Re: [avr-gcc-list] AVRDude serial access problem under WinXP

[Colin O'Flynn]

Hi,

I'll open the bug report. Is it just this though:

  stk500_initpgm(pgm);
  if (pgm->open(pgm, port) >= 0)
    {
      fprintf(stderr,
              "%s: successfully opened stk500v1 device -- please use -c 
stk500v1\n",
              progname);
      return 0;
    }

******** ADD THIS LINE: pgm->close(pgm);

  stk500v2_initpgm(pgm);
  if (pgm->open(pgm, port) >= 0)
    {
  ....

Where we need to add a line to close the port inbetween?

Regards,

 -Colin


On Sunday 25 March 2007 15:59, Joerg Wunsch wrote:
> As Colin O'Flynn wrote:
> > I CAN confirm this bug. However - it is not quite as reported. The
> > output looks like:
> >
> > Sending bytes [some bytes]
> > STK500 Can't find Sync
> > ser_open.c: Can't open com1. Access denied.
> >
> > The important thing here is the error occurs AFTER the port has been
> > opened and some data has been transferred.... so I guess it's an
> > error with how the STK500 v2 detection is done?
>
> Ah, that rings a bell.  That's what happens if the original report is
> too fuzzy.  Thanks for clarifying: I think what happens here is that
> the STK500 v1 code gives up, but doesn't correctly close the port
> then.  When the STK500 v2 code comes along, and wants to open the
> port, the OS refuses that.  (Under Unix, there's no issue with this,
> there can be multiple concurrent opens on the same serial line.
> Philosophical OS differences.)
>
> Please open a bug report for that (someone, not quite necessarily
> Colin, but I don't care).  I'm not sure I'll be able to resolve that
> immediately, because it could require some more restructuring to
> provide a proper strategy of cleaning up before leaving.