I did take a cursory look at it and didn't see anything which would explain such a message. In fact, the implementation of reset_object_address is compiled into the library so I couldn't even see how such a thing could even be detected by just looking at the header. Of course the error message is pretty cryptic to my eye as I don't use that tool myself so it could well be the error message was misinterpreted. Robert Ramey David Abrahams wrote:
on Tue Jul 17 2007, "Robert Ramey"
wrote: Tell me what the change to the source code should be.
Robert,
Valgrind is very reliable (although it is possible for it to report false positives about uninitialized data) and the OP's test case looks simple enough to be considered minimal. If you can't find a problem with it, can you at least be looking at your implementation of reset_object_address to see if you can find a likely use of uninitialized data without making the OP debug unfamiliar code?