[patch remote.c] Re-throw connect fail exception

[Andrew Cagney <ac131313 at cygnus dot com>]

Hello,

The attached tweaks remote.c to so that failed connectsion are 
re-propogated back through the caller.

This patch has been lurking in the wings for sometime only no one could 
think of a test-case demonstrating that it fixed a bug.  Today I got 
lucky, compare:


(gdb)
-target-select remote /dev/null
&"get_tty_state failed: Operation not supported\n"
&"set_tty_state failed: Operation not supported\n"
&"get_tty_state failed: Operation not supported\n"
&"set_tty_state failed: Operation not supported\n"
~"Ignoring packet error, continuing...\n"
~"Ignoring packet error, continuing...\n"
~"Ignoring packet error, continuing...\n"
&"Couldn't establish connection to remote target\n"
&"Malformed response to offset query, timeout\n"
^connected
(gdb)


with:


(gdb)
-target-select remote /dev/null
&"get_tty_state failed: Operation not supported\n"
&"set_tty_state failed: Operation not supported\n"
&"get_tty_state failed: Operation not supported\n"
&"set_tty_state failed: Operation not supported\n"
~"Ignoring packet error, continuing...\n"
~"Ignoring packet error, continuing...\n"
~"Ignoring packet error, continuing...\n"
&"Couldn't establish connection to remote target\n"
&"Malformed response to offset query, timeout\n"
^error,msg="Malformed response to offset query, timeout"
(gdb)


The fix is in.

So, how legitimate is it do do something like:

(gdb) target remote /dev/null

as part of a test case?

Andrew
2002-05-19  Andrew Cagney  <ac131313@redhat.com>
 
	From Fernando Nasser:
	* remote.c (remote_async_open_1): Re-throw the exception when the
	connection fails.
	(remote_cisco_open): Ditto.
	(remote_open_1): Ditto.

Index: remote.c
===================================================================
RCS file: /cvs/src/src/gdb/remote.c,v
retrieving revision 1.83
diff -u -r1.83 remote.c
--- remote.c	19 May 2002 18:32:10 -0000	1.83
+++ remote.c	19 May 2002 20:00:34 -0000
@@ -2314,7 +2314,11 @@
 #endif
 
   /* Start the remote connection.  If error() or QUIT, discard this
-     target (we'd otherwise be in an inconsistent state).
+     target (we'd otherwise be in an inconsistent state) and then
+     propogate the error on up the exception chain.  This ensures that
+     the caller doesn't stumble along blindly assuming that the
+     function succeeded.  The CLI doesn't have this problem but other
+     UI's, such as MI do.
 
      FIXME: cagney/2002-05-19: Instead of re-throwing the exception,
      this function should return an error indication letting the
@@ -2332,7 +2336,7 @@
   if (ex < 0)
     {
       pop_target ();
-      return;
+      throw_exception (ex);
     }
 
   if (extended_p)
@@ -2437,7 +2441,8 @@
 #endif
 
   /* Start the remote connection; if error, discard this target.  See
-     the comments in remote_open_1() for further details.  */
+     the comments in remote_open_1() for further details such as the
+     need to re-throw the exception.  */
   ex = catch_exceptions (uiout,
 			 remote_start_remote, NULL,
 			 "Couldn't establish connection to remote"
@@ -2447,7 +2452,7 @@
     {
       pop_target ();
       wait_forever_enabled_p = 1;
-      return;
+      throw_exception (ex);
     }
 
   wait_forever_enabled_p = 1;
@@ -5555,7 +5560,8 @@
   inferior_ptid = pid_to_ptid (MAGIC_NULL_PID);
 
   /* Start the remote connection; if error, discard this target.  See
-     the comments in remote_open_1() for further details.  */
+     the comments in remote_open_1() for further details such as the
+     need to re-throw the exception.  */
   ex = catch_exceptions (uiout,
 			 remote_start_remote_dummy, NULL,
 			 "Couldn't establish connection to remote"
@@ -5564,7 +5570,7 @@
   if (ex < 0)
     {
       pop_target ();
-      return;
+      throw_exception (ex);
     }
 }