Ward C Travis writes:
Ward> Forgive me if I have overlooked something simple. I have Ward> compiled u-boot using 'config_MPC8272ADS' and, following Ward> Wolfgang's advice, modified as little as possible. I accepted Ward> all the defaults, providing no overrides on the make line.
Ward> Looking at he .srec file, it appears that the image is to be Ward> loaded at 0xfff00000 . The configuration words are based there Ward> and _start is placed beyond them, at the image + 0x0100 .
Ward> The ADS board I am working with has the standard 8MB flash Ward> SIMM, which when the board boots from flash is mapped Ward> beginning at 0xff800000.
Ward> I am unclear on how the MPC8272 reset vectors the PC to Ward> 0xfff00100. Perhaps the simple question is, should the image Ward> be loaded into flash at some other address?
The U-Boot must be loaded at the address for which it's built i.e. 0xFFF00000. It will start running from 0xFFF00100 because the CPU is "told" to do so by hard reset configuration word (HRCW), and it does not matter what the flash start address is. Chapter 5 of MPC8272 reference manual would help.
Ward> My flash is clear - it is a rented board - and so there is no Ward> previous loader or configuration words in flash; hence I must Ward> boot from EEPROM
No, you don't (and must not). This board provides three possible sources for the HRCW: BCSR (the CPLD), flash and EEPROM. Unless somebody erased first sector of the flash, there should be HRCW, but the most reliable way is to use BCSR: it can't be erased. Unfortunately, it uses high boot (0xFFF00100) and not low boot as it should be for such big flashes (more than 1MB). This is the reason the U-Boot is built by default to start at 0xFFF00000. You should not change the jumpers to boot from EEPROM because then CS0 will be mapped to EEPROM and the flash (containing the U-Boot) will be unmapped. Please note that the HRCW which you see at the beginning of the U-Boot image won't be used at all because these addresses can't respond to first four bus transactions.
Ward> and download the code (with Metrowerks, via parallel) while Ward> flash is remapped for EEPROM boot. I currently am Ward> experimenting with programming the image at various offsets Ward> from the beginning of flash. But once I reconfigure to boot Ward> from flash and power-cycle, of course, I see no sign of Ward> activity.