PCI irq affinity fix for 4.12

PCI irq affinity fix for 4.12

[Christoph Hellwig]

Hi Bjorn,

Steve found a case where the !CONFIG_PCI_MSI stub for pci_alloc_irq_vectors
did the wrong thing when used with the newly converted xhci driver.  While
the PCI code has been wrong since day 1 the fact that a newly converted
drivers triggers it makes it a regression in this cycle.

[PATCH] PCI/msi: fix the pci_alloc_irq_vectors_affinity stub

[Christoph Hellwig]

We need to return an error for any call that asks for MSI / MSI-X
vectors only, so that non-trivial fallback logic can work properly.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reported-by: Steven Rostedt <rostedt@goodmis.org>
Fixes: aff17164 ("PCI: Provide sensible IRQ vector alloc/free routines")
---
 include/linux/pci.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/linux/pci.h b/include/linux/pci.h
index 33c2b0b77429..5a7fd3b6a7b9 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -1342,7 +1342,7 @@ pci_alloc_irq_vectors_affinity(struct pci_dev *dev, unsigned int min_vecs,
 			       unsigned int max_vecs, unsigned int flags,
 			       const struct irq_affinity *aff_desc)
 {
-	if (min_vecs > 1)
+	if (min_vecs > 1 || !(flags & PCI_IRQ_LEGACY))
 		return -EINVAL;
 	return 1;
 }
-- 
2.11.0

[PATCH v2] PCI/msi: fix the pci_alloc_irq_vectors_affinity stub

[Christoph Hellwig]

We need to return an error for any call that asks for MSI / MSI-X
vectors only, so that non-trivial fallback logic can work properly.

Also valid dev->irq and use the "correct" errno value based on feedback
from Linus.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reported-by: Steven Rostedt <rostedt@goodmis.org>
Fixes: aff17164 ("PCI: Provide sensible IRQ vector alloc/free routines")
---

Changes from V1:
 - use == comparism
 - return -ENOSPC
 - verify dev->irq

 include/linux/pci.h | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/include/linux/pci.h b/include/linux/pci.h
index 33c2b0b77429..fc2e832d7b9c 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -1342,9 +1342,9 @@ pci_alloc_irq_vectors_affinity(struct pci_dev *dev, unsigned int min_vecs,
 			       unsigned int max_vecs, unsigned int flags,
 			       const struct irq_affinity *aff_desc)
 {
-	if (min_vecs > 1)
-		return -EINVAL;
-	return 1;
+	if ((flags & PCI_IRQ_LEGACY) && min_vecs == 1 && dev->irq)
+		return 1;
+	return -ENOSPC;
 }
 
 static inline void pci_free_irq_vectors(struct pci_dev *dev)
-- 
2.11.0

Re: [PATCH v2] PCI/msi: fix the pci_alloc_irq_vectors_affinity stub

[Christoph Hellwig]

Bjorn,

can you take this regression fix and sent it to Linux before -rc3?
I had hope we could get it into -rc2 but already missed that, and
I haven't heard a comment since..

On Sat, May 20, 2017 at 06:59:54PM +0200, Christoph Hellwig wrote:
> We need to return an error for any call that asks for MSI / MSI-X
> vectors only, so that non-trivial fallback logic can work properly.
> 
> Also valid dev->irq and use the "correct" errno value based on feedback
> from Linus.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> Reported-by: Steven Rostedt <rostedt@goodmis.org>
> Fixes: aff17164 ("PCI: Provide sensible IRQ vector alloc/free routines")
> ---
> 
> Changes from V1:
>  - use == comparism
>  - return -ENOSPC
>  - verify dev->irq
> 
>  include/linux/pci.h | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/include/linux/pci.h b/include/linux/pci.h
> index 33c2b0b77429..fc2e832d7b9c 100644
> --- a/include/linux/pci.h
> +++ b/include/linux/pci.h
> @@ -1342,9 +1342,9 @@ pci_alloc_irq_vectors_affinity(struct pci_dev *dev, unsigned int min_vecs,
>  			       unsigned int max_vecs, unsigned int flags,
>  			       const struct irq_affinity *aff_desc)
>  {
> -	if (min_vecs > 1)
> -		return -EINVAL;
> -	return 1;
> +	if ((flags & PCI_IRQ_LEGACY) && min_vecs == 1 && dev->irq)
> +		return 1;
> +	return -ENOSPC;
>  }
>  
>  static inline void pci_free_irq_vectors(struct pci_dev *dev)
> -- 
> 2.11.0
---end quoted text---

Re: [PATCH v2] PCI/msi: fix the pci_alloc_irq_vectors_affinity stub

[Linus Torvalds]

On Thu, May 25, 2017 at 10:46 PM, Christoph Hellwig <hch@lst.de> wrote:
> Bjorn,
>
> can you take this regression fix and sent it to Linux before -rc3?
> I had hope we could get it into -rc2 but already missed that, and
> I haven't heard a comment since..

I took it directly..

                 Linus