[avrdude-dev] [bug #21954] verify fails for masked ('x') bits

[Joerg Wunsch]

Follow-up Comment #3, bug #21954 (project avrdude):

I suggest you start digging through the code, and have a look
at *how* you would integrate a real fix for it, without
breaking anything else.

I recently made a sweep through avrdude.conf, and unified the 
fuse bit settings for all except archaic devices (which have 
very odd fuse read/write commands, compared to today), so there's 
a consistent state now.  Note that this consistent state means 
the definitions for read and write commands differ: while the 
read command always reads full bytes (8 x 'o') so the users 
get the '1' bits in the unused bit positions returned, the 
write command only has 'i' bits for those bits that are marked 
writable by the datasheet.

To make things worse, all recent AVRs have defined their unused 
upper two lock bits as '1' (rather than 'o') in the read command. 
See bug #43421 for the symptoms.

If you can come up with a patch that can consider all the 
ramifications, without adding too ugly hacks into the generic 
memory layer, you're welcome.

    _______________________________________________________

Reply to this item at:

  <http://savannah.nongnu.org/bugs/?21954>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.nongnu.org/