Re: [PATCH] ata: ahci: Don't call pci_intx() directly

[Niklas Cassel <cassel@kernel.org>]

On Mon, Nov 04, 2024 at 02:23:43PM +0100, Heiner Kallweit wrote:
> On 04.11.2024 09:30, Niklas Cassel wrote:
> > On Fri, Nov 01, 2024 at 11:38:53PM +0100, Heiner Kallweit wrote:
> >> pci_intx() should be called by PCI core and some virtualization code
> >> only. In PCI device drivers use the appropriate pci_alloc_irq_vectors()
> >> call.
> > 
> > Hello Heiner,
> > 
> > as you might or might not know, this patch conflicts with a Philipp's
> > already acked patch:
> > https://lore.kernel.org/linux-ide/20241015185124.64726-10-pstanner@redhat.com/
> > 
> I know, therefore he's on cc. Fully migrating PCI device drivers to the
> pci_alloc_irq_vectors() should be done anyway and is the cleaner
> alternative to changing pci_intx(). However for some drivers this is a rather
> complex task, therefore I understand Philipp's approach to adjust pci_intx()
> first. He's incorporating other review feedback in his series, so with the
> next re-spin he could remove the ahci patch from his series.

Well, if you look at Philipp's patch it:

1) Doesn't only update drivers/ata/ahci.c,
it also updates:
drivers/ata/ata_piix.c
drivers/ata/pata_rdc.c
drivers/ata/sata_sil24.c
drivers/ata/sata_sis.c
drivers/ata/sata_uli.c
drivers/ata/sata_vsc.c

Why don't you update the other drivers in drivers/ata/* ?


2) Doesn't just bother to fix a single subsystem (drivers/ata/),
it is actually part of a series that fixes all affected subsystems.

Why don't you send out this fix as part of a series that fixes all the
affected subsystems?


Kind regards,
Niklas