Re: FW: [avrdude-dev] Windows port (was: Butterfly proposal: Exit upgradefunction)

[E. Weddington]

On 9 Feb 2004 at 21:45, Jan-Hinnerk Reichert wrote:

>> > > I can't see a reason. Perhaps it was, because I still had the
> > > command.exe running while doing the change. Or windows just
> > > forgot to apply the changes ;-(
> >
> > Oh, I bet you that's it! Definitely close the command line box
> > before making the changes. It does not automatically grab those
> > changes when running.
> 
> It's not this way. If I close it and open a new one the changes don't 
> take effect. And technically changing the FIFO-threshold is just a 
> single "out"-command, no OS stuff ;-)
> IMHO the only reason would be, that Windows refuses to set the value, 
> because the command.com holds a lock or something like this...

Yeah, close it, make the change, *then* open the command line box. See if that 
works.

 
> I have already changed the value in my kernel once, it's quite easy. 
> The hard stuff is making an interface to user space or at least 
> implementing a boot-option.
> Writing a program for directly changing the FIFO is not feasable (even 
> with root privileges), since Linux sets FIFO-threshold everytime the 
> baudrate is set ;-(

?! That's odd. What does a baud rate have anything to do with the FIFO buffer 
length? :-/