[remote.c] kill kill_kludge

[remote.c] kill kill_kludge

[Pedro Alves]

[-- Attachment #1: Type: text/plain, Size: 1931 bytes --]

Hi,

kill_kludge in remote.c is no longer needed.  handle_inferior_event is
no longer calling target_kill on a TARGET_WAITKIND_SIGNALLED.

   case TARGET_WAITKIND_SIGNALLED:
      if (debug_infrun)
        fprintf_unfiltered (gdb_stdlog, "infrun: 
TARGET_WAITKIND_SIGNALLED\n");
      stop_print_frame = 0;
      stop_signal = ecs->ws.value.sig;
      target_terminal_ours ();	/* Must do this before mourn anyway */

      /* Note: By definition of TARGET_WAITKIND_SIGNALLED, we shouldn't
         reach here unless the inferior is dead.  However, for years
         target_kill() was called here, which hints that fatal signals aren't
         really fatal on some systems.  If that's true, then some changes
         may be needed. */
      target_mourn_inferior ();

      print_stop_reason (SIGNAL_EXITED, stop_signal);
      singlestep_breakpoints_inserted_p = 0;
      stop_stepping (ecs);
      return;

Because of that, the kill_kludge introduces a bug in this sequence:

target remote :9999
signal 9               (inferior exits signaled, kill_kludge is step,
                        but target_kill isn't called)
(restart stub)
target remote :9999    (kill_kludge isn't cleared)
kill                   (kill_kludge is set, so 'k' is not sent to the stub...

static void
remote_kill (void)
{
  /* For some mysterious reason, wait_for_inferior calls kill instead of
     mourn after it gets TARGET_WAITKIND_SIGNALLED.  Work around it.  */
  if (kill_kludge)
    {
      kill_kludge = 0;
      target_mourn_inferior ();
      return;
    }

  /* Use catch_errors so the user can quit from gdb even when we aren't on
     speaking terms with the remote system.  */
  catch_errors ((catch_errors_ftype *) putpkt, "k", "", RETURN_MASK_ERROR);

  /* Don't wait for it to die.  I'm not really sure it matters whether
     we do or not.  For the existing stubs, kill is a noop.  */
  target_mourn_inferior ();
}

-- 
Pedro Alves

[-- Attachment #2: kill_kill_kludge.diff --]
[-- Type: text/x-diff, Size: 2371 bytes --]

2008-05-17  Pedro Alves  <pedro@codesourcery.com>

	* remote.c (kill_kludge): Delete.
	(remote_wait, remote_async_wait): Don't set it.
	(remote_kill, remote_async_kill): Don't do anything with it.

---
 gdb/remote.c |   24 ------------------------
 1 file changed, 24 deletions(-)

Index: src/gdb/remote.c
===================================================================
--- src.orig/gdb/remote.c	2008-05-17 12:40:39.000000000 +0100
+++ src/gdb/remote.c	2008-05-17 12:41:36.000000000 +0100
@@ -3320,10 +3320,6 @@ remote_async_terminal_ours (void)
   remote_async_terminal_ours_p = 1;
 }
 
-/* If nonzero, ignore the next kill.  */
-
-int kill_kludge;
-
 void
 remote_console_output (char *msg)
 {
@@ -3527,7 +3523,6 @@ Packet: '%s'\n"),
 	  status->kind = TARGET_WAITKIND_SIGNALLED;
 	  status->value.sig = (enum target_signal)
 	    (((fromhex (buf[1])) << 4) + (fromhex (buf[2])));
-	  kill_kludge = 1;
 
 	  goto got_status;
 	case 'O':		/* Console output.  */
@@ -3754,7 +3749,6 @@ Packet: '%s'\n"),
 	  status->kind = TARGET_WAITKIND_SIGNALLED;
 	  status->value.sig = (enum target_signal)
 	    (((fromhex (buf[1])) << 4) + (fromhex (buf[2])));
-	  kill_kludge = 1;
 
 	  goto got_status;
 	case 'O':		/* Console output.  */
@@ -5272,15 +5266,6 @@ getpkt_sane (char **buf, long *sizeof_bu
 static void
 remote_kill (void)
 {
-  /* For some mysterious reason, wait_for_inferior calls kill instead of
-     mourn after it gets TARGET_WAITKIND_SIGNALLED.  Work around it.  */
-  if (kill_kludge)
-    {
-      kill_kludge = 0;
-      target_mourn_inferior ();
-      return;
-    }
-
   /* Use catch_errors so the user can quit from gdb even when we aren't on
      speaking terms with the remote system.  */
   catch_errors ((catch_errors_ftype *) putpkt, "k", "", RETURN_MASK_ERROR);
@@ -5298,15 +5283,6 @@ remote_async_kill (void)
   if (target_is_async_p ())
     serial_async (remote_desc, NULL, 0);
 
-  /* For some mysterious reason, wait_for_inferior calls kill instead of
-     mourn after it gets TARGET_WAITKIND_SIGNALLED.  Work around it.  */
-  if (kill_kludge)
-    {
-      kill_kludge = 0;
-      target_mourn_inferior ();
-      return;
-    }
-
   /* Use catch_errors so the user can quit from gdb even when we
      aren't on speaking terms with the remote system.  */
   catch_errors ((catch_errors_ftype *) putpkt, "k", "", RETURN_MASK_ERROR);

Re: [remote.c] kill kill_kludge

[Daniel Jacobowitz]

On Sat, May 17, 2008 at 12:56:53PM +0100, Pedro Alves wrote:
> 2008-05-17  Pedro Alves  <pedro@codesourcery.com>
> 
> 	* remote.c (kill_kludge): Delete.
> 	(remote_wait, remote_async_wait): Don't set it.
> 	(remote_kill, remote_async_kill): Don't do anything with it.

Gross.  Thanks, this is OK.

-- 
Daniel Jacobowitz
CodeSourcery

Re: [remote.c] kill kill_kludge

[Pedro Alves]

A Thursday 05 June 2008 20:18:29, Daniel Jacobowitz wrote:
> On Sat, May 17, 2008 at 12:56:53PM +0100, Pedro Alves wrote:
> > 2008-05-17  Pedro Alves  <pedro@codesourcery.com>
> >
> > 	* remote.c (kill_kludge): Delete.
> > 	(remote_wait, remote_async_wait): Don't set it.
> > 	(remote_kill, remote_async_kill): Don't do anything with it.
>
> Gross.  Thanks, this is OK.

Murdered.

-- 
Pedro Alves