Re: HD spindown and fs cache writeback

[david.gardner@philips.com]

>Heather <star@betelgeuse.starshine.org> writes:
>> > Any ideas?  I would love to get this working again with a 2.2.15 or
>> > 2.3.x/2.4.x kernel, as it made a fair difference to the battery life
>> > and a large difference to the noise generated on some machines...
>>
>> Ok, I've been looking at different (past) flavors of this same thread
>> recently trying to figure out the real root of the problem.
>
>In my experience, there is something with more recent kernels which
>makes disks spin up.  Without any other changes in my set-up, 2.0.36
>allowed my disk to spin down and 2.2.13 did not.  At least one person
>investigated this further and found (I think) that 2.2.10 works for
>him but 2.2.11 doesn't.
>
>Note:  I have also heard that some people are able to keep their disks
>spun down with more recent kernels, and these people sometimes insist
>that I have a config problem not a kernel problem.  But, as I said,
>just changing the kernel causes the problem, and I've looked into
>various other fixes without success.
>
>I hope this information helps someone.

I also have had this problem since upgrading my kernel (a while ago).  I
will try with 2.2.10 today and see what happens, but I remember reading 
that it is because the bdflush daemon has been integrated into the kernel
instead of being a user-space daemon.*  And on these newer kernels, I find 
that mobile-update now just dies after calling the bdflush syscall the 
first time..  So the solution of setting noatime and running mobile-update
doesn't work for me...

[* This happened between 2.2.10 and 2.2.11, hence mobile-update breaking?]

I haven't yet tried noflushd, but if that does the same job with > 2.2.10
kernels then I'll use that instead.

<wanders off to try noflushd>


Btw, wasn't bdflush also known as `update'?  And mobile-update replaced
this, which isn't really possible now that it is a kernel thread?


	Dave