XIO2001: Boot config failed

Hi Darren,

It is a bit difficult to say where the issue lies at this time. Could you please clarify a couple of items for me:

1. Upon initial boot-up and configuration of the XIO2001, is the PCI device able to be recognized properly?
2. Has the customer's designed followed the recommended XIO2001 Implementation Guide?

Best,
David

Hi David,

Could you please move this topic to public?

I want customers to participate directly in the discussion or I'm opening another public thread.

Thank you.

Darren

Hi Darren,

The thread is now public.

Hello TI expert,

 Here is the Block diagram:

A:  Our customer ‘s  PCIe card： XIO2001+DM648 DSP, which works OK for years on various x86 platforms. (However, A is almost a black box for us as it is not our design, we are not able to get any logs)

B: one supplier’s motherboard. A+B is OK

C: our motherboard.  A+C is NOK ， but if we put B’s BIOS to C, A+C is OK, so there must be some BIOS changes needed for C, but as we don’t know how to modify the BIOS , can TI give some suggestion based on your rich experience?

 Both B&C are Intel Whiskey Lake U platform

Detail fail symptoms for A+C

1. When windows OS is first on  with relative driver/SW  installed , thinks are OK
2. But when our customer try to reload the logical file for the DM648 DSP , issue happens, the PCI device （DSP DM648） is  abnormal , details check show the BAR of DM648 has been corrupted/modified（see the following picture）

       3.If we do the action of Disable/enable XIO driver(or DSP driver)  under windows OS , issue on step 2 will disappear, the DSP will be OK again

       4.This issue will not happen with A+B or A+C(when C with B’s BIOS chip instead)

Thanks

Eric Wang

What happens to the DSP when you load the DSP file? If the DSP resets itself, then from the point of view of the host, there is suddenly a new PCI device. Is your system (hardware, firmware, software) capable of handling hot-plugging?

Sorry but it’s the reality that the DSP is a black box to us and we have no way to know what exactly happens, we have tried to enable the PCIe Hot-plugging support in the x86 BIOS, but the fail symptoms does not change.

I know current information is difficult for TI to address the root cause, but if TI can give some more suggestion based education guess it will be highly appreciated

Thanks

Eric Wang

Hi Eric,

I believe to fix this issue, it needs to be known what is happening to the DSP when you load the DSP file.

It seems that the BIOS from B's board works with C's board, is it possible to continue to use this BIOS on C's board? This seems to be a software issue, not necessarily an XIO2001 issue, as the XIO2001 device has been working for quite some time in the same hardware configuration.

Best,
David

Hi Eric,

Thank you for confirming that you have solved this issue.

Best,
David