On Thu, 15 Mar 2018 17:13:03 +0000 Peter Maydell <peter.mayd...@linaro.org> wrote:
> On 13 March 2018 at 16:55, Andrew Baumann <andrew.baum...@microsoft.com> > wrote: > >> From: Qemu-devel <qemu-devel- > >> bounces+andrew.baumann=microsoft....@nongnu.org> On Behalf Of Peter > >> Maydell > >> Sent: Tuesday, 13 March 2018 08:35 > >> > >> Now we have separate types for BCM2386 and BCM2387, we might as well > >> just hard-code the CPU type they use rather than having it passed > >> through as an object property. This then lets us put the initialization > >> of the CPU object in init rather than realize. > > > What about the default_cpu_type field of MachineClass set in > > raspi[23]_machine_init? That seems irrelevant now... > > Igor, can you help with this question? For a board that always > has one CPU type (because the real hardware only ever has > one SoC, and that SoC QOM object hard codes the CPU type) > does it still need to set the mc->default_cpu_type field in > its MachineClass, or does that become pointless ? Since board ignores whatever were specified on '-cpu' and there aren't any checks in board code for current_machine->cpu_type, removing mc->default_cpu_type won't really affect anything. With current code in vl.c and with default_cpu_type set, user will get error if he specified non existing cpu_model with -cpu. If default_cpu_type were NULL, -cpu is completely ignored. But once http://patchwork.ozlabs.org/patch/870297/ is applied it will error out the same way as if default_cpu_type were set since vl.c won't rely on it anymore for parsing cpu_model. So in short it's ok to remove mc->default_cpu_type here. > > thanks > -- PMM