Mail Archives: djgpp/1994/01/10/07:17:25
We've got a piece of code hanging off the clock interrupt. Under
certain circumstances it appears to crash. The causes are
correspondingly difficult to pin-down, but the following data have
been collected:
1) If the interrupt routine does no fp, then the problem
disappears.
2) If we add a bit of asm code to save and restore the fp
registers as we enter and leave the interrupt routine,
things get better - we've not seen this crash but hesitate
to say that this is definately the cause.
Is it possible that the fp status isn't being correctly saved or should
we look elsewhere for a bug ?
Thanks in advance,
M.
- Raw text -