Re: [Xen-devel] [PATCH v3 2/2] VT-d: Fix vt-d flush timeout issue.

["Xu, Quan" <quan.xu@xxxxxxxxx>]

> On 15.12.2015 at 9:44pm, <JBeulich@xxxxxxxx> wrote:
> >>> On 15.12.15 at 14:31, <quan.xu@xxxxxxxxx> wrote:
> > Copy from pci_hide_device(), which is actually add device to dom_xen
> > and add pdev->domain_list to dom_xen->arch.pdev_list.
> >
> > Quite similar, a second one only with proper justification, I can
> > reassign Device form _domain to dom_xen directly. The below is the how
> > to deal With device_tlb ( is it acceptable? ). It is working to hide device.
> 
> Looks reasonable, but ...
> 
> > +void device_tlb_invalidate_timeout(struct iommu *iommu, u16 did,
> > +                                   u16 seg, u8 bus, u8 devfn) {
> > +    struct domain *d;
> > +    struct pci_dev *pdev;
> > +    struct hvm_iommu *hd;
> > +    int rc;
> > +
> > +    d = rcu_lock_domain_by_id(iommu->domid_map[did]);
> > +    ASSERT(d);
> > +    for_each_pdev(d, pdev)
> > +        if ( (pdev->seg == seg) &&
> > +             (pdev->bus == bus) &&
> > +             (pdev->devfn == devfn) )
> > +        {
> > +            if ( pdev->domain )
> > +            {
> > +                hd = domain_hvm_iommu(d);
> > +                rc =  hd->platform_ops->reassign_device(d,
> > +                      dom_xen, devfn, pdev);
> 
> ... doesn't this have the potential of generating further flushes? You 
> clearly need
> to be certain not to recurse here.

Good consideration. I should continue to enhance it.
Thanks Jan...

Quan



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel