Hald won't start after latest updates

Bug #43962 reported by Matthew Flower
42
This bug affects 1 person
Affects Status Importance Assigned to Milestone
hal (Debian)
Fix Released
Unknown
hal (Ubuntu)
Fix Released
Medium
Martin Pitt

Bug Description

I installed tonight's batch of updates, at the end of which Ubuntu requested a reboot.

Once I rebooted, received the familiar splash screen of Ubuntu. Once the startup got to the "Starting Hardware Abstraction Layer Hald", I ran into some trouble. I received the message "select() to /dev/rtc to wait for clock tick times out"

Once I got into X, I got an internal error about Hald not being able to start once I logged in.

I attempted to roll back to an earlier kernel (2.6.15-20-386). It seemed to have the same behavior.

I loaded system rescue cd, and I noticed that after it booted, I was able to run "date" and see the correct time. I'm assuming it fetched this from the hardware clock, which seems to indicate that my hardware clock is okay.

I googled around a bit, and I found some advice about setting HWCLOCKPARS="-r --directisa" in /etc/init.d/hwclockfirst.sh. Neither this, nor settings it to "--directisa" seemed to help.

Thanks for any workaround you can offer as you research a bug fix.

Revision history for this message
JasonNorwoodYoung (jason-tectonic) wrote :

Similar problem. The error message doesn't come up if I wait about 10 minutes before logging in.

~$ sudo /etc/init.d/dbus restart
 * Stopping NetworkManager dispatcher [ ok ]
 * Stopping NetworkManager daemon [ ok ]
 * Stopping DHCP client manager... [ ok ]
 * Stopping Hardware abstraction layer hald [ ok ]
 * Stopping system message bus dbus [ ok ]
 * Starting system message bus dbus [ ok ]
 * Starting Hardware abstraction layer hald
<Here it sits for a couple of minutes, so I guess that's where the problem is> [ ok ]
 * Starting DHCP client manager... [ ok ]
 * Starting NetworkManager daemon [ ok ]
 * Starting NetworkManager dispatcher [ ok ]

During the restart, gnome-panel stops responding. killall gnome-panel gets it going again, but it takes about a minute. Gnome-panel is also v slow on log-in, even if I wait for a while before log-in.

Revision history for this message
Brandon Holtsclaw (imbrandon) wrote :

same issues as the above two entries, seems something in the hald last evening

Revision history for this message
Simon Law (sfllaw) wrote :

Hi Martin,

There appears to be some useful debugging information in the
associated Debian bug.

Perhaps we got this behaviour when we pulled in the endian
probing patch?

Changed in hal:
status: Unconfirmed → Confirmed
Revision history for this message
Martin Pitt (pitti) wrote :

I'll talk with Sjoerd to debug this. Unfortunately I cannot reproduce it myself.

Changed in hal:
assignee: nobody → pitti
Revision history for this message
Matthew Flower (mattflower) wrote : Hal works after new update

Last night's batch of updates included hal, once I installed that update everything seems to be fixed. I'm not sure about anyone else who chimed in, but the bug can be closed for me.

Revision history for this message
Martin Pitt (pitti) wrote :

Brandon, Jason, do you confirm/deny that it works now, too?

Revision history for this message
Dennis Laumen (dennislaumen) wrote :

My issues aren't fixed yet.

I posted a dupe here: https://launchpad.net/distros/ubuntu/+source/hal/+bug/44091

With all updates installed I still have the same issues. Maybe mine is a different issue and shouldn't have been duped.

Revision history for this message
bartman (bart-jukie) wrote :

My issues are now 1/2 fixed.

I used to get an unusable system -- starting hald would kill all getties and X sessions.

After upgrading, it just runs for a while, and then dies. I am running dapper/amd64 and I see the same output as Dennis.

Revision history for this message
bartman (bart-jukie) wrote :

Actually, after restarting dbus, it seems hald seems to stay up. Previously I ran it manually using --daemon=no.

Revision history for this message
JasonNorwoodYoung (jason-tectonic) wrote :

All working - thanks guys. Had a knock-on effect that gnome-panel is starting as it should. Also file browsing was slow to start and that's fixed too.

Revision history for this message
Matthew Flower (mattflower) wrote :

I think I was incorrect when I said that my problems were fixed. Although I didn't seem to get the error when I restarted my computer this morning, I have gotten the problems while starting up twice this evening.

Additionally, I am getting the "Internal error - failed to initialize HAL" error after I log in and while Gnome is starting up.

Revision history for this message
Martin Pitt (pitti) wrote :

I don't believe that the Debian bug and this bug are about the same issue. In the debian bug, the hald-runner is crashing, here the root cause seems to be the input probing endianess fix. Can anyone please try this:

  sudo apt-get update
  sudo apt-get build-dep hal
  apt-get source hal
  cd hal-0.5.7
  rm debian/patches/13_probe_input_endianess.patch
  debuild -us -uc -b
  sudo dpkg -i ../hal_*.deb

and then reboot and check if it works again?

Revision history for this message
Martin Pitt (pitti) wrote :

I stared at that patch again, and it indeed looks fishy now (incompatibility to the ioctl). I backed out the patch and will discuss this with the author and upstream.

hal (0.5.7-1ubuntu15) dapper; urgency=low
 .
   * Fix name of 14_prove_volume_invalidlabel.patch -> s/prove/probe/.
   * Disable debian/patches/13_probe_input_endianess.patch for now; it is
     incorrect (the ioctl expects chars, not long) and causes crashes.
     Closes: LP#43962 (and half a dozen dups)

Changed in hal:
status: Confirmed → Fix Released
Changed in hal:
status: Unconfirmed → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.