Re: [PATCH v2] iommu: iommu_group_claim_dma_owner() must always assign a domain

From: Robin Murphy <robin.murphy@arm.com>
To: Jason Gunthorpe <jgg@nvidia.com>
Cc: Will Deacon <will@kernel.org>,
	"Tian, Kevin" <kevin.tian@intel.com>,
	"Rodel, Jorg" <jroedel@suse.de>,
	Qian Cai <quic_qiancai@quicinc.com>,
	"iommu@lists.linux-foundation.org"
	<iommu@lists.linux-foundation.org>
Subject: Re: [PATCH v2] iommu: iommu_group_claim_dma_owner() must always assign a domain
Date: Mon, 9 May 2022 10:59:11 +0100	[thread overview]
Message-ID: <0f838b34-460c-ed83-7b98-6cda444b10c2@arm.com> (raw)
In-Reply-To: <20220506135431.GC49344@nvidia.com>

On 2022-05-06 14:54, Jason Gunthorpe wrote:
> On Fri, May 06, 2022 at 02:35:50PM +0100, Robin Murphy wrote:
> 
>>> So you want to say "DMA is always managed" when attaching a domain of
>>> type IOMMU_DOMAIN_UNMANAGED? :)
>>
>> Touché ;) Just goes to confirm the point above that confusion between
>> general concepts and specific API terms is all too easy. An "unmanaged"
>> domain from the PoV of the API just means it's managed by the external
>> caller, but you're right that that's not necessarily so obvious either.
> 
> Yeah, I'm not so keen on the naming used for IOMMU_DOMAIN_*
> 
> I looked for a bit and could not figure out why we need to have
> IOMMU_DOMAIN_DMA either.. I didn't find anthing obvious in the iommu
> drivers that looked like a special case for this? Most drivers treat
> it identically to UNMANAGED in their alloc functions
> 
> Only mtk, arm-smmu and some odd stuff in Intel seemed to be sensitive?

Yes, that's a relatively recent change[1] - prior to that, drivers did 
still have to take (minimal) action to opt into iommu-dma support. 
IOMMU_DOMAIN_DMA still needs to exist as a type for the core code to 
differentiate internally, so driver involvement is mostly now just a 
case of saying "yeah OK fine" if they see it. IOMMU_DOMAIN_DMA_FQ now 
effectively takes over the original semantics of IOMMU_DOMAIN_DMA as the 
one that depends on driver-specific functionality.

FWIW, mtk-iommu doesn't really have any need to reject 
IOMMU_DOMAIN_UNMANAGED, they just don't have any relevant client drivers 
that want it; however arm-smmu definitely does need to continue 
rejecting IOMMU_DOMAIN_DMA when it can't rely on the DT code ensuring 
the correct probe ordering with the old DT binding (otherwise client 
drivers are liable to get broken DMA ops).

Thanks,
Robin.

[1] 
https://lore.kernel.org/linux-iommu/cover.1628682048.git.robin.murphy@arm.com/
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu