Instead of a direct assignment, use pci_irq_vector() to get the
Linux IRQ number.
Signed-off-by: Utkarsh Verma <[email protected]>
---
drivers/tty/serial/8250/8250_mid.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/tty/serial/8250/8250_mid.c b/drivers/tty/serial/8250/8250_mid.c
index efa0515139f8..3a279ce0e904 100644
--- a/drivers/tty/serial/8250/8250_mid.c
+++ b/drivers/tty/serial/8250/8250_mid.c
@@ -299,7 +299,7 @@ static int mid8250_probe(struct pci_dev *pdev, const struct pci_device_id *id)
memset(&uart, 0, sizeof(struct uart_8250_port));
uart.port.dev = &pdev->dev;
- uart.port.irq = pdev->irq;
+ uart.port.irq = pci_irq_vector(pdev, 0);
uart.port.private_data = mid;
uart.port.type = PORT_16750;
uart.port.iotype = UPIO_MEM;
--
2.17.1
On Tue, Aug 17, 2021 at 01:44:01PM +0530, Utkarsh Verma wrote:
> Instead of a direct assignment, use pci_irq_vector() to get the
> Linux IRQ number.
Why is this needed?
> Signed-off-by: Utkarsh Verma <[email protected]>
> ---
> drivers/tty/serial/8250/8250_mid.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/tty/serial/8250/8250_mid.c b/drivers/tty/serial/8250/8250_mid.c
> index efa0515139f8..3a279ce0e904 100644
> --- a/drivers/tty/serial/8250/8250_mid.c
> +++ b/drivers/tty/serial/8250/8250_mid.c
> @@ -299,7 +299,7 @@ static int mid8250_probe(struct pci_dev *pdev, const struct pci_device_id *id)
> memset(&uart, 0, sizeof(struct uart_8250_port));
>
> uart.port.dev = &pdev->dev;
> - uart.port.irq = pdev->irq;
> + uart.port.irq = pci_irq_vector(pdev, 0);
What problem does this solve?
Do not describe what you are doing, but rather, _why_ you are doing it.
thanks,
greg k-h
On Tue, Aug 17, 2021 at 3:59 PM Greg Kroah-Hartman
<[email protected]> wrote:
> On Tue, Aug 17, 2021 at 01:44:01PM +0530, Utkarsh Verma wrote:
> > Instead of a direct assignment, use pci_irq_vector() to get the
> > Linux IRQ number.
>
> Why is this needed?
It's not.
> > + uart.port.irq = pci_irq_vector(pdev, 0);
>
> What problem does this solve?
>
> Do not describe what you are doing, but rather, _why_ you are doing it.
I think I know what was the motivation here, but actually there is no
problem with current code. Indeed, when we enable MSI we have to
update the vIRQ line, but this is done in the platform related
->setup() callbacks (for example, dnv_setup() does it). So we have two
scenarios:
1) there is no MSI enabled and in this case pdev->irq is correct;
2) we have MSI enabled and we need to update previously saved pdev->irq.
What you, Utkarsh, are doing in the patch has no effect, because at
that point MSI will never be enabled and pci_irq_vector(pdev, 0) will
return the very same value as kept in pdev->irq.
--
With Best Regards,
Andy Shevchenko