[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: MCM boot-up problems - possible filesystem or flash issues?
>Further investigation showed that while the client
>bootloader was loading the flash image file fimage (which is 0x200018 bytes
>long) to the RAM starting at address 0xc0004000, the memory was wrapping
>around after 0x200000 bytes and the last 28 bytes were overwriting the start
>of the buffer!!
Good, a classic memory mirroring
>I think that "the problem of not being able to start the 'init' process" is
>possibly linked with this situation. If the kernel expects a separate memory
>at the 0xc0204000 address but this actually wraps around and overwrites the
>memory at 0xc0004000 when accessed, it is normal for the machine to hang
>when accessing this memory.
Yes, I agree.
SDRAM memories are divided in 4 banks. In the MCM each bank would hence
be 2 MB. Which bank to use is selected with the BA pins. If these
pins are not connected correctly you will get exactly the kind of
mirroring that you see. I think the BA pins inside the MCM is
configured with the CONF[0-4] pins on the MCM. You should therefore
verify that these are really correclty connected. The same problem
can also occur with an incorrect R_SDRAM_CONFIG.
Magnus Martensson (email@example.com) will possibly send you more
info about the magics of the CONF pins and R_SDRAM_CONFIG.