TMS320F28388D: CM Configurable Watchdog timer

Hi Hareesh Janakiraman

Thanks for the answer, I was able to solve my problem.

All I was missing was - WDPRECLKDIV and WDPS in the WDCR register configuration, I was not configuring these register values and these register values defaults to WWD expiry in uSec and Hence I was unable to control the reset time. And once I configured above values I was able to control and service the WWD. Thanks for your inputs.

Also after WWD expiry it is generating a NMI to the CMNMIWD module and in CM FW  i'm intentionally not handling/clearing/servicing the NMI interrupt.

Implementation 1 : At this point : In this case, As NMI is getting generated on CPU1 aswell, we've added logic in CPU1 application FW to reset total 3 controllers and it is working as expected.

Implementation 2: But now we dont want this to happen, we want only CM core to get RESET independent of CPU1. 

So I will not enable NMI on CPU1 (or) even if enabled I will service NMI in CPU1 and be idle there (I wont reset), And in CM, as WWD NMI is not serviced., After the WDCNTR will overflow and reset the CM. CORRECT ?

After reset, Do CPU1 again need to BOOT CM as it does during the regular boot-up time using Device_bootCM() function ?

To emulate this, My Custom application is already running as TCP server so I will start PINGING continuously and once WWD expires I will not service WWD making it reset CM core and Hence I should be able to observe PING lost/timed-out right ? 

Implementation 3: I'm trying to use AIRCR Register (Offset = D0Ch) [Reset = FA050000h] to reset CM core based on learning from TRM,

To do this, I'm using below method:

void AIRCRCReset()
{

    uint32_t VECTKEY = 0x5FA00000;
    VECTKEY |= 0x04;
    HWREGH(WD_BASE + 0xD0C) = (HWREGH(WD_BASE + 0xD0C) | (VECTKEY) );

}

But I see that CM is not getting resetted as there is PING always.

Please correct me what I'm missing and suggest me a way to achieve our requirement.

Thanks and regards,

pranay.

Hi Pranay, unfortunately Hareesh is out of the office.  I am trying to ID another subject matter expert who can assist, but worst case, you should hear back from Hareesh by the end of next week.

Hello Joe,

Thanks for reply. Please assign another subject matter expert.

Vivek Singh  Can you please help me out here.

Thanks.

Hi Pranay,

pranay r said:

Implementation 2: But now we dont want this to happen, we want only CM core to get RESET independent of CPU1. 

So I will not enable NMI on CPU1 (or) even if enabled I will service NMI in CPU1 and be idle there (I wont reset), And in CM, as WWD NMI is not serviced., After the WDCNTR will overflow and reset the CM. CORRECT ?

Yes, this is correct.

pranay r said:

After reset, Do CPU1 again need to BOOT CM as it does during the regular boot-up time using Device_bootCM() function ?

Yes, it's needed. Please refer the CM BOOT flow in device TRM (below snapshot)

pranay r said:

To emulate this, My Custom application is already running as TCP server so I will start PINGING continuously and once WWD expires I will not service WWD making it reset CM core and Hence I should be able to observe PING lost/timed-out right ? 

I would say yes, if CM boot is not setup properly by CPU1.

pranay r said:

Implementation 3: I'm trying to use AIRCR Register (Offset = D0Ch) [Reset = FA050000h] to reset CM core based on learning from TRM,

To do this, I'm using below method:

Fullscreen

1

2

3

4

5

6

7

8

void AIRCRCReset()

{

uint32_t VECTKEY = 0x5FA00000;

VECTKEY |= 0x04;

HWREGH(WD_BASE + 0xD0C) = (HWREGH(WD_BASE + 0xD0C) | (VECTKEY) );

}

XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX

void AIRCRCReset()
{

    uint32_t VECTKEY = 0x5FA00000;
    VECTKEY |= 0x04;
    HWREGH(WD_BASE + 0xD0C) = (HWREGH(WD_BASE + 0xD0C) | (VECTKEY) );

}

But I see that CM is not getting resetted as there is PING always.

This does not look correct. VECTKEY value should be 0x05FA. You have it as 0x5FA0. Also you are doing read modify reset. As per spec read value for VECTKEY is 0xFA05 so when you do read modify write this will get updated to different value. Why not just write the fix value. Also can you use SYSRESETREQ bit instead.

Regards,

Vivek Singh

Hi Vivek,

Thanks for the reply.

Case 1:

I tried below 3 tries for resetting CM through AIRCR register but all the time I'm unsuccessful in doing so.

void AIRCRCReset()
{

//Try 1:
    //
    // Write the predivider to the appropriate register.
    //
    uint32_t VECTKEY = ((0x05FA<<16) | 0x04);
//    HWREGH(WD_BASE + 0xD0C) = (HWREGH(WD_BASE + 0xD0C) | (VECTKEY) );
    HWREGH(WD_BASE + 0xD0C) = VECTKEY;//Not Working
    
//Try 2:
    uint32_t VECTKEY = ((0x5FA<<16) | 0x04);
    HWREGH(WD_BASE + 0xD0C) = VECTKEY;//Not Working

//Try 3:  
    uint32_t VECTKEY = ((0x5FA<<16) | 0x04);
    HWREGH(WD_BASE + 0xD0C) = (HWREGH(WD_BASE + 0xD0C) | (VECTKEY) ); //Not working
}

I'm confirming you that I'm not seeing any reset is by :

1. I've kept a LED blinking continuously in CM main while(1) Application FW  -> I've not seen this stopped when I execute above function by sending an API over TCP..

2. By keeping continuous PING and checking if there is any PING lost -> There is no PING lost when I execute above function.

FYI: I'm not running in debug mode. I'm flashing FW all the time and sending commands over TCP.

Need help resetting CM core using this register access method...

Case 2:

//
// NMI ISR
//
 __interrupt void cmnmiISR(void)
{
//     bool wdstatus;
//    //
//    // Get the WWD status and clear it
//    //
//    wdstatus = SysCtl_getWatchdogStatus();

    Tx_cpu1_intr(0xE2);
    Tx_cpu2_intr(0xE2);
    
//    if(wdstatus)
//        SysCtl_clearWatchdogStatus();
//
//    //
//    // NMI was handled after the WWD expired
//    //
//    nmiflagstatus = SysCtl_getNMIFlagStatus();
//    SysCtl_clearAllNMIFlags();

}
//
// NMI ISR
//
 __interrupt void cmnmiISR(void)
{
//     bool wdstatus;
//    //
//    // Get the WWD status and clear it
//    //
//    wdstatus = SysCtl_getWatchdogStatus();

//    Tx_cpu1_intr(0xE2);
//    Tx_cpu2_intr(0xE2);
//    if(wdstatus)
//        SysCtl_clearWatchdogStatus();
//
//    //
//    // NMI was handled after the WWD expired
//    //
//    nmiflagstatus = SysCtl_getNMIFlagStatus();
//    SysCtl_clearAllNMIFlags();

}

This is NMI ISR of CM. This gets executed every time when WWD expires and as I'm not serving the interrupt it supposed to be resetted. Yes in this case I see that LED blinking is stopped and PING also LOST. Means Resetted.

 As you can see in above snippet I tried writing IPC data to CPU1 so that it can execute "Flash_releasePumpSemaphore();" (thought it'd be required just in case) and  "Device_bootCM(BOOTMODE_BOOT_TO_FLASH_SECTOR0); I see that IPC has been received and executed commands aswell.

But I see that as CM is not getting booted after executing Device_bootCM();   function....

Not sure what I'm missing here. Anything else that I'd need to perform ?

Thanks and regards,

Pranay

Pranay,

Why are you writing at this location (at WD base address) ?

WD_BASE + 0xD0C

This should be written to NVIC_BASE location like below -

NVIC_BASE + 0xD0C

See below snapshot -

Please correct that and hope that fixes this issue.

Vivek Singh

HI Vivek,

Case 1:

I tried this but I dont see CM restarting. Can you please try this on your end and let me know how it goes.

Case 2:

Whenever WWD expires I see PING is stopping, But I dont think so CM is rebooting or restarting from SSBL. It is not turning on at all.

Do I need to execute any commands in CPU1 to start booting CM apart from Device_bootCM() I'm already executing this function but dont see CM restart ? Do we any API to restart CM from CPU1 ? even this would help us.

Regards,
Pranay

I am assuming now CM is getting reset. Please confirm. Also please note that when WWD expires, it only fires NMI on CM and after NMIWD counter expires, it fires reset.

Can you read the value from CM IPCBOOTSTATUS register and see what is the value there ? 

HI Vivek,

Vivek Singh said:

Also please note that when WWD expires, it only fires NMI on CM and after NMIWD counter expires, it fires reset.

As main while(1) is not running and PING is lost, I believe CM got restarted because of WWD expiry (case 2), But the question is how do I boot CM again from SSBL ?

Case 1:

This one I'm unable to achive, can you please help me figure this one out ?

--pranay

pranay r said:

Case 1:

This one I'm unable to achive, can you please help me figure this one out ?

Have you corrected the address, as I mentioned in my last post.

For CM Boot, I'll loop-in BOOR expert.

Vivek Singh

Hi Vivek,

Regarding case 1:

Yes, I have corrected address as you suggested. Still I see same behavior.

--pranay

For CM BOOT issue, I am going to assign this to our BOOTROM expert. On the reset case, we'll check and get back to you. 

Regards,

Vivek Singh

Hi Vivek,

Can you please assign to your bootrom expert to this thread and try to solve my issue.

Thanks and regards,

pranay

Let me remined him again.

Hi Pranay,

The bootROM expert for F28388D is out of office and will be back next week. Please expect a reply by mid of next week.

Meanwhile can you check if you are entering CM boot or not by selecting CM wait boot mode. Add CM ROM symbol from C2000Ware\libraries\boot_rom\f2838x\revA\rom_sources\cm\ccs_files\Release and point to source files in C2000Ware\libraries\boot_rom\f2838x\revA\rom_sources\cm\F2838x_ROM\bootROM\source. Then you should be able to debug the ROM boot flow.

Thanks and regards, Pawan