[BUG] Stop caused by cpptools’ own -exec-interrupt (to insert a breakpoint while running) shows ‘Exception has occurred. Signal 0’ and doesn’t resume.

Ouverte
#14,869 1 commentaire 0 réactions 0 personnes assignées Voir sur GitHub

Les mainteneurs répondent en général sous 1 jour

Personne n'a encore pris cette issue.

Évaluation

Difficulté
4/5
Temps estimé
3-5 jours
Accessibilité débutants
45/100
Type d'issue
Bug
Clarté
Plutôt claire
Activité
Active
Stack technique
typescript
Domaine
devtools

Piste de recherche

Start by tracing how cppdbg handles GDB MI stop events with reason="signal-received" and how signal-name="0" is surfaced as an exception. The issue provides reproduction steps and debugger logs, but names no source files or tests; first reproduce with a 64-bit GDB debugging a 32-bit WOW64 process while inserting a breakpoint. Done means dynamic breakpoint insertion no longer shows a blocking false error for this stop.

Rédigé par le modèle d'indexation à partir du texte de l'issue.

Description

debugger
Environment

Environment:

  • OS and Version: Windows 10 x64
  • VS Code Version: 1.146.0
  • C/C++ Extension Version: v1.32.2
  • Debugger: GDB (both x86_64-w64-mingw32-gdb 18.1 and gdb-multiarch 18.1 from MSYS2)
  • Target: 32-bit (i386) executable running under WOW64.
Bug Summary and Steps to Reproduce

When debugging a 32-bit (WOW64) process with a 64-bit GDB, the debugger halts and displays an "Exception has occurred. Signal 0" message. This happens when VS Code's cppdbg extension sends an -exec-interrupt command to pause the running process, typically to insert a new breakpoint.

The stop is reported by GDB with reason="signal-received" and signal-name="0" (an unknown signal), which the extension then surfaces as a blocking error dialog. The call stack is irrelevant to the cause; the program counter is simply wherever the thread happened to be (e.g., in a system or library function like SDL_LogCritical). This prevents adding breakpoints during an active debug session, as the user must manually click "Continue" to dismiss the false alarm every time.

To Reproduce:

  1. On Windows x64, configure launch.json to use a 64-bit GDB (e.g., C:/msys64/ucrt64/bin/gdb-multiarch.exe) to debug a 32-bit (i386) executable.
  2. Start the debug session and let the program run.
  3. While the program is running, add a new breakpoint in a source file via the editor gutter.
  4. VS Code sends -exec-interrupt to pause the process. GDB responds with *stopped,reason="signal-received",signal-name="0".
  5. The extension displays the "Exception has occurred. Signal 0" error.

Expected behavior:

The -exec-interrupt command should result in a normal, silent stop (e.g., reason="signal-received",signal-name="SIGINT"). The extension should not surface this as an exception or error to the user. The debugger should be ready for the subsequent -break-insert command without a blocking dialog.

Actual behavior:

The extension interprets GDB's signal-name="0" as a fatal error and surfaces an "Exception has occurred" dialog, halting the user's workflow.

Debugger Logs (engineLogging: true):

The log clearly shows the sequence: VS Code sends -exec-interrupt in response to a setBreakpoints request, and GDB replies with the spurious signal.

<-- C (setBreakpoints-12): {"command":"setBreakpoints","arguments":{"source":{"name":"Model.cpp",...},"lines":[174],...},"type":"request","seq":12}
1: (6400) <-1027-exec-interrupt
1: (6410) ->~"\nThread 1 stopped.\n"
1: (6415) ->*stopped,reason="signal-received",signal-name="0",signal-meaning="Signal 0",frame={addr="0x6c7be23f",func="SDL_LogCritical",from="...SDL2.dll",arch="i386"},thread-id="1",stopped-threads="all"

Additional context:

  • The issue is not related to a specific C++ exception (a catch throw catchpoint has times="0").
  • The issue is not related to SDL's thread-naming exception (0x406D1388), as setting SDL_WINDOWS_DISABLE_THREAD_NAMING=1 does not change the symptom.
  • The root cause is a known GDB bug where its Windows WOW64 support fails to correctly associate its own internal DebugBreakProcess call with the resulting STATUS_BREAKPOINT exception, misreporting it as Signal 0.
  • This behavior is reproducible with both gdb and gdb-multiarch from MSYS2.
  • A workaround is to set all breakpoints before starting the debug session, which avoids the -exec-interrupt. However, this is not a fix for dynamic breakpoint insertion.
  • Relevant launch.json configuration used:
    {
      "type": "cppdbg",
      "request": "launch",
      "MIMode": "gdb",
      "miDebuggerPath": "C:/msys64/ucrt64/bin/gdb-multiarch.exe",
      "logging": { "engineLogging": true, "trace": true },
      "setupCommands": [ ... ]
    }
    
  • An ignoreFailures or a "continue on signal 0" option in the extension could serve as a potential workaround for users unable to change their GDB version.
Debugger Configurations
see above
Debugger Logs
see above
Other Extensions

No response

Additional Information

No response

Langage dominant
TypeScript
Étoiles
6.2k
Forks
1.7k
Merge moyen
1 j 17 h
PR mergées (30 j)
12

Préparer son environnement

Par où commencer

  1. Lisez l'issue en entier, puis le guide de contribution du projet.
  2. Signalez en commentaire que vous la prenez — cela évite que deux personnes fassent le même travail.
  3. Forkez le dépôt et travaillez sur une branche.
  4. Ouvrez une pull request qui référence le numéro de l'issue.

Autres issues de microsoft/vscode-cpptools

Toutes les issues de microsoft/vscode-cpptools

Issues similaires

Plus d'issues TypeScript

Recevez les nouvelles issues par e-mail

Un résumé court des issues GitHub adaptées aux débutants.