linux-kernel - Re: [PATCH 1/1] iommu/arm-smmu: Implement qcom,skip-init

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: <20200721154415.GA5758@jcrouse1-lnx.qualcomm.com>
Date:   Tue, 21 Jul 2020 09:44:15 -0600
From:   Jordan Crouse <jcrouse@...eaurora.org>
To:     Konrad Dybcio <konradybcio@...il.com>
Cc:     Bjorn Andersson <bjorn.andersson@...aro.org>,
        Will Deacon <will@...nel.org>, skrzynka@...radybcio.pl,
        Robin Murphy <robin.murphy@....com>,
        Joerg Roedel <joro@...tes.org>,
        Rob Herring <robh+dt@...nel.org>,
        linux-arm-kernel@...ts.infradead.org,
        iommu@...ts.linux-foundation.org,
        DTML <devicetree@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        John Stultz <john.stultz@...aro.org>
Subject: Re: [PATCH 1/1] iommu/arm-smmu: Implement qcom,skip-init

On Tue, Jul 21, 2020 at 05:04:11PM +0200, Konrad Dybcio wrote:
> So.. is this a no-no?
> 
> I of course would like to omit this entirely, but SMMUs on sdm630 and
> friends are REALLY picky.. What seems to happen is that when the
> driver tries to do things the "standard" way, hypervisor decides to
> hang the platform or force a reboot. Not very usable.
> 
> 
> This thing is needed for the platform to even boot properly and one
> more [1] is required to make mdss work with video mode panels (the
> fact that CMD-mode panels work is kinda hilarious to me).
> 
> To be honest, there are even more qcom quirks (of which at least
> qcom,dynamic and qcom-use-3-lvl-tables are used on 630).. [2]
> 
> Looking forward to your answers and possibly better solutions.

Nobody is disputing that the qcom SMMUs don't have their share of quirks but it
seems that the community has mostly settled on the agreement that there are
better ways to solve this than a handful of device tree properties. The current
focus has been on moving more of the SMMU specific bits into the arm-smmu-qcom
implementation [1] and I think that is the right way to go.

As for the other quirks we can probably discuss those on a case by case basis.
I doubt you will find much enthusiasm for qcom,use-3-lvl-tables and I've been
working on replacing qcom,dynamic with something much better [2].

[1] https://lists.linuxfoundation.org/pipermail/iommu/2020-July/046304.html
[2] https://lists.linuxfoundation.org/pipermail/iommu/2020-July/046756.html

Jordan

> [1] https://github.com/konradybcio/linux/commit/83ac38af259968f92b6a8b7eab90096c78469f87
> [2] https://github.com/sonyxperiadev/kernel/blob/aosp/LA.UM.7.1.r1/drivers/iommu/arm-smmu.c#L404-L415
> 
> Regards
> Konrad

-- 
The Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project