[ltp] Re: Applying undervolting patches

[David Abrahams]

Andrew Barr <andrew.james.barr@gmail.com> writes:

> On Sun, 2006-08-20 at 07:49 -0400, David Abrahams wrote:
>> How has it been proceeding, specifically?
>
> I would say this: for me at least, it has been getting better with each
> release, especially recently. Back in the days of the single-digit 2.6
> releases things were a bit rough, but these days staying with the latest
> stable 2.6 release hasn't been a problem for me.
>
> Heck, I've been running 2.6.18-rc4 for a week now without issues.

Missing context: The following is from Scott J. Henson.

>> > Also, back porting kernels is kind of a project due to the
>> > changes in the kernel build structure from release to
>> > release.  Ive back ported kernels before, and I must say
>> > that it is a pain.
>
> I'm confused as to what you mean by "backporting".
>
> If you mean drivers and patches, the real solution to that is to
> find a way to minimize the number of extra drivers and patches that
> you need so that you aren't backporting out-of-tree stuff (e.g. use
> uswusp instead of suspend2)
>
> Otherwise, any modern distribution like Ubuntu Dapper should be able
> to run any available 2.6 kernel. The userspace changes required of
> distributions largely happened over the 2.4->2.6 transition.

Well, for me switching kernels has been nightmarish.  I'm surely
missing something very simple, but my wireless card has disappeared,
adn I can't seem to keep the proprietary ATI driver (8.28) working
under both the old and the new kernels.


Missing context: Scott Henderson wrote:

>>> In my opinion you may be better served by waiting the 2
>>> months for edgy to come out.  I'm not sure how much battery
>>> life you're expecting to gain, 
>>>
>> From what I've read, it can be substantial.
>
> It can be, depending on how low you can go with your
> undervolting. In my case, at 600 MHz (lowest frequency) I was able
> to go all the way down to 700 mV, which is the lowest you can go.

Are you talking about substantial gains due to undervolting, or due to
a kernel version upgrade?  Scott was suggesting that the kernel
version wouldn't make much impact.

BTW, how do you decide how low you can drive the voltages?

> The machine also runs cooler too as I mentioned previously.
>
>> Anyway, I still have the edgy kernel sources, so I should be able to
>> do that build here and apply the undervolting patches to that.
>
> Maybe, maybe not. 

Well, they *were* tested against edgy according to the linux-phc
site.  But right now I'm conservatively using the dapper kernel with
patches.

> Like I said, due to the way Ubuntu packages their kernel sources, it
> is not simple to tell what they are actually changing. 
> So you just have to try and see if it works.
>
> You might have better luck too using Debian linux-image and
> linux-source packages, even on Ubuntu, as they are very close to
> upstream.

Why would I have better luck with that?

>> I believe the HOWTOs I've found are all using it.
>
> In my opinion there is no reason to use kernel-package unless you wish
> to share the same kernel image amongst multiple machines, or otherwise
> wish to distribute the image. Installing from a vanilla source tree is
> simple:
>
> make [oldconfig|xconfig|gconfig]
> make
> make install
> make modules_install
> update-initramfs -t -u -k [kernelversion]
> (Be sure you don't have any symlinks to the new kernel files in /boot,
> otherwise you will get duplicate GRUB menu entries)
> update-grub
> (reboot)
>
> Best yet, it's quicker (it does not package documentation and source
> code) and the intermediate object files are preserved so that if you
> make any changes (configuration, patching) the build will only have to
> rebuild affected files.

Good to know, thanks.

>> > I haven't used it since my old
>> > days in Debian and initial forays into Ubuntu, but I would
>> > expect it to work still.  I used to follow the bleeding edge
>> > on kernels to get the latest and greatest, but Ive grown
>> > mundane in my old age and count stability and not having to
>> > waste time building kernels over any perceived gain on the
>> > bleeding edge.
>
> Bleeding edge is -rcX releases or git snapshots (of which the current
> version is working fine for me). The 2.6.xx[.y] releases are called
> "stable" for a reason.

I have a vague memory that odd and even xx numbers have different
meanings (?)

> I have a 2.6.17 image from Debian official on my boot menu should I
> start having problems.
>
>> Ahem.  When you reach true old age, like me, you may find yourself
>> getting fussy about how hot your laptop gets, how quietly it can run,
>> etc.  I bought this machine because there were too many annoying
>> things about a MacBook pro.  I need to make it work better than one of
>> those or it will all be for naught.  So even if I don't end up using
>> the new kernel much, it's important for me to know now how well it
>> *can* work.
>
> You will find that the possibilities with your machine in Linux only get
> better with time. Two or three years ago I was fussing with all kinds of
> ACPI bugs and misfeatures (some of which were due to a cruddy Dell BIOS)
> but those days are long over. Linux has come a long way and isn't
> slowing down.

So far my impression is that although there are ways to bend Linux to
your will if you have deep expertise, the system as a whole is
entirely too fragile.  Pull this lever over here, and something over
there breaks... :(

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