linux-kernel - Re: [PATCH v2 06/15] dt-bindings: ufs: Add "max-device-gear" property for UFS 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]

Date:   Thu, 3 Nov 2022 11:23:17 -0400
From:   Krzysztof Kozlowski <krzysztof.kozlowski@...aro.org>
To:     Manivannan Sadhasivam <manivannan.sadhasivam@...aro.org>
Cc:     martin.petersen@...cle.com, jejb@...ux.ibm.com,
        andersson@...nel.org, vkoul@...nel.org,
        krzysztof.kozlowski+dt@...aro.org, konrad.dybcio@...ainline.org,
        robh+dt@...nel.org, quic_cang@...cinc.com,
        linux-arm-msm@...r.kernel.org, devicetree@...r.kernel.org,
        linux-kernel@...r.kernel.org, linux-phy@...ts.infradead.org,
        linux-scsi@...r.kernel.org, dmitry.baryshkov@...aro.org,
        ahalaney@...hat.com
Subject: Re: [PATCH v2 06/15] dt-bindings: ufs: Add "max-device-gear" property
 for UFS device

On 03/11/2022 08:28, Manivannan Sadhasivam wrote:
> On Wed, Nov 02, 2022 at 03:09:50PM -0400, Krzysztof Kozlowski wrote:
>> On 31/10/2022 14:02, Manivannan Sadhasivam wrote:
>>> The maximum gear supported by the UFS device can be specified using the
>>> "max-device-gear" property. This allows the UFS controller to configure the
>>> TX/RX gear before starting communication with the UFS device.
>>
>> This is confusing. The UFS PHY provides gear capability, so what is the
>> "device" here? The attached memory? How could it report something else
>> than phy?
>>
> 
> This is the norm with any storage protocol, right? Both host and device
> (memory) can support different speeds and the OEM can choose to put any
> combinations (even though it might not be very efficient).
> 
> For instance,
> 
> PHY (G4) -> Device (G3)

Yes and look at MMC - no need to define "max mode" supported by eMMC.
You define the modes supported by controller but the memory capabilities
are being autodetected and negotiated.

> 
> From the host perspective we know what the PHY can support but that's not the
> same with the device until probing it. And probing requires using a minimum
> supported gear. For sure we can use something like G2/G3 and reinit later but
> as I learnt, that approach was rejected by the community when submitted
> by Qualcomm earlier.

It should be then referenced somewhere as it might be a reason to accept
the property.

> 
>> The last sentence also suggests that you statically encode gear to avoid
>> runtime negotiation.
>>
> 
> Yes, the OEM should know what the max gear speed they want to run, so getting
> this info from DT makes sense.

Not really if it is auto-detectable. Just because things are static is
not the sole reason to put them into DT. The reason is - they are not
detectable by OS/firmware thus we must have them in DT to be able to
know it.



Best regards,
Krzysztof