Kernel Bug Tracker – Bug 4919
APM resume: freeze at disk access
Last modified: 2007-04-28 12:49:14 UTC
Distribution: Mandriva LE2005
Hardware Environment: Dell Inspiron 8200 + ATI Radeon 9000 Mobility
Software Environment: Kernel 2.6.11-6mdk / 188.8.131.52 / 184.108.40.206, apmd 3.2.2
After suspending and resuming, the console comes back alright, but the computer
freezes at the first physical disk access.
APM suspend & resume works fine with Mandrake 10.0 + kernel 2.6.3-7mdk + apmd 3.1.0.
The same thing seems to happen with Inspiron 8200 + Fedora Core 3 + Nvidia, see
Steps to reproduce:
1) Boot with acpi=off into console
2) Suspend (apm -s or Fn+Esc or close lid)
3) Resume (press powerbutton)
4) Type any command that involves disk access
How strange - APM resume normally does quie a bit of disk I/O anyway, doesn't it?
re you able to be more precise about when the kernel broke? 2.6.3->2.6.12
is rather a large step.
After lots of tests I finally found out that disabling the haldaemon service
solves the problem, at least from console. So it seems that it wasn't a kernel
issue after all. I still experience similar freezes occasionally when suspending
from X but that is probably another issue.
Well the kernel sholdn't go freezing like that. What
is the haldaemon service??
just testing something...
The haldaemon service is new to me too. I cannot find any real documentation but
there is a homepage at <http://www.freedesktop.org/Software/hal>. The following
is from the MandrivaClub knowledge base:
"HAL is a Hardware Abstraction Layer. The purpose of HAL is to allow programs to
interface with common hardware types in a generic manner. It is a vital
component of Mandrakelinux 10.1, Mandriva Linux Limited Edition 2005 and later
releases, and should not be disabled.
Pertinence: Essential. Package: hal. Doc: files in /usr/share/doc/hal-0.4.8."
I've added haldaemon to the services to be stopped at suspend and restarted at
resume. Also, the occasional hangs when suspending from X have stopped after I
set a switch in the suspend script causing the pcmcia card to eject instead of
suspending. It seems to work now, but I should try it out a few more days I think.
Not sure if this is the same problem, but it certainly sounds very similar, so
I'm going to add my case to this bug.
Thinkpad R51 (1529-EHG), Debian Testing, vanilla kernel.org kernels, with and
without ipw2100 driver.
I'm using ACPI, not APM.
Suspend & resume to memory works fine with 2.6.7.
With 2.6.10, 2.6.11 and 2.6.12, on resume, the system hangs dead with the IDE
activity led on solid.
I'm now running 220.127.116.11 with the radeon suspend patch in bug 3022.
After reading this bug report, I added /etc/dbus-1/event.d/20hal stop and start
to my /etc/acpi/lid.sh. This seems to have fixed the problem.
I've been using the above configuration on my Dell Inspiron 8200 for almost a
year, and then upgraded to Mandriva 2006 which was entirely similar. apm suspend
has been working most of the time, but I still experienced occasional hangs.
I'm now running Ubuntu 6.06 Dapper instead. There are no problems with hal
anymore, but I had to write a small console switching script since APM suspend
only works from console for some reason. This was not the case with Mandriva
LE2005. APM suspend has worked flawlessly for a few weeks now but I don't use
this computer much anymore so it's still too early to tell.
It's strange that Wouter Cloetens had the same problem with hal using ACPI
suspend. I've spent some time trying to get ACPI to work with the Inspiron 8200
but the Radeon 9000 Mobility video never comes back properly. None of the
patches or tricks that I have found works properly. But this was only a video
issue, all other things worked well, hal included.
You can try to use the s2ram tool to solve the video-after-resume problem with
the ACPI suspend (see http://en.opensuse.org/S2ram).
*** This bug has been marked as a duplicate of 7225 ***