linux-kernel - Re: [PATCH 2/8] spi: spi-s3s64xx: Add S3C64XX_SPI_QUIRK_CS

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: <dleftjd03lzj9d.fsf%l.stelmach@samsung.com>
Date:   Thu, 20 Aug 2020 12:47:26 +0200
From:   Lukasz Stelmach <l.stelmach@...sung.com>
To:     Mark Brown <broonie@...nel.org>
Cc:     Krzysztof Kozlowski <krzk@...nel.org>,
        Kukjin Kim <kgene@...nel.org>, Andi Shyti <andi@...zian.org>,
        linux-spi@...r.kernel.org, linux-samsung-soc@...r.kernel.org,
        linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
        m.szyprowski@...sung.com, b.zolnierkie@...sung.com
Subject: Re: [PATCH 2/8] spi: spi-s3s64xx: Add S3C64XX_SPI_QUIRK_CS_AUTO for
 Exynos3250

It was <2020-08-19 śro 20:38>, when Mark Brown wrote:
> On Wed, Aug 19, 2020 at 03:01:21PM +0200, Lukasz Stelmach wrote:
>> It was <2020-08-19 śro 14:39>, when Krzysztof Kozlowski wrote:
>
>>> There is here no commit msg, no explanation.
>
>> As I wrote in the cover letter, this and previous commits make things
>> work on Exynos3250 (ARTIK5 precisely). I can't explain why. I read
>> everything I could about this HW and there were no details about
>> automatic CS handling other than how to turn it on and off.
>
> What is similar about those other SoCs - could you be more specific
> here, or what goes wrong if you don't set this?

Without the quirk set DMA transfers longer than 512 bytes fail. They
simply stop and hit the timeout with a few (<20) bytes pending.

As far as I can tell the SPI controller is the same in different Exynos
SoCs.

> The auto mode (or at least the auto mode that was on the Exynos7) is
> not compatible with many SPI devices if the controller chip select is
> actually in use, the quirk was added for controllers that just don't
> have the manual mode.

According to the manual the auto mode makes the controller toggle CS
between SPI packets (bytes?).

I didn't have any problem transferring data (>512 bytes) from the SPI
device in the polling mode. Only the DMA caused problems.
> See also:
>
>   https://lore.kernel.org/linux-spi/CAAgF-BfGwcNzMx0meFVkJqNMTbQ4_PP1PZ3i6edOm6U3bc26_Q@mail.gmail.com/
>
> for an explanation of the quirk.
>

>> CS can also be controlled automatically by setting AUTO_N_MANUAL to 1
>> in CS_CFG. When it is auto CS automatically toggles between packet to
>> packet. NCS_TIME_COUNT in CS_CFG controls the inactive period. The
>> driver by default uses manual mode. But on exynos7 the manual mode is
>> removed.

I *suspect* that the automatic CS toggling between packets gives better
(?) synchronisation between the SPI device and the controller's
internals and prevents some kind of a deadlock inside the
controller. These are just speculations.

-- 
Łukasz Stelmach
Samsung R&D Institute Poland
Samsung Electronics

Download attachment "signature.asc" of type "application/pgp-signature" (488 bytes)