[cDAC] Add interpreter support for stack walking and diagnostics by max-charlamb · Pull Request #126520 · dotnet/runtime

max-charlamb · 2026-04-03T21:59:19Z

Summary

Adds interpreter support to the cDAC (contract-based Data Access Component), enabling diagnostic tools to correctly walk stacks containing interpreter frames, resolve interpreter precodes, and retrieve method information for interpreted methods.

Changes

Native Data Descriptors

Added interpreter type descriptors to datadescriptor.inc: InterpreterRealCodeHeader, InterpreterPrecodeData, InterpByteCodeStart, InterpMethod, InterpMethodContextFrame, InterpreterFrame
Added InterpreterFrame to frames.h explicit frame list for cDAC visibility

Execution Manager - Interpreter JIT Manager

New ExecutionManagerCore.InterpreterJitManager handles code address lookups for interpreter code heaps
GetCodeBlockHandle now searches interpreter code heaps when JIT heaps don't contain the address
GetMethodDesc resolves MethodDesc from interpreter code headers

Precode Resolution (`GetInterpreterCodeFromInterpreterPrecodeIfPresent`)

New API on IPrecodeStubs matching the native DAC pattern: each call site resolves interpreter precodes before passing addresses to ExecutionManager
Passthrough semantics: returns original address if not an interpreter precode
NOTHROW contract via VirtualReadException catch
Applied at 4 call sites: GetMethodDescData, CopyNativeCodeVersionToReJitData, GetTieredVersions, GetILAddressMap

Stack Walking

FrameIterator handles InterpreterFrame - extracts MethodDesc and native code pointer from InterpMethodContextFrame
StackWalk_1 resolves interpreter frames during enumeration

RuntimeTypeSystem

MethodValidation updated to handle interpreter method descriptors (IsInterpreterStub flag, chunk validation)

Documentation

Updated ExecutionManager.md, PrecodeStubs.md, StackWalk.md with interpreter support details

Tests

Unit tests: ExecutionManagerTests (interpreter JIT manager), FrameIteratorTests (interpreter frame handling), PrecodeStubsTests (interpreter precode resolution), MethodDescTests (interpreter method validation)
Dump tests: InterpreterStackDumpTests - 3 integration tests using a mixed JIT/interpreter stack debuggee (InterpreterStack) that validates interleaved frame layout, precode resolution, and thread enumeration
All 1647 unit tests pass, all 358 dump tests pass (218 passed, 140 skipped)

Copilot

Pull request overview

Copilot reviewed 67 out of 68 changed files in this pull request and generated 1 comment.

Copilot

Pull request overview

Copilot reviewed 63 out of 64 changed files in this pull request and generated 5 comments.

Copilot

Pull request overview

Copilot reviewed 63 out of 64 changed files in this pull request and generated 3 comments.

Copilot

Pull request overview

Copilot reviewed 62 out of 63 changed files in this pull request and generated 6 comments.

Copilot

Pull request overview

Copilot reviewed 63 out of 64 changed files in this pull request and generated 1 comment.

Comments suppressed due to low confidence (3)

src/native/managed/cdac/tests/DumpTests/ISOSDacInterfaceTests.cs:1

ConditionalTheory is used but this file doesn’t import Microsoft.DotNet.XUnitExtensions, so it will fail to compile unless there’s a global using elsewhere. Add using Microsoft.DotNet.XUnitExtensions; (and optionally simplify SkipTestException usage) to make the dependency explicit.
src/native/managed/cdac/tests/DumpTests/DumpTests.targets:1
_DebuggeeEnvVars is already escaped earlier, but it is escaped again when passed through nested MSBuild calls. This double-escaping prevents semicolon-separated env var lists (e.g. A=B;C=D) from being restored correctly when later unescaped for Exec EnvironmentVariables (you’d end up with %3B instead of ;). Pass the value through without re-escaping (or unescape before re-escaping) so multi-variable lists remain semicolon-delimited at the final Exec.
src/native/managed/cdac/Microsoft.Diagnostics.DataContractReader.Contracts/Contracts/StackWalk/Context/ARM64/ARM64Unwinder.cs:1
Similar to the ARM change, returning false on missing unwind info risks leaving the context unchanged if the caller doesn’t propagate this failure (most of the stack walk logic expects unwind errors to surface as exceptions). To avoid stalls or partial walks, either throw an InvalidOperationException here (as before, or consistent with other unwinders) or update the unwind call sites to handle a false return deterministically.

// Licensed to the .NET Foundation under one or more agreements.

Copilot

Pull request overview

Copilot reviewed 63 out of 64 changed files in this pull request and generated 5 comments.

Add cDAC contracts and implementations to walk interpreter frames during stack walks and identify interpreter-managed methods. The cDAC stack walker now mirrors the native DAC behavior: * Yields `InterpreterFrame` as a runtime frame marker (pMD = NULL), matching the native DAC. * Implements interpreter virtual unwind by following the `InterpMethodContextFrame.pParent` chain so each interpreted method is yielded as a frameless frame. * Resolves the top `InterpMethodContextFrame` from the `InterpreterFrame` in the same way as the native runtime (`GetTopInterpMethodContextFrame`). * Adds `InterpreterJitManager` for interpreter `CodeBlockHandle` resolution and a precode-stubs fallthrough that recognizes `InterpreterPrecode`. Includes data descriptors for `InterpreterFrame`, `InterpMethodContextFrame`, `InterpMethod`, `InterpByteCodeStart`, `InterpreterPrecodeData`, and `InterpreterRealCodeHeader`, plus documentation updates and unit tests covering the new contract behavior. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Add dump-test infrastructure for the interpreter scenarios and tests that exercise the cDAC interpreter stack walker against real coredumps: * `InterpreterStack` debuggee: single-threaded debuggee with a JIT->interpreter->JIT->interpreter call chain that triggers a `FailFast` from inside an interpreted method. * `InterpreterStackDoubleWalk` debuggee: multi-threaded debuggee where a worker thread is parked deep in an interpreted call chain while the main thread captures the dump. This exercises walking a thread other than the crashing thread and asserts the cDAC walker does not produce duplicated `InterpreterFrame` markers. The new dump tests verify the interleaved JIT/interpreter frame layout, the absence of doubled `InterpreterFrame` markers, and that `DumpTestStackWalker` adjacency assertions hold across the full interpreter call chain. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

When CheckForSkippedFrames clones the context for an interpreter IP and calls Unwind, the interpreter's GetUnwindInfo returns TargetPointer.Null. AMD64Unwinder already guards against this by checking for null and returning false; ARM and ARM64 unwinders did not, and crashed reading RuntimeFunction at address 0 (VirtualReadException at 0x00000000). Add the same null guard to ARM and ARM64 to make all three platforms behaviorally consistent. Also remove StackWalk_NoDoubledInterpreterFrames_WithDebuggerFilterContext since it cannot run in CI (depends on a manually-collected cdb dump in the local repo) and the scenario is invalid per native PR dotnet#126953. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

- Restore single ternary at StackWalk_1.cs (was needlessly verbose) - Remove unrelated infinite-loop guard around Unwind - Replace unicode arrows/em-dashes with -> and -- in code/comments/docs - Drop direct line-number references in comments - Make FrameHelpers access modifiers consistent: public for externally callable methods, private for helpers Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

- Add SetContextToInterpMethodContextFrame and VirtualUnwindInterpreterCallFrame helpers in FrameHelpers, named to mirror native (frames.cpp / eetwain.cpp) - Add IsFaulting field on InterpreterFrame and Stack field on InterpMethodContextFrame; thread Data.InterpreterFrame.Address through - Add RawContextFlags abstraction across all platform contexts so the helper can OR in CONTEXT_EXCEPTION_ACTIVE for faulting top frames - Gate IExecutionManager.IsFunclet on JitType.Interpreter so interpreter code (no native unwind info) reports false without throwing in GetFuncletStartAddress - Remove flaky InterpreterStackDoubleWalk debuggee + tests (timing-sensitive SpinStep loop) - Address Copilot bot feedback: MSBuild escaping in DumpTests.targets, funclet null-safety for interpreter Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

…pe; hide GetFrameHandler behind helper - SOSDacImpl.GetCodeHeaderData: when JIT type is Interpreter, decode the GC info using DecodeInterpreterGCInfo instead of DecodePlatformSpecificGCInfo. This mirrors native ClrDataAccess::GetCodeHeaderData routing through EECodeInfo::GetCodeManager()->GetFunctionSize, where interpreter code goes through InterpreterCodeManager::GetFunctionSize / InterpreterGcInfoDecoder. - FrameHelpers: make GetFrameHandler private and add a semantic helper ApplyInterpreterFrameTransition(context, interpreterFrameAddress) that encapsulates reading the InterpreterFrame as a FramedMethodFrame and invoking HandleTransitionFrame on it. Update StackWalk_1.InterpreterVirtualUnwind to use the new helper instead of reaching into the frame-handler dispatch directly. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Adds an opt-in Interpreter SOS test leg to runtime-diagnostics.yml that exercises the diagnostics-side SOSInterpreterTests against a Checked CoreCLR drop. FEATURE_INTERPRETER is only compiled into Debug/Checked CoreCLR, so the existing Release build leg cannot run this test. Changes: * eng/pipelines/diagnostics/runtime-diag-job.yml: adds a testInterpreter parameter (default false). When true, _TestInterpreterArgs resolves to -testInterpreter and is forwarded to the diagnostics build script alongside the existing -useCdac / -noFallback flags. Default-false invocation is identical to today. * eng/pipelines/runtime-diagnostics.yml: - New build leg that produces a Checked CoreCLR (libs/SDK stay Release) under a distinct artifact name (..._coreclr_Checked). - New Interpreter test job that depends on the Checked build leg, downloads its artifact, and is the only job that sets testInterpreter: true. - The cDAC / cDAC_no_fallback / DAC test jobs and the existing Release build leg are unchanged. Tested with a manual ADO queue using diagnosticsBranch pointed at the companion diagnostics PR (dotnet/diagnostics#5829). Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

…CallFrame handlers Mirror the per-architecture native InlinedCallFrame::UpdateRegDisplay_Impl logic in the corresponding cDAC frame handlers: - AMD64FrameHandler.HandleInlinedCallFrame: sets Rcx (Windows) or Rdi (Unix SysV) to the InterpreterFrame address when the next Frame is an InterpreterFrame, matching src/coreclr/vm/amd64/cgenamd64.cpp:212-218. - ARM64FrameHandler.HandleInlinedCallFrame: sets X0 to the InterpreterFrame address when the next Frame is an InterpreterFrame, matching src/coreclr/vm/arm64/stubs.cpp:408-414. ARM, x86, LoongArch64, and RISCV64 native InlinedCallFrame::UpdateRegDisplay_Impl do NOT perform this update, so the corresponding cDAC handlers (or BaseFrameHandler inheritance) do not either. A protected GetNextFrame helper is added to BaseFrameHandler so each handler can inspect the chain itself without changing the IPlatformFrameHandler interface (handles the FRAME_TOP all-ones terminator). BaseFrameHandler also constructs its own FrameHelpers from the target so derived handlers can inline calls to GetFrameType. Without this update, cDAC reports the thread's literal saved Rcx for frames between an InlinedCallFrame and its successor InterpreterFrame, while the legacy DAC reports the InterpreterFrame address. This trips Debug.Assert(contextStruct.Equals(localContextStruct)) in ClrDataStackWalk.GetContext during !ClrStack on a thread captured during a P/Invoke from interpreted code. Verified locally against the SOS.InterpreterStackTest.Heap.dmp captured in CI: !ClrStack and !PrintException both succeed with the cDAC parity check enabled, walking through [InlinedCallFrame], JIT IL stub frames, [InterpreterFrame: ...], and the interpreted method frames in correct order. All 2081 cDAC unit tests continue to pass. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

The cDAC's InterpreterJitManager (added in this PR) registers interpreter code as code blocks with an interpreter-format DebugInfo, and DebugInfo_2.GetMethodNativeMap successfully decodes that DebugInfo into an OffsetMapping list. This means cDAC produces real, correct IL offsets for interpreter pseudo-IPs (e.g. for !ClrStack -lines source-line resolution), where the legacy DAC bails out early with E_INVALIDARG because ExecutionManager::GetNativeCodeVersion(address) returns null for interpreter IPs (daccess.cpp:5660-5666). Switch the parity validation in GetILOffsetsByAddress from the default AllowDivergentFailures (which rejects cDAC-success vs DAC-failure) to AllowCdacSuccess so cDAC's better behavior is permitted. Gate the offset-comparison block on hrLocal == S_OK as well, so we don't spuriously compare against a zero-initialized localOffsetsNeeded / localIlOffsets when only cDAC succeeded. Verified locally against the SOS.InterpreterStackTest.Heap.dmp from CI: !ClrStack -lines now resolves source lines for the JIT-compiled EH frames AND for the interpreter frames (InterpTestMethodThrow @ 19, InterpTestMethodRunNested @ 13, InterpreterStackTestApp.Main @ 12), with the cDAC parity check enabled. All 2081 cDAC unit tests pass. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Adds ISOSDacInterfaceTests dump test class mirroring ISOSDacInterface13Tests, with a GetCodeHeaderData test that validates TYPE_INTERPRETER routing through the interpreter precode unwrap path. Also switches the InterpreterStack debuggee from a full dump to a heap dump (107 MB -> 16 MB) since the heap dump captures all memory needed for interpreter stack walking and code-header lookup. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Linux createdump segfaults during heap-dump region enumeration when the interpreter is active, so InterpreterStack switches to full dumps. Tracked by dotnet#128044. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

…-bit targets Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Keep the original catch (VirtualReadException) clause; the NotImplementedException widening will be handled separately. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Missed during the rebase onto main (which removed JitType in favor of the unified CodeKind enum). Updates AssertInterpreted/AssertJitted to call IExecutionManager.GetCodeKind with the resolved code pointer. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot

Pull request overview

Copilot reviewed 63 out of 64 changed files in this pull request and generated 4 comments.

- MockDescriptors.ExecutionManager.cs: Remove AddRangeListSection helper that created an invalid RangeSection (RangeList flag set but RangeList and JitManager pointers null). Switch the one caller in GetCodeBlockHandle_InterpreterPrecode_ReturnsNull to use the AddRangeListRangeSection helper which properly initializes RangeList and JitManager. - ExecutionManager.md: Fix missing space after backticked method name in the NonVirtualEntry2MethodDesc paragraph. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot AI review requested due to automatic review settings April 3, 2026 21:59

github-actions Bot added the area-Diagnostics-coreclr label Apr 3, 2026

dotnet-policy-service Bot assigned max-charlamb Apr 3, 2026

Copilot started reviewing on behalf of max-charlamb April 3, 2026 22:00 View session

This comment was marked as outdated.

Sign in to view

build-analysis Bot mentioned this pull request Apr 4, 2026

System.Runtime.Tests SIGSEGV crash on ARM64 Linux (azurelinux.3.arm64) #126516

Closed

jkoritzinsky reviewed Apr 4, 2026

View reviewed changes

Comment thread ....Diagnostics.DataContractReader.Contracts/Contracts/StackWalk/FrameHandling/FrameIterator.cs Outdated

Copilot AI review requested due to automatic review settings April 8, 2026 14:54

max-charlamb force-pushed the dev/max-charlamb/cdac-interpreter-support branch from 7e1ec2a to cdaefe0 Compare April 8, 2026 14:54

Copilot started reviewing on behalf of max-charlamb April 8, 2026 14:55 View session

This comment was marked as outdated.

Sign in to view

Copilot AI review requested due to automatic review settings April 8, 2026 17:28

Copilot started reviewing on behalf of max-charlamb April 8, 2026 17:29 View session

This comment was marked as outdated.

Sign in to view

Copilot AI review requested due to automatic review settings April 8, 2026 20:29

Copilot started reviewing on behalf of max-charlamb April 8, 2026 20:30 View session

This comment was marked as outdated.

Sign in to view

Copilot AI review requested due to automatic review settings April 9, 2026 16:55

max-charlamb force-pushed the dev/max-charlamb/cdac-interpreter-support branch from d425631 to d5c6cd8 Compare April 9, 2026 16:55

Copilot started reviewing on behalf of max-charlamb April 9, 2026 16:56 View session

This comment was marked as outdated.

Sign in to view

max-charlamb marked this pull request as ready for review April 10, 2026 15:37

Copilot AI review requested due to automatic review settings April 10, 2026 15:37

Copilot started reviewing on behalf of max-charlamb April 10, 2026 15:38 View session

max-charlamb force-pushed the dev/max-charlamb/cdac-interpreter-support branch from 129584c to c19d421 Compare May 7, 2026 17:22

Copilot AI review requested due to automatic review settings May 7, 2026 17:28

max-charlamb force-pushed the dev/max-charlamb/cdac-interpreter-support branch from c19d421 to d600dfb Compare May 7, 2026 17:28

Copilot started reviewing on behalf of max-charlamb May 7, 2026 17:28 View session

Copilot AI reviewed May 7, 2026

View reviewed changes

Comment thread src/native/managed/cdac/tests/DumpTests/DumpTests.targets

max-charlamb force-pushed the dev/max-charlamb/cdac-interpreter-support branch from d600dfb to 57325e4 Compare May 7, 2026 17:40

Copilot AI reviewed May 7, 2026

View reviewed changes

Copilot AI reviewed May 8, 2026

View reviewed changes

Comment thread ...soft.Diagnostics.DataContractReader.Contracts/Contracts/StackWalk/Context/ARM/ARMUnwinder.cs

max-charlamb mentioned this pull request May 11, 2026

createdump SIGSEGVs when generating a heap-type minidump for an interpreter process on Linux #128044

Open

Copilot AI reviewed May 11, 2026

View reviewed changes

max-charlamb and others added 16 commits May 12, 2026 17:08

address comments

4ef488e

fix doubled frame issue

b307e0c

Sign-extend interpreter code address for GetCodeHeaderData test on 32…

cb6f549

…-bit targets Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Revert PrecodeStubs_Common.cs catch widening

d3b715c

Keep the original catch (VirtualReadException) clause; the NotImplementedException widening will be handled separately. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot AI reviewed May 13, 2026

View reviewed changes

Conversation

max-charlamb commented Apr 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Changes

Native Data Descriptors

Execution Manager - Interpreter JIT Manager

Precode Resolution (GetInterpreterCodeFromInterpreterPrecodeIfPresent)

Stack Walking

RuntimeTypeSystem

Documentation

Tests

Uh oh!

This comment was marked as outdated.

Uh oh!

Uh oh!

This comment was marked as outdated.

Uh oh!

This comment was marked as outdated.

Uh oh!

This comment was marked as outdated.

Uh oh!

This comment was marked as outdated.

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

max-charlamb commented Apr 3, 2026 •

edited

Loading

Precode Resolution (`GetInterpreterCodeFromInterpreterPrecodeIfPresent`)