>> If process 1 terminates, UNIX semantics are done for. From the moment >> process 1 dies and the system enters kernel space to handle this death, >> the system never goes into user space again. > Think about SMP systems...
* Any clean shutdown sequence would still have to kill all processes, just like it is done today; but instead of calling reboot() or poweroff(), process 1 would just _exit() at the end. If process 1 is the only remaining process on the machine, there is no race condition, SMP or not. * If process 1 terminates while there are still other processes alive on the machine, it may or may not be a clean shutdown, just like a reboot with other processes alive may or may not be clean today, depending on the disk mounts state. Today's SMP systems know how to handle a shutdown. My suggestion does not change that. It does not remove the need for a proper userland shutdown sequence; it does not assume that process 1's exit does anything magical. (auto-unmount everything! auto-kill everything! wheee!) I am *only* suggesting that process 1 exiting should be taken as a hard reboot or poweroff instruction, instead of a panic instruction. Is there something I have missed that would still be a potential problem with SMP systems ? -- Laurent _______________________________________________ busybox mailing list [email protected] http://lists.busybox.net/mailman/listinfo/busybox
