On Thu, Oct 27, 2011 at 2:12 AM, Dan McGee <dpmcgee@gmail.com> wrote:
1. Is there a need to even call swapoff? I can't believe it is essential for swap partitions. It unfortunately does makes sense for swap files so we can later unmount the file system they live on, but it looks like there is no way to differentiate. Dave, I know you want to submit a patch to util-linux for this... :)
You are right in this, if we can only swapoff swap files that would be more efficient (unless I'm missing something). Dave: thanks for the work on this already!
2. Why do we do anything except unmounting filesystems after the kill_all call? It seems like we could move the random seed, and timezone set above it, and then kill udev, and then we'd be safe from any and all spawned processes. The only things following would be (hooks excluded) a umount call, vgchange/cryptsetup calls, a mount call, and either poweroff/reboot.
The reordering you propose makes sense to me, I'll do this with the next round of patches (unless someone beats me to it).
That seems pretty easy to audit, and I feel like you took a potshot at the killall thing when in fact udevd was the only bad boy in the corner causing trouble.
Our killall works if everyone behaves as they should (and to the best of my knowledge, now they do). However, I guess it would not be difficult to create a program that forks at the right times in such a way that it would escape being killed. I think things are "good enough" as they are now though, unless someone finds a real-life problem, or someone proposes a fool-proof replacement. -t