D456 - A fault report distinguishes the emulator's own bugs from the guest's, and names a trap instruction

Status: decided Date: 2026-09-01

When a guest fault's instruction pointer lies outside the guest image, the report states plainly that the fault is in this project's own code, not the guest's, and that the import named in the header is only context. When the faulting instruction is a privileged or trap instruction, the report names it instead of describing the fault as an ordinary bad memory access.

Why: Without this, a fault raised by this project's own code reads as if the guest's last imported call was where it happened, sending an investigation to the wrong function; and a trap instruction raises the same host-level fault shape as a wild pointer dereference, reading as a guest bug where none exists.

Rejected: leaving the reader to infer these from the raw register dump alone - costly and error-prone, as a real investigation demonstrated.