reset of the CPM

hy,

In linux v2.2.12 release (and till in 2.3.47):
commproc.c is a file of  8xx_io directory
dealing with general purpose functions for the
global management of the MPC860 Communication Processor Module.

In  function void m8xx_cpm_reset(uint host_page_addr) following code is
disabled :
#ifdef notdef
 /* We can't do this.  It seems to blow away the microcode
  * patch that EPPC-Bug loaded for us.  EPPC-Bug uses SCC1 for
  * Ethernet, SMC1 for the console, and I2C for serial EEPROM.
  * Our own drivers quickly reset all of these.
  */

 /* Perform a reset.
 */
 commproc->cp_cpcr = (CPM_CR_RST | CPM_CR_FLG);

 /* Wait for it.
 */
 while (commproc->cp_cpcr & CPM_CR_FLG);
#endif

What is the role of this patch?
and then on a reset or when a NMI exception is generated, does that
mean  that no reset of the CPM must ever be done.

Thanks for your help!
JC


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

duyrat1 wrote:

hy,

In linux v2.2.12 release (and till in 2.3.47):

What is the role of this patch?

This one in particular is on the MBX860 board.  It relocates the
I2C parameter ram when SCC1 is used for Ethernet.

I have since become smarter, and I am now making Linux install
it's own microcode patches.  This is necessary because drivers have
to be modified to know such patches exist, as the programming
interface physically and logically changes.  It is now appropriate
to reset the CPM and remove all existing configurations.

and then on a reset or when a NMI exception is generated, does that
mean  that no reset of the CPM must ever be done.

I don't remember the effect of external resets on the CPM.  I am
travelling right now and don't have the book handy.


	-- Dan

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/