[STUMP] Crash:StumpWM on Ubuntu Trusty

Discussion:

r***@google.com

2017-04-02 19:39:24 UTC

I just rebuilt against ***@head and still reproduce a reliable
crash -- this crash started happening about mid-February

Basically Stumpwm crashes if I do something like this: [12:18]

(setq *message-hook* (list #'(lambda nil (echo "foo"))))

I reduced the crash to the above minimal example: in my actual use
case, I used to have a function on that hook that spoke the message to
. [12:19]

incidentally, the same configuration is working fine on my desktop
running Trusty -- but a 3.x kernel

Looks like the above crashes all of X [12:20]

fatal error encountered in SBCL pid 2614(tid 140737353893696): [12:21]

%PRIMITIVE HALT called; the party is over.

Error opening /dev/tty: No such device or address

(obno-popup-notifier:2534): Gdk-WARNING **: obno-popup-notifier: Fatal
IO error 11 (Resource temporarily unavailable) on X server :0.

xterm: fatal IO error 11 (Resource temporarily unavailable) or
KillClient on X server ":0"

g_dbus_connection_real_closed: Remote peer vanished with error:
Underlying GIOStream returned 0 bytes on an async read
(g-io-error-quark, 0). Exiting.

12:20:44 raman-glaptop2 ~ $

stumpwm version from stumpish: >
1.0.0-88-g7d63123 Compiled On Sun Apr
02 2017 10:53:41 [12:23]

is there somewhere I should email the above? [12:31]
<alezost>
Raman: you can report at
<https://github.com/stumpwm/stumpwm/issues/new>
[12:33]

HMM, if that requires use of a JS-powered browser, it's going to be
painful. Can I email in an issue? [12:34]
ERC>
HMM, if that requires use of a JS-powered browser, it's going to be painful. Can I email in an issue?

--
--

David Bjergaard

2017-04-03 15:39:29 UTC

Permalink

Hi Raman,

I respect your choice to use a browser that doesn't use javascript, but I have
to warn you that it may make resolving this issue more difficult. I have opened
an issue for you here:

https://github.com/stumpwm/stumpwm/issues/343

We will track changes there so if you can at least monitor that page for updates
to your bug that would be great.

The next step in narrowing this down is to use git to track down the commit that
caused the break. Failing that, two things stick out to me as odd:
1. the "Error opening /dev/tty" is strange, I can't find anywhere in the
backtrace for "echo" that needs a pseuod-terminal
2. The obno-popup-notifier is unfamiliar to me, I haven't seen those types of
errors before. Did Ubuntu recently change the notification handler?

Next steps:
Is it possible to get a backtrace when the crash happens?
If not, please set up the debugging output to a file and see what is in the log
when the crash happens.

Revert the commits until you find a version of stumpwm that doesn't crash using
your recipe.

David

Post by r***@google.com
crash -- this crash started happening about mid-February
Basically Stumpwm crashes if I do something like this: [12:18]
(setq *message-hook* (list #'(lambda nil (echo "foo"))))
I reduced the crash to the above minimal example: in my actual use
case, I used to have a function on that hook that spoke the message to
. [12:19]
incidentally, the same configuration is working fine on my desktop
running Trusty -- but a 3.x kernel
Looks like the above crashes all of X [12:20]
fatal error encountered in SBCL pid 2614(tid 140737353893696): [12:21]
%PRIMITIVE HALT called; the party is over.
Error opening /dev/tty: No such device or address
(obno-popup-notifier:2534): Gdk-WARNING **: obno-popup-notifier: Fatal
IO error 11 (Resource temporarily unavailable) on X server :0.
xterm: fatal IO error 11 (Resource temporarily unavailable) or
KillClient on X server ":0"
Underlying GIOStream returned 0 bytes on an async read
(g-io-error-quark, 0). Exiting.
12:20:44 raman-glaptop2 ~ $
stumpwm version from stumpish: >
1.0.0-88-g7d63123 Compiled On Sun Apr
02 2017 10:53:41 [12:23]
is there somewhere I should email the above? [12:31]
<alezost>
Raman: you can report at
<https://github.com/stumpwm/stumpwm/issues/new>
[12:33]
HMM, if that requires use of a JS-powered browser, it's going to be
painful. Can I email in an issue? [12:34]
ERC>
HMM, if that requires use of a JS-powered browser, it's going to be painful. Can I email in an issue?
--

raman

2017-04-03 15:53:22 UTC

Permalink

David Bjergaard <***@gmail.com> writes:

Hi -- I dont believe it was a commit to stumpwm that started this crash
to happen, I actually suspect some set of ubuntu updates that happened
in mid-February.

Here is why I conjecture the above

1. StumpWM was working fine, then suddenly the crash started happening
i.e. it wasn't a stumpwm update that started this.

2. Crashes started happening mid-Feb 2017, and I was able to reproduce
the crash by both going backwards in stumpwm builds --- including the
now ancient stumpwm available via apt-get -- all the way to ***@head.

3. As mentioned, stumpwm does not have the issue on my desktop
workstation.

4. The crash triggers only if one attaches a function to *message-hook*

Hope this is useful. I'll follow the GitHub Issues page.

Post by David Bjergaard
Hi Raman,
I respect your choice to use a browser that doesn't use javascript, but I have
to warn you that it may make resolving this issue more difficult. I have opened
https://github.com/stumpwm/stumpwm/issues/343
We will track changes there so if you can at least monitor that page for updates
to your bug that would be great.
The next step in narrowing this down is to use git to track down the commit that
1. the "Error opening /dev/tty" is strange, I can't find anywhere in the
backtrace for "echo" that needs a pseuod-terminal
2. The obno-popup-notifier is unfamiliar to me, I haven't seen those types of
errors before. Did Ubuntu recently change the notification handler?
Is it possible to get a backtrace when the crash happens?
If not, please set up the debugging output to a file and see what is in the log
when the crash happens.
Revert the commits until you find a version of stumpwm that doesn't crash using
your recipe.
David