Trying to get CFI to work on my board.

David Woodhouse dwmw2 at
Tue Feb 13 11:06:52 EST 2001

subodh at said:
>  Correct?  So when I  execute virt_to_phys on physmap_map.map_priv_1
> shouldn't I get my "original" physical address of 0x28000000 back? but
> I am getting  following output 

No. You should get a 'cookie' back which is to be passed to the kernel's 
read[bwl] functions. This 'cookie' may be coincidence be very similar to a 
virtual address. 

I don't think virt_to_phys() is supposed to work for ioremapped areas - 
only for main memory. If you just use map_rom_probe instead of cfi_probe, 
can you read sensible data from the beginning of the chip (i.e. does it 
match what you know is in there)?

What chips are on the board and how are they arranged? 


To unsubscribe, send "unsubscribe mtd" to majordomo at

More information about the linux-mtd mailing list