Messages in this thread | | | Date | Sun, 10 Jan 1999 04:02:20 -0500 | From | David Feuer <> | Subject | Re: Porting vfork() |
| |
Linus Torvalds wrote: > > On Fri, 8 Jan 1999, Kenneth Albanowski wrote: > > > > > > No need to. If you use sleep_on(), the parent won't be getting any signals > > > anyway (only sleep_on_interruptible() cares about signals). > > > > Of course, but is that a good thing? The parent will be unkillable until > > the child does something. > > Yes. That's basically how vfork() works. > > You could make it a special kind of killable - where you can _only_ kill > it (ie only fatal signals will be serviced), and that would work. I > wouldn't do that until people actually start to complain. It's not a > security issue, as the parent _can_ be killed - you just have to kill the > child first. > > Linus > > - > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in > the body of a message to majordomo@vger.rutgers.edu > Please read the FAQ at http://www.tux.org/lkml/
You mean you just have to kill _all_ the children, and their children _in order_. This seems like a big security hazard to me.... Vfork bombs could become quite difficult to stop.... e.g.
while(vfork());
you'd have to start killing at the high end (pid-wise), but the processes are _created_ at the high end, so you would fail.
-- This message has been brought to you by the letter alpha and the number pi.
David Feuer dfeuer@his.com dfeuer@binx.mbhs.edu Open Source: Think locally; act globally.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |