linux-kernel - Re: [PATCH 2/2] iommu: dma: Use of_iommu_get_resv

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]

Date:   Wed, 27 Nov 2019 17:09:41 +0000
From:   Robin Murphy <robin.murphy@....com>
To:     Thierry Reding <thierry.reding@...il.com>,
        Will Deacon <will@...nel.org>
Cc:     Joerg Roedel <joro@...tes.org>, Rob Herring <robh+dt@...nel.org>,
        Frank Rowand <frowand.list@...il.com>,
        iommu@...ts.linux-foundation.org, devicetree@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2/2] iommu: dma: Use of_iommu_get_resv_regions()

On 27/11/2019 2:16 pm, Thierry Reding wrote:
[...]
> Nevermind that, I figured out that I was missingthe initialization of
> some of the S2CR variables. I've got something that I think is working
> now, though I don't know yet how to go about cleaning up the initial
> mapping and "recycling" it.
> 
> I'll clean things up a bit, run some more tests and post a new patch
> that can serve as a basis for discussion.

I'm a little puzzled by the smmu->identity domain - disregarding the 
fact that it's not actually used by the given diff ;) - if isolation is 
the reason for not simply using a bypass S2CR for the window between 
reset and the relevant .add_device call where the default domain proper 
comes in[1], then surely exposing the union of memory regions to the 
union of all associated devices isn't all that desirable either.

Either way, I'll give you the pre-emptive warning that this is the SMMU 
in the way of my EFI framebuffer ;)

...
arm-smmu 7fb20000.iommu: 	1 context banks (1 stage-2 only)
...

Robin.

[1] the fact that it currently depends on probe order whether getting 
that .add_device call requires a driver probing for the device is an 
error as discussed elsewhere, and will get fixed separately, I promise.