[CUWiN-Dev] Re: [CUWiN-Commits] r3468 -
cuw/trunk/src/boot-image/extras/usr/sbin
Bill Comisky
bcomisky at pobox.com
Thu Sep 8 09:52:12 CDT 2005
On Thu, 8 Sep 2005, dyoung at cuw.ojctech.com wrote:
> Author: dyoung
> Date: 2005-09-08 04:09:49 -0500 (Thu, 08 Sep 2005)
> New Revision: 3468
>
> Modified:
> cuw/trunk/src/boot-image/extras/usr/sbin/hellowdog
> Log:
> Fix a serious regression in hellowdog, which wasn't restarting its
> countdown when it got a SIGHUP. It never got SIGHUP because sh(1)
> will not trap signals that its parent ignored, unless sh(1) runs
> in interactive mode. Now, I bracket the trap statements in set
> -i/set +i statements, so that the trap does get installed.
>
> I'm not quite sure how I convinced myself hellowdog was working,
> but I suspect I tested with src/bin/sh/trap.c:1.32, instead of
> trap.c:1.33 (today's HEAD).
It is working here using pre rev3468 and src/bin/sh/trap.c:1.32. If I
only bring one node up it reboots every 6 minutes or so. With >1 node, it
stays up. With one node I can manually /usr/sbin/tickle the system
occasionally to keep it up.
I'm still building nightlies with the 0.5.8 NetBSD src tarball. I feel
like it makes more sense to track what NetBSD src you are using to pick up
possible regressions like this. Are you tracking NetBSD-current, or do
you have a src tree I can rsync to?
bill
--
Bill Comisky
bcomisky at pobox.com
More information about the CU-Wireless-Dev
mailing list