How to figure out what function calls correspond to in Samply?

help
1

I've narrowed down optimization opportunities to a few obvious costly calls I'm making, but for the life of me I can't figure out what this function call is implying:

core::ptr::drop_in_placealloc::string::String

I think drop_in_place has something to do with calling a destructor on a String type. But why would it be so expensive?

calling to_string is 9.8% of my cpu time, while drop_in_place is 13%.

And how can you reason about flame graphs from Samply when the function call is something more obscure like this?

2

Is the flamegraph showing you what is calling drop_in_place?

Looking at the example flamegraph on Samply's readme, you can ask a similar question:
"Why is std::thread::park_timeout taking up so much time?"
Looking at the bars underneath (in the flamegraph) we can see it's called in a function associated with downloading data. It makes sense that rustup would spend time downloading data, as that's what it supposed to be doing.
Is the application you're running dealing with a lot of owned Strings?

If you start by reducing the to_string calls where you can, then you may see the drop calls also become reduced.

2 Likes
3

The main cost of a String or Vec destructor is freeing its heap-allocated memory. On macOS, there is an added cost of zeroing out the freed memory (source).

2 Likes