His Deeds Are Dust

I have a Samsung X120 notebook. Commiserate with me, please, for ever since F13′s kernel went from 2.6.33 to 2.6.34, I’ve been without ACPI. Anything other than acpi=off in the boot settings produced a giant stack trace that scrolled off the tiny screen so fast and so early in the boot that only videoing the screen would have enabled me to transcribe it. Even boot_delay didn’t work, because that won’t delay every line of output when the kernel is curling up in a corner and dying.

It’s a bug in the Samsung BIOS no doubt, and it has an upstream kernel report, but the bisect points to a sane commit that’s good for everyone else.

Running without ACPI isn’t a realistic option on a laptop. I wouldn’t know when to plug it back into the mains, and shutting down requires rather more persuasion than should be necessary, so last night it was time to patch the kernel.

Fedora Wiki to the rescue. I followed the instructions on building a custom kernel to the letter, and eleven hours later, I had a functioning laptop. I’ll break that down, lest you think me hard of understanding: five minutes to read the page, 20 minutes to grab the kernel src.rpm and make the appropriate patch, and 10 hours and 35 minutes to compile the entire thing on the laptop. I went to bed for the last bit, after ensuring that the mains was plugged in and on so that my laptop wouldn’t do the same.

So, to everyone who has had a hand in that page on the wiki, thank you.

As suspected, the Samsung X120 notebook has the same keyboard quirks as many other Samsung laptops, so that some of the function keys don’t work correctly under Fedora 11. There are two problems and fixes needed. One, the kernel needs patching so that the quirks with missing key release events are also applied to the X120 (kernel bug 14052). Secondly, HAL needs patching to connect the function keys to turn the brightness up and down, etc., in the manner described by Fedora bug 504009 and Ubuntu bug 399911 (and many others).

However, I have confirmed that grabbing and patching kernel-2.6.30.9-90.fc11.src.rpm works me.

I’ve just charged my notebook’s battery fully, then allowed it to discharge while the machine was idle. First of all, I nobbled the power management settings in Fedora 11 to make sure that the screen stayed bright throughout, and it didn’t perform a neat shutdown when it guessed the battery was close to death.

I have the four cell Li-Ion battery, product ID AA-PB0TC4L, which Samsung claims will last four hours, based on BatteryMark tests.

It took 1 hour 53 minutes to discharge.

I present this figure without further comment because I don’t know whether this type of battery takes a few charge/discharge cycles to gain a longer life, or any other possible explanations.