Re: [PATCH v6 1/4] iommu: Generalize default PCIe requester ID PASID - Jean-Philippe Brucker

From: Jean-Philippe Brucker <jean-philippe@linaro.org>
To: Jacob Pan <jacob.jun.pan@linux.intel.com>
Cc: LKML <linux-kernel@vger.kernel.org>,
	iommu@lists.linux.dev, Jason Gunthorpe <jgg@nvidia.com>,
	Lu Baolu <baolu.lu@linux.intel.com>,
	Joerg Roedel <joro@8bytes.org>,
	dmaengine@vger.kernel.org, vkoul@kernel.org,
	Robin Murphy <robin.murphy@arm.com>,
	Will Deacon <will@kernel.org>,
	David Woodhouse <dwmw2@infradead.org>,
	Raj Ashok <ashok.raj@intel.com>,
	"Tian, Kevin" <kevin.tian@intel.com>, Yi Liu <yi.l.liu@intel.com>,
	"Yu, Fenghua" <fenghua.yu@intel.com>,
	Dave Jiang <dave.jiang@intel.com>,
	Tony Luck <tony.luck@intel.com>,
	"Zanussi, Tom" <tom.zanussi@intel.com>,
	narayan.ranganathan@intel.com
Subject: Re: [PATCH v6 1/4] iommu: Generalize default PCIe requester ID PASID
Date: Tue, 23 May 2023 15:47:33 +0100	[thread overview]
Message-ID: <20230523144733.GA4137946@myrica> (raw)
In-Reply-To: <20230519203223.2777255-2-jacob.jun.pan@linux.intel.com>

Hi Jacob,

On Fri, May 19, 2023 at 01:32:20PM -0700, Jacob Pan wrote:
> PCIe Process address space ID (PASID) is used to tag DMA traffic, it
> provides finer grained isolation than requester ID (RID).
> 
> For each RID, 0 is as a special PASID for the legacy DMA (without
> PASID), thus RID_PASID. This is universal across all architectures,
> therefore warranted to be declared in the common header.
> Noting that VT-d could support none-zero RID_PASID, but currently not
> used.
> 
> By having a common RID_PASID, we can avoid conflicts between different
> use cases in the generic code. e.g. SVA and DMA API with PASIDs.
> 
> Signed-off-by: Jacob Pan <jacob.jun.pan@linux.intel.com>
> ---
> v6:
>    - let SMMU code use the common RID_PASID macro
> ---
>  .../iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c   |  2 +-
>  drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c   | 10 ++++----
>  drivers/iommu/intel/iommu.c                   | 24 +++++++++----------
>  drivers/iommu/intel/pasid.c                   |  2 +-
>  drivers/iommu/intel/pasid.h                   |  1 -
>  include/linux/iommu.h                         |  1 +
>  6 files changed, 20 insertions(+), 20 deletions(-)
> 
> diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c
> index a5a63b1c947e..160b31e6239d 100644
> --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c
> +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c
> @@ -80,7 +80,7 @@ arm_smmu_share_asid(struct mm_struct *mm, u16 asid)
>  	 * be some overlap between use of both ASIDs, until we invalidate the
>  	 * TLB.
>  	 */
> -	arm_smmu_write_ctx_desc(smmu_domain, 0, cd);
> +	arm_smmu_write_ctx_desc(smmu_domain, IOMMU_DEF_RID_PASID, cd);

I agree with reserving 0 globally for non-PASID DMA, but could we call
this something more generic, like IOMMU_NO_PASID?  The term "RID_PASID" is
specific to VT-d and "RID" to PCI, so it looks confusing here (this driver
also supports non-PCI). "NO_PASID" would be clearer to someone just trying
to follow this driver code.

Thanks,
Jean