[Debian-eeepc-devel] Super Hybrid Engine on eeepc 901c with kernel 2.6.31-6 [resolved]

[Shawn Lamson]

On Thu, 24 Dec 2009 13:36:30 +0000
Darren Salt <linux at youmustbejoking.demon.co.uk> wrote:

> I demand that Shawn Lamson may or may not have written...
> 
> > On Tue, 22 Dec 2009 23:21:39 +0000
> > Darren Salt <linux at youmustbejoking.demon.co.uk> wrote:
> > [see below]
> >> I demand that Shawn Lamson may or may not have top-posted...
> >>> Thanks Darren - if I use your config on the 2.6.31.9 source do I also
> >>> need to apply that rt2860 patch (yes, thats the driver that's used)?
> >> No; the firmware loading is specific to the Debian kernel (DFSG-freeness
> >> issues).
> >>> The reason I moved to the 2.6.31-r6 and up was that the reports were that
> >>> this was fixed in those versions (again, it seems to be a problem once
> >>> again in 2.6.32, though)...
> >> A different bug was fixed.
> >> [snip]
> >>> Thanks for the input, let me know about that patch and config, and i'll
> >>> report back.
> >> Which compiler version are you using? I ask mainly because there are
> >> presently reasons to avoid 4.4 if certain kernel debugging options are
> >> enabled.
> 
> >> [snip; you don't top-post, I don't remove context]
> 
> > well... pretty much the same results, unfortunately, with the 2.6.31
> > .config you provided Darren - though I must say it's a sweet piece of work.
> > I had a few ups and downs with it, haven't ironed them out yet - at one
> > point it just wouldn't boot at all (after a crash? after i removed
> > something else?)...
> 
> Which is a bit odd. It worked fine here, as did 2.6.32...
> 
> > but i am keeping my stock 2.6.31-6 kernel for backup...
> 
> > Would you entertain a couple of questions?  I noticed that
> > CONFIG_BLK_DEV_INITRD is not set and at first I thought maybe the kernel
> > was small enough not to use initramfs, but i am not certain what's going on
> > there...
> 
> Everything which I need at startup is built in.
> 
> > I wound up enabling this because i do generate initrd/initramfs images...
> > is this covered somewhere else in the config options or is it ubiquitous
> > now and assumed that initrd/initramfs is going to be used?
> 
> It's not assumed at all; some of us choose not to use them.
> 
> [snip]
> > Also, aside from the situation where it wouldn't boot that kernel, the
> > only other 2 issues i had were
> > a) elantech touchpad mouse wouldn't accpet two-finger clicking (emulate 3rd
> 
> > mouse button) ... I tried several compiles, enabling the
> > # CONFIG_MOUSE_PS2_SYNAPTICS is not set
> 
> Not needed. CONFIG_MOUSE_PS2_ELANTECH=y.
> 
> > and
> > # CONFIG_MOUSE_SYNAPTICS_I2C is not set
> 
> Again, not needed.
> 
> > but this didn't resolve it, I read up on it a bit, and i could try fixing
> > it using some HAL setting or synclient ...
> 
> Well... X/evdev default to not enabling tapping...

I have a steep learning curve, I haven't tinkered that much with kernel/OS in
the past few years, and a lot has changed recently ... I didn't know about HAL.
I was able to get this working (Elantech touchpad) okay.  Originally it had 
been identified as a Logitech mouse and it actually worked fine - the new kernel
config just exposed an existing issue... creating /etc/hal/fdi/policy/touchpad.fdi
as you suggested on http://wiki.debian.org/DebianEeePC/HowTo/ElantechTouchpad 
worked.

> 
> > b) with AC plugged in, and SpeedStep enabled in BIOS, no luck, and if I
> > boot up on battery and then plug in the AC while *not* in suspend-to-ram
> > then likewise the system freezes - but with the config-2.6.31-eee you gave,
> > this also happens with the HyperThreading Technology enabled in BIOS (and
> > SpeedStep off) ... so it was actually technically a little worse in that
> > regard.
> 
> Hmm. Again, works fine here...
> 
> > I'm going to play around with a few more .config tweaks, maybe using the
> > 2.6.32 kernel source*, but if you can advise on the Elantech "two finger
> > tap" issue that would probably make your .config and the 2.6.31-6 stock
> > kernel equally functional and since yours is more elegant I'd settle on it.
> 
> http://wiki.debian.org/DebianEeePC/HowTo/ElantechTouchpad has enough
> information.
> 
> > (yours takes under 20 minutes to compile - the ones i come up with take
> > several hours, or else result in something not bootable!)
> 
> Mine has no support for hardware which cannot be present, and support for
> only that extra hardware which I'm likely to plug in.
> 
> > * last time I tried 2.6.32 it was a no-go on suspend-to-ram -> resume but
> > maybe things have changed in a week or so... i do use the rt2860 driver and
> > i keep seeing "back and forth"  on it in the various changelogs and bug
> > reports...
> 
> One thing which you should try: in /etc/default/eeepc-acpi-scripts, set
> PWR_CLOCK_AC=1. This *shouldn't* make any difference, but it may well end up
> making things work normally again.

setting PWR_CLOCK_AC=1 in eeepc-acpi-scripts did the trick, i've been testing
it for the last few days and it's unshakeable now, thanks, that was a long
road.

is there another mailing list (debian eeepc) for "general questions" that are
less development related?  I see there is an IRC channel, I guess I'll lurk
there.

Thanks again Darren - I hope Santa was good to you, and have a happy New Year.

Shawn

> 
> Trying
>   # echo 0 >/sys/devices/platform/eeepc/cpufv
> on a "known-good" OS is probably also worth doing.
> 
> (I'm still thinking "bad hardware".)
> 
> > Merry Christmas!
> 
> Humbug! ;-)
> 
> > P.S. bonus round ... i compile the kernel on my desktop PC, then scp the
> > .deb to the netbook and dpkg -i it there... here is what i get:
> 
> > netbook:~# dpkg -i
> /home/slamson/linux-image-2.6.31.9-eee901_2.6.31.9-eee901-10.00.Custom_i386.deb
> [snip]
> > Running postinst hook script update-grub.
> > Generating grub.cfg ...
> > Found background image: moreblue-orbit-grub.png
> > Found linux image: /boot/vmlinuz-2.6.31.9-eee901
> > Found linux image: /boot/vmlinuz-2.6.31.6
> > Found initrd image: /boot/initrd.img-2.6.31.6
> > done
> 
> > what's the deal with the /lib/modules directory?
> 
> Buy one, get one free? Guaranteed free of symlink farms, or your money back?
> 
> > and why do you think initramfs-tools fails? (see the "Found linux image"
> > statements, missing initrd.img-2.6.31.9-eee901 ... but if i run
> > update-initramfs -c -k 2.6.31.9-eee901 it works fine (in fact i got the
> > commands by tracking down the initramfs-tools file!
> 
> I don't see any failure there.
> 

I misundertood what was intended by /etc/kernel/postinst.d/initramfs-tools

# kernel-package passes an extra arg; hack to not run under kernel-package
[ -z "$2" ] || exit 0

so it's not intended to generate an initramfs image when you run dpkg... 

-- 
Shawn Lamson
shawn.lamson at gmail.com