The MCM2+8 chip has been replaced by its successor the MCM4+16 chip. To move an existing code-tree built for the MCM2+8 to the MCM4+16 a few minor changes are required. This document describes the required changes needed for designs where no additional memory is mounted.
The MCM4+16 has 16 Mbytes of RAM internally compared to the 8 Mbytes of RAM on the MCM2+8. Change the following parameters in the configuration file os/linux/.config to adapt to the changes:
CONFIG_ETRAX_DRAM_SIZE: 8 to 16
CONFIG_ETRAX_DEF_R_SDRAM_CONFIG: 00601515 to 09603636
To fully utilize the 4 Mbytes of Flash (2 Mbytes of flash on MCM2+8) changes to the ptablespec (The Partition Table Specification file) are necessary. This file is generated by make in some of the later releases of the developer board's/device server's SDK. This depends on which release that was used to build the firmware for the MCM2+8. Check the contents of the ptablespec file, in the case it exists, and if says "Generated file" or does not exist the changes should be made to the top-level Makefile.
In the case where it is generated by make, parameters with prefix PSIZE_ needs to be modified, otherwise modifications to the ptablespec file (in the top-level directory) will directly be reflected in the partition table of the image (fimage). Make sure that the hexadecimal sum of all partitions equals less or equal to the total size of available flash memory, i.e. 4 Mbytes (0x400000).
Example: Change top-level Makfile
PSIZE_rescue = 0x010000
PSIZE_kernel = 0x350000
PSIZE_jffs2_0 = 0x0a0000
Example: Change ptablespec file
rescue 0x010000 ro rescue rescue.img
flash1 0x350000 rw kernel flash1.img
flash2 0x0a0000 rw jffs2 flash2.img