[PATCH v8 06/13] binman: Support new op-tee binary format

[Jerome Forissier]

On 1/7/23 19:55, Simon Glass wrote:
> Hi Jerome,
> 
> On Thu, 22 Dec 2022 at 15:20, Jerome Forissier
> <jerome.forissier at linaro.org> wrote:
>>
>>
>>
>> On 12/22/22 21:23, Simon Glass wrote:
>>> (dropping the two bounces from cc)
>>>
>>> On Thu, 22 Dec 2022 at 13:18, Simon Glass <sjg at chromium.org> wrote:
>>>>
>>>> Hi Jerome,
>>>>
>>>> On Thu, 22 Dec 2022 at 08:36, Jerome Forissier
>>>> <jerome.forissier at linaro.org> wrote:
>>>>>
>>>>>
>>>>>
>>>>> On 12/22/22 00:07, Simon Glass wrote:
>>>>>> OP-TEE has a format with a binary header that can be used instead of the
>>>>>> ELF file. With newer versions of OP-TEE this may be required on some
>>>>>> platforms.
>>>>>>
>>>>>> Add support for this in binman. First, add a method to obtain the ELF
>>>>>> sections from an entry, then use that in the FIT support. We then end up
>>>>>> with the ability to support both types of OP-TEE files, depending on which
>>>>>> one is passed in with the entry argument (TEE=xxx in the U-Boot build).
>>>>>
>>>>> So, with:
>>>>>
>>>>> BL31=/path/to/bl31.elf TEE=/path/to/tee.bin make -C u-boot \
>>>>>   CROSS_COMPILE=aarch64-linux-gnu- CC=aarch64-linux-gnu-gcc \
>>>>>   HOSTCC=gcc BINMAN_DEBUG=1 BINMAN_VERBOSE=4
>>>>>
>>>>>
>>>>> ...I get:
>>>>>
>>>>> Entry '/binman/simple-bin/fit/images/@tee-SEQ/tee-os' marked absent: uses v1 format which must be in a FIT
> 
> OK, this is expected. It means that the contents are not provided. I
> will add some more comments.
> 
> We can refine this later, but we are a year into this conversion of
> rk3399 to binman and I really want to land this right away before I
> lose interest and it sits here for another year.

I commented that in my other reply. That would break my use case.

> Do you have an influence on OP-TEE?

I am one of the core maintainers ;)

> I think the way they are adding a
> header here is a really back idea. We should look at packaging these
> blobs in a FIT complete with all the metadata that is needed, with a
> standard schema.

Isn't that what this patch series is trying to accomplish? Using a
different tool (binman instead of a custom python script) to package a
blob into a FIT? As I said in my other reply, tee.bin is the output of
the OP-TEE build, it has to be loaded in memory at a specific address,
that you can discover be reading the binary itself -- no need for
metadata... How infrequent is such a use case for a bootloader? to deal
with? :O 

Thanks,
-- 
Jerome

>>>>>
>>>>>
>>>>> More complete log at https://pastebin.com/UZzZeicQ
>>>>
>>>> Thanks.
>>>>
>>>> Is this file in the v1 binary format (with the custom header),
>>
>> Yes
>>
>>>> or is
>>>> is a plain binary file? At present only the former is supported, as I
>>>> thought that it can only be the elf or the v1 binary format these
>>>> days?
>> Correct
>>
>>>> Actually can you please send me the tee.bin ?
>>
>> Sure, here you go: https://ufile.io/f/ex5oy
>>
>> I have added bl31.elf too, although I suppose it is handled properly
>> being an ELF file.
> 
> It seems to work for me.
> 
> [..]
> 
> Regards,
> Simon