Re: [Qemu-devel] [PATCH v3 19/19] hw: i386: Implement the ACPI builder interface for PC

[Samuel Ortiz <sameo@linux.intel.com>]

Hi Paolo,

On Mon, Oct 29, 2018 at 06:28:39PM +0100, Paolo Bonzini wrote:
> On 29/10/2018 18:01, Samuel Ortiz wrote:
> > @@ -1556,6 +1557,7 @@ void acpi_build(AcpiBuildTables *tables,
> >      GArray *tables_blob = tables->table_data;
> >      AcpiSlicOem slic_oem = { .id = NULL, .table_id = NULL };
> >      Object *vmgenid_dev;
> > +    AcpiBuilder *ab = ACPI_BUILDER(machine);
> >  
> >      acpi_get_pm_info(&pm);
> >      acpi_get_misc_info(&misc);
> > @@ -1605,7 +1607,8 @@ void acpi_build(AcpiBuildTables *tables,
> >      aml_len += tables_blob->len - fadt;
> >  
> >      acpi_add_table(table_offsets, tables_blob);
> > -    build_madt(tables_blob, tables->linker, machine, conf);
> > +    acpi_builder_madt(ab, tables_blob, tables->linker,
> > +                      machine, conf);
> >  
> >      vmgenid_dev = find_vmgenid_dev();
> >      if (vmgenid_dev) {
> 
> Just a quick question before I go and actually apply the patches to look
> at the resulting code: is there any reason why you didn't add the
> MachineState and/or AcpiBuilder to AcpiBuildState, similar to how it was
> in v1?
With v1 I was not adding MachineState to AcpiBuildState, I may be
missing your point.
Do you mean adding an AcpiBuilder pointer to AcpiConfiguration?

Cheers,
Samuel.