AM625: 08.06.00.42 Release: ti-k3-secdev: Invalid revision

Nate Drude said:

$ MACHINE=am62xx-evm bitbake var-default-image

Correction: $ MACHINE=am62xx-evm bitbake tisdk-default-image

Andreas Dannenberg Thank you for the fast reply, and the detailed response to my questions.

I did a fresh build following your steps, but I still get the same error. I tried using both export TI_SECURE_DEV_PKG=./core-secdev-k3 (from your steps) and export TI_SECURE_DEV_PKG=`pwd`/core-secdev-k3 (from https://software-dl.ti.com/processor-sdk-linux/esd/AM62X/08_06_00_42/exports/docs/linux/Overview_Building_the_SDK.html#introduction)

$ MACHINE=am62xx-evm bitbake -k tisdk-default-image
...
ERROR: mc:k3r5-gp:ti-k3-secdev-native-git-r0 do_fetch: Fetcher failure: Unable to find revision d14fbab5ccb21ffeb010ce57792c56a1308ed314 in branch master even from upstream
ERROR: mc:k3r5-hs-fs:ti-k3-secdev-native-git-r0 do_fetch: Fetcher failure: Unable to find revision d14fbab5ccb21ffeb010ce57792c56a1308ed314 in branch master even from upstream
ERROR: mc:k3r5-hs-se:ti-k3-secdev-native-git-r0 do_fetch: Fetcher failure: Unable to find revision d14fbab5ccb21ffeb010ce57792c56a1308ed314 in branch master even from upstream
...
$ ls
arago-tmp-external-arm-glibc  bitbake-cookerdaemon.log  buildhistory  cache  conf  core-secdev-k3  sstate-cache

$ printenv TI_SECURE_DEV_PKG
./core-secdev-k3

# I also retried with:

$ printenv TI_SECURE_DEV_PKG
/home/nate/nvme_4tb/bsps/am62x/08.06.00.42.ti/tisdk/build/core-secdev-k3

Can you test this on your side?

Thanks,

Nate

Andreas Dannenberg said:

As of SDK 8.6 for AM62x the default build is for HS-FS devices (rather than GP devices)

Will images built for the HS-FS SoC boot on the GP SoC?

If not, what changes to the BSP (U-Boot, Linux, and meta layers) are required to build an image for GP devices?

Thanks, Nate

Hi Nate,

Nate Drude said:

Will images built for the HS-FS SoC boot on the GP SoC?

You will need a specific tiboot3.bin initial binary for GP devices, that you will need to swap out, see https://software-dl.ti.com/processor-sdk-linux/esd/AM62X/latest/exports/docs/linux/Foundational_Components_Migration_Guide.html   The rest of the boot chain will work no matter GP or HS-FS device variant. Note that there may be some warnings during boot when booting the new v8.6-based images on GP devices regarding secure headers getting discarded but those can be ignored.

Nate Drude said:

If not, what changes to the BSP (U-Boot, Linux, and meta layers) are required to build an image for GP devices?

As far as I know we don't have those steps documented yet. I did try to derive some steps a while ago but it is not as trivial as it may seem and I ran out of time back then. Should you experiment with it one word of caution here. I was told that it is possible to badly confuse the shared state cache doing certain changes around how the multi-config mc:k3r5:ti-sci-fw artifacts are built, so it will be good in case of issues to completely clean/wipe out any build related folders and caches.

Regards, Andreas

Andreas Dannenberg said:

Nate Drude said:

Will images built for the HS-FS SoC boot on the GP SoC?

You will need a specific tiboot3.bin initial binary for GP devices, that you will need to swap out, see https://software-dl.ti.com/processor-sdk-linux/esd/AM62X/latest/exports/docs/linux/Foundational_Components_Migration_Guide.html   The rest of the boot chain will work no matter GP or HS-FS device variant. Note that there may be some warnings during boot when booting the new v8.6-based images on GP devices regarding secure headers getting discarded but those can be ignored.

Nate Drude said:

If not, what changes to the BSP (U-Boot, Linux, and meta layers) are required to build an image for GP devices?

As far as I know we don't have those steps documented yet. I did try to derive some steps a while ago but it is not as trivial as it may seem and I ran out of time back then. Should you experiment with it one word of caution here. I was told that it is possible to badly confuse the shared state cache doing certain changes around how the multi-config mc:k3r5:ti-sci-fw artifacts are built, so it will be good in case of issues to completely clean/wipe out any build related folders and caches.

Thanks for clarifying and for the word of caution.

In case you missed my previous message, can you please try to build the 08.06.00.42 SDK? It still fails for me following your steps, even after a fresh build. I assume if the CI is passing, I must be doing something wrong. Can you check on your side?

Thanks, Nate

Hi Nate Drude,

The export command should be,

export TI_SECURE_DEV_PKG_K3=`pwd`/core-secdev-k3


Note: It's TI_SECURE_DEV_PKG_K3 and not TI_SECURE_DEV_PKG.

Chirag Shilwant 

Are both TI_SECURE_DEV_PKG and TI_SECURE_DEV_PKG_K3 required, or only TI_SECURE_DEV_PKG_K3?

Is the documentation here wrong, or am I looking at the wrong documentation? https://software-dl.ti.com/processor-sdk-linux/esd/AM62X/08_06_00_42/exports/docs/linux/Overview_Building_the_SDK.html?highlight=bitbake

Thanks!

Nate,

I kicked off a build and saw the same issue. Looks like the documentation is wrong. We should only define TI_SECURE_DEV_PKG_K3, and probably using an absolute path, NOT a relative one. On the other hand TI_SECURE_DEV_PKG  is used internally (and hence shown in the error messages, which is misleading/confusing unfortunately), but if you tear the recipes apart (and look at BB_ENV_EXTRAWHITE, which really determines what can be passed into the build environment) you'll see how it works.

Sorry for having trouble with this, this should be very straightforward. Will get our docs updated asap.

Regards, Andreas

Andreas Dannenberg Chirag Shilwant No problem, thanks for the fast support!

I'm starting a fresh build to test on my side to confirm.

In general, what is the reason for manually cloning core-secdev-k3.git and exporting TI_SECURE_DEV_PKG_K3, instead of using a recipe to do this?

Hi Nate,

Nate Drude said:

In general, what is the reason for manually cloning core-secdev-k3.git and exporting TI_SECURE_DEV_PKG_K3, instead of using a recipe to do this?

Some background on this. We at Sitara are transitioning to HS-FS devices as being our "default" devices, moving away from GP devices over time. Then, expectation is for customers using HS-FS devices to eventually transition to HS-SE devices (by way of programming the secure keys), and at that time TI_SECURE_DEV_PKG_K3 will be used to point to the associated custom set of private keys. So always requiring setting TI_SECURE_DEV_PKG_K3 kind of forces a consistent build flow/process across HS-FS and HS-SE while one transitions to their final production devices/environment.

Regards, Andreas

Hi Andreas, my build just finished successfully also. Thanks for your support!

For future reference, the build steps in the official documentation have since gotten updated/corrected, see https://software-dl.ti.com/processor-sdk-linux/esd/AM62X/08_06_00_42/exports/docs/linux/Overview_Building_the_SDK.html#build-steps

Regards, Andreas