2008

Sun, Feb 24

<01:28 EEST> The week's been pretty rough, and on Thursday the stress energy tensor peaked as the new motherboard died. The machine had been crashing quite consistently due to OpenGL accelerated games, and the chipset turned out to have an improperly attached heatsink. The shims were way too thick, so there was considerable paste-filled space between the chip and the sink. After some of my previous warranty replacement issues, I'm expecting a few months without the machine.

Meanwhile, my laptop is working out perfectly good enough for my entertainment needs, as far as sound and video cables go. I managed to locate a problem with the external display; the 1280x1024 monitor refused to use its full resolution, instead sticking to the laptop's native 1024x768. However, this was only with a VGA extension cable, and the monitor's original cable did give the correct resolution. So the extension cable lacked a DDC pin, as I confirmed by actual inspection.

For gaming (mainly old arcade/PC games using MAME and Dosbox :) I bought a mini USB keyboard, and I might even get a second one. At 15 euros they make pretty versatile and cheap game controllers ;)

Mon, Feb 11

<22:19 EEST> This is hopefully the last post about Linux kernel updates for a while. Not that I don't keep updating all the time anyway, but this one is probably noteworthy. 2.6.24.2 fixed a local root exploit, so I booted into that as soon as possible. Even though I trust my users, there's a very remote possibility (pun intended) of someone stumbling upon one of my users' passwords and logging in. While I do use fail2ban to reduce the possibilities of brute-forcing passwords, a local root exploit represents an increased probability of getting root. A little paranoia never hurts fnord.

I've actually had some Life(TM) today as well, due to a media education project at Norssi. I played my part as an early 1900s teacher to teach a little history via role-playing. It was a kind of culmination point for a relatively long project, and in retrospect not too fun or easy, so I've felt a little extra energy today. Not that I could use it for anything productive, after 3 hours of decent sleep last night.

Sun, Feb 10

<15:14 EEST> Linux 2.6.24.1 came out yesterday, and it appears to have fixed the I/O latency bugs. Thus I've updated both of my machines. I haven't experienced any major improvements over 2.6.23, since the big ideas in 2.6.24 seem to target higher performance (Large Receive Offload for IPv4) and multiuser systems (Fair Group Scheduling). A notable small improvement for my machines is the power saving option for sound cards. Of course, for general development and bugfixes, it's good to stay on the stable leading edge.

Wed, Feb 6

<01:14 EEST> Another 2.6.24 kernel IO slowness report. I've been poking around the issue by toggling some of the new config options that seem relevant, with no success.

Mon, Feb 4

<22:34 EEST> In correction to the previous entry, MPlayer is still not ready for the low-power desktop ;) 720p H.264 playback is now smooth, but there is loss of sync, probably due to not dropping CPU-intensive frames. Otherwise it's been an interesting day after too little sleep, almost charring my soy burgers to PAH. On the plus side, in the morning's study group I ended up showing off my Sigma skills with parabolic inequalities :)

Sun, Feb 3

<19:08 EEST> I'm back with the 2.6.23.14 Linux kernel, and I've even reverted to NFS version 3 after some strange bugs relating to the saved state of mounts. V4 is experimental after all. On the plus side, MPlayer has just improved (in scheduling or otherwise) sufficiently to play 720 H.264 videos on Hoo.

Today was time for the second admissions exam for JYT. Pretty exhausting but rewarding.