AM2634: AM263x System Reset Issues After Debugger Connection

Dominic Rath

Part Number: AM2634
Other Parts Discussed in Thread: UNIFLASH, TMDSCNCD263

Tool/software:

Dear TI Team,

we're experiencing some peculiar behavior with system resets on the AM263x when using a debugger:

* **Debugger Not Connected:** When the debugger is not connected to any core of the device, performing a system reset (e.g., through the debugger interface) works as expected. The device boots normally, going through all stages: Boot ROM, SBL, and application image.

* **Debugger Connected:** After connecting the debugger to the first R5F core, subsequent system resets no longer function correctly. The boot process appears to get stuck at some point, without any output on the console. We can only recover by performing a full power cycle. This behavior differs from our experience with similar devices like the AM64xx, where a system reset always initiates a full boot sequence regardless of debugger connection status.

This issue affects both debugging via CCS AND any automation via the debug server or JTAG Uniflash which we'd like to use for production.

Is there a known workaround or specific procedure to perform a clean system reset on the AM263x after the debugger has connected to a core, without resorting to a power cycle?

Best Regards,

Dominic

11 months ago

0 Brennan Hartigan 11 months ago

TI__Genius 10644 points

Hi Dominic,

What debugger are you using to interface with the AM263x? Is this a TI EVM or a custom PCB platform?

Regards,

Brennan

0 Dominic Rath 11 months ago in reply to Brennan Hartigan

Mastermind 7450 points

Hi Brennan,

we're using the AM263x control card (TMDSCNCD263).

Regards, Dominic

0 Sahana H G 11 months ago in reply to Dominic Rath

TI__Expert 5376 points

Hello,

Could you please confirm if you're using the below System Reset option in CCS to perform the reset in both the cases?

Thanks,

Sahana

0 Dominic Rath 11 months ago in reply to Sahana H G

Mastermind 7450 points

Hello Sahana,

yes, that's the System Reset we're using.

If we use that reset after booting a NULL SBL from QSPI we can see the NULL SBL booting again and again.

If we use that reset after booting a NULL SBL from QSPI AND attaching the debugger, the NULL SBL wont run again. This is true even if we detach the core.

Regards,

Dominic

0 Sahana H G 11 months ago in reply to Dominic Rath

TI__Expert 5376 points

Hi Dominic,

Thanks for the confirmation, let me check on this and get back to you by tomorrow.

Thanks,

Sahana

0 Sahana H G 11 months ago in reply to Dominic Rath

TI__Expert 5376 points

Hi Dominic,

I'm consulting some Debug experts on this topic, and it is a Holiday in India tomorrow. So please expect a response on this by early next week.

Regards,

Sahana

0 Sahana H G 11 months ago

TI__Expert 5376 points

Dominic Rath said:
We can only recover by performing a full power cycle

One observation on this is that we can disconnect from the R5F core, and then perform the reset. This makes the boot process to occur correctly.

+1 Sahana H G 11 months ago in reply to Dominic Rath

TI__Expert 5376 points

Hi Dominic,

If the core0 is kept in halted state (after connecting to the core) and system reset is asserted, the boot won't succeed since the debug state will be retained post the reset and ROM code does not execute. Hence, the correct way would be to unhalt the core and then issue system reset, and the boot should work fine. This would result in a clean system reset on the AM263x after the debugger has connected to a core, without resorting to a power cycle. You could incorporate the same during both debugging via CCS AND any automation via the debug server or JTAG Uniflash. Hope this helps.

Thanks,

Sahana

0 Dominic Rath 10 months ago in reply to Sahana H G

Mastermind 7450 points

Hello Sahana,

thanks for looking into this. Resuming before detaching works.

Regards,

Dominic

Arm-based microcontrollers

Arm-based microcontrollers forum

AM2634: AM263x System Reset Issues After Debugger Connection