2019-03-11 11:47:21

by Bharat Kumar Gogada

[permalink] [raw]
Subject: [PATCH v2] PCI: xilinx-nwl: Fix Multi MSI data programming

The current Multi MSI data programming fails if multiple end points
requesting MSI and multi MSI are connected with switch, i.e the current
multi MSI data being given is not considering the number of vectors
being requested in case of multi MSI.
Due to this if multiple end points are connected and requesting MSI
and multi MSI combination, the multi MSI data is ending up using
wrong MSI data, which might be used by different device.

Fix Multi MSI data programming with required alignment by
using number of vectors being requested.

Fixes: ab597d35ef11 ("PCI: xilinx-nwl: Add support for Xilinx NWL PCIe
Host Controller")
Signed-off-by: Bharat Kumar Gogada <[email protected]>
---
V2:
- Added more description of fix
---
drivers/pci/controller/pcie-xilinx-nwl.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/pci/controller/pcie-xilinx-nwl.c b/drivers/pci/controller/pcie-xilinx-nwl.c
index 81538d7..36669c5 100644
--- a/drivers/pci/controller/pcie-xilinx-nwl.c
+++ b/drivers/pci/controller/pcie-xilinx-nwl.c
@@ -484,7 +484,7 @@ static int nwl_irq_domain_alloc(struct irq_domain *domain, unsigned int virq,

mutex_lock(&msi->lock);
bit = bitmap_find_next_zero_area(msi->bitmap, INT_PCI_MSI_NR, 0,
- nr_irqs, 0);
+ nr_irqs, nr_irqs - 1);
if (bit >= INT_PCI_MSI_NR) {
mutex_unlock(&msi->lock);
return -ENOSPC;
--
2.7.4



2019-04-01 17:01:45

by Bharat Kumar Gogada

[permalink] [raw]
Subject: RE: [PATCH v2] PCI: xilinx-nwl: Fix Multi MSI data programming

Hi All,

Please let me know if anyone has any inputs on this.

Regards,
Bharat
>
> The current Multi MSI data programming fails if multiple end points
> requesting MSI and multi MSI are connected with switch, i.e the current
> multi MSI data being given is not considering the number of vectors being
> requested in case of multi MSI.
> Due to this if multiple end points are connected and requesting MSI and
> multi MSI combination, the multi MSI data is ending up using wrong MSI
> data, which might be used by different device.
>
> Fix Multi MSI data programming with required alignment by using number
> of vectors being requested.
>
> Fixes: ab597d35ef11 ("PCI: xilinx-nwl: Add support for Xilinx NWL PCIe Host
> Controller")
> Signed-off-by: Bharat Kumar Gogada <[email protected]>
> ---
> V2:
> - Added more description of fix
> ---
> drivers/pci/controller/pcie-xilinx-nwl.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/pci/controller/pcie-xilinx-nwl.c
> b/drivers/pci/controller/pcie-xilinx-nwl.c
> index 81538d7..36669c5 100644
> --- a/drivers/pci/controller/pcie-xilinx-nwl.c
> +++ b/drivers/pci/controller/pcie-xilinx-nwl.c
> @@ -484,7 +484,7 @@ static int nwl_irq_domain_alloc(struct irq_domain
> *domain, unsigned int virq,
>
> mutex_lock(&msi->lock);
> bit = bitmap_find_next_zero_area(msi->bitmap, INT_PCI_MSI_NR, 0,
> - nr_irqs, 0);
> + nr_irqs, nr_irqs - 1);
> if (bit >= INT_PCI_MSI_NR) {
> mutex_unlock(&msi->lock);
> return -ENOSPC;
> --
> 2.7.4

2019-04-02 16:11:37

by Lorenzo Pieralisi

[permalink] [raw]
Subject: Re: [PATCH v2] PCI: xilinx-nwl: Fix Multi MSI data programming

On Mon, Apr 01, 2019 at 05:00:40PM +0000, Bharat Kumar Gogada wrote:
> Hi All,
>
> Please let me know if anyone has any inputs on this.
>
> Regards,
> Bharat
> >
> > The current Multi MSI data programming fails if multiple end points
> > requesting MSI and multi MSI are connected with switch, i.e the current
> > multi MSI data being given is not considering the number of vectors being
> > requested in case of multi MSI.
> > Due to this if multiple end points are connected and requesting MSI and
> > multi MSI combination, the multi MSI data is ending up using wrong MSI
> > data, which might be used by different device.
> >
> > Fix Multi MSI data programming with required alignment by using number
> > of vectors being requested.

I still do not understand what you mean I am sorry. An example is worth
two thousand words, I would start with that as it stands this commit
log does not provide any information on what you are actually fixing.

Lorenzo

> > Fixes: ab597d35ef11 ("PCI: xilinx-nwl: Add support for Xilinx NWL PCIe Host
> > Controller")
> > Signed-off-by: Bharat Kumar Gogada <[email protected]>
> > ---
> > V2:
> > - Added more description of fix
> > ---
> > drivers/pci/controller/pcie-xilinx-nwl.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/drivers/pci/controller/pcie-xilinx-nwl.c
> > b/drivers/pci/controller/pcie-xilinx-nwl.c
> > index 81538d7..36669c5 100644
> > --- a/drivers/pci/controller/pcie-xilinx-nwl.c
> > +++ b/drivers/pci/controller/pcie-xilinx-nwl.c
> > @@ -484,7 +484,7 @@ static int nwl_irq_domain_alloc(struct irq_domain
> > *domain, unsigned int virq,
> >
> > mutex_lock(&msi->lock);
> > bit = bitmap_find_next_zero_area(msi->bitmap, INT_PCI_MSI_NR, 0,
> > - nr_irqs, 0);
> > + nr_irqs, nr_irqs - 1);
> > if (bit >= INT_PCI_MSI_NR) {
> > mutex_unlock(&msi->lock);
> > return -ENOSPC;
> > --
> > 2.7.4
>

2019-04-02 16:25:11

by Ard Biesheuvel

[permalink] [raw]
Subject: Re: [PATCH v2] PCI: xilinx-nwl: Fix Multi MSI data programming

On Mon, 11 Mar 2019 at 12:46, Bharat Kumar Gogada
<[email protected]> wrote:
>
> The current Multi MSI data programming fails if multiple end points
> requesting MSI and multi MSI are connected with switch, i.e the current
> multi MSI data being given is not considering the number of vectors
> being requested in case of multi MSI.
> Due to this if multiple end points are connected and requesting MSI
> and multi MSI combination, the multi MSI data is ending up using
> wrong MSI data, which might be used by different device.
>
> Fix Multi MSI data programming with required alignment by
> using number of vectors being requested.
>
> Fixes: ab597d35ef11 ("PCI: xilinx-nwl: Add support for Xilinx NWL PCIe
> Host Controller")
> Signed-off-by: Bharat Kumar Gogada <[email protected]>
> ---
> V2:
> - Added more description of fix
> ---
> drivers/pci/controller/pcie-xilinx-nwl.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/pci/controller/pcie-xilinx-nwl.c b/drivers/pci/controller/pcie-xilinx-nwl.c
> index 81538d7..36669c5 100644
> --- a/drivers/pci/controller/pcie-xilinx-nwl.c
> +++ b/drivers/pci/controller/pcie-xilinx-nwl.c
> @@ -484,7 +484,7 @@ static int nwl_irq_domain_alloc(struct irq_domain *domain, unsigned int virq,
>
> mutex_lock(&msi->lock);
> bit = bitmap_find_next_zero_area(msi->bitmap, INT_PCI_MSI_NR, 0,
> - nr_irqs, 0);
> + nr_irqs, nr_irqs - 1);

Are you sure nr_irqs is guaranteed to be a power of 2?

> if (bit >= INT_PCI_MSI_NR) {
> mutex_unlock(&msi->lock);
> return -ENOSPC;
> --
> 2.7.4
>
>
> _______________________________________________
> linux-arm-kernel mailing list
> [email protected]
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

2019-04-04 11:49:46

by Bharat Kumar Gogada

[permalink] [raw]
Subject: RE: [PATCH v2] PCI: xilinx-nwl: Fix Multi MSI data programming

> On Mon, Apr 01, 2019 at 05:00:40PM +0000, Bharat Kumar Gogada wrote:
> > Hi All,
> >
> > Please let me know if anyone has any inputs on this.
> >
> > Regards,
> > Bharat
> > >
> > > The current Multi MSI data programming fails if multiple end points
> > > requesting MSI and multi MSI are connected with switch, i.e the
> > > current multi MSI data being given is not considering the number of
> > > vectors being requested in case of multi MSI.
> > > Due to this if multiple end points are connected and requesting MSI
> > > and multi MSI combination, the multi MSI data is ending up using
> > > wrong MSI data, which might be used by different device.
> > >
> > > Fix Multi MSI data programming with required alignment by using
> > > number of vectors being requested.
>
> I still do not understand what you mean I am sorry. An example is worth
> two thousand words, I would start with that as it stands this commit log
> does not provide any information on what you are actually fixing.
>
Hi Lorenzo, Thanks for your time.
Yes, will add an example to describe the problem.

Regards,
Bharat
> > > Fixes: ab597d35ef11 ("PCI: xilinx-nwl: Add support for Xilinx NWL
> > > PCIe Host
> > > Controller")
> > > Signed-off-by: Bharat Kumar Gogada
> <[email protected]>
> > > ---
> > > V2:
> > > - Added more description of fix
> > > ---
> > > drivers/pci/controller/pcie-xilinx-nwl.c | 2 +-
> > > 1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/drivers/pci/controller/pcie-xilinx-nwl.c
> > > b/drivers/pci/controller/pcie-xilinx-nwl.c
> > > index 81538d7..36669c5 100644
> > > --- a/drivers/pci/controller/pcie-xilinx-nwl.c
> > > +++ b/drivers/pci/controller/pcie-xilinx-nwl.c
> > > @@ -484,7 +484,7 @@ static int nwl_irq_domain_alloc(struct
> > > irq_domain *domain, unsigned int virq,
> > >
> > > mutex_lock(&msi->lock);
> > > bit = bitmap_find_next_zero_area(msi->bitmap, INT_PCI_MSI_NR, 0,
> > > - nr_irqs, 0);
> > > + nr_irqs, nr_irqs - 1);
> > > if (bit >= INT_PCI_MSI_NR) {
> > > mutex_unlock(&msi->lock);
> > > return -ENOSPC;
> > > --
> > > 2.7.4
> >

2019-04-04 11:52:48

by Bharat Kumar Gogada

[permalink] [raw]
Subject: RE: [PATCH v2] PCI: xilinx-nwl: Fix Multi MSI data programming

> Subject: Re: [PATCH v2] PCI: xilinx-nwl: Fix Multi MSI data programming
>
> On Mon, 11 Mar 2019 at 12:46, Bharat Kumar Gogada
> <[email protected]> wrote:
> >
> > The current Multi MSI data programming fails if multiple end points
> > requesting MSI and multi MSI are connected with switch, i.e the
> > current multi MSI data being given is not considering the number of
> > vectors being requested in case of multi MSI.
> > Due to this if multiple end points are connected and requesting MSI
> > and multi MSI combination, the multi MSI data is ending up using wrong
> > MSI data, which might be used by different device.
> >
> > Fix Multi MSI data programming with required alignment by using number
> > of vectors being requested.
> >
> > Fixes: ab597d35ef11 ("PCI: xilinx-nwl: Add support for Xilinx NWL PCIe
> > Host Controller")
> > Signed-off-by: Bharat Kumar Gogada <[email protected]>
> > ---
> > V2:
> > - Added more description of fix
> > ---
> > drivers/pci/controller/pcie-xilinx-nwl.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/drivers/pci/controller/pcie-xilinx-nwl.c
> > b/drivers/pci/controller/pcie-xilinx-nwl.c
> > index 81538d7..36669c5 100644
> > --- a/drivers/pci/controller/pcie-xilinx-nwl.c
> > +++ b/drivers/pci/controller/pcie-xilinx-nwl.c
> > @@ -484,7 +484,7 @@ static int nwl_irq_domain_alloc(struct irq_domain
> > *domain, unsigned int virq,
> >
> > mutex_lock(&msi->lock);
> > bit = bitmap_find_next_zero_area(msi->bitmap, INT_PCI_MSI_NR, 0,
> > - nr_irqs, 0);
> > + nr_irqs, nr_irqs - 1);
>
> Are you sure nr_irqs is guaranteed to be a power of 2?
Hi Ard, Thanks for your time.
After checking other source codes its not guaranteed, will add extra check and resend patch.

Regards,
Bharat
>
> > if (bit >= INT_PCI_MSI_NR) {
> > mutex_unlock(&msi->lock);
> > return -ENOSPC;
> > --
> > 2.7.4
> >
> >
> > _______________________________________________
> > linux-arm-kernel mailing list
> > [email protected]
> > http://lists.infradead.org/mailman/listinfo/linux-arm-kernel