Experimental Physics and
Industrial Control System

[email protected] (Jeff Hill) · Wed, 13 Sep 2000 08:48:35 -0600

Overheard this today on the WRS mail exploder

> -----Original Message-----
> From: the vxWorks Users Group Exploder [mailto:[email protected]]
> Sent: Wednesday, September 13, 2000 8:12 AM
> To: [email protected]
> Subject: RE: data access exception in net task
> 
> 
> Submitted-by [email protected]  Wed Sep 13 07:05:55 2000
> Submitted-by: Jim Way <[email protected]>
> 
> James,
> 
> > Running on PowerPCs (2431 and 2603) sending/receiving data TCP as fast
> > as possible runs fine for about an hour or so and then I get one of the
> > following conditions:
> > 1) Receives of packets stop happening - DMAs into my receive buffers
> > just seem to stop.
> > 2) A data access exception.  The printout points to the net task.
> > 3) The machine just locks up.
> > 
> > Looking at mbufShow(), memShow(), spy() and my driver status don't show
> > anything that would indicate a problem.
> 
> I'm using PPC860T with VxWorks 5.4. Earlier this year, I had a similar
> sounding problem. After "a while" my Ethernet would stop talking. There were
> no apparently obvious symptoms, just like you're describing.
> 
> The problem was fixed by hiring a WRS consultant who applied two SPR fixes
> (and did not apply another seemingly applicable one). Here is a snippet of
> his email to me:
> 
> > I have integrated the fixes for the following :
> > 
> > SPR 22325 - Race condition between interrupt processing of received
> packets and polled mode processing in END driver
> > SPR 23528 - motCpmEnd driver corrupts tNetTask in system mode
> >
> > I have not integrated the fixes for the following :
> > SPR 21143 - motCmpEnd END driver improperly filters incomming frames when
> promiscuous mode is enabled.
> 
> Now, I'm not saying that these specific SPRs are related to your problem.
> Just that my experience in this area suggests that browsing WindSurf for
> SPRs might be helpful. Your local FAE might also be an excellent resource
> for this problem. I know that mine is a top-notch techical guy. Depending on
> your expertise with VxWorks (mine was very limited at the time, now I'm an
> old-hand <ha ha>) you might want to consider using their consulting
> services. We paid for three days of support and one on-site trip, and I
> think my management all believe it was money very well spent. I know that I
> had other things to work on during that time.
> 
> HTH,
> Jim
> -----------------------------------------
> Jim Way, Software Engineer
> Datum Austin (Austron Inc.)
> voice: 512.721.4170
> fax  : 512.990.9712
> email: jwayATdatumDOTcom (no spam please)
> -----------------------------------------
> 
> 
> 
> 
> **********
> 
>     This is a user group mailing list for vxWorks related topics
>     see http://csg.lbl.gov/vxworks for posting and other info

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System