Re: [avrdude-dev] Re: avrdude 4.2.0 Release

[Theodore A. Roth]

On Fri, 5 Sep 2003, Jan-Hinnerk Reichert wrote:

> On Friday 05 September 2003 08:18, Theodore A. Roth wrote:
> > On Thu, 4 Sep 2003, Brian Dean wrote:
>
> > I've tested the attached patch as best I can since my 8515's don't
> > support SPI access to fuses and my only mega163 is temporarily
> > hosed with the SPIEN fuse disabled (need HiV to fix it). :-\ Seems
> > to me it issues the universal commands correctly.
> >
> > It's a little bit of a tweak to Jan's patch. His patch generated
> > warnings due to avr_{read,write}_byte_default() not having protos
> > in avr.h in addition to the broken encapsulation issue.
> >
> > Essentially the same code gets executed in both patches, but I
> > think mine is a bit more clear as to what is getting called when
> > looking at avr.c (of course, that's open to debate ;-).
> >
> > If there's no objections to my hack, I'll commit it tomorrow.
>
> I don't know the semantics of the return codes from "avr_read_byte"
> and "avr_write_byte". Is it only "not supported" or "OK"? If another
> error occurs, the default routine should not be called.

I did consider that and concluded that if the first attempt fails for
any reason, it probably isn't going to hurt to try the default and see
if it works. In most cases (every case I could come up with, in fact),
the default is a direct SPI attempt.

I still consider this a short term hack. Currently, the only failure
in pgm->write_byte is for unsupported and only avr910 uses the
write_byte method (similar for read_byte). If the cleanup to avr.c I
proposed last night goes as I expect, this hack will disappear in the
next dev cycle.

Ted Roth