| Message ID | 20220623200029.26007-4-nicolinc@nvidia.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | Simplify vfio_iommu_type1 attach/detach routine | expand |
On Thu, Jun 23, 2022 at 01:00:27PM -0700, Nicolin Chen wrote: > The domain->ops validation was added, as a precaution, for mixed-driver > systems. > > Per Robin's remarks, > * While bus_set_iommu() still exists, the core code prevents multiple > drivers from registering, so we can't really run into a situation of > having a mixed-driver system: > https://lore.kernel.org/kvm/6e1280c5-4b22-ebb3-3912-6c72bc169982@arm.com/ > > * And there's plenty more significant problems than this to fix; in future > when many can be permitted, we will rely on the IOMMU core code to check > the domain->ops: > https://lore.kernel.org/kvm/6575de6d-94ba-c427-5b1e-967750ddff23@arm.com/ > > So remove the check in VFIO for simplicity. As discussed we can't remove this check, multiple iommus on different busses are allowed today and VFIO does not restrict its attempts to attach a pre-existing domain to a single group or bus. Please go back to the original version with the ops check in the core code. Jason
On Fri, Jun 24, 2022 at 03:28:33PM -0300, Jason Gunthorpe wrote: > On Thu, Jun 23, 2022 at 01:00:27PM -0700, Nicolin Chen wrote: > > The domain->ops validation was added, as a precaution, for mixed-driver > > systems. > > > > Per Robin's remarks, > > * While bus_set_iommu() still exists, the core code prevents multiple > > drivers from registering, so we can't really run into a situation of > > having a mixed-driver system: > > https://lore.kernel.org/kvm/6e1280c5-4b22-ebb3-3912-6c72bc169982@arm.com/ > > > > * And there's plenty more significant problems than this to fix; in future > > when many can be permitted, we will rely on the IOMMU core code to check > > the domain->ops: > > https://lore.kernel.org/kvm/6575de6d-94ba-c427-5b1e-967750ddff23@arm.com/ > > > > So remove the check in VFIO for simplicity. > > As discussed we can't remove this check, multiple iommus on different > busses are allowed today and VFIO does not restrict its attempts to > attach a pre-existing domain to a single group or bus. > > Please go back to the original version with the ops check in the core > code. Robin convinced me, so never mind :) Jason
diff --git a/drivers/vfio/vfio_iommu_type1.c b/drivers/vfio/vfio_iommu_type1.c index f4e3b423a453..11be5f95580b 100644 --- a/drivers/vfio/vfio_iommu_type1.c +++ b/drivers/vfio/vfio_iommu_type1.c @@ -2277,29 +2277,19 @@ static int vfio_iommu_type1_attach_group(void *iommu_data, domain->domain->ops->enforce_cache_coherency( domain->domain); - /* - * Try to match an existing compatible domain. We don't want to - * preclude an IOMMU driver supporting multiple bus_types and being - * able to include different bus_types in the same IOMMU domain, so - * we test whether the domains use the same iommu_ops rather than - * testing if they're on the same bus_type. - */ + /* Try to match an existing compatible domain */ list_for_each_entry(d, &iommu->domain_list, next) { - if (d->domain->ops == domain->domain->ops) { - iommu_detach_group(domain->domain, group->iommu_group); - if (!iommu_attach_group(d->domain, - group->iommu_group)) { - list_add(&group->next, &d->group_list); - iommu_domain_free(domain->domain); - kfree(domain); - goto done; - } - - ret = iommu_attach_group(domain->domain, - group->iommu_group); - if (ret) - goto out_domain; + iommu_detach_group(domain->domain, group->iommu_group); + if (!iommu_attach_group(d->domain, group->iommu_group)) { + list_add(&group->next, &d->group_list); + iommu_domain_free(domain->domain); + kfree(domain); + goto done; } + + ret = iommu_attach_group(domain->domain, group->iommu_group); + if (ret) + goto out_domain; } vfio_test_domain_fgsp(domain);