diff mbox

[1/1] PCI: check return value of pci_find_ext_capability

Message ID 1492956020-32271-1-git-send-email-bianpan201603@163.com
State Not Applicable
Headers show

Commit Message

Pan Bian April 23, 2017, 2 p.m. UTC
From: Pan Bian <bianpan2016@163.com>

Function pci_find_ext_capability() will returns 0 on failure, and its
return value should be checked before it is used. However, in function
pcie_port_enable_msix(), its return value is not checked. This patch
adds the check.

Signed-off-by: Pan Bian <bianpan2016@163.com>
---
 drivers/pci/pcie/portdrv_core.c | 2 ++
 1 file changed, 2 insertions(+)

Comments

Mika Westerberg April 24, 2017, 10:36 a.m. UTC | #1
On Sun, Apr 23, 2017 at 10:00:20PM +0800, Pan Bian wrote:
> From: Pan Bian <bianpan2016@163.com>
> 
> Function pci_find_ext_capability() will returns 0 on failure, and its
> return value should be checked before it is used. However, in function
> pcie_port_enable_msix(), its return value is not checked. This patch
> adds the check.
> 
> Signed-off-by: Pan Bian <bianpan2016@163.com>
> ---
>  drivers/pci/pcie/portdrv_core.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/drivers/pci/pcie/portdrv_core.c b/drivers/pci/pcie/portdrv_core.c
> index cea504f..001951d 100644
> --- a/drivers/pci/pcie/portdrv_core.c
> +++ b/drivers/pci/pcie/portdrv_core.c
> @@ -103,6 +103,8 @@ static int pcie_port_enable_msix(struct pci_dev *dev, int *irqs, int mask)
>  		 * interrupt message."
>  		 */
>  		pos = pci_find_ext_capability(dev, PCI_EXT_CAP_ID_ERR);
> +		if (!pos)
> +			goto out_free_irqs;

I don't think this can happen because get_port_device_capability() will
enumerate this capability and only if it exists, set PCIE_PORT_SERVICE_AER.
Bjorn Helgaas April 25, 2017, 7:31 p.m. UTC | #2
On Mon, Apr 24, 2017 at 01:36:58PM +0300, Mika Westerberg wrote:
> On Sun, Apr 23, 2017 at 10:00:20PM +0800, Pan Bian wrote:
> > From: Pan Bian <bianpan2016@163.com>
> > 
> > Function pci_find_ext_capability() will returns 0 on failure, and its
> > return value should be checked before it is used. However, in function
> > pcie_port_enable_msix(), its return value is not checked. This patch
> > adds the check.
> > 
> > Signed-off-by: Pan Bian <bianpan2016@163.com>
> > ---
> >  drivers/pci/pcie/portdrv_core.c | 2 ++
> >  1 file changed, 2 insertions(+)
> > 
> > diff --git a/drivers/pci/pcie/portdrv_core.c b/drivers/pci/pcie/portdrv_core.c
> > index cea504f..001951d 100644
> > --- a/drivers/pci/pcie/portdrv_core.c
> > +++ b/drivers/pci/pcie/portdrv_core.c
> > @@ -103,6 +103,8 @@ static int pcie_port_enable_msix(struct pci_dev *dev, int *irqs, int mask)
> >  		 * interrupt message."
> >  		 */
> >  		pos = pci_find_ext_capability(dev, PCI_EXT_CAP_ID_ERR);
> > +		if (!pos)
> > +			goto out_free_irqs;
> 
> I don't think this can happen because get_port_device_capability() will
> enumerate this capability and only if it exists, set PCIE_PORT_SERVICE_AER.

The path is more complicated than I'd like, but I think you're right.
We should also be able to use dev->aer_cap here instead of looking it
up again.

Bjorn
diff mbox

Patch

diff --git a/drivers/pci/pcie/portdrv_core.c b/drivers/pci/pcie/portdrv_core.c
index cea504f..001951d 100644
--- a/drivers/pci/pcie/portdrv_core.c
+++ b/drivers/pci/pcie/portdrv_core.c
@@ -103,6 +103,8 @@  static int pcie_port_enable_msix(struct pci_dev *dev, int *irqs, int mask)
 		 * interrupt message."
 		 */
 		pos = pci_find_ext_capability(dev, PCI_EXT_CAP_ID_ERR);
+		if (!pos)
+			goto out_free_irqs;
 		pci_read_config_dword(dev, pos + PCI_ERR_ROOT_STATUS, &reg32);
 		entry = reg32 >> 27;
 		if (entry >= nr_entries)