2.6.33-rc8 test: Good News

I have compiled the Linux 2.6.33-rc8 kernel and tested it on the Acer Aspire One and the Gecko Edubook baby laptops.

I configured the kernel for uniprocessor, as I have a lingering suspicion that coniguring with SMP support may be causing problems on some uniprocessor systems. I disabled the "rfkill subsystem" as that seems to be the cause of broken Atheros drivers on both my main Acer laptop and the Aspire One. A few other small changes, including making ps/2 mouse as a module rather than builtin.

Aspire One 532h
This is my recent acquisition (2 days ago). Very pleasant surprise, everything works. I didn't have to specify any "acpi=" kernel boot parameter. The ath9k wireless driver worked. It powered off.
Everything works, except for a couple of issues:

1. Screen too bright. Anyone know how I can reduce the brightness?
2. Shutdown without asking to create a 'pupsave'.

Regarding that second one, I do recall some reports about that on the forum, but it never happened to me. Now that it has, I should be able to fix it -- will take a look at that tonight.

Gecko Edubook
Oh wonderful, it powered off! Still no wireless though -- this kernel does have a 'rl8192s_usb' module, same as what I compiled separately in earlier testing on the Gecko -- however this module does not work.
This is good, it is the stock standard kernel that I intend to use in future Quirky builds, don't need to do anything special for it to run on the Gecko -- oh, except that I forgot to apply the patch for sound.

When the 2.6.33 kernel is released, I'll compile it again, and will apply the patch so that sound will work in the Gecko. I'll also compile the 3rd party wireless driver.

Posted on 14 Feb 2010, 15:47


Posted on 14 Feb 2010, 16:49 by BarryK
Screen brightness
Quirky has the 'xgamma' utility, and I found that works nicely to adjust the brightness. I searched the forum to see if anyone has created a GUI for xgamma...

mikeb posted some info about xgamma:

Ah hah, PANZERKOPF has created a GUI:

...thanks for that PANZERKOPF, I'll put it into Woof.

Posted on 14 Feb 2010, 17:31 by pakt
rl8192s_usb firmware?
Barry, if I remember correctly, rl8192s_usb needs firmware. Could it be missing?

Posted on 14 Feb 2010, 17:49 by pakt
More on Realtek 0bda:8171 WiFi
Found this link with clues for making the Realtek (0bda:8171) chipset work, but they may only be relevant to kernel


Posted on 14 Feb 2010, 19:18 by BarryK
Yeah I know it needs firmware, that was present.

When I executed "modprobe r8192s_usb" it loads, but the wlan interface does not become available.

Oh one other thing... I have downloaded the Windows driver, so I'll see if ndiswrapper works. Except that I have not yet compiled the ndiswrapper module.

Ah, I just realised why the r8192s_usb module might not have worked. It needs that patch, thanks for that. The firmware is in a tarball in /lib/modules/all-firmware, but that does not get unpacked just by executing "modprobe r8192s_usb" -- if I had unpacked it, the driver probably would have worked.

I'll grab the 2.6.33 kernel as soon as it is released, and apply the wireless and sound patches for the Gecko, and will build a Quirky 007 that should work very nicely on the Gecko.

I'm wondering about the battery monitor, whether we could cheat somehow, and just display an estimated percentage charge. /proc/apm does show when the external power is applied, so it is possible to theoretically estimate how much charge the battery is getting. Then when running without external power it is possible to calculate theoretically what the percentage charge is likely to be.

Posted on 14 Feb 2010, 20:29 by pakt
Monitoring battery charge...
The Gecko power light turns from green to red when the batteries are low.

I wonder if the LED is connected directly to the power module or if it controlled by the CPU.

If it is controlled by the CPU, its change in state from green to red might somehow be monitored.

Just a thought...

Posted on 14 Feb 2010, 20:58 by tronkel
So it looks as if this Quirky will be the only OS that actually functions properly with the Edubook.

I hope Northtec will appreciate this fact and maybe make a contribution in kind.

I wonder if touchscreen for the Edubook is the way to go from here. I believe touchscreen has a module already in the kernel

Give it a title? EduQuirky? GeckoQuirky? SuperQuirky? PowerQuirky?

Posted on 14 Feb 2010, 22:59 by teddog
dim ACER screen
press and hold the left side Fn key, and the left arrow

Posted on 15 Feb 2010, 3:36 by BarryK
Acer One pupsave
Ah, I found why it is shutting down without asking to create a 'pupsave'.

My new busybox package has /sbin/reboot and /sbin/poweroff, and they overwrite scripts of the same names. Fixed.

Anyone reading this who knows something about what 'acpid' does?

My new busybox has 'acpid' which I have disabled as Quirky has an acpid package provided by shinobar.

I see that that package was created by Patriot for 4.1.1. It has a start/stop script /etc/init.d/rc.acpi for starting/stopping acpid, but what puzzles me is /root/Startup/acpi which has this in it:


[ -x /etc/init.d/rc.acpi ] || exit
/etc/init.d/rc.acpi stop
sleep 1
/etc/init.d/rc.acpi start

Then, I'm wondering what functionality acpid is providing for my Aspire One.

Posted on 15 Feb 2010, 4:53 by BarryK
acpi power button
Ah, I found Patriot's post:


Patriot created the package to enable power-off by pressing the laptop power button. Yes, that is useful.

Patriot does not have that script in /root/Startup, so that is something shinobar has added.

Posted on 15 Feb 2010, 6:49 by Bm
I use the Lin'N'Win method for my frugal installs.
Some time ago I discovered that when on the first run and having this session not saved issue, I had to with psubdir=pup-xxx
I got the panel then to create a pup_save file.
On subsequent boots, I don't use the psubdir=pup-xxx anymore and the save_faile is updated without problem.

Posted on 15 Feb 2010, 15:03 by tempestuous
acpid is a daemon which simply reads ACPI events.
Often there's some other kernel module involved (hardware-specific) which exposes the ACPI events.

Obviously these events are then useful only when combined with relevant acpid scripts to respond to the ACPI events.
And since ACPI events are not standardised, these scripts are usually hardware-specific.

I packaged a set of Eee-specific acpid-hotkey scripts in October 2008. jemimah has recently enhanced these.

Posted on 16 Feb 2010, 18:45 by Anonymous Coward
Enable MMX in kernel?
Barry, if you could, could you enable MMX when building the kernel for the next few distros? Hardinfo and other diagnostics software on Windows have reported that the Edubook's processor supports MMX, and it would be interesting to see if Puppy runs faster as a result of enabling this..

Posted on 16 Feb 2010, 19:47 by BarryK
I have always built the kernel to support mmx.

Posted on 16 Feb 2010, 21:54 by Anonymous Coward