Subject: Re: uVAXII with Emulex 3rd
To: Anders Magnusson <ragge@ludd.luth.se>
From: Tim Shoppa <shoppa@krl.caltech.edu>
List: port-vax
Date: 09/29/1995 11:05:14
> > 
> > 
> > Hi all,
> > 
> > I'm previously reported here my problems with an Emulex controller in
> > a uVAXII. Meanwhile I4ve read the controller manual. I4ve found a table
> > which contains several csr settings for certain hardware. 
> > Additionally, the first UDA50 controller should have an interrupt vector of
> > 154. The NetBSD kernel just before it hangs writes out the message:
> > 	uda0 at uba0 csr 172150 vec 177,ipl 17
> > 	uda0 controller hung
> > Is it possible that the reason is a difference in the interrupt vector ?
> > Is there a way to modify the interrupt vector with the rom monitor ?
> > 
> No, it is not possible. The interrupt vector is dynamically allocated
> by the uba driver from the floating vector space, and told to the
> controller. It that should fail you would get noticed on the console
> about that. Your problem is that the ctlr stops in the middle
> of it's own initialization.
> -- Ragge


Or that it is being polled so often by the driver that the controller
is incapable of completing its initialization.  A month or so ago
Ken Wellsch related his diagnosis of this problem on Dilog controllers,
and gave the differences between his waterloo-uda.c that fixed this 
and the standard netbsd-uda.c.  I wouldn't be surprised if the
Emulex controllers require the same fix, as both the Dilog and
Emulex are microprocessor-based controllers.

Tim. (shoppa@altair.krl.caltech.edu)