Re: [Qemu-devel] [PATCH for-2.2] acpi-build: mark RAM dirty on table update

[Igor Mammedov]

On Thu, 20 Nov 2014 09:49:20 +0530
Amit Shah <address@hidden> wrote:

> On (Wed) 19 Nov 2014 [11:08:46], Igor Mammedov wrote:
> > On Wed, 19 Nov 2014 12:51:00 +0530
> > Amit Shah <address@hidden> wrote:
> 
> > > > -static void *acpi_add_rom_blob(AcpiBuildState *build_state, GArray
> > > > *blob, +static ram_addr_t acpi_add_rom_blob(AcpiBuildState
> > > > *build_state, GArray *blob, const char *name)
> > > >  {
> > > >      return rom_add_blob(name, blob->data, acpi_data_len(blob), -1,
> > > > name, @@ -1777,6 +1781,7 @@ void acpi_setup(PcGuestInfo *guest_info)
> > > >      /* Now expose it all to Guest */
> > > >      build_state->table_ram = acpi_add_rom_blob(build_state,
> > > > tables.table_data, ACPI_BUILD_TABLE_FILE);
> > > > +    assert(build_state->table_ram != RAM_ADDR_MAX);
> > > >      build_state->table_size = acpi_data_len(tables.table_data);
> > > 
> > > Isn't an assert too strong if this happens during hotplug?
> > > 
> > > I'm trying to follow this code, but looks like this isn't called in
> > > the hotplug path - is that right?
> > yep, it's called only at startup
> 
> Thanks; what's the path taken for hotplug, then?  (Ensuring we have
> that case covered too).
In case of hotplug ACPI blob doesn't change at all (it has all
possible device objects in it). It changes on the first time BIOS
fetches ROM blob and on reset.
Later during hotplug, guest gets SCI interrupt and runs related to
hotplug event AML method, which fishes out bits necessary for hotplug
via corresponding mmio interface.


>               Amit