Electronics > KiCad

Version 8 Project News

<< < (7/11) > >>

Kais:
In the mean time the release candidate for version 8 RC1 is out:
https://www.kicad.org/blog/2024/01/KiCad-Version-8-Release-Candidate-1/

shapirus:
Anyone who's using Linux, does the memory leak that I described in https://gitlab.com/kicad/code/kicad/-/issues/16404 reproduce for you? What are you using: X11, Wayland?

It apparently does not reproduce under Windows, at least, per one comment in that ticket.

For me with X11 (and KDE) it reproduces very reliably, which makes me restart kicad from time to time to avoid OOM, and that's quite annoying. And that's even though I have 64GB of RAM on my desktop -- it would've been much worse with 32, let alone 16.

If more people can reproduce it, then we can try to narrow down a specific enviroment where it occurs, which may help the developers fix the leak before 8.0 is released. I think it's a quite serious bug, as is any significant memory leak.

Also, does anyone have any suggestion of a memory leak detection tool that I could use to track what exactly is not freeing memory in my case?

nctnico:
IMHO Valgrind is THE tool of choice to find memory leaks in Linux software. I had to use it a couple of times and it works like a charm after a short RTM session. But you'll likely need unstripped binaries with debugging information enabled for Valgrind to show you which line of code is causing the leak.

shapirus:

--- Quote from: nctnico on January 20, 2024, 03:54:53 pm ---IMHO Valgrind is THE tool of choice to find memory leaks in Linux software. I had to use it a couple of times and it works like a charm after a short RTM session. But you'll likely need unstripped binaries with debugging information enabled for Valgrind to show you which line of code is causing the leak.

--- End quote ---
Yeah I should have known, as I used valgrind for my own programs myself (even though it was 20 years ago).

KiCad is running super extra slow under it though... Will see if it reports anything.

update: it's not just slow, but it appears to be going into an infinite loop once I click the start simulation button. Happens with both memcheck and massif. Meh.

I guess I'll try some other tools. I remember there were a few that would LD_PRELOAD their libs that substitute the normal malloc() calls, now I need to dig them out of my memory or just find them from scratch.

SiliconWizard:
I'd be curious to see what Valgrind finds with Firefox... :-DD

Navigation

[0] Message Index

[#] Next page

[*] Previous page