by Marek Vasut

[permalink] [raw]

Subject: Re: [PATCH v2 00/14] Introduce STM32MP1 RCC in secured mode

On 3/11/21 2:15 PM, Alexandre TORGUE wrote:
> Hi Marek

Hello Alexandre,

> On 3/11/21 12:43 PM, Marek Vasut wrote:
>> On 3/11/21 9:08 AM, Alexandre TORGUE wrote:
>>> Hi ALex
>>
>> Hello everyone,
>>
>> [...]
>>
>>>> Subject: Re: [PATCH v2 00/14] Introduce STM32MP1 RCC in secured mode
>>>>
>>>> On 1/26/21 3:01 AM, [email protected] wrote:
>>>>> From: Gabriel Fernandez <[email protected]>
>>>>>
>>>>> Platform STM32MP1 can be used in configuration where some clocks and
>>>>> IP resets can relate as secure resources.
>>>>> These resources are moved from a RCC clock/reset handle to a SCMI
>>>>> clock/reset_domain handle.
>>>>>
>>>>> The RCC clock driver is now dependent of the SCMI driver, then we have
>>>>> to manage now the probe defering.
>>>>>
>>>>> v1 -> v2:
>>>>> - fix yamllint warnings.
>>>>
>>>> Hi Gabriel,
>>>>
>>>> I don't have much clout with the maintainers, but I have to NAK this
>>>> series
>>>> after finding major breakage.
>>>>
>>>> The problem with series is that it breaks pretty much every board it
>>>> touches.
>>>> I have a DK2 here that I'm using for development, which no longer
>>>> boots with
>>>> this series applied.
>>>>
>>>> The crux of the matter is that this series assumes all boards will
>>>> boot with an
>>>> FSBL that implements a very specific SCMI clock tree. This is major ABI
>>>> breakage for anyone not using TF-A as the first stage bootloader.
>>>> Anyone
>>>> using u-boot SPL is screwed.
>>>>
>>>> This series imposes a SOC-wide change via the dtsi files. So even
>>>> boards that
>>>> you don't intend to convert to SCMI will get broken this way.
>>>> Adding a -no-scmi file that isn't used anywhere doesn't help things.
>>>
>>> You are right. We mainly take care about NO ST (DH/...) boards, but
>>> not really about current usage
>>> Of our stm32 boards. Several options exist:
>>
>> Since a lot of people benefit from the good upstream support for the
>> MP1 _and_ keep updating their machines to get the latest fixes, it is
>> very important to keep the current usage working.
>>
>>> 1- Break the current ABI: as soon as those patches are merged,
>>> stm32mp157c-dk2.dtb will impose to use
>>> A tf-a for scmi clocks. For people using u-boot spl, the will have to
>>> create their own "no-secure" devicetree.
>>
>> NAK, this breaks existing boards and existing setups, e.g. DK2 that
>> does not use ATF. >
>>> 2-As you suggest, create a new "secure" dtb per boards (Not my wish
>>> for maintenance perspectives).
>>
>> I agree with Alex (G) that the "secure" option should be opt-in.
>> That way existing setups remain working and no extra requirements are
>> imposed on MP1 users. Esp. since as far as I understand this, the
>> "secure" part isn't really about security, but rather about moving
>> clock configuration from Linux to some firmware blob.
>>
>>> 3- Keep kernel device tree as they are and applied this secure layer
>>> (scmi clocks phandle) thanks to dtbo in
>>> U-boot.
>>
>> Is this really better than
>> #include "stm32mp15xx-enable-secure-stuff.dtsi"
>> in a board DT ? Because that is how I imagine the opt-in "secure"
>> option could work.
>
> The dtbo usage could avoid to add another st board (actually a secure
> config) in arch/arm/boot/dts.

It isn't even a board, it is a configuration. Could you detect this
secure/non-secure state at runtime, have both clock options in the DT,
and handle it accordingly ? That might be even better option.