>> which wasn't there. Configuring the kernel without LED timers solved >> that issue. > > Oh? Why does the same kernel work in qemu then? It's on my todo-list to find out why... > > That sounds very close to the hlt instruction on x86. PPC has different SPRs > that can set the CPU in a similar mode. > > Just look at the other architectures. There's a pretty cool helper to sleep > until a signal / timer occurs. OK, cool. > >> - Is /dev/console supposed to be created in the initrd, by the init >> process, by the kernel or a combination thereof? > > A combination. The actual device is created by the kernel. The device node > on your initrd fs is just a device node. /linuxrc should create that one. So the sys_open(..., "/dev/console",...) call in main/init.c: init() should succeed even though there is nothing creating the fs device node in the initrd, or? (sorry, I find this somewhat confusing) > >> - If created by the kernel, who takes care of this? >> - How is /dev/console linked to /dev/ttyXXXX? > > Usually using console=xxx on the kernel command line. Some distributions > also ship tools to redirect the output (blogd). > > In your case I guess the kernel couldn't find the specified device to link > /dev/console to. Double-check that the uart init code was successful. It seems that not all of the uart init code is executing - exactly as you suggest. I'll look into it. Thanks! -Christoffer