Re: [hatari-devel] A question about the debugger

[ Thread Index | Date Index | More Archives ]

On 5 Sep 2013 at 10:54, Miro Kropácek wrote:

> On Thu, Sep 5, 2013 at 10:39 AM, Roger Burrows <rfburrows@xxxxxxxxx> wrote:
> > 4. the converter stopped accepting the command, causing a timeout.
> Btw, how would you explain it worked for TOS? Maybe there's a different
> timout handling at least?
It only showed up on SPECIFIC patterns of bits within the commands.  I happened
to see it when copying a very large file in EmuCON.  Most of the time, it
worked even with the double termination.  By putting in debugging statements, I
found where the copy was failing, then wrote a tester so I could reproduce it.

For example, the following read commands failed:
	08 10 ff xx yy 00
	08 08 ff xx yy 00
	08 04 ff xx yy 00
	08 02 ff xx yy 00
where xx, yy are any hex digits.

The following almost-the-same read commands did not fail:
	08 10 fe xx yy 00
	08 0f ff xx yy 00
etc etc.

It just happened that the location of that large file on disk, plus the amount
of sectors read each time by the copy command, caused a read command to have
one of the bad combinations.


Mail converted by MHonArc 2.6.19+