Hi,
branch "cfi" inside
http://git.infradead.org/users/borneoa/openocd.git
contains a set of patches I would propose for being merged in OpenOCD.
My development is mainly a review of memory accesses to CFI flash to
guarantee that every access is performed at CFI bus_width size.
Current version of OpenOCD does not strictly obey to such role, using
mix of 8, 16 and 32 bit read accesses.
As you could imagine, I'm working on a target that does not like so
much all this flexibility.
a) if you want to test my tree, focus on :
1) verify your config file handle properly the line "flash bank
... cfi ...".
This is supposed to work for every TCL script inside OpenOCD,
but a custom configuration could fail.
2) check you get correct flash ID as output of "flash probe 0"
3) check the flash can be properly written
b) I still need to fix the code for erase_check. This is the last
missing piece, but requires some time.
c) I have noticed that the assembly code for flash programming in
cfi_intel_write_block() and cfi_spansion_write_block() is for ARM only,
while CFI is used also for MIPS cores (./tcl/board/x300t.cfg)!
Is this a bug? It is not in the tracker.
Is anyone already working at moving this part in target specific files?
One of the commits is a rewrite of my former patch, posted on March 24 as
> [RFC/PATCH] CFI: simplify bufferwsize computation
now obsoleted.
Best Regards,
Antonio Borneo
_______________________________________________
Openocd-development mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/openocd-development
