On 6/19/21 11:32 PM, Michael S. Tsirkin wrote: > On Sat, Jun 19, 2021 at 10:45:17AM +0200, Philippe Mathieu-Daudé wrote: >> On 6/18/21 1:37 PM, Igor Mammedov wrote: >>> On Wed, 16 Jun 2021 22:43:24 +0200 >>> Philippe Mathieu-Daudé <phi...@redhat.com> wrote: >>> >>>> pc_madt_cpu_entry() is specific to QEMU 'PC' machines, >>>> move it to acpi-pc.c. >>>> >>>> Signed-off-by: Philippe Mathieu-Daudé <phi...@redhat.com> >>>> --- >>>> hw/i386/acpi-common.c | 40 ---------------------------------------- >>>> hw/i386/acpi-pc.c | 39 +++++++++++++++++++++++++++++++++++++++ >>> >>> it's used not only by PC machines but also microvm, >>> which didn't use acpi-build.c (aka acpi-pc.c) >>> it only links fine by virtue that PC machines >>> are object files are also included. >> >> Is that something new? I can't see this in mainstream, >> the microvm machine builds fine without this code. > > I think Igor means this: > > hw/i386/generic_event_device_x86.c: adevc->madt_cpu = pc_madt_cpu_entry; > > and > > hw/i386/microvm.c: DeviceState *dev = qdev_new(TYPE_ACPI_GED_X86);
Oh I missed that, indeed I didn't runtime test. Thanks Michael. > Admittedly given it's not limited to pc the function name is wrong ... >
