lightdm set SIGPIPE disposition to SIG_IGN even for children

[Yves-Alexis Perez]

Hi,

a Debian user reported that his session was running with SIGPIPE set to SIG_IGN, and tracked the issue to lightdm.

The whole investigation is in the downstream bug, but in summary the glib gio functions set the disposition, and nothing reverts it. Using g_spawn_*() function should do it, but they're apparently not used in lightdm.

The reporter also provided a minimal patch, which is attached.

[Yves-Alexis Perez (corsac)]

Hi,

is there some news on this? Can you at least comment on the patch even without integrating it, so I can move forward and include it (or not) in the Debian package?

[Robert Ancell (robert-ancell)]

Hi,

Sorry for the delay. I'm not sure where the downstream bug is but looking through the GLib source code I can confirm the GSocket code does set SIGPIPE to SIG_IGN. So the attached patch will correctly revert it to the default for the child processes. As far as I can tell this is the right thing to do.

[Robert Ancell (robert-ancell)]

http://stackoverflow.com/questions/8369506/why-does-sigpipe-exist has a good description of why this is a stupid signal you don't want to handle...

[Robert Ancell (robert-ancell)]

Commited this patch with the following changes:
- I explicitly disable SIGPIPE when LightDM starts (because I don't want it to occur anywhere).
- I re-enable it just before I exec() the children so it can't affect any of the LightDM code.

[Ubuntu Foundations Team Bug Bot (crichton)]

The attachment "reset SIGPIPE to SIG_DFL after forking" seems to be a patch. If it isn't, please remove the "patch" flag from the attachment, remove the "patch" tag, and if you are a member of the ~ubuntu-reviewers, unsubscribe the team.

[This is an automated message performed by a Launchpad user owned by ~brian-murray, for any issues please contact him.]