Gentoo Linux on D945GCLF with Atom 230
Posted by Akom • Friday, September 26. 2008 • Category: ReviewsI decided to downscale my asterisk server (currently Athlon XP2000) to something that's not quite as power hungry. The server barely does anything most of the time, but consumes just over 100 watts.
$70 later I was staring at this very cute little green square, complete with a chipset fat/heatsink and a very naked CPU (needs no cooling at 5 watts). It looks hilarious in my small tower case, hides entirely behind the vertical ATX power supply (I realize I can run this off a Pico PSU)
Power improvement? (Same case, power supply, hard drive)
A7N8X-VM with Athlon XP 2000+ : about 100-110watts at idle.
D945GCLF with Atom 230: About 45 watts at idle.
(At this rate, the power savings will pay for my expenses in 10 months, roughly - including the $30 for the 2GB DDR2 RAM)
The following is a small list of issues and solutions I've come across while migrating my Gentoo server over to this thing.
<
ol>
This was easily remedied - there was an old genkernel generic kernel around, which allowed me to recompile with i586/i686/etc as my CPU selection. As a precaution, I also took march=i686 out of my /etc/make.conf
- This is fixed in 2.6.25-r7 kernel, though as I use udev, this was not obvious to me until I glanced through dmesg output to see this:
eth0: RTL8169 at 0xf8916000, 00:1c:c0:6f:dd:7b, XID 24a00000 IRQ 11 udev: renamed network interface eth0 to eth3
- Not only did udev remember the old network cards I had as eth0 and eth1, but it also added eth2 for the temporarily installed PCI NIC I had to use, and finally eth3 for the now-working onboard LAN.
- I just kept symlinking /etc/init.d/net.eth1, eth2, eth3 to net.lo, until I figured out that
/etc/udev/rules.d/70-persistent-net.rules has all these relocations stored, confusing everything for me. I kept thinking that the onboard lan was still not working. Deleting this file got me back my lovely eth0
smsc47m1-isa-0680 Adapter: ISA adapter fan1: 0 RPM (min = 1280 RPM, div = 4) ALARM fan2: 6826 RPM (min = 1280 RPM, div = 4) smsc47m192-i2c-0-2d Adapter: SMBus I801 adapter at 3000 +2.5V: +2.53 V (min = +0.00 V, max = +3.32 V) VCore: +1.15 V (min = +0.00 V, max = +2.99 V) +3.3V: +3.33 V (min = +0.00 V, max = +4.38 V) +5V: +5.10 V (min = +0.00 V, max = +6.64 V) +12V: +12.25 V (min = +0.00 V, max = +15.94 V) VCC: +3.35 V (min = +0.00 V, max = +4.38 V) +1.5V: +1.56 V (min = +0.00 V, max = +1.99 V) +1.8V: +1.78 V (min = +0.00 V, max = +2.39 V) Chip Temp: +37.0 C (low = -127.0 C, high = +127.0 C) CPU Temp: +45.0 C (low = -127.0 C, high = +127.0 C) Sys Temp: +40.0 C (low = -127.0 C, high = +127.0 C) cpu0_vid: +2.050 V
The former was not initially detected because I didn't have the Intel 82801 (ICH) i2c module (i2c-i801) enabled in the kernel. I had to enable both SMSC LPC47M* items in the Hardware Monitoring section of kernel config too.
hda_intel: azx_get_response timeout, switching to polling mode: last cmd=0x200f0000
The card is detected, mixer works, but when you try to play - the data simply isn't getting through the card - the player application slows to a crawl and is writing audio at 1/100th of the speed, roughly.
I tried to pass the model= option to the kernel (Tried all the models for ALC662):
3stack-dig 3-stack (2-channel) with SPDIF 3stack-6ch 3-stack (6-channel) 3stack-6ch-dig 3-stack (6-channel) with SPDIF 6stack-dig 6-stack with SPDIF lenovo-101e Lenovo laptop eeepc-p701 ASUS Eeepc P701 eeepc-ep20 ASUS Eeepc EP20 auto auto-config reading BIOS (default)
Some of these cause it to accept data just slightly faster than the rest.
Googling turned up nothing thus far, so I am using the only PCI slot for a Via ENVY 24 that I had laying around (AV-710), now that I got the onboard LAN working. Got a solution? Let me know.
Here is an interesting thing. When I decided to use the PCI sound card, I disabled the oboard audio in the BIOS for simplicity - and the onboard LAN stopped working. I am not joking, and I did not accidentally shut that off too. I flipped audio back on - and LAN came back. This may be an r8169 linux driver bug, or an IRQ conflict, I am not sure.
Then, in my modprobe cluelessness I attempted to remap the Envy to snd-slot-0 ... no such luck. The useless onboard audio was always the first sound card. Eventually I got tired of this game and blacklisted the module (/etc/modprobe.d/blacklist):
blacklist snd_hda_intel blacklist snd_hwdep
Job complete.
Thanks for the helpful post, I just bought an Atom230 and am trying to get Gentoo running on it via PXE / net boot.
I normally use a very low-spec '586 and above' kernel that boots on almost anything, so was surprised when this failed.
Upgrading kernel now to see if that makes a difference, guess it should, your comments are confirmed here: http://www.gossamer-threads.com/lists/linux/kernel/982159