Re: Unresponsive after 'tmpwm' ?

Hi folks,

Antoine Busque <abusque-vg+e7yoeK/dWk0Htik3J/[email protected]> writes:

> On Mon, Nov 16, 2015 at 6:01 PM, Fahri Cihan Demirci
> <[email protected]> wrote:
>> On Mon, Nov 16, 2015 at 12:14:44PM +0100, Andrea Remondini wrote:
>>> Hello! I'm seeing this problem with Ratpoison 1.4.8 on Linux/x86_64:
>>> after a 'tmpwm' command, and after exiting the temporary wm, Ratpoison's
>>> input bar and the 'exec' input bar are unresponsive. No keystrokes are
>>> accepted. Killing Ratpoison seems to be the only way out.
>>>
>>> BTW, if the specified temp wm is inexistent, Ratpoison opens two
>>> unnamed windows that can't be closed. After that, the input bars are
>>> unresponsive.
>>>
>>> Can anyone confirm this behaviour?
>>>
>>> Thanks for your comments,
>>> AR
>>
>> Hello Andrea,
>>
>> For what it's worth, I can't seem to reproduce either behaviour.
>> I tried pekwm and a few tiling window managers in the same session and
>> they all started up normally and delegated the control back to
>> ratpoison upon exit. Inputting a bogus command for tempwm does not
>> seem to cause any problems either.
>>
>> I've tried both the master and 1.4.8 versions on the same platform; 64 bit GNU/Linux. Could the problem be related to the temporary window manager you are trying to run?

64 bits GNU/Linux is a bit broad. :)

> I can reproduce on 1.4.8, installed from the Arch Linux repositories,
> on an x86-64 4.2.5 kernel, with xorg-server 1.17.4. For instance,
> launching openbox (version 3.6.1) via tmpwm and exiting (right-click
> on root, then 'logout') gives the control back to ratpoison, but it
> remains unresponsive. Using a bogus executable as tmpwm argument also
> causes unresponsiveness and creates the two unnamed windows. The only
> way out of both situations seemed to kill X via Ctrl-alt-bksp.

On OpenBSD-current, amd64, I can't reproduce your problem using
ratpoison-1.4.8.  Gotta try on some Linux flavors, do you know if other
distros are affected, aside from Arch?

> I haven't tested on master yet. I know however that the
> unresponsiveness issue when exiting the tmpwm is not new from the
> current stable version, as I've encountered it systematically for a
> while.

For a while, like since ratpoison-1.4.6 in 2013? 'cause I fixed
a very similar problem back then, ratpoison was working nicely on Debian
but tmpwm would always hang on OpenBSD (pkill -HUP helped).  I tested
the fix on both Debian and OpenBSD and both were fine, but things may
have changed since. Could you try the diff below? (ratpoison-1.4.8)

--- src/actions.c.orig	Fri Nov 20 21:36:05 2015
+++ src/actions.c	Fri Nov 20 21:36:05 2015
@@ -4855,7 +4855,7 @@
   XSync (dpy, False);
 
   /* Disable our SIGCHLD handler */
-  set_sig_handler (SIGCHLD, SIG_DFL);
+  set_sig_handler (SIGCHLD, SIG_IGN);
   /* Launch the new WM and wait for it to terminate. */
   pid = spawn (ARG_STRING(0), 0, NULL);
   PRINT_DEBUG (("spawn pid: %d\n", pid));


> I'll try and investigate the issue later this week, if I can reproduce
> it on master.

Hmm, please do, but I doubt you'll find a difference.

Cheers,
-- 
jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF  DDCC 0DFA 74AE 1524 E7EE