On Mon, 2006-05-22 at 22:29, Dubb, Benjamin wrote: > JP, I forgot to mention that we had to replace the original AT25F1024 > part on the EDB9307 with a bigger F2048 device (256 Kbytes). So I tested > > it on the board with that bigger flash chip originally. I just tried > loading my u-boot.bin onto an unmodified edb9307 board, and the download > > utility reports a failure at the last moment. It's possible we are > seeing > the same problem. That is why I modified flasher.S in the first time, just to output the offset and block where the problem occured. Offset of the error was always the same for a given binary. And text file would just be fine. I used a board we have to reread the EEPROM and there was an error at the same offset as detected by flasher.S The error was on the MSB of a byte and my guess was that somehow, there was one clock cycle too many at the end of some transfer. So I implemented SPI flash page write ie 256 byte at once. This time, the error is on a page boundary. I am now working on other things I have to do with this board, since cirrus is supposed to release a new download utility soon, so wait and see :) JP Francois