On Mon, Feb 13, 2017 at 02:28:16PM +0000, Peter Maydell wrote: > From: Michael Davidsaver <mdavidsa...@gmail.com> > > Add a new API cpu_generic_new() which creates a QOM CPU object > (including calling the CPU class parse_features method) but > does not realize it, and reimplement cpu_generic_init() to > simply call cpu_generic_new() and then immediately realize. > the CPU. > > The motivation for this is that there is currently no > way for board code to take advantage of the cpu_generic_init() > convenience function if it needs to set QOM properties > on the created CPU. Instead it has to do it all manually, which > is prone to bugs like that fixed in commit 00909b585861 (where > a board code forgot to call parse_features which meant that > command line +feature,-feature flags were ignored). > > Signed-off-by: Michael Davidsaver <mdavidsa...@gmail.com> > [PMM: renamed new function to cpu_generic_new(), rewrote > commit message] > Signed-off-by: Peter Maydell <peter.mayd...@linaro.org> > --- > include/qom/cpu.h | 17 +++++++++++++++++ > qom/cpu.c | 27 ++++++++++++++++++--------- > 2 files changed, 35 insertions(+), 9 deletions(-) > > diff --git a/include/qom/cpu.h b/include/qom/cpu.h > index 45bcf21..e900586 100644 > --- a/include/qom/cpu.h > +++ b/include/qom/cpu.h > @@ -599,11 +599,28 @@ void cpu_reset(CPUState *cpu); > ObjectClass *cpu_class_by_name(const char *typename, const char *cpu_model); > > /** > + * cpu_generic_new: > + * @typename: The CPU base type. > + * @cpu_model: The model string including optional parameters. > + * > + * Instantiates a CPU, processes optional parameters but does not realize it. > + * This is the recommended way to create a CPU object which needs to be > + * configured by then setting QOM properties on it. The configured CPU can > + * then be realized in the usual way by calling > + * object_property_set_bool(cpuobj, true, "realized", &err); > + * > + * Returns: A #CPUState or %NULL if an error occurred. > + */ > +CPUState *cpu_generic_new(const char *typename, const char *cpu_model);
I am not sure we will stlil add this, but in case we are going to add this helper temporarily until we have Igor's generic cpu_model parsing, it would be useful to make the function signature be: CPUState *cpu_generic_new(const char *typename, MachineState *machine) and parse machine->cpu_model inside the function implementation. This way we could replace the helper with object_new(machine->cpu_typename) more easily once we add Igor's cleanup. -- Eduardo
