Please help with GDB/PSIM and Redhat Linux

[Hoang Ngo]

I made the modification in dummy.c. Recompiled rtems-4.5.0-beta3a
and rtems-packages-19991215. I still got the same error message.

I am somewhat confused on what we are trying to achieve here, so i 
will re-iterate my problems here:

1. The ncurses examples works fine for GDB/PSIM under Slackware
   but not on Redhat-6.2
2. The examples that run under Slackware doesn't display the 
   correct result.

Here is the content of dummy.c after i modified it:

#define CONFIGURE_RTEMS_INIT_TASKS_TABLE
#define CONFIGURE_MAXIMUM_TASKS 10
#define CONFIGURE_USE_IMFS_AS_BASE_FILESYSTEM
#define CONFIGURE_LIBIO_MAXIMUM_FILE_DESCRIPTORS 20
#define CONFIGURE_INIT

Hoang



-----Original Message-----
From: Joel Sherrill [mailto:joel.sherrill at OARcorp.com]
Sent: Tuesday, August 01, 2000 8:09 AM
To: Hoang Ngo
Cc: hoanghngo at home.com; rtems-users at OARcorp.com
Subject: Re: Please help with GDB/PSIM and Redhat Linux


Try this patch to use a more reasonable default configuration:

bash$ cvs diff -u c/src/libmisc/dummy
cvs server: Diffing c/src/libmisc/dummy
Index: c/src/libmisc/dummy/dummy.c
===================================================================
RCS file: /usr1/CVS/rtems/c/src/libmisc/dummy/dummy.c,v
retrieving revision 1.3
diff -u -r1.3 dummy.c
--- c/src/libmisc/dummy/dummy.c	2000/06/12 15:00:07	1.3
+++ c/src/libmisc/dummy/dummy.c	2000/08/01 15:07:51
@@ -26,7 +26,9 @@
 /* configuration information */
 
 #define CONFIGURE_RTEMS_INIT_TASKS_TABLE
-#define CONFIGURE_MAXIMUM_TASKS 1
+#define CONFIGURE_MAXIMUM_TASKS 10
+#define CONFIGURE_USE_IMFS_AS_BASE_FILESYSTEM
+#define CONFIGURE_LIBIO_MAXIMUM_FILE_DESCRIPTORS 20
 
 #define CONFIGURE_INIT
 

This is more like what is used to be around Dec 1999.
Now the defaults are lower to save memory.

The call to "boot_card" in psim/start/start.S needs
to pass argc, argv, and environp.  See the file
libbsp/m68k/shared/start.S for an example of how this
is supposed to be.  I don't see any PPC code that
is setting it up properly.  The best I see at first
glance is 0's for argc/argv -- nothing for environp.

environp must be passed in to make setenv work.

Hoang Ngo wrote:
> 
> I used the same psim-gdb script for both environments.
> The on under slackware works but not the one under redhat.
> 
> Here is the content of psim_tree.hngo
> /#address-cells 2
> /openprom/options/oea-memory-size 8388608
> 
> Here is the result of running examples/ncurses/rain.exe
> core_find_mapping() - access to unmaped address, attach a default map to
> handle this - addr=0x38392d31 nr_bytes=0x1 processor=0x401a5008 cia=0xd828
> 
> The address 0xd828 is in the function <set_curterm>.
> 
> Followed your suggestion, i tried to uncommented some of the devices
> in the psim-gdb script but the simulator doesn;t know about the device
> sem or shm.
> 
> Hoang
> 
> -----Original Message----
> From: Joel Sherrill [mailto:joel.sherrill at OARcorp.com]
> Sent: Sunday, July 30, 2000 7:04 AM
> To: hoanghngo at home.com
> Cc: rtems-users at OARcorp.com
> Subject: Re: Please help with GDB/PSIM and Redhat Linux
> 
> Hoang Ngo wrote:
> >
> > I ran into a perflexing problem. At home i use Slackware7
> > and rtems-4.5.0-beta3a, and rtems-packages-19991215
> > compiled for GDB/PSIM. All the applications under ncurses
> > directory ran fine (except that the terminal is not displaying
> > what it supposed to). At work i use redhat-6.2 with all the same
> > rtems packages but unable to get the ncurses applications to run.
> > The error message indicates that the applications is accessing
> > an unmapped region of memory (core_find_mapping() - access
> > to unmaped address, attach a default map to handle this -
> > addr=0x38392d31 nr_bytes=0x1 processor=0x401a5008 cia=0xf1dc).
> >
> > If you have any suggestions as what to look for please drop me a line.
> 
> I think this is a matter of either one of two things:
> 
>  + You are not starting the simulator with all the device options.
>    See the script psim/psim-gdb included with the BSP in the tools
>    subdirectory.
>  + I am not sure what instruction/inside what function the cia
>    is pointing to but it might be an issue that the shared memory
>    or semaphore is not being initialized correctly.
> 
> More than likely, it is the first since the shared mem/semapahore
> is mapped to different processor address range. See what the
> psim script does to initialize the simulator.
> 
> > TIA,
> > Hoang
> 
> --
> Joel Sherrill, Ph.D.             Director of Research & Development
> joel at OARcorp.com                 On-Line Applications Research
> Ask me about RTEMS: a free RTOS  Huntsville AL 35805
>    Support Available             (256) 722-9985

-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel at OARcorp.com                 On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
   Support Available             (256) 722-9985