[ECOS] vector table

[ECOS] vector table

[Christoph Csebits]

hi,

i am working on a eCos port for a custom PPC860 board,
with a simple "load&go" bootstrapper with no
exception handler installed.

i have a problem setting up the vector tables 

nm gives me:

RE: [ECOS] vector table

[Gary Thomas]

On 03-May-2001 Christoph Csebits wrote:
> hi,
> 
> i am working on a eCos port for a custom PPC860 board,
> with a simple "load&go" bootstrapper with no
> exception handler installed.
> 
> i have a problem setting up the vector tables 
> 
> nm gives me:

What?

For this mode of operation, make sure that you set
  CYGSEM_HAL_POWERPC_COPY_VECTORS
Also, make sure that you turn off
  CYGSEM_HAL_USE_ROM_MONITOR

Re: [ECOS] vector table

[Christoph Csebits]

On Thu, May 03, 2001 at 07:30:02AM -0600, Gary Thomas wrote:
> 
> On 03-May-2001 Christoph Csebits wrote:
> > i have a problem setting up the vector tables 
> > 
> > nm gives me:
> 
> What?
nm - list symbols from object files.

0000000000010900 ? __exception_decrementer

this line shows me the the exception handler for
the decrementer interrupt is located at offset 0x10900
but it should located at 0x00900 (MSR[IP]=0).

In other words:
A decrementer exception is vectored to 0x00900 but
the belonging handler waits at 0x10900.

or am i completely wrong?

> For this mode of operation, make sure that you set
>   CYGSEM_HAL_POWERPC_COPY_VECTORS
turned on

> Also, make sure that you turn off
>   CYGSEM_HAL_USE_ROM_MONITOR
turned off

Re: [ECOS] vector table

[Gary Thomas]

On 03-May-2001 Christoph Csebits wrote:
> On Thu, May 03, 2001 at 07:30:02AM -0600, Gary Thomas wrote:
>> 
>> On 03-May-2001 Christoph Csebits wrote:
>> > i have a problem setting up the vector tables 
>> > 
>> > nm gives me:
>> 
>> What?
> nm - list symbols from object files.
> 
> 0000000000010900 ? __exception_decrementer
> 
> this line shows me the the exception handler for
> the decrementer interrupt is located at offset 0x10900
> but it should located at 0x00900 (MSR[IP]=0).
> 
> In other words:
> A decrementer exception is vectored to 0x00900 but
> the belonging handler waits at 0x10900.
> 
> or am i completely wrong?
> 

Actually, the contents of the exception tables at 0x00000000
get created from the loaded values at 0x10000 (if you tell
eCos to copy the vectors).  Thus, the code for the decrementer
at 0x10900 will get moved to 0x900.

>> For this mode of operation, make sure that you set
>>   CYGSEM_HAL_POWERPC_COPY_VECTORS
> turned on
> 
>> Also, make sure that you turn off
>>   CYGSEM_HAL_USE_ROM_MONITOR
> turned off

OK, so what happens when you run it?  How far does it get?

One thing you could try is to include GDB stubs in your application
and then connect to it via GDB.  Just load & start it and you should
get a GDB prompt $T<<STUFF>> on the debug serial port at 38400/8N1.

Re: [ECOS] vector table

[Christoph Csebits]

On Thu, May 03, 2001 at 08:49:46AM -0600, Gary Thomas wrote:
> Actually, the contents of the exception tables at 0x00000000
> get created from the loaded values at 0x10000 (if you tell
> eCos to copy the vectors).  Thus, the code for the decrementer
> at 0x10900 will get moved to 0x900.

i see, thanks for explanation

> OK, so what happens when you run it?  How far does it get?

not really far. turning leds on at the main entry point
(_start in vector.S) do not work.

> One thing you could try is to include GDB stubs in your application
> and then connect to it via GDB.  Just load & start it and you should
> get a GDB prompt $T<<STUFF>> on the debug serial port at 38400/8N1.

serial port doesnt working yet, and i rather fix that problem first

thanks for your patience,
Christoph