linux-kernel - Re: [PATCH] spi: mediatek: Manually set dma_ops for spi

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:   Thu, 26 Jan 2017 09:26:13 -0800
From:   Dmitry Torokhov <dtor@...omium.org>
To:     Mark Brown <broonie@...nel.org>
Cc:     Daniel Kurtz <djkurtz@...omium.org>,
        Dmitry Torokhov <dtor@...omium.org>,
        Robin Murphy <robin.murphy@....com>,
        Leilk Liu <leilk.liu@...iatek.com>,
        open list <linux-kernel@...r.kernel.org>,
        "open list:SPI SUBSYSTEM" <linux-spi@...r.kernel.org>,
        Matthias Brugger <matthias.bgg@...il.com>,
        "moderated list:ARM/Mediatek SoC support" 
        <linux-mediatek@...ts.infradead.org>,
        Guenter Roeck <groeck@...omium.org>,
        "moderated list:ARM/Mediatek SoC support" 
        <linux-arm-kernel@...ts.infradead.org>
Subject: Re: [PATCH] spi: mediatek: Manually set dma_ops for spi_master device

On Thu, Jan 26, 2017 at 9:07 AM, Mark Brown <broonie@...nel.org> wrote:
> On Fri, Jan 27, 2017 at 12:52:15AM +0800, Daniel Kurtz wrote:
>> On Fri, Jan 27, 2017 at 12:35 AM, Dmitry Torokhov <dtor@...omium.org> wrote:
>
>> > Nobody doing DMA likes unaligned buffers. In fact, memory used for DMA
>> > must be cacheline aligned, or you going to have trouble. What
>> > component supplies unaligned buffers to SPI core for transfers?
>
>> cros-ec-spi
>
> That needs fixing then - I keep thinking we should add a warning to the
> core for this, especially now we do DMA mapping in the core.

OTOH the full buffer used by cros ec *is* DMA-safe (allocated with
kmalloc), it is that cors ec uses it at offset while transferring data
piece by piece. Do we want to support this?