Re: [PATCH 0/8] x86/pci: consolidate x86-specific PCI quirks into arch/x86/pci/fixup.c

[Bjorn Helgaas <helgaas@kernel.org>]

On Fri, Jul 10, 2026 at 01:51:01PM -0700, Rosen Penev wrote:
> On Fri, Jul 10, 2026 at 1:46 PM Bjorn Helgaas <helgaas@kernel.org> wrote:
> > On Tue, Jul 07, 2026 at 03:17:52PM -0700, Rosen Penev wrote:
> > > This series moves all x86-specific PCI quirks from the generic
> > > drivers/pci/quirks.c into arch/x86/pci/fixup.c, where they
> > > architecturally belong. This reduces the ifdef burden in the generic
> > > quirk file and improves build granularity by gating the x86 fixups
> > > behind CONFIG_PCI_QUIRKS.
> > >
> > > The series is organized as a preparatory commit followed by one
> > > commit per vendor/family, keeping each move reviewable.
> > >
> > > Suggested by: https://lore.kernel.org/all/20161221142607.GA8594@infradead.org/
> > >
> > > Rosen Penev (8):
> > >   x86/pci: gate arch fixups with CONFIG_PCI_QUIRKS
> > >   x86/pci: move Intel chipset quirks to arch/x86/pci/fixup.c
> > >   x86/pci: move VIA chipset quirks to arch/x86/pci/fixup.c
> > >   x86/pci: move SiS chipset quirks to arch/x86/pci/fixup.c
> > >   x86/pci: move ALi chipset quirks to arch/x86/pci/fixup.c
> > >   x86/pci: move AMD x86 chipset quirks to arch/x86/pci/fixup.c
> > >   x86/pci: move x86 ifdef-gated quirks to arch/x86/pci/fixup.c
> > >   x86/pci: move remaining x86-specific quirks to fixup.c
> > >
> > >  arch/x86/pci/Makefile |    2 +-
> > >  arch/x86/pci/fixup.c  | 2284 ++++++++++++++++++++++++++++++++++++++++
> > >  drivers/pci/pci.h     |    3 +
> > >  drivers/pci/quirks.c  | 2330 +----------------------------------------
> > >  4 files changed, 2316 insertions(+), 2303 deletions(-)
> >
> > I don't object to any of this from a PCI perspective, so I guess it's
> > up to the x86 folks if they want them.
> >
> > It looks like the typical subject line style in fixup.c is:
> >
> >   x86/PCI: <Capitalized-first-word> ...
>
> I'm guessing this means I should resend.

I'd wait a few days in case others have comments.

While you're waiting you could analyze these to make sure they don't
change any critical ordering.  The quirks are called by
pci_do_fixups(), which I'm sure you didn't change, but the order of
the quirks in each level will change because the tables, e.g.,
__start_pci_fixups_early[], are put together by the linker so the
order within the table depends on the file the quirks are in.