https://bugs.kde.org/show_bug.cgi?id=514025

Sewer56 <[email protected]> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEEDSINFO                   |RESOLVED
         Resolution|WORKSFORME                  |NOT A BUG

--- Comment #3 from Sewer56 <[email protected]> ---
(In reply to Milian Wolff from comment #1)
> when you look at the time range start -> your point of interest, then the
> "leaked" cost gives you the data you are looking for, no?

Anyway, been 3 months since I last checked; but I had another peek quickly.
Seems you're right. I honestly can't believe I didn't think of this back in
December; I feel a bit silly in retrospect.
This issue can be closed.

Only caveat is knowing when to get a good reading.
I wound up just adding a long sleep that would stand out very clearly in the
'consumed' graph (as a flat line)

```rust
sleep(Duration::from_secs(10));
std::process::exit(0);
```

And can check by selecting start of chart to the clearly marked flat segment.
Only caveat is I got 20MB allocated but process reports a PSS of >500MB, but
that's very much a `glibc` thing; swapping out allocator when not profiling
works. It's the good old `MALLOC_ARENA_MAX` and friends.

Thanks!

PS. I can't believe the same person answered on both Hotspot and Heaptrack
right after the other; small world 😅

-- 
You are receiving this mail because:
You are watching all bug changes.

Reply via email to