Address sanitizer unhappy on develop running simple benchmark

Developers
1

hey y’all,

I had an idea about trying to get the benchmark times down, but then this lead to the benchmark running out of memory, which made me think it was now 1) running faster 2) therefore being run more times by the benchmark library 3) therefore finally triggering an OOM segfault. So I fired up the address sanitizer and I’m seeing really weird stuff even on develop and I’m a bit worried, especially given that the Clang team claims that “AddressSanitizer does not produce false alarms.”

Can anyone try this with me and make sure I’m not crazy? I think we should probably get to the bottom of this before releasing 2.19… You can run with the memory sanitizer in clang by adding:
-g -fsanitize=address -fno-omit-frame-pointer

Thanks 😅

2

False alarm! I think this is maybe not a big deal as it doesn’t seem to have problems when I run the same code outside of the benchmark in a for loop.

4 Likes