Problem:

We're seeing EMI spikes at 120MHz & 360MHz that we've isolated to the SOM board. If I program the Flash to do a 'halt' as the first instruction out of reset, the spikes are still there. Hold reset low and the spikes go away. Any pointers to the main sources of EMI and solutions would be helpful.

Solution:

The silver bullet is disabling the MCF5329 Debug Module. The MCF5329 supports real time processor debugging. The side effect is by default the JTAG/PST lines spam out processor status information on every clockcycle (i.e. multiple external signals switching at 120Mhz). Customers should be disabling these signals in end product software.

For customers using uCLinux and Freescale's LTIB, there's a configuration setting in the kernel config, under 'kernel hacking' that sets CONFIG_BDM_DISABLE. That, in turn, enables a bit of code in a config.c file that turns off PSTCLK. So at first glance, it appears as if all that must be done is to set this flag in the kernel configuration.

The catch is that there's a bug, at least in the version of the kernel/LTIB that I'm using (2.6.17.1)...the kernel config calls it "CONFIG_BDM_DISABLE" but the arch/m68knommu/platform/532x/config.c file calls it "MCF_BDM_DISABLE". So this file in the kernel must be changed to "CONFIG_BDM_DISABLE" and then all works. It is the only place in the kernel that uses this flag, as far as I can tell.