[Bug gdb/16188] New: Verify PTRACE_TRACEME succeeded?

[Bug gdb/16188] New: Verify PTRACE_TRACEME succeeded?

[dje at google dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16188

            Bug ID: 16188
           Summary: Verify PTRACE_TRACEME succeeded?
           Product: gdb
           Version: HEAD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: gdb
          Assignee: unassigned at sourceware dot org
          Reporter: dje at google dot com

It would help when debugging certain kinds of problems if one could know
whether the PTRACE_TRACEME call succeeded.

Maybe this could pass a pass/fail code to its caller and the caller could at
least notify the user?

static void
inf_ptrace_me (void)
{
  /* "Trace me, Dr. Memory!"  */
  ptrace (PT_TRACE_ME, 0, (PTRACE_TYPE_ARG3)0, 0);
}

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/16188] Verify PTRACE_TRACEME succeeded?

[palves at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16188

Pedro Alves <palves at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |palves at redhat dot com

--- Comment #1 from Pedro Alves <palves at redhat dot com> ---
Yeah.  In fork_inferior, we handle a similar case, when the fork child fails to
exec:

      /* "Trace me, Dr. Memory!"  */
      (*traceme_fun) ();
...
      environ = env;

      if (exec_fun != NULL)
        (*exec_fun) (argv[0], argv, env);
      else
        execvp (argv[0], argv);

      /* If we get here, it's an error.  */
      save_errno = errno;
      fprintf_unfiltered (gdb_stderr, "Cannot exec %s", exec_file);
      for (i = 1; argv[i] != NULL; i++)
    fprintf_unfiltered (gdb_stderr, " %s", argv[i]);
      fprintf_unfiltered (gdb_stderr, ".\n");
      fprintf_unfiltered (gdb_stderr, "Error: %s\n",
              safe_strerror (save_errno));
      gdb_flush (gdb_stderr);
      _exit (0177);
    }


So it seems like what's missing is making traceme_fun return an error
indication, and then change

      (*traceme_fun) ();

to check for failure and _exit with error.  This will make startup_inferior see
the child exit (before the number of expected traps), and error out with
"During startup program exited with code ...".

-- 
You are receiving this mail because:
You are on the CC list for the bug.