Add support for encoding Continuation types with specific layouts in ReadyToRun #6
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This fixes the System.Runtime.Tests.ControlledExecutionTests.CancelItselfFromFinally libraries test that was failing with the interpreter. The issue was that finally block was being aborted due to the fact that instead of the complex COMPlusCheckForAbort, the interpreter was only checking if abort was requested.
…t#123261) Ref: https://github.com/dotnet/runtime/pull/122414/files#r2698997821. I also took the liberty of removing no-op `add`s from `LCL_ADDR` codegen.
Fixes dotnet#122472 From some inspection, it looks like `PerfMap::Initialize` does not depend on the methods right above it. ``` #ifdef FEATURE_PERFTRACING DiagnosticServerAdapter::Initialize(); DiagnosticServerAdapter::PauseForDiagnosticsMonitor(); #endif // FEATURE_PERFTRACING #ifdef FEATURE_GDBJIT // Initialize gdbjit NotifyGdb::Initialize(); #endif // FEATURE_GDBJIT #ifdef FEATURE_EVENT_TRACE // Initialize event tracing early so we can trace CLR startup time events. InitializeEventTracing(); // Fire the EE startup ETW event ETWFireEvent(EEStartupStart_V1); #endif // FEATURE_EVENT_TRACE InitGSCookie(); #ifdef LOGGING InitializeLogging() #endif ``` `PerfMap::Initialize` depends on `SString::Startup()` which occurs earlier in startup and some PAL/OS services. Instead of adding more logic to handle a racing DiagnosticServer IPC [`EnablePerfMap` command](https://github.com/dotnet/diagnostics/blob/main/documentation/design-docs/ipc-protocol.md#enableperfmap) and `PerfMap::Initialize` (e.g. lazy-init/queued commands), bump the PerfMap initialization ahead of the DiagnosticServer initialization. The DiagnosticServer also doesn't depend on PerfMap being initialized afterwards, and PerfMap's `CrstStatic` and `PerfMap::Enable(type, sendExisting)` suggests that the IPC enable perfmap command should occur after `PerfMap::Initialize`. This way, DiagnosticServer is still early in startup, and its `PerfMap::Enable` will not crash the runtime. As for the IPC command to set environment variables, it is unlikely that it was used to `DOTNET_PerfMapEnabled` because that would either not have applied early enough or have crashed the runtime like in the issue. Instead, the `EnablePerfMap`/`DisablePerfMap` commands should be used to toggle PerfMap status.
…forms (dotnet#123274) ## Description NativeAOT runtime packs for iOS and other Apple mobile platforms were missing required static libraries (libSystem.Globalization.Native.a, libSystem.IO.Compression.Native.a, libSystem.Native.a, libbrotli*.a) starting in version 11.0.0-alpha.1.26065.101. ## Changes Modified `eng/liveBuilds.targets` in the `ResolveLibrariesRuntimeFilesFromLocalBuild` target: - Added `BuildNativeAOTRuntimePack != 'true'` condition to the `ExcludeNativeLibrariesRuntimeFiles` item for Apple mobile platforms - Updated comment to clarify that the exclusion does not apply to NativeAOT **Rationale**: Regular CoreCLR packs retrieve these libraries from the CoreCLR shared framework directory, so they're excluded from LibrariesRuntimeFiles to avoid duplicates. NativeAOT packs do not reach back into CoreCLR shared framework artifacts and require these libraries from LibrariesRuntimeFiles. ```xml <!-- Before --> <ExcludeNativeLibrariesRuntimeFiles Condition="'$(TargetsAppleMobile)' == 'true' and '$(RuntimeFlavor)' == 'CoreCLR'" <!-- After --> <ExcludeNativeLibrariesRuntimeFiles Condition="'$(TargetsAppleMobile)' == 'true' and '$(RuntimeFlavor)' == 'CoreCLR' and '$(BuildNativeAOTRuntimePack)' != 'true'" ``` ## Testing The fix will be validated when NativeAOT runtime packs for Apple mobile platforms are built in CI/CD, verifying that the static libraries are included in the package output. <!-- START COPILOT ORIGINAL PROMPT --> <details> <summary>Original prompt</summary> > > ---- > > *This section details on the original issue you should resolve* > > <issue_title>Latest .NET 11 NativeAOT packs missing some static libraries</issue_title> > <issue_description>The `11.0.0-alpha.1.26065.101` version of the `microsoft.netcore.app.runtime.nativeaot.ios-arm64` pack (and probably other packs for other RIDs as well, I haven't checked) are missing the following files: > > * libSystem.Globalization.Native.a > * libSystem.IO.Compression.Native.a > * libSystem.Native.a > * libbrotlicommon.a > * libbrotlidec.a > * libbrotlienc.a > > They were present in `11.0.0-alpha.1.26062.101` > > Complete file list: https://gist.github.com/rolfbjarne/7f42f3cf3ed3a815e2e9fe2b28c580fb#file-gistfile1-txt-L76-L82 > > This started happening in this maestro PR: dotnet/macios#24508 > > specifically this commit: dotnet/macios@ee15359 > > I tried looking through the relevant dotnet/runtime commits, but I didn't see anything obvious causing this.</issue_description> > > <agent_instructions>In the ResolveLibrariesRefAssembliesFromLocalBuild target in eng/livebuilds.targets, add BuildNativeAOTRuntimePack != 'true' to the condition in the ExcludeNativeLibrariesRuntimeFiles. It should only apply to the item where TargetsAppleMobile == 'true'. > > Additionally, update the comment above the item to indicate that it does not apply to nativeaot because we do not reach back into coreclr sharedFramework artifacts.</agent_instructions> > > ## Comments on the Issue (you are @copilot in this section) > > <comments> > <comment_new><author>@MichalStrehovsky</author><body> > > and probably other packs for other RIDs as well, I haven't checked > > I checked Microsoft.NETCore.App.Runtime.NativeAOT.linux-arm64.11.0.0-preview.1.26065.113 and that one has these. Microsoft.NETCore.App.Runtime.NativeAOT.ios-arm64.11.0.0-preview.1.26065.113 doesn't have these. > > So looks to be iDevice specific. I can't build these things locally as I don't have a mac.</body></comment_new> > <comment_new><author>@steveisok</author><body> > In the local build, it looks like coreclr and nativeaot are supposed to look back to `artifacts/bin/coreclr/ios.<arch>.<config>/sharedFramework` for the native lib artifacts. NativeAOT does not appear to be doing that, hence the reason the static libs aren't included. </body></comment_new> > <comment_new><author>@steveisok</author><body> > `ResolveLibrariesRuntimeFilesFromLocalBuild` typically copies all the libraries and native lib bits. There's a step to exclude some native bits if the runtime is coreclr or nativeaot because they are supposed to be accounted for in `ResolveRuntimeFilesFromLocalBuild`. That's not happening for nativeaot due to `BuildNativeAOTRuntimePack=true`. </body></comment_new> > <comment_new><author>@steveisok</author><body> > @kotlarmilos if you want, I can fix it. It's no problem, so let me know.</body></comment_new> > </comments> > </details> <!-- START COPILOT CODING AGENT SUFFIX --> - Fixes dotnet#123253 <!-- START COPILOT CODING AGENT TIPS --> --- 💬 We'd love your input! Share your thoughts on Copilot coding agent in our [2 minute survey](https://gh.io/copilot-coding-agent-survey). --------- Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: steveisok <[email protected]>
…t slowpatheltleave.sh catched on 2K3000. (dotnet#123250) like [PR#122714](dotnet#122714), use `pData->buffer` to return `struct{single, single}` in `ProfileArgIterator::GetReturnBufferAddr()`.
… platforms (dotnet#122789) We had a couple of places that were using `__builtin_available` to guard against some platform version features. Now that our Apple platform floor is macOS 12 and iOS/tvOS 13, we can remove some of them. Co-authored-by: Jeremy Barton <[email protected]>
Adding 3 constants that represent C#, F#, and Visual Basic to the StringSyntaxAttribute Fix dotnet#122604
…23270) I've decided to do a bespoke function for WASM, mainly due to the awkwardness of porting the register mask code that would be of questionable value. Instead, we place an allocation constraint on the RA. With these changes, we are very close to getting a functional "add(x, y)". The only remaining part is the SP arg, being worked on in dotnet#123262. This is not a fully sound implementation since we have a limit on the number of instructions in the prolog (and here we're generating a potentially unbounded number of instruction). I've put a point about this into dotnet#121865. It seems this single-IG restriction ought to be fixable...
…122758) The `nonExpansive` parameter in `Dictionary::PopulateEntry` was always passed as `FALSE` from its only call site in `jithelpers.cpp`. This change removes the dead parameter and simplifies the implementation by eliminating unreachable code paths. --------- Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: jkotas <[email protected]>
This PR adds a fast path in `IEnumerableSkipTakeIterator.TryGetLast`. When source is an Iterator and `GetCount(onlyIfCheap: true)` returns a valid count (not -1), check if count <= _minIndexInclusive. In this case, immediately return with found=false. --------- Co-authored-by: Stephen Toub <[email protected]>
…t#123333) Sometimes code is placed at a higher native offset than the code at IL offsets adjacent to it. Our debugger algorithm deals with this by, during traversal of mappings, [setting the IL offset back](https://github.com/dotnet/runtime/blob/6afecd4adc64210e5b46767d02a6495af1a08c46/src/coreclr/vm/debugdebugger.cpp#L1321) if one is encountered that is lower than the current IL offset in the traversal. Native AOT does not have this mechanism, which leads to such code being misattributed to higher-than-actual source lines. This change reproduces this mechanism in NativeAOT by inserting a native sequence point whenever the IL offset drops below the previous offset. Repro issue (line attribution of DoWork): ```csharp using System.Runtime.CompilerServices; static class Program { #line 53 [MethodImpl(MethodImplOptions.NoOptimization | MethodImplOptions.NoInlining)] public async static Task Main() { await DoWork(0); } [MethodImpl(MethodImplOptions.NoOptimization | MethodImplOptions.NoInlining)] public async static Task BackgroundWorker1() { // await Task.Delay(10000); // await Task.Yield(); throw new InvalidOperationException("BackgroundWorker1 exception."); } [MethodImpl(MethodImplOptions.NoOptimization | MethodImplOptions.NoInlining)] public async static Task BackgroundWorker2() { await Task.Delay(10); Console.WriteLine("BackgroundWorker2 completed."); } [MethodImpl(MethodImplOptions.NoOptimization | MethodImplOptions.NoInlining)] public async static Task BackgroundWorker3() { await Task.Delay(10); Console.WriteLine("BackgroundWorker3 completed."); } [MethodImpl(MethodImplOptions.NoOptimization | MethodImplOptions.NoInlining)] static async Task DoWork(int i) { Task worker1 = BackgroundWorker1(); Task worker2 = BackgroundWorker2(); Task worker3 = BackgroundWorker3(); await Task.WhenAll(worker1, worker2, worker3); Console.WriteLine("All background workers completed."); } } ``` Config: Windows x64 Debug, .NET 11.0.100-alpha.1.25618.104, runtime-async on <table> <tr> <th>NativeOffset</th> <td>0</td> <td>106</td> <td>107</td> <td>107</td> <td>116</td> <td>124</td> <td>124</td> <td>133</td> <td>141</td> <td>141</td> <td>150</td> <td>158</td> <td>169</td> <td>183</td> <td>192</td> <td>205</td> <td>222</td> <td>231</td> <td>244</td> <td>261</td> <td>270</td> <td>283</td> <td>307</td> <td>312</td> <td>357</td> <td>379</td> <td>380</td> <td>387</td> <td>392</td> <td>393</td> <td>420</td> <td>431</td> <td>867</td> <td>878</td> <td>1067</td> <td>1096</td> </tr> <tr> <th>ILOffset</th> <td>0</td> <td>0</td> <td>1</td> <td>1</td> <td>6</td> <td>7</td> <td>7</td> <td>12</td> <td>13</td> <td>13</td> <td>18</td> <td>19</td> <td>27</td> <td>30</td> <td>35</td> <td>37</td> <td>40</td> <td>45</td> <td>47</td> <td>50</td> <td>55</td> <td>57</td> <td>60</td> <td>65</td> <td>70</td> <td>75</td> <td>76</td> <td>81</td> <td>86</td> <td>87</td> <td>87</td> <td>70</td> <td>87</td> <td>70</td> <td>0</td> <td>87</td> </tr> </table> <table> <tr> <th>ILOffset</th> <td>0</td> <td>1</td> <td>7</td> <td>13</td> <td>19</td> <td>76</td> </tr> <tr> <th>LineNumber</th> <td>80</td> <td>81</td> <td>82</td> <td>83</td> <td>84</td> <td>85</td> </tr> </table> Before: <table> <tr> <th>NativeOffset</th> <td>0</td> <td>106</td> <td>107</td> <td>124</td> <td>141</td> <td>158</td> <td>380</td> <td>1067</td> </tr> <tr> <th>LineNumber</th> <td>80</td> <td>80</td> <td>81</td> <td>82</td> <td>83</td> <td>84</td> <td>85</td> <td>80</td> </tr> </table> Notice in particular that code after native offset 431, the suspension code for DoWork, is mistakenly attributed to line 85 (Console.WriteLine). After: <table> <tr> <th>NativeOffset</th> <td>0</td> <td>106</td> <td>107</td> <td>124</td> <td>141</td> <td>158</td> <td>380</td> <td>431</td> <td>878</td> <td>1067</td> </tr> <tr> <th>LineNumber</th> <td>80</td> <td>80</td> <td>81</td> <td>82</td> <td>83</td> <td>84</td> <td>85</td> <td>84</td> <td>84</td> <td>80</td> </tr> </table> Here we see that 380-431 is attributed to Console.WriteLine, while the suspension code that follows is correctly attributed to the previous line of await. [aot.txt](https://github.com/user-attachments/files/24697543/aot.txt)
Read binary psinfo for System.Diagnostic.Process on SunOS (Solaris or illumos). No failures in System.Diagnostic.Process.Tests (but lots of skip) --------- Co-authored-by: Austin Wise <[email protected]> Co-authored-by: Adeel Mujahid <[email protected]> Co-authored-by: Jan Kotas <[email protected]>
…ensions.Configuration.Functional.Tests (dotnet#123335)
…cand methods (dotnet#123288) ## Description `BFloat16.TryWriteSignificandBigEndian` and `TryWriteSignificandLittleEndian` initially reported `bytesWritten = 4` (sizeof(uint)) instead of the correct value. Upon further investigation, it was discovered that BFloat16's significand should be byte-sized (8 bits: 7 trailing bits + 1 implicit bit), not ushort-sized (16 bits). ## Changes - **BFloat16.cs**: - Changed `Significand` property from `ushort` to `byte` to correctly represent the 8-bit significand - Updated `GetSignificandByteCount()` to return `sizeof(byte)` (1) instead of `sizeof(ushort)` (2) - Updated `TryWriteSignificandBigEndian` to write 1 byte directly (endianness is irrelevant for single byte) - Updated `TryWriteSignificandLittleEndian` to write 1 byte directly (endianness is irrelevant for single byte) - **BFloat16Tests.cs**: Added test coverage for `TryWriteSignificandBigEndian` and `TryWriteSignificandLittleEndian` using `[Theory]` with `[MemberData]` to provide BFloat16 test values directly. Tests verify correct byte count (1 byte) and output values for various BFloat16 values (NegativeInfinity, MinValue, NaN, Epsilon, Zero, MaxValue, PositiveInfinity). Empty destination edge cases are tested in separate `[Fact]` methods. Test data methods use modern C# collection literal syntax `[ ... ]` for cleaner, more maintainable code. **Technical Details**: BFloat16 format has 1 sign bit + 8 exponent bits + 7 trailing significand bits. The `Significand` property includes the implicit leading 1 bit for normal numbers, resulting in 8 bits total (1 byte). For a single byte value, endianness doesn't apply, so both BigEndian and LittleEndian methods write the byte directly to `destination[0]`. **Note**: Tests use `[MemberData]` instead of `[InlineData]` to pass BFloat16 objects directly, avoiding float-to-BFloat16 conversion issues that could produce different values than expected. <!-- START COPILOT ORIGINAL PROMPT --> <details> <summary>Original prompt</summary> > > ---- > > *This section details on the original issue you should resolve* > > <issue_title>BFloat16.TryWriteSignificand{Big,Little}Endian report 4 for bytesWritten</issue_title> > <issue_description>The `BFloat16` implementation of `TryWriteSignificandBigEndian` and `TryWriteSignificandLittleEndian` write 16-bit integers, but report `4` as the number of bytes written (32-bit), even though only two bytes were written: > > https://github.com/dotnet/runtime/blob/0bb28890029bb15c6d08a3ee11a62fc828097f41/src/libraries/System.Private.CoreLib/src/System/Numerics/BFloat16.cs#L1047-L1051 > > I'm not exactly a numerics person but should this be a `ushort`? (That is what `Half` does).</issue_description> > > <agent_instructions>Please fix bytesWritten for BFloat16, add tests</agent_instructions> > > ## Comments on the Issue (you are @copilot in this section) > > <comments> > </comments> > </details> <!-- START COPILOT CODING AGENT SUFFIX --> - Fixes dotnet#123286 <!-- START COPILOT CODING AGENT TIPS --> --- ✨ Let Copilot coding agent [set things up for you](https://github.com/dotnet/runtime/issues/new?title=✨+Set+up+Copilot+instructions&body=Configure%20instructions%20for%20this%20repository%20as%20documented%20in%20%5BBest%20practices%20for%20Copilot%20coding%20agent%20in%20your%20repository%5D%28https://gh.io/copilot-coding-agent-tips%29%2E%0A%0A%3COnboard%20this%20repo%3E&assignees=copilot) — coding agent works faster and does higher quality work when set up for your repo. --------- Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: EgorBo <[email protected]> Co-authored-by: stephentoub <[email protected]> Co-authored-by: Egor Bogatov <[email protected]>
Remove extra redudant load. ins->dreg is always fpr and sreg is always r13 so they can't be aliased as well, so it's a redundant load.
`Rune` had a typo, `SYSTEM_PRIVATE_CORLIB` instead of `SYSTEM_PRIVATE_CORELIB`. This led to the very minor issue that the ArgumentException would not use the expected exception message. We don't typically unit the text of exception of exception messages, but if we want to do something we can.
…#123353) Terser 5.39 introduced an optimization that removes console.assert(true, ...) as "unnecessary" https://github.com/terser/terser/blob/master/CHANGELOG.md#v5390. In runtime we are counting on the assert to keep `base64String`: https://github.com/dotnet/runtime/blob/1af7e1a9b7d659fdb9e2bc7aa52197fd5d272ec6/src/mono/browser/runtime/debug.ts#L37 The assert removal caused the `base64String` to be optimized away, breaking the `BrowserDebugProxy's `, breaking the ability to read them from the paused scope. <img width="2725" height="204" alt="image" src="https://github.com/user-attachments/assets/a55233c9-fe3e-4c5e-99fb-36a5f02b7dba" /> This affects net11 debugging, net10 works correctly. This PR replaces `true` with `!!Date.now()` which Terser cannot statically evaluate.
…ips more elements than exist (dotnet#123295) Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: stephentoub <[email protected]> Co-authored-by: eiriktsarpalis <[email protected]>
Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: stephentoub <[email protected]> Co-authored-by: tannergooding <[email protected]> Co-authored-by: Tanner Gooding <[email protected]>
Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: stephentoub <[email protected]> Co-authored-by: MihaZupan <[email protected]> Co-authored-by: Miha Zupan <[email protected]>
## Description Adds `*.exe` to `.gitignore` to prevent accidental commits of executable files like `nuget.exe`. **Changes:** - Added `*.exe` pattern to ignore all executable files - Added exceptions for existing test resource `.exe` files that must remain tracked: - `src/libraries/System.Diagnostics.FileVersionInfo/tests/.../NativeConsoleApp.exe` - `src/libraries/System.Reflection.Metadata/tests/Resources/NetModule/AppCS.exe` - `src/libraries/System.Reflection.Metadata/tests/Resources/Misc/Signed.exe` <!-- START COPILOT CODING AGENT SUFFIX --> <!-- START COPILOT ORIGINAL PROMPT --> <details> <summary>Original prompt</summary> > copilot has an unfortunate habit of trying to merge nuget.exe into the repo. Please ensure that the .gitignore file ignores all .exe files. </details> <!-- START COPILOT CODING AGENT TIPS --> --- 💬 We'd love your input! Share your thoughts on Copilot coding agent in our [2 minute survey](https://gh.io/copilot-coding-agent-survey). --------- Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: stephentoub <[email protected]>
…tnet#123345) `FromStack(1)` and `FromStackDMI(1)` rely on the test entry points to not be inlined into the test executor, so ensure these are NoInlining.
…123213) Marshalled calli is not supported with the interpreter
…22430) This is dotnet#122134 but with the NuGet source moved from the test directory to the repo root's NuGet.config. I briefly explored using `dotnet-trace` from the dotnet-tools nuget source, but extra hurdles were popping up with that approach, from installing the tool which also brought `.store/dotnet-trace/9.0.660901/dotnet-trace/9.0.660901/tools/net8.0/any/` and invoking it would need to resolve sdk location. Even if all were resolved, the added benefit is fairly minimal, since dotnet-trace is just a wrapper around record-trace. ---------------------------------------------------------------- # Repeating dotnet#122134's Description With user_events support added in dotnet#115265, this PR looks to test a few end-to-end user_events scenario. ## Alternative testing approaches considered ### **Existing EventPipe runtime tests** Existing EventPipe tests under `src/tests/tracing/eventpipe` are incompatible with testing the user_events scenario due to: 1. Starting EventPipeSessions through DiagnosticClient ❌ DiagnosticClient does not have the support to send the IPC command to start a user_events based EventPipe session, because it requires the user_events_data file descriptor to be sent using SCM_RIGHTS (see https://github.com/dotnet/diagnostics/blob/main/documentation/design-docs/ipc-protocol.md#passing_file_descriptor). 2. Using an EventPipeEventSource to validate events streamed through EventPipe ❌ User_events based EventPipe sessions do not stream events. Instead, events are written to configured TraceFS tracepoints, and currently only RecordTrace from https://github.com/microsoft/one-collect/ is capable of generating `.nettrace` traces from tracepoint user_events. ### **Native EventPipe Unit Tests** There are Mono Native EventPipe tests under `src/mono/mono/eventpipe/test` that are not hooked up to CI. These unit tests are built through linking the shared EventPipe interface library against [Mono's EventPipe runtime shims](https://github.com/dotnet/runtime/tree/main/src/mono/mono/eventpipe) and using [Mono's test runner](https://github.com/dotnet/runtime/tree/main/src/mono/mono/eglib/test). To update these unit tests into the [standard runtime tests structure](https://github.com/dotnet/runtime/tree/main/src/tests), a **larger investment** is needed to either migrate EventPipe from using runtime shims to a OS Pal source shared by coreclr/nativeaot/mono (see dotnet#118874 (comment)) or build an EventPipe shared library specifically for the runtime test using a runtime-agnostic shim. As existing mono unit tests don't currently test IPC commands, coupled with no existing runtime infrastructure to read events from tracepoints, there would be even more work on top of updating mono native eventpipe unit tests to even test the user_events scenario. ## End-to-End Testing Added A low-cost approach to testing .NET Runtime's user_events functionality leverages RecordTrace from https://github.com/microsoft/one-collect/, which is already capable of starting user_events based EventPipe sessions and generating `.nettrace`s. (Note: [dotnet-trace wraps around RecordTrace](dotnet/diagnostics#5570)) Despite adding an external dependency which allows RecordTrace failures to fail the end-to-end test, user_events was initially added with the intent to depend on RecordTrace for the end-to-end scenario, and there are no other ways to functionally test a user_events based eventpipe session. ### Approach Each scenario uses the same pattern: 1. **Scenario invokes the shared test runner** User events scenarios can differ in their tracee logic, the events expected in the .nettrace, the record-trace script used to collect those events, and how long it takes for the tracee to emit them and for record-trace to resolve symbols and write the .nettrace. To handle this variance, UserEventsTestRunner lets each scenario pass in its scenario-specific record-trace script path, the path to its test assembly (used to spawn the tracee process), a validator that checks for the expected events from the tracee, and optional timeouts for both the tracee and record-trace to exit gracefully. 2. **`UserEventsTestRunner` orchestrates tracing and validation** Using this configuration, UserEventsTestRunner first checks whether user events are supported. It then starts record-trace with the scenario’s script and launches the tracee process so it can emit events. After the run completes, the runner stops both the tracee and record-trace, opens the resulting .nettrace with EventPipeEventSource, and applies the scenario’s validator to confirm that the expected events were recorded. Finally, it returns an exit code indicating whether the scenario passed or failed. ### Dependencies: - Environment with a kernel 6.4+, .NET 10, glibc 2.35+ - Microsoft.OneCollect.RecordTrace (transitively resolved through a dotnet diagnostics public feed) - Microsoft.Diagnostics.Tracing.TraceEvent 3.1.24+ (to read [NetTrace V6](https://github.com/microsoft/perfview/blob/main/src/TraceEvent/EventPipe/NetTraceFormat.md)) ## Helix Nuances UserEvents functional runtime tests differ from other runtime tests because it depends on OneCollect's Record-Trace tool to enable a userevents-based eventpipe session and to collect events. By design, Record-Trace requires elevated privileges, so these tests invoke a record-trace executable with sudo. When tests run on Helix, test artifacts are stripped of their permissions, so the test infrastructure was modified to give record-trace execute permissions (helix-extra-executables.list). Moreover, to avoid having one copy of record-trace per scenario, which in turn requires re-adding execute permissions for each, more modifications were added to copy over a single record-trace executable that would be used by all scenarios (OutOfProcess marker). Additionally, in Helix environments, TMPDIR is set to a helix specific temporary directory like /datadisks/disk1/work/<id>/t, and at this time, record-trace only scans /tmp/ for the runtime's diagnostic ports. So as a workaround, the tracee apps are spawned with TMPDIR set to /tmp. Lastly, the job steps to run tests on AzDO prevents restoring individual runtime test projects. Because record-trace is currently only resolvable through the dotnet-diagnostics-tests source, userevents_common.csproj was added to the group of projects restored at the beginning of copying native test components to restore Microsoft.OneCollect.RecordTrace.
Generate src/coreclr/vm/wasm/callhelpers-pinvoke.cpp and use it instead of hardcoded native libs entrypoints.c --------- Co-authored-by: Copilot <[email protected]> Co-authored-by: Pavel Savara <[email protected]> Co-authored-by: Aaron R Robinson <[email protected]>
Fixes dotnet#123238 --------- Co-authored-by: Radek Zikmund <[email protected]>
Fixes dotnet#123361 The CI probably won't be happy to run the test repro
| _ASSERTE(objRefsSize == (dataSize + (TARGET_POINTER_SIZE - 1)) / TARGET_POINTER_SIZE); | ||
| LoaderAllocator* allocator = asyncMethod->GetLoaderAllocator(); | ||
| AsyncContinuationsManager* asyncConts = allocator->GetAsyncContinuationsManager(); | ||
| MethodTable* result = asyncConts->LookupOrCreateContinuationMethodTable((unsigned)dataSize, objRefs, asyncMethod); |
There was a problem hiding this comment.
We should update LookupOrCreateContinuationMethodTable to work off of a Module owner, not from a MethodDesc as an owner. From looking at LookupOrCreateContinuationMethodTable, it only needs the MethodDesc so that it can get a LoaderModule to stuff the ContinuationMethodTable into. The existing jit scenario can continue to use the Module of the MethodDesc being compiled, but the R2R scenario can just use the System module, since we don't support R2R of collectible assemblies. In addition, we won't need #if READYTORUN logic in AsyncContinuationType.cs, and I don't see why we'll need the AllocContinuationMethod/AllocContinuationClass R2R fixups.
If you make this change, then, the R2R format will not need to record a MethodDesc with each ContinuationType, which will allow us to get better file size for R2R images.
…ocumentation (dotnet#123518) Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: MihaZupan <[email protected]> Co-authored-by: Miha Zupan <[email protected]>
## Description Updates Windows build pool images from VS2022 to VS2026 preview across CI pipelines, including main build infrastructure and stress test pipelines. ## Changes - **`eng/pipelines/common/xplat-setup.yml`**: - Official builds: `windows.vs2022.amd64` → `windows.vs2026preview.scout.amd64` - Public builds: `windows.vs2022.amd64.open` → `windows.vs2026preview.scout.amd64.open` - **`eng/pipelines/libraries/fuzzing/deploy-to-onefuzz.yml`**: - Fuzzing jobs: `windows.vs2022.amd64` → `windows.vs2026preview.scout.amd64` - **`eng/pipelines/libraries/stress/ssl.yml`**: - SSL stress tests: `1es-windows-2022-open` → `windows.vs2026preview.scout.amd64.open` - **`eng/pipelines/libraries/stress/http.yml`**: - HTTP stress tests: `1es-windows-2022-open` → `windows.vs2026preview.scout.amd64.open` All VS2022 and 1es-windows-2022 image references in pipeline configurations have been migrated to the VS2026 preview scout image. <!-- START COPILOT ORIGINAL PROMPT --> <details> <summary>Original prompt</summary> > > ---- > > *This section details on the original issue you should resolve* > > <issue_title>Switch from VS2022 to VS2026 image</issue_title> > <issue_description>At least in main but probably also in release/10.0.</issue_description> > > <agent_instructions>Let's start by changing the public and official build windows pools to windows.vs2026preview.scout.amd64. That should be found in eng/pipelines/common/xplat-setup.yml. We'll figure out whatever else is needed as we go.</agent_instructions> > > ## Comments on the Issue (you are @copilot in this section) > > <comments> > </comments> > </details> <!-- START COPILOT CODING AGENT SUFFIX --> - Fixes dotnet#123550 <!-- START COPILOT CODING AGENT TIPS --> --- 💬 We'd love your input! Share your thoughts on Copilot coding agent in our [2 minute survey](https://gh.io/copilot-coding-agent-survey). --------- Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: steveisok <[email protected]> Co-authored-by: hoyosjs <[email protected]>
This PR addresses some review feedback in dotnet#122301 and fills in more codegen for casts on Wasm. Notably: - Casts to and from all int types (byte, short) are now supported with proper sign extension - Some generic casting code, in particular `GenIntCastDesc` was moved out of ifdefs and adapted for Wasm to support this. We don't yet support containment of casts in loads (i.e., the `GenIntCastDesc::LOAD_*` type casts). - float->int casts are implemented using the saturating non-trapping conversion instructions - This required changing the type of our Wasm opcodes to be 2 bytes instead of 1 --------- Co-authored-by: Katelyn Gadd <[email protected]> Co-authored-by: Copilot <[email protected]> Co-authored-by: SingleAccretion <[email protected]>
…nericLookupResult` (dotnet#123512) We broke outerloops in dotnet#122012. Looks like the outerloop run was already broken in that PR but nobody went over the results. ``` Process terminated. Assertion failed. !owningType.IsCanonicalSubtype(CanonicalFormKind.Any) at System.Diagnostics.DebugProvider.Fail(String, String) + 0x37 at System.Diagnostics.Debug.Fail(String, String) + 0x49 at ILCompiler.DependencyAnalysis.TypeGenericDictionaryNode..ctor(TypeDesc, NodeFactory) + 0x4c ... D:\a\_work\1\s\artifacts\bin\coreclr\windows.x64.Checked\build\Microsoft.NETCore.Native.targets(332,5): error MSB3073: The command ""D:\a\_work\1\s\artifacts\bin\coreclr\windows.x64.Checked\x64\ilc\\ilc" @"D:\a\_work\1\s\artifacts\obj\Microsoft.Extensions.Logging.Tests\Release\net11.0-windows\native\Microsoft.Extensions.Logging.Tests.ilc.rsp"" exited with code 57005. [D:\a\_work\1\s\src\libraries\Microsoft.Extensions.Logging\tests\Common\Microsoft.Extensions.Logging.Tests.csproj::TargetFramework=net11.0-windows] ``` This is because we're trying to obtain generic dictionaries of something in a canonical form from here. The callsite identifier parameter is not necessary when referencing from NonConcrete methods because we're not going to build an associated generic dictionary anyway.
Now a small amount of matches are stack allocated and the array is only optionally allocated. --------- Co-authored-by: Tarek Mahmoud Sayed <[email protected]> Co-authored-by: Jan Kotas <[email protected]>
…ef token tables (dotnet#123557) We have logic which loads a typedef, and if the typedef is loaded then we assume the method can be loaded via lookup in the MethodDef/FieldDef token map tables. HOWEVER, what this boils down to is we do a load from the TypeDef table, and if that succeeds we will do a load from the FieldDef/MethodDef token map tables, but that second load is done without an explicit memory barrier. Unfortunately, through the wonder of memory ordering behavior, its legal (and not actually all that uncommon) for the CPU to do the load from the FieldDef/MethodDef tables BEFORE it actually loads from the TypeDef table. So what can happen is that the fielddef table read can happen, then the typedef table read can happen, then the CPU does the if check to see if its ok to do the fielddef table read, then we use the value from the fielddef table read which happened earlier. This fix should fix that problem by using a VolatileLoad for all reads from the lookup maps. This is slightly slower but avoids the extremely easy to mess up dependency ordering rules. Fixes dotnet#120754 --------- Co-authored-by: Copilot <[email protected]> Co-authored-by: Aaron R Robinson <[email protected]>
…g unneeded method parameters and simplifying type resolution logic.
Co-authored-by: Jackson Schuster <[email protected]>
…hods (dotnet#123265) ## Description Refactored JIT codebase to use existing GenTree helper methods (`IsUnsigned()`, `SetUnsigned()`, `ClearUnsigned()`) instead of direct GTF_UNSIGNED flag manipulation. ## Changes **Replaced 55 explicit flag operations across 22 files:** - `gtFlags |= GTF_UNSIGNED` → `SetUnsigned()` (13 occurrences) - `gtFlags &= ~GTF_UNSIGNED` → `ClearUnsigned()` (3 occurrences) - `(gtFlags & GTF_UNSIGNED)` → `IsUnsigned()` (39 occurrences) **Moved SetUnsigned/ClearUnsigned to gentree.cpp:** - Implementations moved from inline methods in gentree.h to gentree.cpp - gentree.h now contains only method declarations for cleaner separation - Added debug assertions that print the node operator type on failure for easier debugging - Assertions validate that GTF_UNSIGNED is only set on appropriate node types: - Arithmetic operations: GT_ADD, GT_SUB - Casts: GT_CAST - All comparison operators (using `OperIsCompare()` helper): GT_EQ, GT_NE, GT_LE, GT_LT, GT_GT, GT_GE, GT_TEST_EQ, GT_TEST_NE, and on XARCH also GT_BITTEST_EQ, GT_BITTEST_NE - Multiply operations (via `OperIsMul()` helper): GT_MUL, GT_MULHI, GT_MUL_LONG ## Testing - CoreCLR builds successfully - Enhanced assertions provide helpful diagnostic output during development - No functional changes <!-- START COPILOT CODING AGENT SUFFIX --> <!-- START COPILOT ORIGINAL PROMPT --> <details> <summary>Original prompt</summary> > # A small clenaup task in JIT codebase around GTF_UNSIGNED > in `src/coreclr/jit/*` replace explicit usages of GTF_UNSIGNED over tree objects (e.g. `tmp->gtFlags |= GTF_UNSIGNED;`) with the following methods on GenTree: > > ``` > bool GenTree::IsUnsigned() const > void GenTree::SetUnsigned() > void GenTree::ClearUnsigned() > ``` > > Make sure code compiles and runtime tests pass. </details> <!-- START COPILOT CODING AGENT TIPS --> --- ✨ Let Copilot coding agent [set things up for you](https://github.com/dotnet/runtime/issues/new?title=✨+Set+up+Copilot+instructions&body=Configure%20instructions%20for%20this%20repository%20as%20documented%20in%20%5BBest%20practices%20for%20Copilot%20coding%20agent%20in%20your%20repository%5D%28https://gh.io/copilot-coding-agent-tips%29%2E%0A%0A%3COnboard%20this%20repo%3E&assignees=copilot) — coding agent works faster and does higher quality work when set up for your repo. --------- Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: EgorBo <[email protected]> Co-authored-by: Egor Bogatov <[email protected]>
… and remove obsolete availability checks (dotnet#123200) # Description macOS 12 is out of support. Updated minimum deployment target to 14.0 and Mac Catalyst to 17.0 (which shipped with macOS 14.0 Sonoma) across build configuration, native compilation, installers, and tests. Additionally, removed obsolete runtime availability checks (`__builtin_available` in Objective-C and `@available` in Swift) that are no longer necessary with the new minimum requirements. **Core configuration:** - `Directory.Build.props`: `macOSVersionMin` and `MacCatalystVersionMin` properties - `eng/native/configurecompiler.cmake`: `CMAKE_OSX_DEPLOYMENT_TARGET` and Mac Catalyst compiler targets - `eng/native/build-commons.sh`: Mac Catalyst deployment target **Build targets:** - `src/coreclr/nativeaot/BuildIntegration/Microsoft.NETCore.Native.Unix.targets` - `src/tasks/Crossgen2Tasks/Microsoft.NET.CrossGen.targets` **Mono AOT compilation:** - `src/mono/msbuild/common/MonoAOTCompiler.props`: Mac Catalyst mtriple arguments - `src/mono/msbuild/apple/build/AppleBuild.targets`: Mac Catalyst mtriple arguments **Installers:** - Distribution templates (arm64 and x64): `os-version min` requirement **Tests:** - 8 Swift interop CMakeLists.txt files **Removed obsolete availability checks:** - `src/native/libs/System.Security.Cryptography.Native.Apple/pal_networkframework.m`: Removed checks for macOS 12.3, iOS 15.4, tvOS 15.4 - `src/native/libs/System.Security.Cryptography.Native.Apple/pal_rsa.c`: Removed check for macOS 10.15, iOS 13, tvOS 13 - `src/native/libs/System.Security.Cryptography.Native.Apple/pal_sec.c`: Removed check for iOS 11.3, tvOS 11.3 - `src/native/libs/System.Security.Cryptography.Native.Apple/pal_ecc.c`: Removed check for macOS 10.15, iOS 13, tvOS 13 - `src/native/libs/System.Security.Cryptography.Native.Apple/pal_swiftbindings.swift`: HKDF functions retain `@available(iOS 14, tvOS 14, *)` annotations (macOS annotation removed since HKDF was introduced in macOS 11, below our new minimum) - `src/native/minipal/memorybarrierprocesswide.c`: Removed check for macOS 10.14, iOS 12, tvOS 9 and fallback code - `src/mono/mono/utils/mono-codeman.c`: Removed checks for macOS 11 in write protection functions - `src/mono/mono/utils/mono-mmap.c`: Removed check for macOS 11 in MAP_JIT flag handling # Customer Impact Customers on macOS 12 will no longer be able to run .NET. Customers on Mac Catalyst versions older than 17.0 will also be affected. This aligns with Apple's current support lifecycle. # Regression No. This is a forward-looking change to drop support for an EOL operating system version and align Mac Catalyst with the corresponding macOS release. # Testing Configuration changes updated version constants from 12.0 to 14.0 for macOS across 14 files and from 15.2 to 17.0 for Mac Catalyst across 7 files. Additionally, removed obsolete runtime availability checks across 8 files, simplifying the codebase by eliminating unnecessary fallback code paths for OS versions below the new minimums. HKDF functions retain appropriate `@available` annotations for iOS 14 and tvOS 14 where these APIs are above the minimum supported versions. # Risk Low. Changes consist of version constant updates and removal of obsolete runtime checks for OS versions we no longer support. All updates are consistent and the removed checks were for versions below our new minimums, so the previously guarded code will now always execute. HKDF availability annotations are correctly maintained for iOS and tvOS where needed. <!-- START COPILOT CODING AGENT SUFFIX --> <!-- START COPILOT ORIGINAL PROMPT --> <details> <summary>Original prompt</summary> > The repo currently sets > <macOSVersionMin>12.0</macOSVersionMin>. 12 is out of support. We should update to 14. </details> <!-- START COPILOT CODING AGENT TIPS --> --- 💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more [Copilot coding agent tips](https://gh.io/copilot-coding-agent-tips) in the docs. --------- Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: agocke <[email protected]> Co-authored-by: jkoritzinsky <[email protected]>
The count of switch targets does not include the default.
…otnet#123613) Trying to make it more deterministically build before trying to test and commit.
…es (dotnet#123600) Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: stephentoub <[email protected]>
…atch tracking. (dotnet#117148) (By default) Linux allows 128 inotify instances per user. By sharing the inotify instance between the FileSystemWatchers we reduce contention with other applications. Fixes dotnet#62869.
…ments (dotnet#123599) Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: stephentoub <[email protected]>
…um when isFinalBlock=false (dotnet#123313) Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: MihaZupan <[email protected]> Co-authored-by: Miha Zupan <[email protected]>
In such cases we won't have set up a stack pointer.
## Description `AND` between a mask and constant zero was incorrectly folded to return the mask value instead of zero in the JIT's `gtFoldExprHWIntrinsic`. ```csharp // Expected: <0, 0, 0, ...> Actual (buggy): <-1, -1, -1, ...> Vector512.Equals(v, v) & Vector512.LessThan(v.AsUInt32(), Vector512<uint>.Zero).AsInt32(); ``` ### Changes - **`src/coreclr/jit/gentree.cpp`**: For mask types in `GT_AND` case, changed `resultNode = otherNode` to `resultNode = gtWrapWithSideEffects(cnsNode, otherNode, GTF_ALL_EFFECT)` — matching the non-mask vector case - **`src/tests/JIT/Regression/JitBlue/Runtime_123399/`**: Added regression test (requires AVX-512) <!-- START COPILOT ORIGINAL PROMPT --> <details> <summary>Original prompt</summary> > > ---- > > *This section details on the original issue you should resolve* > > <issue_title>JIT: Mask AND with zero incorrectly folded</issue_title> > <issue_description>### Description > > `AND` between a mask and a constant zero is incorrectly folded to the mask value in net10.0. > > https://github.com/dotnet/runtime/blob/2075e8ffaa8cf598dac9c09f5f690778ef1a6c8e/src/coreclr/jit/gentree.cpp#L32988-L32992 > > ### Reproduction Steps > > ```c# > using System.Runtime.CompilerServices; > using System.Runtime.Intrinsics; > > Console.WriteLine(AndMaskZero(Vector512<int>.One)); > > [MethodImpl(MethodImplOptions.NoInlining)] > static Vector512<int> AndMaskZero(Vector512<int> v) => > Vector512.Equals(v, v) & Vector512.LessThan(v.AsUInt32(), Vector512<uint>.Zero).AsInt32(); > ``` > > ### Expected behavior > > ```bash > > dotnet run -c release -f net9.0 > ``` > > `<0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0>` > > ### Actual behavior > > ```bash > > dotnet run -c release -f net10.0 > ``` > `<-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1>` > > ### Regression? > > Yes, worked in 9.0 > > ### Configuration > > Requires AVX-512 > > </issue_description> > > <agent_instructions>Fix the bug in jit (see comments in the issue) + add a regression test (JitBlue)</agent_instructions> > > ## Comments on the Issue (you are @copilot in this section) > > <comments> > <comment_new><author>@JulieLeeMSFT</author><body> > Thanks, @saucecontrol, for the report. > @anthonycanino, can your team please take a look? > CC @kg. </body></comment_new> > </comments> > </details> <!-- START COPILOT CODING AGENT SUFFIX --> - Fixes dotnet#123399 <!-- START COPILOT CODING AGENT TIPS --> --- 💬 We'd love your input! Share your thoughts on Copilot coding agent in our [2 minute survey](https://gh.io/copilot-coding-agent-survey). --------- Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: EgorBo <[email protected]>
Fixes dotnet#123583 ## Description Adds a FileCheck test that verifies codegen doesn't contain bounds checks (`CORINFO_HELP_RNGCHKFAIL`) for patterns where the JIT should prove access is in-bounds. ### Test Methods - **ComplexBinaryOperators** - Masking/shifting restricts index to 0-63 for 65-element span - **LastCharCheck** - String length comparison before indexed access - **CountDigits** - `ulong.Log2()` result (0-63) indexes 64-element span - **AndByConst** - Bitwise AND with constant - **AndByLength** - Bitwise AND with `(span.Length - 1)` pattern ### Files Added - `src/tests/JIT/opt/RangeChecks/ElidedBoundsChecks.cs` - Test methods with `X64-NOT` and `ARM64-NOT` FileCheck annotations - `src/tests/JIT/opt/RangeChecks/ElidedBoundsChecks.csproj` - Project with `HasDisasmCheck=true` ```csharp [MethodImpl(MethodImplOptions.NoInlining)] static byte AndByLength(int i) { // X64-NOT: CORINFO_HELP_RNGCHKFAIL // ARM64-NOT: CORINFO_HELP_RNGCHKFAIL ReadOnlySpan<byte> span = new byte[] { 1, 2, 3, 4 }; return span[i & (span.Length - 1)]; } ``` <!-- START COPILOT CODING AGENT SUFFIX --> <!-- START COPILOT ORIGINAL PROMPT --> <details> <summary>Original prompt</summary> > Add a test that uses FileCheck to varify that codegen doesn't contain bounds checks at all > > the test should be located in `src\tests\JIT\opt\RangeChecks` folder > > Check on x64 and arm64, here are the asm examples how bounds checks look like: > ``` > ; x64: > call CORINFO_HELP_RNGCHKFAIL > ; arm64: > bl CORINFO_HELP_RNGCHKFAIL > ``` > > Use .cs and .csproj in `C:\prj\runtime-main2\src\tests\JIT\opt\Subtract\` as a reference on how to create FileCheck tests (HasDisasmCheck) > > Here are the functions I want to make sure they don't have bounds checks under optimizations: > > ```csharp > public class ElidedBoundsChecks > { > [MethodImpl(MethodImplOptions.NoInlining)] > static int ComplexBinaryOperators(byte inData) > { > ReadOnlySpan<byte> base64 = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/="u8; > return base64[((inData & 0x03) << 4) | ((inData & 0xf0) >> 4)]; > } > > [MethodImpl(MethodImplOptions.NoInlining)] > static bool LastCharCheck(string prefix, string path) > { > if (prefix.Length < path.Length) > return (path[prefix.Length] == '/'); > return false; > } > > [MethodImpl(MethodImplOptions.NoInlining)] > static nint CountDigits(ulong value) > { > ReadOnlySpan<byte> log2ToPow10 = > [ > 1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, > 6, 6, 6, 7, 7, 7, 7, 8, 8, 8, 9, 9, 9, 10, 10, 10, > 10, 11, 11, 11, 12, 12, 12, 13, 13, 13, 13, 14, 14, 14, 15, 15, > 15, 16, 16, 16, 16, 17, 17, 17, 18, 18, 18, 19, 19, 19, 19, 20 > ]; > return log2ToPow10[(int)ulong.Log2(value)]; > } > > [MethodImpl(MethodImplOptions.NoInlining)] > static byte AndByConst(int i) > { > ReadOnlySpan<byte> span = new byte[] { 1, 2, 3, 4 }; > return span[i & 2]; > } > > [MethodImpl(MethodImplOptions.NoInlining)] > static byte AndByLength(int i) > { > ReadOnlySpan<byte> span = new byte[] { 1, 2, 3, 4 }; > return span[i & (span.Length - 1)]; > } > } > ``` </details> <!-- START COPILOT CODING AGENT TIPS --> --- ✨ Let Copilot coding agent [set things up for you](https://github.com/dotnet/runtime/issues/new?title=✨+Set+up+Copilot+instructions&body=Configure%20instructions%20for%20this%20repository%20as%20documented%20in%20%5BBest%20practices%20for%20Copilot%20coding%20agent%20in%20your%20repository%5D%28https://gh.io/copilot-coding-agent-tips%29%2E%0A%0A%3COnboard%20this%20repo%3E&assignees=copilot) — coding agent works faster and does higher quality work when set up for your repo. --------- Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: EgorBo <[email protected]> Co-authored-by: Egor Bogatov <[email protected]>
…otnet#123319) Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: stephentoub <[email protected]> Co-authored-by: adamsitnik <[email protected]> Co-authored-by: EgorBo <[email protected]>
…oryFile (dotnet#122947) Co-authored-by: copilot-swe-agent[bot] <[email protected]> Co-authored-by: Stephen Toub <[email protected]> Co-authored-by: jozkee <[email protected]> Co-authored-by: stephentoub <[email protected]>
davidwrighton
approved these changes
Jan 26, 2026
davidwrighton
left a comment
davidwrighton
left a comment
There was a problem hiding this comment.
LGTM for now. Please file an issue to unify the ContinuationMethodTable creation path so we don't have a path for jitting and a path for R2R, but this will do for now. Assign the issue to me or @VSadov
Emits Task-returning async method thunks in crossgen2 as MethodSignatures with the AsyncVariant flag set Adds `GetPrimaryMethodDesc` to get the most natural "primary" method desc for MethodDesc's that point to the same Ecma/Instantiated method. This is used in the places where we cast to EcmaMethod to get relevant metadata information. GCRefMap emission also needed to be updated to account for the continuation parameter. Runtime's `FakeGcScanRoots` (used to verify the ReadyToRun images GCRefMap in debug config) also needed to updated to handle the continuation parameter. In R2R exception info clauses, a new flag is added to use when the handler catches System.Exception. It's not guaranteed that a token for System.Exception exists in the context of the EH info struct, and there is no equivalent of a module override for the structs. In the MethodWithToken constructor, it's expected that callers pass the Token that points to the exact method that is being called on the correct instantiated type. It's again not guaranteed that we have valid tokens for the method being called in the context of an instantiated type (technically any call to `G<MyClass>.M()` will have a token for G<Class>, but when we compile the canonicalize method, we call the async `G<!0>.M()` in the IL, and we don't have a token for `G<!0>`). For this reason, we wrap the MethodIL in a ManifestModuleWrappedMethodIL, and when we construct a MethodWithToken, we resolve the OwningType with the MutableModule token, but swap out the Token for the typical method definition token and force emission of a full MethodSignature in the Fixup. --------- Co-authored-by: David Wrighton <[email protected]> Co-authored-by: Jan Kotas <[email protected]>
Owner
Author
|
Made a new PR to target the dotnet repo: dotnet#123643 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Encode Continuation types in R2R images using a new fixup kind (READYTORUN_FIXUP_Continuation_Layout = 0x37):
Add runtime support to read continuation type fixups from R2R images:
Also adds helper constants for future continuation allocation helpers.
Note: These code paths are not yet exercised as async2 methods are not currently compiled to ReadyToRun images.