bug#43389: 28.0.50; Emacs memory leaks

[Russell Adams]

On Fri, Nov 27, 2020 at 02:38:07PM +0200, Eli Zaretskii wrote:
> > Unfortunately the Emacs that was 8GB has since been stopped, I killed
> > it before working with the trace files. My laptop was rebooted later
> > when the trace statistics utils ate all the RAM (my error, wrong input
> > file).
> >
> > This list of timers is from a new instance, but the configuration
> > hasn't changed.
> >
> > Are the 50+GB of trace files I have of any value?
>
> I don't think Carlos and others saw your reports, because they were
> not CC'ed.  I'm CC'ing them now; please make sure to reply to all of
> them next time.
>
> Carlos, please read
>
>   https://debbugs.gnu.org/cgi/bugreport.cgi?bug=43389#554
>
> for the details posted by Russel about his data points.  If you can
> instruct him how to produce some analysis from the mtrace files, or
> how to make them available for your analysis, please do.

I find particularly of interest the growth of Emacs processes while
idle.

Yesterday I restarted Emacs and over the next 18 hours I left my
laptop idle with Emacs as the focused application. My Emacs has grown
to 3GB and every time I select my Emacs window it will grow by a few
MB while I watch in htop.

I will restart it again tonight and leave it focused, and see if I can
reproduce the growth. It also appears that the growth is not linear,
slower at first and hard to see, but in the multiple MB at a time
later when the total is in GB.

Again I use emacs in daemon mode with one or more emacsclient
processes connected (x11 and terminal). I use StumpWM in full screen
mode with my emacsclient, and if it's focused it seems the growth
continues despite xscreensaver coming on and dimming the screen.

------------------------------------------------------------------
Russell Adams                            RLAdams@AdamsInfoServ.com

PGP Key ID:     0x1160DCB3           http://www.adamsinfoserv.com/

Fingerprint:    1723 D8CA 4280 1EC9 557F  66E8 1154 E018 1160 DCB3