Re: [PATCH] PCI: Fix devres regression in pci_intx()

From: Alex Williamson
Date: Wed Sep 04 2024 - 17:11:58 EST


On Wed, 4 Sep 2024 23:24:53 +0300
Andy Shevchenko <andy.shevchenko@xxxxxxxxx> wrote:

> Wed, Sep 04, 2024 at 12:07:21PM -0600, Alex Williamson kirjoitti:
> > On Wed, 04 Sep 2024 15:37:25 +0200
> > Philipp Stanner <pstanner@xxxxxxxxxx> wrote:
> > > On Wed, 2024-09-04 at 17:25 +0900, Damien Le Moal wrote:
>
> ...
>
> > > If vfio-pci can get rid of pci_intx() alltogether, that might be a good
> > > thing. As far as I understood Andy Shevchenko, pci_intx() is outdated.
> > > There's only a hand full of users anyways.
> >
> > What's the alternative?
>
> From API perspective the pci_alloc_irq_vectors() & Co should be used.

We can't replace a device level INTx control with a vector allocation
function.

> > vfio-pci has a potentially unique requirement
> > here, we don't know how to handle the device interrupt, we only forward
> > it to the userspace driver. As a level triggered interrupt, INTx will
> > continue to assert until that userspace driver handles the device.
> > That's obviously unacceptable from a host perspective, so INTx is
> > masked at the device via pci_intx() where available, or at the
> > interrupt controller otherwise. The API with the userspace driver
> > requires that driver to unmask the interrupt, again resulting in a call
> > to pci_intx() or unmasking the interrupt controller, in order to receive
> > further interrupts from the device. Thanks,
>
> I briefly read the discussion and if I understand it correctly the problem here
> is in the flow: when the above mentioned API is being called. Hence it's design
> (or architectural) level of issue and changing call from foo() to bar() won't
> magically make problem go away. But I might be mistaken.

Certainly from a vector allocation standpoint we can change to whatever
is preferred, but the direct INTx manipulation functions are a
different thing entirely and afaik there's nothing else that can
replace them at a low level, nor can we just get rid of our calls to
pci_intx(). Thanks,

Alex