[ECOS] problem with running eCos kernel image on Redboot

["Chandrashekhar Padiyar" <chandrashekhar dot padiyar at xaltedinc dot com>]

Hello

We have ported the eCos Rattler package for our board, which is based on the
MPC8260 CPU. We have managed to initialize the serial and Ethernet
interfaces and have got the Redboot prompt on the serial port with ROMRAM
startup.

Our board uses SMC2 for the UART and FCC2 for Ethernet. We have 32MB of
Flash from 0xFE000000 to 0xFFFFFFFF (as two 16-bit 16MB chips in parallel).

After this, we build the eCos kernel image (ecos.srec) from 'libtarget.a'
and 'target.ld' (with 'powerpc-eabi-ld' and 'powerpc-eabi-objcopy') and
download it into RAM at 0x00040000 using the 'load' command (TFTP). Now,
when we try to write it to Flash at 0xFFF00000 using 'fis create', we get an
error somewhat like this: 'Can't program Flash at 0xFFF00000 from
0x00040000...'.

When we try to execute the downloaded code from 0x00040000 using 'go', it
executes some instructions and then, when it gets to an 'eieio' instruction,
it simply branches to some 0x000000800-odd location, at which there is
garbage (although there should be exception vectors there [ROMRAM]), and
stops executing. We don't know how these memory locations show garbage. Then
when we check locations 0x40000 (our code), it too shows garbage!!!

However, when we restart the board (hard reset) and check 0x40000 and 0x800,
we see the correct code and exception vectors. How does it then become
garbage in the middle of execution??? We are using VisionClick software and
VisionICE emulator for emulation and single-stepping.

Regards

Chandrashekhar Padiyar


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