mips and libgloss/mips/tx39 simulator questions

[Joel Sherrill]

"Frank Ch. Eigler" wrote:
> 
> Joel Sherrill <joel.sherrill@OARcorp.com> writes:
> 
> : [...]
> : I am trying to put together a simple RTEMS BSP based
> : on the tx39 code in libgloss.  [...]
> : target sim --board=jmr3904pal \
> :   --memory-region 0xffff8000,0x900 \
> :   --memory-region 0xffffe000,0x4 \
> :   --memory-region 0xb2100000,0x4 \
> :   --memory-region=0x8800000,0x1000000
> : [...]
> : /usr1/rtems/build/build-mips-rtems/mips-rtems/jmr3904/tests/hello.exe
> : mips-core: 8 byte write to unmapped address 0x80009ff8 at 0x0
> 
> Well, what memory region is supposed to sit at 0x80009ff8?  It's hard
> to help without some idea of what your BSP code assumes about the
> target board; what the code was trying to do at the time (--trace).

I think that's my problem.  I am trying to be blind about the "board."
I simply want to use the tx39 simulator in gdb as a test target for
RTEMS.  I did not really care what kind of mips3 it was.  I just liked
the fact it had real peripherals.  I used the support code in libgloss
as "reference material" making as few changes as possible to cast
it into the form of an RTEMS BSP.  Now I have a linked RTEMS hello.exe
that seems to run on the simulator until it tries to do IO.  Then it
is in a loop waiting toa status bit in the SIO controller to change but
I believe I did not tell gdb the right stuff to enable it.

What is the proper magic to run that code?

What I am trying to match is (I think) equivalent to what is called
tx39-sim in dejagnu.

> : Also any comments on using what cpu models the simulator supports
> : would be welcome.
> 
> See the "*cpu:" tags in the src/mips/*.igen files, the configure.in
> target filtering lines, and perhaps some TARGET-run command-line
> option.

Thanks.  That I can do. :)

> - FChE

--joel