linux-kernel - Re: [PATCH] iommu/io-pgtable: Make IOMMU_IO_PGTABLE

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:   Tue, 27 Sep 2022 16:50:03 +0100
From:   Robin Murphy <robin.murphy@....com>
To:     Geert Uytterhoeven <geert@...ux-m68k.org>
Cc:     Joerg Roedel <joro@...tes.org>, Will Deacon <will@...nel.org>,
        Sven Peter <sven@...npeter.dev>, Janne Grunau <j@...nau.net>,
        Hector Martin <marcan@...can.st>, iommu@...ts.linux.dev,
        linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] iommu/io-pgtable: Make IOMMU_IO_PGTABLE_DART invisible

On 2022-09-27 16:29, Geert Uytterhoeven wrote:
> Hi Robin,
> 
> On Tue, Sep 27, 2022 at 5:09 PM Robin Murphy <robin.murphy@....com> wrote:
>> On 2022-09-27 15:48, Geert Uytterhoeven wrote:
>>> On Tue, Sep 27, 2022 at 4:15 PM Robin Murphy <robin.murphy@....com> wrote:
>>>> On 2022-09-27 14:36, Geert Uytterhoeven wrote:
>>>>> There is no point in asking the user about both "Apple DART Formats" and
>>>>> "Apple DART IOMMU Support", as the former is useless without the latter,
>>>>> and the latter auto-selects the former.
>>>>>
>>>>> Fixes: 745ef1092bcfcf3b ("iommu/io-pgtable: Move Apple DART support to its own file")
>>>>> Signed-off-by: Geert Uytterhoeven <geert+renesas@...der.be>
>>>>> ---
>>>>> Should IOMMU_IO_PGTABLE_LPAE and IOMMU_IO_PGTABLE_ARMV7S be made
>>>>> invisible, too?
>>>>> Are there users that do not select them?
>>>>
>>>> The aim was for formats to be independently selectable for COMPILE_TEST
>>>> coverage. The Arm formats are manually selectable for the sake of their
>>>> runtime self-tests, which are self-contained, but since DART format
>>>> doesn't do anything by itself I'd agree there's no need to prompt when
>>>> !COMPILE_TEST here.
>>>
>>> IOMMU_IO_PGTABLE_LPAE and IOMMU_IO_PGTABLE_ARMV7S are
>>> selected by other symbols that can be enabled when compile-testing, so
>>> the tests can still be enabled in those cases, too
>>
>> Sure, but when you want to compile-test a thing, what would you rather
>> do: enable the thing, or go hunting to find some other thing that
>> happens to select the thing you actually want, then potentially have to
>> figure out *that* thing's dependencies, and so on?
> 
> Agreed.
> 
>> Coverage isn't solely about whether it's technically possible to ever
>> reach somewhere at all, it's just as much about how easily and/or often
>> you can get there in practice. I don't see who benefits from making
>> COMPILE_TEST harder to use :/
> 
> So perhaps the visibility of IOMMU_IO_PGTABLE_LPAE and
> IOMMU_IO_PGTABLE_ARMV7S should depend on COMPILE_TEST?
> Normal users would still get it through select when needed.

As I say those still offer functionality beyond compile-testing, but now 
you've got me suspecting that it's already suboptimal that one has to 
enable the format to make the self-test option appear... Perhaps what we 
want is a separate master option to enable io-pgtable self-tests in 
general, then rejig the rest around that.

Of course the self-tests would be even more useful if the harness was at 
the level of the core io-pgtable API so it could cover new formats 
automatically as long as they provide the configuration parameters, but 
that's a separate matter for someone with sufficient free time and 
enthusiasm :)

Cheers,
Robin.