Fixed code to reenable some warnings in CoreCLR #33902

@Tratcher

- add debug information for `Assert` failures in `MaximumOSVersionTest` - update src/TestingUtils/Microsoft.AspNetCore.Testing/test/MaximumOSVersionTest.cs - co-Authored-By: @Tratcher - try with VS2019 queues - skip `MaximumOSVersion` tests on .NET due to xunit issue - co-authored-by: @Tratcher Commit migrated from dotnet/extensions@26ff835

* Remove Serilog dependency in extensions * Add xunit logging for shutdown tests * Need to remove dependency on AspNetCore.Testing and remove DumpCollector Commit migrated from dotnet/extensions@540b4e8

Commit migrated from dotnet/extensions@6171c14

…x-logging-event-source Fix LoggingEventSource to handle null strings Commit migrated from dotnet/extensions@8d0fa05

Commit migrated from dotnet/extensions@5bcbaa2

…dotnet/extensions#2940) * Add environment var to allow disabling live reload in default builder * Switch to using hostingContext, rename flag to fit nomenclature. * Change flag to memory data source. Change content root default instead of runtime default. * Update config key to be hierarchical. Change await on positive case to be longer and cancel using the reload token. * Uncomment a development test case. * Apply suggestions from code review Co-authored-by: Chris Ross <Tratcher@Outlook.com> Commit migrated from dotnet/extensions@64140f9

…y with mono_gchandle_new_weakref_internal. (#34153) Looks like this was missed in: Unity-Technologies/mono@941a335 Co-authored-by: UnityAlex <UnityAlex@users.noreply.github.com>

Because the formerly uninitialized scratch context is now an input.

Co-authored-by: BrzVlad <BrzVlad@users.noreply.github.com>

Keeping the download of corelib artifacts to check if just corelib is sufficient to build libraires.

…34201) * Unwrap the TargetInvocationException thrown during an IDispatch. This matches the behavior of .NET Framework.

* Add notifyInstructionSetUsage api to jit interface- Add api to be used when the jit is to notify the VM of any usage of higher level instruction sets- Note that this change does nothing using the api. This is merely a scaffolding change to prepare for functional changes later - There is also scaffoling in the jit to JITDUMP any usage of the api * Code review feedback

…4221) This helper is idempotent and exception free, so enable it for value numbering. Also, no need to spill the entire eval stack for a virtual stub calls. Addresses part of #7723.

Revert the warn as error as it's causing the coreclr outerloop build to break. Investigated fixing the warnings but there are a mix of C# and IL warnings. The latter I'm less sure of the fix. Reverting to unblock and will investigate the warnings in parallel. closes #34220

* Add RuntimeInformation.RuntimeIdentifier This value returns the Runtime Identifier (RID) of the current machine. Contributes to #26780 * Include XML doc comments for the new property. * Consistently check for AppContext strings.

… doing starg.0. (#34081)  Co-authored-by: vargaz <vargaz@users.noreply.github.com>

Also unifying the style with the others that just throw.

Contributes to #33680

…4 (#34237) - ILLink.Tasks - 5.0.0-preview.3.20177.4 Co-authored-by: dotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com>

…34199) Because the key object generation was done in the algorithm-specific base class, the triplet of interface types was using the key instances in parallel. By moving the static variable (and initialization thereof) to each of the derived classes, the key objects are unique per class, which matches the test parallelism. Making the classes be part of the same test collection would also solve this problem, which would save on a few random keygens, but would likely overall take more time due to the number of tests that would be moved to sequential execution.

Hoping this fixes the test build.

This contains outdated information about tiered compilation preview

* Update README.md * Update homepage and add section to dogfooding * Link homepage to existing dogfooding.md page * Add section for Obtaining nightly builds of NuGet packages * Incorporate dotnet CLI commands

…4244) These are calling ToString on a StringBuilder, but StringBuilder.Append has a StringBuilder-based overload in netcoreapp. On other builds this will just fall back to the object-based overload, which will call ToString.

Left-over from Razzle build

Most places use TargetArchitecture already. These two properties seems to be always set to the same value.

Fixes #33581

…s.proj pretest is required to generate the Microsoft.NETCore.App.deps files for CoreRun to work correctly.

- Delete unnecessary msbuild properties (left-overs from the conversion to SDK projects) - Delete jit.settings.targets

…34262) Fixes mono/mono#18127.  Co-authored-by: vargaz <vargaz@users.noreply.github.com>

…needed. (#33955) In normal llvm mode, these variables are marked as indirect and stored to a stack slot so the EH code can access them. Co-authored-by: vargaz <vargaz@users.noreply.github.com>

* [interp] Intrinsify Type.IsValueType * [interp] Intrinsify InitBlockUnaligned

… mode. (#34265)  Co-authored-by: vargaz <vargaz@users.noreply.github.com>

…34095) * Bump maximum ICU library version we scan in System.Globalization.Native on Unix The maximum major version of ICU library that we start scanning at was set to 70 (50 as the minimum supported version + 20) long time ago. The available version of ICU is approaching that limit (it is at 67 now). This change bumps the max scanned version to 80 to give us enough head space for a couple of years (guessing based on the current release rate) * Reflect PR feedback

since pretest needs installer.tasks.dll

* Switch back to only handling synchronous exceptions in host components Add native hosting tests for components with unhandled exceptions. * Switch setting /EHa to be for coreclr instead of shared default

- __TargetOS -> TargetOS - __BuildType, BuildType -> Configuration - __BuildArch, BuildArch -> TargetArchitecture

Allow phases to declare what parts of jit state they might have modified, and suppress dumps and checks based on these declarations. Existing phases that don't know about this are handled by defaulting to reporting that everything might have been modified. Changed over all explicit phases to return status, as well as a handful of phases that are methods on the Compiler object. Also converted a few lambda phases into Compiler phase methods.

* Vectorise BitArray for ARM64 * Make algorithms endianness agnostic * Change loop counter to uint to prevent overflow * Fix CopyTo(bool[]) * Move constant variables outside the loop * More unsigned goodness * Use span.Clear instead of Fill(0)

* fix TLS13 procesing on windows * fix test * rework disabled renegotiation * update platform detection Co-authored-by: Tomas Weinfurt <furt@GokceOncu.middleeast.corp.microsoft.com>

* Fix debug symbol generation The debug symbol generation got recently broken. For most of the shared libraries, the debug symbols were stripped twice due to the fact that install_clr for them was invoked twice - once for the default install location and once for the sharedFramework location and the stripping was executed in both of them. First stripping stripped the symbols off the target binary and set so called debuglink in the binary to point to the symbol file. This debuglink includes a crc32 of the dbg symbols file. The second stripping tried to strip symbols from the already stripped binary. That resulted in a small dbg symbols file that didn't actually contain any useful symbols. Moreover, it is not possible to set a debuglink in a binary if it is already set there. So the second attempt failed and the crc was left set to the crc of the previous debug. Thus when debugger loads such a binary, it cannot find the debug symbols file, as the crc doesn't match. And even if it matched, the data would have no value. The fix is to modify install_clr so that it has an extra optional argument to specify the secondary install location and use just one install_clr per target. The function then does the stripping just once and the actual installation once or twice depending on the secondary location argumenbt presence. * Fix libraries and installer * Fix installer placement of lib files

Today, for volatile variable access, JIT generates "dmb oshld" which is a memory barrier with outer shareable domain. Coreclr uses [atomics and interlocked APIs](https://github.com/dotnet/runtime/blob/fcd862e06413a000f9cafa9d2f359226c60b9b42/src/coreclr/src/inc/volatile.h#L153) for memory safety and msvc / clang both generate inner shareable domain for them. Hence it is sensible for JIT that is compiled with [msvs](https://godbolt.org/z/8XEWYr) ./ [clang](https://godbolt.org/z/DMZi26) to generate inner shareable domain for volatile access. [gcc](https://godbolt.org/z/zSf4fS) too emits inner shareable domain for API that takes full system memory barrier.

…thods (#34270) * Allow coreclr_create_delegate to return pointers to NativeCallable methods Fixes #34266 * Simplify CorHost implementation * Updated error message

* Hoist volatile variable's access outside the loop `_tables` is a volatile object and it has `_countPerLock` volatile field. On ARM64, JIT generates memory barrier instruction "dmb" for every volatile variable access which could be expensive. This PR caches the volatile variables outside the loop and use cached local variables instead. Fixes: #34198

* Add System.Type converter for JsonSerializer * Address review feedback * Remove throw helper

…rReader (#34046) * Implement tag support for CborWriter and CborReader * Implement CBOR special value support * add nested special value tests * implement half-precision float decoding; address feedback * address style * remove dead code * add checks for CBOR tags in indefinite-length collections * Update src/libraries/System.Security.Cryptography.Encoding/tests/Cbor/CborWriter.Integer.cs Co-Authored-By: Jeremy Barton <jbarton@microsoft.com> Co-authored-by: Jeremy Barton <jbarton@microsoft.com>

A recent change caused this break :)

Improve LLVM 9 compatibility. Several intrinsics were removed in LLVM 7 and above; replace them with IR sequences that generate the desired instructions. Fix some type mismatch problems with the generated IR (in Sse2.StoreAlignedNonTemporal, Vector128.CreateScalarUnsafe, and Sse2.MaskMove); these were caught by running the coreclr hardware intrinsic JIT tests. Support the two-argument forms of Sse.SqrtScalar, Sse.ReciprocalScalar, and Sse.ReciprocalSqrtScalar. llvm::linkCoreCLRGC(), in llvm/CodeGen/GCs.h, was removed and replaced with llvm::linkAllBuiltinGCs(), in llvm/CodeGen/BuiltinGCs.h. In mono_llvm_dump_value and mono_llvm_dump_module, flush the c library's stdout stream before using (and flushing) LLVM's own buffered output functions.

There is an assert in FindCgroupPath that fires when hierarchy_root and cgroup_path_relative_to_mount are equal, which is the case for cgroups that are not named. This assert checks that the common path in those two variables ends with / which is only the case with named groups. We have never seen this assert to fire because cgroups initialization happens before the debugger support initialization in PAL and so asserts are disabled at that point. I am going to fix that in a separate PR. This problem was discovered with the standalone GC where the assert actually fires as it uses a plain C assert function. This change fixes the assert to account for the case when both the paths are the same.

* Extract `fgMergeBlockReturn`. Morph is already a very vague verb in the Jit, try to use a more precise one. * Add a function header. * Make `compCurBB` available for `fgMorphBlockReturn`. When we generate an assignment we could need to create a new assertion, that requires `compCurBB` to be available. * Delete `INVALID_POINTER_VALUE`. I would like to remove it because: 1) it was debug only; 2) there were no null checks for `compHndBBtab`, because it is a dependent variable so there was no need to distinguish valid null pointer from a bad invalid pointer; 3) that is the only place where this mechanism was used. * Allow to CSE the merge return ASG. I can't see a reason why it should not, there are no diffs. The issue with the previous version was that we did not actually know what we were marking: GT_ASG, GT_COMMA, something else? Was the idea to mark individual ASG under COMMA?

…d.Rent/Array constructor. Because it can behave differently in OSX, where it supports int.MaxValue array size. (#34040)

Fixes #34292

…34306) This reverts commit 4e30ff0.

With this, the IL SDK will continue to work when it comes to either single or multitargeting without restoring the projects in multi-targeting mode. This helps to not introduce a breaking change for 3.x SDKs.

Moved runtime restore + pretest.proj to teh tests.proj build.

- Updates InternalsVisibleTo files as well - Setup Logging.EventSource.Tests - Fix three failing EventSourceLoggerTest tests - Delete file: IntegrationTesting csproj not needed. - Rename folder test -> tests (Logging/tests/DI.Common/tests) - Cleanup other csproj fix compile issue

Extensions commit mop-up

- Remove unnecessary layers in heap allocation APIs - Delete HeapCreate/Alloc/Free/Destroy from the PAL

Co-authored-by: vargaz <vargaz@users.noreply.github.com>

Fixes mono/mono#17024 Co-authored-by: fanyang-mono <fanyang-mono@users.noreply.github.com>

* Use HostArch for LLVM packages * define HostArch for mono

Small change to correctly check for null in one of the overloads of Expression.Field.

…4312)

* Implement Environment.GetFolderPath on iOS * Address feedback * Move GetFolderPathCore to Environment.Unix.GetFolderPathCore.cs * Fix build issue * Address feedback * cache all special directories * Fix build issue * remove a whitespace * Fix UserProfile issue * undo changes in GetEnvironmentVariableCore * Update Environment.Unix.Mono.cs * Extract to InternalGetEnvironmentVariable * Fix build issue * Return emtpy string if underlying native function returns null * Add nullability

…on DOM exceptions (#34167)

…#34307)

* Guard against deserializing types with open generics * Include serialization tests * Address review feedback * Address review feedback

#34248) * Cache parameterized ctor delegates in class info rather than converter * Address review feedback - move delegate assignment to start of deserialization * Address review feedback - nullability

Adding static to ss_req_acquire and ss_req_count Fixes mono/mono#19389 Co-authored-by: thaystg <thaystg@users.noreply.github.com>

- Fix race condition in JIT_ReversePInvokeEnter - Disable R2R for x86 on all platforms

* Initial proposal for P/Invokes via Source Generators.

* add support for FreeBSD cross-compilation * feedback from review * update detection * add openssl to docs * feedback from reviews * feedback from reviews * final cleanup * feedback from reviews * simplify detection of route.h * correct __CrossBuild detection * use TargetOS for installer

… stack (#33908) * Fix ProfileTailCallNaked so it doesn't write xmm1 past the end of its stack, and do some documentation/refactoring. * code review feedback

This fix is similar to artifact download logic done in build-base.yml

Co-authored-by: Mitchell Hwang <mihw@microsoft.com>

* fixing intel for ref and src * fixing ref and src intell * renaming props file * remove extra property * fixing the intellisense in tests * add extra line * addressing feedback * revert shim path * fix allconfig restore * adding comment

* Disable test based on 34178 * Try with */* * Fix diff

* Fix System.Data.SqlClient version Fixes #34351 * Update Microsoft.Windows.Compatibility.pkgproj

* Fix filename typo * Disable Bitmap round-tripping tests for old libgdiplus versions Old libgdiplus versions have uninitialized stack variable bug that makes Bitmap round-tripping unreliable. The bug causes Bitmap.Flags, Bitmap.HorizontalResolution and Bitmap.VerticalResolution properties to be set to bogus values if the stack happens to contain certain bit patterns. This bug was fixed for libgdiplus 6 by mono/libgdiplus@81e45a1#diff-c96a8261ecb168c12b44248208da21c0R118. * Fix and simplify gdiplus library loading

)

Turns out `llvm::Triple(llvm::sys::getDefaultTargetTriple()).getArchName()` doesn't work as expected on all targets so the only option is to set arch via defines. NOTE: it accepts "x86-64", not "x86_64

…34392)

* Disable test on windows and enable on other platforms * Remove mac from test

…4251)" (#34306)" (#34315) This reverts commit 94c4b52.

Stephen Bonikowsky is no longer on the WCF team, updating ownership.

Starting with mono/mono#16907 , the runtime ends in GC-Safe state (mode) after mono_jit_init_version() is called. mono_gc_init_finalizer_thread() expects the GC to not already be in safe mode.

…serting. (#34347)  Co-authored-by: vargaz <vargaz@users.noreply.github.com>

@joncham

…age (#34346) sets. This fixes: https://issuetracker.unity3d.com/issues/editor-crash-when-entering-and-exiting-playmode-for-a-few-times-with-microsoft-maps-sdk @joncham and I were working on this we hit the two asserts that I've removed. We were wondering if they were still needed?  Co-authored-by: UnityAlex <UnityAlex@users.noreply.github.com>

* free list bucket skipping * couple more tweaks * More size buckets in POH free lists * static_data_table * PR feedback * off-by-one error in BASE_LOH_ALIST_BITS

Add missing constants to handle ppc64/ppc64le and s390x platforms. Related: mono/mono@68a8b92 Fixes: mono/mono#19402 Co-authored-by: sharkcz <sharkcz@users.noreply.github.com>

Co-authored-by: Thomas Levesque <thomaslevesque@users.noreply.github.com>

)

The code in CorUnix::InternalSetThreadDescription is writing behind the end of the allocated memory in case the name is shorter than 16 characters. That is causing memory heap corruption.

* Eliminate dead configuration code * Remove more dead code * Eliminating EEConfig::GetConfiguration_DontUse_ * Eliminate EEConfig::GetConfigValueCallback * Eliminate ConfigSearch

Ubuntu 14.04 is no longer supported.

…nagedDll (#34425) * Reorganize the NativeLibrary tests to be in one root folder and build/use the same native library for testing * Add test for AssemblyLoadContext.LoadUnmanagedDll

…#34428)

@trylek

In crossgen2 we convert all methods to their canon versions in CreateMethodEntrypointNodeHelper, which causes a problem for delegates set to generic static methods. When these methods are called via a delegate, at runtime it gets a hidden generic argument that is a MethodDesc of a canon version of the static method instead of the specific instantiation. I've investigated why we convert all methods to their canon versions in the CreateMethodEntrypointNodeHelper and I've found a PR by @trylek in CoreRT repo that has introduced it. I've discussed that with him, but he wasn't able to figure out why the change was made and it seems like it was mostly accidental or fixing some problem that no longer exists. Removing this conversion fixes the problem and I have verified that it doesn't introduce new ones by running both pri 1 CoreCLR tests and the libraries tests. It has also made the diff between code generated by crossgen1 and crossgen2 e.g. for System.Private.CoreLib.dll smaller.

* Enhance PAL initialization order The CGroups initialization has _ASSERTE that is not asserting because debugging support was being initialized after the CGroups initialization. This change moves the debugging support initialization to the earliest possible point. It also modifies it so that it doesn't depend on the PAL env functions and uses getenv instead. And finally, initialization of the critical section for debug printf is moved to the very end of DBG_init_channels just as a little cleanup.

Command for opening Visual Studio works from root folder, but not inside library folder as was implied by ordering in the instructions.

1. Fixed several inconsistencies w.r.t. signature context in recursive descent into generic signatures based on offline investigation with JanV. 2. Don't crash the dump upon invalid UTF8 characters. Thanks Tomas

This change fixes two bugs in field offset computation where the results that crossgen2 was getting was different from what runtime computes. In both cases, the problem was caused by alignment of a derived class being done differently. The first issue was happening for the case when the base and derived classes are in different assemblies. Runtime detect if two assemblies are in the same version bubble using the native manifest metadata table containing a list of assemblies that was supposed to contain all assemblies that the assembly being compiled was found to reference. However, it contained only assemblies that were not in the original assembly reference list, e.g. ones pulled in by inlining. So runtime wasn't getting the same view on what's in the bubble. The second issue happened for the case when both the base and derived class were from the same assembly, but one of the ancestor classes had a field of a value class type that was from another assembly and could be transitively decomposed to fields of types from the same assembly or types like primitive types, object, pointer or enums. The alignment of a derived class members is determined based on that and runtime decision is to align if there is any type from another assembly in the type hierarchy of a class or in fields of any ancestors. For example, the decision would be different for the following scenario: Assembly A: struct AA { int a; } Assembly B: class B1 { AA aa; } class B2 : B1 { int x; } Here crossgen2 would not align the first member but runtime would. So the layout of B2 produced by crossgen2 would be: ``` Offset Field 0 MethodTable 8 a 12 x ``` Layout produced by the runtime would be ``` Offset Field 0 MethodTable 8 a 16 x ``` The fix for the first issue is to put all referenced assemblies into the native manifest metadata. The fix for the second issue is to stop decomposing members of value classes once we hit a value class that's from another module.

Move to an Ubuntu image which has the `en_US.UTF-8` locale available. This should remove the spurious warning that we're seeing during PR / CI closes #34280

Add support for building S.P.CoreLib in a separate job

We have seen a failure in the CI where the OnHijackWorkerTripThread was called in preemptive mode and so a contract in Thread::SetFrame down its call chain has fired when checking that the thread is in cooperative mode. I have found that the issue is caused by hijacking a NativeCallable method. Such methods switch to preemptive mode before they return and so when they are hijacked, the OnHijackWorkerTripThread is called in preemptive mode. The fix is to prevent return address hijacking for NativeCallable methods.

Co-authored-by: radical <radical@users.noreply.github.com>

)

This is needed by Xamarin.Android to be able to round-trip with names fetched via reflection, as the only public API for this (mono_type_get_name) uses the IL format instead.

This adds support for generating Microsoft.NETCore.App.Runtime packs for ios-x64, ios-arm, and ios-arm64.

* Experimental pgo tool - Experiment and rough design going forward for providing profile data to the compiler * Update build so dotnet-pgo is built

Reflects PR feedback.

This enables warn as error for the libraries jobs now that our builds are warning free.

@ericstj

cc @ericstj

* Disable iOS arm PR builds Infrastructure issue: #34465 * Add tracking issue and disable installer iOS arm build too

* Fix two issues detected by Valgrind When I have used Valgrind to investigate a memory corruption issue recently, I've noticed that it has also reported two cases when a conditional jump was using an uninitialized variable as one of the inputs to the condition. This change fixes these.

…asses (#34212) Follow CoreCLR's example and add the RTSpecialName type attribute to any class field with the 'value__' name when , used as the value store for enumeration class instances. Second part of the "RTSpecialName fix" that started with #33389 - before, our behavior was to add the RTSpecialName immediately, as soon as the EnumBuilder was constructed. Contributes to #2389

* [sgen] Add support for allocating pinned array These objects will end up allocated either in LOS (which is by design not moved around) or into special major blocks where compaction doesn't take place. * [sgen] Be consistent with coreclr on GC api * Enable GC tests for pinned array allocation * Skip large array allocation test on mono

Otherwise fails with: ``` MSBUILD : error MSB1001: Unknown switch. Switch: --subsetCategory ``` on Windows

…ono (#34445)

* Port CoreCLR's TypeNameBuilder to C#, and use it in Mono too Mono's System.Reflection.Emit creates type names that fail to be normalized or shaped in all ways that CoreCLR does. Port CoreCLR's mixed-mode thread-unsafe implementation to thread-safe C#, and start using it in Mono for names in TypeBuilder. Fixes issues with e.g. null-delimited type names being passed to different Reflection.Emit builders. Contributes to #2389.

Fixes #34478

* simplify libraries build dependencies. * FIx depends for tests * Fix artifact name for webAssembly

These changes involve targeted fixes to issues seen due to the new build mode and generic signature encoding fixes identified in my offline investigation with JanV. While I locally see the tests passing with just a handful of failures, I still seem unable to make the tests run in the lab; I keep investigating that but I believe it's useful to merge this delta in to let us parallelize follow-up efforts. Thanks Tomas

It started failing with the same issue as #34465

* Add a CborReader.SkipValue() method * revert reader state format error changes * refactor skip from recursive to imperative * use non-allocating skips for string data items * add a test for extremely nested values * remove blank line * consolidate SkipString() methods

…nd Unix sockets (#33674) Socket.Windows: support ConnectAsync(SocketAsyncEventArgs) for non-stream protocols

* Enable nullable reference types for mono's Corelib Corelib had nullable reference types enabled, but all warnings were suppressed. This fixes the thousands of warnings that were being hidden and fully enables nullable reference types for mono's corelib, as it is for coreclr's corelib. I copied over the public API annotations from coreclr's corelib and tried to annotate the mono source as accurately as possible. However, I'm 100% sure there are issues, both places I misexpressed the contract of an internal API and places where I suppresed warnings with `!` where it likely could actually null ref. Along the way, when there was an obvious null ref where coreclr was doing a null check and providing different behavior, I fixed it or at least added a comment, but I expect there are more issues lurking. * Fix iOS build for GetSpecialFolder * Address PR feedback

Ports GCBasic, ReJIT, and EventPipe tests as well as adds a test that verifies profilers can call MetaDataGetDispenser.

- Add support for the --instruction-set parameter as described in #226 . NOTE: As the abi for Vector parameters is not yet stable, support for the --instruction-set parameter is only enabled if --inputbubble is also enabled. Parallel work to stabilize the abi is in progress, but is not complete. ALSO NOTE: The names of the instruction sets are shared with mono, and don't follow the names in issue #226 - Add concept of baseline instruction set support to R2R file format - Can be applied at a per method level or at the entire R2R file level - R2RDump support for dumping the extra data - Refactor how support for hardware intrinsics beyond SSE2 support are handled in crossgen2 - Add feature to the JIT to detect which hardware features are actually used - Tell the JIT unconditionally that SSE42+Lzcnt+Popcnt+Pclmulqdq are supported - But if support beyond the --instruction-set specified baseline is used, notate the method with a per-method instruction set support fixup. - This enables usage of many intrinsics in corelib with greater efficiency than today - This enables usage of SSE42 and below intrinsics safely in non-CoreLib code. Use of higher level intrinsics in non CoreLib code will generate code which does not use the higher level intrinsic, and note that the method's code should not be used in the presence of hardware which does support greater CPU capabilities. - In the future a logical enhancement of this work would be to generate multiple bodies of code to handle these more complex cases. - In combination with the --instruction-set argument, if Avx2 is enabled, then the logic gracefully adds a dependency on Avx2 capability and Vector<T> becomes useable by crossgen'd code.

…#34349) * follow up * other occurences * fixing compiler services failure

)

… and Log2 (#34486) ARM64 intrinsic for: - LeadingZeroCount - Log2 - TrailingZeroCount

* Move DependencyModel to libraries Moving DependencyModel to the same folder and infrastructure as the rest of our libraries, and out of the installer folder. I also dropped support for anything below netstandard2.0 at this time. Contributes to #3470 Fix #3425 * Exclude DependencyModel in shims and package checks since it has a duplicated type with System.Collections. * Harvest previous TFM assets from the previously shipped package. This also means we start building DependencyModel for net461 to ensure full framework support works correctly, and doesn't pick up the old net451 asset. * Fix unit tests to consistently pass on core and netfx. * Add back HashCodeCombiner as obsolete. * Fix DependencyModel pkg build to reference the correct version of Newtonsoft.Json for the harvested package assets. * Adding IgnoredTypes for Serialization.Primitives in netcoreapp1.0 and netcoreapp1.1 Co-authored-by: Jose Perez Rodriguez <joperezr@microsoft.com>

…ies (#34516)

TODO: viz for multidim arrays Co-authored-by: radical <radical@users.noreply.github.com>

…icial builds (#34469) * Fix testhost publish condition and omptimize assets we publish in official builds * Publish ref/ folder in official builds as it is needed for the installer build * Just validate all assets paths when IncludeOOB* is true

Right now WebAssembly does not have an implementation for TimeZoneInfo and instead throws exceptions. This is an alternate implementation of mono/mono#17617 that loads the zone information into the webassembly virtual file system using a pre-loaded file package. __BCL Changes__ TimeZoneInfo implementation for WebAssembly to read from WASM VFS `/zoneinfo`. - Remove `TimeZoneInfo.WebAssembly.cs` file - Remove source dependencies on `TimeZoneInfo.WebAssembly.cs`. - Modify TimeZoneInfo for specific WASM functionality - Default root directory is `zoneinfo` - Add icall `mono_timezone_get_local_name` for WASM in CreateLocal - Add icall implementation - ves_icall_System_TimeZoneInfo_mono_timezone_get_local_name __WASM Runtime Build__ - Add `-s FORCE_FILESYSTEM=1` to mono wasm runtime build - Required so that it includes support for loading pre preload packages * Message when generating the zoneinfo data. > Remember to build the main file with -s FORCE_FILESYSTEM=1 so that it includes support for loading this file package - Add targets to run time zone tests. - make run-tzd-sample - make run-tzd-interp-sample - make run-tzd-aot-sample - Modify `runtime-tests.js` to load and parse the zoneinfo data manually to load into the VFS. __WASM Driver - aka `driver.c`__ - Add implementations of the following: - mono_timezone_get_local_name These will be called from the managed bcl module `TimeZoneInfo.cs` module. __packager.exe__ - Add `-s FORCE_FILESYSTEM=1` - Required so that it includes support for loading pre preload packages * Message when generating the zoneinfo data. > Remember to build the main file with -s FORCE_FILESYSTEM=1 so that it includes support for __Zone Info data and support files__ - Three files for zoneinfo VFS support. - `mono-webassembly-zoneinfo-fs-smd.js.metadata` - This is the separate metadata output by emscripten file-packager. Used to parse the zoneinfo data in non browser environments. See `runtime-tests.js` - `mono-webassembly-zoneinfo-fs.js` - Output by emscripten file-packager that will be referenced by browser environments to load the zoneinfo data into the VFS at `/zoneinfo` - `zoneinfo.data` - The binary file output by emscripten file-packager. The file contains the actual zoneinfo data loaded into the VFS and parsed by `TimeZoneInfo`. The packages are generated automatically and the repo is temporarily at: https://github.com/kjpou1/mono-webassembly-zoneinfo There are some limitations to this method where the file packages generated by the emscripten [file_packager](https://emscripten.org/docs/porting/files/packaging_files.html#packaging-using-the-file-packager-tool) only runs within a web browser. Special consideration is needed to load the VFS with the data outside of browser. See modification for `runtime-tests.js` where the zoneinfo data is parsed manually and loaded into the VFS. We may need to create our own file packager to handle some of the other cases. Co-authored-by: kjpou1 <kjpou1@users.noreply.github.com>

…34479) A few more dependencies on removed packages. System.Reflection.Emit*

Minor code cleanup to make the logic clearer.

)

Co-authored-by: Thomas Levesque <thomaslevesque@users.noreply.github.com>

@jbevain

#34531) As suggested by @jbevain we should bump the protocol to implement multi threaded single step. Co-authored-by: thaystg <thaystg@users.noreply.github.com>

Scripting guide for Shell, PowerShell and CMD files for styling or other guidance. Fixes #9317

* Fix host linker options to have -Bsymbolic * Remove -Bsymbolic-functions

Co-authored-by: vargaz <vargaz@users.noreply.github.com>

Introduced by change to Heap APIs. Also fixes a long-existing memory leak on Linux. Introduce a small, simple class to keep track of memory allocations associated with the CompileResult that we need to free. In the replay case, SuperPMI allocates these (such as for the JIT calling allocMem). In the case of collection, the VM allocates memory for allocMem (and related), so that memory doesn't need to be tracked.

Share more code between LLVM 6 and LLVM 9. Mangler::getNameWithPrefix takes a Twine, not a std::string, so use a StringRef to avoid an unnecessary allocation. InstructionCombiningPass grew a dependency on ProfileSummaryInfoWrapperPass in dotnet/llvm-project@09e539f, which depends on the enclosing FunctionPassManager having a valid reference to a module. This commit adds a dummy module to satisfy this, instead of recreating a new FunctionPassManager on every compilation. (See also http://lists.llvm.org/pipermail/llvm-dev/2019-March/130690.html.) Co-authored-by: imhameed <imhameed@users.noreply.github.com>

If D (private key) is supplied but not the public key (Q), permit this and allow the platform to re-calculate the public key from the private key. * Windows uses CNG blobs with the Q.X and Q.Y values set to (0,0). * LInux uses the ECC math module to recompute Q from D and G. * macOS is TBD.

…base for 64-bit targets (#34526)

@alnikola

- Use RestoreUseStaticGraphEvaluation which improves no-op restore by 10-15x down to 10-20 seconds. - .builds msbuild files renamed to .proj as RestoreUseStaticGraphEvaluation throws for non .proj files without an env var set. - Introducing subsets for libraries and mono and replacing -buildtests switch which was only working for libraries in favor of the subset switch -subset tests which works consistently. - Fixing the Microsoft.DotNet.CodeAnalysis analyzer which wasn't running and adding missing exclusions. - Separating restore and build phases in different parts in the repo (ie for installer.tasks) as generated props and targets need to be imported which requires a reevaluation in the build phase. - Fix eng/docker/build-docker-sdk.ps1 by using the official build entrypoints (cc @alnikola) - Remove a few depprojs in favor of project restore (faster restore :)) - Fix root code coverage measurement not working correctly - Traversal support instead of dir.traversal.targets or manual build target defines. - Introduce a root Build.proj entrypoint which is responsible for building and restoring the repository. This is necessary to enable the new NuGet fast restore which works best and fastest with a single entrypoint. - Avoid binclashes in libraries and between libraries and installer (netstandard.depproj vs netstandard.csproj) - Upgrading the SDK to 5.0 latest - Code cleanup

* Remove entry point helper scripts The entry point helper scripts were useful during the consolidation of the repositories. This work is mostly done and the build.cmd/sh script should be used, going forward.

Upstream cgroup v2 documentation is available at: https://www.kernel.org/doc/Documentation/cgroup-v2.txt Some notable differences between cgroup v1 and v2, from a coreclr point of view, include: - cgroup v2 has a single hierarchy, so we just look for a single "cgroup2" entry in /proc/self/mountinfo (without looking for a subsystem match). - Since cgroup v2 has a single hierarchy, /proc/self/cgroup generally has a single line "0::/path". There's no need to match subsystems or hierarchy ids here. - "memory.limit_in_bytes" is now "memory.max". It can contain the literal "max" to indicate no limit. - "memory.usage_in_bytes" is now "memory.current" - "cpu.cfs_quota_us" and "cpu.cfs_period_us" have been combined into a single "cpu.max" file with the format "$MAX $PERIOD". The max value can be a literal "max" to indicate a limit is not active. It is possible to have both cgroup v1 and v2 enabled on a host (but not inside a container, AFAIK). In that case, this change will pick one based on /sys/fs/cgroup.

* Fix Linux x86 build Related to #33005 * Fix Linux x86 build Related to #33653, #33005 * Fix Linux x86 build Related to #32250

…#34556) * Fix contructor attributes nullability in S.R.E.CustomAttributeBuilder Fixes #819 * Finalise nullable * Updates ref

* [installer] Include Android into installer * [eng] Provide Android with just Microsoft.NETCore.App.Runtime * Attempt to make same changes from iOS PR Co-authored-by: Mitchell Hwang <mihw@microsoft.com>

…34559)

@optional

string and Oid types on Choice can be null. Reference types that are @optional can be null. Also, remove #nullable enable, because all projects that include the ASN.1 generate files should already have nullable enable project-wide.

…hod. Pass this symbol to the init functions instead of the method index. (#34009) This removes a dependency the generated code has on the method index, hopefully allowing separate compilation etc. in the future.  Co-authored-by: vargaz <vargaz@users.noreply.github.com>

…34430)

* Adds deserialization support for JsonDocument This change adds support to `System.Text.Json` for deserializing `JsonDocument`. Specifically, an internal converter `JsonDocumentConverter` is added to the default converter dictionary. I have created a basic test, but I feel more could be done here, and it may not be in the right file/class - some guidance would be helpful here. Fixes #1573 * Adds JsonDocumentTests * Dispose JsonDocument

* Implement Arm.ArmBase and Arm.Crc32 intrinsics

* moving more tfm dependent properties to targets file * correct allconfig build * shimrefpath -> runtime and indentation

* Block route trip of function pointer as Delegate field. Add tests for scenario. * Update MarshalStructAsParamDLL.cpp

…#34611) StubPrecode and NDirectImportPrecode precoded used undefined instruction encodings on x64 due to copy&paste from x86. Change them to use valid instructions.

Fixes #33680

The issue is that we are automatically assuming that types are blittable until proven otherwise while traversing the list of fields declared on the type. In the case of a type with no fields, we have to also check if the parent type is blittable, before blindly setting the flag to true (Ex: what if the base type has non-blittable fields?)

Upstream cgroup v2 documentation is available at: https://www.kernel.org/doc/Documentation/cgroup-v2.txt Some notable differences between cgroup v1 and v2, from a coreclr point of view, include: - cgroup v2 has a single hierarchy, so we just look for a single "cgroup2" entry in /proc/self/mountinfo (without looking for a subsystem match). - Since cgroup v2 has a single hierarchy, /proc/self/cgroup generally has a single line "0::/path". There's no need to match subsystems or hierarchy ids here. - "memory.limit_in_bytes" is now "memory.max". It can contain the literal "max" to indicate no limit. - "memory.usage_in_bytes" is now "memory.current" - "cpu.cfs_quota_us" and "cpu.cfs_period_us" have been combined into a single "cpu.max" file with the format "$MAX $PERIOD". The max value can be a literal "max" to indicate a limit is not active. It is possible to have both cgroup v1 and v2 enabled on a host (but not inside a container, AFAIK). In that case, this change will pick one based on /sys/fs/cgroup.

…jack Prevent return address hijacking of NativeCallable method

* Remove dead yml file * Fix installer prepare for publish

Synchronize access to disposing the SafeHandle, and clean up some code distributed oddly between the ctor and Start, so as to make more fields readonly and only register for cancellation once everything has started, to avoid race conditions that could otherwise result between cancellation and starting happening concurrently.

* Feed cleanup Move away from blob feeds. * Add dotnet5 feed * Update version of System.CommandLine.Experimental * Revert to released versions in SuperILC and R2RDump * Don't remove public feeds for tests

…n Linux.

…/runtime into warning_compliance_code

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fixed code to reenable some warnings in CoreCLR #33902

Fixed code to reenable some warnings in CoreCLR #33902

Commits on Mar 27, 2020

Commits on Mar 28, 2020

Commits on Mar 29, 2020

Commits on Mar 30, 2020

Commits on Mar 31, 2020

Commits on Apr 1, 2020

Commits on Apr 2, 2020

Commits on Apr 3, 2020

Commits on Apr 4, 2020

Commits on Apr 5, 2020

Commits on Apr 6, 2020

Commits on Apr 7, 2020