adding a "__builtin_assume(offs==0);" shows the same problem, and
adding a "__builtin_assume(offs=0);" avoids the problem but shows a new warning ...
test.c:14:20: warning: the argument to '__builtin_assume' has side effects that will be discarded __builtin_assume(offs=0); ^~~~~~ warning: Trimmed ExplodedGraph is empty. Warning: dumping graph requires assertions 1 warning generated.
So for the case of "__builtin_assume(offs=0);", one of two things may be occurring:
1) There really may be a side effect even though the warning says something different.
2) There really is no side effect, but the SA is picking up on the constraint.
These are contradictory, and I'm not sure which is correct :/
Does any one have knowledge of this particular issue and where to dig further into this? Or maybe I'm just missing something obvious :/