How to find the location of a crash using XGDB

  • version

    1.0.1

  • scope

    Example.

    This code is provided as example code for a user to base their code on.

  • description

    How to find the location of a crash using XGDB

  • boards

    Unless otherwise specified, this example runs on the SliceKIT Core Board, but can easily be run on any XMOS device by using a different XN file.

If an exception occurs in the target code, XGDB can be used to find the location of the exception and diagnose its root cause. For example, compile the following code ensuring that debug is enabled (-g):

int divide(int x, int y) {
  return x / y;
}

int main() {
  divide(5, 0);
  return 0;
}

From within xTIMEcomposer Studio

Create a new debug configuration via Run->Debug Configurations->xCORE Applications, then start debugging. Execution will now break when the exception occurs. The Debug view will display the trap type information, in this case, ET_ARITHMETIC (caused by the divide by zero). It will also show the pc of the excepting instruction. For more details the Registers view can be used to find the contents of the exception data register, which depending on the type of exception that occurred, can provide more information as to its root cause.

From the command line

For example, start XGDB, connect to the simulator and start debugging. Execution will now break when the exception occurs, and the exception type will be displayed on the console. The stack trace of how this location in the program was reached can be found using the backtrace command. You can then look at the register contents to attempt to diagnose the root cause of the trap:

> xgdb a.xe
...etc...
(gdb) connect -s
0xffffc04e in ?? ()
(gdb) run
...etc...
Program received signal ET_ARITHMETIC, Arithmetic exception.
0x000100ba in divide (x=5, y=0) at find_the_location_of_a_crash.xc:10
10    return x / y;
(gdb) backtrace
#0  0x000100ba in divide (x=5, y=0) at find_the_location_of_a_crash.xc:10
#1  0x000100ce in main () at find_the_location_of_a_crash.xc:14
(gdb) print /x $spc
$1 = 0x100ba
(gdb) print /x $ssr
$2 = 0x0
(gdb) print /x $et
$3 = 0x7
(gdb) print /x $ed
$4 = 0x0

Note: In the above, the spc (saved program counter) and the ssr (saved status register) will contain the values in the pc and the sr at the time of the exception. The et and ed registers contain the exception type and exception data details. In this case, the exception type is 0x7, which corresponds to ET_ARITHMETIC (see xs1b_user.h). The exception data register contents in general depend on the type of exception. For example, for an ET_RESOURCE exception the exception data will contain the id of the resource causing the trap. However, in the case of arithmetic exceptions, the ed will be set to zero.