[PATCH, RFC] PCI: use pci_intx_for_msi in pci_alloc_irq_vectors

[PATCH, RFC] PCI: use pci_intx_for_msi in pci_alloc_irq_vectors

[Christoph Hellwig]

Alex pointed out that we might want to use pci_intx_for_msi in
pci_alloc_irq_vectors for the case where allocating MSI or MSI-X vectors
fails completely.  I tried to read up the history behind pci_intx_for_msi,
but the changelogs seem a bit confusing, e.g. the comment in the tg3
around that code that was replaced by this quirk only talk about
_disabling_ INT-X.  Maybe someone can help me to shed some light about
this workaround, and help answering the question wether we need it
in this case.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: David Miller <davem@davemloft.net>
Cc: Shane Huang <shane.huang@amd.com>
---
 drivers/pci/msi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/pci/msi.c b/drivers/pci/msi.c
index 9da5ecb..a4b443f 100644
--- a/drivers/pci/msi.c
+++ b/drivers/pci/msi.c
@@ -1215,7 +1215,7 @@ int pci_alloc_irq_vectors(struct pci_dev *dev, unsigned int min_vecs,
 
 	/* use legacy irq if allowed */
 	if ((flags & PCI_IRQ_LEGACY) && min_vecs == 1) {
-		pci_intx(dev, 1);
+		pci_intx_for_msi(dev, 1);
 		return 1;
 	}
 
-- 
2.1.4

Re: [PATCH, RFC] PCI: use pci_intx_for_msi in pci_alloc_irq_vectors

[Bjorn Helgaas]

Hi Christoph,

On Sun, Sep 11, 2016 at 09:09:27AM +0200, Christoph Hellwig wrote:
> Alex pointed out that we might want to use pci_intx_for_msi in
> pci_alloc_irq_vectors for the case where allocating MSI or MSI-X vectors
> fails completely.  I tried to read up the history behind pci_intx_for_msi,
> but the changelogs seem a bit confusing, e.g. the comment in the tg3
> around that code that was replaced by this quirk only talk about
> _disabling_ INT-X.  Maybe someone can help me to shed some light about
> this workaround, and help answering the question wether we need it
> in this case.

If we want to do this, I'd like to have a changelog that explains
*why*, so I'm ignoring it for now.  I'm not nacking it; I just don't
understand why we should do it yet, and I'm too lazy to do all the
research myself.

> Signed-off-by: Christoph Hellwig <hch@lst.de>
> Cc: David Miller <davem@davemloft.net>
> Cc: Shane Huang <shane.huang@amd.com>
> ---
>  drivers/pci/msi.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/pci/msi.c b/drivers/pci/msi.c
> index 9da5ecb..a4b443f 100644
> --- a/drivers/pci/msi.c
> +++ b/drivers/pci/msi.c
> @@ -1215,7 +1215,7 @@ int pci_alloc_irq_vectors(struct pci_dev *dev, unsigned int min_vecs,
>  
>  	/* use legacy irq if allowed */
>  	if ((flags & PCI_IRQ_LEGACY) && min_vecs == 1) {
> -		pci_intx(dev, 1);
> +		pci_intx_for_msi(dev, 1);
>  		return 1;
>  	}
>  
> -- 
> 2.1.4
> 
> --
> 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

Re: [PATCH, RFC] PCI: use pci_intx_for_msi in pci_alloc_irq_vectors

[Christoph Hellwig]

On Tue, Sep 20, 2016 at 02:46:14PM -0500, Bjorn Helgaas wrote:
> If we want to do this, I'd like to have a changelog that explains
> *why*, so I'm ignoring it for now.  I'm not nacking it; I just don't
> understand why we should do it yet, and I'm too lazy to do all the
> research myself.

Can't blame you.  Alex pointed me to the function, and it seems like a
good idea, but research in git history just doesn't give me an conclusive
reason on why to do it either.