[RFA] testsuite/gdb.base/bang.exp: Allow trailing text after "Program exited normally"

[RFA] testsuite/gdb.base/bang.exp: Allow trailing text after "Program exited normally"

[Corinna Vinschen]

Hi,

while testing I found that a test for ending the program in
gdb.base/bang.exp failed, even though the target had actually
exited normally.  The cause of the FAIL is that bang.exp doesn't
expect anything after the string "Program exited normally.".
On the sh target at least, the output after exiting contains a
bit more text:

  Program exited normally.
  [Switching to process 0] 

Since that trailing output after a normal exit doesn't change
anything with respect to the testresult, I propose to allow
trailing characters:

	* testsuite/gdb.base/bang.exp: Allow trailing text after
	"Program exited normally." string.

Index: testsuite/gdb.base/bang.exp
===================================================================
RCS file: /cvs/src/src/gdb/testsuite/gdb.base/bang.exp,v
retrieving revision 1.1
diff -u -p -r1.1 bang.exp
--- testsuite/gdb.base/bang.exp 24 Jun 2003 22:04:06 -0000      1.1
+++ testsuite/gdb.base/bang.exp 8 Sep 2003 11:29:38 -0000
@@ -36,6 +36,6 @@ gdb_load ${binfile}

 # Verify that we can run the program and that it terminates normally.
 gdb_test "run" \
-         ".*Program exited normally\." \
+         ".*Program exited normally\..*" \
          "run program"


Corinna

-- 
Corinna Vinschen
Cygwin Developer
Red Hat, Inc.

Re: [RFA] testsuite/gdb.base/bang.exp: Allow trailing text after "Program exited normally"

[Daniel Jacobowitz]

On Mon, Sep 08, 2003 at 01:42:01PM +0200, Corinna Vinschen wrote:
> Hi,
> 
> while testing I found that a test for ending the program in
> gdb.base/bang.exp failed, even though the target had actually
> exited normally.  The cause of the FAIL is that bang.exp doesn't
> expect anything after the string "Program exited normally.".
> On the sh target at least, the output after exiting contains a
> bit more text:
> 
>   Program exited normally.
>   [Switching to process 0] 
> 
> Since that trailing output after a normal exit doesn't change
> anything with respect to the testresult, I propose to allow
> trailing characters:

But isn't the [Switching to process 0] a bug?  We aren't switching
processes, the inferior is dead.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: [RFA] testsuite/gdb.base/bang.exp: Allow trailing text after "Program exited normally"

[Corinna Vinschen]

On Mon, Sep 08, 2003 at 09:27:17AM -0400, Daniel Jacobowitz wrote:
> On Mon, Sep 08, 2003 at 01:42:01PM +0200, Corinna Vinschen wrote:
> > Hi,
> > 
> > while testing I found that a test for ending the program in
> > gdb.base/bang.exp failed, even though the target had actually
> > exited normally.  The cause of the FAIL is that bang.exp doesn't
> > expect anything after the string "Program exited normally.".
> > On the sh target at least, the output after exiting contains a
> > bit more text:
> > 
> >   Program exited normally.
> >   [Switching to process 0] 
> > 
> > Since that trailing output after a normal exit doesn't change
> > anything with respect to the testresult, I propose to allow
> > trailing characters:
> 
> But isn't the [Switching to process 0] a bug?  We aren't switching
> processes, the inferior is dead.

Actually I don't understand where's the bug.  The above string is
printed if previous_inferior_ptid != inferior_ptid in normal_stop().
previous_inferior_ptid is reproducible == {pid = 42, lwp = 0, tid = 0}
while inferior_pid is set to all zero.  The pid 42 is coming from
remote-sim.c, function gdbsim_create_inferior, which always generates
a pid 42.  What's the exact problem with this?

Corinna

-- 
Corinna Vinschen
Cygwin Developer
Red Hat, Inc.

Re: [RFA] testsuite/gdb.base/bang.exp: Allow trailing text after "Program exited normally"

[Daniel Jacobowitz]

On Mon, Sep 08, 2003 at 04:29:59PM +0200, Corinna Vinschen wrote:
> On Mon, Sep 08, 2003 at 09:27:17AM -0400, Daniel Jacobowitz wrote:
> > On Mon, Sep 08, 2003 at 01:42:01PM +0200, Corinna Vinschen wrote:
> > > Hi,
> > > 
> > > while testing I found that a test for ending the program in
> > > gdb.base/bang.exp failed, even though the target had actually
> > > exited normally.  The cause of the FAIL is that bang.exp doesn't
> > > expect anything after the string "Program exited normally.".
> > > On the sh target at least, the output after exiting contains a
> > > bit more text:
> > > 
> > >   Program exited normally.
> > >   [Switching to process 0] 
> > > 
> > > Since that trailing output after a normal exit doesn't change
> > > anything with respect to the testresult, I propose to allow
> > > trailing characters:
> > 
> > But isn't the [Switching to process 0] a bug?  We aren't switching
> > processes, the inferior is dead.
> 
> Actually I don't understand where's the bug.  The above string is
> printed if previous_inferior_ptid != inferior_ptid in normal_stop().
> previous_inferior_ptid is reproducible == {pid = 42, lwp = 0, tid = 0}
> while inferior_pid is set to all zero.  The pid 42 is coming from
> remote-sim.c, function gdbsim_create_inferior, which always generates
> a pid 42.  What's the exact problem with this?

Haven't we exited at this point?  I'd think we should skip the
Switching message in that case.  If you take a look at normal_stop, it
is trying to, but target_has_execution is still true for the sim at
this point.

I was thinking of this patch.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

2003-09-08  Daniel Jacobowitz  <drow@mvista.com>

	* infrun.c (normal_stop): Don't print a message if the inferior
	has exited.

Index: infrun.c
===================================================================
RCS file: /cvs/src/src/gdb/infrun.c,v
retrieving revision 1.111
diff -u -p -r1.111 infrun.c
--- infrun.c	7 Sep 2003 15:15:01 -0000	1.111
+++ infrun.c	8 Sep 2003 14:50:59 -0000
@@ -3054,14 +3054,21 @@ print_stop_reason (enum inferior_stop_re
 void
 normal_stop (void)
 {
+  struct target_waitstatus last;
+  ptid_t last_ptid;
+
+  get_last_target_status (&last_ptid, &last);
+
   /* As with the notification of thread events, we want to delay
      notifying the user that we've switched thread context until
      the inferior actually stops.
 
      (Note that there's no point in saying anything if the inferior
-     has exited!) */
+     has exited!)  */
   if (!ptid_equal (previous_inferior_ptid, inferior_ptid)
-      && target_has_execution)
+      && target_has_execution
+      && last.kind != TARGET_WAITKIND_SIGNALLED
+      && last.kind != TARGET_WAITKIND_EXITED)
     {
       target_terminal_ours_for_output ();
       printf_filtered ("[Switching to %s]\n",

Re: [RFA] testsuite/gdb.base/bang.exp: Allow trailing text after "Program exited normally"

[Corinna Vinschen]

On Mon, Sep 08, 2003 at 10:52:13AM -0400, Daniel Jacobowitz wrote:
> I was thinking of this patch.

Looks good.  Do you check this in?

Corinna

> 2003-09-08  Daniel Jacobowitz  <drow@mvista.com>
> 
> 	* infrun.c (normal_stop): Don't print a message if the inferior
> 	has exited.
> 
> Index: infrun.c
> ===================================================================
> RCS file: /cvs/src/src/gdb/infrun.c,v
> retrieving revision 1.111
> diff -u -p -r1.111 infrun.c
> --- infrun.c	7 Sep 2003 15:15:01 -0000	1.111
> +++ infrun.c	8 Sep 2003 14:50:59 -0000
> @@ -3054,14 +3054,21 @@ print_stop_reason (enum inferior_stop_re
>  void
>  normal_stop (void)
>  {
> +  struct target_waitstatus last;
> +  ptid_t last_ptid;
> +
> +  get_last_target_status (&last_ptid, &last);
> +
>    /* As with the notification of thread events, we want to delay
>       notifying the user that we've switched thread context until
>       the inferior actually stops.
>  
>       (Note that there's no point in saying anything if the inferior
> -     has exited!) */
> +     has exited!)  */
>    if (!ptid_equal (previous_inferior_ptid, inferior_ptid)
> -      && target_has_execution)
> +      && target_has_execution
> +      && last.kind != TARGET_WAITKIND_SIGNALLED
> +      && last.kind != TARGET_WAITKIND_EXITED)
>      {
>        target_terminal_ours_for_output ();
>        printf_filtered ("[Switching to %s]\n",

-- 
Corinna Vinschen
Cygwin Developer
Red Hat, Inc.

Re: [RFA] testsuite/gdb.base/bang.exp: Allow trailing text after "Program exited normally"

[Andrew Cagney]

> On Mon, Sep 08, 2003 at 01:42:01PM +0200, Corinna Vinschen wrote:
> 
>> Hi,
>> 
>> while testing I found that a test for ending the program in
>> gdb.base/bang.exp failed, even though the target had actually
>> exited normally.  The cause of the FAIL is that bang.exp doesn't
>> expect anything after the string "Program exited normally.".
>> On the sh target at least, the output after exiting contains a
>> bit more text:
>> 
>>   Program exited normally.
>>   [Switching to process 0] 
>> 
>> Since that trailing output after a normal exit doesn't change
>> anything with respect to the testresult, I propose to allow
>> trailing characters:
> 
> 
> But isn't the [Switching to process 0] a bug?  We aren't switching
> processes, the inferior is dead.

Yes.  The message is both a bug and a very long standing regression.

Andrew

PS: Suggest a reminder comment that SIGNALED is really "terminated with 
signal" :-)

Re: [RFA] testsuite/gdb.base/bang.exp: Allow trailing text after "Program exited normally"

[Daniel Jacobowitz]

On Mon, Sep 08, 2003 at 02:52:28PM -0400, Andrew Cagney wrote:
> >On Mon, Sep 08, 2003 at 01:42:01PM +0200, Corinna Vinschen wrote:
> >
> >>Hi,
> >>
> >>while testing I found that a test for ending the program in
> >>gdb.base/bang.exp failed, even though the target had actually
> >>exited normally.  The cause of the FAIL is that bang.exp doesn't
> >>expect anything after the string "Program exited normally.".
> >>On the sh target at least, the output after exiting contains a
> >>bit more text:
> >>
> >>  Program exited normally.
> >>  [Switching to process 0] 
> >>
> >>Since that trailing output after a normal exit doesn't change
> >>anything with respect to the testresult, I propose to allow
> >>trailing characters:
> >
> >
> >But isn't the [Switching to process 0] a bug?  We aren't switching
> >processes, the inferior is dead.
> 
> Yes.  The message is both a bug and a very long standing regression.
> 
> Andrew
> 
> PS: Suggest a reminder comment that SIGNALED is really "terminated with 
> signal" :-)

Good idea.  WIFSIGNALLED confuses me every time.  Checked in to HEAD.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

2003-09-08  Daniel Jacobowitz  <drow@mvista.com>

	* infrun.c (normal_stop): Don't print a message if the inferior
	has exited.

Index: infrun.c
===================================================================
RCS file: /cvs/src/src/gdb/infrun.c,v
retrieving revision 1.111
diff -u -p -r1.111 infrun.c
--- infrun.c	7 Sep 2003 15:15:01 -0000	1.111
+++ infrun.c	8 Sep 2003 20:12:29 -0000
@@ -3054,14 +3054,22 @@ print_stop_reason (enum inferior_stop_re
 void
 normal_stop (void)
 {
+  struct target_waitstatus last;
+  ptid_t last_ptid;
+
+  get_last_target_status (&last_ptid, &last);
+
   /* As with the notification of thread events, we want to delay
      notifying the user that we've switched thread context until
      the inferior actually stops.
 
-     (Note that there's no point in saying anything if the inferior
-     has exited!) */
+     There's no point in saying anything if the inferior has exited.
+     Note that SIGNALLED here means "exited with a signal", not
+     "received a signal".  */
   if (!ptid_equal (previous_inferior_ptid, inferior_ptid)
-      && target_has_execution)
+      && target_has_execution
+      && last.kind != TARGET_WAITKIND_SIGNALLED
+      && last.kind != TARGET_WAITKIND_EXITED)
     {
       target_terminal_ours_for_output ();
       printf_filtered ("[Switching to %s]\n",