Fix debugger hangs due to runtime deadlock by using patch DJI by Copilot · Pull Request #123651 · dotnet/runtime

Copilot · 2026-01-27T00:24:27Z

Description

This PR implements a targeted fix for the debugger deadlock issue where the debugger hangs while waiting for a deadlocked debuggee process. The deadlock occurs when DebuggerController::BindPatch calls GetJitInfo while holding the DebuggerController lock. GetJitInfo can trigger HashMap lookups that perform GC mode transitions, leading to lock ordering issues.

The fix modifies BindPatch to check if the patch already has a DebuggerJitInfo (DJI) stored on it via patch->HasDJI(). When the DJI is available, it uses patch->GetDJI() instead of calling the problematic GetJitInfo method. This approach is cleaner than adding a new parameter since the patch already stores the DJI.

Changes Made

Updated BindPatch implementation to check patch->HasDJI() before calling GetJitInfo
When DJI exists on the patch, use patch->GetDJI() to avoid the deadlock-prone GetJitInfo call
Maintains backward compatibility - patches without DJI continue to call GetJitInfo as before

The implementation uses a ternary operator for concise, idiomatic C++ code:

DebuggerJitInfo *info = patch->HasDJI() ? patch->GetDJI() : g_pDebugger->GetJitInfo(pMD, (const BYTE *)startAddr);

Customer Impact

Without this fix, customers can experience debugger hangs when debugging applications with ReadyToRun methods under specific timing conditions. This occurs when:

A breakpoint is set on a ReadyToRun method that hasn't executed yet
One thread is stepping through code while another thread first executes the ReadyToRun method with the breakpoint

The hang requires the debugger and IDE to be force-closed, resulting in loss of debugging session state and developer productivity.

Regression

This is not a regression from the most recent release. The root cause has existed for a while, but timing changes in the code may have impacted the likelihood of hitting the race condition.

Testing

Full CoreCLR build completed successfully with no errors or warnings
Code review passed with no issues identified
No security vulnerabilities introduced

The changes are minimal (1 file changed, 3 insertions, 1 deletion) and surgical to reduce regression risk for potential backporting to prior .NET releases.

Risk

Low Risk. The changes are minimal and highly targeted:

No API changes - the BindPatch signature remains unchanged
The logic flow remains unchanged - we're just avoiding an unnecessary lookup when the DJI is already available on the patch
The patch already stores the DJI via AddPatchForMethodDef, so we're using existing data
Full build validation confirms no compilation issues
The fix addresses the specific deadlock scenario without affecting other debugger functionality
Maintains backward compatibility - patches without DJI still call GetJitInfo as before

Original prompt

This section details on the original issue you should resolve

<issue_title>Debugger hangs due to runtime deadlock. HashMap takes ThreadStore out of order</issue_title>
<issue_description>### Description

The debugger is blocked in mscordbi code at:

      [Inline Frame] mscordbi.dll!SafeWaitForSingleObject(CordbProcess *) Line 257  C++
>     mscordbi.dll!CordbProcess::StopInternal(unsigned long dwTimeout, VMPTR_Base<AppDomain,void> pAppDomainToken) Line 3611  C++
      [Inline Frame] mscordbi.dll!StopContinueHolder::Init(CordbProcess *) Line 208 C++
      mscordbi.dll!CordbModule::GetFunctionFromToken(unsigned int token, ICorDebugFunction * * ppFunction) Line 1387    C++

mscordbi is waiting on the debuggee to respond which it won't do because it is deadlocked.

Thread 0x9314 in the debuggee (trying to acquire DebuggerController lock):

      [Inline Frame] coreclr.dll!CrstBase::AcquireLock(CrstBase *) Line 174   C++
      [Inline Frame] coreclr.dll!FunctionBase<CrstBase *,&CrstBase::AcquireLock,&CrstBase::ReleaseLock>::DoAcquire() Line 694 C++
      [Inline Frame] coreclr.dll!BaseHolder<CrstBase *,FunctionBase<CrstBase *,&CrstBase::AcquireLock,&CrstBase::ReleaseLock>,0,&CompareDefault<CrstBase *>>::Acquire() Line 266    C++
      [Inline Frame] coreclr.dll!BaseHolder<CrstBase *,FunctionBase<CrstBase *,&CrstBase::AcquireLock,&CrstBase::ReleaseLock>,0,&CompareDefault<CrstBase *>>::{ctor}(CrstBase *) Line 233 C++
      [Inline Frame] coreclr.dll!Holder<CrstBase *,&CrstBase::AcquireLock,&CrstBase::ReleaseLock,0,&CompareDefault<CrstBase *>,1>::{ctor}(CrstBase *) Line 729    C++
>     coreclr.dll!DebuggerController::DispatchPatchOrSingleStep(Thread * thread, _CONTEXT * context, const unsigned char * address, SCAN_TRIGGER which, DebuggerSteppingInfo * pDebuggerSteppingInfo) Line 3063   C++
      coreclr.dll!DebuggerController::DispatchNativeException(_EXCEPTION_RECORD * pException, _CONTEXT * pContext, unsigned long dwCode, Thread * pCurThread, DebuggerSteppingInfo * pDebuggerSteppingInfo) Line 4618   C++
      coreclr.dll!Debugger::FirstChanceNativeException(_EXCEPTION_RECORD * exception, _CONTEXT * context, unsigned long code, Thread * thread, int fIsVEH) Line 5470    C++
      coreclr.dll!IsDebuggerFault(_EXCEPTION_RECORD * pExceptionRecord, _CONTEXT * pContext, unsigned long exceptionCode, Thread * pThread) Line 5811 C++

And the DebuggerController lock is held by thread 0x1E38 which is here:

[Inline Frame] coreclr.dll!CLREventBase::Wait(unsigned long) Line 412 C++
coreclr.dll!Thread::WaitSuspendEventsHelper() Line 4458 C++
coreclr.dll!Thread::RareDisablePreemptiveGC() Line 2185 C++
[Inline Frame] coreclr.dll!Thread::DisablePreemptiveGC() Line 1288 C++
[Inline Frame] coreclr.dll!GCHolderBase::EnterInternalCoop_HackNoThread(bool) Line 4619 C++
[Inline Frame] coreclr.dll!GCCoopHackNoThread::{ctor}(bool) Line 4834 C++
[Inline Frame] coreclr.dll!HashMap::LookupValue(unsigned __int64) Line 552 C++
[Inline Frame] coreclr.dll!PtrHashMap::LookupValue(unsigned __int64 key, void *) Line 607 C++
[Inline Frame] coreclr.dll!ReadyToRunInfo::GetMethodDescForEntryPointInNativeImage(unsigned __int64) Line 376 C++
[Inline Frame] coreclr.dll!ReadyToRunInfo::GetMethodDescForEntryPoint(unsigned __int64 entryPoint) Line 104 C++
coreclr.dll!ReadyToRunJitManager::JitCodeToMethodInfo(RangeSection * pRangeSection, unsigned __int64 currentPC, MethodDesc * * ppMethodDesc, EECodeInfo * pCodeInfo) Line 6451 C++
coreclr.dll!EECodeInfo::Init(unsigned __int64 codeAddress, ExecutionManager::ScanFlag scanFlag) Line 15025 C++
[Inline Frame] coreclr.dll!EECodeInfo::{ctor}(unsigned __int64) Line 2877 C++
[Inline Frame] coreclr.dll!ExecutionManager::GetCodeStartAddress(unsigned __int64) Line 4980 C++
coreclr.dll!EEDbgInterfaceImpl::GetNativeCodeStartAddress(unsigned __int64 address) Line 416 C++
coreclr.dll!Debugger::GetJitInfoWorker(MethodDesc * fd, const unsigned char * pbAddr, DebuggerMethodInfo * * pMethInfo) Line 2711 C++
[Inline Frame] coreclr.dll!Debugger::GetJitInfo(MethodDesc *) Line 2650 C++
coreclr.dll!DebuggerController::BindPatch(DebuggerControllerPatch * patch, MethodDesc * pMD, const unsigned char *) Line 1432 C++

coreclr.dll!DebuggerController::AddBindAndActivatePatchForMethodDesc(MethodDesc * fd, DebuggerJitInfo * dji, unsigned __int64 nativeOffset, DebuggerPatchKind kind, FramePointer fp, AppDomain * pAppDomain) Line 2242 C++
coreclr.dll!DebuggerController::AddBindAndActivateILReplicaPatch(DebuggerControllerPatch * primary, DebuggerJitInfo * dji) Line 2006 C++
coreclr.dll!Debugger::MapPatchToDJI(DebuggerControllerPatch * dcp, DebuggerJitInfo * djiTo) Line 4963 C++
coreclr.dll!Debugger::MapAndBindFunctionPatches(DebuggerJitInfo * djiNew, MethodDesc * fd, const unsigned char *) Line 4879 C++
coreclr.dll!Debugger::JIT...

Fixes Debugger hangs due to runtime deadlock. HashMap takes ThreadStore out of order #123650

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.

noahfalk · 2026-01-27T08:31:14Z

@copilot - it looks like you had an error earlier and no files have changed. Can you please try again?

Modified BindPatch to accept an optional DebuggerJitInfo parameter. When DJI is provided by the caller, use it instead of calling GetJitInfo, which can trigger a deadlock through HashMap lookups with GC mode transitions. Updated two call sites that already have the DJI available: - AddBindAndActivatePatchForMethodDesc - MapPatchToDJI This is a targeted fix to avoid the deadlock described in issue where DebuggerController lock is held while calling GetJitInfo, which can try to acquire locks in the wrong order. Co-authored-by: noahfalk <6243776+noahfalk@users.noreply.github.com>

Copilot · 2026-01-27T08:53:54Z

@copilot - it looks like you had an error earlier and no files have changed. Can you please try again?

I've implemented the targeted fix. The changes pass DJI to BindPatch to avoid the deadlock-causing GetJitInfo call. Build completed successfully with no errors. Commit: 2c6edbe

AaronRobinsonMSFT · 2026-02-04T00:11:47Z

@copilot Instead of passing in a new DebuggerJitInfo argument, you should use the one that is on the patch itself. It can retrieved by calling GetDJI() on the DebuggerControllerPatch. You can check if the DebuggerJitInfo field is valid by calling HasDJI().

Per code review feedback, simplified the fix to use the DJI already stored on the patch rather than passing it as a new parameter. This achieves the same goal of avoiding the deadlock-causing GetJitInfo call while being cleaner and more aligned with existing code patterns. Co-authored-by: AaronRobinsonMSFT <30635565+AaronRobinsonMSFT@users.noreply.github.com>

Per code review feedback, use a ternary operator instead of if-else block for more concise and idiomatic C++ code. Co-authored-by: AaronRobinsonMSFT <30635565+AaronRobinsonMSFT@users.noreply.github.com>

Copilot · 2026-02-04T00:29:06Z

@copilot Instead of passing in a new DebuggerJitInfo argument, you should use the one that is on the patch itself. It can retrieved by calling GetDJI() on the DebuggerControllerPatch. You can ch...

Updated to use patch->GetDJI() instead of passing DJI as a parameter. The implementation now checks patch->HasDJI() and uses patch->GetDJI() when available. Commits: 581a7bb, 6a6a4d6

noahfalk

LGTM

Copilot

Pull request overview

This PR fixes a debugger deadlock issue where the runtime hangs when DebuggerController::BindPatch calls GetJitInfo while holding the DebuggerController lock. The GetJitInfo method can trigger HashMap lookups that perform GC mode transitions, leading to lock ordering violations and deadlocks.

Changes:

Modified BindPatch to check if the patch already has a DebuggerJitInfo (DJI) stored via patch->HasDJI() and use patch->GetDJI() instead of calling the deadlock-prone GetJitInfo method
Added explanatory comments documenting the deadlock issue and why the optimization is necessary
Maintained backward compatibility by falling back to GetJitInfo when no DJI is available on the patch

max-charlamb

Tested locally and looks good to me

steveisok · 2026-02-05T00:59:33Z

/ba-g Known issues #123796 #122874

steveisok · 2026-02-05T00:59:59Z

/backport to release/10.0

github-actions · 2026-02-05T01:00:08Z

Started backporting to release/10.0 (link to workflow run)

@steveisok

…tch DJI (#124022) Backport of #123651 to release/10.0 /cc @steveisok @Copilot Backport of #123651 to release/10.0 ## Customer Impact - [ ] Customer reported - [X] Found internally Without this fix, customers can experience debugger hangs when debugging applications with ReadyToRun methods under specific timing conditions. This occurs when: 1. A breakpoint is set on a ReadyToRun method that hasn't executed yet 2. One thread is stepping through code while another thread first executes the ReadyToRun method with the breakpoint The hang requires the debugger and IDE to be force-closed, resulting in loss of debugging session state and developer productivity. ## Regression - [ ] Yes - [X] No The root cause has existed for a while, but timing changes in the code may have impacted the likelihood of hitting the race condition. ## Testing Manually verified. ## Risk Low The changes are minimal and highly targeted: - No API changes - the BindPatch signature remains unchanged - The logic flow remains unchanged - we're just avoiding an unnecessary lookup when the DJI is already available on the patch - The fix addresses the specific deadlock scenario without affecting other debugger functionality **IMPORTANT**: If this backport is for a servicing release, please verify that: - For .NET 8 and .NET 9: The PR target branch is `release/X.0-staging`, not `release/X.0`. - For .NET 10+: The PR target branch is `release/X.0` (no `-staging` suffix). ## Package authoring no longer needed in .NET 9 **IMPORTANT**: Starting with .NET 9, you no longer need to edit a NuGet package's csproj to enable building and bump the version. Keep in mind that we still need package authoring in .NET 8 and older versions. --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: noahfalk <6243776+noahfalk@users.noreply.github.com> Co-authored-by: AaronRobinsonMSFT <30635565+AaronRobinsonMSFT@users.noreply.github.com>

Initial plan

a238f6c

Copilot AI assigned Copilot and noahfalk Jan 27, 2026

Copilot started work on behalf of noahfalk January 27, 2026 00:25 View session

Copilot AI requested a review from noahfalk January 27, 2026 01:24

Copilot stopped work on behalf of noahfalk due to an error January 27, 2026 01:24
Copilot had to stop work due to a timeout.

Copilot started work on behalf of noahfalk January 27, 2026 08:31 View session

Copilot AI changed the title ~~[WIP] Fix debugger hangs due to runtime deadlock~~ Fix debugger hangs due to runtime deadlock by passing DJI to BindPatch Jan 27, 2026

Copilot finished work on behalf of noahfalk January 27, 2026 08:55

Copilot started work on behalf of AaronRobinsonMSFT February 4, 2026 00:12 View session

Copilot AI and others added 2 commits February 4, 2026 00:26

Simplify to use ternary operator for cleaner code

6a6a4d6

Per code review feedback, use a ternary operator instead of if-else block for more concise and idiomatic C++ code. Co-authored-by: AaronRobinsonMSFT <30635565+AaronRobinsonMSFT@users.noreply.github.com>

Copilot AI changed the title ~~Fix debugger hangs due to runtime deadlock by passing DJI to BindPatch~~ Fix debugger hangs due to runtime deadlock by using patch DJI Feb 4, 2026

Copilot AI requested a review from AaronRobinsonMSFT February 4, 2026 00:30

Copilot finished work on behalf of AaronRobinsonMSFT February 4, 2026 00:30

This was referenced Feb 4, 2026

iOS.Device test WorkItemExecutions #122874

Open

XHarness package install failure on iOS due to devicectl NSPOSIXErrorDomain error 49 #123796

Open

noahfalk approved these changes Feb 4, 2026

View reviewed changes

steveisok marked this pull request as ready for review February 4, 2026 13:35

Copilot AI review requested due to automatic review settings February 4, 2026 13:35

Copilot started reviewing on behalf of steveisok February 4, 2026 13:36 View session

Copilot AI reviewed Feb 4, 2026

View reviewed changes

max-charlamb approved these changes Feb 5, 2026

View reviewed changes

steveisok enabled auto-merge (squash) February 5, 2026 00:58

steveisok merged commit 2e76447 into main Feb 5, 2026
104 of 107 checks passed

steveisok deleted the copilot/fix-debugger-deadlock-issue branch February 5, 2026 01:00

github-actions bot mentioned this pull request Feb 5, 2026

[release/10.0] Fix debugger hangs due to runtime deadlock by using patch DJI #124022

Merged

4 tasks

dotnet-maestro bot mentioned this pull request Feb 5, 2026

[main] Source code updates from dotnet/runtime dotnet/dotnet#4680

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix debugger hangs due to runtime deadlock by using patch DJI#123651

Fix debugger hangs due to runtime deadlock by using patch DJI#123651
steveisok merged 4 commits intomainfrom
copilot/fix-debugger-deadlock-issue

Copilot AI commented Jan 27, 2026 •

edited

Loading

Uh oh!

noahfalk commented Jan 27, 2026

Uh oh!

Copilot AI commented Jan 27, 2026

Uh oh!

AaronRobinsonMSFT commented Feb 4, 2026

Uh oh!

Copilot AI commented Feb 4, 2026

Uh oh!

noahfalk left a comment

Uh oh!

Copilot AI left a comment

Uh oh!

max-charlamb left a comment

Uh oh!

steveisok commented Feb 5, 2026

Uh oh!

Uh oh!

steveisok commented Feb 5, 2026

Uh oh!

github-actions bot commented Feb 5, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Conversation

Copilot AI commented Jan 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Changes Made

Customer Impact

Regression

Testing

Risk

Uh oh!

noahfalk commented Jan 27, 2026

Uh oh!

Copilot AI commented Jan 27, 2026

Uh oh!

AaronRobinsonMSFT commented Feb 4, 2026

Uh oh!

Copilot AI commented Feb 4, 2026

Uh oh!

noahfalk left a comment

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

max-charlamb left a comment

Choose a reason for hiding this comment

Uh oh!

steveisok commented Feb 5, 2026

Uh oh!

Uh oh!

steveisok commented Feb 5, 2026

Uh oh!

github-actions bot commented Feb 5, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Copilot AI commented Jan 27, 2026 •

edited

Loading