linux-kernel - Re: [PATCH v2 05/11] clk: qcom: gcc-msm8998: Mark gpu_cfg_ahb

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:   Fri, 15 Jan 2021 00:05:45 +0100
From:   AngeloGioacchino Del Regno 
        <angelogioacchino.delregno@...ainline.org>
To:     Jeffrey Hugo <jeffrey.l.hugo@...il.com>
Cc:     MSM <linux-arm-msm@...r.kernel.org>, konrad.dybcio@...ainline.org,
        marijn.suijten@...ainline.org, martin.botka@...ainline.org,
        phone-devel@...r.kernel.org, lkml <linux-kernel@...r.kernel.org>,
        Andy Gross <agross@...nel.org>,
        Bjorn Andersson <bjorn.andersson@...aro.org>,
        Michael Turquette <mturquette@...libre.com>,
        Stephen Boyd <sboyd@...nel.org>,
        Rob Herring <robh+dt@...nel.org>, linux-clk@...r.kernel.org,
        DTML <devicetree@...r.kernel.org>
Subject: Re: [PATCH v2 05/11] clk: qcom: gcc-msm8998: Mark gpu_cfg_ahb_clk as
 critical

Il 14/01/21 23:37, Jeffrey Hugo ha scritto:
> On Thu, Jan 14, 2021 at 3:13 PM AngeloGioacchino Del Regno
> <angelogioacchino.delregno@...ainline.org> wrote:
>>
>> The GPU IOMMU depends on this clock and the hypervisor will crash
>> the SoC if this clock gets disabled because the secure contexts
>> that have been set on this IOMMU by the bootloader will become
>> unaccessible (or they get reset).
>> Mark this clock as critical to avoid this issue when the Adreno
>> GPU is enabled.
>>
> 
> You should go review the last attempt to do this -
> https://lkml.org/lkml/2019/12/17/881
> 

Thanks for the tip, but unfortunately this isn't possible on the 
gpu_cfg_ahb_clk, as it is also needed for the Adreno IOMMU, which has 
secure contexts that are set up from one of the bootloader stages and if 
you reset/"mess up" one of them (by - in this case - un-clocking the 
MMU), then the hypervisor will kick in and generate a fault, rebooting 
the SoC.

Of course, this scenario is for the case in which you want to boot the 
device without any gpucc nor any runtime pm user of that.. and the 
aforementioned issue makes that solution not really usable.

Again, unfortunately.