Date: Sun, 23 Oct 2011 15:04:15 -0400
From: Christopher Faylor <cgf-use-the-mailinglist-please AT cygwin DOT com>
To: cygwin AT cygwin DOT com
Subject: Re: gdb problem
Message-ID: <20111023190414.GB13536@ednor.casa.cgf.cx>
Reply-To: cygwin AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com
References: <4EA329A3 DOT 30800 AT cornell DOT edu>
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="ZGiS0Q5IWpPtfppv"
Content-Disposition: inline
In-Reply-To: <4EA329A3.30800@cornell.edu>
User-Agent: Mutt/1.5.20 (2009-06-14)
Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm
Precedence: bulk
Sender: cygwin-owner AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com

--ZGiS0Q5IWpPtfppv
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline

On Sat, Oct 22, 2011 at 04:37:55PM -0400, Ken Brown wrote:
>The attached testcase illustrates a problem with `gdb -i=mi'.  I've 
>tested both gdb 7.3.50-1 and 7.3.50-2, with cygwin 1.7.9 as well as with 
>several recent snapshots (including 2011-10-22).
>
>Under some circumstances, if gdb -i=mi is started and given several 
>input lines at once, it only prints part of the output before stopping. 
>  I've been able to reproduce this once in a while while working 
>interactively (by copying and pasting the whole bunch of input lines); 
>in this case one can press Return to get the rest of the output.  But 
>the problem happens consistently with the attached test case, which runs 
>gdb in a subprocess.  One has to kill the gdb process before the main 
>program exits.
>
>The STC runs as expected on Linux.

Thanks for the STC.  I had to tweak it to actually see how it was supposed
to work on Linux since only a limited number of lines from the pty were
being output.  I've included the revised test case below.

I made a simple change to Cygwin which will probably cause subtle
problems somewhere down the line.  At least for now it seems to make gdb
operate as expected.

I'm building a new snapshot now.

cgf

--ZGiS0Q5IWpPtfppv
Content-Type: text/x-c; charset=us-ascii
Content-Disposition: attachment; filename="gdbstc.cc"

#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <pty.h>
#include <string.h>
#include <sys/wait.h>

void get_output (int fd);

int
main (int argc, const char **argv) 
{
  int master;
  pid_t pid;

  if ((pid = forkpty (&master, NULL, NULL, NULL)) < 0)
    {
      perror ("forkpty");
      exit (1);
    }
  /* child */
  if (pid == 0) 
    {
      const char *av[100];
      // putenv ("HOME=/tmp");
      int i = 0;
#ifdef STRACE_GDB
      av[i++] = "strace";
      av[i++] = "-o";
      av[i++] = "/tmp/strace.out";
#ifdef __CYGWIN__
      av[i++] = "--mask=all+paranoid";
#endif
#endif
      av[i++] = argv[1] ?: "gdb";
      fprintf (stderr, "*** using %s\n", av[0]);
      av[i++] = "-i=mi";
      av[i] = NULL;
      execvp (av[0], (char * const *) av);
      /* shouldn't get here */
      exit (1);
    }
  /* parent */
  const char *input[20];

  int i = 0;
  input[i++] = "1-inferior-tty-set /dev/pty3\n";
  input[i++] = "2-gdb-set height 0\n";
  input[i++] = "3-gdb-set non-stop 1\n";
  input[i++] = "4-file-list-exec-source-files\n";
  input[i++] = "5-file-list-exec-source-file\n";
  input[i++] = "6-gdb-show prompt\n";
  input[i++] = "7-stack-info-frame\n";
  input[i++] = "8-thread-info\n";
  input[i++] = "9-break-list\n";
  input[i++] = "q\n";
  input[i] = NULL;

  for (int i = 0; input[i]; ++i)
    {
      write (master, input[i], strlen (input[i]));
      // sleep (1);
    }
  get_output (master);
  wait (NULL);
}

void
get_output (int fd)
{
  char buf[4096];

  while (1)
    {
      int nread = read (fd, buf, sizeof (buf));
      if (nread > 0)
	write (STDOUT_FILENO, buf, nread);
      else
	{
	  printf ("No more output.  nread %d\n", nread);
	  break;
	}
    }
}


--ZGiS0Q5IWpPtfppv
Content-Type: text/plain; charset=us-ascii

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
--ZGiS0Q5IWpPtfppv--