Michael L. VanLoon writes (after removing references to previous
posters):

>>> If you have an Adaptec or BusLogic PCI SCSI controller you can borrow,
>>> you might also try that in the machine.  Note that the Adaptec driver
>>> also has some bugs of its own.

>>If this means following problem
>>	ahc0: Can't malloc SCB
>>	sd2: not queued, error 5
>>BusLogic driver also has same problem, and IMHO, this is not driver problem,
>>but MI SCSI problem.

Whether this is a MI SCSI problem is open to debate.

>I was not aware that this was a general problem with all SCSI
>adapters.  I had never heard of it happening on other cards.

I've never been bitten by the sendmail `wiz' attack.  Therefore I
don't believe that this security hole existed.


>Is this a new problem?

No.

>What does it take to trigger it?  I've certainly never seen anything
>like that happen on my BT747.  Am I just not pushing it hard enough?

The specific problem reported in current-users seems to come from the
driver telling the MI SCSI code lies about the number of concurrent
requests it can handle on some AIC7XXX cards.  To be more specific, it
tells the MI SCSI code that it has resources for more requests then it
actually has allocated.  The driver tries to allocate misssing
resources dynamically, but that can fail and the MI SCSI system is
just not prepared to deal with this.

Try recompiling you BusLogic driver with 4 CCBs.  That should enable
you to verify that problem yourself.

-- 
Christoph Badura	bad@flatlin.ka.sub.org

You don't need to quote my .signature.  Everyone has seen it by now.
Besides, it doesn't add anything to the current thread.