Kernel compilation: Don't Panic... *PLEASE* DON'T PANIC!!

[catsidhe]

Decided to set up my desktop with the 2.6.20 kernel. Maybe they've fixed the bug where plugging my mobile into the computer crashed the USB module. Did some fiddling with the .config, to hopefully enable some hardware I had not previously been able to get working (like the framebuffer console).

Compile.

Install.

Reboot.

Yay! Framebuffer works! I get a reasonable resolution, and see a penguin up the top corner. Thus, I get a wonderful opportunity to really look at the kernel not being able to find the root partition, and panicking.

Ah, fsck.

At least the previous kernel works fine. Giggling shows that one problem may be with the initramfs. Eh. Recompilation in progress. Let's see if that fixes it.

Edit: Nope. Telling it not to use initramfs in the kernel did nothing. Just tried mkinitfs, let's see how this goes.

Edit: Nope. Bugger. Why does this kernel not work where the previous one does? Something to do with its handling of initrd? Some problem with the hard drive drivers? Gah.

Comments

[araquel.livejournal.com]

You know, I hear Microsoft have a new OS out that's guaranteed to fix everything short of world peace...

[catsidhe.livejournal.com]

Yeah, about that.

[sjl.livejournal.com]

And people wonder why I jumped ship from Linux to OS X. Too much faffing around to get things to work (it was X in my case, but I've had similar problems every fucking time I've recompiled a kernel lately.)

Damnit, I understood the OS in 2.2. What the hell happened?

[brong.livejournal.com]

udev

[tau-iota-mu-c.livejournal.com]

Bah.

Greg KH can go stand in a pit of molten sodium.

[brong.livejournal.com]

That's why I stuck at a single word rather than ranting at length. It's a good idea, and I understand the motivation behind it - but the execution has been plagued with stupidities beyond what you'd expect of a bright guy like Greg. Consistency and robustness across different usage patterns _matters_ for a boot-essential component damnit.