linux-kernel - Re: [PATCH 07/11] iommu/arm-smmu: Make use of the iommu

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <35cd4154-df22-bd8c-6c0b-0a074ab0c179@arm.com>
Date:   Fri, 10 Feb 2017 17:07:47 +0000
From:   Robin Murphy <robin.murphy@....com>
To:     Joerg Roedel <joro@...tes.org>
Cc:     Will Deacon <will.deacon@....com>,
        Lorenzo Pieralisi <lorenzo.pieralisi@....com>,
        Alex Williamson <alex.williamson@...hat.com>,
        David Woodhouse <dwmw2@...radead.org>,
        iommu@...ts.linux-foundation.org, linux-kernel@...r.kernel.org,
        Joerg Roedel <jroedel@...e.de>
Subject: Re: [PATCH 07/11] iommu/arm-smmu: Make use of the iommu_register
 interface

On 10/02/17 15:25, Joerg Roedel wrote:
> On Fri, Feb 10, 2017 at 02:20:34PM +0000, Robin Murphy wrote:
>>> @@ -1795,8 +1798,10 @@ static int arm_smmu_add_device(struct device *dev)
>>>  	}
>>>  
>>>  	group = iommu_group_get_for_dev(dev);
>>> -	if (!IS_ERR(group))
>>> +	if (!IS_ERR(group)) {
>>>  		iommu_group_put(group);
>>> +		iommu_device_link(&smmu->iommu, dev);
>>
>> Given the coupling evident from this and the other patches, might it
>> work to simply do the linking/unlinking automatically in
>> iommu_group_{add,remove}_device()?
> 
> Yes, this is one of the goals too. But currently we don't have a generic
> device->hw_iommu mapping in the iommu-code which would allow to call
> the link/unlink functions in generic code too.

At some point we should change the iommu_ops pointer in iommu_fwspec for
an iommu_device pointer - that would then give us an easy
dev->fwpec->hw_iommu relationship which is mostly managed by core code
already. In the meantime I was imagining just passing it around,
something like iommu_group_add_device(hw_iommu, group, dev), but now I
suspect that'd be running up against a similar objection to before ;)

Robin.

> 
> But changing this is one of the next things on my list :)
> 
> 
> 
> 	Joerg
>