Code banking on upsd3454E

[mikesponk]

Posted on October 18, 2008 at 17:21

Code banking on upsd3454E

[mikesponk]

Posted on May 17, 2011 at 11:52

Hi, does anyone has any suggestion for code banking on 3454E?

The ride emulator is always giving me ''conflict result'', like if the flash memory was misprogrammed.

The csiop is at 0x7f00 both in psd and in ride setup, and I am using the latest l51_bank assembly file, the one with the prefetch correction.

I am using the main flash in program space:

- fs0 : 0x0000<=address<=0x7fff (common bank)

- fs1 (page=0) and 0x8000<=address<=0xffff

- fs2 (page=1) and 0x8000<=address<=0xffff

...

- fs7 (page=6) and 0x8000<=address<=0xffff #- fs1 (page=7) and 0x8000<=address<=0xffff

- csiop at 0x7f00

- the csboot is in data space

What am I doing wrong?

Thx

[mikesponk]

Posted on May 17, 2011 at 11:52

Ok, after a lot of digging, I have my code banking running...

First of all, the ''conflict result'' message has been solved merging the pld/code from psdsoft and programming with it the flash. In my Ride IDE, I just start debugging with no erase/download.

Second: the l51_bank.a51 cannot be used as is for upsd3454. The movx a,@Rx eg won't run, so U cannot use external stack. I have reserved from assembly the 256 pdata space for my stack, and I have manually changed the routine in l51_bank.a51.

Total: 2 days of hard working, and when I was starting with code banking, 2 days ago, I was thinking about a simple placement of a flag in a ''bank switching'' tab...

Now, I will write to Ride... :(