Bug in mudflap?

[Frank Ch. Eigler]

Hi -


Benjamin Redelings I <bredelin@ucla.edu> writes:

> [...]
> While trying to figure out why my C++ program crashes under g++ 4.0
> (cvs), I tried using mudflap.
> 
> However, the following extremely short program generates several
> 'violations', which doesn't make sense.
> 
> #include <vector>
> int main() {
>    std::vector<int> v;
>    v.push_back(1);
> }
> Is there something obvious that I am missing?
> [...]

There might be an instrumentation problem, or even conceivably a real
libstdc++ problem.  Running the instrumented executable with 

% env MUDFLAP_OPTIONS=-trace-calls  ....

will show a trace of registration calls, which should correspond to
objects going in and out of scope.  Often something obvious shows up
there.  I'm investigating, but would you mind opening a bugzilla
report with your little program in it?


- FChE