[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <fb1005b6-e610-407f-39bb-7cd6a198f9d6@somainline.org>
Date: Fri, 10 Dec 2021 17:28:53 +0100
From: AngeloGioacchino Del Regno
<angelogioacchino.delregno@...ainline.org>
To: Marijn Suijten <marijn.suijten@...ainline.org>,
Alex Elder <elder@...aro.org>, phone-devel@...r.kernel.org,
Andy Gross <agross@...nel.org>,
Bjorn Andersson <bjorn.andersson@...aro.org>,
~postmarketos/upstreaming@...ts.sr.ht,
Konrad Dybcio <konrad.dybcio@...ainline.org>,
Martin Botka <martin.botka@...ainline.org>,
Jami Kettunen <jami.kettunen@...ainline.org>,
Pavel Dubrova <pashadubrova@...il.com>,
Kalle Valo <kvalo@...eaurora.org>,
Arnd Bergmann <arnd@...db.de>,
Daniel Lezcano <daniel.lezcano@...aro.org>,
Thara Gopinath <thara.gopinath@...aro.org>,
Elliot Berman <eberman@...eaurora.org>,
linux-arm-msm@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2/3] firmware: qcom: scm: Add function to set the maximum
IOMMU pool size
Il 09/12/21 00:44, Marijn Suijten ha scritto:
> On 2021-12-08 07:30:28, Alex Elder wrote:
>> On 12/8/21 2:34 AM, Marijn Suijten wrote:
>>> From: AngeloGioacchino Del Regno <angelogioacchino.delregno@...ainline.org>
>>>
>>> This is not necessary for basic functionality of the IOMMU, but
>>> it's an optimization that tells to the TZ what's the maximum
>>> mappable size for the secure IOMMUs, so that it can optimize
>>> the data structures in the TZ itself.
>>
>> Are there no users of this function? -Alex
>
> I should have probably mentioned in the cover letter that this function
> and the one introduced in patch 3/3 are going to be used in upcoming
> patches for IOMMUs found in msm8976, msm8974 and related SoCs (with the
> side-note that I don't see this particular set_cp_pool_size used in the
> branch that this was submitted from, but it's most likely used elsewhere
> or planned ahead to be used in the near future - I expect Angelo to be
> able to comment on that more accurately).
>
This function is used in the secured iommu pagetable setup, but not for
all of the "SCM feature versions" (only for version >= 1.1.1, downstream
reads it with a call to scm_feat_version()).
It's not strictly necessary for functionality, hence why Marijn isn't
seeing any call to this in the branch that he was browsing: the spirit here
is to first introduce code that does a minimal (but, of course, working)
setup of the IOMMUs found in MSM8956/76 (which can be adapted with very
minimal changes to other SoCs), and *only then* to add these performance
enhancements to the mix.
...and this is why this commit is here :))
By the way, if my memory isn't failing me, that SCM call should be usable
on all AArch64 SoCs (so, 8956, 8994 and the rest).
Powered by blists - more mailing lists