[ltp] Re: [PATCH] 2.6.17: Unload disks heads before powering down

David Abrahams linux-thinkpad@linux-thinkpad.org
Sat, 23 Sep 2006 10:32:58 -0400


Henrique de Moraes Holschuh <hmh@hmh.eng.br> writes:

> On Fri, 22 Sep 2006, Shem Multinymous wrote:
>> On 9/22/06, Henrique de Moraes Holschuh <hmh@hmh.eng.br> wrote:
>> >Normal unloads are controlled, and do a lot less damage (as in "normal
>> >wear") to the head assembly and arms.  Hitachi is very explicit that
>> >emergency unloads are 100 times more stressful to their hardware than
>> >regular unloads, for example.  Given that you can actually *hear* the
>> >difference quite easily between the two unloads...
>> 
>> Ah, I didn't know that (got a reference by any chance?). It would be
>> well to push this for mainline acceptance, then; and to mention it on
>> the relevant ThinkWiki pages.
>
> Not to mention that *every* linux distro took high pains to cover up for the
> kernel lack of proper handling and tries to unload disk heads in userspace
> (look at the code for the halt binary).  But that is impossible in S4 sleep,
> and doesn't quite always work in some scenarios even for a simple shutdown.
>
> BTW: I know 2.6.17+suspend2 did not do things right for S4.  But the normal
> S3 *does* things right, so chances are 2.6.18 will do things right on S4
> too, even with suspend2.  S5 (power off) is *not* doing things right in
> 2.6.18 and all previous kernels: I have tested it and I could hear the
> "CLICK!" of the emergency unload quite loud and clear.

Is that true even for 2.6.17+SUSE patches?  I guess I hope so, because
I don't want any ambiguity about what kernel I should use ;-).

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com