-
-
Notifications
You must be signed in to change notification settings - Fork 697
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Segmentation fault from xcffib when quitting #626
Comments
Any chance you can run the whole thing in GDB so that it gives us a stack trace? |
Yep
|
Huh, interesting. It looks like that should be protected in xcffib, but apparently it's not :). I'll take a closer look at some point; does this keep it from segfaulting at least? http://paste.ubuntu.com/10502050/ |
Nope, that patch doesn't avoid the segfault :( |
Huh, ok. Thanks, I'll keep thinking about it :) |
Can you post your config that crashes, by chance? |
This is a minimal config that produces the segfault when quitting with
Tested with qtile 989f0f8, Python 3.4.3, xcffib 0.2.0, Xorg 1.17.1, Intel graphics (driver 2.99.917) in Arch Linux x86_64 up-to-date as of 2015-03-04 (Linux 3.18.6). Commenting the |
Hmm. So I tried reproducing this myself and I couldn't get it to fail. What I suspect is happening is destructor ordering strangeness; something like python is deleting globals from out from under xcffib before the del methods in qtile are called. However, that's just a guess. The stack trace you posted is a little confusing to me, because as far as I know xcffib should never call back into python, but it seems to by (viz. the PyLong_Type call at the top of the stack). Any chance you can install xcb's debug symbols and run it again? |
Um... sorry, this bug only happens in a very particular scenario, i.e. when launching qtile from an already started Xorg session. What I do for testing is having a
Once X is started with |
Hi, @luxrck's patch seems to indicate it's a problem with gobject, does tych0@d33b200 fix it? |
Hmm, is this still a problem? If so, any chance #702 fixes it? |
TBH I still have problems when testing as explained in my post above: if you want I can post outputs and backtraces, maybe it's the symptom of some other problem that might affect something else, however you may instead decide that testing qtile that way is not supported, and I'd be fine with that ^^ |
Ah, I see, this is a different segfault then you mentioned here, I think I fixed that one, but I haven't seen this one. I can try to reproduce it later tonight. |
The generated exceptions/segfaults seem to vary quite a bit for me depending on the checked-out commit. For example, with the latest commit, after starting X as explained above I get an exception when trying to launch qtile with |
This issue is stale because it has been open 90 days with no activity. Remove stale label or comment or this will be closed in 30 days. |
Tested with qtile 3eadbbf (latest commit while I'm writing) on Python3 with python-xcffib 0.2.0 in Arch Linux.
When exiting qtile, a segfault happens systematically if the screen has a Bar, regardless of the added widgets:
Sometimes, more or less randomly, better tracebacks are output:
No segfaults happen if the screen doesn't have a Bar or if it only has a Gap.
The text was updated successfully, but these errors were encountered: