deandra.homeip.net

Today I spent a little more time cleaning out my home directory on my server. It was like an archaeological dig, unearthing stuff from five or six years ago. Ran across a bunch of old photos, including some of Tigger when he was a kitten; you can find those in The Tigger Gallery.

On to other stuff. TuxOnIce isn't yet working 100% reliably with my Pavilion tx1000 (no surprise there.) Regulars to my site will remember how much trouble I've had with this thing running Linux; it just plain sucks at it. But I keep trying, out of bullheaded stubbornness, probably. And I seem to be making headway.

Suspending the machine always works OK. However, resuming only works about half the time. Sometimes you can try the resume again with the same image, and other times you can't. And there seems to be several different problems; the resume does not always fail in the same way.

I've read somewhere that these kinds of issues can be caused by drivers. In fact, the hibernate script that ships with Debian tries to unload a bunch of modules before doing a suspend, to avoid this sort of trouble. Normally, these modules will be reloaded after the resume is complete. However, since I'm running an initrd kernel, the modules get reloaded before the resume starts. I've run into problems with initrds and modules before, when trying to get ndiswrapper working. It's not enough to blacklist the ssb module to avoid a conflict; you also have to rebuild your initrd without that module, or it will simply be reloaded by the initrd on the next boot, and your wireless will stop working. Anyway, so to stop these modules from being loaded before the resume, I've changed a setting in my /etc/initramfs-tools/initramfs.conf (modules=most to modules=list), and made a list of modules in /etc/initramfs-tools/modules that I want loaded. The changes mean that only those modules will be included in the initrd, so don't forget the critical ones (like the SATA and filesystem drivers. Think of the things you would have to build into your kernel if you weren't running an initrd.) Then I rebuilt the initrd with "mkinitramfs -u" and rebooted the machine.

So far, I've hibernated and resumed twice with success. However, I'm not going to call it fixed yet; I'll see how things look in a week or so. If I haven't had any problems, I'll update my page on this topic.