Prolixium dot com: News >> Blog

We've all heard about Michael Phelps, and the record-breaking olympics he's had this year. Just to recap (because I keep forgetting), here are the eight (8) gold medals he's gotten:

August 10: 400 IM (4:03.84, WR)
August 11: 4x100 Freestyle Relay (3:08.24, WR)
August 12: 200 Free (1:42.96, WR)
August 13: 200 Fly (1:52.03, WR)
August 13: 4x200 Freestyle Relay (6:58.56, WR)
August 15: 200 IM (1:54.23, WR)
August 16: 100 Fly (50.58, OR)
August 17: 4x100 Medley Relay (3:29.34, WR)

The best finish was the 4x100 freestyle relay on the 11th:

4x100 freestyle relay win

Yep, that's how it feels to beat the French team, who earlier said they would "smash" you.

If I may put it so humbly: Phelps is da' bomb.

So, I upgraded the kernel on my laptop (Gentoo) to 2.6.25-gentoo-r7, thinking the suspend-to-ram might start working (broke in 2.6.24-gentoo-r8). Big mistake. Here's what broke:

Wireless

The Atheros AR5212 is supported in the new ath5k driver built into the kernel. Unfortunately, it doesn't work correctly. I switched wpa_supplicant to use the wext driver instead of madwifi, and after a couple worrying ioctl error messages DHCP ended up succeeding and I was apparently online. iwconfig said I was stuck at a bitrate of 1Mbit/sec, though, and I was getting heavy (30%+) packet-loss. No amount of fooling with the settings fixed this.

I tried to go back to the madwifi drivers, so I emerge'ed madwifi-ng again. After changing wpa_supplicant back to use the madwifi driver, I started up net.ath0. Yep, system hard-locked. The kernel was still alive, but I couldn't get anything useful out of the memory or task information provided by hitting some SysRq keys.

Bad.

fglrx

Yes, ATI sucks, and fglrx is a horrible non-free piece of junk. However, it's the only thing that will give me fast video on the Mobile Radeon 9600 video controller in the IBM T42.

The module built fine (from the ati-drivers ebuild), but wouldn't load due to the init_mm symbol being gone from the kernel. It didn't look like I forgot to enable any deprecated options in menuconfig. I guess the Gentoo developers just didn't include a patch to make this work.

Yep, fail.

Debian, here I come

I just got fed up with waiting for my laptop to recompile everything, too, so I installed Debian, and within an hour or two, am already up and running with 2.6.25, the madwifi drivers, and fglrx.

Since it's been a little bit since I've configured Xorg on a Debian system, I was a little surprised to see that dpkg-reconfigure -plow xserver-xorg produced a completely plain xorg.conf. It only asked me about my keyboard, nothing about the video card drivers or monitor specifications. I didn't really care about that (Xorg is moving to a hotplug/dynamic-type of system, anyway), but it kept trying to use the old radeon driver that didn't provide 3D or quick video.

I used module-assistant to build the fglrx drivers (the binary ones didn't work), and then ran:

# aticonfig --ovt Xv --initial -i /etc/X11/xorg.conf

Presto, working xorg.conf. No tweaking/hacking needed. However, any subsequent reinstalls of xserver-xorg will probably revert the xorg.conf back to the blank state. Not a big deal.

The ath5k driver that came with Debian's 2.6.25 (2.6.25-2-686) kernel image didn't work, either. Same packet-loss problem, but this time it at least tried to run at 24Mbit/sec (I'm not that close to the AP). So, I just used module-assistant to build the madwifi drivers and created some custom debs. Surprisingly, udev did the right thing, and loaded ath_pci instead of ath5k. Works fine!

So, two more Gentoo systems need converting: my work laptop (Lenovo T60) and desktop at home.

Oh yeah, I also turned 27 today. Bleh.

I haven't used a Bluetooth keyboard or mouse with my laptop (Gentoo) in awhile, and decided to use my Apple Wireless Keyboard (the tiny one; w/out PgUp/PgDn, etc.) today. Unfortunately, I realized that hidd and friends no longer existed. Here are the versions I'm using:

[I--] [ ~] net-wireless/bluez-libs-3.32 (0)
[I--] [ ~] net-wireless/bluez-utils-3.32 (0)

Figuring this was a mistake, I Googled around and found a ton of related Ubuntu forum posts and bug reports filed against the bluez-utils package in the "hardy" and "intrepid" (upcoming) releases.

Apparently, the Bluez folks decided to do away with the old daemons (hidd, etc.) and move everything to the dbus service-type (input.service, audio.service, etc.) model. Unfortunately, they left everyone with only a couple bad replacements for the easy "hidd --search" command:

Issue a couple cryptic dbus-send commands and pray everything will work
Use gnome-bluetooth and click your way to success
Write a bunch of Python scripts to manipulate dbus

Sure, the Gnome way sounds preferred, but not everyone uses Gnome (or X). What happens if I need to use a Bluetooth keyboard on a box w/out X? The dbus-send commands (described here) are fairly complex for such a routine task such as connecting a wireless keyboard.

It does seem that there's a configure option to compile the old daemons (and USE flag in Gentoo named "old-daemons") when the dbus-send commands fail, but this is just a temporary fix since it appears that these will be permanently removed and completely unsupported in the future.

I really hate deprecation, especially deprecation that involves GUI-centric replacements. Don't even get me started about NetworkManager.

Relevant links:

Ubuntu bug 191704: hidd binary removed form bluez-utils package unable to connect as a result
Ubuntu Forums post: Apple Bluetooth Keyboard in Hardy (dbus-send commands - didn't work for me)
HOWTO/InputDevices - BlueZ - Trac (Python and command-line examples)

Update: What I did to get this whole pile of junk working:

(00:20:E0:79:0B:1F is my laptop's bdaddr and 00:1E:52:FA:6D:94 is my keyboard)

# echo "00:1E:52:FA:6D:94 1234" > /var/lib/bluetooth/00:20:E0:79:0B:1F/pincodes
% dbus-send --system --type=method_call --print-reply --dest=org.bluez /org/bluez \
org.bluez.Manager.ActivateService string:input
method return sender=:1.0 -> dest=:1.8 reply_serial=2
   string "org.bluez"
% dbus-send --system --type=method_call --print-reply --dest="org.bluez" /org/bluez/input \
org.bluez.input.Manager.CreateDevice string:00:1E:52:FA:6D:94

I hit 1 2 3 4 <ENTER> on the keyboard…

method return sender=:1.0 -> dest=:1.4 reply_serial=2
   string "/org/bluez/input/keyboard6"
% dbus-send --system --type=method_call --print-reply --dest="org.bluez" \
/org/bluez/input/keyboard6 org.bluez.input.Device.Connect
method return sender=:1.0 -> dest=:1.7 reply_serial=2

Keyboard then started to work.

I dropped my E71 on a tile floor, today. Compared to some other Motorola phones I've owned in the past, this one didn't flinch. Nice design - although there's a slight scratch on the bottom, now :-(

…of course, my E61i was pretty sturdy, too - dropped that during the ceremony at my friend's wedding.

Yeah, I appear to be getting clumsy in my old age.

Originally posted to IRC:

21:18 < prox> HA
21:18 < prox> We can't take off because the cover is missing from one of the flourescent lights in the cabin.
21:19 < prox> <3 flying!
21:19 < prox> .. of course, with Newark, it's always something.
21:20 < prox> And, of course, we can't use any personal electronic devices - however cellphones are allowed!
21:20 < prox> This all makes perfect sense!
21:22 < prox> Dammit.
21:29 < prox> HA
21:29 < prox> Now they just kicked off all the standby passengers.
21:31 < prox> This is great.  Everyone on the plane juSt boo-ed.
21:32 < prox> Hilarious.
21:32 < cnj> define "cell phone"
21:32 < cnj> and how it differs from a PED
21:33 < cnj> (... if my laptop has a UMTS card in it, and I'm using it to make a SIP call ... that's a cell phone?)
21:33 < cnj> yeah, don't bother.
21:33 < prox> Yeah, they'll just get confused and ban cell phones as well.
21:33 < prox> Then I'd get attacked.
21:36 < prox> "If you're in the lavatory, please return to your seat."
21:38 < prox> WHOO
21:38  * prox &

Yeah. My flight from Charlotte to Newark on US Airways was supposed to depart at 20:25, but wheels-up ended up being 21:58. It wasn't the delay that bothered me - I'm used to these kind of delays flying into Newark. It was the stupidity. Also, sitting next to this girl who claimed she's the "most impatient person I'll ever meet."