AM623: Assistance Required for USB DFU Issue

Hello Vikash

We really do not support GP much now, esp with SDK 10.x onwards I do not expect the SDK teams to test/regress GP devices. Can you please evaluate moving to HSFS devices and HSFS EVMs for development. Otherwise you will continue to run into challenges on development and support 

Regarding the GP die, is there a requirement to transition to HS-FS or better? What if customers are content with the GP device? Specifically is there a plan to EOL it before 2034?

If a customer is using GP, can the continue to build with the new SDK and perhaps use old tiboot3.bin/tisbl.bin images in conjunction with the new SDK. Or is this PROHIBITED because the new SDK is tightly integrated with security features of HS-FS or HS-SE hardware?

That is what level of support will GP devices get? And if none should we move to HS-FS? If we don't utilize the Field Securable setting in fielded devices could the devices be bricked later by malware or accidental enabling the FS feature? That is, if we continue to utilize the GP die with the latests 10.x SDK what are the possible downsides. Can you enumerate?

Note: https://software-dl.ti.com/processor-sdk-linux/esd/AM62X/10_00_07_04/exports/docs/linux/Foundational_Components_Migration_Guide.html

Indicates GP devices can still boot signed images, and that HS-FS can boot un-signed images - which leads me to belive that we could move to HS-FS without issue - unless there are potential downsides of bricking. Please expand on this?

RG,

the subject matter expert this is assigned to is out of office this week (Thanksgiving Holiday). Thanks for your patience.

Regards, Andreas

Thanks for the update Andreas. Will look to next week.

Hi Bin Liu,

There seems to be a disconnect between TI sales and TI E2E as they selling the GP device:

https://www.mouser.com/ProductDetail/Texas-Instruments/AM6232ATCGGAALW?qs=IPgv5n7u5QZdMZZnEmvZzA%3D%3D

And earlier work with TI showed longevity out to 2034. 

1. Lets resolve this ASAP.

2. Then lets go back and look at the workarounds I proposed above (if we can use 10.x on GP with workarounds, or if we move to HS-FS the pros and cons of not locking down a field securable device).

Also, we should move this to a new thread. "AM6xx SDK10.0 GP or non high security chip support" ?

Hello RG

We sell GP, but we have gradually stopped testing GP in our test farms , all our EVMs and most of our customers are on HSFS now. So this is not a continuity of supply issue, it is more of a what we test and prefer to support 

The underlying reason is similar to what was explained on AM64 

https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1158018/faq-processor-sdk-am64x-am64x-am243x-sr-2-0-device-type-transition

Subsequent device family like AM62A and AM62P are only offered in HSFS , not GP.

If you are starting a new development we strongly recommend that you migrate to HSFS silicon now, as that gets you best software support for new design and development. 

RG said:

And earlier work with TI showed longevity out to 2034. 

i am not sure what this is referring to , can you clarify where you see this?

If you are asking us to debug by GP is failing with latest SDK , I want to warn you that the support will be best effort basis and may not always be timely  as team predominantly has HSFS EVMs and SDK testing is now predominantly done with HSFS EVMs too.

Hope that clarifies. Let me know if you have more questions. 

Regards

Mukul 

Hi Mukul, 

Just want to confirm you checked that link to a distributor selling GP versions of AM62 parts?

I'll look at the HS-FS documents regarding what Field Securable actually does, and if there are dangers related to not securing the device. Time to review.

I would have thought TI would have issued AM6232B* part number nomenclature to distinguish between the AM6232A* to indicate die & functionality change between GP and HSFS features (sort of what happened with sister product line AM6442A* and AM6442B* as of October 2023)

RG

Sorry if I was not clear, we are selling GP version , Ti or distributors , as we have customers who are in production with it so we have not stopped making these parts etc. 

However we are recommending that if you are starting a new design and starting development, please use HSFS instead of GP, as this is what our latest software is primarily tested on and it is what is on our development EVMs (our original EVM were on GP, but those have been phased out and no longer available) 

HSFS devices behave like GP and not needed to be secured if you do not need secure boot in your application. There are some minor migration aspects, that are listed in the SDK docs, but if you are using latest SDK and latest EVM, default experience is on HSFS. 

Hope this helps. 

Jim

Thanks for your input. Hopefully you chiming in on the thread does not further confuse RG. 

FYI we also have an AM62A family of devices. 

https://www.ti.com/product/AM62A3 (this is different then AM62) 

Nomenclature table clarifies GP , HS FS please use the datasheet (sharing AM64 link for you)

https://www.ti.com/document-viewer/AM6442/datasheet#GUID-CD6E613A-70C1-4449-8B74-B692CA434DE6/TITLE-SPRSP56INT_SPRS858_PACK_00005

Understand that the "EVMs" have been named to differentiate GP vs HSFS

SK-AM62 = GP with discrete power. 

SK-AM62B-P1 - HSFS and PMIC 

I am hoping RG has the clarification needed from the guidance I provided 

I have two questions,

1. One thing I don't understand ( I get 10.x not supporting GP devices) in later revisions. 

But if SSBL (tspl and u-boot) are both signed, and the customer can code u-boot to either enforce root certificates and signatures however they want, why would Linux built by 10.x + SDK not run/launch? Has the latest 10.x Linux kernel had modules added that check for security hardware or some such?

2. I need to read more, but do most people secure the FS devices in factory to enforce valid image loads? Ok 3 questions, and 3. Are fuses blown when securing FS devices and if so If the boot0/boot1 or Filesys DOS or SD card are corrupted wiping out the tiboot3/tispl and u-boot images. Can you only load these images with the same exact keys or are there internal keys that allow roll forward but no roll backs of FSBL code? I know this is a complicated question. The page : https://dev.ti.com/tirex/explore/node?node=AAz9Dxht0WlNjAge02XsaQ has A, P, X references in the links no B? 

3b. https://dev.ti.com/tirex/explore/content/am62x_academy_10_00_00_00/am62x_academy_10_00_00_00/source/security/ti-overview.html If you ship HS-FS without performing factory security, can a malicious backup boot load code that then sets the fuses or security mode?

Thanks,

Robert

Thanks Mukul, I now see that part numbering nomenclature differs between AM62x and AM64x (and AM62Ax for that matter). Brian Berner had a slide set that showed differences between AM62x and AM64x last year; now with the other AM62* variants, it's confusing -- Brian might need a "Rosetta Stone" to map out all "AM6*"

regards Jim

Any quick responses to questions above?