MVME5500 not working with RTEMS 4.7 (CVS)

Phillip Sorensen pas37 at cornell.edu
Thu Jul 28 22:50:54 UTC 2005


 >> 1. c/src/lib/libbsp/powerpc/mvme5500/startup/bspstart.c
 >> I added the Cpu_table.exceptions_in_RAM = TRUE; before the
 >> initialize_exceptions() call since the exception handler address
 >> routine
 >> know uses it.
 >
 >
 >
 > May  I know what  the symptom was without this patch ?
 > I  did'nt experience  any problem with RTEMS 4.6.0.


The indication on the console is a failure when setting up the exception 
handler for the decrement exception, due to the fact that the current 
handler is not equal to the default handler.

The problem is that the mpc60x_get_vector_addr() function now references 
the Cpu_table.exceptions_in_RAM variable to calculate the vector offset. 
    If it is not defined the mpc60x_init_exceptions() function (called 
by initialize exceptions()) puts the default handlers at the wrong 
memory locations.

This change was made in CVS sometime in November 2004.  I'm not sure if 
it is in the 4.6.2 release or not.


 >> 2. c/src/lib/libbsp/powerpc/shared/clock/p_clock.c
 >> I removed the comments preventing the clockIrqData structure from
 >> being
 >> initialized.
 >
 >
 >
 > Why was it commented out in 4.7 CVS ?  This was not commented
 > out  in  RTEMS 4.6.0.


It seems to have been done in May.  The log message only refers to a 
change of a header file.  I think it may have been an unintentional change.

 >> c/src/lib/lipcpu/powerpc/mpc6xx/clock/c_clock.c
 >> Added the & to the assembly call in the clockIsq routine as shown in
 >> PR#773.
 >
 >
 >
 > Yes, this is needed for gcc4.0+
 >
 > Kate
 >
Phil





More information about the users mailing list