Re: [avrdude-dev] support for bitbanging GPIO lines through /sys/class/gpio

[Radoslav Kolev]

On Mon, Apr 12, 2010 at 1:17 AM, Joerg Wunsch <address@hidden> wrote:
> The question is probably whether there is *any* known application that
> really runs AVRDUDE in such an (embedded) application right now, or
> can be expected in the near future.  So far, I've only seen AVRDUDE
> operating in classical PC (or alike) desktop environments.

It started as a discussion thread about the Bifferboard embedded
device I mentioned in my first message:

http://groups.google.com/group/bifferboard/browse_thread/thread/7c4835cb1773b5d4/1ae4238738a7c763?lnk=gst&q=avrdude

To sum up the discussion:

Ivan implemented an AVR programmer in Perl to use his Bifferboard GPIO
lines to program an AVR. I suggested that it would be better to add
this as part of avrdude to reuse all the MCU definitions. Biff
expressed an interest and I decided to do it as an exercise during the
weekend. So currently there are 2-5 potential users :).

> At the least though, the patch should auto-probe to see whether the
> required generic GPIO interface is available in the target system or
> not, i.e. it must contain the necessary autoconf glue logic.

I'm not good with autoconf, but that may be a good occasion to do some
reading. I don't know if it will be possible to automatically detect
this though. Usually for embedded systems you cross-compile so testing
about sysfs-gpio support in the kernel of the system doing the build
is not of much use. Probably a better approach would be to provide a
configure option or something to enable/disable this programmer if
needed.

How is this usually handled for other similar cases in avrdude?

This uses nothing more than file open/read/write/close functions so it
should build on all platforms. If you try to use it and the expected
/sys/class/gpio directory is not there it will exit with an error.