Asm("sc;"); system call exception

cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

Asm("sc;"); system call exception

Jump to solution
1,123 Views
pierre-antoinem
Contributor III

Hi,

As i was reconfigurating the DDR on a default codewarrior project i came up with à little problem.

In the default configuration, the DDR(interleaved controllers) is mapped  from 0x0000_0000 to 0x7FFF_FFFF.

What I did was to remap the DDR from 0x0000_0000 to 0x3FFF_FFFF (which is 1GB or 512MB per controller)

I also disabled the interleaving mode which means each DDR controller is assigned 512MB of space (DDR1 from 0x0000_0000 to 0x1FFF_FFFF and DDR2 from 0x2000_0000 to 0x3FFF_FFFF)

The code start addresses were changed in the .lcf file according to the new DDR mapping. DDR1 unchanged but DDR2 changed from 0x40XX_XXXX to 0x20XX_XXXX (code, heap and stack, etc.)

When i run the default main(), core1 is stuck at adresse 0x...X_4000_0700 when executing asm("sc;"), command.

When I run the project without the asm("sc";); command on core1 it works fine. (the command does work on core0 because it jumps to adress 0x...X_0000_0700)

Does any one know what is missing to make it work ?

Thanks

Labels (1)
1 Solution
857 Views
pierre-antoinem
Contributor III

Hi,

First thanks for responding marius.

So yes this is what I have in the lcf file.

But I have just found the solution and every thing works.

The problem was the interupt vector address.

In the .tcl file you have :

if {$Ret} {
   reg ${SPR_GROUP}IVPR = 0x[expr {${proc_id} << 2}]0000000

    } else {

   # SMP project, same interrupt vectors for all cores
   reg ${SPR_GROUP}IVPR = 0x00000000

   }

whick means that for core1, proc_id = 1  then left shift of 2 bits multiplies it by 4 so we end up with 0x4000_0000.

But the DDR was remapped and consequently the code was displaced to 0x2000_0000 for core 1.

So I had to change the 2 by a 1 which makes a shif of 1 bit and so on multiplies the proc_id value (1) by 2 and we end up with 0x2000_0000.

So here is the correct tcl code :

if {$Ret} {
   reg ${SPR_GROUP}IVPR = 0x[expr {${proc_id} << 1}]0000000

    } else {

   # SMP project, same interrupt vectors for all cores
   reg ${SPR_GROUP}IVPR = 0x00000000

   }

cheers.

View solution in original post

2 Replies
857 Views
marius_grigoras
NXP Employee
NXP Employee

Hi,

So, you updated the lcf file for the second core, right?

You'll need to have something like this:

  _stack_addr = 0x203dfff0;

  _stack_end  = 0x203d7ff0;

 

  _heap_addr  = 0x203cfff0;

  _heap_end   = 0x203d7ff0;

  .intvec 0x20000000 :

  {

    *(.intvec)

  } = 0xffff

  . = 0x20100000;

  .newstart :

  {

    *(.newstart)

  }

/* the rest of the file */

If this still doesn't work, please make some hardware diagnostics for DDR (right click in the target task view).

Regards,

Marius

858 Views
pierre-antoinem
Contributor III

Hi,

First thanks for responding marius.

So yes this is what I have in the lcf file.

But I have just found the solution and every thing works.

The problem was the interupt vector address.

In the .tcl file you have :

if {$Ret} {
   reg ${SPR_GROUP}IVPR = 0x[expr {${proc_id} << 2}]0000000

    } else {

   # SMP project, same interrupt vectors for all cores
   reg ${SPR_GROUP}IVPR = 0x00000000

   }

whick means that for core1, proc_id = 1  then left shift of 2 bits multiplies it by 4 so we end up with 0x4000_0000.

But the DDR was remapped and consequently the code was displaced to 0x2000_0000 for core 1.

So I had to change the 2 by a 1 which makes a shif of 1 bit and so on multiplies the proc_id value (1) by 2 and we end up with 0x2000_0000.

So here is the correct tcl code :

if {$Ret} {
   reg ${SPR_GROUP}IVPR = 0x[expr {${proc_id} << 1}]0000000

    } else {

   # SMP project, same interrupt vectors for all cores
   reg ${SPR_GROUP}IVPR = 0x00000000

   }

cheers.