Re: [systemd-devel] Missing signal / dbus-fd while waiting for dbus-method response

2018-10-22 Thread Jan Mueller
Hello folks,
we’re facing the following issue:
An application does not receive DBUS signals via the DBUS file descriptor while 
waiting for a dbus-method response.
The basic setup: Two applications are communicating via DBUS, using a select() 
loop waiting for events on the dbus-fd.
The faulty procedure:

-   Triggered by a timer event, Application A calls the dbus-method 
“transmit” of application B.

-   During processing of the “transmit”-method, application B broadcasts a 
dbus-signal on which application A is matched on.

-   Afterwards, application B sleeps for 5ms and then finishes its 
“transmit”-method by sending a response message towards application A.

-   The dbus-fd of Application A does not trigger the select() afterwards 
(same for epoll() if implemented so).
Though …

o   … when triggering sd_bus_process() manually, the signal is being receved in 
application A, meaning that the signal was in fact received and queued, but the 
event was not signalled.

o   … other processes, not being “blocked” by waiting for a response-message, 
do also receive the signal.

We saw this behavior on multiple platforms, e.g. Linux kubuntu 
4.4.0-137-generic with 229-4ubuntu21.1 and Debian Buster 4.18.0-2-am64 with 
systemd 239-10.
Attached you can find a simple example, including application A / B, a makefile 
and the correlating *.conf-files.

Thanks in advance and best regards,
Jan Mueller
<>
___
systemd-devel mailing list
systemd-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/systemd-devel


Re: [systemd-devel] Strange problem: Failed to get properties: Failed to activate service

2018-10-22 Thread Jan Synacek
On Thu, Oct 18, 2018 at 3:29 PM Cecil Westerhof  wrote:
> I found out that in reality no reboot had taken place. For one reason or 
> another that did not work. Doing:
> systemctl reboot
> systemctl poweroff
> init 0
> halt
>
> All did not work. Those gave messages like:
> Transport endpoint is not connected

This is very familiar to me. Try this:

1) Boot the computer.
2) Run "systemctl daemon-reexec" as root.
3) Run "dmesg" as root.
4) Look for a pid 1 segfault or a message that says "Freezing execution".

If you see a SIGSEGV, let me know. Note that this problem is *not*
present in the upstream code base.

Cheers,
-- 
Jan Synacek
Software Engineer, Red Hat
___
systemd-devel mailing list
systemd-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/systemd-devel