Re: [PATCH] hw/i386: Add the possibility to disable the 'isapc' machine

[Thomas Huth]

On 07/01/2022 19.54, Philippe Mathieu-Daudé wrote:
> On 1/7/22 17:07, Thomas Huth wrote:
> > We already have a CONFIG_ISAPC switch - but we're not using it yet.
> > Add some "#ifdefs" to make it possible to disable this machine now.
> >
> > Signed-off-by: Thomas Huth <thuth@redhat.com>
> > ---
> >   hw/i386/pc_piix.c        | 5 ++++-
> >   tests/qtest/cdrom-test.c | 2 +-
> >   2 files changed, 5 insertions(+), 2 deletions(-)
> >
> > diff --git a/hw/i386/pc_piix.c b/hw/i386/pc_piix.c
> > index 1999190276..5147e1ee74 100644
> > --- a/hw/i386/pc_piix.c
> > +++ b/hw/i386/pc_piix.c
> > @@ -357,10 +357,12 @@ static void pc_compat_1_4_fn(MachineState *machine)
> >       pc_compat_1_5_fn(machine);
> >   }
> >   
> > +#ifdef CONFIG_ISAPC
> >   static void pc_init_isa(MachineState *machine)
> >   {
> >       pc_init1(machine, TYPE_I440FX_PCI_HOST_BRIDGE, TYPE_I440FX_PCI_DEVICE);
> >   }
> > +#endif
> >   
> >   #ifdef CONFIG_XEN
> >   static void pc_xen_hvm_init_pci(MachineState *machine)
> > @@ -916,6 +918,7 @@ void igd_passthrough_isa_bridge_create(PCIBus *bus, 
> > uint16_t gpu_dev_id)
> >       pci_config_set_revision(bridge_dev->config, pch_rev_id);
> >   }
> >   
> > +#ifdef CONFIG_ISAPC
> >   static void isapc_machine_options(MachineClass *m)
> >   {
> >       PCMachineClass *pcmc = PC_MACHINE_CLASS(m);
> > @@ -935,7 +938,7 @@ static void isapc_machine_options(MachineClass *m)
> >   
> >   DEFINE_PC_MACHINE(isapc, "isapc", pc_init_isa,
> >                     isapc_machine_options);
> > -
> > +#endif
>
> Wouldn't it be cleaner to extract the isapc machine to a new C unit
> (after renaming/declaring pc_init1 public)? 

That would be cleaner, of course, but that would also mean a major 
refactoring: We could rearrange the code so that it is possible to compile a 
binary that has isapc, but not i440fx... but that likely means a complete 
rework of pc_piix.c so that the code that is common between i440fx and isapc 
needs to be moved into a separate file, too. And that's likely not worth the 
effort, I guess, since people rather want to compile without isapc than 
compiling without i440fx, I think. So let's go with this low-hanging fruit 
here for now, and keep the possible clean-up in mind for later.

> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>

Thanks!

 Thomas