Usb mass storage Hard fault Handler?

[its]

Posted on December 03, 2012 at 06:42

hi:

    after receiving DCMI data i am writing it to USB. after writing some files Hard fault handler occurs... why is it so?

Firstly I capture the DCMI data to buffer, after one frame capture I stop the DCMI and start to transmit data over serial. I repeat this in while(1) loop. e.g

              while(1)

     {

      frame_capture();

      //when frame is capture

      Send_usart(Buffer);

     }     // it works fine it keep on running

secondly, I capture a single frame, then write it on USB mass storage device, repeatedly. e.g like this

 

  frame_capture();

  while(1)

  {

     fopen(......);

    fwrite(buffer);

     fclose(....);

  }   // it works also fine. it keep on running

but whne I combine both codes , Hard fault Handler occurs. like this

while(1)

{

  frame_capture();

     fopen(......);

    fwrite(buffer);

     fclose(....);

 

}  // i get error after random file write after at 15, at 90 at 110  etc

  // some time i even go 400 write file but then fault handler occurs.

why is this happening? I also have set interrupt priority differently. but no effect !!

thanks

#usb-mass-storage #hard-falt-hanlder #dcmi

[Tesla DeLorean]

Posted on December 03, 2012 at 15:39

Examine the stacked state, registers, core, and then look at the faulting instruction(s).

Use a Hard Fault Handler that displays/decodes the failing state.

At a guess I'd say your stack size is insufficient.

[its]

Posted on December 04, 2012 at 06:36

hi:

 ''Use a Hard Fault Handler that displays/decodes the failing state.''

I have put a code to transmit a single byte over USART in Hard fault handler routine, thats why i get to know hard fault is occuring. whtat do you else mean by ''Use a Hard Fault Handler ''.

and how can I monitor stack status? what information should I updated here in post after debugging?

Can you guide me through this?

  

[Tesla DeLorean]

Posted on December 04, 2012 at 07:47

From Joseph Yiu's book and forum postings

http://blog.frankvh.com/2011/12/07/cortex-m3-m4-hard-fault-handler/

For the stack, start by seeing how big it currently is, and what your code is allocating. If you have several 512 byte buffers as local/automatic variables, and only a 1K stack allocation things will clearly go pear shaped. Understand your code.

Fill the stack with a suitable marker character, and see how much is used when it faults by looking at a memory view in the debugger. You can also write code to quantify this.

[its]

Posted on December 14, 2012 at 07:37

I am getting Hard Fault in my program occasionally.

=>could  the Clock provided to image sensor be the problem , i.e. some thing like synchronization between Pclk or Vsync etc.

=> could be the function or variable Scope be the problem?

if any of above statement can be the problem then why the error occurs at random time?

After debugging I came across the function during which hard fault occurs i.e frame_capture() . in this function I am just enabling DCMI and wait for a frame to be received , now the Hard Fault occurs always when I was leaving the frame_capture() function  .

like this ....

while(1)

{

 

     fopen(......);

   frame_capture();    //  Hard Fault occurs when I am leaving from this function

   Usart_send(usart2,'J');    

   fwrite(buffer);

     fclose(....);

 

}   

and

void frame_capture()

{

 ....

....

....

....

Usart_send(Usart2,'L');

}

 i receive L on the hyper terminal just before Hard fault handler, but not J. and when ever Hard fault handler occurs this happens always.

and the Core register values at the time of hard fault are

R0 0x40004400

R1 0x00000040

R2 0xffffffbf

R3 0x000000c0

R4 0x83177217

R5 0x86177316

R6 0x82127316

R7 0x20000a6c

R8 0x00000000

R9 0x200005bc

R10 0x00000000

R11 0x00000000

R12 0x50003000

R13(sp) 0x20014ba8

R14(LR) 0x080028eb

R15(PC) 0x080028ec

I get this values during debugging session. when hard fault occurs i stop the program and these values apears. I do it several times and one thing i noticed is R2 every time = 0xffffffbf.....

any help?

 

Regards

haris

[frankmeyer9]

Posted on December 14, 2012 at 08:27

I am getting Hard Fault in my program occasionally.

 

...

 

if any of above statement can be the problem then why the error occurs at random time?

 

That sounds suspiciously like stack overflow.

As interrupt handlers involve stack operations, the 'overflow' is driven by such events. And in case of external communication (like USB, ETH, UART), it appears to be random.

[Tesla DeLorean]

Posted on December 14, 2012 at 15:39

Look at the assembler code at/around the faulting instructions.

Look at the stack, and if it is in a region it should be.

Make sure you capture code isn't overwriting critical code or data. DMA can write all over things very easily.

[its]

Posted on December 19, 2012 at 06:03

actually I want a character array of length 5, that will increment like integer e.g

00001 will be equal to 0x30 0x30 0x30 0x30 0x31

and

00002 will be equal to 0x30 0x30 0x30 0x30 0x32

etc

name=['0','0','0','0','0']

I want this because file open function requires a string, a character array like this

if ((HCD_IsDeviceConnected(&USB_OTG_Core) != 1) || (f_open(&file, name, FA_CREATE_ALWAYS | FA_WRITE) != FR_OK))

is there any problem with that? where name is global character array initialize with '0's.

void name_rec()

{

  name[4]=name[4]+1;

  if(name[4]>'9')

   {

    name[4]='0';

    name[3]=name[3]+1;

   if(name[3]>'9')

    {

     name[3]='0';

     name[2]=name[2]+1;

     if(name[2]>'9')

      {

      name[2]='0';

      name[1]=name[1]+1;

      if(name[1]>'9')

       {

        name[1]='0';

        name[0]=name[0]+1;

        if(name[0]>'9')

         {

        name[0]='0';

        

      

       

         }

   

       }

   

      }

 

     }

   }

 }

i am calling this function after every file write.I got Hard fault Occasionally when I use name in f_open function. it might be that hard fault occurring due to some other reason (as  previous in this thread but somehow i manage to get rid of it).

 but on contrary, when I change character array name[] with ''haris'' my program runs well and no error occurs even after 1500 write, as below

if ((HCD_IsDeviceConnected(&USB_OTG_Core) != 1) || (f_open(&file, ''haris'', FA_CREATE_ALWAYS | FA_WRITE) != FR_OK))

where as I am still calling the function name_rec() in my  routine. (only change the string in f_open()).

what can be the problem?

[Tesla DeLorean]

Posted on December 19, 2012 at 15:21

what can be the problem?

NUL termination of the string in C?

[its]

Posted on December 20, 2012 at 08:11

sorry I didnt mention in my last post. I am already doing that with null character . like name[6]={'0','0','0','0','0','\0'}

if i use ''haris'' for file name it works fine, but with variable incrementing character array the fault comes.

but the fault is occurring on the same point as i describe in my second last post.

while(1)

{

 

     fopen(......);

   frame_capture();    //  Hard Fault occurs when I am leaving from this function

   Usart_send(usart2,'J');    

   fwrite(buffer);

     fclose(....);

 

}   

and

void frame_capture()

{

 ....

....

....

....

Usart_send(Usart2,'L');

}

 i receive L on the hyper terminal just before Hard fault handler, but not J. and when ever Hard fault handler occurs this happens always.

1-are the registers r0 r1...r12 are stack register? if so, then I have put marker addresses in these registers and at the time of fault occurring some of the stack registers are not changed.(is this mean that i have no problem of stack over flow?)

2- how can i came to know faulting instruction? after the hardfault occurs my program goes into infinite while,  if i  stops the debugger and see my pointer in dissasembler it correspond to pointer in C file at while(1), now if i go up in dissasembler the cursor in ISR code of ----_it.c goes up with it on one to one correspondence.

I want to ask that if I goes up further in dissassembler then does it mean that the instruction above the ISR will be the instruction from where this interrupt generated?

3-the link about hardfault(the from frank wespage) , i tried it, but I can not manage to incorporate it with stm32Fxxx.s

 i got error about opcoeds like handler_args is not defines etc how can i use that function in keil?

4- how can I detect the faulting insrtuction?

5-my capture code is not overwriting,as i use DMA in circular mode, and I verifies it with checking memory afetr 32768 bytes ,it remain preserve from which I had initilaize with.

finally i want to make clear things that i got error every time when frame_capture()is called, although the occuring of fault is at random,but when ever it occurs I am in frame_capture() or u can say leaving from it as u can see in my second last post.

thanks for ur support

[Tesla DeLorean]

Posted on December 20, 2012 at 14:21

Please go look at some books and manuals on the M3 core.

Joseph Yiu has a good book on the core, and has posted proper Hard Fault handlers (ie not a while(1) loop), which decode core registers, and dig back through the stacked state. You need to look at the faulting PC, and LR. Look at the instructions around these addresses to know where and what subroutine the fault occurred in. Use the registers to understand what the instructions were doing/accessing. R0..R15 are processor registers, R13 (SP) is the stack pointer by convention, R14 (LR), R15 (PC) In Keil startup.s

HardFault_Handler PROC
; Determine correct stack
TST lr, #4
ITE EQ
MRSEQ R0, MSP ; Read MSP (Main)
MRSNE R0, PSP ; Read PSP (Process)
EXTERN hard_fault_handler_c
B hard_fault_handler_c
ENDP

In app.c

void hard_fault_handler_c(unsigned int * hardfault_args)
{
printf (''[Hard Fault]
'');
printf (''r0 = %08X, r1 = %08X, r2 = %08X, r3 = %08
X'',
hardfault_args[0], hardfault_args[1], hardfault_args[2], hardfault_args[3]);
printf (''r12= %08X, lr = %08X, pc = %08X, psr= %08
X'',
hardfault_args[4], hardfault_args[5], hardfault_args[6], hardfault_args[7]);
printf (''bfar=%08X, cfsr=%08X, hfsr=%08X, dfsr=%08X, afsr=%08
X'',
*((volatile unsigned long *)(0xE000ED38)),
*((volatile unsigned long *)(0xE000ED28)),
*((volatile unsigned long *)(0xE000ED2C)),
*((volatile unsigned long *)(0xE000ED30)),
*((volatile unsigned long *)(0xE000ED3C)) );
while(1);
return;
}