Program suddenly exit when call `si` or `n` or other continue command in repl.

[Katzeee]

Debug adapter definition and debug configuration

Installed adapter via Mason, I tried cpptools and codelldb, both have this issue.
Here is my config, should the same as wiki's.

return {
   adapters = {
     codelldb = {
       type = "server",
       port = "${port}",
       executable = {
          command = codelldb_path,
          args = { "--port", "${port}" },
        },
      },
      cppdbg = {
           id = "cppdbg",
           type = "executable",
          command = install_root_dir .. "/bin/OpenDebugAD7",
       },
  },
  configurations = {
    cpp = {
      name = "Launch File",
      type = "cppdbg",
      -- type = "codelldb",
      request = "launch",
      program = function() return vim.fn.input("Path to executable: ", vim.fn.getcwd() .. "/", "file") end
      cwd = "${workspaceFolder}",
      stopOnEntry = false,
    },
  },
}

Debug adapter version

No response

Steps to Reproduce

#include <iostream>
#include <memory>
int f() {
  static int x = 0;
  x++;
  return 1;
}
int main() {
  int x = 0;
  for (int i = 0; i < 5; i++) {
    x += 1;
  }
  f();
  auto node = std::make_unique<int>(1);
  std::cout << &node << " " << node.get() << std::endl;
}

I set the breakpoint at line 5, then dap.continue, dap.repl.open, call -exec si, the program suddenly exit.

Expected Result

I expect the program stop at this line, repl output something like

0x00005555555551c3    6    x++;

as it output when I debug it using gdb or lldb.
I find the output message and a stop event in dap.log, but it is wierd that I also got this session debug message in it.

[ DEBUG ] 2023-03-15T08:30:02Z+0800 ] .../xac/.local/share/nvim/lazy/nvi
m-dap/lua/dap/session.lua:932 ] 1       {
  body = {
    category = "stdout",
    output = "0x00005555555551c3\t6\t  x++;\n"
  },
  event = "output",
  seq = 43,
  type = "event"
}
[ DEBUG ] 2023-03-15T08:30:02Z+0800 ] .../xac/.local/share/nvim/lazy/nvi
m-dap/lua/dap/session.lua:932 ] 1       {
  body = {
    allThreadsStopped = true,
    column = 1,
    line = 6,
    reason = "step",
    source = {
      checksums = {},
      name = "x.cpp",
      path = "/home/xac/x.cpp",
      sources = {}
    },
    threadId = 198705
  },
  event = "stopped",
  seq = 46,
  type = "event"
}
[ DEBUG ] 2023-03-15T08:30:02Z+0800 ] .../xac/.local/share/nvim/lazy/nvi
m-dap/lua/dap/session.lua:648 ] "Received stopped event, but a.out [1987
05] is already stopped. Resuming newly stopped thread. To disable this s
et the `auto_continue_if_many_stopped` option to false."

seems the program exit before the stopped event? I'm new to dap so I can't figure out whether it's a bug of dap.

Actual Result

There's no need other logs. I found that I can get it work by setting auto_continue_if_many_stopped to false.

[Katzeee]

But there is another problem, after the debugger send me the result of executing si or n, the sign indicates where the program stopped doesn't change.

Expected Result

The stop sign should at line 6 after I excecute -exec n. Seems ui doesn't refresh until I call dap.step_over or something.

Actual Result

The stop sign stays at line 5 until the program exit.

[Katzeee]

1. Found that you set should_jump to false at session.lua, line 659, but why? I'm wondering at what time self.stopped_thread_id and should jump should be both true.
2. Both dap.step_over and -exec n will recieve a stop event which reason is step, but clear_running is called in session:_step, which set session.stopped_thread_id to nil. Then should_jump will not be set to false in session:event_stopped, and the sign will be redraw. Is there some difference between them?

Maybe there should be more coditional statenment in event_stopped to handle this?