[Nut-upsdev] Asking hard questions about the NUT architecture

[Eric S. Raymond]

Charles Lepple <clepple at gmail.com>:
> I'm getting delayed whiplash from your time-warp comment at the start
> of the thread.

:-)
 
> I'll admit that the idea of a working SIGPWR in Linux* is growing on me. 

I'm pretty sure, from the behavior of my systems and Linux signal(7),
that it already works.  But you've raised some doubt in my mind, and I
will check... googling now...

Aha.  Here's a post from 2002 that implies that SIGPWR was not delivered
in (unspecified) early Linux versions but came in with RT support:
<http://gcc.gnu.org/ml/java/2002-01/msg00164.html>

Here's a knowledge base entry that speaks of the Linux kernel detecting 
power failure: <http://www.applieddata.net/forums/topic.asp?TOPIC_ID=479>.

Just talked to a friendly kernel hacker who says for sure the kernel 
registers the PCI power fail, but thinks acpid rather than the kernel proper
might be what's shipping SIGPWR to userspace in some ports.  

Yoicks.  *That* raises the possibility that whether SIGPWR is delivered
might depend on the distro integrators.  I'm going top chase this some
more, after I sleep.
-- 
		<a href="http://www.catb.org/~esr/">Eric S. Raymond</a>