Re: devfs patch v21 available

Richard Gooch (rgooch@atnf.CSIRO.AU)
Wed, 1 Apr 1998 09:18:06 +1000

Miquel van Smoorenburg writes:
> In article <199803312213.IAA26698@vindaloo.atnf.CSIRO.AU>,
> Richard Gooch <rgooch@atnf.CSIRO.AU> wrote:
> >Ack! In my suggested userspace fix, I should have prefixed an exec:
> >
> ># cd /sbin
> ># mv init init.real
> ># echo '#! /bin/sh\nmount -n -t devfs none /dev\nexec /sbin/init.real' > init
> ># chmod a+x init
> >
> >This will ensure that the init.real programme runs with PID=1. I think
> >that will solve the problem.
>
> No, init does NOT create the /dev/initctl fifo. It just assumes it's there.
> Therefor the correct solution is to create it in some startup script and do
> a kill -USR1 1 to tell init to open it.

OK, I should have been more explicit. I was addressing two points:

1) the above fix would solve the /dev/initctl problem. As you've
shown me, it doesn't. So, the fix below should work, correct?

# cd /sbin
# mv init init.real
# cat > init
#! /bin/sh
mount -n -t devfs none /dev
mknod /dev/initctl p
exec /sbin/init.real
^D
# chmod a+x init

2) the generic problem of running a script which then runs the real
init binary (by using exec that should work).

> Oh, and please do not create it in devfs itself - I hope to convince

I wouldn't want to do that in devfs itself: devfs (and the kernel in
general) should not be making assumptions about what lies in
userspace.

Regards,

Richard....

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu