On 07/01/2013 12:51 PM, Wolfgang Denk wrote:
Dear Mike Dunn,
In message 51D1C455.9010801@newsguy.com you wrote:
But there's a good motivation for wanting to turn off optimization.
I disagree here. If you are hunting down a problem, you want to be as close at the original code as possible. Disabling optimization is such a dramatic change to the generated code that you actually debug a different program.
I guess it depends on what problem being debugged... I think your point is valid for things like tracking down race conditions or misbehaving hardware. But for more mundane problems like logical errors in the code, optimizations get in the way. It would be nice to be able to experiment with optimizations off. Just knowing that a particular problem still exists when unoptimized code is run will provide some clues.
Single-stepping with a debugger at the C source level is almost useless. I've since gotten better at single-stepping at the assembly level while using the mixed c and assembly view of gdb.
Hm... Did you read up the documentation, say [1], and try out these recommendations?
No, but I will. Many thanks! Mike