Re: [PATCH v11 1/8] PCI: initialize and release SR-IOV capability

[Yu Zhao <yu.zhao@intel.com>]

On Fri, Mar 20, 2009 at 03:53:12AM +0800, Matthew Wilcox wrote:
> On Wed, Mar 11, 2009 at 03:25:42PM +0800, Yu Zhao wrote:
> > +config PCI_IOV
> > +	bool "PCI IOV support"
> > +	depends on PCI
> > +	help
> > +	  PCI-SIG I/O Virtualization (IOV) Specifications support.
> > +	  Single Root IOV: allows the creation of virtual PCI devices
> > +	  that share the physical resources from a real device.
> > +
> > +	  When in doubt, say N.
> 
> It's certainly shorter than my text, which is nice.  But I think it
> still has too much spec-ese and not enough explanation.  How about:
> 
> 	help
> 	  I/O Virtualization is a PCI feature supported by some devices
> 	  which allows them to create virtual devices which share their
> 	  physical resources.
> 
> 	  If unsure, say N.

Yes, it's more user-friendly.

> > +	list_for_each_entry(pdev, &dev->bus->devices, bus_list)
> > +		if (pdev->is_physfn)
> > +			break;
> > +	if (list_empty(&dev->bus->devices) || !pdev->is_physfn)
> > +		pdev = NULL;
> 
> This is still wrong.  If the 'break' condition is not hit, pdev is
> pointing to garbage, not to the last pci_dev in the list.

Yes, you are right. I should think it over after you commented on it
last time.

So it looks like we need to make it as:

	ctrl = 0;
	list_for_each_entry(pdev, &dev->bus->devices, bus_list)
		if (pdev->is_physfn)
			goto found;

	pdev = NULL;
	if (pci_ari_enabled(dev->bus))
		ctrl |= PCI_SRIOV_CTRL_ARI;

found:
	pci_write_config_word(dev, pos + PCI_SRIOV_CTRL, ctrl);
	...

> > @@ -270,6 +278,7 @@ struct pci_dev {
> >  	struct list_head msi_list;
> >  #endif
> >  	struct pci_vpd *vpd;
> > +	struct pci_sriov *sriov;	/* SR-IOV capability related */
> 
> Should be ifdeffed?

Yes, will do.


Thank you for reviewing it. The patch series was applied on Xen Domain0
tree 2 days ago, and I'll carry your comments back to Xen tree too.