Re: [Qemu-devel] [PATCH v2 1/1] i386: expose floppy-related objects in SSDT

[Roman Kagan]

On Wed, Dec 16, 2015 at 11:15:55PM +0100, Igor Mammedov wrote:
> On Wed, 16 Dec 2015 20:34:55 +0300
> Roman Kagan <address@hidden> wrote:
> 
> > On Wed, Dec 16, 2015 at 05:46:57PM +0100, Igor Mammedov wrote:
> > > On Wed, 16 Dec 2015 10:45:09 +0300 "Denis V. Lunev" <address@hidden>
> > > wrote:
> > > > @@ -1125,6 +1183,26 @@ build_ssdt(GArray *table_data, GArray
> > > > *linker, aml_append(ssdt, scope);
> > > >      }
> > > >  
> > > > +    {
> > > add this block only if floppy controller is present
> > 
> > ATM both qemu master (and stable branches) and your patchset define
> > FDC0 unconditionally, making its _STA depend on FDEN field of the
> > containing ISA device rather than any sort of qemu internal state.  I
> > think it's more consistent also to always define its _FDE.  The whole
> > definition of FDC0 can then be made contidional on the presence of
> > the floppy controller later on top of your patchset.
> Ok, I don't insist on it as I'll have to do anyway for whole FDC0 after
> merging SSDT into DSDT.
> 
> The reason why I haven't done it in conversion patchset, is to keep
> DSDT exactly the same as an original one, so 'make check' could prove
> conversion correctness.

I just realized that bios-tables-test used to report differences between
the expected and the actual DSDT and SSDT even before the patch but that
would only cause warnings and not a test failure.  With the patch
applied (which adds yet another difference) nothing changes: it still
reports warnings but passes.

Roman.