Alignment/memory map issue?

daneichenwald2 · ‎2009-11-19

Posted on November 19, 2009 at 21:37

daneichenwald2 · ‎2011-05-17

Posted on May 17, 2011 at 13:31

We are experiencing a very strange issue for the second time, both instances completely unrelated. It boils down to the fact that we can add No-ops into functions that are not even on the code path (they are definitely being compiled in though) and our code begins to randomly behave strangely. So things work perfectly before adding it and things start randomly misbehaving after. Even when it starts misbehaving, the majority of the system works just fine, so its not that the whole linking got screwed up. We suspect that there is a memory map issue where adding just one no-op causes a large ripple effect in the map. We've tried a binary search to change where the no-op was being added to try to narrow down an addr before or after which causes a large change, but haven't had any success yet. Has anyone else experienced anything similar? Any other suggestions?

Thanks,

Dan

[ This message was edited by: daneichenwald1 on 19-11-2009 23:44 ]

picguy · ‎2011-05-17

Posted on May 17, 2011 at 13:31

I had a problem not unlike this. For reasons I did not understand the tbb (Table Branch Byte) instruction had to be on an even or odd half-word boundary. Put an org mod 4 into each of your modules. Then if the problem is still around add one nop into each module one at a time. I hope this helps.

Whatever you find please let us know.