Anyone using an IBM RISCWatch ICE?

Anyone using an IBM RISCWatch ICE?

[Wright, David]

Hi,

I've been attempting to use an IBM RISCWatch ICE on a Walnut system
and I'm having problems.  (Spare me any lectures about how I should
be using a BDI.  I'm just trying to find out how usable the IBM is,
since it's already here.)

The specific problem I'm trying to solve is this:  how do I set a
breakpoint early in kernel startup, say at start_kernel (just as
the MMU is enabled)?  I've tried booting the Walnut so that it
comes up into its ROM monitor, then stopping it with the ICE, and
setting a hardware breakpoint at the start_kernel virtual address.
Problem is, when I start the CPU running again, the Walnut is no
longer responding to the keyboard (serial line).  Can't do anything
except reset it at that point.  Using the ICE reveals that the CPU
is still running, but I have no real idea what it's doing.  It sure
isn't responding to the serial port.

As an alternative, I tried setting a hardware breakpoint at the
physical address of "start_here".  That doesn't hang, but the
breakpoint is never hit, either.

Any ideas gratefully accepted.  I've tried this kind of thing with
a BDI and had no problems.

  -- David Wright, InfiniSwitch Corp.


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: Anyone using an IBM RISCWatch ICE?

[Frank Rowand]

"Wright, David" wrote:
>
> Hi,
>
> I've been attempting to use an IBM RISCWatch ICE on a Walnut system
> and I'm having problems.  (Spare me any lectures about how I should
> be using a BDI.  I'm just trying to find out how usable the IBM is,
> since it's already here.)
>
> The specific problem I'm trying to solve is this:  how do I set a
> breakpoint early in kernel startup, say at start_kernel (just as
> the MMU is enabled)?  I've tried booting the Walnut so that it
> comes up into its ROM monitor, then stopping it with the ICE, and
> setting a hardware breakpoint at the start_kernel virtual address.

The disk in the PC that hosts my RISCWatch crashed yesterday so I'm
doing this from memory.

I've done that sort of thing quite often.  OpenBios is sitting
at the menu prompt,

  RW> stop
  RW> bp set ihw 0xc015a644
  RW> run

(0xc015a644 is start_kernel)

Then tell OpenBios to boot the kernel.

> Problem is, when I start the CPU running again, the Walnut is no
> longer responding to the keyboard (serial line).  Can't do anything
> except reset it at that point.  Using the ICE reveals that the CPU
> is still running, but I have no real idea what it's doing.  It sure
> isn't responding to the serial port.

If you stop it with RISCWatch, what is the program counter?


>
> As an alternative, I tried setting a hardware breakpoint at the
> physical address of "start_here".  That doesn't hang, but the
> breakpoint is never hit, either.
>
> Any ideas gratefully accepted.  I've tried this kind of thing with
> a BDI and had no problems.
>
>   -- David Wright, InfiniSwitch Corp.
>

--
Frank Rowand <frank_rowand@mvista.com>
MontaVista Software, Inc

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: Anyone using an IBM RISCWatch ICE?

[David Gibson]

On Wed, Aug 15, 2001 at 11:12:17AM -0400, Wright, David wrote:
>
> Hi,
>
> I've been attempting to use an IBM RISCWatch ICE on a Walnut system
> and I'm having problems.  (Spare me any lectures about how I should
> be using a BDI.  I'm just trying to find out how usable the IBM is,
> since it's already here.)
>
> The specific problem I'm trying to solve is this:  how do I set a
> breakpoint early in kernel startup, say at start_kernel (just as
> the MMU is enabled)?  I've tried booting the Walnut so that it
> comes up into its ROM monitor, then stopping it with the ICE, and
> setting a hardware breakpoint at the start_kernel virtual address.
> Problem is, when I start the CPU running again, the Walnut is no
> longer responding to the keyboard (serial line).  Can't do anything
> except reset it at that point.  Using the ICE reveals that the CPU
> is still running, but I have no real idea what it's doing.  It sure
> isn't responding to the serial port.

I've encountered the same problem on a Walnut here, but not reliably -
I generally find that it works for a few times, then I try another
breakpoint in just the same way and the board stops responding to
serial.  I've found that resetting the board and the processor probe
and restarting the RISCWatch software tends to help, in the sense that
it will then work again for a reasonable while.

> As an alternative, I tried setting a hardware breakpoint at the
> physical address of "start_here".  That doesn't hang, but the
> breakpoint is never hit, either.
>
> Any ideas gratefully accepted.  I've tried this kind of thing with
> a BDI and had no problems.

--
David Gibson			| Microsoft: Making the easy things hard
david@gibson.dropbear.id.au	| and the hard things buggy
http://www.ozlabs.org/people/dgibson


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: Anyone using an IBM RISCWatch ICE?

[John F Davis]

Hello

I am not a riscwatch expert, but its the only thing I've ever used.  Maybe
I can help.


I don't know how you could use the virtual address.  But, here is a simple
thing you can do.  Insert a tight loop in the code where you want to break,
like:

jump:
     if (foo) {
          goto jump;
     }

Then run your code and stop it. When things are to your liking, clear
foo and restart it.

Also, when you stop the code, examine the IAR.  Maybe the physical address
you were using wasn't the same one
the IAR shows.

Lastly, when you are stopped, be careful about examining memory.  By
default the memory window will try to examine memory
at location 0.  On the 405 system I use, this causes a page fault and thus
oops my kernel when I restart it.  I don't if this is happening to
you when you reach your breakpoint or not.  I fixed this problem my using a
TLB entry to map memory address zero to a non used
location.  That way I can use the memory window.  If you know how to bring
up the memory window at an address other than
zero I would like to know.

I am sure that when Chip gets off vacation, he can help you more.

Good luck,

John



"Wright, David" <dwright@infiniswitch.com>@lists.linuxppc.org on 08/15/2001
11:12:17 AM

Sent by:  owner-linuxppc-embedded@lists.linuxppc.org


To:   <linuxppc-embedded@lists.linuxppc.org>
cc:
Subject:  Anyone using an IBM RISCWatch ICE?




Hi,

I've been attempting to use an IBM RISCWatch ICE on a Walnut system
and I'm having problems.  (Spare me any lectures about how I should
be using a BDI.  I'm just trying to find out how usable the IBM is,
since it's already here.)

The specific problem I'm trying to solve is this:  how do I set a
breakpoint early in kernel startup, say at start_kernel (just as
the MMU is enabled)?  I've tried booting the Walnut so that it
comes up into its ROM monitor, then stopping it with the ICE, and
setting a hardware breakpoint at the start_kernel virtual address.
Problem is, when I start the CPU running again, the Walnut is no
longer responding to the keyboard (serial line).  Can't do anything
except reset it at that point.  Using the ICE reveals that the CPU
is still running, but I have no real idea what it's doing.  It sure
isn't responding to the serial port.

As an alternative, I tried setting a hardware breakpoint at the
physical address of "start_here".  That doesn't hang, but the
breakpoint is never hit, either.

Any ideas gratefully accepted.  I've tried this kind of thing with
a BDI and had no problems.

  -- David Wright, InfiniSwitch Corp.


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

RE: Anyone using an IBM RISCWatch ICE?

[Wright, David]

Sigh.  As is all too often the case, it was an example of user error
(mine).  I thought I was setting a hardware breakpoint, but I
misunderstood the RISCWatch interface and was actually trying to set
a software breakpoint.  Once I followed Frank's directions, things
worked great.  (I figured I must be doing SOMETHING wrong; this one
was too severe not to have been seen already if it were a real bug.)

Thanks,

  -- David Wright


> > Hi,
> >
> > I've been attempting to use an IBM RISCWatch ICE on a Walnut system
> > and I'm having problems.  (Spare me any lectures about how I should
> > be using a BDI.  I'm just trying to find out how usable the IBM is,
> > since it's already here.)
> >
> > The specific problem I'm trying to solve is this:  how do I set a
> > breakpoint early in kernel startup, say at start_kernel (just as
> > the MMU is enabled)?  I've tried booting the Walnut so that it
> > comes up into its ROM monitor, then stopping it with the ICE, and
> > setting a hardware breakpoint at the start_kernel virtual address.
>
> The disk in the PC that hosts my RISCWatch crashed yesterday so I'm
> doing this from memory.
>
> I've done that sort of thing quite often.  OpenBios is sitting
> at the menu prompt,
>
>   RW> stop
>   RW> bp set ihw 0xc015a644
>   RW> run
>
> (0xc015a644 is start_kernel)
>
> Then tell OpenBios to boot the kernel.

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/