Hi Jeremy, On Tue, 18 May 2021 05:28:56 +0100, Jeremy Linton <jeremy.linton@xxxxxxx> wrote: > > Hi, > > > On 3/30/21 10:11 AM, Marc Zyngier wrote: > > We have now three ways of ending up with NO_MSI being set. > > Document them. > > > > Acked-by: Bjorn Helgaas <bhelgaas@xxxxxxxxxx> > > Signed-off-by: Marc Zyngier <maz@xxxxxxxxxx> > > --- > > drivers/pci/msi.c | 11 +++++++++-- > > 1 file changed, 9 insertions(+), 2 deletions(-) > > > > diff --git a/drivers/pci/msi.c b/drivers/pci/msi.c > > index d9c73c173c14..217dc9f0231f 100644 > > --- a/drivers/pci/msi.c > > +++ b/drivers/pci/msi.c > > @@ -871,8 +871,15 @@ static int pci_msi_supported(struct pci_dev *dev, int nvec) > > * Any bridge which does NOT route MSI transactions from its > > * secondary bus to its primary bus must set NO_MSI flag on > > * the secondary pci_bus. > > - * We expect only arch-specific PCI host bus controller driver > > - * or quirks for specific PCI bridges to be setting NO_MSI. > > + * > > + * The NO_MSI flag can either be set directly by: > > + * - arch-specific PCI host bus controller drivers (deprecated) > > + * - quirks for specific PCI bridges > > + * > > + * or indirectly by platform-specific PCI host bridge drivers by > > + * advertising the 'msi_domain' property, which results in > > + * the NO_MSI flag when no MSI domain is found for this bridge > > + * at probe time. > > I have an ACPI machine with a gicv2 (no m), and a MSI region that > isn't described by ACPI because its non-standard. In the past this > tended to work because PCIe device drivers would fall back to legacy > pci intx silently. But, with 5.13, it seems this series now triggers > the WARN_ON_ONCE() in arch_setup_msi_irq, because duh, no MSI support. This is nothing new, and you could get the exact same warning if you didn't have legacy drivers compiled in (any of the 3 drivers that were fixed in this series). This series now makes sure you definitely know about this issue. And look, it worked! :-) > Everything of course continues to work, it just gets this ugly splat > on bootup telling me basically the machine doesn't support MSIs. So, I > considered a few patches, including just basically setting nomsi if > gicv2 && acpi, or eek a host bridge quirk. > > None of these seem great, so how can this be fixed? The host bridge quirk seems the most likely route to address this, but you could just as well advertise msi_domain in the ACPI PCI path, *and* check for IORT mappings in pci_register_host_bridge(), similarly to what Jean-Philippe has proposed for DT in [1]. Thanks, M. [1] https://lore.kernel.org/r/20210510173129.750496-1-jean-philippe@xxxxxxxxxx -- Without deviation from the norm, progress is not possible.
