Conversation
Makefile.am
Outdated
| @sed '/^# AUTO-GENERATED START/q' prof.supp > $@ | ||
| @cat /usr/share/glib-2.0/valgrind/glib.supp >> $@ | ||
| @wget -O- https://raw.githubusercontent.com/python/cpython/refs/tags/v`python3 --version | cut -d' ' -f2`/Misc/valgrind-python.supp >> $@ | ||
| @wget -O- https://raw.githubusercontent.com/GNOME/gtk/refs/tags/`pkg-config --modversion gtk+-3.0`/gtk.supp >> $@ |
There was a problem hiding this comment.
This one isn't working: https://raw.githubusercontent.com/GNOME/gtk/refs/tags/gtk.supp
This one is though: https://raw.githubusercontent.com/GNOME/gtk/refs/heads/main/gtk.supp
There was a problem hiding this comment.
This one is though:
https://raw.githubusercontent.com/GNOME/gtk/refs/heads/main/gtk.supp
You most likely don't want the supp file for main, but for "your installed gtk version".
This one isn't working:
https://raw.githubusercontent.com/GNOME/gtk/refs/tags/gtk.supp
This is wrongly extended from
https://raw.githubusercontent.com/GNOME/gtk/refs/tags/`pkg-config --modversion gtk+-3.0`/gtk.supp
, so it seems like pkg-config --modversion gtk+-3.0 does not work for you.
Do you know where this fails for you?
What do the following commands return?
pkg-config --modversion gtk+-3.0
pkg-config --modversion gtk+-2.0
There was a problem hiding this comment.
Ah indeed. gtk3-devel wasn't installed. With that it works fine.
There was a problem hiding this comment.
Researched this a bit further and removed the dependency on gtk3 + the part for gtk2, since there's no Valgrind supp file provided yet.
|
Seems to work also with reconnects. I'll keep this branch running for some time to see if I find more. |
sjaeckel
force-pushed
the
debug-1946
branch
2 times, most recently
from
ed95f4a to
e3817bb
Compare
This seems to be some kind of standard script name for bootstrapping autotools. Signed-off-by: Steffen Jaeckel <s@jaeckel.eu>
This reverts commit d7e46d6.
They're only used in `window.c`. Signed-off-by: Steffen Jaeckel <s@jaeckel.eu>
This creates a "as personal as possible" Valgrind suppressions file. Signed-off-by: Steffen Jaeckel <s@jaeckel.eu>
Add all elements to an array, in order to easily be able to run operations in bulk. Signed-off-by: Steffen Jaeckel <s@jaeckel.eu>
`Py_Initialize()` does it since 3.7, it's a noop and deprecated since 3.9. https://docs.python.org/3/c-api/init.html#c.PyEval_InitThreads Signed-off-by: Steffen Jaeckel <s@jaeckel.eu>
Signed-off-by: Steffen Jaeckel <s@jaeckel.eu>
Before this change issuing `/quit` directly called `exit(0)` and did not invoke all the graceful shutdown routines. Now we first try to exit from the event loop, which includes cleaning up everything. In case the event loop is stuck for some reason, you could try to issue a second `/quit`, which will then directly call `exit(0)`. Signed-off-by: Steffen Jaeckel <s@jaeckel.eu>
* Fix some linter suggestions. * Change some defines to `const` vars. * Free buffer entries in the reverse order they were allocated. Signed-off-by: Steffen Jaeckel <s@jaeckel.eu>
|
Not too sure about the changes in prof.supp. But I want to move things forward. |
2 participants
This is a starting point while debugging #1946.
The memory usage on current develop has increased significantly since 0.14.0.
The main memory saving was done by reverting d7e46d6 which is not in 0.14.0.
I did not yet conclude on what the real problem is, but my guess is that the memory growth can only be tackled by rewriting some internal parts and we're not really leaking memory. AFAICT most memory is accounted for and is free'd on program termination.
NB: The valgrind output of a default build, executed with "our default valgrind settings" is kind of misleading. Each time a gnupg API is invoked it forks which causes the valgrind output to blow up and be hard to analyze. One way to improve the debug experience would be to disable following forks when running valgrind, another is to disable gnupg.
I've extended our default
prof.suppby quite a bit, which must still be reviewed whether those excludes should be there or not.