Licensed to the .NET Foundation under one or more agreements. The .NET Foundation licenses this file to you under the MIT license. See the LICENSE file in the project root for more information. -->
<!-- Copyright (c) Microsoft. All Rights Reserved. Licensed under the Apache License, Version 2.0. See License.txt in the project root for license information. -->
# Please remember to update the documentation if you make changes to these parameters!
parameters:
HelixSource:'pr/default'# required -- sources must start with pr/, official/, prodcon/, or agent/
HelixType:'tests/default/'# required -- Helix telemetry which identifies what type of data this is; should include "test" for clarity and must end in '/'
HelixBuild:$(Build.BuildNumber) # required -- the build number Helix will use to identify this -- automatically set to the AzDO build number
HelixTargetQueues:''# required -- semicolon delimited list of Helix queues to test on; see https://helix.dot.net/ for a list of queues
HelixAccessToken:''# required -- access token to make Helix API requests; should be provided by the appropriate variable group
HelixConfiguration:''# optional -- additional property attached to a job
HelixPreCommands:''# optional -- commands to run before Helix work item execution
HelixPostCommands:''# optional -- commands to run after Helix work item execution
WorkItemDirectory:''# optional -- a payload directory to zip up and send to Helix; requires WorkItemCommand; incompatible with XUnitProjects
WorkItemCommand:''# optional -- a command to execute on the payload; requires WorkItemDirectory; incompatible with XUnitProjects
WorkItemTimeout:''# optional -- a timeout in seconds for the work item command; requires WorkItemDirectory; incompatible with XUnitProjects
CorrelationPayloadDirectory:''# optional -- a directory to zip up and send to Helix as a correlation payload
XUnitProjects:''# optional -- semicolon delimited list of XUnitProjects to parse and send to Helix; requires XUnitRuntimeTargetFramework, XUnitPublishTargetFramework, XUnitRunnerVersion, and IncludeDotNetCli=true
XUnitWorkItemTimeout:''# optional -- the workitem timeout in seconds for all workitems created from the xUnit projects specified by XUnitProjects
XUnitPublishTargetFramework:''# optional -- framework to use to publish your xUnit projects
XUnitRuntimeTargetFramework:''# optional -- framework to use for the xUnit console runner
XUnitRunnerVersion:''# optional -- version of the xUnit nuget package you wish to use on Helix; required for XUnitProjects
IncludeDotNetCli:false# optional -- true will download a version of the .NET CLI onto the Helix machine as a correlation payload; requires DotNetCliPackageType and DotNetCliVersion
DotNetCliPackageType:''# optional -- either 'sdk' or 'runtime'; determines whether the sdk or runtime will be sent to Helix; see https://raw.githubusercontent.com/dotnet/core/master/release-notes/releases.json
DotNetCliVersion:''# optional -- version of the CLI to send to Helix; based on this: https://raw.githubusercontent.com/dotnet/core/master/release-notes/releases.json
EnableXUnitReporter:false# optional -- true enables XUnit result reporting to Mission Control
WaitForWorkItemCompletion:true# optional -- true will make the task wait until work items have been completed and fail the build if work items fail. False is "fire and forget."
IsExternal:false# [DEPRECATED] -- doesn't do anything, jobs are external if HelixAccessToken is empty and Creator is set
Creator:''# optional -- if the build is external, use this to specify who is sending the job
DisplayNamePrefix:'Run Tests'# optional -- rename the beginning of the displayName of the steps in AzDO
condition:succeeded() # optional -- condition for step to execute; defaults to succeeded()
continueOnError:false# optional -- determines whether to continue the build if the step errors; defaults to false
# Report error - the developer environment is initialized with incompatible VS version.
throw"Developer Command Prompt for VS $($env:VisualStudioVersion) is not recent enough. Please upgrade to $vsMinVersionStr or build from a plain CMD window"
}
}
# Locate Visual Studio installation or download x-copy msbuild.
1. Pick a revision of roslyn to use. Ideally, use a revision that Visual Studio is using, but any revision should do.
2. Make a new folder in roslyn-binaries for this revision. Copy over the contents of the *previous folder* into this folder.
3. Make a temporary commit.
4. Find that revision in the 'roslyn compilers' nuget feed: https://dotnet.myget.org/feed/roslyn/package/vsix/7922692f-f018-45e7-8f3f-d3b7c0262841
5. Download the vsix file and unpack it (it's a ZIP file).
6. Inside the vsix you will find lots of metadata, along with executables for this build of roslyn. We want to copy most of these into a new folder in roslyn-binaries, so...
7. Copy all the dll, exe and rsp files from the vsix into our new folder, overwriting the ones from the previous version. It's time to look at the diff.
8. The vsix contains DLLs we don't need to ship. Ideally we will not need to add any new files, but roslyn sometimes picks up new dependencies. Figuring out what to keep will require looking at what facades we ship (note that this is profile-specific - check Facades/subdirs.make) and which of these DLLs are actually used by roslyn instead of Visual Studio.
9. Remove any DLLs from the new folder that we don't need. In practice your commit should be adding 1-2 DLLs at most or something very bad happened.
10. Examine the diffs for the .rsp files. We ship our own versions of the .rsps, so we don't want to carry over the vsix versions, but we *do* want to make note of anything new they have added to the rsp files and add it to ours if necessary.
11. Add and commit your new folder. Push it to a branch and make a PR against roslyn-binaries, then get it merged.
----
11. Okay, we have a new roslyn-binaries. The next step is to prepare a mono bump. Do this by updating the roslyn-binaries submodule to point to the new revision of roslyn-binaries. You can temporarily point it to your own branch for testing and run CI builds against that, just don't merge that into master.
12. Once you commit the submodule bump to your local repository, you can do test builds. **If you do test builds without committing the bump, the build process will erase the bump.**
13. Now we need to change the roslyn paths used by the build process - we're still using the older roslyn folder. We do this by changing the CSC_LOCATION and VBCS_LOCATION variables in configure.ac. They should be something like ```Microsoft.Net.Compilers/3.1.0/csc.exe```, change the version number in those paths, save, and make a local commit. Then do the full autogen, make, make install routine and do some basic tests. This should work.
14. Next, make sure the bump didn't break csi. If you managed the .rsp files correctly in the last step, it should work. This has test coverage, so just pay attention to the automated tests for it.
15. Ensure the 'Linux x64 - prefix sanity checks' lane passed on CI, even if it's currently disabled or not required. This is required for a bump. If it isn't visible on the PR statuses, manually trigger the lane using Jenkins.
16. Ensure that 'make install' output has working csc. Install to a prefix is fine.
17. Verify that the compiler server works, even if it is currently off by default. You can do this by editing your ```mcs/build/config.make``` and changing ENABLE_COMPILER_SERVER to ?=1, doing make clean, and then make -j 1. Use 'top' with the O->COMMAND=mono filter to identify whether the compiler server is being used and working. Also, for safety's sake make sure that all three of the roslyn-binaries paths in there match the expected version.
18. Double-check that you have a clean (points to roslyn-binaries master) PR ready to commit. Now it's msbuild time.
----
19. Prepare a PR against mono/msbuild's xplat-master branch:
20. Modify the CompilerToolsVersion and MicrosoftNetCompilersVersion properties in ```eng/Packages.props``` to point to the same new roslyn packages. The version strings have weird textual info in them so you'll have to figure out the new version string.
21. Verify that a build of this new msbuild succeeds.
22. Verify that you can manually run the MSBuild.dll from this new build using a new source build of mono (i.e. with new roslyn installed to a prefix)
23. Get your PR merged.
----
24. Now that msbuild is updated, we make a final change to the mono PR:
25. Modify ```packaging/MacSDK/msbuild.py```, updating the ```revision =``` line's commit hash to be the commit hash of your post-merge xplat-master commit.
26. Get your PR merged.
----
27. Now we need to update the linux packaging and deploy new packages (this involves changing the https://github.com/mono/linux-packaging-msbuild repository)
28. import upstream xplat-master changes by giving the CI package job a URL to a tarball containing the contents of the latest xplat-master revision.
29. trigger CI lane builds to generate a new package
30. identify necessary changes to the debian/rules file for the build to succeed, if step 29 failed
31. identify any new patches required to make latest xplat-master compatible with debian, including build script changes (see above)
32. eventually the lane succeeds. you should have updated linux packages for debian and ubuntu now.
----
33. Follow up with the VSfM team to make sure they know that roslyn is being bumped, it may break them on the next integration. Ensure the rest of the team knows roslyn has been bumped. Keep an eye out for any new bug reports about issues with csc or msbuild.
@@ -246,12 +246,6 @@ program that comes with Mono, and calling it like this:
.fi
.TP
.Illcopts=[options]
Use this option to add more flags to the built-in set of flags passed to the
LLVM system compiler. If not provided, it will fallback to
.I-mcpu=generic
on x86 systems.
.TP
.Illvm-outfile=[filename]
Gives the path for the temporary LLVM bitcode file created during AOT.
.Idedup
@@ -273,6 +267,9 @@ Print the architecture the AOT in this copy of Mono targets and quit.
Generates all required wrappers, so that it is possible to run --interpreter without
any code generation at runtime. This option only makes sense with \fBmscorlib.dll\fR.
Embedders can set
.TP
.Idepfile=[filename]
Outputs a gcc -M style dependency file.
.nf
mono_jit_set_aot_mode (MONO_AOT_MODE_INTERP);
@@ -646,6 +643,10 @@ running a 1.1 program on a 2.0 version:
Configures the virtual machine to be better suited for server
operations (currently, allows a heavier threadpool initialization).
.TP
\fB--use-map-jit\fR
Instructs Mono to generate code using MAP_JIT on MacOS. Necessary for
bundled applications.
.TP
\fB--verify-all\fR
Verifies mscorlib and assemblies in the global
assembly cache for valid IL, and all user code for IL
@@ -658,6 +659,9 @@ mscorlib and assemblies located on the global assembly cache.
\fB-V\fR, \fB--version\fR
Prints JIT version information (system configuration, release number
and branch names if available).
.TP
\fB--version=number\fR
Print version number only.
.SHDEVELOPMENTOPTIONS
@@ -1175,23 +1179,26 @@ If set, tells mono to attempt using native asynchronous I/O services. If not
set, a default select/poll implementation is used. Currently epoll and kqueue
are supported.
.TP
\fBMONO_THREADS_SUSPEND\fR
Selects a mechanism that Mono will use to suspend threads. May be set to
"preemptive", "coop", or "hybrid". Threads may need to be suspended by the
debugger, or using some .NET threading APIs, and most commonly when the SGen
garbage collector needs to stop all threads during a critical phase of garbage
collection. Preemptive mode is the mode that Mono has used historically, going
back to the Boehm days, where the garbage collector would run at any point and
suspend execution of all threads as required to perform a garbage collection.
The cooperative mode on the other hand requires the cooperation of all threads
to stop at a safe point. This makes for an easier to debug garbage collector.
As of Mono 4.3.0 it is a work in progress, and while it works, it has not been
used extensively. This option enables the feature and allows us to find spots
that need to be tuned for this mode of operation. Hybrid mode is a combination
of the two that retains better compatability with scenarios where Mono is
embedded in another application: threads that are running managed code or code
that comprises the Mono runtime will be cooperatively suspended, while threads
running embedder code will be preemptively suspended.
\fBMONO_THREADS_SUSPEND\fR Selects a mechanism that Mono will use to suspend
threads. May be set to "preemptive", "coop", or "hybrid". Threads may need to
be suspended by the debugger, or using some .NET threading APIs, and most
commonly when the SGen garbage collector needs to stop all threads during a
critical phase of garbage collection. Preemptive mode is the mode that Mono
has used historically, going back to the Boehm days, where the garbage
collector would run at any point and suspend execution of all threads as
required to perform a garbage collection. The cooperative mode on the other
hand requires the cooperation of all threads to stop at a safe point. This
makes for an easier to debug garbage collector and it improves the stability of
the runtime because threads are not suspended when accessing critical
resources. In scenarios where Mono is embedded in another application,
cooperative suspend requires the embedder code to follow coding guidelines in
order to cooperate with the garbage collector. Cooperative suspend in embedded
Mono is currently experimental. Hybrid mode is a combination of the two that
retains better compatability with scenarios where Mono is embedded in another
application: threads that are running managed code or code that comprises the
Mono runtime will be cooperatively suspended, while threads running embedder
code will be preemptively suspended. Hybrid suspend is the default on some
desktop platforms.
Alternatively, coop and hybrid mode can be enabled at compile time by using the
--enable-cooperative-suspend or --enable-hybrid-suspend flags, respectively,
Loaded 100 of 2088 files, more files were not shown because too many files have changed in this diff.
Show more
Reference in new issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.