Kumar Gala wrote:
This is on purpose, the boards in question are using the default value of SCCR and have been. Kim added the CFG_SCCR_* but it was never actually used by the boards.
I got an email a way back from someone complaining that because the USB clocks were not being initialized on an 8349EMDS, the USB Linux driver was hanging on initialization. I have not seen this myself, and I don't remember who sent the email, but whoever did was convinced that programming the clocks fixed the problem.
You can also probably get rid of these macros in mpc83xx.h:
#define SCCR_ENCCM_0 0x00000000 #define SCCR_ENCCM_1 0x01000000 #define SCCR_ENCCM_2 0x02000000 #define SCCR_ENCCM_3 0x03000000 #define SCCR_TSEC1CM_0 0x00000000 #define SCCR_TSEC1CM_1 0x40000000 #define SCCR_TSEC1CM_2 0x80000000 #define SCCR_TSEC1CM_3 0xC0000000 #define SCCR_TSEC2CM_0 0x00000000 #define SCCR_TSEC2CM_1 0x10000000 #define SCCR_TSEC2CM_2 0x20000000 #define SCCR_TSEC2CM_3 0x30000000 #define SCCR_USBCM_0 0x00000000 #define SCCR_USBCM_1 0x00500000 #define SCCR_USBCM_2 0x00A00000 #define SCCR_USBCM_3 0x00F00000
Don't see any harm in having them.
Someone might seem them and think they should be used to initialize the clocks. With the changes to cpu_init_f(), these macros will never be used.