Free as in speech

So I’ve got my shiny new laptop which should complement my 5 years old PowerBook G4 500 MHz and had a bit of fun installing Debian AMD64 on it. I did it the hard way, not using debian-installer as it was a given for me that both the kernel and debian-installer itself were unable to deal with the machine.

Installation method:

• debootstrap an amd64 system and an i386 system on a FireWire disk
• build a suitable x86_64 kernel for the MacBook Pro, handpicking patches from mactel-linux
• reinstall a fresh and stripped down Mac OS X system on a smaller partition
• boot on the mactel-linux live CD
• partition using parted
• sync GPT and MBR with gptsync
• mkfs, mkswap
• transfer the amd64 system using xfsdump/xfsrestore
• chroot into the i386 system, edit lilo.conf, run lilo
• reboot, hold down the option key, choose the infamous “Windows” icon
• ENJOY

Note that you do not need to install BootCamp on Mac OS X. On the Core2 Duo machines, the firmware already includes the SMBIOS stuff, so you’d only install a useless Mac OS X GUI which you can’t use because you’ve repartitioned before running BootCamp Assistant :-)

Current status: most of what should work with Linux sort of works (that is, to say, everything but the WiFi).

Things that do not work:

• external VGA monitor: a DVI-D monitor works fine, looks like fglrx does not enable the DVI-I output on the DVI connector (it even reads the EDID data from the VGA monitor). If you know someone at ATI, please point them to this post. My bug report to ATI has been discarded with this comment: “we do not supply drivers for laptops, it’s up to the manufacturer” (roughly). Yeah, right. Not buying ATI anymore if this doesn’t get fixed.
• WiFi: new Atheros5008 a/b/g/pre-n chip, unsupported by madwifi and future support is uncertain at best as of now. Happy. NOT. (madwifi ticket #1001)
• sound: sort of works with Linux 2.6.18, sound works on both headphones and speakers. Broken in 2.6.19. (ALSA issue #2198)
• no keyboard in LILO: strange, but the keyboard doesn’t work at boot time in LILO when run through rEFIt, though it does sometime work when run directly from the Apple chooser. lilo -R to the rescue, but this is painful. Maybe GRUB would work better.
• applesmc: has issues, the SMC times out frequently. Works nonetheless, and it does not happen when first booted into OS X. Probably an initialisation issue.
• trackpad: doesn’t work as good as in Mac OS X, some synaptics config tuning should fix that
• battery life sucks: power consumption is a steady 26 W for a 55 Wh battery, even with both cores running at 1 GHz instead of 2.33 GHz. Power consumption doesn’t vary depending on the CPU frequency. Bug ?

Everything else works, and works fine. Wrote a small daemon to handle the special keys for LCD backlight, sound, keyboard backlight and CD ejection. Still needs a bit of polishing, after what I’ll release and package it.

rEFIt isn’t available in Debian AMD64 yet, so gptsync isn’t available and must be run from an i386 system for the time being (plus gptsync is buggy on 64bit architectures). Patches ready, waiting for GNU-EFI on amd64 (#383801).

Suspend to RAM and suspend to disk both work, use s2ram -f -p -m until the machine makes it into the built-in whitelist.

The machine is pretty nice, though I’d appreciate it if the fans weren’t running at 5500 rpm when I have the laptop on my laps. Might be related to both the applesmc and power consumption issues.

WiFi remains the biggest issue, battery life comes second ex-aequo with DVI-I output and sound is third.

I’m quite unhappy with having to taint my kernel by using the fglrx (and madwifi in the future) proprietary crap, and even more when it doesn’t work as good as it should. Maybe RMS was right about ATI.

[For background information on the whole debian-installer/Sven Luther situation, please read the Debian wiki.]

The whole debian-installer/Sven Luther issue has been running for months now, and has seen little to no progress. A number of DPL mediation attempts have utterly failed, and there’s just no authority whatsoever above the debian-installer people that could help settle the issue.

I’m just not interested in discussing who’s at fault, as both sides are at fault for something or something else.

Now, looking back, there is something that bothers me. It makes me sick, even.

At the end of march this year, Sven lost his mother in very difficult circumstances, in the middle of the flamewar opposing him to the debian-installer team. He sent a private mail to Frans Pop which you can read on this wiki page explaining what was happening to him, and asking him to give him a break.

Reading this mail, it’s pretty clear that Sven was in shock when he wrote it. It’s so obvious that you can’t miss it, it doesn’t take a psychologist to tell you that. Nonetheless, the bashing went on, and nobody cared about what was happening to him.

Now, compare to what happened to dato. How did the DDs react to that ? They rushed to debian-private, both the mailing-list and the IRC channel, to get an update on his condition, for something like two weeks.

In both cases, some people knew what was happening (in Sven’s case, others DDs knew what was going on too, Frans Pop wasn’t the only one). I just can’t explain why the reaction was different, and it really bothers me.

I spoke IRL with Sven about the debian-installer situation. We rehashed most of what had been told already, and he mentionned the death of his mother just as he did mention it on the mailing-lists. He has been and still is very affected by what happened to his mother. Reading his mail to Frans Pop, it’s pretty easy to understand I think.

The behaviour of some DDs in this flamewar, knowing that Sven was deeply affected by this event, is inexcusable. You really should be ashamed of what you’ve done to him. He needed the kind words from the Debian family, the same kind words the Debian family told to dato, his friends and family.

It’s time to clear up the mess, folks.