linux-kernel - Re: [PATCH 1/7] arm-nommu: remove the partial DMA_ATTR_NON

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: <20190625061332.GC28986@lst.de>
Date:   Tue, 25 Jun 2019 08:13:32 +0200
From:   Christoph Hellwig <hch@....de>
To:     Vladimir Murzin <vladimir.murzin@....com>
Cc:     Christoph Hellwig <hch@....de>, Vineet Gupta <vgupta@...opsys.com>,
        Jonas Bonn <jonas@...thpole.se>,
        Stefan Kristiansson <stefan.kristiansson@...nalahti.fi>,
        Stafford Horne <shorne@...il.com>,
        Helge Deller <deller@....de>,
        linux-snps-arc@...ts.infradead.org,
        linux-arm-kernel@...ts.infradead.org,
        openrisc@...ts.librecores.org, linux-parisc@...r.kernel.org,
        linux-xtensa@...ux-xtensa.org, iommu@...ts.linux-foundation.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/7] arm-nommu: remove the partial
 DMA_ATTR_NON_CONSISTENT support

On Mon, Jun 24, 2019 at 03:23:08PM +0100, Vladimir Murzin wrote:
> On 6/14/19 3:44 PM, Christoph Hellwig wrote:
> > The arm-nommu DMA code supports DMA_ATTR_NON_CONSISTENT allocations, but
> > does not provide a cache_sync operation.  This means any user of it
> > will never be able to actually transfer cache ownership and thus cause
> > coherency bugs.
> 
> By the way, Documentation/DMA-attributes.txt doesn't specify cache_sync() as
> requirement for DMA_ATTR_NON_CONSISTENT it only states that it is responsibility
> of the driver to have all the correct and necessary sync points.

True.  dma_cache_sync has always been a rather odd interface, as it
doesn't specify in what direction we need to sync and doesn't
participate in our ownership protocol.  So my mid-term plan is to kill
it off and replace it with the existing dma_sync_* helpers.  This
series is the first step towards that.