linux-kernel - Re: dma_mask limited to 32-bits with OF platform device

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: <20200212175756.GA6034@lst.de>
Date:   Wed, 12 Feb 2020 18:57:56 +0100
From:   Christoph Hellwig <hch@....de>
To:     Robin Murphy <robin.murphy@....com>
Cc:     Roger Quadros <rogerq@...com>, Christoph Hellwig <hch@....de>,
        Péter Ujfalusi <peter.ujfalusi@...com>,
        Murali Karicheri <m-karicheri2@...com>,
        "Nori, Sekhar" <nsekhar@...com>, "Anna, Suman" <s-anna@...com>,
        stefan.wahren@...e.com, afaerber@...e.de, hverkuil@...all.nl,
        Rob Herring <robh+dt@...nel.org>, devicetree@...r.kernel.org,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        Nishanth Menon <nm@...com>,
        "hdegoede@...hat.com" <hdegoede@...hat.com>
Subject: Re: dma_mask limited to 32-bits with OF platform device

On Wed, Feb 12, 2020 at 02:04:31PM +0000, Robin Murphy wrote:
>> For now, let's say that we limit dma-ranges to 4GB size. with "dma-ranges 
>> = <0x00000000 0x00000000 0x1 0x00000000>;"
>> Then, dma_bus_limit is set correctly to 0xffffffff, SATA driver sets masks 
>> to 64-bit as IP supports that.
>>
>> [   13.306847] ahci 4a140000.sata: dma_mask 0xffffffffffffffff, 
>> coherent_mask 0xffffffffffffffff, dma_bus_limit 0xffffffff
>>
>> However, the SATA controller still tries to do DMA above 32-bits.
>> dma_alloc() doesn't seem to be taking dma_bus_limit into account?
>
> Yay ARM LPAE... Peter and Christoph have already been playing whack-a-mole 
> with other bugs under that config - is this with or without SWIOTLB? (and 
> whichever way, does the other work any better?)

Hmm.  ARM LPAE still uses the arm legacy dma alloc coherent, and that
might not be taking the dma_bus_limit into account.  Let me check..