Re: Experiencing hangs on SMP box with no console messages given for clues. Details inside.

On Sat, Mar 10, 2007 at 02:39:47PM -0500, Kris Kennaway wrote:
> On Sat, Mar 10, 2007 at 10:14:23AM -0600, Tillman Hodgson wrote:
> 
> > If I'm understanding what I'm reading, this isn't all that helpful
> > because it's just saying that I broke into the kernel debugger and then
> > issued a manual panic (which is true, that's how I got the core). Is
> > there a way to obtain information on what was happening before I broke
> > into the debugger?
> 
> Well you really needed to investigate some things in DDB before
> rebooting, e.g. the lock states.  I thought this was described in the
> developers handbook - did you look for it?

Kris, I greatly respect your work and I'm glad to have you involved in
the thread.

I'm familiar with your email style and while I could be reading you
wrong I'm detecting curtness and a bit of hostility in your posts to
this thread. I sincerely apologize for being inexperienced at this. I'm
a fast learner and I've been involved with FreeBSD for a some time, but
that obviously can't compensate for my lack of knowledge in this area.

In response to your question, I've already posted to this thread the
documents that I looked for and was following and the Developers
Handbook wasn't among them. I'll read over it now, of course. Please
keep in mind that it has been 56 hours between when I first posted and
this post from you. If I'm inexperienced and followed the wrong
documentation, great -- someone with kernel debugging experience could
point that out to me and I can get to work getting the information the
developers need. I don't think it's reasonable to expect someone to
leave a machine in the debugger for 56 hours before even getting a
pointer to the correct docs though, especially when the problem
re-occurs somewhat regularly.

Now that I have the documentation that you'd like me to follow (I assume
it's section 11.5 and especially 11.9 of the Developers Handbook), it's
easiest to simply wait for it to hang again. Shouldn't take more than a
day or two to get the info requested in 11.9.


I suggest that the doc team be brought into the loop so that the FAQ 5.8
and 18.14 can be updated to point to the correct documentation. This
should make it easier for other folks who are looking for the same
information, especially since those FAQ entries are both official
documentation and one of the top Google hits.

-T


-- 
If the world was without any natural evil and suffering we wouldn't have the
opportunity, or nearly as much opportunity, to show courage, patience and
sympathy.
    -- Richard Swinburne, _The Philosophers' Magazine_