| Message ID | 20230302005959.2695267-3-jacob.jun.pan@linux.intel.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | Re-enable IDXD kernel workqueue under DMA API | expand |
On 2023/3/2 8:59, Jacob Pan wrote: > For in-kernel DMA, use non-privileged access for all PASIDs to be > consistent with RID_PASID. > There's no need to differentiate user and kernel for in-kernel DMA. > > Signed-off-by: Jacob Pan <jacob.jun.pan@linux.intel.com> > --- > drivers/iommu/intel/iommu.c | 2 -- > 1 file changed, 2 deletions(-) > > diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c > index a0cb3bc851ac..9e3c056e392d 100644 > --- a/drivers/iommu/intel/iommu.c > +++ b/drivers/iommu/intel/iommu.c > @@ -2334,8 +2334,6 @@ static int domain_setup_first_level(struct intel_iommu *iommu, > if (level != 4 && level != 5) > return -EINVAL; > > - if (pasid != PASID_RID2PASID) > - flags |= PASID_FLAG_SUPERVISOR_MODE; With above removed, PASID_FLAG_SUPERVISOR_MODE is not used anywhere? Perhaps you can cleanup it to avoid dead code? > if (level == 5) > flags |= PASID_FLAG_FL5LP; > Best regards, baolu
Hi Baolu, On Thu, 2 Mar 2023 22:11:36 +0800, Baolu Lu <baolu.lu@linux.intel.com> wrote: > On 2023/3/2 8:59, Jacob Pan wrote: > > For in-kernel DMA, use non-privileged access for all PASIDs to be > > consistent with RID_PASID. > > There's no need to differentiate user and kernel for in-kernel DMA. > > > Signed-off-by: Jacob Pan <jacob.jun.pan@linux.intel.com> > > --- > > drivers/iommu/intel/iommu.c | 2 -- > > 1 file changed, 2 deletions(-) > > > > diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c > > index a0cb3bc851ac..9e3c056e392d 100644 > > --- a/drivers/iommu/intel/iommu.c > > +++ b/drivers/iommu/intel/iommu.c > > @@ -2334,8 +2334,6 @@ static int domain_setup_first_level(struct > > intel_iommu *iommu, if (level != 4 && level != 5) > > return -EINVAL; > > > > - if (pasid != PASID_RID2PASID) > > - flags |= PASID_FLAG_SUPERVISOR_MODE; > > With above removed, PASID_FLAG_SUPERVISOR_MODE is not used anywhere? > Perhaps you can cleanup it to avoid dead code? good point, we could remove pasid_set_sre() related code for FL,SL, and PT. Thanks, Jacob
diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c index a0cb3bc851ac..9e3c056e392d 100644 --- a/drivers/iommu/intel/iommu.c +++ b/drivers/iommu/intel/iommu.c @@ -2334,8 +2334,6 @@ static int domain_setup_first_level(struct intel_iommu *iommu, if (level != 4 && level != 5) return -EINVAL; - if (pasid != PASID_RID2PASID) - flags |= PASID_FLAG_SUPERVISOR_MODE; if (level == 5) flags |= PASID_FLAG_FL5LP;
For in-kernel DMA, use non-privileged access for all PASIDs to be consistent with RID_PASID. There's no need to differentiate user and kernel for in-kernel DMA. Signed-off-by: Jacob Pan <jacob.jun.pan@linux.intel.com> --- drivers/iommu/intel/iommu.c | 2 -- 1 file changed, 2 deletions(-)