• State TI Thinks Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 2 answers
  • Subscribers 100 subscribers
  • Views 646 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TMS320C6205: PCI lockup or PC down issue..

Ernest Cho
Expert 3341 points

Part Number: TMS320C6205

Hello team,

One of my factory automation customer have an old PCI platform which is developed using TMS320C6205 have an issue regarding PCI driver loading between C6205 customer board and Desktop PC.

Sometimes, the issue causes C6205 board locked during PC booting or PC is locked when trying driver initialization.

 

For this issue, they checked below :

-       They have seen “C6205 Errata” but no related issue found.

-       They have seen “TMS320C6000 DSP PCI Reference Guide(SPRU581C)”, but it is hard to get their issue related points.

 

But since C6205 does not have any reference schematic and does not have much information on the web, so they request of PCI related documents.

Would you please to support PCI materials as customer requested below?

-       TI TMS320C6205 PCI related schematics

-       Any reference documents for PCI trouble shoot (not PCIe).

 

Best Regards,

Ernest Cho

  • 0
    TI__Expert 3341 points
    Hello team,
    During their test evaluation, they split it as two issues and under testing with their workaround.
    1. PCI reset
    They thought PCI reset sequence is mainly caused their current issue. so they modified to do POR reset when PCI reset is asserted.
    Could you please check and confirm it would be a possible workaround?
    2. PCI 3.3V_AUX
    They found some PCI slot in the PC side do not support PCI 3.3V_AUX power, so they would like to connect 3.3V and 3.3V_AUX power source directly.
    Is it possible to do in the application level?

    Hope your great supports with fastest response.
    Thank you.
    B.R.
    Ernest Cho
  • 0 in reply to Ernest Cho
    TI__Expert 3341 points
    One more additional question.

    TMS320C6205 device specified to support PCI Rev2.2. Is this means backward compatible with PCI Rev2.1?

    Thank you.
  • 0 in reply to Ernest Cho
    TI__Mastermind 46460 points

    Ernest,

    As you know this is a very old device and our ability to support it is very limited.  We highly recommend that new implementations be done with newer DSPs or Processors from TI.  Full software and hardware support is available on more modern devices that cost less and run much faster.  Additionally, the assembly or C-code written for the C62xx devices will easily port to newer devices.

    You surmised that reset or power may be contributing to the issue.  Either or both might be causing problems.  Have you done any experiments with the proposed solutions?  I know with PCIe that reset response can cause enumeration failures and/or cause system hangs.  Do they see any change in behavior by connecting the PCI Reset to the POR on the C6205?

    Main 3.3V and 3.3V_AUX are meant to be separate and sourced at certain times.  Obviously, different motherboards may or may not follow all of the recommendations of the standards.  It is reasonable that sourcing the board circuitry for both domains from the main 3.3V pins may work in some cases.  However, it is a spec violation and cannot be used in all installations.  Your board would need to be customizable.  Have you see this solve the problems in your testing?

    Tom

  • 0 in reply to Tom Johnson 16214
    TI__Mastermind 30755 points
    Ernest,

    There appears to be no further communication on this, I assume the issue has been resolved.
  • 0 in reply to PaulM
    TI__Guru*** 125430 points
    I've seen some issues with old c6000 designs where the vectors.asm file had an issue. Specifically, the reset vector inside vectors.asm was pushing a register to the stack before writing c_int00 address into it. The problem is that at this point the stack register is not initialized and so you end up writing to a random address. I've seen that result in very early boot lockups. Similarly, I've seen accesses to specific address ranges lockup the PCI bus. If nothing else, I recommend looking at your vectors.asm to make sure this issue is not in your reset vector.