linux-kernel - Re: [PATCHv4 1/6] iommu/io-pgtable-arm: Add support to use system cache

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: <20200921222329.GA4409@willie-the-truck>
Date:   Mon, 21 Sep 2020 23:23:30 +0100
From:   Will Deacon <will@...nel.org>
To:     Robin Murphy <robin.murphy@....com>
Cc:     Sai Prakash Ranjan <saiprakash.ranjan@...eaurora.org>,
        freedreno@...ts.freedesktop.org, linux-arm-msm@...r.kernel.org,
        iommu@...ts.linux-foundation.org, linux-kernel@...r.kernel.org,
        Akhil P Oommen <akhilpo@...eaurora.org>,
        dri-devel@...ts.freedesktop.org,
        "Kristian H . Kristensen" <hoegsberg@...gle.com>,
        linux-arm-kernel@...ts.infradead.org
Subject: Re: [PATCHv4 1/6] iommu/io-pgtable-arm: Add support to use system
 cache

On Mon, Sep 21, 2020 at 11:03:49PM +0100, Robin Murphy wrote:
> On 2020-09-21 19:03, Will Deacon wrote:
> > On Fri, Sep 11, 2020 at 07:57:18PM +0530, Sai Prakash Ranjan wrote:
> > > Add a quirk IO_PGTABLE_QUIRK_SYS_CACHE to override the
> > > attributes set in TCR for the page table walker when
> > > using system cache.
> > 
> > I wonder if the panfrost folks can reuse this for the issue discussed
> > over at:
> > 
> > https://lore.kernel.org/r/cover.1600213517.git.robin.murphy@arm.com
> 
> Isn't this all hinged around the outer cacheability attribute, rather than
> shareability (since these are nominally NC mappings and thus already
> properly Osh)? The Panfrost issue is just about shareability domains being a
> bit wonky; the cacheability attributes there are actually reasonably normal
> (other than not having a non-cacheable type at all, only a choice of
> allocation policies...)

Hmm, yes, this quirk _also_ changes the cacheability settings which isn't
what we need. It's a bit grotty having two different ways to configure these
TCR bits (i.e. a quirk and a format), but at least the mali format rejects
all of the quirks so I suppose it's not the end of the world.

Will