On Mon, 26 Apr 2021 at 10:23, Philippe Mathieu-Daudé <f4...@amsat.org> wrote: > > On 4/25/21 8:33 PM, Peter Maydell wrote: > > On Sat, 24 Apr 2021 at 14:04, Philippe Mathieu-Daudé <f4...@amsat.org> > > wrote: > >> I now understand better the diag288 case, but I still don't understand > >> the TYPE_APIC one. It has no DeviceClass::reset(), its abstract parent > >> TYPE_APIC_COMMON register apic_reset_common() but being TYPE_DEVICE it > >> is not on a qbus. It is somehow connected to the X86CPU object, but the > >> single call to apic_init_reset() is from do_cpu_init() - not a reset > >> method -. > > > > pc_machine_reset() calls device_legacy_reset(cpu->apic_state) > > which is to say it invokes the DeviceState::reset method, > > which is either kvm_apic_reset or apic_reset_common. > > Oh, thanks! I guess "convoluted" is the proper adjective to describe > this reset logic. I suppose APIC is a very old device, part of the > Frankenstein PC, so hard to rework (because we are scared of the > implications of changing old & heavily used devices).
This is mostly just another instance of "our reset logic doesn't deal well with devices which aren't on buses". The APIC isn't on a bus, so the machine that uses it has a local workaround to manually arrange for it to reset, just as it does for the CPU. thanks -- PMM
