Bug 243021

Eck! Not good, got this here too. kernel-2.6.21-1.3194.fc7 just told the
notebook to power down and simply yanked the plug on the hard disc. I've an
Intel 82801DBM (ICH4-M) controller using the ata_piix driver connected to a
FUJITSU MHT2060AT. I'm switching back to a kernel using the old IDE code in the
meantime.

This may not strictly be a kernel issue: see http://linux-ata.org/shutdown.html
; possibly a problem in sysvinit or initscripts?

I think you are right. I'm changing the component for this bug to SysVInit
(which contains /sbin/shutdown)

Understood. Although reading through the kernel Bugzilla
http://bugzilla.kernel.org/show_bug.cgi?id=7838 it seems that the 2.6.22 release
upstream will have stuff to fix this as well. I'm going to try the suggested
patch and see what happens.

  Roughly how many emergency parkings are desktop and/or laptop HDs rated for? 
(I have a desktop.)  Just to get a sense of how fast this needs to be fixed.

Please be fast :-)

Dang, tried patch, didn't work (apparently doesn't cover PATA). Should've read
the notes more carefully! 

Can anybody raise the priority of this bug from "low"?

I'm setting the component back to "kernel", as it's where the problem is
_happening_.
Doesn't matter if it can be worked around in userspace, kernel or both. 
I looked at the kernel bugzilla and the bug (7838) is marked as
resolved_code_fix, though I can't find the patch itself. (only the scsi one, but
not the libata-specific one).
I've looked in git (scsi-misc-2.6) too, but I can't find anything related to
this to test.

PS: this is *NOT* a forum, so add a comment only if you know something new. 
If you just need help or something use http://www.fedoraforum.org/
Comments like "be fast" just spams the developers which are fixing bugs,
including this one!

I found some patches for 2.6.21 to solve this issue:
http://marc.info/?l=linux-ide&m=118047865916766&w=2

kernel-2.6.21-1.3228.fc7 still have the bug.

Nice find, Stefano. Those patches worked for me (patched against vanilla
2.6.21.4 and 2.6.21.5).

Can't someone from redhat implement these changes in Fedora?

http://linux-ata.org/shutdown.html

This also happens when hibernating. 

On my Toshiba Satellite A200 I get a loud crack noise from disk heads, which
clearly means that hardware is suffering. 

This could produce hard disk failure, just out of software misbehaviour. We'd be
so glad to see a fix.

I think this bug should be marked as a duplicate of Bug 242799.

They are related but not the same. It would really be nice if we could get
someone to step up an fix this problem since it been known about for a couple of
months.

Kernel 2.6.21-1.3255 is in testing now with patches to fix this. Please test.

(Update kernels are in the updates-testing repository.)

As root:

  yum --enablerepo=updates-testing update kernel

  A message just appeared in fedora-package-announce notifying of a test update
for this kernel.  Normally, this mailing list is only for updates-released.  Is
this a mistake, or has it actually been released?

https://www.redhat.com/archives/fedora-package-announce/2007-July/msg00156.html

(In reply to comment #19)
>   A message just appeared in fedora-package-announce notifying of a test update
> for this kernel.  Normally, this mailing list is only for updates-released.  Is
> this a mistake, or has it actually been released?
> 

That was a glitch in the announcement processing. The kernel is in testing.

[root@joker4 ~]# yum --enablerepo=updates-testing update kernel
Loading "installonlyn" plugin
Setting up Update Process
Could not find update match for kernel
No Packages marked for Update/Obsoletion

current kernel is:
[root@joker4 ~]# uname -r
2.6.21-1.3228.fc7

Kernel may not have mirrored out yet.

I found the i386 "normal" kernel here, so it is released:
ftp://download.fedora.redhat.com/pub/fedora/linux/updates/testing/7/i386/kernel-2.6.21-1.3255.fc7.i686.rpm

In my testing it fixed my problem. I tried both a shutdown -r now and a shutdown
-h now - and my 
192 Power-Off_Retract_Count 0x0032   100   100   000    Old_age   Always       -
      87 - stayed at 87.
I did a smartctl to find out what it was. Then did a shutdown -r now. Checked
and it was still 87 - I then did a shutdown -h now. Powered back on and checked
again and it was still 87. In the past it had always incremented , plus made a
loud click.

Now I haven't ran the kernel at all but it fixed this problem for me.

This testing kernel fixed the problem for me. My Power-Off_Retract_Count remains
now the same as stated above (comment #23) and the click is not heard anymore.

I have to run the testing kernel (if I run the current one, I still get the clicks)

Thanks very much

Hey where did this kernel go 3255 - i have another system with the same problem
but this kernel is
gone!?ftp://download.fedora.redhat.com/pub/fedora/linux/updates/testing/7/i386/kernel-2.6.21-1.3255.fc7.i686.rpm

It's now :
ftp://download.fedora.redhat.com/pub/fedora/linux/updates/testing/7/i386/kernel-2.6.22.1-20.fc7.i686.rpm

2.6.22.1-20.fc7 fixes the problem on Compaq Presario 900

  The kernel-2.6.22.1-27.fc7 in updates-released fixes it for me on a Dell
Dimension B110.

The kernel-2.6.22.1-27.fc7 in updates-released fixes it on a Toshiba Satellite A200

It may be a good idea to add Fedora to the list of fixed distros here (none as
of yet):

http://linux-ata.org/shutdown.html

2.6.22.1-27.fc7 fixes it for me too on my Asus A6km.
I think I can close this bug now.