[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: PCI CLKS disabled relating to FPGA controllers ??




Marco,

For the debug case, check out the following procedure:

     1.  Power-on unit - FPGA not recognized by host, and clocks in bridge
are turned off.
     2.  Configure your FPGA using the JTAG port - It was you that had the
JTAG initialization, wasn't it?
     3.  Reset the host (reset button) - This should cause the host to go
out and re-intialize the PCI bus again, but this time the PCI device  will
be recognized (as long as a system reset doesn't cause your FPGA to lose
it's contents)

This was the procedure I plan on using.  Is there some reason why this
won't work?

     Tom Yunghans
     Rockwell Collins
     Pomona, CA
     (909) 868-2694





"Marco Brambilla (Wireline)" <marco-tpa.brambilla@st.com>@eux100.sgp.st.com
on 06/29/2000 09:22:22 AM

Sent by:  marco@eux100.sgp.st.com


To:   pci-sig@znyx.com
cc:   pci-sig@znyx.com

Subject:  Re: PCI CLKS disabled relating to FPGA controllers ??


Hi everyone, and thanks to all thos (you were really lots) that dropped
their mails.

To recap the latest findings:

- the newer motherboards (actually the PCI2PCI bridge) can disable the
clock for individual slots if the slots themselves are not occupied. The
bridge usually has N clk outputs, one for every slot, so no additional
logic is required. An example of this is
http://developer.intel.com/design/bridge/datashts/27810601.pdf

(thanks to Leif Erik Laerum for pointing me to the file)

In our specific case we have a device that during the debug phase cannot
be programmed until after the PC finishes the boot phase.

At power on reset the bridge chip will reset the clock disable bits,
then it will be the CPU that will do a bus scan and disable the slots
where no devices are present.

This reset is only done at power on, so in order to have your clock back
you cannot simply do a reset of the machine, but you must write a
register in the config space of the pci bridge.

When detection fails for a production FPGA device that could mean that
either
- the PC doesn't wait for the pci 2.2 one second "boot" period for pci
devices (on our board the clock is seen present even on empty slots - no
boards at all - for some seconds, where some was not precisely measured,
but is 2 - 5.

- the fpga takes longer than 1 second to load after power good



Our last problem is that although we have a pci bus excerciser, we
cannot access the clock enabling register, because the configuration
space of the bridge can only be accessed from the primary bus.

So we now just need to understand how to access, via BIOS or whatever,
the configuration space of the pci bridge.

Thanks for your support and patience.

Marco.
--
--------------------------------------------------
 Marco BRAMBILLA

 STMicroelectronics
 Via C. Olivetti, 2
 20041 Agrate Brianza (MI)
 ITALY

 TPA - Wireline Communications Division
 tel   : +39 039 603.5064   (ST Agrate - TINA 050)
 tel   : +33 (0)4 7658.5063 (ST Meylan - TINA 041)
 fax   : +39 039 603.5060
 mailto:marco-tpa.brambilla@st.com
--------------------------------------------------