[Coco] NitrOS-9 and TCC1014
robert.gault at worldnet.att.net
Wed Jan 13 21:37:52 EST 2010
Mark Marlette wrote:
> The sparkles are REALLY present in test case a.
> The point of crash is random. Once in 10 it will boot.
> VERY strange. Again only the GIME changes and it works.
> I will try a 6809 boot same test case tomorrow. Doubt there will be an issue.
> I do recall the boot problem prior to .06.
If one assumes the crash is during the loading of the kernel and os9boot
and then looks at the source code for boot and rb1773, you will find
that timing constants are different in Level2 for 6809 and 6309 systems.
That is not a surprise as the number of cycles for op-codes can change
if run in 6309 native mode.
There is a small possibility that the older GIME is not able to handle
the data flow at the faster speed of native 6309 code.
I guess a test of this possibility would be to use the same 6309
assembly but forcing it to be in emulation rather than native mode for
Boot and rb1773. Only problem with this test is the stack is handled
differently in emulation vs native mode on interrupts.
Well let's see what you get for the 6809 code.
More information about the Coco