Re: LOCK# accesses

[Danny Pierini <Danny.Pierini@matrox.com>]

That's what I'm trying to do.  I have a system that generates LOCK# cycles 
and I'm trying to figure out why & how it does that.  My goal is to get rid 
of the LOCK# cycles.

BTW, P2P & host bridges are required to support LOCK# cycles (legacy issues).

Danny


At 10:04 AM 11/23/2001 +0530, Monish Shah wrote:
> > Hi.  Can someone educate me as how to generate a LOCK# read access to my
> > PCI device from software, ie: how does a software command generate a LOCK#
> > read access on the PCI bus.  The device I am reading from is behind a P2P
> > bridge.
>
>The best advice on how to generate LOCK# is: "don't".  It is really bad for
>system performance.  And, it is a pain for the host to support.  Not all
>hosts support it.
>
>Also, I have a vague recollection that P2P bridges are not required to
>support LOCK.  I may be wrong about that, though.
>
> > Thanks
> >
> > Danny
>
>Monish

----------------------------------------------------------------
Danny Pierini                   email:  Danny.Pierini@matrox.com
Hardware Architect              tel:    (514) 685-7230 ext 2788
Video Products Group            fax:    (514) 822-6024
Matrox Electronic Systems       WWW:    http://www.matrox.com

Head Office:                    US Mail:
1025 St. Regis                  625 State Route 3, Unit B
Dorval, Quebec, Canada          Plattsburgh, NY, USA
H9P-2T4                         12901-6530
----------------------------------------------------------------