Re: [ECOS] AT91 EB40(A) board serial port issues?

[Tim Drury <tdrury at siliconmotorsports dot com>]

Scott - thanks.  I setup redboot to print out "U" intead of "redboot>"
so I'd get a nice 0x55 (01010101) pattern out of the serial port and
measured the baud rate to be about 80k.  After looking through ecos
and reading the EB40A docs I found out that the AT40 uses a 
32.768MHz clock and the EB40A uses a 66MHz clock.  Changing
CYGNUM_HAL_ARM_AT91_CLOCK_SPEED to 66000000 in
hal_arm_at91.cdl fixed it and know I can run redboot out of RAM.
Woohoo!

On to the flash....

-tim

To summarize what is needed to get redboot running in RAM on the
EB40A:

I'm running gcc-3.0.4, binutils-2.10.1, insight/gdb-5.1.1.

fix CYGNUM_HAL_ARM_AT91_CLOCK_SPEED as above and
build redboot.

Upon running arm-elf-gdb -nw redboot.elf I issue the commands:
tar rdi s=/dev/ttyS0
set $ps=0xd3
set *0xffe00004=0x02002525  <- I added this to active onboard SRAM
lo
c
then hit ctrl-z to suspend gdb (redboot continues to run on eb40a)

run 'minicom -o' with baud rate at 38400, N81 and hit return to see
the 'redboot>' prompt.  run 've' command to see version info.



On Monday 17 June 2002 11:16 am, Scott Dattalo wrote:
> On Mon, 17 Jun 2002, Tim Drury wrote:
> > Okay, I inserted 'das blinkenlights' throughout the ecos hal and redboot
> > and it all looks good.  The code enter's redboot's while(true) loop where
> > the prompt ("redboot>") is printed to the serial port.  I forced it to
> > print to the serial port on each loop iteration so I could try to see
> > something in minicom.  I run it from gdb, ctrl-z out of gdb and the
> > serial-out led is still steadily blinking assuring me that redboot is
> > still pumping out prompts.
> >
> > I run 'minicom -o' but all I see is garbage.  I used 8N1 with no flow
> > control and tried every baud rate from 9600 to 115200.  Could the clock
> > freq of this the AT91R80004 (used in the EB40A) be such that the usart
> > freq is 'just off' from what the pc can read?  I find this very unlikely.
> >  Can anyone offer other suggestions to try and fix this?
>
> Did you measure the bit timing on the RS232 lines with an oscilloscope?
>
> Scott


-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss