Actions
Forum Thread Details
  • State Not Answered
  • Replies 2 replies
  • Subscribers 328 subscribers
  • Views 1037 views
  • Users 0 members are here
Related

Can I use XScuGic and XScuTimer at the same time?

Former Member

Hi, all,
  I have an IP generated by vivado hls as the accelerator of my project and I also use the scu timer to get the execution cycles of other functions which are implemented in software. The hardware IP contains the interrupt interface.

  I find that if I seperate the accelerator and test it alone, it behave correct. (I do not use timer here). However, When I put
it into my whole design, the accelerator always busy waiting. My whole design use scu timer to estimate the performance of different functions.

  Is it the problem that I can not use scu timer and interrupt (when hardware finished, it will generate an interrupt) at the same time?  (PS: I do not mean to implement a scu timer with interrupt feature)
  I am using Zedboard and Xilinx 14.5

  Any one could give me some advice?

Thanks

Parents
  • 0

    I use both in my application. As you said, the GIC is the Interrupt Controller, and stores which ISRs to call when an interrupt is triggered.

    Also, be aware that the ScuTimer is a down counter. Perhaps you can use global timer - then you won't have to worry about expiring timers AND it's an up counter AND it's a 64 bit counter AND it's the same frequency as the ScuTimer. See xtime_l.h for info.

    There's a few things you'll have to do to enable interrupts in your application:

    To enable interrupts through the GIC:
    1) XScuGic_CfgInitialize (The device ID can be found in xparameters.h)
    2) Xil_ExceptionInit (Initialize the exception vector table)
    3) Xil_ExceptionRegisterHandler (Just register the XScuGic_InterruptHandler if you wish - it's already written for you)
    4) Xil_ExceptionEnable() (Enable interrupts on the Zedboard)

    From here, the GIC is ready to call ISRs for whichever device sets their interrupt flag. It sounds like you need to figure out the interrupt ID of the accelerator and register that to the GIC.

    To enable interrupts for a device:
    1) Configure the device for interrupts (this might involve setting a bit to allow interrupts in the device - most xilinx third party libraries have functions written for this; for the XScuTimer it is XScuTimer_EnableInterupt - who knew!?)
    2) Register an interrupt handler to the GIC, so it knows what ISR to call when the interrupt has fired. Do this with XScuGic_Connect
    3) Enable interrupts for that specific deivce using XScuGic_Enable (notice that the second parameters for XScuGic_Connect and XScuGic_Enable must match).

    From there you should be good to go. If you don't register a handler to the GIC for a device that has interrupts enabled, then it will call the default handler, StubHandler, found in xscugic.c, which will do nothing except increment the UnhandledInterrupts variable in the XScuGic struct and if that struct has not been properly initialized, Xil_AssertVoid will probably go into an infinite loop.

    Hope this helps.

Reply
  • 0

    I use both in my application. As you said, the GIC is the Interrupt Controller, and stores which ISRs to call when an interrupt is triggered.

    Also, be aware that the ScuTimer is a down counter. Perhaps you can use global timer - then you won't have to worry about expiring timers AND it's an up counter AND it's a 64 bit counter AND it's the same frequency as the ScuTimer. See xtime_l.h for info.

    There's a few things you'll have to do to enable interrupts in your application:

    To enable interrupts through the GIC:
    1) XScuGic_CfgInitialize (The device ID can be found in xparameters.h)
    2) Xil_ExceptionInit (Initialize the exception vector table)
    3) Xil_ExceptionRegisterHandler (Just register the XScuGic_InterruptHandler if you wish - it's already written for you)
    4) Xil_ExceptionEnable() (Enable interrupts on the Zedboard)

    From here, the GIC is ready to call ISRs for whichever device sets their interrupt flag. It sounds like you need to figure out the interrupt ID of the accelerator and register that to the GIC.

    To enable interrupts for a device:
    1) Configure the device for interrupts (this might involve setting a bit to allow interrupts in the device - most xilinx third party libraries have functions written for this; for the XScuTimer it is XScuTimer_EnableInterupt - who knew!?)
    2) Register an interrupt handler to the GIC, so it knows what ISR to call when the interrupt has fired. Do this with XScuGic_Connect
    3) Enable interrupts for that specific deivce using XScuGic_Enable (notice that the second parameters for XScuGic_Connect and XScuGic_Enable must match).

    From there you should be good to go. If you don't register a handler to the GIC for a device that has interrupts enabled, then it will call the default handler, StubHandler, found in xscugic.c, which will do nothing except increment the UnhandledInterrupts variable in the XScuGic struct and if that struct has not been properly initialized, Xil_AssertVoid will probably go into an infinite loop.

    Hope this helps.

Children