Bogus Write/Write Data Race problem reports

Bogus Write/Write Data Race problem reports

I am using Inspector 2017 Update 4 on Windows. Running a "Detect Deadlocks and data races" analysis of some multi-threaded C++ .

My reports are including so many "bogus" write/write data race problems that real problems get lost in the noise.

These bogus reports are all similar. One "write" is on ::malloc() (or other low level C++ allocator) and the other "write" will be somewhere else in my C++ where some fields are being set.  They are completely bogus reports - the report would indicate that in one thread #1  malloc is writing into a memory location shared by  a valid object operation happening in thread #2 . The software would be crashing immediately if this were true.

It's very frustrating as I have to forensically examine every report to decide "bogus vs real".

1 post / 0 new
For more complete information about compiler optimizations, see our Optimization Notice.