On Monday, July 15, 2013 10:57:40 AM Prarit Bhargava wrote:
> Driver probe's currently do the following
>
> pci_enable_device();
> /* ... do some other init stuff, and eventually call ... */
> request_irq();
>
> After pci_enable_device() is called it is assumed that the device's irq
> value (pci_dev->irq) has been appropriately set on success. This value
> is passed into the request_irq() call.
>
> In the case that ACPI is used to determine the irq value, it is possible
> that the ACPI IRQ look up for a specific device fails and success is
> returned by pci_enable_device().
>
> The call sequence is:
>
> pci_enable_device();
> -> pci_enable_device_flags();
> ->do_pci_enable_device();
> -> pcibios_enable_device() which, if the device
> does not use MSI calls
> -> pcibios_enable_irq() which maps to
> acpi_pci_irq_enable()
> -> acpi_pci_irq_lookup()
>
> If acpi_pci_irq_lookup() cannot map the device's IRQ value it returns NULL
> as an error. The error is returned to acpi_pci_irq_enable(), but is not
> propagated further. This can result in the driver returning success for
> pci_enable_device() and the driver probe attempting to call request_irq()
> with dev->irq = 0.
>
> This patch modifies acpi_pci_irq_enable() to return an error in the case
> that an entry is not found in the ACPI tables.
Is there any known system on which this leads to observable misbehavior?
If so, what's that system?
Rafael
> Signed-off-by: Prarit Bhargava <prarit@xxxxxxxxxx>
> Cc: Len Brown <lenb@xxxxxxxxxx>
> Cc: "Rafael J. Wysocki" <rjw@xxxxxxx>
> Cc: "Bjorn Helgaas" <bhelgaas@xxxxxxxxxx>
> Cc: "Myron Stowe" <mstowe@xxxxxxxxxx>
> Cc: linux-pci@xxxxxxxxxxxxxxx
> ---
> drivers/acpi/pci_irq.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/acpi/pci_irq.c b/drivers/acpi/pci_irq.c
> index 41c5e1b..9681847 100644
> --- a/drivers/acpi/pci_irq.c
> +++ b/drivers/acpi/pci_irq.c
> @@ -430,6 +430,7 @@ int acpi_pci_irq_enable(struct pci_dev *dev)
> } else {
> dev_warn(&dev->dev, "PCI INT %c: no GSI\n",
> pin_name(pin));
> + return -ENOENT;
> }
>
> return 0;
>
--
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.
--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
