linux-kernel - Re: [PATCH] sdhci: Add quirk and device tree parameter to force SD test mode

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] [day] [month] [year] [list]

Message-ID: <bd59af70-ebd8-2c92-e6ab-e1f0bf9acce7@intel.com>
Date:   Wed, 24 Aug 2016 09:16:48 +0300
From:   Adrian Hunter <adrian.hunter@...el.com>
To:     Zach Brown <zach.brown@...com>
Cc:     ulf.hansson@...aro.org, mark.rutland@....com, robh+dt@...nel.org,
        linux-mmc@...r.kernel.org, devicetree@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] sdhci: Add quirk and device tree parameter to force SD
 test mode

On 24/08/16 01:14, Zach Brown wrote:
> On Tue, Aug 23, 2016 at 09:22:29AM +0300, Adrian Hunter wrote:
>> On 23/08/16 01:55, Zach Brown wrote:
>>> From: Jaeden Amero <jaeden.amero@...com>
>>>
>>> On some devices, CD is broken so that we must force the SDHCI into test
>>> mode and set CD, so that it always detects an SD card as present.
>>>
>>> In order to get a device with broken CD working, we had previously
>>> always set the SDHCI into test mode. Unfortunately, this had the side
>>> effect of making all SD cards used with our Linux kernels undetectable
>>> and non-removable.
>>>
>>> By making this "SD test mode" setting optional via a quirk, we can avoid
>>> this side effect for devices other than the device with broken CD.
>>> Additionally, we add a device parameter to sdhci-pltfm to allow all
>>> SDHCI drivers to enable this quirk.
>>
>> Generally new quirks are not acceptable, but I don't see how test mode helps
>> very much since you still don't get any card detection events.  If you
>> really need test mode, please explain more about how it helps (as opposed to
>> polling for example).
>>
>>
> 
> Polling doesnt work in our case, since the issue lies with the SD controller
> itself. The SD controller in xilinx zynq devices requires the CDn signal to
> work. http://www.xilinx.com/support/answers/61064.html We have a situation 
> where the controller requires the signal, but we can't use any of our pins to
> supply it. With test mode we can get the controller into a state where it works
> despite not having the signal.
> 

Doesn't sound like something we need in sdhci.c.  Have you considered adding
it to the driver, perhaps using the ->reset() callback?