http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57489
--- Comment #4 from Jim Hand <jhand at austin dot rr.com> --- One of my coworkers, a former Intel employee, made the point that signed integer overflow is precisely defined for X86, in that overflowing and then underflowing will produce the correct value 100% of the time. Another point that was made is that when the test fails, there is no other way to see why it should fail: silent failure. Moreover, adding instrumentation would change the results. The few of us that have discussed this believe that the compiler is producing the incorrect results in this case.
