Author uses jmalloc to confirm malloc allocations are unfreed, then later speculates the allocations might be something not visible to Valgrind e.g. mmap. I’ve

Author uses jmalloc to confirm malloc allocations are unfreed, then later speculates the allocations might be something not visible to Valgrind e.g. mmap. I’ve often done similar mistakes, where data from step 1 already rules out a hypothesis for step 2 - but I’m too sleep-deprived and desperate to realize it. Debugging production issues is the worst.