We don't have relocations for these that can support the large types, but we do often have the size information available at compile time. For these cases, we can emit the (possibly complex) sequence of instructions to materialise the size and insert the CSetBounds instruction at compile time, without involving the linker in the bounds (the linker is still responsible for fixups in finding the base).
We don't have relocations for these that can support the large types, but we do often have the size information available at compile time. For these cases, we can emit the (possibly complex) sequence of instructions to materialise the size and insert the CSetBounds instruction at compile time, without involving the linker in the bounds (the linker is still responsible for fixups in finding the base).