On Mon, Nov 14, 2016 at 03:33:22PM -0500, Peter Xu wrote:
> On Thu, Nov 10, 2016 at 08:24:19PM +0100, Andrew Jones wrote:
>
> [...]
>
> > > +void pci_scan_bars(struct pci_dev *dev)
> > > +{
> > > + int i = 0;
> > > +
> > > + for (i = 0; i < PCI_BAR_NUM; i++) {
> > > + if (!pci_bar_is_valid(dev, i))
> > > + continue;
> > > + dev->bar[i] = pci_bar_get_addr(dev, i);
> >
> > What happens when you get_addr a 64-bit bar in the middle?
> > Shouldn't we skip that?
>
> Hmm yes... Do you like this?
>
> ---------8<----------
>
> diff --git a/lib/pci.c b/lib/pci.c
> index 0593699..2a58b30 100644
> --- a/lib/pci.c
> +++ b/lib/pci.c
> @@ -230,11 +230,15 @@ void pci_print(void)
>
> void pci_scan_bars(struct pci_dev *dev)
> {
> - int i = 0;
> + int i;
> +
> + memset(&dev->bar[0], 0, sizeof(dev->bar));
I'm not sure what this memset is for. Anyway, if you need to zero a
bar, why not just
bar = (phys_addr_t)0;
>
> for (i = 0; i < PCI_BAR_NUM; i++) {
> if (!pci_bar_is_valid(dev, i))
> continue;
> dev->bar[i] = pci_bar_get_addr(dev, i);
> + if (pci_bar_is64(dev, i))
> + i++;
> }
yes, this is consistent with how Alex did it in pci_probe. So
looks good to me.
> }
>
> --------->8----------
>
> pci_bar_is64() is called twice per bar, but I think it's okay here
> since it's during init and the whole scan is called only once for 6
> bars.
yeah, the "inefficiency" is no biggy.
Thanks,
drew
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
