XIO2001

The detection of 32 PCI devices is now solved… The 32 devices were detected as we didn’t connect the IDSEL line to one of the ADx lines, so in configuration phase the OS was detecting 32 devices.

The only problem we are observing right now is that interrupt do not seem propagated from the INTAN input line of the bridge to the PC.. Even if INTAN is asserted low, our PCI SW driver does not receive any interrupt.

Communication-wise we have not seen any problem on what tested till now… Target transactions are properly translated by the PCIExpress-to-PCI bridge

YongJie,

Can you verify that the interrupts are being passed to the PCIe bus via the use of a PCIe bus analyzer?

Usually these cases are caused by issues with the host drivers, but it would be helpful to ensure that the interrupt is actually being signaled via PCIe.

Best,

-DK

Hi DK

I don’t have a PCIe bus analyzer, and I am not inclined to buy one as we are not going to do PCIe development. If u have some software that can do the job, please provide it to me…

Anyhow, I have more interesting evidences to complete the observation…

We have tried to use the INTAN INTBN INTCN INTDN and we have verified that those signals were properly asserted on the PCI bus…

Strangely enough only the INTDN is producing an interrupt at application level, while INTAN, INTBN and INTCN are not.

Same behavior have been observed on both desktop PCs (on 1x, 8x and 16x PCIe slots) and laptops.

Do you have a possible explanation for this or (ideally) do you have an alternative driver for the PCI bridge beside the standard Windows driver that I can try out?

thanks and regards,

YJ