| Message ID | 20240527161450.326615-18-herve.codina@bootlin.com (mailing list archive) |
|---|---|
| State | Superseded |
| Delegated to: | Bjorn Helgaas |
| Headers | show |
| Series | Add support for the LAN966x PCI device using a DT overlay | expand |
On Mon, May 27, 2024 at 06:14:44PM +0200, Herve Codina wrote: > PCI devices and bridges DT nodes created during the PCI scan are created > with the interrupt-map property set to handle interrupts. > > In order to set this interrupt-map property at a specific level, a > phandle to the parent interrupt controller is needed. On systems that > are not fully described by a device-tree, the parent interrupt > controller may be unavailable (i.e. not described by the device-tree). > > As mentioned in the [1], avoiding the use of the interrupt-map property > and considering a PCI device as an interrupt controller itself avoid the > use of a parent interrupt phandle. > > In that case, the PCI device itself as an interrupt controller is > responsible for routing the interrupts described in the device-tree > world (DT overlay) to the PCI interrupts. > > Add the 'interrupt-controller' property in the PCI device DT node. > > [1]: https://lore.kernel.org/lkml/CAL_Jsq+je7+9ATR=B6jXHjEJHjn24vQFs4Tvi9=vhDeK9n42Aw@mail.gmail.com/ > > Signed-off-by: Herve Codina <herve.codina@bootlin.com> No objection from me, but I'd like an ack/review from Rob. > --- > drivers/pci/of_property.c | 24 ++++++++++++++++++++++++ > 1 file changed, 24 insertions(+) > > diff --git a/drivers/pci/of_property.c b/drivers/pci/of_property.c > index 03539e505372..5a0b98e69795 100644 > --- a/drivers/pci/of_property.c > +++ b/drivers/pci/of_property.c > @@ -183,6 +183,26 @@ static int of_pci_prop_interrupts(struct pci_dev *pdev, > return of_changeset_add_prop_u32(ocs, np, "interrupts", (u32)pin); > } > > +static int of_pci_prop_intr_ctrl(struct pci_dev *pdev, struct of_changeset *ocs, > + struct device_node *np) > +{ > + int ret; > + u8 pin; > + > + ret = pci_read_config_byte(pdev, PCI_INTERRUPT_PIN, &pin); > + if (ret != 0) > + return ret; > + > + if (!pin) > + return 0; > + > + ret = of_changeset_add_prop_u32(ocs, np, "#interrupt-cells", 1); > + if (ret) > + return ret; > + > + return of_changeset_add_prop_bool(ocs, np, "interrupt-controller"); > +} > + > static int of_pci_prop_intr_map(struct pci_dev *pdev, struct of_changeset *ocs, > struct device_node *np) > { > @@ -336,6 +356,10 @@ int of_pci_add_properties(struct pci_dev *pdev, struct of_changeset *ocs, > ret = of_pci_prop_intr_map(pdev, ocs, np); > if (ret) > return ret; > + } else { > + ret = of_pci_prop_intr_ctrl(pdev, ocs, np); > + if (ret) > + return ret; > } > > ret = of_pci_prop_ranges(pdev, ocs, np); > -- > 2.45.0 >
On Thu, Jun 06, 2024 at 02:26:12PM -0500, Bjorn Helgaas wrote: > On Mon, May 27, 2024 at 06:14:44PM +0200, Herve Codina wrote: > > PCI devices and bridges DT nodes created during the PCI scan are created > > with the interrupt-map property set to handle interrupts. > > > > In order to set this interrupt-map property at a specific level, a > > phandle to the parent interrupt controller is needed. On systems that > > are not fully described by a device-tree, the parent interrupt > > controller may be unavailable (i.e. not described by the device-tree). > > > > As mentioned in the [1], avoiding the use of the interrupt-map property > > and considering a PCI device as an interrupt controller itself avoid the > > use of a parent interrupt phandle. > > > > In that case, the PCI device itself as an interrupt controller is > > responsible for routing the interrupts described in the device-tree > > world (DT overlay) to the PCI interrupts. > > > > Add the 'interrupt-controller' property in the PCI device DT node. > > > > [1]: https://lore.kernel.org/lkml/CAL_Jsq+je7+9ATR=B6jXHjEJHjn24vQFs4Tvi9=vhDeK9n42Aw@mail.gmail.com/ > > > > Signed-off-by: Herve Codina <herve.codina@bootlin.com> > > No objection from me, but I'd like an ack/review from Rob. Given it is more DT patches in the series, how about I take them and this one with your ack instead? Rob
On Mon, Jun 10, 2024 at 03:37:35PM -0600, Rob Herring wrote: > On Thu, Jun 06, 2024 at 02:26:12PM -0500, Bjorn Helgaas wrote: > > On Mon, May 27, 2024 at 06:14:44PM +0200, Herve Codina wrote: > > > PCI devices and bridges DT nodes created during the PCI scan are created > > > with the interrupt-map property set to handle interrupts. > > > > > > In order to set this interrupt-map property at a specific level, a > > > phandle to the parent interrupt controller is needed. On systems that > > > are not fully described by a device-tree, the parent interrupt > > > controller may be unavailable (i.e. not described by the device-tree). > > > > > > As mentioned in the [1], avoiding the use of the interrupt-map property > > > and considering a PCI device as an interrupt controller itself avoid the > > > use of a parent interrupt phandle. > > > > > > In that case, the PCI device itself as an interrupt controller is > > > responsible for routing the interrupts described in the device-tree > > > world (DT overlay) to the PCI interrupts. > > > > > > Add the 'interrupt-controller' property in the PCI device DT node. > > > > > > [1]: https://lore.kernel.org/lkml/CAL_Jsq+je7+9ATR=B6jXHjEJHjn24vQFs4Tvi9=vhDeK9n42Aw@mail.gmail.com/ > > > > > > Signed-off-by: Herve Codina <herve.codina@bootlin.com> > > > > No objection from me, but I'd like an ack/review from Rob. > > Given it is more DT patches in the series, how about I take them and > this one with your ack instead? Sure. There's very little PCI content here, so I didn't plan to take them; I just thought this needed at least your ack :) Acked-by: Bjorn Helgaas <bhelgaas@google.com>
diff --git a/drivers/pci/of_property.c b/drivers/pci/of_property.c index 03539e505372..5a0b98e69795 100644 --- a/drivers/pci/of_property.c +++ b/drivers/pci/of_property.c @@ -183,6 +183,26 @@ static int of_pci_prop_interrupts(struct pci_dev *pdev, return of_changeset_add_prop_u32(ocs, np, "interrupts", (u32)pin); } +static int of_pci_prop_intr_ctrl(struct pci_dev *pdev, struct of_changeset *ocs, + struct device_node *np) +{ + int ret; + u8 pin; + + ret = pci_read_config_byte(pdev, PCI_INTERRUPT_PIN, &pin); + if (ret != 0) + return ret; + + if (!pin) + return 0; + + ret = of_changeset_add_prop_u32(ocs, np, "#interrupt-cells", 1); + if (ret) + return ret; + + return of_changeset_add_prop_bool(ocs, np, "interrupt-controller"); +} + static int of_pci_prop_intr_map(struct pci_dev *pdev, struct of_changeset *ocs, struct device_node *np) { @@ -336,6 +356,10 @@ int of_pci_add_properties(struct pci_dev *pdev, struct of_changeset *ocs, ret = of_pci_prop_intr_map(pdev, ocs, np); if (ret) return ret; + } else { + ret = of_pci_prop_intr_ctrl(pdev, ocs, np); + if (ret) + return ret; } ret = of_pci_prop_ranges(pdev, ocs, np);
PCI devices and bridges DT nodes created during the PCI scan are created with the interrupt-map property set to handle interrupts. In order to set this interrupt-map property at a specific level, a phandle to the parent interrupt controller is needed. On systems that are not fully described by a device-tree, the parent interrupt controller may be unavailable (i.e. not described by the device-tree). As mentioned in the [1], avoiding the use of the interrupt-map property and considering a PCI device as an interrupt controller itself avoid the use of a parent interrupt phandle. In that case, the PCI device itself as an interrupt controller is responsible for routing the interrupts described in the device-tree world (DT overlay) to the PCI interrupts. Add the 'interrupt-controller' property in the PCI device DT node. [1]: https://lore.kernel.org/lkml/CAL_Jsq+je7+9ATR=B6jXHjEJHjn24vQFs4Tvi9=vhDeK9n42Aw@mail.gmail.com/ Signed-off-by: Herve Codina <herve.codina@bootlin.com> --- drivers/pci/of_property.c | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+)