Well, i don't understand where the deadlock could appear. the init scripts don't run in parallel (at least in debian), so the only init script currently in charge is /etc/init.d/udev. Udev daemon AFAICU sits on hotplug event handler and spawns children initializing devices as the hotplug events go. The children are instances of udevd again, and they should be forced to timeout instance by instance device-wise. While the loop in the initscript should wait for them to terminate forever. The reason is: following init scripts' success heavily depends on existence of devices created by the udev daemon, and timeout means you leave the success to a chance. What if my super duper RAID controller with 126 LUNs initializes for a minute? Or my mouse farm? Or my wifi controller, which happens to run before the ides? (i believe i was lucky when my box booted into RL1, and i can imagine scenario where the only solution is liveCD and edit of some cfgs.)
Thanks for the care and the speedup, anyway.
Dominik Sauer
- Bug#336189: acknowledged by developer (Bug#336189: fixed in ... Dominik Sauer
