[pkg-wpa-devel] Bug#753345: wpasupplicant hook for pm-utils breaks pm-suspend when used with wicd

[Stefan Lippers-Hollmann]

reassign 753345 wicd-daemon
notfound 753345 1.1-1
notfound 753345 1.7.0+ds1-5+squeeze3
found 753345 1.7.2.4-4
found 753345 1.7.2.4-4.1
thanks

Hi

On Monday 30 June 2014, Stefan Lippers-Hollmann wrote:
> Hi
> 
> [ CC'ing the wicd maintainers ]
> 
> On Monday 30 June 2014, fzacarias3k at gmail.com wrote:
> > Package: wpasupplicant
> > Version: 1.1-1
> > Severity: important
> > 
> > Dear Maintainer,
> > 
> > After installing wicd and wpasupplicant, suspending the computer
> > with pm-suspend stops working.
> > The problem lies in the wicd and wpasupplicant installed hooks for
> > pm-utils colliding.
> > wicd's hook runs first and disconnects my wireless device, causing
> > wpa_supplicant process to stop. Then the wpasupplicant's hook runs and
> > tries to execute the suspend command but the hook fails because wpa_cli
> > can't find the control socket (because it was removed by the previous
> > hook!).
> > Extract from /var/log/pm-suspend.log:
> > ---
> > Running hook /usr/lib/pm-utils/sleep.d/60_wpa_supplicant suspend
> > suspend:
> > Failed to connect to wpa_supplicant - wpa_ctrl_open: No such file or
> > directory
> > /usr/lib/pm-utils/sleep.d/60_wpa_supplicant suspend suspend: Returned
> > exit code 255.
> > 
> > Mon Jun 30 20:03:15 CEST 2014: Inhibit found, will not perform suspend
> > Mon Jun 30 20:03:15 CEST 2014: Running hooks for resume
> > ---
> > 
> > This can be reproduced every time. I temporarily fixed it by appending
> > "exit 0" to wpasupplicant's hook.
> > Better solutions might be:
> >   * run this hook before wicd's
> >   * verify wpa_supplicant's control socket existance
> > 
> > -- System Information:
> > Debian Release: jessie/sid
> >   APT prefers testing
> >   APT policy: (500, 'testing')
> [...]
> 
> Unless there is a reason to run wicd's pm-utils hook at 55, rather 
> than >= 61[1], I'd tend to reassign this bug to wicd. The reason being
> that wicd is the leaf package here, while changing the order for
> wpasupplicant might introduce subtile ordering problems with the other 
> frontends that are possible to use with wpasupplicant, like ifupdown,
> network-manager, networkd (systemd >= 209), connman, et al.

As indicated above, I consider this (potential) ordering issue to be a
bug in wicd rather than wpasupplicant. Especially as changing it in the
leaf package, wicd-daemon, can be done without risking regressions in 
the alternative frontends to wpasupplicant.

Given that this has apparently gone unnoticed since the wheezy 
development cycle and upower probably being the dominant option for
suspending, I'd suggest lowering the severity of this bug - but I leave
that to the wicd maintainers.
 
> [1]	according to pm-action(8), the 50-74 range seems to be recommended
> 	for these kinds of services.
 
Regards
	Stefan Lippers-Hollmann
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.alioth.debian.org/pipermail/pkg-wpa-devel/attachments/20140720/b78ca716/attachment-0001.sig>