[PATCH] ACPI / PCI: fix a memory leak bug

[PATCH] ACPI / PCI: fix a memory leak bug

[Wenwen Wang]

In acpi_pci_irq_enable(), 'entry' is allocated by invoking
acpi_pci_irq_lookup(). However, it is not deallocated if
acpi_pci_irq_valid() returns false, leading to a memory leak. To fix this
issue, free 'entry' before returning 0.

Signed-off-by: Wenwen Wang <wenwen@cs.uga.edu>
---
 drivers/acpi/pci_irq.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/acpi/pci_irq.c b/drivers/acpi/pci_irq.c
index d2549ae..dea8a60 100644
--- a/drivers/acpi/pci_irq.c
+++ b/drivers/acpi/pci_irq.c
@@ -449,8 +449,10 @@ int acpi_pci_irq_enable(struct pci_dev *dev)
 		 * No IRQ known to the ACPI subsystem - maybe the BIOS /
 		 * driver reported one, then use it. Exit in any case.
 		 */
-		if (!acpi_pci_irq_valid(dev, pin))
+		if (!acpi_pci_irq_valid(dev, pin)) {
+			kfree(entry);
 			return 0;
+		}
 
 		if (acpi_isa_register_gsi(dev))
 			dev_warn(&dev->dev, "PCI INT %c: no GSI\n",
-- 
2.7.4

Re: [PATCH] ACPI / PCI: fix a memory leak bug

[Bjorn Helgaas]

The subject line should give a clue about where the leak is, e.g.,

  ACPI / PCI: fix acpi_pci_irq_enable() memory leak

On Thu, Aug 15, 2019 at 11:33:22PM -0500, Wenwen Wang wrote:
> In acpi_pci_irq_enable(), 'entry' is allocated by invoking
> acpi_pci_irq_lookup(). However, it is not deallocated if
> acpi_pci_irq_valid() returns false, leading to a memory leak. To fix this
> issue, free 'entry' before returning 0.

I think the corresponding kzalloc() is the one in
acpi_pci_irq_check_entry().

> Signed-off-by: Wenwen Wang <wenwen@cs.uga.edu>
> ---
>  drivers/acpi/pci_irq.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/acpi/pci_irq.c b/drivers/acpi/pci_irq.c
> index d2549ae..dea8a60 100644
> --- a/drivers/acpi/pci_irq.c
> +++ b/drivers/acpi/pci_irq.c
> @@ -449,8 +449,10 @@ int acpi_pci_irq_enable(struct pci_dev *dev)
>  		 * No IRQ known to the ACPI subsystem - maybe the BIOS /
>  		 * driver reported one, then use it. Exit in any case.
>  		 */
> -		if (!acpi_pci_irq_valid(dev, pin))
> +		if (!acpi_pci_irq_valid(dev, pin)) {
> +			kfree(entry);
>  			return 0;
> +		}

Looks like we missed this when e237a5518425 ("x86/ACPI/PCI: Recognize
that Interrupt Line 255 means "not connected"") was merged.

You could add:

Fixes: e237a5518425 ("x86/ACPI/PCI: Recognize that Interrupt Line 255 means "not connected"")

>  		if (acpi_isa_register_gsi(dev))
>  			dev_warn(&dev->dev, "PCI INT %c: no GSI\n",
> -- 
> 2.7.4
> 

Re: [PATCH] ACPI / PCI: fix a memory leak bug

[Wenwen Wang]

On Mon, Aug 19, 2019 at 5:23 PM Bjorn Helgaas <helgaas@kernel.org> wrote:
>
> The subject line should give a clue about where the leak is, e.g.,
>
>   ACPI / PCI: fix acpi_pci_irq_enable() memory leak
>
> On Thu, Aug 15, 2019 at 11:33:22PM -0500, Wenwen Wang wrote:
> > In acpi_pci_irq_enable(), 'entry' is allocated by invoking
> > acpi_pci_irq_lookup(). However, it is not deallocated if
> > acpi_pci_irq_valid() returns false, leading to a memory leak. To fix this
> > issue, free 'entry' before returning 0.
>
> I think the corresponding kzalloc() is the one in
> acpi_pci_irq_check_entry().
>
> > Signed-off-by: Wenwen Wang <wenwen@cs.uga.edu>
> > ---
> >  drivers/acpi/pci_irq.c | 4 +++-
> >  1 file changed, 3 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/acpi/pci_irq.c b/drivers/acpi/pci_irq.c
> > index d2549ae..dea8a60 100644
> > --- a/drivers/acpi/pci_irq.c
> > +++ b/drivers/acpi/pci_irq.c
> > @@ -449,8 +449,10 @@ int acpi_pci_irq_enable(struct pci_dev *dev)
> >                * No IRQ known to the ACPI subsystem - maybe the BIOS /
> >                * driver reported one, then use it. Exit in any case.
> >                */
> > -             if (!acpi_pci_irq_valid(dev, pin))
> > +             if (!acpi_pci_irq_valid(dev, pin)) {
> > +                     kfree(entry);
> >                       return 0;
> > +             }
>
> Looks like we missed this when e237a5518425 ("x86/ACPI/PCI: Recognize
> that Interrupt Line 255 means "not connected"") was merged.
>
> You could add:
>
> Fixes: e237a5518425 ("x86/ACPI/PCI: Recognize that Interrupt Line 255 means "not connected"")
>
> >               if (acpi_isa_register_gsi(dev))
> >                       dev_warn(&dev->dev, "PCI INT %c: no GSI\n",
> > --
> > 2.7.4
> >

Thanks for your comments and suggestions! I will rework the patch.

Wenwen