[PATCH 3/3] bootm: add initial FIT support

Yegor Yefremov yegorslists at googlemail.com
Tue Jan 5 05:05:28 PST 2016

On Tue, Jan 5, 2016 at 12:54 PM, Marc Kleine-Budde <mkl at pengutronix.de> wrote:
> On 01/05/2016 11:40 AM, Yegor Yefremov wrote:
>> On Tue, Jan 5, 2016 at 11:32 AM, Marc Kleine-Budde <mkl at pengutronix.de> wrote:
>>> On 01/05/2016 11:28 AM, Yegor Yefremov wrote:
>>>> Hi Marc,
>>>> thanks for reposting the patches.
>>>> On Tue, Jan 5, 2016 at 9:11 AM, Marc Kleine-Budde <mkl at pengutronix.de> wrote:
>>>>> From: Jan Luebbe <jlu at pengutronix.de>
>>>>> This implementation is inspired by U-Boot's FIT support. Instead of
>>>>> using libfdt (which does not exist in barebox), configuration signatures
>>>>> are verified by using a simplified DT parser based on barebox's own
>>>>> code.
>>>>> Currently, only signed configurations with hashed images are supported,
>>>>> as the other variants are less useful for verified boot. Compatible FIT
>>>>> images can be created using U-Boot's mkimage tool.
>>>> What about unsigned images?
>>> That's not our use case. We use plain zImages instead.
>> The solution would be to introduce an option like in U-Boot?
>> This option enables signature verification of FIT uImages,
>> using a hash signed and verified using RSA. If
>> CONFIG_SHA_PROG_HW_ACCEL is defined, i.e support for progressive
>> hashing is available using hardware, RSA library will use it.
>> See doc/uImage.FIT/signature.txt for more details.
> Technically possible, but I'm not sure what are the benefits of using
> fit images, if you don't need signatures. barebox implements
> freedesktop.org's bootspec and this is IMHO the way to go.

For me FIT is just a way to have a kernel and a bunch of device tree
blobs in one file. Signed or not signed is an option for me. Just like
U-Boot implements it. This is user responsibility.

In my use case I just read device ID from EEPROM, load my
kernel-fit.itb and select needed DTB via this ID. This way I have only
one SD card image, that can be run on more, than 10 different devices
using the same core module.

>>>> I also get: unsupported algo crc32
>>>> Is it intended to be supported?
>>> Not for our usecase - feel free to add crc32 support.
>> OK.
>> But what about FIT configuration selection syntax?
> What's this?

Have you seen my comments to this patch regarding
fit_open_configuration() routine?



More information about the barebox mailing list