• State Resolved
  • Locked Locked
  • Replies 2 replies
  • Subscribers 98 subscribers
  • Views 467 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.

PCIe MSI-Interrupt with different events

Bernd Erbe
Expert 2065 points

Hello,

I'm using a TMDXEVM6678LE with a AMC-to-PCIe-Adaptor Card. I've tested successfully the functionality of generating an MSI-Interrupt by the DSP (EP) to the OS (RC). This works for example by writing the data word 0x4191 to the PCIe-Address 0x6000f00c.

First problem is, that in the MSI0_IRQ_STATUS no bit is set. How is this done? I'll check tomorrow if one of the other MSIx_IRQ_STATUS is set, but the MSI0_IRQ_STATUS is zero -> so no acknowdlege can be done.

Second problem is that I want to signal with the same interrupt another event. With one MSI-Interrupt it is possible to generate 4 different events (like described in the PCIe-Doku) by changing the LSBs of the data word. But this doesn't work. Changing the data word 0x4191 to 0x4198 or 0x4191 no IRQ arrives at the RC.

Is it the right way to modify the data word or is there a misunderstanding? How can I achieve to generate the IRQ with another source?

Best Regards,
Bernd

  • 0
    TI__Genius 12065 points

    Bernd,

    1. MSIn_IRQ_STATUS is being set if the DSP RECEIVES MSI interrupt. Since DSP is generating the MSI interrupt to OS/RC, you can check the similar status register on the OS/RC side.

    2. I am not sure how the the data word (0x4191) and PCIe address (0x6000f00c) are assigned for generating the MSI interrupt from DSP. In C6678, the MSI vectors are grouped with step of 8, such as MSI0 includes vectors 0,8,16,24 and MSI7 includes vectors 7,15,23,31. For example, if you can use 0x4191 to generate interrupt, can you check with 0x4191+8=0x4199 instead of 0x4198 please? It is better to know how the OS/RC handles the MSI vectors/interrupt as well.

  • 0 in reply to Steven Ji
    Expert 2065 points

    Hello Steven,

    thanks for your answer. To be similar to your other PCIe-Components, I will define a new status register for IRQ-reason and acknowdlege.

    To 2.) I also tried it with 0x4199, but it also hasn't worked. But with the new status register, we're free to signal the IRQ-reason

    Regards,
    Bernd