• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 30 subscribers
  • Views 1186 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

ESM1.90 - SYS - Register Soft Error

Peter Fidelman
Prodigy 200 points
Other Parts Discussed in Thread: RM57L843

Hi all,

Within the span of a few minutes, I experienced a few Group1 - Channel90 - "SYS - Register Soft Errors" on my RM57L843 HDK.  It was a transient event and has not recurred.


What is a register soft error?  Researching them, it seems to mean a temporary single-bit-flip in the SYS module - is that correct?  Did I have a run-in with some cosmic rays?

Peter

  • 0
    TI__Guru**** 191906 points

    Hi Peter,

      Yes, your understanding is correct. In RM57L device there are registers which will self-correct. These registers from the programmer model perspective are single bits but in actual implementation they are multi-bits. For example, the CPURSTCR has only one visible bit from a programmer's model. This bit is used to generate a CPU reset. However, there are actually several physical bits implemented with self-correction logic such that when one bit flipped it will signal an error to the ESM and also correct it back. This is a safety diagnostic to prevent and detect any unintentional generation of CPU reset due to soft errors. I'm surprised you are able to see this error in a normal bench environment. Also for your information, the SYSECR register and the CDDIS register are also implemented with self-correcting logic in the SYS module.

  • 0 in reply to Charles Tsai
    Prodigy 200 points

    Okay, thanks for confirming.

  • 0 in reply to Charles Tsai
    Prodigy 200 points
    One more follow-up question:

    I am still getting ESM 1.90's from time to time. They are not readily reproducible, but I normally get at least a few a day. I have confirmed this behavior on multiple boards, both HDK's and custom PCB's, using RM57's from SIlicon Revisions A and B.

    This is indeed a benchtop environment with no (known?) radiation sources nearby to cause legitimate bit flips.

    Is there any additional information available, such as an "error details" register that I can read that would specifically tell me which SYS register(s) flipped bits to trigger the ESM1.90? I understand that in the case of a SRAM, CPU, flash or interconnect fault, I could read the fault address from the EPC, but the EPC's section of spnu562 does not mention any diagnostics of the SYS module.

    Basically, I'm looking for ways to confirm that the ESM1.90 indicates a real soft error in the SYS module (not an error in the diagnostics), and then identify any patterns in the fault details that might help me find the root cause.

    Any suggestions?