Detected Parity Error bits on multi-function devices

[frank.story@tempe.vlsi.com]

I have a corner case sort of a question about bit 15 in the 
status register. This bit should be set whenever a parity 
error is detected by a device.

The description I'm having trouble nailing down is the second 
to the last paragraph of Section 3.8.1, which states in part 
"Agents that support parity checking must always set the Detected 
Parity Error bit in the Configuration Space Status register 
(refer to Section 6.2.3) when a parity error is detected."

In the case of a multi-function device, each function has its 
own status register with its own Detected Parity Error bit in 
bit 15. Its pretty clear that any address parity error should 
set bit 15 of the status register in each and every function.

Its not so clear that a data parity error should do the same. 
I can't find a requirement that detection of data errors is 
required by targets that are not selected. It seems more 
intuitive that only the selected target would detect data 
errors. Therefore, I believe that data parity errors should 
set status bit 15 only in the target that returned DEVSEL#. 
However, the way I read the spec, *all* targets should set 
status bit 15 in the event of data parity errors. In terms 
of the multi-function device, any data parity error should 
set bit 15 of the status register in each and every function.

I do recognize the fact that this wouldn't affect the need 
for a seperate status bit 15 for each function. The ability 
of software to clear the status bits individually must be 
preserved. I'm just not sure whether the set mechanism can 
be common.

So, do I set bit 15 of all status registers on data parity 
errors, or only on the selected target?

Thanks,
Frank


Frank Story                    frank.story@tempe.vlsi.com
VLSI Technology                602-752-6098
Computing Products Group
LèØ