Reset Reason Status Register?

Hi Tom,

The only status flag I know of is the "WDFLAG" in the SysCtrlRegs.WDCR Register. Check out System Control and interrrupts Guide Section 3.4.5

Best regards

Andreas

Hi!

I have the same question. So far I found the WDFLAG, as Andreas said, and the MCLKSTS in PLLSTS  register (NMI Watchdog reset). Are there other flags you found?

Thanks,

Monica

Hi!

Thanks! I took a short look at that discussion. I wonder if the same happens with MCLKSTS. But until then, I must understand how you solved the issue. I'm not an expert.... yet :)

Regards,

Monica

 

You define and reserve a RAM segment in the linker such as (make sure that you adjust the RAM segment to remove what you are adding)

MEMORY

{

...

PAGE 1 :

...

HEALTHRAM :

origin = 0x0007F0, length = 0x000010 /* preserved RAM between resets  assume power to chip was not lost */

}

SECTIONS {

....

statusram : > HEALTHRAM PAGE = 1

....

 

In your code define a set of variables:

#pragma

DATA_SECTION(statushigh, "statusram");

#pragma

DATA_SECTION(statuslow, "statusram");

u16 statushigh;

u16 statuslow;

Add two functions:

u16 GetCookie(u16 *cookie)

{

u16 rval=0;

if ( statushigh == ~statuslow)

{

rval = 1;

}

*cookie = statushigh;

return rval;

}

void

SetCookie(u16 cookie)

{

statushigh = cookie;

statuslow = ~statushigh;

}

During power up you can check the status of the variable by calling GetCookie().  You can put snap shots of registers, or other things by calling SetCookie()

Hope this helps.

 

Thanks again!

So I could SetCookies() in the interrupts  (I use interrupt mode for both Watchdog and NMIWatchdog) and after reset I read the RAM to get the reason of reset.

Regards,

Monica

Yes.  Just be careful, since you need to ensure atomic access to the variables if you access in background (non interupt) code, you may need to add a disabble interrupt.  Or, rewrite in assembly to avoid read/modify/write.

If you only set in interrupt, and during a power up interrupts are off, you will be ok.

 

Tom

Thank you Tom!

Was really helpful!

I still have one question: what's the purpose of *cookie in GetCookie? I've always had troubles with pointers...

In my code I have defined:

Uint16 WD_reset;
Uint16 *WD_rst;

I don't know if it's correct, was the only way I could compile.

Then in INT:

WD_reset = any_value;
SetCookie(WD_reset);

And I check:

flag = GetCookie(WD_rst);

I see statushigh, statuslow get the correct value, the flag also. But WD_rst is a memory location which I cannot read. In the debugger appears: Memory map prevented reading on target memory.

I'm not sure I read the code correctly, but I thought at the memory address WD_rst I  would find 'any_value'.

I've tried to write:

WD_rst = &WD_reset; // GetCookie would overwrite WD_reset with statushigh, right?

But I get a compilation error.

Regards,

Monica

Monica, you want to pass an address to the function as follows, (note example of how I'm using this)

u16 rval;

u16 cookie=0;

rval = GetCookie(&cookie);

    if ( rval == 0)
    {
    	SetCookie( COOKIE_POWERON);

    }

Tom

Thanks a million Tom!

This was of GREAT help! Now I can have a different code for each reset cause ( so far, WD reset, NMI reset and power-on). For NMI reset I've posted a question http://e2e.ti.com/support/microcontrollers/tms320c2000_32-bit_real-time_mcus/f/171/t/172402.aspx#629940 , but I didn't get an answer yet. Have you worked with it too?

Regards,

Monica

sorry I haven't worked with this. if your use an external crystal you can instrument a board so you can pull the crystal out during runtime. I've done this before.

In the moment I use the ControlCard as provided by Texas, without any crystal. But anyway, the procedure you showed me works great with NMI reset too.

Regards,

Monica