New dbus port breaks hald

[Kevin Oberman]

> Date: Wed, 27 Oct 2010 01:34:50 -0400
> From: Joe Marcus Clarke <marcus at freebsd.org>
> 
> On 10/26/10 7:26 PM, Kevin Oberman wrote:
> >> Date: Tue, 26 Oct 2010 16:07:24 -0700
> >> From: "Kevin Oberman" <oberman at es.net>
> >> Sender: owner-freebsd-gnome at freebsd.org
> >>
> >>> Date: Tue, 26 Oct 2010 01:38:24 -0400
> >>> From: Joe Marcus Clarke <marcus at freebsd.org>
> >>>
> >>> On 10/25/10 4:04 PM, Kevin Oberman wrote:
> >>>> This weekend I upgraded dbus from 1.2.24_2 to 1.4.0. The upgrade
> >>>> reported no problems and dbus appears to start up just fine, but hald
> >>>> does not. It exits during initialization with:
> >>>> 11:52:44.493 [I] hald_runner.c:110: runner_server_message_handler: destination=(null) obj_path=/org/freedesktop/DBus/Local interface=org.freedesktop.DBus.Local method=Disconnected
> >>>> 11:52:44.493 [I] hald_runner.c:151: runner process disconnected
> >>>> *** [DIE] hald_runner.c:runner_died():204 : Runner died
> >>>>
> >>>> I have had this happen on at least 10 attempts to start hald, but on ONE
> >>>> occasion, hald started normally, so I suspect a race.
> >>>>
> >>>> I have submitted a PR, ports/151725, which has more detail and I have
> >>>> rolled back to 1.2.24_2. 
> >>>>
> >>>> Any other information I can supply to help track this down?
> >>>
> >>> Works for me.  It looks like hal may be crashing.  Are you seeing
> >>> anything in dmesg?  Are you seeing any core dumps?  What happens if you
> >>> run hald from gdb with --daemon=no?
> >>
> >> I included the tail of the output of 'hald --daemon=no --verbose=yes'.
> >> Looks like it is simply exiting with an error (exit 1) after reporting
> >> that hal-runner had died. I get:
> >>
> >> Run started hald-addon-mouse-sysmouse (0) (0) 
> >> !  full path is '/usr/local/libexec/hald-addon-mouse-sysmouse', program_dir is '/usr/local/libexec'
> >> 15:29:04.198 [I] hald_runner.c:663: running_processes 0x34283dc0, num = 1
> >> 15:29:04.198 [I] hald.c:120: Started addon hald-addon-mouse-sysmouse for udi /org/freedesktop/Hal/devices/psm_0
> >> dumpcis: CIS code read: Inappropriate ioctl for device
> >>
> >> [various device additions elided]
> >>
> >> Run started hald-addon-input (0) (0) 
> >> !  full path is 'hald-addon-input', program_dir is '(null)'
> >> 15:29:04.231 [E] hald_runner.c:676: Error running 'hald-addon-input': org.freedesktop.HalRunner.Failed: Start request failed
> >> 15:29:04.231 [E] hald_dbus.c:3579: Cannot start singleton addon hald-addon-input for udi /org/freedesktop/Hal/devices/atkbd_0
> >>
> >> [Lots of probe results that look "normal" elided]
> >>
> >> 15:29:04.274 [I] hald_runner.c:110: runner_server_message_handler: destination=(null) obj_path=/org/freedesktop/DBus/Local interface=org.freedesktop.DBus.Local method=Disconnected
> >> 15:29:04.274 [I] hald_runner.c:151: runner process disconnected
> >> *** [DIE] hald_runner.c:runner_died():204 : Runner died
> >>
> >> While I don't see anything else that looks even slightly odd, I can
> >> provide the full output.
> > 
> > To follow up my own message, the dumpcis error is occurring on both the
> > old (1.2.24) and new (1.4.0) dbus systems. I will try to get a little
> > time tonight to compare the outputs for REAL diffs, but it is clearly
> > the failure of hal-runner that is causing hald to exit.
> 
> What were ALL of the messages leading up the termination?

As I promised, the full logs of 'hald --daemon=no --verbose=yes' with
dbus-1.2.24 (good) and 1.4.0 (new) are at:
http://home.comcast.net/~k.oberman/FreeBSD/hald.good.log
http://home.comcast.net/~k.oberman/FreeBSD/hald.new.log
-- 
R. Kevin Oberman, Network Engineer
Energy Sciences Network (ESnet)
Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab)
E-mail: oberman at es.net			Phone: +1 510 486-8634
Key fingerprint:059B 2DDF 031C 9BA3 14A4  EADA 927D EBB3 987B 3751