On 18/06/20 16:29, Peter Maydell wrote: >> - tz_ppc_dummy_ops which is broken and should just use NULL ops > Why is it broken? The intention is to create a MemoryRegion > which asserts if it's ever used (because it is a QEMU bug if > board code ever actually maps that MemoryRegion into anything). > NULL ops doesn't do that, it creates a MemoryRegion whose accesses > all report an error to the guest.
I meant "broken by the assertions" but yeah the NULL ops suggestion is wrong. I misread tz_ppc_dummy_accepts as "return false" (which is exactly what you get from NULL ops) instead of "g_assert_not_reached()". Paolo
