Hi Bill,The checksum is Csmith's way of summarizing the values left over in global variables after the random code finishes. Checksum differences often indicate miscompilations.
I still have troubles getting my compiler to compile the test codes. It crashed.
Well, it is not totally unusual for 100% of test cases generated by Csmith to crash a random embedded compiler that has not seen Csmith before. If that is what is going on, you'll want to generate reduced test cases (that are still valid C) and discuss them with your tool vendor.
If your vendor claims to be compliant with ANSI C or C99 and you hand them a valid ANSI C or C99 program that crashes their tool, they're supposed to fix it.
I suspecting that typedefs are not correctly defined for my test setup. Can someone show me how to define types for my setup? The types for my system are: char 8 bits int 16 bits long 32 bits
The typedefs are trivial, for example: typedef long int int32_t; Or you can probably just use the ones we defined for AVR: csmith/runtime/stdint_avr.h John