SIGSTOP/SIGCONT problem ? (scheduler or make(1) ?)

[Danny Braniss]

> Kris Kennaway wrote:
> > On Mon, Jan 03, 2005 at 08:04:24PM +0100, Poul-Henning Kamp wrote:
> > 
> >>In message <20050103183736.GA93409 at xor.obsecurity.org>, Kris Kennaway writes:
> >>
> >>>--cWoXeonUoKmBZSoM
> >>>Content-Type: text/plain; charset=us-ascii
> >>>Content-Disposition: inline
> >>>Content-Transfer-Encoding: quoted-printable
> >>>
> >>>On Mon, Jan 03, 2005 at 02:34:41PM +0100, Poul-Henning Kamp wrote:
> >>>
> >>>>=20
> >>>>I ran a "make -j 48" on my testbox and suspended the task after
> >>>>a while so I could start systat(1)
> >>>>=20
> >>>>=3D=3D=3D> usr.sbin/sysinstall (obj,build-tools)
> >>>>=3D=3D=3D> gnu/usr.bin/cc/cc_tools (obj,depend,all)
> >>>>*** Stopped -- signal 18
> >>>>Child (25164) not in table?
> >>>
> >>>I've seen this too.
> >>
> >>With make or with other apps ?
> > 
> > 
> > With make, I think.
> > 
> > Kris
> 
> I have this odd feeling that make needs an explicit SIGSTOP and/or
> SIGTSTP handler.  There is a conditional in the source for USE_PGRP that
> might be worth looking at since it contains just such a handler.
> 
> Scott

i've noticed something similar, but different :-)
	su
	...
	suspend

sometimes the shell does not get the SIGCONT, doing a 'kill -CONT shellpid'
fixes the problem, so, IMHO, it's not just make.

danny