Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Panics & -strintercept-: what is this!?

1 view
Skip to first unread message

Aaron Mahler

unread,
Jan 8, 1996, 3:00:00 AM1/8/96
to
Could someone please tell us what the heck this is:

Jan 8 16:39:03 soulcage unix: WARNING: strintercept: dropping message
Jan 8 16:39:03 soulcage unix: WARNING: strintercept: dropping message
Jan 8 16:39:04 soulcage unix: WARNING: strintercept: dropping message
Jan 8 16:40:37 soulcage unix: WARNING: strintercept: dropping message
Jan 8 16:40:38 soulcage unix: WARNING: strintercept: dropping message
Jan 8 16:40:38 soulcage unix: WARNING: strintercept: dropping message
Jan 8 16:40:38 soulcage unix: WARNING: strintercept: dropping message
Jan 8 16:41:53 soulcage unix: WARNING: strintercept: dropping message

We cannot seem to find it in any kind of reference materials anywhere - not even
hints or hits in all the Novell website databases.

We get these often - sometimes hundreds of them in a matter of seconds and
our machine panics within a short period of them. They also do not always appear
when a panic happens.

We have had a known problem between UW and MST PPP for a year now and are going
through various tests with both parties to resolve it. I am assuming this might
be related but we are at wits end!

As of this message we have panicked 8 times today - with times between ranging from
7 hours to a rash of them 15-30 minutes apart.

There appear to be NO other warnings of relevance in the logs anywhere. The only
other report I can offer is that users say that the machine often becomes VERY
slow just prior to death.

In a nutshell: our config is UW 2.03, P90, 96 Megs RAM, 3 gigs of drive space on
two Adaptec controlled drives, 3COM ethernet (3C509 I think)... our most common
activites are lots of rlogin and telnet sessions coming from our termservers, many
MST PPP processes, and a couple of web daemons.

Any clues would be welcome - we will flood you and your family with gifts.

Thanks!
- Aaron

--
Aaron Mahler - In Mind, Inc. - 110 Vista Centre Dr. - Forest, Va 24551
ama...@inmind.com - W3 http://www.inmind.com/people/amahler.html
Office: (804) 385-4087 - Pager: (804) 948-9960
What the world really needs is a good Automatic Bicycle Sharpener.

Steve

unread,
Jan 9, 1996, 3:00:00 AM1/9/96
to
In article <30F196...@inmind.com> Aaron Mahler <ama...@inmind.com> writes:
>Could someone please tell us what the heck this is:
>
>Jan 8 16:39:03 soulcage unix: WARNING: strintercept: dropping message
<rest deleted>

A little explanation will help explain this. In 2.x, the semantics of
kmem_alloc changed for various reasons. The most notable things are
that allocated memory may not be physically contiguous and the memory
may not be "16 bit DMA-able". There are drivers that are written assuming
that these characteristics are present (since they used to be). The
new DDI defines this point as ddi 5 compliance. ddi 5 or earlier gets
these semantics, 6 and beyond get the new.

So, what's happening? Well, you have a driver that is ddi 5 or earlier,
which is not a problem, per se. What STREAMS does when it sees this
situation is that it peeks at each message going into the driver to see
if it matches the expected constraints, and if not, it alters the message
on the fly to provide the correct semantics. The message you are seeing
on the console means that a message that needed to be altered could not be
because of a failed memory allocation. At that point, STREAMS has 2
options - drop the message or pass on a message that might panic the system.
So, it drops the message and hopes for the best.

Of course, it's not clear if the message really needs to be altered. There
is no way to tell so a worst case assumption has to be made. This intercept
only occurs for hardware drivers that might be diddling in physical memory.
If you know which driver is involved, and you are absolutely sure that
it has no embedded assumptions about memory (and it doesn't use any obsolete
or non-ddi interfaces), you could try changing the $interface line for
the driver to be a 6. This is not guaranteed to work! I don't know the
driver involved so this could cause panics and/or some other breakage, but
it's perhaps worth a try.
Steve Albert
s...@novell.com

0 new messages