[PATCH] PCI: designware: reject MSI-X irqs

[PATCH] PCI: designware: reject MSI-X irqs

[Lucas Stach]

The DW PCIe MSI hardware does not support MSI-X irqs. Setting those up
failed as a side effect of a bug which was fixed by 91f8ae823f2b
(PCI: designware: Setup and clear exactly one MSI at a time).

Now that this bug is fixed MSI-X irqs need to be rejected explicitly,
otherwise devices trying to use them may end up with incorrectly working
interrupts.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Cc: <stable@vger.kernel.org> # 3.18
---
 drivers/pci/host/pcie-designware.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/pci/host/pcie-designware.c b/drivers/pci/host/pcie-designware.c
index df781cdf13c1..17ca98657a28 100644
--- a/drivers/pci/host/pcie-designware.c
+++ b/drivers/pci/host/pcie-designware.c
@@ -283,6 +283,9 @@ static int dw_msi_setup_irq(struct msi_controller *chip, struct pci_dev *pdev,
 	struct msi_msg msg;
 	struct pcie_port *pp = sys_to_pcie(pdev->bus->sysdata);
 
+	if (desc->msi_attrib.is_msix)
+		return -EINVAL;
+
 	irq = assign_irq(1, desc, &pos);
 	if (irq < 0)
 		return irq;
-- 
2.1.4

Re: [PATCH] PCI: designware: reject MSI-X irqs

[Jingoo Han]

On Monday, January 26, 2015 7:55 PM, Lucas Stach wrote:
> 
> The DW PCIe MSI hardware does not support MSI-X irqs. Setting those up
> failed as a side effect of a bug which was fixed by 91f8ae823f2b
> (PCI: designware: Setup and clear exactly one MSI at a time).
> 
> Now that this bug is fixed MSI-X irqs need to be rejected explicitly,
> otherwise devices trying to use them may end up with incorrectly working
> interrupts.

Right.

> 
> Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
> Cc: <stable@vger.kernel.org> # 3.18

Acked-by: Jingoo Han <jg1.han@samsung.com>

Best regards,
Jingoo Han

> ---
>  drivers/pci/host/pcie-designware.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/drivers/pci/host/pcie-designware.c b/drivers/pci/host/pcie-designware.c
> index df781cdf13c1..17ca98657a28 100644
> --- a/drivers/pci/host/pcie-designware.c
> +++ b/drivers/pci/host/pcie-designware.c
> @@ -283,6 +283,9 @@ static int dw_msi_setup_irq(struct msi_controller *chip, struct pci_dev *pdev,
>  	struct msi_msg msg;
>  	struct pcie_port *pp = sys_to_pcie(pdev->bus->sysdata);
> 
> +	if (desc->msi_attrib.is_msix)
> +		return -EINVAL;
> +
>  	irq = assign_irq(1, desc, &pos);
>  	if (irq < 0)
>  		return irq;
> --
> 2.1.4

Re: [PATCH] PCI: designware: reject MSI-X irqs

[Bjorn Helgaas]

On Mon, Jan 26, 2015 at 11:54:44AM +0100, Lucas Stach wrote:
> The DW PCIe MSI hardware does not support MSI-X irqs. Setting those up
> failed as a side effect of a bug which was fixed by 91f8ae823f2b
> (PCI: designware: Setup and clear exactly one MSI at a time).
> 
> Now that this bug is fixed MSI-X irqs need to be rejected explicitly,
> otherwise devices trying to use them may end up with incorrectly working
> interrupts.
> 
> Signed-off-by: Lucas Stach <l.stach@pengutronix.de>

Applied with Jingoo's ack to pci/host-designware for v3.20, thanks!

> Cc: <stable@vger.kernel.org> # 3.18
> ---
>  drivers/pci/host/pcie-designware.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/drivers/pci/host/pcie-designware.c b/drivers/pci/host/pcie-designware.c
> index df781cdf13c1..17ca98657a28 100644
> --- a/drivers/pci/host/pcie-designware.c
> +++ b/drivers/pci/host/pcie-designware.c
> @@ -283,6 +283,9 @@ static int dw_msi_setup_irq(struct msi_controller *chip, struct pci_dev *pdev,
>  	struct msi_msg msg;
>  	struct pcie_port *pp = sys_to_pcie(pdev->bus->sysdata);
>  
> +	if (desc->msi_attrib.is_msix)
> +		return -EINVAL;
> +
>  	irq = assign_irq(1, desc, &pos);
>  	if (irq < 0)
>  		return irq;
> -- 
> 2.1.4
> 

Re: [PATCH] PCI: designware: reject MSI-X irqs

[Lucas Stach]

Hi Bjorn,

Am Montag, den 26.01.2015, 12:12 -0600 schrieb Bjorn Helgaas:
> On Mon, Jan 26, 2015 at 11:54:44AM +0100, Lucas Stach wrote:
> > The DW PCIe MSI hardware does not support MSI-X irqs. Setting those up
> > failed as a side effect of a bug which was fixed by 91f8ae823f2b
> > (PCI: designware: Setup and clear exactly one MSI at a time).
> > 
> > Now that this bug is fixed MSI-X irqs need to be rejected explicitly,
> > otherwise devices trying to use them may end up with incorrectly working
> > interrupts.
> > 
> > Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
> 
> Applied with Jingoo's ack to pci/host-designware for v3.20, thanks!
> 
I would be happier if this could still find its way into v3.19 if you
are going to send another fixes pull, as this is clearly a regression
even if it was already introduced with v3.18.

> > Cc: <stable@vger.kernel.org> # 3.18
> > ---
> >  drivers/pci/host/pcie-designware.c | 3 +++
> >  1 file changed, 3 insertions(+)
> > 
> > diff --git a/drivers/pci/host/pcie-designware.c b/drivers/pci/host/pcie-designware.c
> > index df781cdf13c1..17ca98657a28 100644
> > --- a/drivers/pci/host/pcie-designware.c
> > +++ b/drivers/pci/host/pcie-designware.c
> > @@ -283,6 +283,9 @@ static int dw_msi_setup_irq(struct msi_controller *chip, struct pci_dev *pdev,
> >  	struct msi_msg msg;
> >  	struct pcie_port *pp = sys_to_pcie(pdev->bus->sysdata);
> >  
> > +	if (desc->msi_attrib.is_msix)
> > +		return -EINVAL;
> > +
> >  	irq = assign_irq(1, desc, &pos);
> >  	if (irq < 0)
> >  		return irq;
> > -- 
> > 2.1.4
> > 

Re: [PATCH] PCI: designware: reject MSI-X irqs

[Bjorn Helgaas]

On Tue, Jan 27, 2015 at 3:22 AM, Lucas Stach <l.stach@pengutronix.de> wrote:
> Hi Bjorn,
>
> Am Montag, den 26.01.2015, 12:12 -0600 schrieb Bjorn Helgaas:
>> On Mon, Jan 26, 2015 at 11:54:44AM +0100, Lucas Stach wrote:
>> > The DW PCIe MSI hardware does not support MSI-X irqs. Setting those up
>> > failed as a side effect of a bug which was fixed by 91f8ae823f2b
>> > (PCI: designware: Setup and clear exactly one MSI at a time).
>> >
>> > Now that this bug is fixed MSI-X irqs need to be rejected explicitly,
>> > otherwise devices trying to use them may end up with incorrectly working
>> > interrupts.
>> >
>> > Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
>>
>> Applied with Jingoo's ack to pci/host-designware for v3.20, thanks!
>>
> I would be happier if this could still find its way into v3.19 if you
> are going to send another fixes pull, as this is clearly a regression
> even if it was already introduced with v3.18.

Added to for-linus for v3.19 and dropped from pci/host-designware.

In general, I apply changes for the next release unless it's clear
that they need to be in the current release.  In this case, I couldn't
tell from the changelog whether this was a theoretical issue or a
"device X worked on v3.17 but stopped working on v3.18 and this patch
fixes it" issue.

>> > Cc: <stable@vger.kernel.org> # 3.18
>> > ---
>> >  drivers/pci/host/pcie-designware.c | 3 +++
>> >  1 file changed, 3 insertions(+)
>> >
>> > diff --git a/drivers/pci/host/pcie-designware.c b/drivers/pci/host/pcie-designware.c
>> > index df781cdf13c1..17ca98657a28 100644
>> > --- a/drivers/pci/host/pcie-designware.c
>> > +++ b/drivers/pci/host/pcie-designware.c
>> > @@ -283,6 +283,9 @@ static int dw_msi_setup_irq(struct msi_controller *chip, struct pci_dev *pdev,
>> >     struct msi_msg msg;
>> >     struct pcie_port *pp = sys_to_pcie(pdev->bus->sysdata);
>> >
>> > +   if (desc->msi_attrib.is_msix)
>> > +           return -EINVAL;
>> > +
>> >     irq = assign_irq(1, desc, &pos);
>> >     if (irq < 0)
>> >             return irq;
>> > --
>> > 2.1.4
>> >
>
>