David C. Rankin wrote:
On 07/06/2019 02:07 AM, Per Jessen wrote:
I've read the longish kde report - I don't see any difference in valgrind's behaviour, it just reports what it sees? Are you saying valgrind would previously ignore or suppress allocations made by library calls? (possibly directed by some blacklist). The two devs don't seem to think valgrind ever had such functionality.
Yes, exactly. Since time began for valgrind, up until the point I filed the bug reports, valgrind always shipped with proper exclusion files that masked the non-user allocations that take place behind the scene.
Okay, understood.
It's really simple, look at the bytes reported below:
Oh, I understand the problem. I'm just thinking about how it might be solved with a (simple?) patch to those suppression files. On a memcheck, you can add your own with '--suppression='. ISTR valgrind will even help you produce those files.
In an actual real-world program it makes the number valgrind reports completely batshit worthless. There will be scores of invisible allocations reported and tens of thousands or hundreds of thousands of system allocations mixed in with the total making that number completely meaningless for any type of analysis of user allocation requirements.
That sounds like a bit of a stretch. I am assuming the standard set of suppression files are still in place, but need updating to suit the libraries ? -- Per Jessen, Zürich (18.7°C) http://www.dns24.ch/ - your free DNS host, made in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org