[lldb] Limit the number of warnings emitted for unsupported DWARF form values

[da-viper]

warnings_spamming.mp4

This is mostly unsupported dwarf DW_FORM values.

[llvmbot]

@llvm/issue-subscribers-lldb-dap

Author: Ebuka Ezike (da-viper)

https://github.com/user-attachments/assets/1bd31fb7-9151-413e-ba99-7177f646508b

This is mostly unsupported dwarf DW_FORM values.

[llvmbot]

@llvm/issue-subscribers-lldb

Author: Ebuka Ezike (da-viper)

https://github.com/user-attachments/assets/1bd31fb7-9151-413e-ba99-7177f646508b

This is mostly unsupported dwarf DW_FORM values.

[JDevlieghere]

This is not an lldb-dap specific issue. The underlying problem is that we shouldn't be using the asynchronous diagnostics for unsupported from values or at least limit how often this is emitted.

[da-viper]

There are two ways I can think of to handle this issue.

• Only show errors as popups and warnings in the debug console.
• hash the call once flag for modules based on the actual payload leaving out the module information.
  This will only show the warning for the first module that reports it. (relevant code path).
  This may also have the effect of warnings that you want to show more than once.
  

  llvm-project/lldb/source/Core/Module.cpp

  Lines 1116 to 1123 in f7fdc8d

  	void Module::ReportWarning(const llvm::formatv_object_base &payload) {
  	StreamString strm;
  	GetDescription(strm.AsRawOstream(), lldb::eDescriptionLevelFull);
  	std::string msg = payload.str();
  	strm << ' ' << msg;
  	Debugger::ReportWarning(strm.GetString().str(), {},
  	GetDiagnosticOnceFlag(msg));
  	}