[interp] Fix doubled interpreter frames in DAC/DBI stack walks by kotlarmilos · Pull Request #126953 · dotnet/runtime

kotlarmilos · 2026-04-15T17:09:36Z

Description

When DAC/DBI seeds StackFrameIterator from a CONTEXT pointing into interpreted code, m_crawl.pFrame is initialized to the thread's top explicit Frame, which is the InterpreterFrame that owns the executing InterpMethodContextFrame chain.

ResetRegDisp reads the owning InterpreterFrame* from the CONTEXT's first-arg register and advances m_crawl.pFrame past it before ProcessCurrentFrame runs.

Tests

Fixes the following interpreter debugger test failures:

StackWalking.NestedException
StackWalking.ChildParentTest

Copilot

Pull request overview

Fixes two interpreter-specific debugger stack-walk issues in CoreCLR that caused duplicated frames and incorrect source line mapping when unwinding interpreter call frames.

Changes:

Adjust StackFrameIterator to advance the explicit frame chain past InterpreterFrame after the DummyCallerIP transition to native code, preventing the debugger path from re-entering the interpreter chain and duplicating frames.
Adjust interpreter unwinding to set the caller IP to (pFrame->ip - 1) so return-address mapping lands within the call instruction range (aligning with JIT-style return-address adjustment).

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
src/coreclr/vm/stackwalk.cpp	Skips past `InterpreterFrame` in the explicit frame chain after transitioning out of the last interpreted frame, avoiding duplicate interpreter frame reporting in debugger walks.
src/coreclr/vm/eetwain.cpp	Adjusts the unwound interpreter caller IP by `-1` to correct non-leaf frame line mapping in stack traces/debugger views.

dotnet-policy-service · 2026-04-15T17:17:06Z

Tagging subscribers to this area: @steveisok, @tommcdon, @dotnet/dotnet-diag
See info in area-owners.md if you want to be subscribed.

Fix two issues in the StackFrameIterator for FEATURE_INTERPRETER: 1. Double-frame bug (stackwalk.cpp): After the DummyCallerIP transition from the last interpreter frame to native code, advance m_crawl.pFrame past the InterpreterFrame. Without this, the DAC/DBI debugger path (which starts with context already pointing into the interpreter chain) would encounter the InterpreterFrame again and re-enter the chain, reporting each interpreter frame twice. The GC path is unaffected because it already advances pFrame when entering the chain via the explicit frame handler. 2. Line number off-by-one (eetwain.cpp): In VirtualUnwindInterpreterCallFrame, subtract 1 from pFrame->ip when unwinding to a parent frame. The interpreter stores ip after advancing past the call instruction (ip += N), so the raw IP points to the start of the next instruction. Subtracting 1 places the IP within the call instruction, ensuring the debug info maps to the correct IL offset and source line for non-leaf frames. This is analogous to how JIT return addresses are adjusted for native code. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

The stack walker should return the real IP. IP adjustment for call sites belongs on the consumer side (EH already does it; debugger-side handling for the interpreter line-number off-by-one is tracked separately). Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot

Pull request overview

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

noahfalk · 2026-04-28T09:59:20Z

+                // frame, because the stack frame iterator assumes that when it is walking interpreted frames, it has
+                // already processed the interpreter frame. Without this skip, the stack walk would end up walking the
+                // interpreted frames twice.
+                m_crawl.GotoNextFrame();


The normal invariant enforced by NextRaw() is that m_crawl moves past the InterpreterFrame before calling ProcessCurrentFrame() with a CONTEXT pointing at interpreted code. Also in general it is the caller's job to advance the Frame, not ProcessCurrentFrame() (the comments on this method say this is called after we've stopped). I'd suggest:

In StackFrameIterator::Init() assert that the CONTEXT doesn't point at interpreted code.

In StackFrameIterator::ResetRegDisp - there is already a bunch of logic that tries to start the Frame pointer at the correct position in the Frame chain. Currently that logic uses stack pointer comparisons but it probably needs to be updated to handle CONTEXTs that reference interpreter code. I'm guessing we'd need to map the interpreter code to its logically associated InterpreterFrame, then advance the Frame pointer to the next Frame in the chain.

Remove this code here because now ResetRegDisp() will have initialized the state correctly.

Done - Init asserts the CONTEXT isn't interpreted, ResetRegDisp reads the owning InterpreterFrame* from the CONTEXT's first-arg register and advances m_crawl.pFrame past it before ProcessCurrentFrame runs.

Move the doubled-interpreter-frames fix from ProcessCurrentFrame to ResetRegDisp so the StackFrameIterator starts from a correct state when DAC/DBI seeds it with a CONTEXT pointing into interpreted code. - Init: assert the CONTEXT does not reference interpreted code; that path is for OS-thread-top contexts and only DAC/DBI seeds interpreter contexts (via ResetRegDisp). - ResetRegDisp: when CONTEXT references interpreted code, locate the owning InterpreterFrame whose InterpMethodContextFrame chain contains the SP, and advance m_crawl.pFrame past it. - ProcessCurrentFrame: drop the post-hoc workaround that skipped the re-encountered InterpreterFrame; no longer needed. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

When the StackFrameIterator is seeded from a CONTEXT pointing into interpreted code, InterpreterFrame::SetContextToInterpMethodContextFrame already writes the owning InterpreterFrame pointer into the first-arg register. Read it from the CONTEXT instead of searching the explicit Frame chain for an InterpMethodContextFrame whose address matches SP. Drops the chain search and removes the implicit assumption that the seeded context always belongs to the topmost InterpreterFrame, which also generalizes correctly to mid-stack seeds. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

janvorli

LGTM, thank you!

noahfalk

LGTM!

Copilot AI review requested due to automatic review settings April 15, 2026 17:09

github-actions Bot added the area-CodeGen-Interpreter-coreclr label Apr 15, 2026

dotnet-policy-service Bot assigned kotlarmilos Apr 15, 2026

Copilot started reviewing on behalf of kotlarmilos April 15, 2026 17:10 View session

Copilot AI reviewed Apr 15, 2026

View reviewed changes

kotlarmilos changed the title ~~[Interpreter] Fix double frames and line number off-by-one in debugger stack walks~~ [clr-ios] Fix double frames and line number off-by-one in debugger stack walks Apr 15, 2026

kotlarmilos added area-Diagnostics-coreclr os-ios Apple iOS and removed area-CodeGen-Interpreter-coreclr labels Apr 15, 2026

This was referenced Apr 15, 2026

Unable to pull image from mcr.microsoft.com #117164

Open

System.Net.NameResolution.Tests DNS failures: Name or service not known #126641

Open

kotlarmilos force-pushed the interp-fix-stackwalking branch from 8113c17 to 123c90a Compare April 16, 2026 09:02

janvorli reviewed Apr 16, 2026

View reviewed changes

Comment thread src/coreclr/vm/eetwain.cpp Outdated

Proper fix of the doubled stack trace issue

beab360

Copilot AI review requested due to automatic review settings April 16, 2026 17:05

Copilot started reviewing on behalf of janvorli April 16, 2026 17:06 View session

Copilot AI reviewed Apr 16, 2026

View reviewed changes

Comment thread src/coreclr/vm/eetwain.cpp Outdated

kotlarmilos mentioned this pull request Apr 22, 2026

[clr-ios] Enable debugging on CoreCLR mobile platforms #124744

Open

18 tasks

kotlarmilos mentioned this pull request Apr 22, 2026

[clr-interp] Line number off-by-one for non-leaf interpreter frames in debugger stack traces #127277

Open

kotlarmilos changed the title ~~[clr-ios] Fix double frames and line number off-by-one in debugger stack walks~~ [interp] Fix doubled interpreter frames in debugger stack walks Apr 22, 2026

kotlarmilos requested a review from janvorli April 22, 2026 13:06

kotlarmilos marked this pull request as ready for review April 22, 2026 13:06

kotlarmilos requested review from Copilot and noahfalk April 22, 2026 13:06

Copilot started reviewing on behalf of kotlarmilos April 22, 2026 13:07 View session

Copilot AI reviewed Apr 22, 2026

View reviewed changes

Comment thread src/coreclr/vm/stackwalk.cpp Outdated

build-analysis Bot mentioned this pull request Apr 22, 2026

browser-wasm linux Release LibraryTests queues timing out #117974

Open

janvorli mentioned this pull request Apr 23, 2026

[cDAC] Add interpreter support for stack walking and diagnostics #126520

Open

noahfalk requested changes Apr 28, 2026

View reviewed changes

kotlarmilos changed the title ~~[interp] Fix doubled interpreter frames in debugger stack walks~~ [interp] Fix doubled interpreter frames in DAC/DBI stack walks Apr 28, 2026

kotlarmilos requested a review from noahfalk April 28, 2026 12:58

Copilot AI review requested due to automatic review settings April 28, 2026 14:47

janvorli approved these changes Apr 28, 2026

View reviewed changes

noahfalk approved these changes Apr 28, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[interp] Fix doubled interpreter frames in DAC/DBI stack walks#126953

[interp] Fix doubled interpreter frames in DAC/DBI stack walks#126953
kotlarmilos wants to merge 5 commits intodotnet:mainfrom
kotlarmilos:interp-fix-stackwalking

kotlarmilos commented Apr 15, 2026 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

dotnet-policy-service Bot commented Apr 15, 2026

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

noahfalk Apr 28, 2026

Uh oh!

kotlarmilos Apr 28, 2026 •

edited

Loading

Uh oh!

janvorli left a comment

Uh oh!

noahfalk left a comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Conversation

kotlarmilos commented Apr 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Tests

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

dotnet-policy-service Bot commented Apr 15, 2026

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!

noahfalk Apr 28, 2026

Choose a reason for hiding this comment

Uh oh!

kotlarmilos Apr 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

janvorli left a comment

Choose a reason for hiding this comment

Uh oh!

noahfalk left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

kotlarmilos commented Apr 15, 2026 •

edited

Loading

kotlarmilos Apr 28, 2026 •

edited

Loading