Messages in this thread | | | Date | Tue, 5 Sep 2023 10:32:58 -0300 | From | Jason Gunthorpe <> | Subject | Re: [PATCH v1 0/3] Clean-up arm-smmu-v3-sva.c: remove arm_smmu_bond |
| |
On Tue, Sep 05, 2023 at 09:24:22PM +0800, Michael Shavit wrote: > On Tue, Sep 5, 2023 at 8:35 PM Jason Gunthorpe <jgg@nvidia.com> wrote: > > > > On Tue, Sep 05, 2023 at 07:49:11PM +0800, Michael Shavit wrote: > > > > > > Note that arm-smmu-v3-sva performs a second level of normalization by > > > mapping multiple bonds (now SVA domains) attached to devices with the > > > same SMMU (if those devices have the same RID domain attached) to a > > > single arm_smmu_mmu_notifier. This is not affected by these patches. > > > > Ultimately the notifier should be per-iommu_domain as well. > > Speaking of, I'm questioning whether the multi-SMMU domain patchseries > and Tina's sva domain sharing are really prerequisites to get rid of > the notifier sharing. Is anyone really depending on or taking > advantage of this?
Currently I don't see any in-tree user of SVA on ARM except for uacce, which seems to work with only one device (HiSillicon QM).
So, yes, you can probably make it slightly less efficient for a time. It will be functionally correct still with multiple registered notifiers.
Jason
| |