I am using a MK10DX256VLL7 and am seeing instances of the core locking up that do not result in a LOCKUP reset. From my reading of the documentation this would not seem possible. If I modify my fault handlers to deliberately create a core lockup then I do get a LOCKUP reset. Also, when using a JTAG debugger (BDI3000) it is not possible to halt the core once in the locked up state.
On a working build of my software the locked up state is seen very infrequently, in the order of once in a month of continuous testing. But occasionally I have produced builds that lock up every time, again there is no LOCKUP reset. These builds appear to show extreme sensitivity to the location of critical pieces of code. In one case the difference between a working build and one that locks up was reduced to the position of a section of code just 6 instructions long. This section was moved around by the insertion of NOPs. This piece of code was related to SPI interrupt servicing.
So to summarize, does anyone know of:-
1) Cases of core lockup that do not result in a LOCKUP reset?
2) Cases of extreme sensitivity to code location?
3) Any known issues around SPI interrupt servicing?