The TestEquality instance for LLVMPointer is suspicious

[travitch]

The TestEquality instance for LLVMPointer returns true if the inputs are the same width, with no additional checks:

crucible/crucible-llvm/src/Lang/Crucible/LLVM/MemModel/Pointer.hs

Line 129 in 3db599e

instance IsExprBuilder sym => TestEquality (LLVMPointer sym) where

This seems unsafe and uses of this instance should probably just be testing the width reprs explicitly.

[robdockins]

Agreed. The contract for TestEquality is ambiguous at best, but we're pretty uniformly using it as a value equality check, and not just a type equality check.

[langston-barrett]

Here's the issue that appears to be tracking/discussing the ambiguity: GaloisInc/parameterized-utils#63