Re: RE: [pci] SOS: about turnaround cycle

[wuyunsheng <wuyunsheng@mprc.pku.edu.cn>]

thanks a lot for your timely response. now let me describe what i meet in more details:

my chip is the default bus owner which will drive the PCI bus when no others drive it. for example, my chip drives 0x00000000 to AD( r these the idle signals u mentioned?) . Now the other master get the grant to to start transaction, for example, it drives 0xffff0000 to AD. due to design mistake, there is no cycles which all devices stop driving PCI bus. that is :
cycle1: AD=0x00000000(my chip drives), FRAME# is pulled up on system board(no driver)
cycle2: AD=0xFFFF0000, FRAME# is drived to 0(the other master drives)

will this work?

Best Regards

>Hi
>
>I would warmly recommend you to change that, but as you wrote, you don't
>have much time.
>So, to understand you better:
>
> - turn-around cycles are there, but instead of "no bus driver" your chip
>still drives bus for one more cycle
> OR
> - turn-around cycles are there, but instead of "no bus driver" your chip
>starts driving bus with "idle" signals
> OR
> - turn-around cycles are left out, what means that your chip starts driving
>bus one cycle too soon (every PCI compatible device expects turn-around
>cycle at this time - what about your device?)
>
>Well in the last case, if your chip expects active signals that soon, then
>"PCI" bus with your chips will work, they just might not live that long due
>to possible bus contentions.
>
>I hope this helps.
>
>Regards,
>	Tadej
>
>
>
>-----Original Message-----
>From: owner-pci@opencores.org [mailto:owner-pci@opencores.org]On Behalf Of
>wuyunsheng
>Sent: 20. marec 2003 7:38
>To: pci@opencores.org
>Subject: [pci] SOS: about turnaround cycle
>
>
>hi all,
>
>my design containing a pci core(i design it myself) will be taped out soon.
>but just now i found out that in some circumstances there are no turn around
>cycles on AD and CBE signals when the bus driver changes. since it is on
>schedule to tape out, i am anxious to know whether i must modify rtl code to
>fix it.
>
>any advices are deeply appreciated.
>
>best regards.
>
>sumnow
>
>--
>To unsubscribe from pci mailing list please visit
>http://www.opencores.org/mailinglists.shtml
>
>
>--
>To unsubscribe from pci mailing list please visit http://www.opencores.org/mailinglists.shtml
>.


            wuyunsheng
            wuyunsheng@mprc.pku.edu.cn

--
To unsubscribe from pci mailing list please visit http://www.opencores.org/mailinglists.shtml