AM623: AM623: RTC driver crash : rtc-ti-k3 2b1f0000.rtc: [Hardware Error]: Erratum i2327/IRQ

Hi Alexis,

Is this a follow-up to an existing E2E thread? If so, can you please provide the link to the thread?

Hi Bin,

Please see this: https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1247816/am623-rtc-driver-crash-rtc-ti-k3-2b1f0000-rtc-hardware-error-erratum-i2327-irq

Thank you.

Best regards,
François.

Thanks Francois,

I am routing the query to our RTC expert for comments.

Hi Nic , 

please one last  question :  as per errata if  you are <1sec   RTC  should be working  . so if customer using RTC only to wake up from time to time  can they use it ?   how to set properly the clock in case is possible ?

sorry for  asking but crucial  to see any solution for the customer  to avoid redesign PCB .

thank you 

BR
Carlo

Hello Carlo,

No problem. Will take a closer look and get back to you soon.

Regards,

Nick

Hi Nic ,

please any news ?   

thank you 

BR

Carlo

Hello Carlo,

Apologies for the continued delay here. I have a meeting scheduled for tomorrow to discuss RTC within the context of wakeup from sleep, and I will respond after that meeting with what I have learned.

Regards,

Nick

Hello Carlo & Alexis,

Questions

What is the boot method being used? And what version of the SDK are you using?

Tentative update

As far as I can tell, the interrupt storm is caused when the RTC does not get unlocked within 1 second during the boot process. The code to unlock the RTC is within the uboot file tiboot3.bin. We are checking separately to see if there was a specific software release that the code was written or updated.

That means that if tiboot3.bin can be loaded and the code can run within 1 second, then (with the information I currently have), we would not expect to see the RTC issue. So the hypothesis is that SD, EMMC, OSPI boot should not run into this issue. However, if the boot files need to be downloaded from somewhere else, like with USB boot, UART, ethernet, etc, then there is a chance that the code does not download and run within 1 second. In that case, the errata will happen.

If we can guarantee that the RTC gets unlocked within 1 second during the boot process, it sounds like the RTC can be used later during runtime (e.g., to wake from sleep).

However, if we CANNOT guarantee that the RTC gets unlocked within 1 second, then the RTC should not be used later during runtime. For waking from sleep, you would want to use an external trigger (e.g., GPIO, UART, etc), or an external RTC. Off the top of my head I am not sure how an external RTC would interact with sleep mode. If that is an avenue you want to explore, I can check with the developers.

Regards,

Nick

This is the commit that adds the uboot fix:
https://git.ti.com/cgit/ti-u-boot/ti-u-boot/commit/?id=fb3474bef0b2a14cfa1295f90ee4eb5337765e83

I see that code starting in the AM62x SDK 8.3.

Regards,

Nick