When you're shutting down your computer, trashed displays are irrelevant
anyway. Shutting down to single user mode is of course an exception, though,
so you do have a point.
> >
> > This is wrong too. shutdown sends a SIGTERM or some other "light"
> > signal, which X receives and then shuts down cleanly.
> >
> > Anyone who says that shutting down your machine from within' X can be
> > expected to crash your machine, and that this is OK, is smoking crack.
> > That's %100 a bug in my book.
Agree wholeheartedly. Also bringing me back to my original example: I don't
think running a second server is `stupid'. X servers were designed to allow
themselves to be run multiple times. If the framebuffer driver doesn't let
this happen, it's a bug in the framebuffer driver.
>
> And if X terminates slowly, SIGKILL goes in few seconds...
>
> I found much more serious bug: xinit sometimes sends SIGKILL to xserver
> when terminating. If your machine is under heavy load (make -j zImage) and
> you correctly terminate windowmanager (not xserver), xinit kills xserver
> with -9 and console is dead.
This is actually supposed to be a feature: it's to protect against the server
crashing and refusing to terminate, which will definitely leave your console
unusable.
(as an aside, make -j is completely useless unless you have 10-12 processors
in your machine, except for stress-testing the kernel)
>
> I think the kernel shouldn't allow users to send unmaskable signals to
> suid programs.
suid programs should be written to be able to deal with such circumstances.
Unexpected events can happen due to any number of causes. suid programs don't
have the excuse that they were run by root, so he can be blamed for anything
that goes wrong.
-- arvind
-
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/