Regression for gdb.trace/backtrace.exp [Re: [PATCH 0/4 v3] Add support for static and SystemTap probes]

[Sergio Durigan Junior]

On Saturday, April 28 2012, Jan Kratochvil wrote:

> On Mon, 23 Apr 2012 06:53:11 +0200, Sergio Durigan Junior wrote:
>> Regtested on Fedora x86_64, no regressions.  Comments are welcome.
>
> -PASS: gdb.trace/backtrace.exp: run trace experiment
> +ERROR: Process no longer exists
> 1786		loc->probe->pops->set_semaphore (loc->probe, loc->gdbarch);
> (gdb) p loc->probe
> $1 = (struct probe *) 0x0
>
> It happens only with gdbserver.

Thanks for catching this.  I forgot to check if there is a probe before
setting/clearing the semaphore.  This code fixes the problem.  I am
committing it under the obvious rule.

Again, thanks for catching.

-- 
Sergio

Index: gdb/ChangeLog
===================================================================
RCS file: /cvs/src/src/gdb/ChangeLog,v
retrieving revision 1.14185
diff -u -p -r1.14185 ChangeLog
--- gdb/ChangeLog	28 Apr 2012 23:22:11 -0000	1.14185
+++ gdb/ChangeLog	28 Apr 2012 23:58:31 -0000
@@ -1,3 +1,8 @@
+2012-04-28  Sergio Durigan Junior  <sergiodj@redhat.com>
+
+	* tracepoint.c (start_tracing, stop_tracing): Checking for NULL
+	probes, fixing a regression inserted by my previous patchset.
+
 2012-04-28  Doug Evans  <dje@google.com>
 
 	Initial support for Fission.  http://gcc.gnu.org/wiki/DebugFission
Index: gdb/tracepoint.c
===================================================================
RCS file: /cvs/src/src/gdb/tracepoint.c,v
retrieving revision 1.253
diff -u -p -r1.253 tracepoint.c
--- gdb/tracepoint.c	27 Apr 2012 20:47:56 -0000	1.253
+++ gdb/tracepoint.c	28 Apr 2012 23:58:31 -0000
@@ -1783,7 +1783,8 @@ start_tracing (char *notes)
       t->number_on_target = b->number;
 
       for (loc = b->loc; loc; loc = loc->next)
-	loc->probe->pops->set_semaphore (loc->probe, loc->gdbarch);
+	if (loc->probe != NULL)
+	  loc->probe->pops->set_semaphore (loc->probe, loc->gdbarch);
     }
   VEC_free (breakpoint_p, tp_vec);
 
@@ -1878,7 +1879,8 @@ stop_tracing (char *note)
 	     but we don't really care if this semaphore goes out of sync.
 	     That's why we are decrementing it here, but not taking care
 	     in other places.  */
-	  loc->probe->pops->clear_semaphore (loc->probe, loc->gdbarch);
+	  if (loc->probe != NULL)
+	    loc->probe->pops->clear_semaphore (loc->probe, loc->gdbarch);
 	}
     }