Hi all,

With the help of the wisdom of blue_z, I have managed to load AT91Bootstrap and U-boot into the SRAM and SDRAM of the custom 9G20 board I'm working on. I've also been able to load a second copy of U-boot into SDRAM, and then copy that to NAND Flash. At present, I can load AT91Bootstrap via JTAG, and it will pull U-boot out of NAND Flash and execute it. This is great progress.

Here's where I'm stuck: AT91Bootstrap is in SRAM, which means I need to reload it after every reset. Strangely, the method I used to copy U-boot to Flash does not appear to work for AT91Bootstrap.

Here's what worked for U-boot.

GDB session


U-boot session over DBGU serial connection


When I try the same trick with AT91Bootstrap, the board does not boot from Flash.




I have considered the possibility that maybe the copying to Flash is working, but that the booting is not. However, I don't think this is the problem-- if I read out the first few bytes of Flash into SDRAM using U-boot's nand read function, I don't get see the interrupt vector table that I can see at the start of the AT91Bootstrap binary. Here's the evidence:



You can see that the first few bytes have changed, which suggests that the nand read command is working. But those bytes don't match the starting bytes of AT91Bootstrap.

Any ideas as to what I'm doing wrong? Should I expect the nand write command to work the way I'm using it? Perhaps it's an issue of loading an ELF file rather than a .bin?

Thanks,
Brandon

Yes, you must program the binary variant into NAND, check before writing to flash whether the image has the vectors at the beginning, including the proper size entry.

Since you are initially going to boot from NAND you should be aware of the following errata:

44.3.5.1 RSTC: Reset During SDRAM Accesses
When a User Reset occurs during SDRAM read access, the SDRAM clock is turned off while
data are ready to be read on the data bus. The SDRAM maintains the data until the clock
restarts.
If the user Reset is programmed to assert a general reset, the data maintained by the SDRAM
leads to a data bus conflict and adversely affects the boot memories connected on the EBI:
• NAND Flash boot functionality, if the system boots out of internal ROM.
• NOR Flash boot, if the system boots on an external memory connected on the EBI CS0.

I think that errata that seems to be shared by (all?) AT91SAM9 devices makes them unreliable to boot from any EBI connected storage.

Reinhard

is it resolved? thanks.

Most of you failing here are not correctly formatting the image to pass the RomBOOT sanity checks.

Reinhard basically states the issue here: