Re: [PATCH v2 3/4] iommu/arm-smmu-v3: Add support for dirty tracking in domain alloc

From: Jason Gunthorpe <jgg@nvidia.com>
To: Shameerali Kolothum Thodi <shameerali.kolothum.thodi@huawei.com>
Cc: Joao Martins <joao.m.martins@oracle.com>,
	"joro@8bytes.org" <joro@8bytes.org>,
	"kevin.tian@intel.com" <kevin.tian@intel.com>,
	"nicolinc@nvidia.com" <nicolinc@nvidia.com>,
	"iommu@lists.linux.dev" <iommu@lists.linux.dev>,
	"mshavit@google.com" <mshavit@google.com>,
	"robin.murphy@arm.com" <robin.murphy@arm.com>,
	"will@kernel.org" <will@kernel.org>,
	jiangkunkun <jiangkunkun@huawei.com>,
	zhukeqian <zhukeqian1@huawei.com>, Linuxarm <linuxarm@huawei.com>,
	"linux-arm-kernel@lists.infradead.org"
	<linux-arm-kernel@lists.infradead.org>
Subject: Re: [PATCH v2 3/4] iommu/arm-smmu-v3: Add support for dirty tracking in domain alloc
Date: Thu, 22 Feb 2024 09:11:01 -0400	[thread overview]
Message-ID: <20240222131101.GZ13330@nvidia.com> (raw)
In-Reply-To: <d6442be07c9d4f999286e073ed221984@huawei.com>

On Thu, Feb 22, 2024 at 12:24:06PM +0000, Shameerali Kolothum Thodi wrote:

> > > Ok. But do we really need to check this in attach()? As dirty_ops are added only
> > > if it is requested in alloc_user() and there we return err when hardware doesn't
> > > have the capability.  So not sure how this matters in attach() path. May be I am
> > > missing something.
> > 
> > That's when you create the domain with dev A. Afterwards that dev A is
> > attached,
> > but later on you can attach another device B to the domain. So this check is
> > there such that a domain with dirty tracking ops set will only have devices in
> > there that support dirty tracking.
> 
> But that only matters if dev B is on  different smmu without dbm capability, right?
> In that case we already fail the attach with,
> 
> >>>  	} else if (smmu_domain->smmu != smmu)
> > >>>  		ret = -EINVAL;
> > >>>
> 
> Is that right?

Right. Relaxing this is on my list :\

The other two drivers don't have this limitation so needed the check
that Joao described.

Jason