Date: Tue, 07 Sep 1999 17:46:13 -0700 From: Mike Smith <mike@smith.net.au> To: Jeremy McMillan <aphor@228-121.ppp.ripco.net> Cc: stable@freebsd.org Message-ID: <199909080046.RAA01782@dingo.cdrom.com>
next in thread | raw e-mail | index | archive | help
Fcc: sent
Subject: Re: newly introduced repeatable SMP hang in STABLE/3.3-RC
In-reply-to: Your message of "Tue, 07 Sep 1999 19:30:25 CDT."
<Pine.BSF.4.05.9909071919430.38952-100000@www.notrecords.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
--------
> Could this be that one kernel thread is wedged in a state where it has a
> lock on a resource, but it can still do low level functions like handle
> ICMP? Maybe the other thread hung, and thus cannot release the lock on
> whatever resource the working (ICMP reachable kernel thread) needs to
> resume process managment services? What kernel resource do threads take
> turns locking?
A 'wedged' host that is pingable is locked in the kernel for some
reason, but without at least some interrupts masked. ICMP echo replies
are generated in interrupt context.
If the machine is pingable but you can't get to DDB, chances are that
you're in some code running at spltty, which may be code related to
serial I/O, console output, or printer I/O. In particular, turning off
an attached printer can cause this. (I don't think this is the
plaintiff's problem though.)
--
\\ The mind's the standard \\ Mike Smith
\\ of the man. \\ msmith@freebsd.org
\\ -- Joseph Merrick \\ msmith@cdrom.com
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199909080046.RAA01782>