On Fri, Aug 31, 2018 at 09:21:32AM +0100, Chris Wilson wrote:
> Quoting Lucas De Marchi (2018-08-29 01:35:31)
> > The 2 PCI IDs that are used for the command line overrid mechanism
> > were left defined. The rest can be gone and then we just use the kernel
> > defines.
> >
> > Signed-off-by: Lucas De Marchi <lucas.demarchi@xxxxxxxxx>
> > ---
> > intel/intel_chipset.c | 5 ++
> > intel/intel_chipset.h | 187 +-----------------------------------------
> > 2 files changed, 6 insertions(+), 186 deletions(-)
> >
> > diff --git a/intel/intel_chipset.c b/intel/intel_chipset.c
> > index 0c2ba884..c984d8ac 100644
> > --- a/intel/intel_chipset.c
> > +++ b/intel/intel_chipset.c
> > @@ -36,6 +36,11 @@ static const struct pci_device {
> > } pciids[] = {
> > INTEL_ICL_11_IDS(11),
> > INTEL_CNL_IDS(10),
> > + INTEL_CFL_IDS(9),
> > + INTEL_GLK_IDS(9),
> > + INTEL_KBL_IDS(9),
> > + INTEL_BXT_IDS(9),
> > + INTEL_SKL_IDS(9),
>
> The gradual conversion lgtm. But why stop here? :)
>From cover letter:
Initially my plan was to convert all gens, back to gen2, but
that proved slightly difficult since there are some corner cases
to cover and I didn't want to block the important part, i.e.:
for recent gens, there's no risk of missing a PCI ID.
With the last approach moving the implementation to a .c file I think it
will be easier to implement for older gens, but there's no point in
doing the manual boring labor of converting all gens just to have to
change the approach in a v2, v3 of the patch set. Like I did for v1 ->
v2. I can convert the rest if we agree the current approach is
okish
I'm even ok with letting older ones as is since I hope we won't add a
new pci id for e.g. gen3, so I won't have to touch that.
Lucas De Marchi
> -Chris
_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/dri-devel
