Prototype gchandle

.github/workflows/check-no-merge-label.yml

@@ -5,7 +5,7 @@ permissions:
 
 on:
   pull_request_target:
-    types: [labeled, unlabeled]
+    types: [opened, edited, reopened, labeled, unlabeled, synchronize]
     branches:
       - 'main'
       - 'release/**'

.github/workflows/check-service-labels.yml

@@ -5,7 +5,7 @@ permissions:
 
 on:
   pull_request_target:
-    types: [labeled, unlabeled]
+    types: [opened, edited, reopened, labeled, unlabeled, synchronize]
     branches:
       - 'release/**'
 

docs/design/coreclr/jit/porting-ryujit.md

@@ -23,13 +23,18 @@ The following components need to be updated, or target-specific versions created
 * The basics
   * target.h
 * Instruction set architecture:
-  * registerXXX.h
-  * emitXXX.h, emitfmtXXX.h
-  * instrsXXX.h, emitXXX.cpp and targetXXX.cpp
-  * lowerXXX.cpp
-  * lsraXXX.cpp
-  * codegenXXX.cpp and simdcodegenXXX.cpp
-  * unwindXXX.cpp
+  * registerXXX.h - defines all registers used by the architecture and any aliases for them
+  * emitXXX.h - defines signatures for public instruction emission methods (e.g. "emit an instruction which takes a single integer argument") and private architecture-specific helpers
+  * emitXXX.cpp - implementation for emitXXX.h
+  * emitfmtXXX.h - optionally defines validity rules for how instructions should be formatted (e.g. RISC-V has no rules defined)
+  * instrsXXX.h - defines per-architecture instructions in assembly
+  * targetXXX.h - defines architectural constraints used elsewhere, such as "bitmask for all integer registers where callee is saved" or "size in bytes of a floating point register"
+  * targetXXX.cpp - implements ABI classifier for this architecture
+  * lowerXXX.cpp - implements [Lowering](https://github.com/dotnet/runtime/blob/main/docs/design/coreclr/jit/ryujit-overview.md#lowering) for this architecture
+  * lsraXXX.cpp - implements register requirement setting based on [GenTree Nodes](https://github.com/dotnet/runtime/blob/main/docs/design/coreclr/jit/ryujit-overview.md#gentree-nodes)
+  * codegenXXX.cpp - implements main codegen for this architecture (i.e. generating per-architecture instructions based on [GenTree Nodes](https://github.com/dotnet/runtime/blob/main/docs/design/coreclr/jit/ryujit-overview.md#gentree-nodes))
+  * hwintrinsic\*XXX.\* and simdashwintrinsic\*XXX.h - defines and implements hardware intrinsic features, e.g. vector instructions
+  * unwindXXX.cpp - implements public unwinding API and unwind info dumping for debug use
 * Calling Convention and ABI: all over the place
 * 32 vs. 64 bits
   * Also all over the place. Some pointer size-specific data is centralized in target.h, but probably not 100%.

docs/design/datacontracts/StackWalk.md

@@ -16,6 +16,9 @@ IEnumerable<IStackDataFrameHandle> CreateStackWalk(ThreadData threadData);
 byte[] GetRawContext(IStackDataFrameHandle stackDataFrameHandle);
 // Gets the Frame address at the given stack dataframe. Returns TargetPointer.Null if the current dataframe does not have a valid Frame.
 TargetPointer GetFrameAddress(IStackDataFrameHandle stackDataFrameHandle);
+
+// Gets the Frame name associated with the given Frame identifier. If no matching Frame name found returns an empty string.
+string GetFrameName(TargetPointer frameIdentifier);
 ```
 
 ## Version 1
@@ -323,3 +326,8 @@ If the Frame is not valid, returns `TargetPointer.Null`.
 TargetPointer GetFrameAddress(IStackDataFrameHandle stackDataFrameHandle);
 ```
 
+
+`GetFrameName` gets the name associated with a FrameIdentifier (pointer sized value) from the Globals stored in the contract descriptor. If no associated Frame name is found, it returns an empty string.
+```csharp
+string GetFrameName(TargetPointer frameIdentifier);
+```

docs/design/features/hw-intrinsics.md

@@ -18,7 +18,7 @@ The hardware intrinsics operate on and produce both primitive types (`int`, `flo
 
 ### Platform-agnostic vector types
 
-The vector types supported by one or more target ISAs are supported across platforms, though they extent to which operations on them are available and accelerated is dependent on the target ISA. These are:
+The vector types supported by one or more target ISAs are supported across platforms, though the extent to which operations on them are available and accelerated is dependent on the target ISA. These are:
 
 * `Vector64<T>` - A 64-bit vector of type `T`. For example, a `Vector64<int>` would hold two 32-bit integers.
   * Note that `Vector64<T>` intrinsics are currently supported only on Arm64, and these are not supported for `double`. Support could be added for this, but would require additional handling.
@@ -51,7 +51,7 @@ The JIT depends on the VM and configuration settings to determine what target pl
 
 Hardware intrinsics are built on RyuJIT's `NamedIntrinsic` mechanism to identify method calls that should be recognized as intrinsics (see https://github.com/dotnet/runtime/blob/main/src/coreclr/jit/namedintrinsiclist.h). In the incoming IL, intrinsic invocations are just method calls, so the JIT must distinguish intrinsic calls from ordinary call-sites and map them to its IR representation: the `GenTreeHWIntrinsic` node.
 
-The [Intrinsic] attribute was added to eliminate the need to check each call-site. It [Intrinsic] attribute has a different meaning on each attribute target:
+The [Intrinsic] attribute was added to eliminate the need to check each call-site. It has a different meaning on each attribute target:
 
 * Method: call targets marked with [Intrinsic] will be checked by the JIT when importing call-sites. If the method's (namespace, class name, method name) triple matches a record in the Hardware Intrinsics Table, it will be recognized as an intrinsic call.
 
@@ -98,7 +98,7 @@ The register allocator has three main passes.
 
 The `LinearScan::buildNode` method is responsible for identifying all register references in the IR, and constructing the `RefPosition`s that represent those references, for each node. For hardware intrinsics it delegates this function to `LinearScan::buildHWIntrinsic()` and the `LinearScan::getKillSetForHWIntrinsic()` method is responsible for generating kill `RefPositions` for these nodes.
 
-The other thing to be aware of is that the calling convention for large vectors (256-bit vectors on x86, and 128-bit vectors on Arm64) does not preserve the upper half of the callee-save vector registers. As a result, this require some special modeling in the register allocator. See the places where `FEATURE_PARTIAL_SIMD_CALLEE_SAVE` appears in the code. This code, fortunately, requires little differentiation between the two platforms.
+The other thing to be aware of is that the calling convention for large vectors (256-bit vectors on x86, and 128-bit vectors on Arm64) does not preserve the upper half of the callee-save vector registers. As a result, this requires some special modeling in the register allocator. See the places where `FEATURE_PARTIAL_SIMD_CALLEE_SAVE` appears in the code. This code, fortunately, requires little differentiation between the two platforms.
 
 ## Code Generation
 

docs/design/specs/PortablePdb-Metadata.md

@@ -26,6 +26,7 @@ The ECMA-335-II standard is amended by an addition of the following tables to th
     * [SourceLink](#source-link-c-and-vb-compilers)
     * [CompilationMetadataReferences](#compilation-metadata-references-c-and-vb-compilers)
     * [CompilationOptions](#compilation-options-c-and-vb-compilers)
+    * [TypeDefinitionDocument](#type-definition-document-c-and-vb-compilers)
 
 Debugging metadata tables may be embedded into type system metadata (and part of a PE file), or they may be stored separately in a metadata blob contained in a .pdb file. In the latter case additional information is included that connects the debugging metadata to the type system metadata.
 
@@ -598,3 +599,22 @@ The order of the options in the list is insignificant.
 > The purpose of this data is to allow a tool to reconstruct the compilation the module was built from.
 > The source files for the compilation are expected to be recovered from the source server using [SourceLink](#source-link-c-and-vb-compilers) and/or from [sources embedded](#embedded-source-c-and-vb-compilers) in the PDB.
 > The metadata references for the compilation are expected to be recovered from a file indexing service (e.g. symbol server) using information in [Compilation Metadata References](#compilation-metadata-references-c-and-vb-compilers) record.
+
+##### Type Definition Document (C# and VB compilers)
+Parent: TypeDef
+
+Kind: {932E74BC-DBA9-4478-8D46-0F32A7BAB3D3}
+
+Stores document information for type definitions that would otherwise not have document information stored through other means, for example in the [MethodDebugInformation](#methoddebuginformation-table-0x31) table.
+
+The blob has the following structure:
+
+    Blob ::= (document-ordinal)*
+
+| terminal         | encoding                    | description                            |
+|:-----------------|:----------------------------|:---------------------------------------|
+| document-ordinal | compressed unsigned integer | row id of the Document table. |
+
+> The purpose of this data is to allow a tool navigate to the source of a type definition where there would
+> otherwise not be any data available. For example an empty class definition, or an enum, where there are
+> no methods to provide the data in the `MethodDebugInformation` table.

docs/project/list-of-diagnostics.md

@@ -115,6 +115,7 @@ The PR that reveals the implementation of the `<IncludeInternalObsoleteAttribute
 |  __`SYSLIB0058`__ | KeyExchangeAlgorithm, KeyExchangeStrength, CipherAlgorithm, CipherStrength, HashAlgorithm and HashStrength properties of SslStream are obsolete. Use NegotiatedCipherSuite instead. |
 |  __`SYSLIB0059`__ | SystemEvents.EventsThreadShutdown callbacks are not run before the process exits. Use AppDomain.ProcessExit instead. |
 |  __`SYSLIB0060`__ | The constructors on Rfc2898DeriveBytes are obsolete. Use the static Pbkdf2 method instead. |
+|  __`SYSLIB0061`__ | The Queryable MinBy and MaxBy taking an IComparer\<TSource\> are obsolete. Use the new ones that take an IComparer\<TKey\>. |
 
 ## Analyzer Warnings
 
@@ -185,7 +186,7 @@ The diagnostic id values reserved for .NET Libraries analyzer warnings are `SYSL
 |  __`SYSLIB1059`__ | Marshaller type does not support allocating constructor |
 |  __`SYSLIB1060`__ | Specified marshaller type is invalid |
 |  __`SYSLIB1061`__ | Marshaller type has incompatible method signatures |
-|  __`SYSLIB1062`__ | Project must be updated with '<AllowUnsafeBlocks>true</AllowUnsafeBlocks>' |
+|  __`SYSLIB1062`__ | Project must be updated with '\<AllowUnsafeBlocks\>true\</AllowUnsafeBlocks\>' |
 |  __`SYSLIB1063`__ | _`SYSLIB1063`-`SYSLIB1069` reserved for Microsoft.Interop.LibraryImportGenerator._ |
 |  __`SYSLIB1064`__ | _`SYSLIB1063`-`SYSLIB1069` reserved for Microsoft.Interop.LibraryImportGenerator._ |
 |  __`SYSLIB1065`__ | _`SYSLIB1063`-`SYSLIB1069` reserved for Microsoft.Interop.LibraryImportGenerator._ |

eng/Signing.props

@@ -43,8 +43,12 @@
     <FileSignInfo Include="mscordaccore.dll" CertificateName="None" />
     <FileSignInfo Include="mscordbi.dll" CertificateName="None" />
 
-    <!-- On MacOS, we need to sign a number of our executables with the Mac developer cert with hardening enabled. -->
-    <FileSignInfo Condition="'$(TargetsOSX)' == 'true'" Include="dotnet;apphost;corerun;createdump" CertificateName="MacDeveloperHarden" />
+    <!-- On MacOS, we need to sign a number of our executables with the Mac developer cert with hardening enabled.
+         Avoid doing this on Linux, which has the same executable names -->
+    <FileSignInfo Condition="'$(TargetsOSX)' == 'true'" Include="dotnet;apphost;corerun;createdump;singlefilehost;crossgen2;ilasm;ilc;ildasm;llc;mono-aot-cross;opt;Mono" CertificateName="MacDeveloperHarden" />
+    <!-- Additionally, we need to notarize any .pkg files -->
+    <MacOSPkg Include="$(ArtifactsPackagesDir)**/dotnet-runtime*.pkg" Exclude="$(ArtifactsPackagesDir)**/dotnet-runtime-internal*.pkg" />
+    <FileSignInfo Include="@(MacOSPkg->'%(Filename)%(Extension)')" CertificateName="MacDeveloperWithNotarization" />
 
     <!-- We don't need to code sign .js files because they are not used in Windows Script Host. -->
     <!-- WARNING: Needs to happed outside of any target -->

eng/SourceBuildPrebuiltBaseline.xml

@@ -36,5 +36,18 @@
     <UsagePattern IdentityGlob="System.IO.Pipelines/7.0.0" />
     <UsagePattern IdentityGlob="System.IO.Pipelines/8.0.*" />
     <UsagePattern IdentityGlob="System.Security.Cryptography.Pkcs/5.0.0" />
+
+    <!-- Dependencies of Roslyn, replaced with live packages during full source build
+         https://github.com/dotnet/runtime/pull/112937#discussion_r2022951143 -->
+    <UsagePattern IdentityGlob="Microsoft.Bcl.AsyncInterfaces/9.0.0" />
+    <UsagePattern IdentityGlob="System.Collections.Immutable/9.0.0" />
+    <UsagePattern IdentityGlob="System.Composition/9.0.0" />
+    <UsagePattern IdentityGlob="System.Composition.AttributedModel/9.0.0" />
+    <UsagePattern IdentityGlob="System.Composition.Convention/9.0.0" />
+    <UsagePattern IdentityGlob="System.Composition.Hosting/9.0.0" />
+    <UsagePattern IdentityGlob="System.Composition.Runtime/9.0.0" />
+    <UsagePattern IdentityGlob="System.Composition.TypedParts/9.0.0" />
+    <UsagePattern IdentityGlob="System.IO.Pipelines/9.0.0" />
+    <UsagePattern IdentityGlob="System.Reflection.Metadata/9.0.0" />
   </IgnorePatterns>
 </UsageData>