[patch] typo in gdb/testsuite/config/gdbserver.exp

[patch] typo in gdb/testsuite/config/gdbserver.exp

[Felix Lee]

when doing host-x-host testing to a remote gdbserver,
gdbserver.exp will send an executable to the target host (using
rsh or whatever), and then it tries to run it with gdbserver on
the remote target, but it uses the name of the local executable,
not the remote executable, so that fails.

gdb/testsuite/ChangeLog:

2004-09-30  Felix Lee  <felix+log1@specifixinc.com>

        * config/gdbserver.exp (gdb_load): Use right filename on remote.

Index: gdb/testsuite/config/gdbserver.exp
diff -u -p -r1.1.1.2 gdbserver.exp
--- gdb/testsuite/config/gdbserver.exp  8 Jul 2004 13:06:50 -0000       1.1.1.2
+++ gdb/testsuite/config/gdbserver.exp  1 Oct 2004 07:33:56 -0000
@@ -207,7 +207,7 @@ proc gdb_load { arg } {
        }
     }
 
-    set res [gdbserver_gdb_load $host_exec]
+    set res [gdbserver_gdb_load $server_exec]
     set protocol [lindex $res 0]
     set gdbport [lindex $res 1]
 

Re: [patch] typo in gdb/testsuite/config/gdbserver.exp

[Daniel Jacobowitz]

On Fri, Oct 01, 2004 at 12:36:59AM -0700, Felix Lee wrote:
> when doing host-x-host testing to a remote gdbserver,
> gdbserver.exp will send an executable to the target host (using
> rsh or whatever), and then it tries to run it with gdbserver on
> the remote target, but it uses the name of the local executable,
> not the remote executable, so that fails.
> 
> gdb/testsuite/ChangeLog:
> 
> 2004-09-30  Felix Lee  <felix+log1@specifixinc.com>
> 
>         * config/gdbserver.exp (gdb_load): Use right filename on remote.

I seem to remember this breaking my local gdbserver setup, i.e. running
gdbserver on localhost.  It's entirely possible that that only happened
because I've goofed my board files, though.  I've learned a painful
amount about writing DejaGNU board files in the last couple of months,
so I'll try again...

> 
> Index: gdb/testsuite/config/gdbserver.exp
> diff -u -p -r1.1.1.2 gdbserver.exp
> --- gdb/testsuite/config/gdbserver.exp  8 Jul 2004 13:06:50 -0000       1.1.1.2
> +++ gdb/testsuite/config/gdbserver.exp  1 Oct 2004 07:33:56 -0000
> @@ -207,7 +207,7 @@ proc gdb_load { arg } {
>         }
>      }
>  
> -    set res [gdbserver_gdb_load $host_exec]
> +    set res [gdbserver_gdb_load $server_exec]
>      set protocol [lindex $res 0]
>      set gdbport [lindex $res 1]
>  
> 

-- 
Daniel Jacobowitz

Re: [patch] typo in gdb/testsuite/config/gdbserver.exp

[Felix Lee]

Daniel Jacobowitz <drow@false.org>:
> I seem to remember this breaking my local gdbserver setup, i.e. running
> gdbserver on localhost.  It's entirely possible that that only happened
> because I've goofed my board files, though.  I've learned a painful
> amount about writing DejaGNU board files in the last couple of months,
> so I'll try again...

I just looked at that, it took me a few tries to get it to work.
this board file works for me:

    load_generic_config "gdbserver"

    unset_board_info isremote
    set_board_info isremote 0

    set_board_info compiler "[find_gcc]"
    set_board_info gdb,do_reload_on_run 1
    set_board_info gdb,no_hardware_watchpoints 1
    set_board_info noargs 1
    set_board_info sockethost "localhost:"
    set_board_info use_gdb_stub 1
--

Re: [patch] typo in gdb/testsuite/config/gdbserver.exp

[Daniel Jacobowitz]

On Mon, Oct 04, 2004 at 01:40:05AM -0700, Felix Lee wrote:
> Daniel Jacobowitz <drow@false.org>:
> > I seem to remember this breaking my local gdbserver setup, i.e. running
> > gdbserver on localhost.  It's entirely possible that that only happened
> > because I've goofed my board files, though.  I've learned a painful
> > amount about writing DejaGNU board files in the last couple of months,
> > so I'll try again...
> 
> I just looked at that, it took me a few tries to get it to work.
> this board file works for me:
> 
>     load_generic_config "gdbserver"
> 
>     unset_board_info isremote
>     set_board_info isremote 0
> 
>     set_board_info compiler "[find_gcc]"
>     set_board_info gdb,do_reload_on_run 1
>     set_board_info gdb,no_hardware_watchpoints 1
>     set_board_info noargs 1
>     set_board_info sockethost "localhost:"
>     set_board_info use_gdb_stub 1

But not very well.  Here's a cleaned up version of mine...

In any case, your patch to config/gdbserver.exp appears correct to me,
and doesn't break my testing.  Please check it in.

# gdbserver running native.

load_generic_config "gdbserver"

set_board_info compiler "[find_gcc]"
set_board_info gdb,do_reload_on_run 1
set_board_info gdb,no_hardware_watchpoints 1
set_board_info noargs 1
set_board_info sockethost "localhost:"
set_board_info use_gdb_stub 1

process_multilib_options ""

set_board_info gdb_protocol "remote"
set_board_info gdb_server_prog "../gdbserver/gdbserver"

set_board_info gdb,noinferiorio 1

proc ${board}_spawn { board cmd } {
  global board_info

  set baseboard [lindex [split $board "/"] 0]

  set board_info($baseboard,isremote) 0
  set result [remote_spawn $board $cmd]
  set board_info($baseboard,isremote) 1

  return $result
}

proc ${board}_download { board host dest } {
  return $host
}


-- 
Daniel Jacobowitz