[PATCH] PCI/MSI: pci_irq_get_affinity() should cope with NULL affinity vector

[PATCH] PCI/MSI: pci_irq_get_affinity() should cope with NULL affinity vector

[Jan Beulich]

msi_setup_entry() only logs a message when the affinity vector can't be
allocated, and hence pci_irq_get_affinity() indexing entry->affinity is
wrong without a prior check.

Fixes: ee8d41e53efe "pci/msi: Retrieve affinity for a vector"
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Christoph Hellwig <hch@lst.de>
---
 drivers/pci/msi.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

--- 4.9-rc4/drivers/pci/msi.c
+++ 4.9-rc4-PCI-MSI-no-affinity/drivers/pci/msi.c
@@ -1294,7 +1294,8 @@ const struct cpumask *pci_irq_get_affini
 	} else if (dev->msi_enabled) {
 		struct msi_desc *entry = first_pci_msi_entry(dev);
 
-		if (WARN_ON_ONCE(!entry || nr >= entry->nvec_used))
+		if (WARN_ON_ONCE(!entry || !entry->affinity ||
+				 nr >= entry->nvec_used))
 			return NULL;
 
 		return &entry->affinity[nr];

Re: [PATCH] PCI/MSI: pci_irq_get_affinity() should cope with NULL affinity vector

[Christoph Hellwig]

On Tue, Nov 08, 2016 at 12:43:54AM -0700, Jan Beulich wrote:
> msi_setup_entry() only logs a message when the affinity vector can't be
> allocated, and hence pci_irq_get_affinity() indexing entry->affinity is
> wrong without a prior check.

Looks fine:

Reviewed-by: Christoph Hellwig <hch@lst.de>

But just curious: did you run into that case in practice?

Re: [PATCH] PCI/MSI: pci_irq_get_affinity() should cope with NULL affinity vector

[Jan Beulich]

>>> On 08.11.16 at 15:59, <hch@lst.de> wrote:
> On Tue, Nov 08, 2016 at 12:43:54AM -0700, Jan Beulich wrote:
>> msi_setup_entry() only logs a message when the affinity vector can't be
>> allocated, and hence pci_irq_get_affinity() indexing entry->affinity is
>> wrong without a prior check.
> 
> Looks fine:
> 
> Reviewed-by: Christoph Hellwig <hch@lst.de>
> 
> But just curious: did you run into that case in practice?

No, I've just noticed it while re-basing other changes onto 4.9-rc.

Jan

Re: [PATCH] PCI/MSI: pci_irq_get_affinity() should cope with NULL affinity vector

[Bjorn Helgaas]

On Tue, Nov 08, 2016 at 12:43:54AM -0700, Jan Beulich wrote:
> msi_setup_entry() only logs a message when the affinity vector can't be
> allocated, and hence pci_irq_get_affinity() indexing entry->affinity is
> wrong without a prior check.
> 
> Fixes: ee8d41e53efe "pci/msi: Retrieve affinity for a vector"
> Signed-off-by: Jan Beulich <jbeulich@suse.com>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: Christoph Hellwig <hch@lst.de>

Applied to pci/msi for v4.10 with Christoph's reviewed-by and the following
changelog:

  PCI/MSI: Check for NULL affinity mask in pci_irq_get_affinity()
  
  If msi_setup_entry() fails to allocate an affinity mask, it logs a message
  but continues on and allocates an MSI entry with entry->affinity == NULL.
  
  Check for this case in pci_irq_get_affinity() so we don't try to
  dereference a NULL pointer.

> ---
>  drivers/pci/msi.c |    3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> --- 4.9-rc4/drivers/pci/msi.c
> +++ 4.9-rc4-PCI-MSI-no-affinity/drivers/pci/msi.c
> @@ -1294,7 +1294,8 @@ const struct cpumask *pci_irq_get_affini
>  	} else if (dev->msi_enabled) {
>  		struct msi_desc *entry = first_pci_msi_entry(dev);
>  
> -		if (WARN_ON_ONCE(!entry || nr >= entry->nvec_used))
> +		if (WARN_ON_ONCE(!entry || !entry->affinity ||
> +				 nr >= entry->nvec_used))
>  			return NULL;
>  
>  		return &entry->affinity[nr];
> 
> 
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-pci" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html